X SIMPOSIO

IBEROAMERICANO
SOBRE PROTECCIÓN
DE SISTEMAS ELÉCTRICOS

DE POTENCIA

MEMORIA TÉCNICA
Monterrey. México. Mayo 2011
 X SIMPOSIO
IBEROAMERICANO
SOBRE PROTECCIÓN
DE SISTEMAS ELÉCTRICOS

DE POTENCIA
Mayo de 2011 Monterrey, Nuevo León. México

Lunes 23 de mayo
Operation and Design of a Protection Relay for Transformer Condition Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. T. Yip, S. Hosseini, M. Moscoso, G. J. Lloyd, y K. Liu (Reino Unido) 1
Improvements to Transformer Differential Protection—Design and Test Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. T. Yip, M. Moscoso, G. Lloyd, K. Liu, y Z. Wang (Reino Unido) 9
Nuevo Algoritmo Para Protección Diferencial de Transformadores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. Davila (Canadá) 17
Modern Line Current Differential Protection Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. Miller, J. Burger, N. Fischer, y B. Kasztenny (Estados Unidos) 29
Smarter Time Sync: Applying the IEEE PC37.238 Standard to Power System Applications . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Brunner, y G. S. Antonova (Estados Unidos) 55
An Experimental Analysis of High-Speed-Distance Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Dzienis, M. Kereit, J. Blumschein, M. Claus (Alemania) 65
Medición de la Impedancia de Línea y Tierra para una Estimación Correcta del Factor de Tierra (K0) en Líneas
de Transmisión y Mejora de la Confiabilidad de la Protección de Distancia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Gutiérrez, y M. Krugger (Estados Unidos, Austria) 73
Empleo de Nomogramas para Localización de Fallas en Redes de Distribución . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. González, A. Martínez, R. Mejía, y J. J. Mier (México) 81
Real-Time Multiterminal Fault Location System for Transmission Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Y. Gong, M. Mynam, A. Guzmán, G. Benmouyal, y B. Shulim (Estados Unidos) 89
Protección de Líneas de Transmisión de Doble Circuito con Relevadores de Sobre-Corriente . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Nájera, y D. Sebastián (México) 95
Metodología para el Análisis de Perturbaciones en Sistemas de Potencia Mediante ATPDRAW—Caso de Puente
Abierto en la Línea Esmeralda–Yumbo 2 a 230 kV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. A. Tejada, y S. S. Urbaez (Colombia) 103
Continúa en la página v. . .

iii
X SIMPOSIO IBEROAMERICANO SOBRE PROTECCIÓN DE SISTEMAS ELÉCTRICOS DE POTENCIA

UNIVERSIDAD AUTÓNOMA DE NUEVO LEÓN

Dr. Jesús Ancer Rodríguez, Rector

FACULTAD DE INGENIERÍA MECÁNICA Y ELÉCTRICA

M. C. Esteban Báez Villarreal, Director

COMISIÓN FEDERAL DE ELECTRICIDAD

Mtro. Antonio Vivanco Casamadrid, Director General
Ing. José Abel Valdéz Campoy, Director de Operación
Ing. Noe Peña Silva, Subdirector de Transmisión
Ing. Luis Carlos Hernández Ayala, Subdirector de Generación
Ing. Jesús Eliseo Ramos, Subdirector de Distribución
Ing. Eduardo Meraz Ateca, Subdirector del Centro Nacional de Control de Energía
Ing. Román Ramírez Rodríguez, Coordinador de Protecciones, Comunicación y Control

COMITÉ ORGANIZADOR
Dr. Ernesto Vázquez Martínez, Presidente
Ing. José Abel Valdéz Campoy, Secretario Ejecutivo
Dr. Arturo Conde Enríquez, Responsable, Programa Técnico
Dr. Manuel Antonio Andrade Soto, Responsable, Logística
Dr. Marco Tulio Mata Jiménez, Logística
Dra. Gina María Idárraga Ospina, Responsable, Exposición Industrial
Ing. Ramiro Patiño Bedolla, Exposición Industrial
Ing. Sergio David González Cantú, Exposición Industrial
Ing. Gerardo manuel Robledo Leal, Exposición Industrial

c Copyright
Memoria del X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia (ISBN XXXX-XXXX). ˆ
2011 por la Universidad Autónoma de Nuevo León, todos los derechos reservados.
ISBN XXXX-XXXX
. . . continúa de la página iii
Martes 24 de mayo
Protecting Distribution Feeders for Simultaneous Faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. Betanzos, H. E. Lemus, E. Alcázar, D. Sánchez, y H. J. Altuve (México) 113
Communications for the Smart Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Adamiak, y P. Sanza (Estados Unidos) 123
Wide Area Monitoring with Phasor Measurement Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Wache, y H.-J. Herrmann (Alemania) 133
Implementación de la Smartgrid en Redes de Distribución Existentes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Arzuaga, y R. Quintanilla (España) 137
Esquema de Proteccion y Control para la Interconexion en 13.8 kV del Permisionario Mexico-Carbon. .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. A. Ortiz, L. del Á. Diego, y P. Martínez (México) 145
El Mantenimiento Centrado en Confiabilidad con el Apoyo de los Sistemas de Protecciones de las
Subestaciones de Distribución . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. Betanzos, y J. C. López (México) 153
Using Fault Tree Analysis to Evaluate Protection Scheme Redundancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. R. Sandoval, C. A. Ventura, H. J. Altuve, R. A. Schwartz, D. A. Costello, D. A. Tziouvaras, y D. Sánchez 159
(México, Estados Unidos)
Experiencias Operativas y Criterios para Determinar Fallas de Alta Impedancia en Lineas de Transmi-
sión de Energía a Alto Voltaje . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. S. Urbaez, J. A. Calderon, y L. E. Llano (Colombia) 179
Abatimientos de Voltaje Prolongados en la Red de Media Tensión Ante Fallas en la Red de 230 Y 115
kV en Zona de Distribución Culiacán División Noroeste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Acosta, e I. Soto (México) 199
Energy Production Using the Co-Generation of Methane Gas Connected to Copel’s Distribution Grid—
Tests and Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. J. Rufato, C. M. V. Tahan, y C. C.B. de Oliveira (Brasil) 205
Miércoles 25 de mayo
Optimal Energy Resource Allocation in a Microgrid Environment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . A. Hajimiragha, A. Palizban, S. Allan, M. Adamiak, y M. Zadeh (Estados Unidos, Canadá) 209
CALBECK: Un Software para Calibraciones Totalmente Automatizadas de Relevadores de Protección en
Centrales Generadoras. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . M. V. González, A. Ahumada, C. Moreno, J. Lizárraga, Á. E. Gómez, D. I. Zamorano, y J. A. Castro 219
(México)
Increased Sensitivity of Loss of Field Protection Based on Admittance Measurement. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H.-J. Herrmann, y A. Smit (Alemania, Estados Unidos) 225
New Algorithm for Transformer Diferential Protection That Will Detect Turn-To-Turn Faults in Windings
Transformers in a Secure Manner and Faults Within the Diferential Protection Zone in Transformers
and Generators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. A. Cordova (México) 235
Why We Should Measure Line Impedance? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Vandiver, y A. Apostolov (Estados Unidos) 247
Event-Driven Load Shed Scheme and Operation Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Adamiak, M. Schiefen, G. Schauerman, y B. Cable (Estados Unidos) 257
Protection of Battery Energy Storage Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. Hedding, y P. Hayes (Estados Unidos) 263
Power System Protection Testing—Exploring the IEEE C37.233 Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . V. Madani, H. DoCarmo, M. Adamiak, R. Beresh, D. Novosel, P. McLaren, J. Verzosa, M. Kezunovic, 269
A. Apostolov, J. Ariza, E. Udren, y G. Henneberg (Estados Unidos)
IEC 61850 Goose Applications to Distribution Protection Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Apostolov, y B. Vandiver (Estados Unidos) 283
IEC 61850 - Impact on Protection Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .M. C. Janssen (Países Bajos) 295
Ethernet as a Channel for Protective Relaying. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. Benckenstein (Estados Unidos) 301
Testing Ethernet Network for Goose Tripping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Espinosa, R. Aguilar, J. Ariza, y J. Mendoza (México, Estados Unidos) 317

v
Mensaje del Comité Organizador
El Comité Organizador del X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos
de Potencia, SIPSEP, da la más cordial bienvenida a los participantes. Es un honor para
nosotros recibirlos en la ciudad de Monterrey y tener la posibilidad de compartir estos días
con ustedes.
El Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia es un
evento en el cual los ingenieros de toda Iberoamérica pueden intercambiar experiencias entre
sí y con especialistas de los países de mayor nivel de desarrollo tecnológico. Se mantienen
vigentes los objetivos originales del Simposio y su intención de promover el acercamiento
entre universidades, centros de investigación, fabricantes de equipos, y empresas eléctricas
de diversos países.
En las nueve ediciones anteriores del evento, han participado poco más de 1500 ingenieros
de 25 países, se han presentado 313 ponencias, se han dictado 23 conferencias magistrales y se
han desarrollado 10 meses redondas. La exposición de equipos ha sido un elemento importante
del evento, en la que se ha contado con la participación de las firmas más reconocidas en la
especialidad, a nivel mundial. Más de 600 ingenieros de una diversidad de países han asistido
a los 19 cursos tutoriales que se han impartido en el marco de estos nueve Simposios.
En el X SIPSEP se presentarán 33 ponencias por especialistas de 11 países, se dictarán tres
conferencias magistrales y un panel, que estarán a cargo de personalidades reconocidas inter-
nacionalmente y versarán sobre conceptos de Smart Grid y experiencias en la implementación
de redes inteligentes. Nuevamente, se contará con una exposición industrial, en la que partici-
parán los fabricantes más innovadores del mundo. Además, se dictarán dos cursos tutoriales,
dedicados a los temas de protección de líneas de transmisión y calidad de la energía.
El Comité Organizador desea hacer patente su reconocimiento y gratitud a la Universidad
Autónoma de Nuevo León, la Facultad de Ingeniería Mecánica y Eléctrica, y a la Comisión
Federal de Electricidad por el apoyo y estímulo que han brindado en la organización de este
evento.
Esperamos que el efecto satisfaga las expectativas de todos los participantes y constituya
una experiencia útil para su desarrollo profesional.

Comité Organizador,
Monterrey, México. Mayo de 2011.

vii
Empresas Representadas en la Exposición Industrial
• ABB México, S. A. de C. V.

• Allectro, S. A. de C. V.

• ALSTOM GRID (AREVA)

• ENOSERV

• ERL

• Fonkel Mexicana, S. A. de C. V.

• INDUSTRIA SIGRAMA, S. A. de C. V.

• INELAP, S. A. de C. V., una Empresa de Grupo Arteche

• Ingeniería e Instrumentación Internacional, S. A. de C. V. (AMETEK)

• ISA S. R. L.

• Ivesco, S. A. de C. V.

• MABREX, S. A. de C. V.

• OMICRON electronics Corp. USA

• Ontario Instruments, S. A. de C. V.

• Oropeza Ingenieros, S. A. de C. V.

• PROTECSA Ingeniería, S. A. de C. V.

• Schweitzer Engineering Laboratories, S. A. de C. V.

• Siemens, S. A. de C. V.

• Sistemas Eléctricos de Potencia Computarizada, S. A. de C. V.

• Transmisión Y Distribución, S. A. de C. V. (GE Digital Energy – Multilin)

• ZIV Aplicaciones y Tecnología, S. L.

ix
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 1

Operation and Design of a Protection Relay for Transformer Condition

Monitoring

H.T. YIP, S. HOSSEINI, M. MOSCOSO, G.J. LLOYD, K. LIU

ALSTOM GRID
UK

e-mail: graeme.lloyd@alstom.com

KEYWORDS

Transformer condition monitoring, protection.

1 INTRODUCTION
Power transformers are a critical link in the ac transportation of electricity from the generating stations
to end users. In terms of total investment, electric power utilities invest at least as much in
transformers as they do in generating stations. Transformers are expected to last 20 to 30 years, and in
many cases, even longer.
The application of monitoring systems to transformers offers benefits in more cost-effective utilization
by detecting faults that can be fixed before causing irreparable damage or in diagnosing faults and
determining if they can be repaired, lived with, or require transformer replacement.
Overloading power transformers beyond the nameplate rating can cause a rise in temperature of both
transformer oil and windings. If the winding temperature rise exceeds the transformer limits, the
insulation will deteriorate and may fail prematurely. Prolonged thermal heating weakens the insulation
over time, resulting in accelerated transformer loss-of-life. Through-fault currents create forces within
the transformer that can eventually weaken the winding integrity. Transformer overexcitation causes
transformer heating and increases exciting current, noise, and vibration.
A comprehensive transformer protection scheme needs to include protection against transformer
overload, through-fault and overexcitation, as well as protection for internal faults.
Measurable indicators of transformer serviceability include electrical load; top-oil, hottest-spot and
ambient temperatures; fault history; and measured excitation. Utilities that use these indicators can
make intelligent profit/risk decisions and plan optimal transformer loading and maintenance. Also
smart IEDs with standard flexible communications can ease integration of these devices into the
customer control system.
Although transformer monitoring systems have been on the market for some years and have achieved
success in detecting and diagnosing problems, their adoption is far from universal. The perceived
barriers to the more general application of transformer monitoring include high costs of installation
and maintenance, incompatibility between systems and transformers of different manufacture and the
difficulty of collecting an assimilating the output of diverse systems across a network.
Including many of the transformer monitoring features in a standard protection relay reduces the cost
barriers. Also, the flexibility and standardization of the protection relay communications allows data
easier collection and interfacing to the customers existing SCADA systems.
This paper discusses the operation and design of a transformer protection relay that includes many of
the common transformer condition monitoring features required making condition monitoring a more
economical option.
2

2 TRANSFORMER RELAY DESIGN FOR CONDITION MONITORING

A comprehensive transformer protection relay for protecting two and three winding transformers
(including autotransformers), with up to 5 sets of three-phase CT inputs and 4 voltage inputs has been
designed. The relay includes protection against transformer overload, through-fault and
overexcitation, as well as standard protection functions such as differential, overcurrent and earth fault
etc for internal faults.
The relay includes a number of measurable indicators of transformer serviceability monitoring the
electrical load; top-oil, hottest-spot and ambient temperatures; fault history; and measured excitation.
Utilities that use these indicators can make intelligent profit/risk decisions and plan optimal
transformer loading and maintenance.
The protection relay includes many of the common transformer condition monitoring features
required, as listed below and described in more detail later in the paper:
• Transformer top oil and hot spot temperature calculation taking into account the ambient
temperature, load variations and oil thermal parameters, based on IEEE Standard C57.91-1995
[2]
• Transformer loss of life calculation, based on IEEE Standard C57.91-1995 [2], providing data
on accumulated loss of life, rate of loss of life, aging acceleration factor and residual life hours
• Transformer through fault monitoring based on I2t calculation of the maximum fault current
and fault duration per phase
• Temperature monitoring for example of top oil, bottom oil, cooler inlet/outlet oil, tap changer
compartment oil and winding hotspot temperature via the relays ten RTD inputs or four
current loop inputs (CLIO - current loop inputs and outputs)
• Monitoring and alarming of other sensor quantities, such as tap changer position, oil level,
bushing oil pressure via the relay current loop inputs
• Load and short circuit current and system voltages monitoring via the protection relay current
and voltage transformer inputs. The relay can also provide CT and VT supervision of these
inputs.
The relay includes flexible industry standard communication options to interface the relay protection
and condition monitoring functions to the customers SCADA system. For example the relay supports a
number of standard protocols such as MODBUS, IEC60870-5-103, DNP 3.0 and IEC61850 and rear
port communication port options such as RS485, fibre optic, Ethernet or redundant Ethernet plus a
front communications port for local access.
The relay also includes programmable leds for user friendly local indication and programmable
scheme logic for easy scheme customization. Sequence of event and disturbance recording features
are also included that can help diagnosing faults and alarms.
As described above the relay includes four transducer (current loop) inputs with flexible ranges of 0-
1mA, 0-10mA, 0-20mA or 4-20mA which can be used with a number of external monitoring sensors
to indicate temperature, oil level etc. Associated with each input there are two time delayed protection
stages, one for alarm and one for trip. Each stage can be set for 'Over' or 'Under' operation. Four
current loop outputs are also provided with ranges of 0-1mA, 0-10mA, 0-20mA or 4-20mA which can
alleviate the need for separate transducers. These may be used to feed standard moving coil ammeters
for analog indication of certain measured quantities or for input to SCADA using an existing analog
RTU. Up to 24 digital inputs are available to provide status information from external devices such as
CBs or external sensors and up to 24 output contacts are available to provide alarm and trip outputs.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 3

Figure 1: Transformer protection and condition monitoring functional diagram

3 TRANSFORMER THERMAL MODEL FUNDUMENTALS

Factors that influence the transformer thermal characteristic and generate heat, which causes a rise of
temperature are mainly Core losses and Conductor resistance loss (usually referred to as Copper loss
i.e. I2R loss). The losses in a transformer are shown in the following diagram.

Losses of a
Transformer

Load No-Load
Losses Losses

Copper Stray Apparent Core Losses

Losses Losses Losses (iron losses)

Hysteresis Losses

Eddy-Current Losses

P4318ENa
Figure 2: Transformer Losses

The rating of a transformer usually depends upon the maximum permitted working temperature of the
transformer insulation [1]. The steady-state temperature rise is reached when these losses are balanced
4

by the transformer cooling due to radiation, convection and conduction of heat. The losses increase
with output and the maximum output or transformer rating is based on the losses that give the
maximum permitted temperature Tp. The permitted temperature rise θp also depends upon the ambient
temperature of the surrounding air Ta:
Maximum temperature rise θp = Tp - Ta
During energization, the transformer inrush current causes the transformer to be overloaded for a short
duration until its temperature reaches Tp. The inrush current is well in excess of the full-load current
(perhaps up to 10 times) but the transient time is sufficiently short to ensure that θp is not exceeded.

4 THERMAL MONITORING AND PROTECTION

The relay thermal overload monitoring and protection is based on IEEE Standard C57.91-1995 [2].
The thermal overload trip is based on the hot spot temperature ΘH or top oil temperature ΘTO, each
with its own exclusive time constant setting. The top oil temperature can be calculated or can be
measured directly by one of the current loop or RTD inputs. The hot spot temperature can only be
calculated. The ambient temperature ΘA can be measured directly or an average value can be set by
the user (average ambient temperatures covers 24 hour time periods). Since transformer loads are
becoming increasingly non-linear, RMS current values are used to replicate the winding hotspot
temperature. The winding, the temperature of which is under supervision, can be selected as either HV
or LV or TV or biased current. It is recommended to select the biased current option so that an overall
thermal condition monitoring of the transformer is carried out.
The biased current used by the thermal protection element is not the same as the biased current used
by the differential protection. To calculate the bias current, the thermal element considers the
maximum rms phase current of each transformer winding. The thermal bias current calculation is as
follows.
MaxHVI rms + MaxLVI rms + MaxTVI rms
I bias =
2
The relay has up to three hot spot stages and up to three top oil stages. The Top Oil tripping signal is
asserted when the top oil (measured or calculated) temperature is above the setting and the time delay
has elapsed. Also, the Hot Spot tripping signal is asserted when the hottest-spot (calculated only)
temperature is above the setting and the time delay has elapsed.
If a more accurate representation of the thermal state of the transformer is required, the use of
temperature monitoring inputs (RTDs or Current Loop Inputs (CLI)) which target specific areas is
recommended. Also, for short time overloads, the application of RTDs/CLI and overcurrent
protection can provide better protection.
4.1 Top oil temperature
If the top oil temperature is not available as a measured input quantity, it is calculated every cycle by
the following equation:

ΘTO = Θ A + ∆ΘTO

ΘTO = Top oil temperature

ΘA = Ambient temperature
∆ΘTO = Top oil rise over ambient temperature due to a step load change
ΘA can be acquired by measurement directly or can be set by the user through the user interface.

4.2 Hot spot temperature

The hot spot temperature can only be obtained by calculation. The following equation is used to
calculate the hot spot temperature every cycle:
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 5

Θ H = ΘTO + ∆Θ H

Where:
ΘH = Hot spot (winding) temperature
ΘTO = Top oil temperature
∆ΘH = Hot spot rise above top oil temperature
4.3 Cooling modes
Four cooling modes are available in the relay - Natural, Forced A ir, Forced O il, Forced Air O il.
The oil exponent and winding exponent can be set independently for each mode. These exponents are
empirically derived and are required to calculate the variation of ∆ΘH and ∆ΘTO with load changes.
The oil and winding exponent values changes for each mode of cooling to approximately account for
changes in the resistance and viscosity with changes in load.
The cooling mode selection can be made in the relay by using digital inputs or can be set to a fixed
mode in the setting file. The digital inputs enable the thermal characteristic to be changed to match the
different cooling modes as fans and pumps are switched on and off.

5 LOSS OF LIFE STATISTICS

As stated in IEEE Std. C57.91-1995 [2], aging of insulation is a time function of temperature, moisture
and oxygen content. The moisture and oxygen contributions to insulation deterioration are minimized
due to modern oil preservation systems. Therefore, temperature is the key parameter in insulation
ageing. Temperature distribution is not uniform; the part with the highest temperature undergoes the
greatest deterioration. Therefore the hottest spot temperature is considered in loss of life calculations.
As indicated in IEEE Std. C57.91-1995 the aging acceleration factor is the rate at which transformer
insulation aging for a given hottest spot temperature is accelerated compared with the aging rate at a
reference hottest spot temperature. For 65°C average winding rise transformers, the reference hottest
spot temperature is 110°C. For 55°C average winding rise transformers, the reference hottest spot
temperature is 95°C. For hottest spot temperatures in excess of the reference hottest spot temperature
the aging acceleration factor is greater than 1. For hottest spot temperatures lower than the reference
hottest spot temperature, the aging acceleration factor is less than 1.
The relay includes a single stage definite time delay alarm based on aging acceleration factor (FAA)
and loss of life (LOL). Measurements are also available for Accumulated loss of life (LOL) in hours,
Residual life at reference hottest spot temperature, Rate of loss of life (ROLOL) in %, Aging
acceleration factor (FAA), Mean aging acceleration factor (FAA,m), and Residual life hours at FAA,m
(Lres(FAA,m)).
The model used for loss of life statistics is based on the equations for Loss of Life LOL and
Accelerated Aging Factor FAA as stated below.
Loss of life is calculated every hour according to the following formula:
LOL = L(Θ H ,r ) − Lres (Θ H ,r )
Where:
L(ΘH,r) = life hours at reference winding hottest-spot temperature. This parameter is set by the user.
Lres(ΘH,r) = residual life hours at reference winding hottest-spot temperature.
The insulation aging acceleration factor FAA, for each time interval Δt, is calculated according to the
following formula:
6

 B 
A +   B B 
L(Θ H , r ) e  ΘH ,r + 273   − 
 ΘH ,r + 273 ΘH + 273 
FAA = = =e
L(Θ H ) 
A +
B 

 ΘH + 273 
e
Where:

L(ΘH) = life hours at winding hottest-spot temperature.

ΘH = hottest-spot temperature as calculated in thermal overload protection.
ΘH,r = hottest-spot temperature at rated load.
B = constant B from the life expectancy curve. This parameter is set by the user. IEEE Std.
C57.91-1995 [2] recommends a B value of 15000.

The residual life hours at reference hottest-spot temperature is updated every hour as follows:
3600
∑ FAA,i (Θ H )
i =1
Lres (Θ H ,r ) = Lres,p (Θ H ,r ) −
3600
Where:
Lres,p(ΘH,r) = residual life hours at reference temperature one hour ago.
FAA,i(ΘH) = mean aging acceleration factor, as calculated above. It is calculated every second.
The accumulated loss of life (LOL) is updated in non-volatile memory once per hour. It is possible to
reset and set a new loss of life figure, in the event that a relay is applied in a new location with a pre-
aged resident transformer.
Daily rate of loss of life (ROLOL) in percentage per day is given as follows and it is updated every
day:
24
ROLOL = ⋅ FAA,m (Θ H ) ⋅100%
L (Θ H , r )

The mean aging acceleration factor FAA,m is updated per day and is given by:

N N
∑ FAAn ⋅ ∆t n ∑ FAAn
n =1 n =1
FAA, m = N
=
N
∑ ∆t n
n =1
Where:
FAAn is calculated every cycle
∆tn = 1cycle
FAA, m states the latest one-day statistics of FAA. When the relay is energized for the first time, FAA, m
default value is 1.
The residual life in hours at FAA,m is updated per day and is given by:
Lres (Θ H , r )
Lres (FAA, m ) =
FAA, m

6 THROUGH FAULT MONITORING AND CONTROL

According to statistics [3], through faults is the main culprit in causing transformer failure and
according to IEEE Std. C57.109-1993 [4] mechanical effects are more significant than thermal
effects for fault-current magnitudes near the design capability of the transformer. However, at
fault-current magnitudes close to the overload range, mechanical effects are less important unless the
frequency of fault occurrence is high. Figure 3 below depicts an example initiation of a through fault.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 7

Through fault
current for fault 2
Equivalent Source 1 Through fault Equivalent source 2
0.024∠84° pu 275kV 132kV current for fault 2 0.096∠84° pu
10000MVA HV LV 25
00 MVA
600:1 1200:1

CT1 CT2

Through fault 2
Through fault 1 Single phase fault
Single phase fault 275/132/33 kV Three phase fault
Three phase fault 240MVA
Positive and Negative sequence
impedances:
XHV = j0.134pu
XLV = j0.075pu
XTV = -j0.024pu
Zero sequence impedance:
XHV = j0.067pu
XLV = j0.0375pu
XTV = -j0.012pu

Figure 3: Through fault monitoring example

As fault duty and feeder exposure increase, the incidence and severity of through faults
experienced by a transformer will tend to go up over time. IEEE Std. C57.12 [5] provides
construction guidelines for short-circuit withstand for transformers. The standard states that a
transformer shall withstand 2 seconds of a bolted fault at the transformer terminals.
In the relay design the through fault monitoring element can monitor the HV, the LV or the TV
winding. In three winding applications, the winding through which the highest current would flow
during an external fault should be monitored. The through fault current monitoring function gives the
fault current level, the duration of the faulty condition and the date and time for each through fault.
An I2t calculation based on the recorded time duration and maximum current is performed for each
phase. This calculation is only performed when the current is above the provided setting and if the
differential protection element has not started. Cumulative stored calculations for each phase
can be monitored so that the user may schedule the transformer maintenance based on this
data. This may also justify possible system enhancement to reduce through fault level.
A single stage alarm is provided for through-fault monitoring. The alarm is issued if the
maximum cumulative I2t in the three phases exceeds the relevant setting. A through fault event
is recorded if any of the phase currents is bigger than a trigger setting. This setting should be
greater than the overload capability of the transformer. According to IEEE Std. C57.109-1993
[4], values of 3.5 or less multiples of normal base current may result from overloads rather than
faults. IEEE Std. C57.91-1995 [2], states that the suggested limit of load for loading above the
nameplate of a distribution transformer with 65°C rise is 300% of rated load during short-time
loading (0.5 hours or less). On the other hand, the suggested limit of load for loading above the
nameplate of a power transformer with 55°C rise is 200% maximum.

7 OVEREXCITATION MONITORING AND PROTECTION

The magnetic flux inside the transformer core is directly proportional to the applied voltage and
inversely proportional to the system frequency. Overvoltage and/or underfrequency conditions can
produce flux levels that saturate the transformer core. This could cause stray flux to be induced in
unlaminated components that have not been designed to carry the flux. The resulting eddy currents in
solid components (core bolts and clamps) and end of core laminations can cause rapid overheating and
damage. These abnormal operating conditions can exist in any part of the power system, so any
transformer may be exposed to overexcitation. A severely overexcited transformer should be
disconnected to avoid transformer damage.
For overfluxing protection and alarming the relay includes three definite time setting stages and one
DT/IDMT setting stage plus a single stage alarm setting that can be combined to create a combined
multi-stage V/Hz trip operating characteristic using programmable scheme logic. Reference should be
made to the manufacturers’ withstand characteristics before formulating these settings.
8

Transient overfluxing conditions can cause unwanted operation of the transformer differential
protection if stabilization is not provided. Overexcitation of a power transformer is a typical case of
a.c. saturation of the core that produces odd harmonics in the exciting current. The third harmonic is
the most suitable for detecting overexcitation conditions, but either the delta connection of the CTs or
the delta connection compensation of the differential relay filters out this harmonic. The fifth
harmonic, however, is a reliable quantity for detecting overexcitation conditions.
The relay filters the differential current and determines the fundamental component and the fifth
harmonic component. If the ratio (5th harmonic component)/(fundamental component) exceeds the set
value in at least one phase in two consecutive calculations, and if the differential current is larger than
0.1 pu, differential tripping is blocked selectively per phase.
The 5th harmonic element can also be used to raise an alarm due to persistent overfluxing that may be
caused by geomagnetic disturbances that may not be detected by the conventional a.c. overfluxing
protection.

8 CONCLUSION
Power transformers are a critical link in the ac transportation of electricity from the generating stations
to end users. The application of monitoring systems to transformers offers benefits in more cost-
effective utilization by detecting faults that can be fixed before causing irreparable damage or in
diagnosing faults and determining if they can be repaired, lived with, or require transformer
replacement. Although transformer monitoring systems have been on the market for some years and
have achieved success in detecting and diagnosing problems, their adoption is far from universal. The
perceived barriers to the more general application of transformer monitoring include high costs of
installation and maintenance, incompatibility between systems and transformers of different
manufacture and the difficulty of collecting an assimilating the output of diverse systems across a
network.
This paper discusses the operation and design of a transformer protection relay that includes many of
the common transformer condition monitoring features required making condition monitoring a more
economical option.

9 REFERENCES
[1] Guide for Loading Mineral-Oil-Immersed Power Transformers Rated in Excess of 100
MVA (65ºC Winding Rise), IEEE Std. C57.115-1991.
[2] IEEE Guide for Loading Mineral-Oil-Immersed Transformers, IEEE Std C57.91-1995.
[3] W. H. Bartley, “An Analysis of Transformer Failures—1988 through 1997,” The
Locomotive, Hartford Steam Boiler Inspection and Insurance Company.
[4] Guide for Liquid-Immersed Transformer Through-Fault-Current Duration, IEEE Std
C57.109-1993 (R2008).
[5] IEEE Standard General Requirements for Liquid-Immersed Distribution, Power, and
Regulating Transformers, IEEE Std. C57.12.00-1993.
[6] CIGRE Technical Brochure Ref. 343, Recommendations for Condition Monitoring and
Condition Assessment Facilities for Transformers, Working Group A2.27, April 2008.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 9

IMPROVEMENTS TO TRANSFORMER DIFFERENTIAL PROTECTION – DESIGN

AND TEST EXPERIENCE

H.T. YIP, M MOSCOSO, G LLOYD, K LIU, Z WANG

ALSTOM Grid
United Kingdom
tony.yip@alstom.com

KEYWORDS

New algorithms, transformer differential protection, CT saturation, magnetising inrush.

1 INTRODUCTION

Percentage Bias characteristics have been commonly applied for transformer differential
protection. Traditional protection schemes use dual slope or multiple slope characteristics
to provide stability during CT saturation, together with second harmonic as a restraining
or blocking quantity for inrush conditions. However, studies have shown that there is a
limitation with the dual slope characteristic for stability, to avoid requiring excessive over-
dimensioning of the CTs. The second harmonic blocking has also been shown to slow
down the protection for internal faults if CT saturation occurs [1], [2]. This is because CT
saturated waveforms can also contain a high second harmonic component. Early studies
on magnetizing inrush indicates that the second harmonic content might be 15% or more
of the fundamental current [3]. Recent studies indicate that improvements in the
transformer design and core steel result in low content of second harmonic (as low as
7%) [3].

This paper presents new algorithms that have been developed to improve the
performance of transformer biased differential protection. A transient bias algorithm has
been implemented which increases the operating threshold momentarily when there is a
sudden increase in the bias current, thus enhancing stability during external faults. This
approach will not affect the performance during internal faults, since the differential
current will rise much faster than the bias currents. A novel CT saturation detection
algorithm has been developed. It is based on the fact that magnetising inrush and
saturated current waveforms are distorted and discontinuity patterns can be observed on
both waveforms. This new technique emphasizes the discontinuity pattern by
differentiating the differential current making it possible to distinguish magnetising inrush
from CT saturation within a cycle. A no gap detection algorithm has been applied to
complement the CT saturation detection. It considers the number of samples above a
dynamic threshold within the last “N” executions. The assertion of CT saturation or no
gap detection unblocks the differential protection from the second harmonic blocking. As
a result, the restrained differential element is allowed to trip even during CT saturation,
giving faster tripping times for internal faults.
10

2 TRANSIENT BIAS TECHNIQUE

To provide further stability for external faults, additional measures are considered on the
calculation of the bias current: delayed bias, maximum bias and transient bias.

The delayed bias is calculated on a per phase basis and it is the maximum of the
fundamental bias quantities calculated within the last cycle. The delayed bias provides
added stability when an external fault is cleared and the fault currents drop off.

Ibias A_delayed = Maximum [Ibias, A(n), Ibias,A (n-1), ........., Ibias,A (n – (k-1))] Equation 1

Ibias B_delayed = Maximum [Ibias, B(n), Ibias,B (n-1), ........., Ibias,B (n – (k-1))] Equation 2

Ibias C_delayed = Maximum [Ibias, C(n), Ibias,C (n-1), ........., Ibias,C (n – (k-1))] Equation 3

The maximum bias is the maximum of the delayed bias currents from all three phases.
The maximum bias is used to prevent mal-operation under external faults conditions,
when spill current could flow into a healthy phase the CT of which is partially saturated.
Ibias,max = Maximum [IbiasA_delayed, Ibias,B_delayed, Ibias,C_delayed] Equation 4

The maximum bias is used to calculate the differential operating current Iop, using the
following equations by considering the multiple slope characteristic shown in Figure 1:
I s1
Characteristic equation for the range: 0 ≤ I bias ,max ≤ Equation 5
K1
Iop = I s1 Equation 6

I s1
Characteristic equation for the range: ≤ I bias , max ≤ I s 2 Equation 7
K1
Iop = K1 ⋅ I bias , max Equation 8

Characteristic equation for the range: I bias ≥ I s 2 Equation 9

Iop = K1 ⋅ I s 2 + K 2 (I bias , max − I s 2 ) Equation 10

I s1
K1: characteristic slope in range ≤ I bias , max ≤ I s 2 Equation 11
K1
K2: characteristic slope in range I bias , max ≥ I s 2 Equation 12

Idiff (pu)

Restraint region

K2
Operating region

K1
Is1

Is2 Ibias (pu)

Figure 1 Tripping characteristic of differential protection

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 11

The transient bias is introduced when there is a sudden increase in the mean bias
measurement. The transient bias decays exponentially and it resets to zero once the
relay trips, or if the mean bias quantity is below the restrained element setting. The
operating current threshold is calculated at the maximum bias current. The transient
bias is calculated on a per phase basis and it is added to the operating current
calculated at the maximum bias. Therefore, the following differential current thresholds
are available:
Diff threshold phase A = Iop at max bias + transient bias_phase A Equation 13

Diff threshold phase B = Iop at max bias + transient bias_phase B Equation 14

Diff threshold phase C = Iop at max bias + transient bias_phase C Equation 15

The differential current is compared against the differential current threshold given above
on a per phase basis. If the differential current is above the threshold, then the biased
differential element might trip as long as the second harmonic and fifth harmonic
blockings are not asserted.
The transient bias technique considers a time decay constant, stability coefficients and
the differential function settings to provide a dynamic bias characteristic. Figure 2 shows
the differential thresholds calculated by the relay on a per phase basis for a AN external
fault on the star side of a Ynd11 transformer. It can be observed that the transient bias
has enhanced relay stability. For the relay to trip, the differential current should be
above the operating current at max bias + transient bias.

2 2
Transient bias - phaseB
Transient bias - phase A
Ibdiff - Phase B differential
1.5 Iadiff - Phase A Diff current 1.5 current
Differential threshold Phase B

Differential threshold phase A

I (pu)
I (pu)

1 1

0.5 0.5

0 0
0 0.5 1 1.5 0 0.5 1 1.5
t(s) t(s)

2
Transient bias - phase C

1.5 Icdiff - Phase C Diff current

Differential threshold phase C

I (pu)

0.5

0
0 0.5 1 1.5
t(s)

Figure 2 Transient bias – external fault

Figure 3 shows the nominal biased characteristic, the dynamic threshold and the
differential current loci. The differential currents increment from A to B, meanwhile the
differential current thresholds also increase from A’ to B’. It can be observed that
considering the nominal characteristic for phases A and B results in a mal-operation
because Iadiff and Ibdiff are above the nominal characteristic. Taking into account the
transient bias quantities prevents an unwanted trip because Iadiff and Ibdiff are below
the dynamic threshold. From B to C the differential currents are below the dynamic
threshold from B’ to C’.
12

3.00 3.00
Nominal characteristic
2.50 2.50 Nominal characteristic
Iadiff - Phase A diff current
Ibdiff - Phase B Diff Current
2.00 B' Differential threshold phase A 2.00

Diff (pu)

Diff (pu)
Differential threshold phase B
B B'
1.50 1.50
B
1.00 1.00

C' A'
0.50 A' 0.50
C'
C C
A A
0.00 0.00
0.00 1.00 2.00 3.00 4.00 5.00 0.00 1.00 2.00 3.00 4.00 5.00
Bias (pu) Bias (pu)

3.00

Nominal characteristic
2.50
Icdiff - Phase C Diff Current
2.00
Diff (pu) Differential threshold phase C

1.50
B'

1.00

A' B
0.50
C'
C
A
0.00
0.00 1.00 2.00 3.00 4.00 5.00
Bias (pu)

Figure 3 Differential characteristic with transient bias – external fault

The transient bias enhances the stability of the differential element during external
faults and allows for the time delay in CT saturation caused by small external fault
currents and high X/R ratios. For single-end or double-end fed faults the differential
current is dominant and the transient bias has no effect.
Figure 4 shows the differential thresholds calculated by the relay on a per phase
basis for a BN internal fault on the delta side of a Ynd11 transformer. The transient
bias does not prevent the operation of the differential element as the differential
current is above the operating current at max bias + transient bias.
2
6
transient bias - Phase A transient bias - Phase B
5
1.5 Iadiff - Phase A Diff Current Ibdiff - Phase B diff current

4
Differential threshold phase A Differential threshold phase B
I (pu)

I (pu)

1 3

2
0.5
1

0 0
0 0.5 1 1.5 0 0.5 1 1.5
t(s) t(s)

transient bias - Phase C

5

Icdiff - Phase C Diff current

4

Differential threshold phase C

I (pu)

0
0 0.5 1 1.5
t(s)

Figure 4 Transient bias – internal fault

Figure 5 shows that the dynamic threshold coincides with the nominal characteristic
because the transient bias for phase A is zero. For phases B and C the operating
currents have been increased due to the transient bias, but since this is an internal
fault the dynamic thresholds are well below the differential current loci.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 13

6.00
3.00
Nominal characteristic
B
2.50 5.00
Ibdif - Phase B diff current

2.00 4.00 Differential threshold phase B

Diff (pu)
Diff (pu)
1.50 3.00
Nominal characteristic B'

1.00 2.00
Iadiff - Phase A diff current

0.50 1.00 C'

Differential threshold phase A A'

0.00 0.00 A
C
0.00 1.00 2.00 3.00 4.00 5.00 0.00 1.00 2.00 3.00 4.00 5.00
Bias (pu) Bias (pu)

6.00
Nominal characteristic
B
5.00
Icdiff - Phase C diff current

4.00
Differential threshold phase C

Diff (pu)
3.00
B'

2.00

1.00
C'A'

0.00 A
C
0.00 1.00 2.00 3.00 4.00 5.00
Bias (pu)

Figure 5 Transient bias – differential characteristic – internal fault

No transient bias is produced under load switching conditions. Also, no transient

bias is generated when the CT comes out of saturation.

3 CT SATURATION AND NO GAP DETECTION TECHNIQUES

When second harmonic blocking is used to stabilize the restrained differential element
during inrush conditions, the CT saturation detection technique can be used to enhance
the restrained differential element operating time during internal faults with CT
saturation. The content of second harmonic may be above the second harmonic
threshold during internal faults with CT saturation. As a result, the biased differential
element may be blocked during an internal fault. If the fault level is not high enough,
then the unrestrained differential element will not be asserted; thus a fast trip cannot be
achieved. When the CT comes out of saturation, the biased differential element would
be operative. The result is a delayed differential trip [1], [2]. The delayed trip can be
avoided either by unblocking the biased differential element during internal faults or by
using a CT with a sufficiently high knee point voltage to avoid CT saturation (which might
not be possible).
The CT saturation detection technique distinguishes between magnetising inrush and
saturation; therefore, the relay stability is maintained during inrush conditions. To detect
a CT saturation condition the differential current samples on a per phase basis are
considered. The relay analyzes the differential current waveforms considering their
derivatives and dynamic and fixed thresholds that were determined by RTDS (real time
digital simulator) tests. The CT saturation detection logic unblocks the low set
differential element during CT saturation to allow the biased differential element
operation.
The no gap detection technique compliments the CT saturation technique. It considers
the number of samples on a per phase basis above a dynamic threshold within the last
N executions. This threshold is a function of the second harmonic content. If the
summation of the number of samples above the dynamic threshold is above or equal to
a fixed threshold (determined by RTDS tests) then there is no gap, and the blocking of
low set differential element is prevented.
The differential biased trip is affected by the CT saturation and no gap detection
techniques. If the second harmonic blocking is asserted and either the CT saturation
detection or no gap detection technique is asserted, then the biased differential trip is
14

unblocked. A biased differential trip will occur if the fifth harmonic blocking is not
asserted and the bias differential start signal is asserted. The differential biased trip logic
is described in Figure 6.

Is CrossBlock enabled?
5th Har Blk A
Yes = 1
No = 0

Id Bias Start A & Id Bias Trip A

2nd Har Blk A

& ≥
CT Saturation A &

≥
No Gap A 5th Har Blk B
&

External fault A Id Bias Start B & Id Bias Trip B

2nd Har Blk B

& ≥
CT Saturation B &

≥
5th Har Blk C
No Gap B
&

External fault B Id Bias Start C & Id Bias Trip C

&
2nd Har Blk C
≥

CT Saturation C &

≥ ≥
&
No Gap C
&

External fault C

Figure 6 Differential biased trip logic

Extensive real time simulation testing has been done to evaluate these techniques.
Figure 7(a) is the disturbance record for a AN internal fault from a relay with CT
saturation and No gap detection enabled, the operating time of the low set differential
element is 28 ms. It can be observed that both the CT saturation logic and the no gap
detection logic are asserted. The CT saturation and No gap detection logics
complement each other. Figure 7(b) is the disturbance record for a AN internal fault
from a relay with CT saturation and No gap detection disabled. The operating time is 57
ms. Hence, the relay is 29 ms faster if the CT saturation and No gap detection is
enabled.

(a) Enabled (b) Disabled

Figure 7 (a) CT saturation and no gap detection enabled (b) CT saturation and no
gap detection disabled
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 15

4 CT REQUIREMENTS
To determine the CT requirements for the differential function through fault injection tests
were performed under different X/R ratios, CT burdens, fault currents, fault types and
point on wave fault inception. The fault current waveforms considered a six fault
sequence, where the second set of three faults was controlled to generate the opposite
flux movement in the CT than the first set [4].
To establish the CT requirements the burden was incremented and the six fault
sequence repeated, until either the burden limit had been reached, or the relay mal-
operated.
Figure 8 shows the CT requirements when the transient bias technique is applied and
when it is not. It can be observed that when the transient bias technique is not included
the K factor is more than 5 times the K factor when the transient bias technique is
included.

Transient bias technique applied

Transient bias not applied
45.0 300.0
X/R = 5
40.0 X/R = 5
X/R = 10
250.0 X/R = 10
35.0 X/R = 20
X/R = 20
30.0 X/R = 40
200.0 X/R = 40
X/R = 80 X/R = 80
25.0 X/R = 120 X/R = 120
K

K
150.0
20.0
15.0 100.0
10.0
5.0 50.0

0.0 0.0
0 8 16 24 32 40 48 56 64 0 8 16 24 32 40 48 56 64
If If

Figure 8 CT Requirements with and without transient bias

Considering the results shown in Figure 8 when the transient bias technique is applied
the CT requirements might be quoted as follows:

System K
Knee point voltage
conditions (CT dimensioning factor)
In < IF ≤ 64In K = 25 VK ≥ 25 × In × (RCT + 2RL + Rr)(1)
5 ≤ X/R ≤ 20 VK ≥ 25 × In × (RCT + RL + Rr)(2)
In < IF ≤ 64In K = 40 VK ≥ 40 × In × (RCT + 2RL + Rr) (1)
20 <X/R ≤ 120 VK ≥ 40 × In × (RCT + RL + Rr) (2)
Table1 CT requirements

Where:
K = CT dimensioning factor
IF = maximum external fault current
X/R = primary system X/R ratio
In = Relay rated current
Rr = resistance of any other protective relays sharing the current transformer (Ω)
(1)
= single phase fault or phase-phase-ground fault
(2)
= three phase fault or phase-phase fault
To ensure that through fault stability is achieved the following ratios:
16

Vk-HV / Rtot-HV : Vk-LV / Rtot-LV

Vk-HV / Rtot-HV : Vk-TV / Rtot-TV
Vk-LV / Rtot-LV : Vk-TV / Rtot-TV
should not exceed a maximum disparity of 4:1. Fulfilling the disparity ratios ensures that
during a through fault condition the flux density in the current transformers is not greatly
different [5].
Where:
Vk-HV = Knee point voltage of CT at HV side
Rtot-HV = Total burden connected to CT at HV side = (RCT+ 2Rl + Rr) or (RCT + Rl + Rr)
Vk-LV = Knee point voltage of CT at LV side
Rtot-LV = Total burden connected to CT at LV side = (RCT+ 2Rl + Rr) or (RCT + Rl + Rr)
Vk-TV = Knee point voltage of CT at TV side
Rtot-TV = Total burden connected to CT at TV side = (RCT+ 2Rl + Rr) or (RCT + Rl + Rr)

5 CONCLUSIONS

It has been demonstrated that the transient bias algorithm improves the stability of the
biased differential element because the biased characteristic varies dynamically taking
into consideration the increase in bias current as well as the differential settings. In
addition, the CT requirements for a biased differential element are much lower when the
transient bias technique is included.

The CT saturation and no gap detection techniques improve the operating times as the
biased differential element is not longer blocked during internal faults with CT saturation.

6 REFERENCES

[1] IEEE Std C37.91-2000 Guide for Protective Relay Application to Power Transformers

[2] ALSTOM Transformer Differential Protection Relay MBCH publication R6070L

[3] BLACKBURN J. L., 1997. Protective Relaying Principles and Applications. 2nd ed.
New York: New York

[4] NGTS 3.24.18 Issue 1 – 2000 Unit Feeder Main Protection

[5] ALSTOM KBCH Manual KBCH/EN-M-G11

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 17

Nuevo Algoritmo Para Protección Diferencial

de Transformadores
Hugo Davila P.Eng.
ERLPhase Power Technologies Ltd


Resumen—Este documento presenta un nuevo algoritmo, sus Las técnicas comunes aplicadas para la protección de
fundamentos, resulta dos de p ruebas de validación p or transformadores serán rápidamente revisadas en la Sección-II.
simuladores en tiemp o real (RTDS) , además de los conceptos
básicos de protección convencional, est a t écnica comb ina los En los últimos años, la tecnología de protección basada en
principios de la prot ección diferencial porcentu al y protección microprocesadores ha ayudado al desarrollo y mejoras de las
direccional de corriente, y usando conceptos muy fundamentales
técnicas de protección y han ganado confiabilidad debido a los
como la ley de Kirchoff (corriente que entra=corriente que sale),
estas técnicas p odrán s er t ambién im plementada para la avances tecnológicos (doble procesador, comunicaciones por
protección dif erencial de barras de b aja impedan cia, se verá fibra óptica), haciendo de la protección mas sofisticada.
claramente como los efectos tradicionales de saturación de TCs y
niveles de cor riente directa (DC of fset) no tendrán mayor
incidencia en la decisión final de la operación del rele, con la cual II. PROTECCION DIFERENCIAL CONVENCIONAL
el reto de ignorar las operaciones en falso debido a los fenómenos
indicados sean de ay uda a la confiabilidad y seguridad de la A. Diferencial Porcentual
operación de los transformadores de potencia dentro del sistema,
estas técni cas mant ienen la característica porcentual diferencial
de doble pendiente y es su pervisado por el Nuevo alg oritmo Los reles de protección diferencial con restricción porcentual
Delta-Phase/ROCOD. han estado en servicio por muchos años. La Fig. 1 muestra el
diagrama de conexiones del rele diferencial típico. Los reles
Índice Términos—Transformador, Des fase Angular, Delta- diferenciales suman las Corrientes de cada entrada o salida
Phase, ROCOD, Registro de fallas, Diferencial, Oscilografias, IO asociado al dispositivo, para determinar la diferencia entre las
(Corriente de operación), IR (Corriente de restricción). Corrientes entrantes y salientes al rele. Una diferencia
sustancial indicara una falla en el elemento protegido o entre
los transformadores de corriente (TCs) localizados alrededor
I. INTRODUCION del transformador.
La protección de transformadores de potencia, es muy critica,
dado su importancia dentro del sistema, así como la magnitud
de la inversión que el mismo demanda, por tanto su protección
deberá ser segura, confiable y altamente sensible para la
detección de fallas de alta impedancia dentro de la zona de
protección, la presencia de saturación debido a fallas externas
de alta capacidad, la presencia de contenido de corriente
directa, y el bloqueo frente a fenómenos de inserción y
sobreexcitación.

El falso disparo de un relé que protege un transformador de

Fig-1 Esquema convencional de protección diferencial
potencia puede ocasionar apagones de gran magnitud, que
puedan incluir varias líneas de transmisión o alimentadores
Convencionalmente los valores de medida adoptados están
conectados a la unidad, además la desconexión de un
sujetos a evaluación numérica de acuerdo al principio de
transformador de potencia puede hacer colapsar el sistema,
medición de protección diferencial.
ocasionando sobrecarga en el sistema, o incidir en la
estabilidad del mismo. Por tanto es imperativo tener una
protección altamente confiable y segura.

Hugo Davila works for ERLPhase Power Technologies, Winnipeg, Canada (e-mail: hdavila@erlphase.com).
18

La característica de arranque (pick-up), tiene tres etapas, que

se usa en la forma convencional para protección numérica.
La protección diferencial tiene que enfrentar diferentes casos
que pueden causar la operación errónea y compensarlos según
sea necesario, entre estos factores tenemos básicamente:
 Errores de los TCs
 Relación de transformación en los TCs
 Relación variable de los transformadores debido a la
presencia de TAPs en el LTC para la regulación.
 Desfase angular de la Corrientes, debido al grupo de
conexión de transformador.
 Corrientes de Inserción debido a la energización del
transformador (2da y 4ta armónica).
 Corrientes altas de excitación, debido a fenómenos de
sobrexcitación (5ta armónica).
 Saturación de TCs.
 Existencia del componente de corriente directa en Fig-3, Muestra una ampliación de la zona-1
transformadores (DC Offset).

III. CONSIDERACIONES EN LA PROTECCION DIFERENCIAL

Aquí mencionaremos los ya muy conocidos casos que se
presentan durante la ocurrencia de fallas y que han sido
motivo para la aplicación de diversas técnicas y la técnica
presente que hace de la protección segura contra estos efectos
que son La Saturación de TCs y presencia de Corriente
Directa en la corriente de falla (DC Offset).

Saturación de TCs:
Si un TC es sometido a corrientes altas en el lado primario que
causa la saturación de la corriente secundaria, esta corriente
secundaria no será representativa de la corriente primaria y
por tanto puede causar la operación diferencial incorrecta, un
ejemplo de la saturación de TC se muestra en la figura
siguiente.

Fig-2 Característica típica de Operación porcentual de un rele

diferencial numérico.
Característica de doble pendiente Fig-2, muestra la
característica de doble pendiente, donde se pueden observar
claramente tres zonas de operación según se presente la falla.
Donde denominaremos IDIFF=IO y IRES=IR.

Zona-1: Considera el mínimo valor de arranque de la función

diferencial y contempla los errores de TCs, presencia de
LTC, Corriente de excitación y margen de ajuste.

Zona-2: O pendiente-1, considera básicamente la protección

del transformador bajo consideraciones lineales sin saturación
de los TCs.

Zona-3: O Pendiente-2, considera básicamente la protección

contra fallas con niveles de corriente altos y presencia de
saturación en los TCs. Fig-4, Característica típica de saturación en transformadores
de corriente
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 19

Como se puede ver en la Fig-4 la forma de onda senoidal de la
corriente secundaria esta distorsionada durante varios ciclos.
Este error en la corriente secundaria puede crear una cantidad
de corriente IO para una falla externa y tener como
consecuencia un falso disparo. Varias técnicas se han usado
para detector la saturación de TCs para bloquear el rele de
falsos disparos.
 El desfase angular es usado para supervisar la
característica tradicional de doble pendiente para
protección diferencial.
 El ángulo entre las entradas es calculada muy
rápidamente y fácilmente usando el producto
vectorial I1.I2=I1xI2 Cos ().

Componente de Corriente Directa (DC Offset):

Como se puede ver en la Fig-4 la corriente DC o componente

directa de corriente (DC Offset) puede acomodar la saturación
del TC. Esta presencia de corriente directa (DC Offset) puede
aparecer en el secundario del TC aún si no existe saturación.
Algunos reles de transformador usan técnicas en el software
para reducir o remover el DC Offset por el uso de algún
proceso de filtrado, introduciendo desfases de corriente y
retardos de tiempo en recomponer las corrientes.

IV. REVISION DEL NUEVO ALGORITMO

La protección diferencial implementa dos algoritmos
asociados a la característica tradicional diferencial de doble
pendiente, hablamos de Delta-Phase y ROCOF.

A. DELTA-PHASE ( Δ-Ø)

El Nueva técnica denominado como Delta Pase (Δ-ø),

esencialmente compara la información del ángulo de fase de Fig-A, Plano de operación para “Falla Interna”
todas las corrientes que entran y salen de la zona protegida
(por fase) para distinguir si la falla ocurrida es externa o
interna a la zona de protección. El umbral del ángulo de fase
usado para comparar fallas internas (Fig-A) es < 90º entre
cualquier corriente de entrada o de salida a la zona de
protección y si el ángulo de fase de cualquiera de las
corrientes que entra con respecto a la corriente que sale de la
zona de protección es > 90º, entonces la falla es declarada
como externa (Fig-B) a la zona de protección y el Rele será
bloqueado. Este Nuevo principio asegura significativamente la
operación de la función diferencial contra operaciones en falso
durante energizacion y fallas externas de una manera muy
elegante y sencilla comparada al método tradicional
“característica de Pendiente”. Es importante notar que la
técnica Δ-ø , por diseño requiere de por lo menos dos (2)
cantidades fasoriales (I1, I2) para comparación, como se
indica en las figuras adjuntas, en caso que se tenga disponible
solo una cantidad fasorial, la supervisión de el método Delta-
Phase se desactivara, y la protección se volverá al método
tradicional de operación “ característica de Pendiente”.
El Algoritmo  en Resumen: Fig-B, Plano de Bloqueo para “Falla Externa”
 Usa el concepto de la ley de Kirchoff (corriente que
entra=corriente que sale)
 No considera las desviaciones de las magnitudes de
corriente por efectos de Saturación de TC y
contenido de corriente continua (DC offset), debido a
que solo considera la diferencia angular para
comparación.
20
B. RAZON DE CAMBIO DE DIFERENCIAL (ROCOD)
La técnica Rate Of Change Of Differential
(ROCOD) esencialmente utiliza la primera derivada
o razón de cambio de la secuencia positiva de la
corriente de operación IO, y la corriente de
restricción IR, que entra y sale de la zona de
protección.
El principio Matemático usa la derivada de la
DIFERENCIAL (u OPERACION) y corriente de
RESTRICCION, la cual es una extensión del
principio que se esta usando en muchos reles
modernos basados en microprocesadores como
protección “diferencial de baja impedancia
porcentual”.
Considere el principio diferencial porcentual
aplicado para un dispositivo de protección de “n”
terminales:
…. EC-01
Es la corriente diferencial o de operación, la cual es
la suma fasorial absoluta de las corrientes que entran
y salen de la zona de protección.
La correspondiente corriente de restricción absoluta
individual (corriente de polarizacion) implementada
es:
.…EC-02
Normalmente, el siguiente criterio es usado para
disparo de reles diferenciales porcentuales de baja
impedancia:
…EC-03
Donde “f” implica alguna función de IR, la cual es la
“Característica de Pendiente”.
Luego diferenciando la Ecuación-3 (la cual es la
misma si diferenciamos las ecuaciones 1 y 2) con
respecto al tiempo, en ambos casos da el algoritmo
ROCOD:
….EC-04
La lógica ROCOD a continuación es usada para
Habilitar o Bloquear el disparo como se ilustra en el
diagrama lógico funcional (Fig-1). El principio
derivado usado en la técnica ROCOD elimina la
dependencia (efecto de restricción) de las
condiciones de carga en la corriente de restricción
polarizada de los terminales del transformador. Esto
es porque la derivada de la carga constante
(magnitud constante de los fasores según Ecuación-
2) previa a la falla será teóricamente cero.
Por tanto la Ecuación-4 refleja las condiciones para
identificar la posibilidad de falla interna incluso bajo
condición de transformador sobrecargado o falla
interna con muy alta impedancia.
La nueva Lógica ROCOD (Rate Of Change Of
Differential) que supervisa el algoritmo Δ-ø, es
rápida, confiable y sensitiva para todas las fallas
internas, se ha probado su efectividad en diferentes
fallas internas y se ha obtenido el 100% de
operaciones correctas con Resistencia de fallas en
el rango desde 0.1 hasta 200 Ohm.
El siguiente esquema booleano indica la operación de
un rele de protección diferencial.
Fig-5, Diagrama Lógico Funcional del algoritmo de
protección que incluye las técnicas Delta-Phase y
ROCOD
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 21

Las ventajas del algoritmo ROCOD es como sigue:

A.-Caso (i) Energizacion del Transformador – sin carga
 Trabaja con la misma velocidad como actúa el Descripción: En este caso de prueba, El transformador es
diferencial tradicional con “característica de pendiente” excitado en el lado HV, sin carga en el lado LV. El rele no
para fallas internas. debería operar para este caso debido la corriente de inserción
 Habilita Δ-ø para todas las fallas externas, corriente de en los lados LV y HV.
magnetización, saturación de TC, etc., incrementado la
seguridad.
 No requiere ningún ajuste nuevo.
 No hay impacto en la performance de los recursos del
procesador.
 No depende de la funcionalidad 87N. la función 87N es
independiente y disponible para ser usado según se
requiera.
 La técnica es independiente de los efectos de carga del
transformador para fallas internas.

V. RESULTADO DE PRUEBAS DE VALIDACION DEL NUEVO

ALGORITMO

Pruebas intensivas de validación del nuevo algoritmo se han llevado

acabo usando un simulador en tiempo real (RTDS), de la que algunos
de sus resultados se muestran a continuación.

Figure A1.0: Energizacion de transformador sin carga,-

operación estable durante energizacion.

A) Energizacion de transformador sin carga.

Casos de Prueba

Varios casos de pruebas fueron realizados, los siguientes son

solo algunos de ellos.

i) Pruebas de energizacion del transformador (con y sin carga)

Como se muestra.

ii) Conectar dos transformadores en paralelo, uno en servicio

alimentando la carga, y uno fuera de servicio. Energizar el
transformador fuera de servicio para verificar que ningún
transformador se dispara.
. Figure A2.: Característica Diferencial sin carga-Observe las
iii) El mismo que (i) pero creando una falla interna durante la corrientes de operación y restricción.
energizacion para mostrar que el rele opera.
iv) Conectar un transformador alimentando la carga radial,
luego crear una falla interna para probar el disparo del rele.
v) Aplicar fallas externas con y sin saturación de TCs
vi) Aplicar fallas internas de baja impedancia y de alta
impedancia(200 Ohm).
vii) Chequear la operación del rele para todo tipo de fallas
(AG, BG, CG, AB, BC, CA, ABG, BCG, CAG, ABC)
22

Figure A3.: Corrientes en los lados AT y BT durante la

energizacion Figure B2: Característica Diferencial durante energizacion con
carga pasiva en el lado de BT.
Observaciones:
1. El rele opera como se espero en la energizacion sin carga
2. Restricción Armónica (ajustado a 20%) ha prevenido
satisfactoriamente el disparo.

B.- Caso(i) Energizacion del Transformador-Con carga

Descripciones: En este caso de prueba, el transformador del

lado AT es excitado con una carga pasiva en el lado BT. El
rele no debería operar en este caso debido a la presencia de
corriente de inserción en los lados BT y AT.
Figura

Figure B3: Relación de fases de las corrientes en los lados AT

y BT durante la energizacion del transformador con carga
pasiva en el lado BT.

Figure B1.: Energizacion de transformador con carga,-

operación estable durante energizacion.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 23

C.-Caso (ii) Energizacion del Transformador en Paralelo-

Con Carga

Figure C1: Energizacion del transformador en paralelo con carga Figure C3: Ángulos de las corrientes de fase en AT y BT durante la
pasiva en el lado BT. energizacion de transformador en paralelo con carga pasiva en el
lado de BT.

D.- Caso (iii) Energizacion de Transformador – con falla interna –

Sin carga

Figure C2: Característica Diferencial durante energizacion del

transformador en paralelo con carga pasiva en el lado de BT.

Figure D1: Energizacion de transformador con falla interna sin carga

en el lado de BT.
24

E.-Caso (iii) Energizacion de Transformador-con falla interna-con

Carga

Figure D2: Característica Diferencial durante energizacion de

transformador con falla interna sin carga en el lado de BT. Figure E1: Energizacion de transformador con falla interna y con
carga pasiva en el lado de BT.

Figure D3: Corrientes en el lado AT y BT durante energizacion de Figure E2: Característica Diferencial durante energizacion de
transformador con falla interna sin carga en el lado de BT. transformador con falla interna y con carga pasiva en el lado de BT.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 25

Figure E3: Ángulos de las corrientes de fase en AT y BT durante la Figure F2: Corrientes en los lados AT y BT durante falla interna en
energizacion de transformador en paralelo con carga pasiva en el lado AT (100% del debando, Fase A) con carga pasiva en el lado de
lado de BT. BT.

F.- Caso (iv) Transformador Alimentando una carga radial-Con falla G.- Caso (v) Falla Externa sin saturación de CTs
interna

Figure G1: Falla externa sin saturación de CT, el rele 87 no se activa

.

Figure F1: Resumen de falla interna en el lado AT (100% de

devanado, Fase A), con carga pasiva en el lado BT, con resistencia
de falla de 0.1 ohm.

Figure G2: Falla externa sin saturación de CT, el rele 87 no se activa

26

Figure G3: Falla externa sin saturación de CT, La corriente de

operación IO es irrelevante o muy minimo.
Figura H3: Falla externa con saturación de CT, se ve claramente el
contenido de corriente de operación IO.
H.- Caso (v) Falla Externa con saturación de CTs
I.- Caso (vi) Falla Interna de alta impedancia (200Ohm).

Figura H1: Falla externa con saturación de CT, la función 87 Trip se

Figura I1: Falla interna de alta impedancia (200Ohm), podemos ver
mantiene inactiva..
como la función 87Trip se activa

Figura H2: Falla externa con saturación de CT, se ve claramente

como la falla ingresa a la zona de operación, la cual es bloqueada por
el algoritmo DP. Figura I2: Falla interna de alta impedancia (200Ohm), podemos ver
como la falla entra en la región de IO min.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 27

Figura I2: Falla interna de alta impedancia (200Ohm), podemos ver Figura J2: Falla interna de baja impedancia, se puede observa como
la presencia de corriente de operación. la corriente de operación es muy significativa.

J.- Caso (vi) Falla Interna de baja impedancia.

Figura J1: Falla interna de baja impedancia, podemos ver como la

función 87Trip se activa

Figura J2: Falla interna de baja impedancia, se puede observa como

la corriente de operación entra francamente en la región de disdparo.
28

VI. CONCLUSIONES
Dentro de las conclusiones de la aplicación de este Nuevo
algoritmo podemos mencionar lo siguiente:
 La combinación de las tecnicas Delta-Phase/ROCOD
aplicado a la caracterisitca diferencial porcentual,
incrementa la confiabilidad para el caso de fallas
internas de alta impedancia
 Incrementa la Seguridad para el caso de fallas de alta
magnitud externas y cercanas a la zona de protección
en condición de alto cometido de corriente directa
(DC offset) , la cual corresponde a la región de la
pendiente-2 en el esquema tradicional de doble
pendiente.
 No hay dependencia en las magnitudes diferenciales
solamente. En el pasado aun con reles diferenciales
numéricos se han tenido falsos disparos o falta de
disparos durante fallas.
 La técnica Delta-Phase proporciona información si la
falla es interna o externa, mientras que la técnica
ROCOD determina la sensibilidad de la falla dentro
de la zona de operación.
 La combinación de estas técnicas mejora la
Confiabilidad, Sensibilidad y Seguridad tanto para
Fallas externas, internas de baja y alta impedancia.

VII. REFERENCES

[1] ERLPhase Power Technologies, Transformer and Bus Bar Protection

Techniques, Winnipeg, Manitoba, Canada.
[2] RTDS test results, Winnipeg, Manitoba, Canada

[3] Secured Busbar Differential Protection Using A Computationally

Efficient Dot Product Technique, Power System Protection and
Automation 2010, New Delhi, India.

VIII. BIOGRAPHY

Hugo Davila
Tiene el grado de Inegniero Mecanico-Electrico de la
Universidad Nacional de Ingenieria (UNI) en Lima-Peru.
Hugo es un miembro activo del IEEE, actualmente es Gerente
Regional de ERLPhase Power Technologies Ltd(inicialmente
NXTPhase) para la Region Centro America, Mexico y El
Caribe, Antes de unirse a NXTPhase, El trabajo para
Beckwith Electric Company como Ingeniero de Soporte
Tecnico en Proteccion y Control. Tambien tiene experiencia
trabajando como Ingeniero de campo en proteccion y control
para las principales empresas utilitarias de su pais, y es autor y
coauthor de varios documentos tecnicos sobre aplicaciones en
proteccion de sistemas de potencia. (e-mail:
hdavila@erlphase.com).
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Modern Line Current Differential
Protection Solutions
Hank Miller and John Burger, American Electric Power
Normann Fischer and Bogdan Kasztenny, Schweitzer Engineering Laboratories, Inc.
Abstract—Line current differential protection creates
challenges for relay design and application. From a design
perspective, the distributed nature of the line current differential
system imposes limits on the amount of data that can be
exchanged between the system terminals and calls for data
alignment schemes to enable the differential protection principle.
From the application perspective, line current differential
schemes are concerned with CT saturation, particularly in dual-
breaker applications; in-zone reactors and line-charging current;
in-line and tapped transformers; sensitivity to high-resistive
faults; single-pole tripping; security on channel impairments;
application to lines with more than three terminals; and so on.
This paper reviews technical solutions to the line current
differential design and application, addressing the common
design constraints and utility-driven application needs. The
paper is a tutorial in this challenging area where protection
principles and applications mix with communications and signal
processing.
I. INTRODUCTION
As a unit protection having its zone delimited by location
of current transformers (CTs), the differential protection
principle is considered superior with respect to selectivity,
sensitivity, and speed of operation as compared with
directional comparison, phase comparison, or stepped distance
schemes.
The differential function responds to the sum of all the
currents of its zone of protection. Ideally, this sum equals zero
under all events except for internal faults. Practically,
measurement errors and shunt elements inside the zone may
create a spurious differential signal, calling for adequate
countermeasures. These countermeasures became more
sophisticated with advancements in the field of differential
protection and progressed from adding an intentional time
delay, percentage restraint, and harmonic restraint and
blocking to sophisticated external fault detection algorithms
and adaptive restraining techniques.
As applied to line protection, the differential principle
faced the limitations of line length. Analog schemes using
pilot wires can only be applied to very short lines because of
signal attenuation due to series resistance and the shunt
capacitance of the pilot. These applications are still beneficial
because the very short lines cannot be adequately protected
with distance relays.
The development of microprocessor-based line current
differential schemes utilizing digital communications channels
redefined the field of line protection.
When suitable long-haul digital communications channels
became more readily available because of the deployment of
29
digital microwave and direct fiber-optic connections as well as
synchronous optical network (SONET) or synchronous digital
hierarchy (SDH) systems, applications of line current
differential schemes kept expanding.
The key benefits of differential protection as applied to
power lines include good performance on multiterminal and
series-compensated lines and lines of any length as compared
with distance or directional comparison schemes; considerable
immunity to changing system conditions, long-term evolution
of the system, or nontraditional short-circuit current sources,
such as wind generators, photovoltaic sources, or power
electronic-based sources in general; good sensitivity; and
simplicity of application, at least from the protection
perspective.
The art of microprocessor-based line current differential
protection advanced with the first generation of relays
bringing achievements in data synchronization, working with
wide-area communications equipment, enhancing protection
principles (e.g., the Alpha Plane line differential element), or
standardizing physical interfaces between relays and
multiplexers (IEEE C37.94).
Many lessons were learned during this period. Originally
designed for direct fiber connections, line current differential
schemes were mostly deployed over multiplexed channels
because high-bandwidth fiber pairs were utilized for shared
data traffic. Protection engineers needed to learn new skills
related to digital communications. The communications
equipment originally designed around carrying voice data
needed some adjustments to support protection applications.
Post-event analysis related to communications impairments
was difficult because of the lack of recording and data-
gathering facilities at the interface between relays and
multiplexers or modems.
Looking at both the lessons learned from the first
generation of line differential relays and at the emerging needs
for new functions and features, this paper outlines general
design directions for a next generation line current differential
protection system.
The paper provides a utility perspective on needs and
expectations for a new line current differential relay. These
new functions include better security against CT saturation in
applications with dual-breaker terminals, line-charging current
compensation, multiterminal applications, a redefined role of
backup functions, and accommodating in-line transformers, to
name a few.
From the relay design point of view, this paper addresses
the stated requirements and describes an optimized line
30
current differential scheme working with limited bandwidth
channels, while providing for high performance in terms of
operation speed, sensitivity, and security under CT saturation
and channel impairments.
II. REQUIREMENTS FOR NEXT GENERATION LINE CURRENT
DIFFERENTIAL PROTECTION
Microprocessor-based line current differential schemes
found their broad field applications only a decade or so ago.
Many lessons have been learned from these first generation
schemes. When looking forward at the next generation of line
current differential schemes, the following requirements have
been identified from a utility perspective.
A. High Performance
Advances in bus and transformer differential protection
brought to life high-performance products. This pertains to
speed of operation, sensitivity, and immunity to CT saturation.
These improvements were possible because of the availability
of high-performance microprocessors, high-speed sampling,
optimum internal data buses, and new algorithms. Subcycle
trip times became common, while requirements for CTs were
considerably relaxed by the implementation of external fault
detectors and better restraining techniques.
Equally high performance is expected from the next
generation of line current differential relays.
B. Protection Security
Even though their application constantly broadens, line
current differential schemes tend to be used on higher voltage,
critical lines and those relatively shorter in length. Whether
protecting lines carrying gigawatts of power in the expanding
765 kV network or protecting transmission lines connecting
distributed generation, line current differential schemes must
be exceptionally secure.
At least three areas affect the security of a line current
differential scheme:
• The robustness of the relay hardware and firmware.
• The robustness of the applied algorithms and logic.
• The ability to deal with channel impairments and long
fiber circuits.
The first two categories apply to any microprocessor-based
relay. Manufacturers have developed significant knowledge in
designing for relay and algorithm performance.
The latter category is specific to line current differential
relays and relates to relay interaction with a communications
network—a relatively complex system that is out of the
control of relay designers and only partially under the control
of the protection engineers at the user organization. Lessons
learned in this respect should be incorporated in the next
generation schemes, including unexpected channel
asymmetry, channel switching, error detection capabilities,
accidental loopback, accidental cross-connection of relays,
and so on.
In addition, a modern relay should provide for recording or
evidence-gathering capabilities to aid post-event analysis in
areas related to communications impairments.
C. Channel Requirements
Both direct point-to-point fiber and multiplexed
connections should be supported.
Direct point-to-point fiber can be made available for
critical applications. In this respect, users expect relays to
support longer distances. This allows the elimination of
amplifiers along the fiber path and the related infrastructure
(e.g., housing, redundant power supply, physical security
perimeter, access roads, cybersecurity requirements).
Multiplexed connectivity is required for the majority of
regular applications when direct fiber connections cannot be
spared and instead only 64 kbps or a multiple thereof is made
available for line current differential protection.
A modern relay should provide for redundant channel
options at least in two-terminal applications. In particular, one
channel may be a direct fiber connection, and the standby
channel may be a multiplexed connection.
D. Charging Current Compensation
Line-charging current compensation is an expected feature
of a modern line current differential relay.
In networks operated with a considerable unbalance,
charging current compensation may be highly beneficial, even
if using sequence differential elements, such as a negative-
sequence differential element (87LQ).
In some 765 kV installations, the positive-sequence
charging current is in the range of several hundred to over
1,000 A, and it may be higher than either the load or fault
level.
E. Redefined Role of Backup Functions
Traditionally, a line current differential relay provides basic
distance backup to cover cases of unavailability of the
differential function because of problems with the
communications channels. Typically, these distance functions
are of a common design and slower speed and may face some
performance issues.
A new trend has emerged that calls for a “line relay” rather
than a line current differential relay, a full-featured distance
relay, a distance backup relay, or a simple overcurrent backup
relay.
In this concept, a multipurpose line relay is provided with
the same form factor, wiring, and ordering/procurement,
regardless of the served function. This allows a generic panel
design and expands the standardization of panels, wiring, and
substation integration, leading to reduced cost and shortened
design and production cycles.
Functions of such a relay are still differentiated to allow for
controlling the value and cost to the user.
In one aspect of this concept, there is little or no
differentiation between a full-featured distance relay and a line
current differential relay. The application may be decided late
into the project, or it may change at some point in time, but
the relay mounted in the panel is capable of providing both
functions, even though not necessarily simultaneously. Some
functions may only be operational when the communications
are not in service.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
F. Multiterminal Applications
A modern relay should provide protection for lines with
more than three terminals. This need is driven by tapped loads
or distributed generation tapped to transmission lines.
G. Auxiliary Functions
A modern relay should support modern and emerging
functions in a variety of areas. This includes but is not limited
to state-of-the-art Ethernet connectivity, synchrophasors,
IEC 61850, secure point-to-point digital signaling, user-
programmable math, customizable user interfaces, and cross-
tripping.
The relay must also support current CIP (Critical
Infrastructure Protection) requirements for passwords,
security, and disturbance monitoring, including digital
disturbance recorder (DDR), Sequential Events Recorder
(SER), and digital fault recorder (DFR) requirements.
H. Single-Pole Tripping and Reclosing
We expect that under congested transmission, increased
penetration of distributed generation, and the overall
expansion of the North American power system, more and
more transmission lines will be protected using the single-pole
tripping and reclosing philosophy.
A modern line current differential relay should be capable
of single-pole tripping and reclosing. This applies to tripping
from the differential, as well as backup functions.
I. Self-Monitoring and Relay Maintenance
A new trend of extending maintenance intervals based on
the amount of self-monitoring in microprocessor-based relays
has gained momentum.
The concept supported by the emerging NERC (North
American Electric Reliability Corporation) standards
encourages relay manufacturers to quantify the extent and
strength of self-monitoring in their microprocessor-based
relays so that adaptive maintenance programs can be formally
established at the user organizations.
As a logical consequence of this trend, the next generation
relays should be designed with the ease of quantification of
the embedded self-monitoring in mind and with the goal of
making the self-monitoring as strong as possible without
impacting the availability of the device. The ultimate goal is
the option to “run to fail” with all relay failures being fail safe
and detectable by the self-monitoring tests.
Line current differential schemes offer extra opportunities
in this respect. Because multiple, independent relays constitute
the protection system, data are shared and can be crosschecked
between the relays in the differential system.
III. A TYPICAL LINE CURRENT DIFFERENTIAL SYSTEM
With reference to Fig. 1, the most distinctive feature of a
line current differential protection system is that it consists of
multiple relays operating independently and linked by a digital
communications channel.
31
Fig. 1. Simplified architecture of a typical line current differential system.
The multiple relays in the system depend on each other in
order to serve their purpose of providing the line current
differential protection but are located at different substations
and operate autonomously in terms of sampling, frequency
tracking, transmission and reception of data, filtering, and
protection calculations.
Each relay samples its analog input currents via an analog-
to-digital (A/D) converter to obtain a digital representation
(iD) of the analog inputs (iA). The sampling rate may vary from
a few kilohertz to less than 20 samples per cycle, depending
on the design. Typically, the same digital data feed the line
current differential subsystem of the relay, as well as all the
other local functions—metering, fault recording, and
protection (distance backup, breaker failure, overcurrent, and
so on). Therefore, the sampling is typically of high resolution
and rate, even though the line current differential function may
utilize these data at lower sampling rates.
The sampling may be performed at a constant rate or at a
variable rate, tracking the power system frequency. The
samples may be taken asynchronously with the local relay
time or synchronously with the absolute time to simplify
implementation of synchrophasor measurements.
Some implementations allow the relays in the differential
system to sample asynchronously from each other, while some
implementations use the communications channel to force
synchronization of the relay sampling clocks.
To facilitate the line current differential function, the local
current data must be communicated to the remote terminals
( i TX → i RX in Fig. 1). Each relay that receives a full set of
data from all its remote peers can align the data, run its
differential trip equations, and operate autonomously in the
so-called master mode. Each relay that only serves the data
but does not receive all the remote data because of a
permanent lack of communication (channel not installed) or a
temporary loss of communication is referred to as working in
the slave mode. Direct transfer tripping (DTT) from masters
allows the slave relays to issue the trip command to their
breakers.
32
Channels traditionally used for line current differential
protection are limited in bandwidth (64 kbps is a typical
value), and therefore only limited amounts of data can be
exchanged between the relays at different line terminals. In
this respect, various line current differential designs differ
considerably.
Some relays exchange samples of currents, some relays
work on phasors, and yet others work on nonconventional
quantities such as a positive or negative current charge
between consecutive zero crossings of the current.
This variety of approaches is driven by the substantial
design challenge of conveying sufficient amounts of data to
facilitate adequate protection performance, meeting channel
bandwidth limitations, and providing for robust data
synchronization, all at the same time.
When working with current samples, the system can collect
and exchange samples at a rate that is high enough for
accurate interpolation (e.g., 1 kHz). As a result, the design
challenge of data synchronization can be met by measuring
the data latency between the remote and local relays and
interpolating (resampling) the remote current samples to align
them with the local samples. While resampling, the relay can
perform frequency tracking (i.e., take new virtual samples at
the rate following the actual system frequency). As an extra
advantage, the actual sampling clocks of the individual relays
in the system do not have to be synchronized.
When working with phasors, the system faces a
disadvantage that twice as much bandwidth is required to send
the real and imaginary parts of the currents. As a result, the
phasor exchange rate cannot be high enough to facilitate
phasor interpolation, and the relay sampling clocks must be
synchronized. This complicates the design, particularly in
multiterminal applications where more than two relays need to
stay synchronized. Also, in order to track system frequency,
this kind of design must control relay sampling clocks with
the concurrent goals of staying synchronized and following
the power system frequency. Additionally, filtered phasors
suppress high-fidelity information in the input currents, such
as harmonics or rate of change of currents, making certain
applications more difficult, such as in-line transformers or fast
detection of CT saturation.
Data synchronization (alignment) is one of the key
elements of a line current differential system. Commonly, a
channel-based synchronization method known as a “ping-
pong” algorithm is used to estimate clock offset between two
relays working via a communications channel. This well-
established timing method for communications networks
measures a two-way travel time by time-stamping the sending
and receiving of communications events and exchanging some
of the time stamps. Assuming that the channel is symmetrical
(meaning it is of the same latency in both directions), the total
channel time can be recalculated into the clock offset between
the two relays (see Section XI for more information). With the
knowledge of this offset, the system is capable of aligning the
data.
One method of alignment is to use the measured clock
offset to control the local sampling time at both ends with the
goal to null out the offset, meaning to force synchronization of
the two clocks. This approach works satisfactorily in two-
terminal applications but becomes complex in multiterminal
applications.
Another method is to time-stamp the transmitted data with
the local time and, knowing the clock difference between the
two relays, resample the data to align them to the same time
instants. This method works naturally for any number of
terminals as it allows each relay to freewheel without forcing a
larger group of relays into relative synchronism.
The channel-based synchronization method works well if
the channel is symmetrical. If the transmit and receive delays
differ, the alignment becomes less accurate, leading to a
phantom shift between the local and remote current
measurements. Various protection algorithms have different
immunity to such shifts if they are minor, but with major
asymmetry, both dependability and security are impacted to
the point of a complete loss of function.
In such cases, line current differential relays employ
synchronization, if needed, based on an external time
reference to augment or substitute for the channel-based
synchronization. Until now, the Global Positioning System
(GPS) has been the only practical way to provide for common
timing across wide areas. Reliance on a satellite system and
additional devices is not an ideal protection solution.
Therefore, applications with symmetrical channels have been
considered superior, while the need for GPS assistance has
been considered the least preferred solution.
However, terrestrial systems are emerging for providing a
common time reference [1]. These systems are synchronized
to GPS—for the benefit of using true time under normal
conditions—via multiple receivers located at multiple
geographical locations but continue to provide common
timing independently should all the GPS inputs be lost. These
systems are a part of the communications infrastructure that is
used for line current differential communications in the first
place. Therefore, they are a safe means of providing for
external time reference without introducing more devices,
unnecessary interdependencies, or reduced reliability or
availability.
There are two distinct communications channel
applications for line current differential protection: dedicated
point-to-point channels (typically fiber based) and multiplexed
channels.
Dedicated point-to-point channels remove any third-party
devices between the two communicating relays and, with
them, all the associated failure modes. Such channels are
inherently symmetrical, and any communications impairments
are caused only by the relays themselves or the passive media
between the relays.
As long-haul-dedicated channels are more expensive and
are usually available only under special circumstances,
multiplexed channels are used, typically within a
SONET/SDH infrastructure. Multiplexed channels put third-
party devices between the communicating relays with all their
complexity, failure modes, and ride-through attempts often
designed for nonprotection applications. This requires the line
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
current differential relays to be designed for a variety of
failure modes caused by the active communications
infrastructure between the relays.
This paper elaborates on a number of aspects briefly
introduced in this section, while reviewing design directions
for a new line current differential system.
IV. RELAY DESIGN CONSTRAINTS BECAUSE OF AVAILABLE
CHANNEL BANDWIDTH
From the relay design perspective, the primary constraint
of a microprocessor-based line current differential system is
the requirement to work with a communications channel of a
limited bandwidth.
Today, line current differential relays must work with
64 kbps channels. Even though direct point-to-point fiber
connections allow bandwidths in the range of tens of megabits
per second, and multiplexed channels can be requested with a
bandwidth of Nx64 kbps, the 64 kbps bandwidth continues to
be a common application scenario.
To realize the amount of data that can be conveyed for
protection purposes over a 64 kbps channel, let us assume a
sampling rate of 16 samples per cycle and review the
following:
• 64,000 bits per second =
• 1,067 bits per a 60 Hz power cycle =
• 267 bits per quarter of a 60 Hz power cycle =
• 66 bits per each of 16 sample sets in a 60 Hz power
cycle
The 267 bits available every quarter of a cycle or 66 bits
available 16 times a cycle may seem sufficient. However, as
with any digital communications scheme, there is always
certain overhead in the communications packet on top of the
actual payload. In a line current differential system, the
digitally encoded values of currents are the payload. The
major components of the overhead include the following:
• A header is required to tell consecutive packets apart
at the receiving end. A total of 15 bits may be needed.
• The integrity of the data must be protected by
redundancy checks such as Bose, Ray-Chaudhuri,
Hocquenghem (BCH) Code or Cyclic Redundancy
Code (CRC). A total of 32 bits is typically needed.
• Channel-based synchronization methods need to
append certain time values to the packet. A total of
16 bits or more may be required.
• The packet must support basic addressing to prevent
accidental cross-connection of line differential relays.
A total of 4 to 8 bits may be required for basic
addressing.
• DTT and other flags must be supported. A total of 4 to
8 bits may be used for this purpose.
The above can add 50 to 80 bits of overhead.
33
Note that when sending packets 16 times a 60 Hz power
cycle over a 64 kbps channel, we can only use 66 bits, having
practically no room for payload even when significantly
optimizing the payload and the overhead.
Still, when designing relays for high-speed operation, it is
beneficial to keep the rate at which fresh data are passed from
subsystem to subsystem high so that the total data latency is
minimized. Therefore, it is advantageous to exchange the
analog data between line current differential terminals
multiple times per cycle.
As shown by our simple calculations, the task of passing
the right data at a high rate is not trivial. It is clear that the
protection-driven payload and the communications-driven
constraints must be addressed in a concurrent design in order
to yield a high-performance scheme. It is very important
which quantities are sent, how often, and how they are
encoded, packetized, and protected for integrity.
When moving data within a single relay using internal data
buses designed adequately, we do not have to apply as much
optimization. When working with a low-bandwidth channel in
a line current differential system, however, an important part
of protection design is to select protection algorithms that
maximize the available bandwidth.
In general, the following solutions aid the design:
• Smart encoding. Properties of the sent data, if studied
carefully, may allow reducing the number of bits
required to convey their values. For example, a
negative-sequence restraint can be sent as per unit of
the highest phase current restraint. Or the value of
current can be encoded on a log-based scale rather
than a linear scale to recognize the wide range of
current signals.
• Interleaving, or sending small fragments of slowly
changing data in consecutive packets. For example,
the channel-based synchronization calculations can be
run at a rate lower than the packet rate.
• Sending various pieces of data at optimum rates
required by the applied protection equations.
• Increasing the packet size so that the payload-to-
overhead ratio becomes more favorable.
• Selecting the payload in a way that maximizes the
information content in it, given the intended protection
algorithms.
In order to illustrate the concept, consider the packet and
protection equations presented in this paper. This solution
works with 1 kHz samples of currents and utilizes proven
Alpha Plane protection equations [2].
34

Table I summarizes the packet content. • Restraint quantities are magnitudes that can be
TABLE I encoded using fewer bits. In addition, the restraint
PACKET PAYLOAD DEFINITION quantities are auxiliary terms and can be encoded with
lower accuracy without sacrificing security.
Symbol Meaning
• The five restraint terms can be interleaved, saving
iA(k) Present (k-th) sample of the line current, A-phase extra communications bandwidth.
iA(k-1) k-1 sample of the line current, A-phase • The negative- and zero-sequence restraint terms can
iA(k-2) k-2 sample of the line current, A-phase be encoded as per-unit values with respect to the
highest phase restraint, further reducing the bandwidth
iB(k)
requirement.
iB(k-1) As above, B-phase • The packet format makes the solution scalable, as it
iB(k-2) works with any number of local currents at a given
iC(k)
line terminal. The packet always contains the partial
differential and partial restraint terms.
iC(k-1) As above, C-phase
The following sections explain how this packet format
iC(k-2) enables fast detection of saturated CTs, provides for proper
IAR Restraint term, A-phase restraint in dual-breaker applications, and supports line-
charging current compensation and protection of in-line power
IBR Restraint term, B-phase
transformers.
ICR Restraint term, C-phase
IQR Negative-sequence restraint term V. DUAL-BREAKER TERMINALS AND PROTECTION SECURITY
IGR Zero-sequence (3I0) restraint term A. Introduction
Dual breaker refers to a line configuration where the line is
The payload of Table I can be encoded using slightly more
terminated in a double-bus double-breaker, breaker-and-a-
than 100 bits, allowing us to send packets every
half, or ring-bus substation (Fig. 2). Modern line protection
3 milliseconds (3 milliseconds at 64 kbps allows 192 bits).
relays support two three-phase sets of current inputs and
The following sections of this paper explain in detail how
measure the two currents independently. These relays work
the individual data items in the packet are calculated prior to
with the internally summed current for the main protection
transmission and consumed upon reception. In short, the
function—distance, ground directional overcurrent in a pilot-
instantaneous values are total line currents at the sending
assisted scheme, or the line current differential. At the same
terminal (a sum of all the local currents, such as from the two
time, it provides for two independent breaker failure functions,
breakers of a dual-breaker termination), while the restraint
two independent autoreclosers, metering, recording, and time-
terms are sums of magnitudes of all the local currents (such as
coordinated backup, all responding to the individual breaker
from the two breakers of a dual-breaker termination). Simply
currents.
put, the instantaneous values are partial line differential
currents, and the restraint terms are partial line restraint
currents.
The selected payload provides the following benefits:
• Fresh data are sent every 3 milliseconds, or more than
five times a 60 Hz cycle, minimizing latencies and
speeding up operation of the relay.
• A packet lost just before or during an internal fault
erases only 3 milliseconds of data, allowing for fast
recovery and preventing delayed operation of the Σ
relay.
• Working with 1 kHz samples offers good fidelity of
differential current measurements and allows the
calculating of harmonics for in-line transformer
applications and fast detection of saturated CTs.
• Sending three samples of instantaneous current per
packet improves the payload-to-overhead ratio.
• Sending one value of a restraint per packet (or per
three samples of instantaneous values) reduces Fig. 2. Dual-breaker line termination and application of dual CT input
relays.
bandwidth requirements, while it is sufficient for
protection applications.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 35

As a result, dual CT input line relays enable a complete differential functions. Two approaches are used independently
protection, metering, monitoring, and fault recording solution and simultaneously in order to secure these functions.
for dual-breaker line terminals. First, the notion of a through-fault current is developed via
Dual-breaker applications working with breaker currents proper restraint terms. These terms are inserted into the
may create some protection challenges. communications packet, utilizing a minimum possible
First, the two CTs may be rated much higher as compared bandwidth. The restraints are not used directly for tripping but
with the load of the protected line, challenging protection feed into generalized Alpha Plane trip equations (see
sensitivity and calling for low settings compared with the CT Section VII).
secondary values. Second, an external fault detection logic is implemented to
Second, a through fault across the two breakers may signal the occurrence of an external fault and increase security
challenge protection security, particularly if the remote line beyond the natural level provided by the restraint terms.
terminals are weak and/or the applied settings are sensitive.
B. Partial Differential and Restraint Terms
With reference to Fig. 3, an external fault below CT-2
draws a fault current from the local system via CT-1 and from Consider the three-terminal line configuration of Fig. 4,
the remote terminal(s) via the protected line. The local relay with each line end terminated as a dual-breaker connection.
responds to the internally summed iCT-1 and iCT-2 currents T1 T2
(iL(MEAS)). With no CT errors, this virtual current is the line
current at the local terminal. When considering CT errors, i1 CT-1 CT-3 i3
however, this current is measured with a finite accuracy as
i(T1) i(T2)
compared to the real line current (iL(TRUE)).
Strong
system i2 CT-2 CT-4 i4
i(T3)

iCT-1 CT-1 CT-5 CT-6

T3
iL(MEAS)
Weak
system i5 i6
iL(TRUE)
Fig. 4. Sample three-terminal, dual-breaker line configuration.
iCT-2 CT-2
In the proposed design, each relay calculates its partial
differential and partial restraint term as follows:
i A ( T1) = i1A + i 2A (1a)

Fig. 3. A through-fault scenario challenging protection security of a dual- i B( T1) = i1B + i 2B (1b)
breaker line application.
i C( T1) = i1C + i 2C (1c)
If the fault current is high and the CT carrying the current
away from the line terminal saturates (CT-2, in this case), I AR ( T1) = I1A + I2A (2a)
enough error can develop in the measured line current, and its
direction may be reversed with respect to the true current, I BR ( T1) = I1B + I2B (2b)
especially if the true current is low because the remote line
ICR ( T1) = I1C + I 2C (2c)
terminal(s) are relatively weak.
Close-in reverse faults depress the voltages, while CT IQR ( T1) = I1Q + I2Q (3a)
errors can reverse the measured line current. With enough CT
saturation, any line protection method can be defeated in this IGR ( T1) = I1G + I2G (3b)
situation: distance, ground directional overcurrent, line current
differential, phase comparison, and so on, unless the relay where lowercase symbols stand for instantaneous values, and
design and/or the application logic address the issue explicitly. uppercase symbols denote magnitudes.
Ground (zero-sequence) and negative-sequence elements Similar terms are calculated for Terminals T2 and T3.
are particularly vulnerable during faults that do not produce a If required, CT ratio matching between the local CTs and
sequence component in the primary currents (three-phase remote CTs of the protected line is performed prior to the
symmetrical faults and line-to-line faults for ground elements, calculations.
and three-phase faults for negative-sequence elements). Under The above quantities constitute the core protection payload
these fault conditions, the true primary sequence current is per Table I. Each terminal calculates its partial terms and
zero, but a CT error will generate a secondary sequence sends them to its peers.
current. Assume now that an external fault occurs at the T1
This paper is concerned with the phase (87LP), ground terminal. Under CT saturation, the partial differential current
(87LG), and negative-sequence (87LQ) line current sent by this terminal may have a considerable error in it.
However, at the same time, this terminal sends a restraint term
36
that reflects the external fault current, feeding the Alpha Plane
trip equations with information to counterbalance the errors in
the differential signal.
Upon receiving and aligning all the partial terms, each
relay calculates the total line differential and restraint currents
for the 87LP function (A-phase is shown; B- and C-phases are
similar):
i ADIF = i A ( T1) + i A ( T2 ) + i A ( T3) = ...
(4)
... = i1A + i 2A + i3A + i 4A + i5A + i 6A
I ARST = I AR ( T1) + I AR ( T2) + I AR ( T3) = ...
(5)
... = I1A + I 2A + I3A + I 4A + I5A + I6A
and for the 87LQ and 87LG functions:
IQRST = IQR ( T1) + IQR ( T2) + IQR ( T3) = ...
(6)
... = I1Q + I2Q + I3Q + I4Q + I5Q + I6Q
IGRST = IGR ( T1) + IGR ( T2) + IGR ( T3) = ...
(7)
... = I1G + I 2G + I3G + I 4G + I5G + I6G
In this way, each relay in the line current differential
system derives the true value of the restraint current,
regardless of the location of the fault and the short-circuit
capacity behind any given relay. For example, the T3 terminal
may be very weak, therefore producing very little restraint for
a fault at T1. However, it will receive the T1 partial restraint
values to counterbalance possible errors in the T1 partial
differential current.
Note that the proposed design is scalable and works with
any number of local currents without the need to modify the
communications package or increase the bandwidth. The other
local currents can be line reactor currents, calculated line-
charging currents, or currents of a small bus included in the
line protection zone, as long as the relay hardware supports
extra current inputs.
The line differential and restraint currents feed into
generalized Alpha Plane trip equations, as explained in
Section VII.
C. External Fault Detection Logic
In addition to relying on natural levels of restraint, the
solution described in this paper incorporates an explicit
external fault detection (EFD) logic with the intent to increase
security further during external faults and relax the CT
requirements and related engineering effort of verifying the
CTs.
Fig. 5 presents the EFD logic. This method is successfully
used in bus and transformer differential relays [3] [4]. An
increase in the instantaneous restraining signal (above the
threshold P) without a similar increase in the differential
current (multiplier q) signifies an external fault. The dropout
timer (DPO) ensures security throughout the fault duration.
Σ
Σ
Fig. 5. External fault detection logic.
A simplistic implementation of this method would require
instantaneous values of all the currents of the line differential
zone to be communicated between the relays. This
requirement is not practical and has been overcome as follows.
Each terminal uses (8) to derive its own instantaneous
restraining current based on its local currents and the partial
differential terms from the remote terminals (refer to Fig. 4;
the phase index—A, B, or C—is omitted).
i RST ( T1) = i1 + i 2 + i( T2 ) + i( T3) (8a)
i RST ( T2 ) = i3 + i 4 + i( T1) + i( T3) (8b)
i RST ( T3) = i5 + i 6 + i( T1) + i ( T2) (8c)
where x stands for the absolute value of an instantaneous
signal x.
The above terms are not communicated between the relays
but used locally. Each relay executes the logic of Fig. 5 using
its instantaneous local restraint (8) and the line differential
current (4).
The relay at the terminal with an external fault is
guaranteed to detect the fault because it measures the through-
fault current. The other relays may or may not detect the fault,
depending on the current flow between the terminals. If a
given terminal is weak, it may not detect the external fault
located at the remote strong terminal.
Therefore, the EFD flag derived locally is added to the
communications payload and consolidated with the other
terminals upon reception, as shown in Fig. 6.
Stub bus To outgoing packets
EFDA
Local terminal EFDB
EFD
EFDC
Remote terminals EFD1
(incoming packets)
EFD2
EFD3
...
Fig. 6. Consolidating EFD flags between phases and line terminals.
In this way, all the relays work with the same amount of
information, and all behave correctly. A weak terminal, T3 for
example, may receive an incorrect value of the partial
differential current from T1 because of an external fault at T1
and considerable CT saturation. On its own, the T3 terminal
may not be able to detect the event as an external fault, but it
receives the EFD flag from the T1 terminal informing it
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
explicitly about the external fault. As a result, a differential
term in any given packet with a potential error because of an
external fault and CT saturation is always accompanied by the
EFD flag and an elevated restraint term. In our example, the
T3 terminal is fully prepared to deal with the event at T1, even
though it does not have any direct visibility into this event
(i.e., it does not measure the i1 and i2 currents directly).
Fig. 7 presents a simulation example illustrating the
restraint terms and the EFD logic. The terminal with the
external AB fault asserts the EFD flag safely before CT
saturation impacts accuracy and produces considerable errors
in the A-phase current sent by this terminal. Note that the
restraint term in the A-phase is considerable, already
stabilizing the 87LP function.
iCT-2, A
iCT-1, A
i, A
IR, A
i, A-phase, A
i, B-phase, A
Fig. 7. Critical signals at the dual-breaker terminal for an external AB fault:
currents across the two breakers, partial differential current (measured and
actual), partial restraint, delta changes in differential and restraint currents and
the EFD flags.
The described logic delivers similar performance to that of
a modern, low-impedance bus or transformer relay. It is
capable of detecting external faults even if considerably
underrated CTs are installed. As long as the CTs perform with
no or only minor errors in the first 3 milliseconds of an
external fault, the logic works correctly. This level of
performance is achieved at the expense of adding just 1 bit to
the communications payload.
37
The EFD logic is scalable and works seamlessly with any
number of local currents at any given terminal without the
need to modify or expand the packet.
Upon asserting the EFD flag, the relay increases security
by intentionally raising its natural restraint terms. This may
include harmonic restraint—adding harmonics in the
differential current to the restraint or adding a portion of the
phase restraints to the negative- and zero-sequence restraint
terms to secure these elements under external faults that do not
produce any natural sequence restraint. The intentionally
amplified auxiliary restraint term feeds into the Alpha Plane
trip equations as explained, further increasing protection
security.
VI. MULTITERMINAL LINE APPLICATIONS
Line current differential relays have been primarily
designed to cover two- and three-terminal line applications.
Typically, two communications ports are provided on a relay.
These ports can be used to communicate with two remote
relays in a nonredundant way, allowing for a three-terminal
line application. Or the two ports can be used for redundant
communication between two relays in a two-terminal
application.
Recently, reasons to provide for more than three-terminal
line applications became more compelling.
Some utilities tap subtransmission and transmission lines to
feed distribution stations and larger industrial loads. In some
cases, five taps can be encountered, constituting a seven-
terminal application. Instantaneous fault clearing in such cases
is typically accomplished using distance protection in a pilot-
assisted scheme, with the pilot installed only between the two
stations. The distance functions are set to overreach the far-
end bus but set short of the low-voltage buses of the tapped
transformers. This setting rule may be difficult to meet for
longer lines and/or when a large transformer is located close to
one of the substations. In such cases, extra pilot channels may
be required between some of the taps and the main terminals
to allow explicit blocking for faults in the lower voltage
system.
The application is usually complicated. The line current
differential protection principle is a better alternative for lines
with multiple taps. With a constantly progressing deployment
of digital communications channels and advancements in relay
design, line current differential schemes became more readily
available.
More recently, distributed generation has been tapped into
subtransmission or transmission lines the same way loads used
to be tapped. This poses an even greater challenge, as the
short-circuit response of these nontraditional sources may
cause problems or uncertainty for traditional protection
principles. These sources often include power electronics with
fast controllers designed to provide a low-voltage ride-through
capability or just protect the source itself from damage.
Protection engineers struggle with modeling these generators
for short-circuit studies, and as a result, some level of
uncertainty remains in these applications.
38
Because lines tapping new generation are essential to
having the generation available in the first place, they may
apply single-pole tripping and reclosing. This complicates
protection applications further.
Very often, however, lines connecting distributed
generation have digital communications channels available.
Line current differential protection is a natural solution in this
case, eliminating complexity and uncertainty related to the
behavior of nontraditional sources, while reducing the danger
of overtripping and increasing availability of the added
generation.
In order to provide multiterminal (N-terminal) line current
differential protection, a differential relay must first resolve
the following issues:
• Consolidation of all the terminal currents for
protection calculations.
• Connectivity between N terminals.
• Synchronization between N terminals, given their
actual connectivity pattern.
The solution outlined earlier in this paper of sending partial
differential and partial restraint terms from each relay solves
the first problem. Regardless of the number of local currents at
each line terminal (single breaker, dual breaker, small bus),
the same optimized packet conveys all the information
required to provide a high-performance line current
differential function. The next section explains how the Alpha
Plane principle is applied to an arbitrary number of terminals.
Another challenge is the connectivity between N terminals.
Direct point-to-point connections require each relay to support
N – 1 communications ports for exchange of data with all its
peers (e.g., six ports for a seven-terminal application). This
increases the relay part count and, as a result, impacts cost,
availability, and reliability.
In addition, the user needs to provide a number of point-to-
point connections: 1 channel for 2 terminals, 3 for 3 terminals,
6 for 4 terminals, 9 for 5 terminals, and so on. This of course
assumes all relays are masters. A hybrid solution allows some
relays to be slaves (i.e., only serve the data) with a few other
relays being masters capable of receiving all the data,
asserting a trip, and sending it via DTT to the slaves. This
reduces the number of required connections, at the cost of
slowing down the operation.
Channel-based synchronization may be an even greater
challenge, depending on the method applied. Solutions that
force synchronization of relay sampling clocks face more
difficulties.
The challenge of connectivity and data synchronization is
easy to solve from the point of view of a modern
communications system such as SONET/SDH. Recently,
these communications systems have provided for
deterministic transport mechanisms, including “deterministic
Ethernet” for protection applications as well as common
timing with the capacity of being a GPS-independent common
time source [1].
10
In this solution, the external time reference synchronization
method is used (see Section XI), while deterministic Ethernet
with guaranteed bandwidth and latency becomes the data
transport mechanism (Fig. 8). In a way, the next generation
SONET/SDH built for utility applications, including
protection and synchrophasors, becomes a part of the line
current differential system. This is a simple and robust
solution for N-terminal line current differential relaying
because of the following:
• The SONET/SDH network is entirely under the
control of the user. It can be deployed and configured
for redundancy and is operated and maintained as a
protection-grade system.
• The common time reference provided by the system is
independent from GPS. Even if all dispersed receivers
are lost, the system maintains common time across the
wide area.
• With the SONET/SDH deterministic communications
backbone, a portion of Ethernet traffic is isolated for
deterministic delivery across a wide area. In this way,
Ethernet is more of a connectivity media between the
line current differential relays and the multiplexers,
while the actual transport is rigorously controlled by
the next generation SONET, using its deterministic
backbone.
• Using the same system for timing and data transport
improves reliability by reducing the total number of
devices and associated failure modes.
M
UX
MUX
MUX
MU
X
Fig. 8. Multiterminal line current differential application with next
generation SONET/SDH systems.
Multiterminal line current differential applications can be
implemented in a simple and robust way by utilizing a next
generation SONET/SDH system and tasking it with providing
a common wide-area time reference independent from GPS
and deterministic Ethernet connectivity in the relay-
multiplexer-fiber-multiplexer-relay path.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia11 39
VII. GENERALIZED ALPHA PLANE FOR MULTITERMINAL
APPLICATIONS
A. Introduction
The Alpha Plane current differential protection principle
compares individual magnitudes and angles of the zone
currents. The principle is easy to understand for the case of a
two-terminal line. Under balanced conditions, the two currents
of the zone (IL and IR) are equal in magnitude and opposite in
phase. This yields an operating point on the Alpha Plane of
k = 1∠180°. Under internal faults, the complex current ratio,
k, departs from this ideal blocking point, allowing the Alpha
Plane element to operate. Blocking and operating regions are
shaped as pictured in Fig. 9, with typically two easy-to-set
parameters.
Im(k)
RESTRAIN
OPERATE
g le
An
Re(k)
–1
s
diu
Ra
Fig. 9. Typical Alpha Plane characteristic.
Typically, the phase (87LP) and either the negative-
sequence (87LQ) or ground (87LG) Alpha Plane elements are
applied.
Because of its many advantages, the principle has been
successfully implemented and proven by a long field record
[5]. The most important advantages are: significant tolerance
to CT saturation and synchronization errors, excellent
sensitivity and speed of the 87LQ and 87LG elements,
significant tolerance of the 87LQ and 87LG elements to the
line-charging current, and good performance in series-
compensated lines.
While intuitive and straightforward in two-terminal
applications, the Alpha Plane is less intuitive in a general N-
terminal case. Complex current flow patterns can be
encountered, such as a circulating current—a current leaving
the zone at one terminal to reenter it at the other. These
patterns must be analyzed carefully in order to avoid a failure
to trip by responding to one of the currents flowing out of the
zone to feed a load or circulating to the other line terminal.
Many possible permutations of ratios between many possible
currents complicate understanding, implementation, testing,
and post-event analysis.
This paper introduces a generalized N-terminal Alpha
Plane concept. This method calculates a two-terminal
equivalent for a general N-terminal case and applies the tried-
and-true Alpha Plane principle to the two equivalent currents.
B. Mapping N-Terminal Currents Into Two-Terminal
Currents
The general N-terminal Alpha Plane works with two
equivalent currents (IL EQ and IR EQ) calculated from the N
currents of an arbitrary N-terminal zone, using the differential
and restraint terms as a transformation tool.
Consider a general N-terminal differential zone of
protection, as shown in Fig. 10a. The classical differential
principle derives the following differential and restraining
currents for this zone (all currents are phasors in the following
equations, and | | stands for the phasor magnitude):
I DIF( N ) = ∑ nN=1 In (9a)
I RST ( N ) = ∑ nN=1 I n (9b)
(a) (b)
I2 IL EQ
I1
I3
IDIF(N) IDIF(2)
IRST(N) IRST(2)
IN
I4 IR EQ
Fig. 10. The principle of a generalized Alpha Plane.
Now consider the two-terminal case of Fig. 10b—the
natural application for the Alpha Plane.
The two virtual currents in the two-terminal equivalent are
sought such that the same differential and the same restraining
currents are measured in the two-terminal equivalent as in the
actual N-terminal application.
I DIF( 2 ) = I DIF( N ) (10a)
I RST( 2) = I RST ( N ) (10b)
The two currents of the two-terminal equivalent have a
total of four degrees of freedom (two magnitudes and two
angles), while we have a total of three boundary equations: the
real and imaginary parts of the differential current (10a) and
the magnitude of the restraining current (10b).
In this method of representing N currents of a differential
zone by two equivalent currents for the Alpha Plane, the
fourth balance equation calls for one of the two sought
equivalent currents to be along the line of a specific current of
the N-terminal zone. This specific zone current (IP) is selected
as the one that has the largest projection on the differential
current phasor.
The rationale behind this choice is that during external
faults with CT saturation, the spurious differential signal, if
40

significant, will be approximately located along the line of the
fault current. Therefore, by selecting the reference current IP
that is closest in phase to the differential current, we position
the two equivalent Alpha Plane currents along the lines of the
current flowing in and out of the zone.
To select the reference current IP, the following auxiliary
numbers are calculated first:
(
R n = real In • I*DIF( N ) ) (11)
where n = 1..N.
The current with the highest value of R becomes the
Solving for the two-current equivalent, we obtain:
IL EQ = 11.1 A∠11.7° and IR EQ = 18.9 A∠160°
Note that when calculated for this two-terminal equivalent,
the differential and restraining currents are 11.2 A∠128° and
30 A, exactly as in the original three-terminal system.
The above two equivalent currents give the operating point
on the Alpha Plane of 1.71∠148.3°. Fig. 11 presents this case
graphically.
Three-terminal application Two-terminal equivalent
IRST IRST

reference current IP.

IDIF IDIF
Let us denote the angle of this current as β:

Imaginary

Imaginary
I3 IR EQ
I1 IL EQ
β = angle ( Ip ) (12) I2

The differential current is shifted for the convenience of

subsequent calculations as follows:
I X = IDIF( N ) •1∠ ( – β ) (13) Real Real

And the two currents of the two-terminal equivalent are Fig. 11. Graphical illustration of the numerical Example 1.
now calculated as follows: 2) Example 2
I L EQ = Continue the simulation example of Fig. 7, and refer to
Fig. 12, presenting the zone differential and restraining
⎛
( ) ⎞
2
⎜ Im ( I X ) – I RST( N ) – Re ( IX )
2
currents in the A-phase. The equivalent Alpha Plane yields an
⎟ (14a)
⎜ + j• Im ( IX ) ⎟ •1∠β
( )
operating point of about 0.5∠170°. Note that in this case, the
⎜ 2 • I RST ( N ) – Re ( IX ) ⎟ IP current is selected with some approximation, as the line
⎝ ⎠
current differential system does not work directly with the
(
I R EQ = IRST ( N ) – IL EQ •1∠β ) (14b) individual currents at the faulted terminal. Still, the large
restraint term compared with the spurious differential keeps
The traditional Alpha Plane protection principle takes over the equivalent Alpha Plane in the blocking region.
from here, working with the IL EQ and IR EQ currents. 100
The classical differential principle has been used in this
0
iCT-2, A

method as a mathematical mapping tool to project the general

case of an N-terminal differential zone into an equivalent two- –100

terminal zone, requiring the differential and restraining –200

0.2 0.25 0.3 0.35
currents be identical between the N-terminal application and EFD
its two-terminal equivalent. 0.2 0.25 0.3 0.35
The principle is applied to the phase, negative-sequence,
IADIF, IARST, A

150
and ground differential elements with the filtered differential
100
and restraint currents obtained via (4) through (7) from the
partial terms communicated between the relays. 50

0
C. Numerical Examples 0.2 0.25 0.3 0.35

1) Example 1
kMAG (87LP)

1
Consider the following three-terminal application:
0.5 Set point:
I1 = 10.0 A∠160° Radius = 5
I2 = 8.0 A∠–175° 0
0.2 0.25 0.3 0.35
I3 = 12.0 A∠30°
200
kANG (87LP)

The differential current is IDIF(N) = 11.2 A∠128°. 150

Set point:
Following the proposed method, we calculate: 100 Stability angle = 120°
R1 = 97.37A2, R2 = 49.50A2, R3 = 20.14A2 50
0
Therefore, the first current is selected as the reference, 0.2 0.25 0.3 0.35
time, s
IP = 10.0 A∠160°. This means one of the equivalent currents
will be located on the line of 160° or –20°. Fig. 12. Secondary current of the saturated A-phase CT for the case of
Fig. 7, line differential and restraint currents, and the equivalent Alpha
The restraining current is IRST(N) = 30.0 A. Plane—magnitude and angle.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
D. Discussion
The generalized Alpha Plane allows implementation of the
tried-and-true two-terminal principle to multiterminal lines,
retaining key advantages while enabling new applications. The
following are worth noticing:
• The generalized principle is transparent in two-
terminal applications—the two equivalent currents
always equal the two actual currents.
• In three-terminal applications, the principle behaves
nearly identically, with the heuristic implementation
of the two-terminal rule developed by analyzing
possible current flow patterns.
• Any case with a balanced differential current yields an
ideal blocking point on the Alpha Plane (1∠180°).
Decreasing the differential current—such as by line-
charging current compensation (Section VIII)—brings
the Alpha Plane point closer to the ideal blocking
position.
• Any case with higher restraint brings the Alpha Plane
point closer to the ideal blocking point. The method
allows applications where the restraint term is
intentionally increased, such as when using harmonic
restraint in transformer protection (Section IX).
• The principle works well without the need to
communicate all local currents from all terminals. The
line differential and restraint terms enabled by the
proposed communications package map well into the
generalized Alpha Plane.
• The principle works exceptionally well during external
faults under CT saturation. First, by relying on the true
restraint term, the calculated Alpha Plane point shows
a strong blocking tendency. Second, extra security is
added by the nature of the Alpha Plane itself.
• The principle works very well for the 87LQ and 87LG
elements. Under internal faults, the elements’ currents
are close in phase and differ only by the system
nonhomogeneity angles. The generalized Alpha Plane
returns a strong unblocking indication in this case,
regardless of the magnitudes of the compared currents.
Under external faults, including faults that do not
produce any natural restraint (phase-to-phase faults for
the 87LG, for example), a cross-phase restraint may
be used upon detecting an external fault (EFD bit
asserted). With increased restraint, the equivalent
Alpha Plane point shifts safely toward blocking.
• By reducing a differential zone of protection with any
number of terminals to a single operating point on the
Alpha Plane, the principle simplifies implementation,
testing, and post-event analysis.
VIII. LINE-CHARGING CURRENT COMPENSATION
A. Impact of the Line-Charging Current and Benefits of
Compensation
High-voltage overhead lines draw about 1 A per each
kilometer of length (about 1.6 A per mile). For high-voltage
cables and extra-high-voltage overhead lines, the total
41
charging current can amount to hundreds of primary amperes.
In some cases, the charging current can be comparable with
the fault current.
Line-charging current leaks from the differential
measurement and affects both the security and dependability
of differential protection. The phase differential element
(87LP) is impacted more than the sequence elements (87LQ
and 87LG).
Consider line energization, external faults, and internal
faults in the context of the line-charging current.
1) Line Energization
During line energization, the energizing terminal draws the
total charging current. If the energizing voltage is balanced
and the line well transposed, the charging current is composed
predominantly of positive-sequence current and therefore only
affects the phase elements. If the line and/or the supply
voltage are unbalanced, an unbalanced charging current may
flow. As the charging current is fed from one terminal only, it
appears as a single-feed current, and as such, it cannot be
addressed by any restraint means. Increasing the pickup
permanently above the charging current or using the remote
breaker status to temporarily boost security are practical ways
of dealing with the line-charging current during energization.
If the line is not well transposed and the total charging
current is high, we may have to increase the pickup of the
87LQ and 87LG functions considerably, potentially
diminishing their natural protection sensitivity.
2) External Faults
During external faults, changes in voltages induce extra
charging current. This includes both fault inception and
clearance. As the fault voltages are unbalanced, asymmetrical
charging currents are induced, potentially impacting the 87LQ
and 87LG functions.
This situation is not a major concern, however, because the
external fault current produces restraints for all three 87LP,
87LQ, and 87LG functions. These restraint terms allow
counterbalancing of the charging current component in the
differential current. Weak systems may pose some challenge
as they do not generate large currents that would boost
restraint terms but allow voltages to depress considerably, thus
creating larger charging currents.
3) Internal Faults
During internal faults, the charging current caused by the
change in voltages subtracts from the fault current, potentially
reducing sensitivity. Typically, this is not a major concern for
the 87LQ and 87LG functions because high sensitivity is
required during high-resistance ground faults, and these faults
do not change voltages much. Therefore, only small zero- and
negative-sequence charging currents are generated during
high-resistance faults. However, if the pickup settings are
increased to deal with energization of a poorly transposed line,
sensitivity may be impacted.
In general, line-charging current is not a major concern for
the 87LQ and 87LG functions, unless the line is not well
transposed or is operated under considerable unbalance (e.g.,
caused by single-phase reactor operation).
42

If high sensitivity is required from the 87LP function, line- The average line voltage can be approximated by the
charging current is, however, a significant concern. average terminal voltage, and therefore:
Line-charging current can be compensated for in a line 1 d
current differential relay using voltage signals. This section i C _ TOTAL = CTOTAL • • ( v1 + v 2 + v3 ) (15b)
3 dt
explains and illustrates the basic principles of compensation
and presents a method to make the compensation immune to Rearranging further:
problems with voltage transformers (VTs). 1 d 1 d
i C _ TOTAL = • CTOTAL • v1 + • CTOTAL • v 2 + ...
B. Principle of Line-Charging Current Compensation 3 dt 3 dt
(15c)
The purpose of line-charging compensation is to remove 1 d
... + • CTOTAL • v3
the charging current from the differential current. Ideally, the 3 dt
method applied should work under balanced and unbalanced In other words, the total charging current is the sum of
conditions, transposed and nontransposed lines, energization, three components:
external faults, internal faults, and other events.
i C _ TOTAL = i C1 + i C2 + i C3 (16a)
With reference to Fig. 13, a multiterminal line draws a
charging current through its distributed capacitances. The each derived from a single line terminal voltage:
exact distribution of this current depends on the line and 1 d
system parameters, as well as on the voltage profile along the i C1 = • CTOTAL • v1 (16b)
3 dt
line and its segments. Higher voltages draw larger charging
currents. Open-ended lines develop an overvoltage at the open 1 d
i C2 = • C TOTAL • v 2 (16c)
end while not drawing any current from that end. During 3 dt
faults (internal or external) voltages change and become
1 d
unbalanced, causing changes in the charging current, with the i C3 = • CTOTAL • v3 (16d)
charge flowing out and into the line. 3 dt
The above explains that each terminal can calculate a share
of the total charging current based on local voltage and a
portion of the total line capacitance proportional to the number
of line terminals that perform the compensation at any given
time.
These fractions of the charging current when summed up
by the final differential current will null out the actual
Fig. 13. Distributed capacitance three-terminal line. charging current.
This principle is key—the line current differential system
From the differential protection point of view, however, the
effectively calculates the charging current based on the
total charging current is of primary interest. Contributions to
average voltage from all line terminals, without sending any
the differential current from the individual line terminals are
voltages between the relays. Instead, each terminal subtracts
secondary. If so, the total line-charging current can be well
its share of the charging current from the measured current
approximated as a current drawn by the total line capacitance
and sends it to its peers (TX is transmitted to the peers; RX is
under the average line voltage. The former is known and
received from the peers):
becomes a user setting. The latter can be calculated from the
measured line terminal voltages. i TX = i MEASURED – i C (17a)
With reference to Fig. 14, the line capacitance can be and calculates the line differential current as:
represented by a lumped parameter model at each terminal of
i DIF = i TX + ∑ i RX (17b)
the line that allows the line current differential relay to
measure the voltage. Using Fig. 14 as an example, the differential current is
calculated as follows:
i DIF = ( i MEASURED1 – i C1 ) + ( i MEASURED2 – i C2 ) +
(18a)
( iMEASURED3 – iC3 )
or
i DIF = i MEASURED1 + i MEASURED2 + i MEASURED3 –
(18b)
Fig. 14. Lumped parameter three-terminal line. ( iC1 + iC2 + iC3 )
or
d
i C _ TOTAL = CTOTAL • vAVERAGE (15a) i DIF = i MEASURED1 + i MEASURED2 + i MEASURED3 – iC _ TOTAL (18c)
dt
The share of the charging current estimated at a given
terminal may not equal the actual charging current supplied by
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 43

this terminal. The open line end is the ultimate case—the

actual current supplied by the open terminal is zero, while this

v1, kV
terminal estimates its share of the total charging current based
on the voltage at the open end, provided line-side VTs are
installed. At the same time, the closed terminal underestimates
its share. When added up in the differential calculations, all
the estimates of the charging current will, however, match the
actual total charging current of the line.
In general, for a line with N terminals performing charging

iC1, A
current compensation, each terminal uses 1/N of the total line
capacitance and its own voltage to estimate its share of the
charging current.
C. Three-Phase Implementation
As indicated by (15), the charging current is calculated as a Fig. 15. Sample line energization: Terminal 1 voltage and the calculated
derivative of the voltage signal, assuming known capacitance share of the charging current.
of the line. Fig. 16 shows the voltage and the calculated share of the
A microprocessor-based relay uses digital filtering to charging current for the second terminal. Again, the actual
calculate the time derivative. current supplied by this terminal is zero.
In general, the following equation is used to calculate the
phase-charging currents:
⎡i A ⎤ ⎡C AA CAB CAC ⎤ ⎡ vA ⎤
v2, kV
⎢ ⎥ ⎢ ⎥ d ⎢ ⎥
i
⎢ B⎥ = ⎢ CBA CBB CBC ⎥ • ⎢ v B ⎥ (19)
dt
⎢⎣ i C ⎥⎦ ⎢⎣ CCA CCB CCC ⎥⎦ ⎢⎣ vC ⎥⎦
C

For fully transposed lines, the matrix is symmetrical and is

composed of the self- and mutual capacitances calculated from
the zero- and positive-sequence susceptances of the line.
For nontransposed lines, the matrix is not symmetrical, and
iC2, A

extra settings (in addition to the zero- and positive-sequence

susceptances) are required to determine the matrix and
perform accurate compensation.
Data required for the compensation are typically available
from the commissioning records of the protected line. Fig. 16. Sample line energization: Terminal 2 voltage and the calculated
Symmetrical components of the charging current are share of the charging current.
compensated for automatically by compensating the phase Fig. 17 shows the voltage and the calculated share of the
currents using (19). charging current for the third terminal. Note that this terminal
Equation (19) is the time domain implementation of the supplies the actual charging current, and its calculated share is
method. Not only the fundamental frequency component but only about one-third of the actual current.
also the instantaneous values of the differential current are
compensated. This allows for various algorithms that respond
to signal features other than the fundamental frequency
component to work well.
Note that the implementation method (19) works well
under a variety of system conditions: energization, faults,
open-pole conditions, and so on.
D. A Simulation Example
Fig. 15 through Fig. 17 illustrate a case of energizing a
three-terminal 275 kV line, with a total length of
300 kilometers and the steady-state positive-sequence
charging current of 230 A. The line is energized from the third
terminal.
Fig. 15 shows the voltage and the calculated share of the
Fig. 17. Sample line energization: Terminal 3 voltage and the calculated
charging current for the first terminal. Note that the actual share of the charging current.
charging current drawn from this terminal is zero (open
breaker).
44

Fig. 18 compares the actual and calculated charging
currents. The actual charging current is the current measured
at the energizing terminal (Terminal 3, in this example). The
calculated current is the total of the charging current shares
calculated at each of the line terminals.
iC (actual), A
E. Charging Current Compensation and Alpha Plane
The operating point on the Alpha Plane is derived from the
differential and restraining auxiliary terms, as explained in
Section VII.
By reducing the differential signal, the charging current
compensation shifts the Alpha Plane operating point toward
the blocking position. In an ultimate case of removing all the

1000

0 charging current, it yields the operating point of 1∠180° on

–1000
the Alpha Plane. This is not only the desired effect, but it
happens without the knowledge of how the actual charging
0.05 0.1 0.15 0.2 current divides between the terminals of the line.
During internal faults, the charging current compensation
increases the differential signal by not allowing its inductive
iC (calculated), A

1000
0 This compensation shifts the operating point on the Alpha
–1000
0.05 0.1 0.15
time, s
Fig. 18. Sample line energization: actual and calculated charging currents.
To illustrate the effect and accuracy of the compensation,
Fig. 19 compares the differential currents without and with
compensation. The differential current without compensation
is the current measured at the energizing terminal. The
differential current with compensation is the current calculated
following (17). Note that a vast portion of the charging current
is removed from the differential signal, and the remainder is of
higher frequencies and will be further suppressed by the relay
filtering algorithms.
iDIFA, A
component to be reduced by the capacitive charging current.
Plane away from the blocking region as expected.
As a result, the differential elements (phase elements, in
0.2
particular) can be set more sensitively when the charging
current compensation is enabled.
An interesting design question arises regarding adding the
calculated charging current to the restraint terms. On the
surface, this is yet another current that feeds into differential
calculations and as such should be included in the restraining
terms that normally mirror the differential calculations. In this
case, however, the proper design choice is not to include the
calculated charging current in the restraint terms for the
following reasons.
The actual charging current is measured by the relays and
already contributes to the restraint terms. The calculated
charging current is the countermeasure and should not be
included, or else the charging current would be counted twice

in the restraining terms.

In order to illustrate this issue, consider the case of line
energization shown in Fig. 15 to Fig. 17. The actual charging
current is measured at the third terminal and, in steady state,
produces about 230 A of phase restraint current. At the same
iDIFB, A

time, the compensated charging current is near zero. These

values yield an operating point of 1∠180° on the Alpha Plane,
or a solid blocking indication.
This shows that the charging current compensation allows
the trip equations to restrain properly rather than to block via
iDIFC, A

an elevated pickup setting.

Fig. 19. Sample line energization: differential current without (solid line)
and with (dashed line) charging current compensation.
F. Accuracy of Compensation
The applied compensation method uses a lumped
parameter model to estimate and subtract the line-charging
current. This lumped model represents actual transmission
lines well for frequencies up to a few hundred hertz.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Fig. 20 presents a frequency response of an admittance of a
sample overhead line for several different line lengths,
superimposed on the lumped parameter model. As we can see,
the actual line and the lumped parameter model can differ
considerably at higher frequencies, particularly for long lines
(and it is long lines that benefit most from the compensation).
Fig. 20. Frequency response (admittance) of the actual line and the lumped
parameter model.
In theory, we may develop a more accurate model of the
line (compared with the C • dv/dt) and implement it in a line
current differential relay. The following are the disadvantages
of this approach:
• The behavior of long transmission lines at higher
frequencies is difficult to model without considerable
amounts of data. There are differences between
overhead transmission lines and cables. Mixed lines
with some cable and overhead sections are difficult to
cover. Line geometry, skin effects, and zero-sequence
resistance have impact on the model. This approach
would put extra burden on the user.
• The frequency response of the VTs would have to be
factored in, making the problem even more complex.
The under-/overcompensated higher frequency components
are not used for protection purposes and therefore do not have
to be measured correctly. Instead, they are noise impacting
other measurements and must be dealt with as such using
known protection approaches, if they cannot be eliminated
easily in a practical way.
One way to deal with the under-/overcompensated high-
frequency charging current components is to produce an extra
restraint from the high-frequency spectrum in the differential
signal. With reference to Fig. 21, a high-pass filter is applied
to extract high-frequency components from the instantaneous
differential signal, and a root-mean-square (rms) measurement
is used to boost the fundamental frequency restraint. In this
way, if the high-frequency component affects measurements
of the differential signal, it also automatically increases the
corresponding restraint term. As a result, the equivalent Alpha
Plane is kept secure.
45
Σ
Fig. 21. Extra restraint term from the high-frequency spectrum in the
differential current.
The desired approach is simply to remove the charging
current for frequencies near the fundamental frequency
because these frequencies affect the transient and steady-state
response of the relay, while providing an extra restraint
derived from higher frequencies in the differential signal.
G. Treatment of Line Reactors
Long lines that benefit from the charging current
compensation feature of the relay often have shunt reactors
installed inside the protection zone bounded by the location of
the CTs connecting the line current differential scheme. The
capacitive current of the line and the inductive current of the
reactors do not cancel transiently as far as fast protection is
concerned. The transient nature (frequency response) is
different between an inductor and a capacitor, their positive-
to zero-sequence reactance ratios can be different, and reactor
saturation makes the inductance nonlinear. In addition,
reactors are switched on and off as a part of voltage/reactive
power control in the power system and can be operated in an
unbalanced way (one or two phases).
In order to keep protection applications simple, the reactor
current is typically taken out of the measuring zone by
paralleling its CTs with the line CTs. At any given time, the
line may or may not be compensated, but the relay always
measures the entire charging current and compensates for it.
With reference to Fig. 22, when applying the line-charging
current compensation, the line differential zone excludes both
the reactors and the charging current itself.
iCT-1
iC
Differential zone
iCT-2 iREACTOR
Fig. 22. Line reactors and line capacitance are removed from the differential
protection zone.
H. Immunity to VT Problems
Charging current compensation makes the line current
differential function partially dependent on availability of the
voltage sources and exposes it to problems with the voltage
signals. This is not a best solution, and the dependence on
voltage should be remedied as much as possible.
The proposed scheme uses an extra bit in the
communications payload to signal if the terminal actually
46
performs the compensation (charging current compensation in
progress, CCC).
A relay at a given terminal may not calculate and subtract
its share of the charging current for the following reasons:
VTs are not installed or not wired to the relay; a problem is
detected with the VTs (loss of potential), and the charging
current compensation logic suspends compensation based on
this voltage; or bus-side VTs are used while the line breaker is
opened, and the logic suspends usage of this voltage as not
representative for the line.
Each set of received currents is either compensated or not
as per the accompanying CCC flag. Each receiving terminal
therefore knows how many terminals actually subtract their
share of the charging current and can calculate its own share
of compensation in order to make up for the full charging
current of the line.
In this way, the system works with a variable number of
compensating terminals and is partially immune to the loss of
one or more sets of VTs.
Consider the following examples.
1) Example 1
Assume a three-terminal application and one terminal with
no voltage source. The relay with no voltage does not
compensate its currents and permanently sends CCC = 0 to the
other peers. Each of the receiving peers sees that the said
terminal does not compensate, and there are a total of two
terminals that do compensate. As a result, each of them will
apply a multiplier of 1/2 in their equations for the share of the
charging current. In this way, the charging current is
compensated fully, using an average of the two terminal
voltages that are available to the line current differential
system. Note that the terminal with no voltage still works with
a fully compensated differential current.
2) Example 2
Assume a three-terminal application and one terminal with
a bus-side voltage. Normally, each terminal performs the
compensation by applying the multiplier of 1/3 and asserting
the CCC bit. At the moment the terminal that uses a bus-side
voltage detects any pole open, it deasserts its CCC flag and
ceases to compensate its current. The other two terminals
recognize that from the received CCC flag, switch their
multipliers from 1/3 to 1/2, and the compensation continues,
using the two voltages that are representative of the line
voltage profile.
3) Example 3
Continue the second example, and assume that
subsequently one of the other terminals detects loss of
potential. It deasserts the CCC flag and stops compensating its
outgoing current. The only terminal with a healthy voltage
now knows that the second peer lost the compensation and
switches its multiplier from 1/2 to 1. In this way, the
compensation is done with the single voltage that is still
available. As such, it will be less accurate but still useful. The
other two terminals still receive CCC = 1 from the last
terminal capable of compensation and know that the line
differential current is fully compensated.
This scheme makes the charging current compensation
feature more immune to problems with VTs. In a way,
individual relays in the line current differential system work as
remote backup voltage sources for each other. If all voltages
are lost, the relay automatically desensitizes to ensure security
and continues working with currents only.
IX. IN-LINE TRANSFORMERS
A. Introduction
In some applications, a line and a transformer are installed
without a breaker in between to separate the two pieces of
equipment (see Fig. 23). This is often driven by economics,
particularly in cases when the line is not tapped or
multiterminal, and the transformer does not feed any other
loads. Thus, installing a separating breaker does not add any
operational flexibility. The lack of a separating breaker puts
both the line and the transformer into the same trip zone,
regardless of which element actually requires isolation from
the rest of the system.
It is still beneficial to apply two measuring zones in this
case, as shown in Fig. 23a, even if they both trip both the line
and the transformer. By using relays designed for a given type
of apparatus, we maximize the overall performance of the
protection system, sensitivity to transformer faults in
particular.
(a)
Communications
87 channel 87
87 L L
T
DTT
(b)
87 Communications channel 87
L+T L+T
Fig. 23. Transformer and line protection—individual (a) and combined (b)
measuring zones.
There is a trend, however, to combine the two measuring
zones in a line current differential relay with an in-line
transformer feature, as shown in Fig. 23b. This approach
simplifies the application because fewer relays, less wiring,
and less labor are required.
This section elaborates on the relay design to support in-
line transformers.
B. Transformer Protection Using the Alpha Plane Differential
Element
Being a differential principle, the Alpha Plane technique
can provide short-circuit protection for power transformers.
First, the local currents are compensated for transformer
connections to follow the art of transformer differential
protection. Commonly referred to as ratio (tap) matching,
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 47

zero-sequence removal, and vector group compensation, the For example, the line terminal associated with the delta
differential current equations actually match ampere-turns for winding in this example performs the following operations to
a healthy transformer. For example, for a delta/wye obtain its instantaneous currents for transmission:
transformer, the following equations describe the differential i A = k Δ • i ΔA (22a)
signal:
i B = k Δ • i ΔB (22b)
1
i DIF(1) = k Δ • i ΔA + k Y • • ( i YA – i YB ) (20a) i C = k Δ • i ΔC (22c)
3
1 while the terminal associated with the wye winding obtains its
i DIF( 2 ) = k Δ • i ΔB + k Y • • ( i YB – i YC ) (20b) terminal currents using the following equations:
3
1
1 iA = k Y • • ( i YA – i YB ) (23a)
i DIF( 3) = k Δ • i ΔC + k Y • • ( i YC – i YA ) (20c) 3
3
1
The taps, kΔ and kY, are based on the CT and transformer iB = k Y • • ( i YB – i YC ) (23b)
ratios, and indices 1, 2, and 3 refer to the loops of the 3
differential function. 1
The restraining terms are created as mirror equations to the iC = k Y • • ( i YC – i YA ) (23c)
3
differential terms (20):
After operations (22) and (23), the resulting currents can be
1
i RST (1) = k Δ • i ΔA + k Y • • i YA – i YB (21a) processed as if there were no in-line transformer. In particular:
3 • The charging current is subtracted after being
1 compensated for transformer connections.
i RST ( 2 ) = k Δ • i ΔB + k Y • • i YB – i YC (21b) • The phase and sequence restraint terms are calculated.
3
• The partial differential and restraint terms are created
1 based on the number of local currents.
i RST ( 3) = k Δ • i ΔC + k Y • • i YC – i YA (21c)
3 • The partial differential and restraint terms are sent,
For a general case on an N-winding transformer, the received, and aligned.
differential and restraining signals are sums of ratio-matched • The total line differential and restraint terms are
and vector-compensated winding currents. The above calculated.
observation leads to a simple relay implementation of At this point, the nature of transformer differential
compensating the individual currents for transformer protection needs to be recognized by addressing the inrush and
connections based on the position of the current with respect overexcitation phenomena.
to the various transformer windings and running the regular Inrush and overexcitation harmonic blocking are naturally
line current differential communications and processing performed by measuring the second and fifth harmonics in the
algorithms based on the transformer-compensated currents differential current and comparing them with user thresholds.
(Fig. 24). Cross-phase blocking or the fourth harmonic can be used,
depending on the design and user preferences [4] [6].
Harmonic restraint is also easy to implement using the
generalized Alpha Plane principle of Section VII. The
harmonics of interest in the differential current (second,
Local terminal currents

fourth, and fifth) are added to the fundamental frequency

restraint terms using appropriate multipliers as per the
Σ
Transmit and align

principles of harmonic restraint.

Subsequently, the generalized Alpha Plane calculations are
executed. If the restraint terms are increased sufficiently by
the harmonics in the differential signal, the boosted restraint
Σ shifts the Alpha Plane toward the blocking point and restrains
the differential function during inrush conditions.
In order to provide adequate transformer protection, the
blocking region of the Alpha Plane needs to be set smaller in
applications with in-line transformers as compared with
Fig. 24. Compensating currents for transformer connections in a line current applications for transmission lines.
differential scheme.
48

C. 87LQ and Sensitivity to Transformer Faults

87L-1
The negative-sequence transformer differential function VY
IW IX
(87TQ) has proven to be very sensitive and capable of
detecting turn-to-turn faults [4].
The 87LQ function works naturally with in-line
IW
transformers within the implementation described above. It
87L-3
can be set sensitively because of the external fault detection
IX VY
logic.
The 87LQ function provides good sensitivity to turn-to-
turn transformer faults and high-resistance line faults. As a
IW IX
result, the combined transformer and line protection VY
87L-2
application using the presented approach can be considered
adequate.
D. Application Considerations
A separate transformer relay may still be a better option to
IW
provide other transformer functions, such as overexcitation, 87L-1
thermal, or restricted earth fault protection. For smaller- and VY IX
medium-sized transformers, these functions may not be
required or may be provided in a nonredundant fashion using IW
the second protection system. 87L-3
If the line generates considerable charging current (cables, IX VY
for example), the line capacitance may affect harmonics in the
differential current and cause some problems related to VY IW
87L-2
harmonic blocking or restraining. The charging current
IX
compensation feature remedies the problem, but nonetheless,
care should be taken when considering in-line transformer
applications with considerable line-charging current.
Using dual CT input relays capable of multiterminal
applications may provide protection for combinations of lines,
transformers, or small buses, as illustrated in Fig. 25. For
cases where two relays are located in the same substation, it is
IW IW
not necessary to equip all relays with two communications
87L-1 87L-2
channels—one of two relays in the same substation is
VY IX IX VY
equipped with a communications link toward the remote relay,
and the other relay as well as the remote relay can operate in
the slave mode in order to save one communications channel
between the two stations.

VY IX
87L-3
IW

Fig. 25. Sample applications with in-line transformers.

X. SPEED OF OPERATION
The following delays limit the operating speed of a line
current differential relay:
• Internal relay data latencies
• Algorithm delay
• Channel delay
Relay designers have no control over the last component.
Direct fiber connections add negligible delay to the scheme.
Multiplexed channels can add a delay in the range of a few
milliseconds to tens of milliseconds, depending primarily on
the number of devices passing the data between the relays and,
to a degree, on the length of the fiber path linking the
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 49

multiplexers or modems. Selection of a communications
channel in terms of the worst-case latency is one of the
engineering steps needed to fulfill the required fault clearance
time.
Internal relay data latency adds to the trip time as well.
This includes collecting and passing samples from the A/D
converter, processing the local data for transmission,
assembling outgoing packets, accepting incoming packets,
aligning the data, processing the global differential
calculations, and asserting the outputs. Modern relay
platforms optimize the above processes for speed. The
presented solution sends and receives packets every
3 milliseconds, allowing for lower relay latencies and faster
operation.
The algorithm time refers to the length of a data window
that must be available before a trip decision can be made. It
should not be mistaken for the length of the filter data
windows—it is simply a point in time when enough
information is available to make a reliable trip decision.
In this respect, the 87LQ and 87LG functions are very fast.
Even when using full-cycle filters, these functions assert their
outputs in a fraction of a cycle (see Fig. 26, for example). This
short response time is because they are not biased by the load
current prior to the fault. Owing to the proper restraining
techniques, the external fault detection logic, and the charging
current compensation, these functions can be operated at their
natural speed without introducing much intentional delay for
security purposes.
10 IARST
IADIF, IARST, A
differential protection is feasible in the proposed scheme
because the key signals are available as samples. In a
differential scheme, most of the transients cancel in the
operating signal, and therefore it is secure to make trip
decisions based on less than a full cycle of data. CT saturation
may cause problems, but this issue is addressed by fast and
reliable EFD logic. If the EFD asserts, the high-speed path is
inhibited.
As a result of careful design, the relay architecture
minimizes data latencies. The applied protection algorithms
are fast under internal fault conditions because they are
restrained in a way that does not penalize speed. Overall, the
presented solution provides for subcycle trip times if used
with fast communications channels.
XI. DATA HANDLING AND SYNCHRONIZATION
A. Introduction
Proper data handling and synchronization are the
foundation on which a line current differential relay is built.
This is as important as the protection algorithms and logic. A
careful design is required because this part of the relay relates
to the channel and associated third-party equipment that are
often only under limited control of protection engineers.
The following requirements apply to the data handling and
synchronization subsystem:
• Unified packet structure and near-identical processing
for the channel-based synchronization and the external
time reference synchronization methods.
• Minimum requirement for extra payload to

communicate sequence numbers, time stamps, and

5 IADIF other timing and data-tagging information.
• Security under and fast recovery from lost packets and
0
0.2 0.25 0.3 0.35 channel brownout conditions.
3.5
3 IQRST
• Immunity to step changes and variations in channel
IQDIF
delay.
IQDIF, IQRST, A

2.5
2 • Ability to work with channels having a total round trip
1.5 delay of 80 milliseconds or less.
1
0.5
• Accuracy of data alignment better than about 1.0
0 electrical degree (this yields a spurious differential
0.2 0.25 0.3 0.35
current of less than 1 percent of the through current).
1 • Quick startup, in the order of a few tens to few
kMAG (87LQ)

0.5 Set point:

hundreds of milliseconds, without the need for clock
Radius = 5 synchronization or similar mechanisms.
0
0.2 0.25 0.3 0.35
• In applications with external time reference, ability to
200
measure actual channel delays independently in the
transmit and receive directions.
kANG (87LQ)

150
100 Set point: The presented solution is based on estimating the clock
50 Stability angle = 140° offset and compensating the data time stamps for this offset.
0 The sent current data are time-stamped according to the time
0.2 0.25 0.3 0.35
87LQ PKP of the relay that took the data. This time is synchronized to an
87LQ
7 ms
external source (typically GPS) if such an external source is
0.2 0.25 0.3 0.35 available. Otherwise, the relay time is freewheeling, and the
time, s
time difference between any two relays may drift.
Fig. 26. Operation of the 87LQ function for a 100 Ω AG fault. The line current differential system measures the clock
In addition, the 87LP element uses a high-speed element in offset and augments the time stamps to express both the local
parallel with the full-cycle filtered path. The high-speed and the remote data in the same consistent time.
50
Under the channel-based synchronization method, the time
offset is truly measured. Because the estimated offset changes
very slowly, heavy averaging of the raw measurements is
applied, allowing the scheme to ride through a temporary
channel loss, corrupted packets, channel switching, and other
impairments.
Under the external time reference method, the clock offset
is known and equals zero.
When configured to use the channel-based synchronization
method, the system measures the clock offset and uses it even
if the two relays are synchronized to an external time
reference and the calculations return the clock offset of zero.
The method does not depend on the availability or precision of
the external time source, even if the latter is connected and
available, unless an explicit user setting mandates using the
external time reference synchronization method.
When compared to a method that forces synchronization of
the relay sampling clocks, the approach presented in this paper
is both simpler and more robust. By estimating a slowly
changing parameter (the clock offset), the scheme applies
averaging and benefits from the resulting advantages. By not
having to synchronize sampling clocks, the scheme can be
naturally extended on multiterminal applications and allow a
mixed synchronization mode in which some data are
synchronized based on symmetrical channels and some are
synchronized externally. The latter approach limits exposure
of the scheme to problems with the external time reference.
This section explains the two synchronization methods in
more detail and discusses fallback strategies for the loss of the
external time reference.
B. Channel-Based Synchronization
Refer to Fig. 27. In the presented design, Relay 1 collects
three fresh current samples for transmission, forms the packet,
and at time t0, precisely tTX after the newest sample was taken,
sends the packet out. The packet is marked with a sequence
number to identify it at the time of usage. The time t0 is
captured by Relay 1 using its own local time. An explicit time
stamping for the outgoing message can be used, or the t0 time
can be derived from the time stamp of the newest sample in
the packet and the tTX design constant.
Relay 1 Relay 2
Data
tDATA
tTX
t0
t3 – tCH – tTX in Relay 1 time = … t1
… = tDATA in Relay 2 time
tH
tDATA
tTX tTX
t2
tH, tDATA
tCH
t3
Fig. 27. Illustration of the channel-based synchronization method.
The packet arrives at Relay 2 after the unknown channel
delay time (few milliseconds to tens of milliseconds). Relay 2
captures the packet arrival time t1 using its own local clock.
This clock is asynchronous from the clock of Relay 1. Time t1
is required to measure the message hold time (turnaround
time) at Relay 2 in order to facilitate the ping-pong algorithm
for estimation of the channel delay.
Some time afterwards, Relay 2 has collected three fresh
samples of its own currents and is ready to send them to
Relay 1. Again, precisely tTX after the newest sample has been
taken, a message is launched by Relay 2 to Relay 1. The
message goes out at time t2. The hold time tH = t2 – t1 is
included in the payload of the message. This time is known in
advance because the message is started after a constant delay
tTX from the newest sample in the packet, and the current
samples are located at known points in time. Therefore, the
hold time can be precalculated at some point after capturing t1
and be conveniently put in the packet ahead of the
transmission time. Relay 2 returns the message sequence
number, letting Relay 1 know that the hold time returned to
Relay 1 was for the message that originated at t0.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 51

In its packet, Relay 2 includes a time stamp for the current
samples tDATA (assume the time stamp of the newest sample in
the set of three). In a practical implementation, the packet
sequence number and this time stamp can be the same
number.
Relay 1 receives the packet after the channel delay (few
milliseconds to tens of milliseconds). It captures the time of
reception as t3 using its own clock. From the sequence number
received, Relay 1 knows this is a reply to the message sent out
at time t0.
At this point, Relay 1 can finish the key calculations related
to channel delay, clock offset, and data alignment.
Assuming symmetrical channel delay, the one-way channel
delay is:
( t3 – t0 ) – tH
t CH = (24)
2
Note that the difference between t3 and t0 is the time
elapsed at the local relay, and the hold time is the time
measured by the remote relay and communicated back
The clock offset value is used to correct the remote time
stamp into the local time:
t DATA(@ relay1) = t DATA + t OFFSET (29)
C. External Time Reference-Based Synchronization
Refer to Fig. 28. With both relays synchronized to the same
external time source, their local times are mutually
synchronized. The relays take samples at the same points in
time (relative to the top of a second) and assign the same time
stamps to the simultaneously taken data. The data are
therefore used by the scheme directly based on the time
stamps, applying the same equations as in the channel-based
synchronization mode, except for the clock offset not
calculated, but known:
t OFFSET ≡ 0 (30)
Relay 1 Relay 2
Data

explicitly. Therefore, (24) makes sense even though its

components were derived from two asynchronously running tDATA
tTX
clocks. t0
Backdating t3 by the channel delay time, we get the tCH-TX
transmission time at Relay 2 expressed in the local time of t1 t1
Relay 1:
t 2(@ relay1) = t 3 – t CH (25) tH

tDATA
Backdating further by the known delay in transmitting a tTX tTX
t2
t2
packet after capturing the data, we obtain the data time stamp
expressed in time of Relay 1:
t DATA(@ relay1) = t 3 – t CH – t TX (26) tH, tDATA
tCH-RX
The data time stamp expressed in Relay 2 time is included
in the packet. This allows calculating of the time offset (i.e.,
the difference in time between the two relays):
t3
t OFFSET = t DATA(@ relay1) – t DATA = ...
(27)
... = t 3 – t CH – t TX – t DATA
Fig. 28. Illustration of the external time reference-based synchronization
Positive values of the offset time mean the local clock method.
(Relay 1) is leading the remote clock; negative offset means
The rest of the scheme works the same way as in the
the remote clock is ahead.
channel-based synchronization mode. In particular, the
Inserting (24) into (27) gives the following key equation:
message transmit and receive times are captured, and the hold
1
• ( t 0 + t 3 + t H ) – t TX – t DATA
time is communicated.
t OFFSET = (28)
2 These values have a different usage, however. They allow
Note that the clock offset value is a very stable number the relay to characterize the communications channel as
because it reflects a difference between clocks of the two follows.
relays, regardless of data latency and therefore regardless of The channel delay in the receive direction at Relay 1 is
the channel delay at any given moment. This number may calculated as:
change at a rate of a few parts per million, depending on the t CH – RX = t 3 – t 2 = t 3 – ( t DATA + t TX ) (31)
stability of the oscillators used in the relay hardware.
Therefore, it is both possible and recommended to average the The channel delay in the transmit direction at Relay 1 is
relay clock offset given by (28) over a number of calculated as:
measurements. This allows riding through channel t CH –TX = t 3 – t 0 – t H – t CH – RX (32)
impairments and increases accuracy by letting the rounding-up The two channel times are made available to the user for
errors and jitter average out. alarming and overall channel monitoring.
52
Also, these channel measurements can be used to control
the response of the differential system to the loss of the
external time reference, as discussed later.
D. Sources of External Time Reference
Historically, the common time reference has been provided
via an IRIG-B connection from a GPS-synchronized
substation clock. This dependency on time reference used to
create some concerns for protection applications in terms of
relying on extra equipment and exposing the differential
system to the failure modes thereof. The failure points
included the connection between the relay and the clock,
including copper-to-fiber conversion, if any, the clock itself,
its cabling and antenna, and finally the broadcast of the GPS
signal itself.
Availability of the GPS signal concerned some users,
particularly outside of North America, while the GPS
infrastructure that was originally designed for military use
migrated into civilian applications. Presently, GPS
applications are so widely spread that the long-term viability
of GPS is not questioned anymore.
Still, good protection solutions utilize as little equipment
and connections as possible in order to maximize availability
and remove unnecessary failure modes.
In this respect, it is worth noticing that terrestrial wide-area
systems emerged recently that provide for precise timing
independently from the GPS time. One solution uses the
internal precise timing of a SONET system to serve common
time at individual multiplexer nodes. Normally, this common
time is synchronized to GPS via an array of receivers placed at
different geographical locations, but if all GPS receivers are
lost or the GPS system itself becomes unavailable, the
common wide-area time continues to be generated internally
by the SONET system [1].
This enhances the availability of line current differential
schemes operating in the external time reference
synchronization mode.
Even with this improvement, we need to consider a failure
mode of losing the time reference because of problems with
connections for the timing signal (IRIG-B or IEEE 1588 via
Ethernet).
E. Fallback Strategies for Loss of External Time Reference
In the external time reference mode, the line current
differential system loses synchronization if the external time
reference is lost for an asymmetrical channel.
Misoperation is not an option, and the complete loss of
protection is not a preferred solution. Therefore, a fallback
strategy should be considered in order to provide security and
some protection functionality under such a contingency.
The following options could be considered:
• The 87L function can be left as is for a certain period
of time after losing the common time reference. It will
take some time for the internal relay clocks to drift
apart, and the system will stay synchronized at least
for few seconds. During that time, the clock offset
remains zero even if one of the relays does not
synchronize to the common time. If the external time
source does not recover after some time, the danger of
the clocks drifting apart increases, and the scheme
needs to take other steps, as explained below.
• In applications with only some channels being
asymmetrical, the line current differential system can
mark the asymmetrical channels that lost a time
reference at either end as unavailable. With enough
remaining connections between the relays, a master-
slave operation may be possible, retaining the
functionality of the system at the expense of slightly
delayed tripping at the slave sites.
• The 87L function can disable itself automatically if
any of the required time references are lost.
• The 87L function can desensitize itself to a degree if
any of the required time references are lost.
• If the used channel was symmetrical just prior to
losing the external time reference, the line current
differential system may switch to the channel-based
synchronization mode. This mode continues
indefinitely, assuming the channel stays symmetrical
until the time reference recovers. Or this mode can
stay in place until the total (round trip) channel time
changes, signifying the fact that the channel has been
switched and may become asymmetrical as a result of
the switching.
• Similarly, if the channel displayed a stable limited
asymmetry prior to losing the time reference, this
asymmetry can be used to predict the worst-case
measurement errors and put in place appropriate
protection countermeasures (stability angle setting for
the Alpha Plane). The channel may switch into
channel-based synchronization, applying enough
protection countermeasures to fight the possible
spurious differential current. This situation may
continue indefinitely or until the total channel time
changes, signifying channel switching and possible
increase in asymmetry.
• Similarly to the above option, the worst-case channel
asymmetry can be tracked and stored in the relay. This
worst-case asymmetry can be used to calculate secure
87L settings, and the system can switch to channel-
based synchronization upon losing the external time
reference. The initial value of the worst-case
asymmetry is a user estimate, but during the life of the
installation, channels with common time references at
both ends can be characterized by the relays for the
worst-case asymmetry.
XII. CONCLUSIONS
This paper outlines general design directions for a next
generation line current differential protection scheme.
Regarding synchronization and data alignment, the
presented solution is unified for the channel-based and the
external time reference-based synchronization methods. By
not forcing the sampling clocks to sample synchronously, the
presented solution works naturally and is more robust in
multiterminal applications.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
A terrestrial, GPS-independent time source was introduced,
making line current differential applications with
asymmetrical channels safer and more dependable.
A number of fallback strategies were presented to allow the
system to ride through the temporary loss of the timing source.
A reliable Alpha Plane restraining technique was proposed
for multiterminal applications with any number of local
current inputs to the relays. The size and structure of the
communications payload are independent from the number of
terminals or the number of local currents in the system,
making the implementation simple and thus the solution more
robust.
The system works with high-fidelity current information,
allowing a fast external fault detection logic as well as
harmonic measurements for in-line transformer applications.
The combination of the high-performance external fault
detection logic capable of detecting external faults based on as
little as 3 milliseconds of unsaturated current waveforms and
the proper restraining techniques makes the relay both very
secure and fast. As in the case of modern bus or transformer
relays, application of the external fault detection relaxes CT
requirements considerably and simplifies application by
eliminating a cumbersome analysis of suitability of the
applied CTs and settings.
The presented solution applies the Alpha Plane differential
trip equations, carrying forward all tried-and-true advantages
of this approach, but enhances the original concept to
multiterminal applications and allows for harmonic restraining
of the Alpha Plane to facilitate in-line transformer protection.
Line-charging current compensation was incorporated,
enhancing security during line energization and improving
sensitivity to internal faults. A novel concept of dynamic
selection of voltage sources for the compensation was
introduced, minimizing dependence of the scheme on VTs.
Communications packets and data handling internal to the
relays were designed for low latencies. Combined with high-
speed performance of the applied protection equations, this
enables subcycle trip times in applications with fast
communications channels.
XIII. REFERENCES
[1] K. Fodero, C. Huntley, and D. Whitehead, “Wide-Area Time
Synchronization for Protection,” proceedings of the 36th Annual
Western Protective Relay Conference, Spokane, WA, October 2009.
[2] J. Roberts, D. Tziouvaras, G. Benmouyal, and H. Altuve, “The Effect of
Multiprinciple Line Protection on Dependability and Security,”
proceedings of the 55th Annual Georgia Tech Protective Relaying
Conference, Atlanta, GA, May 2001.
[3] A. Guzmán, C. Labuschagne, and B.-L. Qin, “Reliable Busbar and
Breaker Failure Protection With Advanced Zone Selection,”
proceedings of the 31st Annual Western Protective Relay Conference,
Spokane, WA, October 2004.
[4] A. Guzmán, N. Fischer, and C. Labuschagne, “Improvements in
Transformer Protection and Control,” proceedings of the 62nd Annual
Conference for Protective Relay Engineers, College Station, TX, March
2009.
[5] G. Benmouyal, “The Trajectories of Line Current Differential Faults in
the Alpha Plane,” proceedings of the 32nd Annual Western Protective
Relay Conference, Spokane, WA, October 2005.
53
[6] K. Behrendt, N. Fischer, and C. Labuschagne, “Considerations for Using
Harmonic Blocking and Harmonic Restraint Techniques on Transformer
Differential Relays,” proceedings of the 33rd Annual Western Protective
Relay Conference, Spokane, WA, October 2006.
XIV. BIOGRAPHIES
Henry (Hank) Miller has a BSEE and a BEE from The Ohio State University
and a BA degree in philosophy from The Pontifical College Josephinum. He
is a registered professional engineer in the state of Ohio and holds a patent for
a substation steel design. Hank has over 31 years of utility experience in
station and line protection and control. Hank is currently working as a staff
engineer and supervisor in the protection control asset engineering group of
American Electric Power (AEP) with responsibilities for developing
protection and control standards and application guides and supporting the
relay setting project work. With John Burger, he shares the responsibility for
ensuring that devices new to the AEP system are protected and controlled
properly. Hank is a member of the IEEE.
John Burger has a BSEE from Case Institute of Technology and an MSEE
from Fairleigh Dickinson University. He is a registered professional engineer
in the states of Ohio and New Jersey. John has over 35 years experience in
station and line relay protection and control. He has worked for American
Electric Power (AEP), primarily in the protection and control group, for the
last 29 years. John is currently serving as a staff engineer and supervisor in
the protection control asset engineering group, with responsibilities for
developing protection and control standards and application guides and
supporting the relay setting project work.
He shares the responsibility for ensuring that devices new to the AEP system
are protected and controlled properly with Hank Miller. John is a Senior
Member of the IEEE, past chairman of the Columbus Chapter of the PES, a
member of the IEEE Power System Relay Main Committee, Substation and
Communications Subcommittees and chairman of Working Group H6. He is
also currently serving as Chairman of the UCA International Users Group,
providing technical support for IEC 61850.
Normann Fischer received a Higher Diploma in Technology, with honors,
from Witwatersrand Technikon, Johannesburg in 1988, a BSEE, with honors,
from the University of Cape Town in 1993, and an MSEE from the University
of Idaho in 2005. He joined Eskom as a protection technician in 1984 and was
a senior design engineer in Eskom’s protection design department for three
years. He then joined IST Energy as a senior design engineer in 1996. In
1999, he joined Schweitzer Engineering Laboratories, Inc. as a power
engineer in the research and development division. Normann was a registered
professional engineer in South Africa and a member of the South Africa
Institute of Electrical Engineers. He is currently a member of IEEE and
ASEE.
Bogdan Kasztenny is a principal systems engineer in the research and
development division of Schweitzer Engineering Laboratories, Inc. He has 20
years of experience in protection and control, including his ten-year academic
career at Wroclaw University of Technology, Poland, Southern Illinois
University, and Texas A&M University. He also has ten years of industrial
experience with General Electric, where he developed, promoted, and
supported many protection and control products.
Bogdan is an IEEE Fellow, Senior Fulbright Fellow, Canadian member of
CIGRE Study Committee B5, and an Adjunct Professor at the University of
Western Ontario. He has authored about 200 technical papers and holds 16
patents. He is active in the Power System Relaying Committee of the IEEE
and is a registered professional engineer in the province of Ontario.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Smarter Time Sync: Applying the IEEE PC37.238 Standard to
Power System Applications
Christoph Brunner
it4power
christoph.brunner@it4power.com
Keywords: IEEE 1588, precise time synchronization,
IEC 61850, process bus, synchrophasors.
Abstract
Synchronization plays an important role for many functions
of the electric power systems. The PC37.238 draft standard
specifies precise time distribution for power system
applications. As the standard is nearing completion, its usage
for specific applications needs to be better understood. This
paper explains how this new time distribution applies to the
most demanding applications, namely synchrophasors and
IEC 61850 process bus.
1 Introduction
Synchronization requirements for power system applications
differ significantly. Some applications require relative
synchronization (synchronization of merging units), others
need traceability to universal time coordinated (UTC) either
locally or over a wide area (event recorders, synchrophasors).
Synchronization of sampled analogue values and
measurements used for protection is critical and demands
high security and dependability.
1.1 Timing requirements for IEC 61850 process bus
IEC 61850 introduced the concept known as “process bus” to
substation automation systems. Although not necessarily a
separate communication bus, the concept introduces the
possibility of connecting process equipment, such as
switchgear and sensors, directly via a digital interface to the
rest of the system.
Such approach offers many benefits to the user, for example
simplified engineering and commissioning and a reduction of
copper wiring within a substation. As an overall benefit, life
cycle costs are expected to be reduced. The approach however
presents some challenges, which have so far prevented a wide
scale implementation of this process bus.
An important aspect while using sampled values of a power
system is the phase relationship between the different
measured signals, in particular between current and voltage.
For some applications the accuracy in the time domain needs
to be in the range of 1 microsecond. That means, when
samples are acquired by different devices, the relation
between the time bases must be with an inaccuracy of less
than 1 microsecond.
55
Galina S. Antonova
ABB Inc.
galina.s.antonova@ca.abb.com
There are basically two solutions for this problem. The first is
the approach of a constant acquisition delay. In that case, the
delay from the acquisition of the sampled value (the moment,
where the value is sampled at its source) until the reception in
the data sink (e.g. the protection equipment) needs to be
constant and known with a jitter below the required accuracy.
The receiving unit can then put the values in a time relation to
each other based on the reception time and the known delay.
The second approach is to use synchronized or time related
sampling. All units performing sampling are globally
synchronized with the required accuracy. The samples are
either taken all at the same time or the samples are tagged
with a timestamp indicating their sampling time. Only this
approach can deal with variable communication delays that
are inevitable when using a network topology for the
communication. IEC 61850 has therefore chosen this second
approach. According to the concepts of IEC 61850, the
samples are all taken at the same time and each sample is
identified by a number that provides the time reference. The
approach is illustrated in Figure 1.
n-1 n
n-4 n-3 n-2 n-3 n-2
n-2 n-1 n
n-4
n-4 n-3
n-1 n
Subscriber
Publisher
Figure 1. The approach to transmit sampled values
The challenge here is to synchronize the devices performing
the sampling with an accuracy in the range of 1 microsecond.
IEC 61850-9-2 that defines the mapping for the transmission
of the sampled values may use an external signal to
synchronize the devices. In today's prototype
implementations [7], a one pulse per second (1 PPS) signal is
used to synchronize the devices. This requires in addition to
the communication network a physically separate
synchronization network. Such an additional synchronization
network is of course a source of failure. However, not all
functions depend on these high synchronization requirements.
Mission critical functions do not need information from many
56
different acquisition points. With elaborated system
architectures the dependency of the critical functions on the
synchronization network can be removed. The drawback of
such approach is that this limits the benefits from a process
bus implementation.
1.2 Timing requirements for synchrophasors
Phasors have been commonly used for representations of
periodic signals in computations and power system analysis.
The use of synchrophasors was first specified in
IEEE 1344-1995 Standard for Synchrophasors for Power
Systems, which after a revision became IEEE C37.118-2005.
This standard is currently being revised and harmonized with
IEC 61850.
Synchrophasor, as defined in [6], is the estimate of magnitude
and instantaneous phase angle of a signal relative to the
cosine function at nominal system frequency synchronized to
UTC time. UTC second rollovers coincide with rising edges
of 1 PPS signal. These are used for estimations of 0 degrees
and -90 degrees synchrophasors. Synchrophasor definition is
illustrated on Figure 2.
v(t) = 2 V cos (ω0 t + ϕ)
v
Start of the second
2V
t distribution mechanism is provided in Section 2. Section 3
V ∠ 0°
v
Start of the second
2V
t
Figure 2. Synchrophasor definition
Phasor measurement units (PMUs) estimate synchrophasor
values and transmit them in a defined data frame serially or
over an Ethernet network. If desired, this data can be merged
with data from other PMUs using Phasor Data
Concentrator(s) (PDCs).
Synchrophasors provide real-time information on power
system currents and voltages, and can be used for various
applications, including local or wide area monitoring,
synchrocheck and protection functions. A major statement on
the usefulness of synchrophasors is the 2003 blackout
synchrophasor data collected over Eastern US area that
clearly indicated faulty conditions, which, if timely analyzed,
could have prevented this major outage.
Synchrophasor measurements are normally relative to UTC
time, and their precision greatly depends on time
synchronization accuracy. Note that these measurements can
be collected over small or very large geographical area, and
measurements from remote devices need to be correlated for
system analysis and protection decisions. Time sources and
time distribution must be reliable, available and accessible by
PMU devices in the whole system.
Accuracy limits for the synchrophasor measurements, as
defined in [6] shall not exceed 1% Total Vector Error (TVE).
This translates to a maximum time error of
± 26 microseconds for 60 Hz system and
± 31 microseconds for 50 Hz system. To provide
synchrophasor measurements with specified measurement
accuracy, much tighter requirements are placed on system
time synchronization. It has been generally agreed that
± 1 microsecond time accuracy of UTC is desired at device
inputs to achieve the specified measurement accuracy of
1% TVE.
Both IEC 61850 process bus and IEEE C37.118
synchrophasors require reliable and available time
synchronization with down to ± 1 microsecond time
accuracy. Both currently use separate synchronization
networks and 1 PPS signals.
This paper describes how time synchronization requirements
for IEC 61850 process bus and IEEE C37.118
synchrophasors, outlined in this section, can be met using the
PC37.238 draft standard. A basic description of the new time
describes the use of PC37.238 draft standard for
synchrophasor and IEC 61850 process bus applications,
including known challenges and issues.
2 IEEE 1588 and Power Profile
IEEE Standard for a Precision Clock Synchronization
Protocol for Networked Measurement and Control System
was initially released in 2002 and revised in 2008 [1]. The
standard describes a protocol for distributing time with sub-
microsecond time accuracy over various communication
technologies, such as Ethernet, UDP/IP, DeviceNet, etc. The
key advantages of this standard are that timing can be
distributed over the same network as application data, and
time accuracies generally not achievable by other time
distribution protocols become possible.
2.1 IEEE 1588 Profiles
The second version of the IEEE 1588 Standard specifies
many new features of the Precision Time Protocol (PTP),
mandatory and optional, including conceptually new devices
(transparent clocks), new message formats, etc. Development
of this version was driven by five main industries: test and
measurement, telecom, industrial automation, power and
military. As a result, it was impossible to specify a single set
of interoperable functions, as requirements for these
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
industries differ significantly. To address this challenge a
concept of IEEE 1588 profile was introduced, as a clearly
defined subset of protocol features, the use of which will meet
specific industry requirements. Two default profiles are
defined in the Annex J of the second version of the IEEE
1588 Standard [4]. Industries are expected to define their own
profiles to address specific requirements of their applications.
IEEE 1588 profile development work was launched for
different industries. Among the most developed profiles are
the LAN eXtensions for Instrumentation (LXI) profile for test
and measurements [12], the profile for time-sensitive
application in bridged local area networks, specified in
P802.1AS/D7.2 [15], and the power profile, specified in
PC37.238/D5.5 [5]. Other profiles are being developed for
telecom and Internet communities under ITU-T and IETF
respectfully [8, 16]. Four paragraphs below provide more
information about IEEE 1588 profiles, and clause 2.2
describes IEEE 1588 power profile in more detail.
The LXI profile was developed to address the needs of test
and measurement community (these applications may require
nanosecond and sub-nanosecond levels of time accuracy). It
is interesting to note that this profile uses IEEE 1588
management messages for configuration, status and control.
The LXI consortium also developed comprehensive
documented test procedures, a big effort that should be
acknowledged.
IEEE 1588 profile for time-sensitive applications in bridged
Local Area Networks is being developed by the Audio/Video
Bridging Task force of the IEEE 802.1, the main standards
body of the IEEE 802.3/Ethernet standards. The initial
applications included time synchronization for home theatre
and audio / video entertainment systems. The latest version of
this profile is specified in the P802.1AS / D7.2 Draft
Standard, which passed the initial IEEE ballot and is
undergoing comment resolution before the final approval. The
profile defines a Simple Network Management Protocol
(SNMP) Management Information Database (MIB) for
configuration, status and control.
IEEE 1588 profile for telecom industry is being developed
under ITU-T Study Group 15. An approach with separate
frequency profile and phase profile has been chosen, where
the frequency profile is being defined first. Results of these
developments, including use cases, performance metrics
analysis, etc can be found in ITU-T G.8261 [8].
The Internet community has also launched IEEE 1588 profile
development in TICTOC group, where various PTP protocol
mappings, including mapping into Multiprotocol Label
Switching (MPLS) have been discussed [16]
2.2 IEEE 1588 Power Profile
The IEEE Power System Relaying Committee (PSRC) and
Substation Committee (SUB), both operating under the IEEE
Power and Energy Society (PES) formed a joined Working
57
Group to develop IEEE 1588 power profile. The PSRC
H7/Sub C7 Working Group functions in close co-ordination
with IEC TC57 WG10 and other committees with similar
interests. The group develops PC37.238 Draft Standard for
Use of IEEE Std. 1588 in Power System Applications [5].
The PC37.238/D5.5 is completed and released for IEEE
Sponsor ballot, which will close on November 21, 2010.
Comment resolution and recirculation ballot will follow prior
to the final approval. During standard’s development profile
interoperability tests, called plug-fests, were conducted [3].
The profile as well has been evaluated for the IEEE 1588
Testbed project [9].
PSRC WG H7/Sub C7 plans to write a summary paper on the
PC37.238 standard that will describe in detail the selected
IEEE 1588 features and their justification. The paragraphs
below provide a summary of the supported features plus why
and how the PC37.238 draft standard can be used for power
system applications, namely synchrophasors and IEC 61850
process bus.
Main Function
PC37.238 specifies how to distribute time from a device that
is synchronized to universal time (e.g. to UTC via GPS) to
devices that require time to support their application functions
(e.g. IEDs with line differential protection, phasor
measurement units, merging units, digital event recorders).
Devices
A device that is synchronized to the universal time is called a
grandmaster clock. Devices that can only receive time from
grandmaster clock are called slaves or slave-only clocks.
Network devices that interconnect grandmaster and slave
clocks are called transparent clocks, these are usually
Ethernet switches.
Performance requirements
Normative performance requirements are specified in
PC37.238 Annex B. As depicted in Figure 3 time at a slave
device connected to grandmaster clock over 16 network hops
shall be within +/-1 microsecond from grandmaster time, with
an out-of-range probability of 10-4. Note that all devices,
including intermediate nodes have to support PC37.238.
GPS
800 ns
200 ns
Hop 1 Hop 2 Hop 3 ... Hop 15 Hop 16
Time
Switch Switch Switch IED
Source
50 ns
± 1 µ s time accuracy with 10 −4 probability of failure
Figure 3. PC37.238 performance requirements
58
Operation
Time is distributed using Sync messages that contain the time
at the grandmaster at the instant a given Sync message was
issued. A best grandmaster selection is performed using
Announce messages. Path delay measurements are preformed
using PDelay messages.
Mapping
All PC37.238 messages are mapped directly into
Layer 2 / Ethernet frames and are transmitted over Ethernet
cables (fibre or copper). Multicast Destination MAC address
and IEEE 802.1Q VLAN tags are used for per-VLAN traffic
separation, if desired.
Best grandmaster selection and redundancy
For redundancy there should be two or three devices capable
of being grandmasters. These devices are called grandmaster-
capable clocks. On start-up, failure or planned maintenance
outage, an automatic selection of the best master is performed
that results in one device being a system’s grandmaster.
Selection of the grandmaster is based on device
qualifications, including achievable clock accuracy.
Grandmaster identification
A feature unique to PC37.238 is the addition of an extension
to communicate grandmaster identification. This information
helps to minimize the effects of transient states, when
distribute time is not traceable, on synchronization and
operation. It communicates to applications if data they require
used the time from the same grandmaster.
Local Time
Local time extension is also provided, as some applications
may require this, e.g. for reporting (not operation) functions.
Management
A PC37.238 SNMP MIB is defined for device configuration
and status information. MIB support is mandatory only for
grandmaster-capable devices.
“Smart” features
Time accuracy in sub-microsecond range can be achieved
because of the following “smart” features offered by
IEEE 1588 and supported by PC37.238 draft standard:
(1) Correction for variable delays in the protocol stack
(2) Correction for cable delay
(3) Correction for residence time in intermediate devices
All three above remove sources of non-deterministic latency,
making more precise time synchronization possible. Figure 4
illustrates these features.
Switch with Transparent clock
Switch Fabric
Grandmaster clock Slave clock
Store and Forward
Queues
PTP PTP
MAC MAC MAC MAC
MII
Cable delay Residence time Cable delay
MII
PHY PHY PHY PHY
Timestamp Pont
Figure 4. Compensation for variable delay sources
Variable delay introduced by communication protocol stack
(1) is avoided by timestamping Sync messages just before
they are transmitted into physical media. Timestamp point is
between Media Access Control (MAC) block and Physical
interface, typically called PHY at the Media Independent
Interface (MII), in case of 100Mb/s Ethernet. Some PC37.238
devices can support “on-the-fly” timestamping, i.e. inserting
the precise actual grandmaster time while Sync message is
being transmitted over MII interface. Other devices may
support two-step operation mode with a Follow_up message
transmitted right after Sync message. Follow_up message
contains the precise grandmaster time when its associated
Sync messages was transmitted over grandmaster’s MII
interface.
Two other variable delay sources are dealt with using peer-to-
peer transparent clock functionality. These devices are
typically Ethernet switches that measure cable delay and
residence time then provide these in a special field, called
CorrectionField, to the end devices, so that they can add these
delays to the received grandmaster time before adjusting their
clocks.
Cable delay (2) is measured using peer-to-peer delay
measurement mechanism. It uses PDelay message exchange
to measure the delay on each point-to-point link. Point-to-
point delay measurements enable faster convergence times
upon grandmaster changes, as cable delays are measured for
all links, including those blocked by Spanning Tree Protocols,
regardless of whether the link is currently on time distribution
path or not. Cable delay correction is very important as each
100m copper link introduces approximately 500ns delay,
which is half of the full time accuracy budget of
1 microsecond.
Transparent clocks, which are Ethernet switches supporting
PC37.238, also measure residence time, the time that Sync
message carrying grandmaster time actually spends inside the
switch. This time is variable and depends on traffic loads that
lead to varying store-and-forward queuing delays.
Transparent clock and residence time concepts are shown in
Figure 4.
The above features make the IEEE 1588 power profile
specified in PC37.238 draft standard different from other
IEEE 1588 profiles. Comparison between PC37.238 and the
P802.1AS, a profile for time-sensitive application for bridged
local area networks is given in [10].
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 59

3 The use of PC37.238 for power applications
The PC37.238 draft standard was developed with
requirements of power applications in mind. Time
synchronization can now be performed using the same
communication network that provides data exchange for the
applications. This eliminates the need for a physically
separate synchronization network. This new time distribution
mechanism also delivers the time quality and traceability to
universal time information required by core applications. This
information is made available via management parameters,
and can be mapped directly into parameters, specific to a
given application. The mapping of these parameters is
specified in PC37.238/D5.5 Annex C [5]. Finally, local time
and grandmaster identification are provided. The later was
added specifically to minimize the effect of transient states
when distributed time is not traceable to absolute time on
synchronization and operation, as well as to support system
separation into operation islands. Two sections below
describe the use of PC37.238 draft standard for IEC 61850
process bus and IEEE C37.118 applications.
3.1 IEC 61850 process bus
connected to the IEC 61850 communication or in the case of
intelligent switchgear, the electronics may be directly
included in the switchgear. IEC 61850 GOOSE messages can
be used to transmit a trip signal from the protection relay to
the circuit breaker.
With IEC 61850 sampled value transmission, it is possible to
replace the analog signals from CTs (current transformers)
and VTs (voltage transformers) with serial communication.
Instrument transformers based on new technologies like
optical sensors can directly be equipped with a
communication interface according to IEC 61850. However,
also conventional instrument transformers may be connected
to a communication network using a “merging unit” as a
communication interface. A merging unit can transmit the
sampled values of the three voltages and the three currents in
one communication message. The concept of a merging unit
is shown in Figure 6.
Line
Protection
IEC 61850-9-2

The standard IEC 61850 "Communication networks and Logical Device

“Merging Unit”
systems in substations" has been published by IEC between
2003 and 2005. The purpose of IEC 61850 is, to provide all LNPCTR
LN PCTR
LN TVTR
the necessary specifications required to achieve LNPCTR
PCTR
LN
LN TCTR
interoperability between the equipment of an integrated
substation automation system.

Figure 5 shows an example of a substation automation system

(SAS) with a process bus connection. With a traditional
approach, copper wires connect the process equipment with
Figure 6. Concept of a merging unit
the bay level devices. These copper wires are used, e.g. to
connect the trip output from the protection equipment with
Note that despite of the term “process bus”, it does not
the trip coil of the circuit breaker, the position indications of
necessarily imply a separate communication network. It is
the different switches with the bay controllers or the analogue
perfectly possible to combine the communication between
signals from the current and voltage transformers with the
station level and bay level devices (typically called “station
protection relay.
bus”) with the process bus communication as this is shown in
the example of Figure 7.
Control Center HMI, Station
controller

Router Control Center HMI Engineering

Station bus
Router
Bay Relay Relay Bay Relay Relay
Controller X1 X2 Controller X1 X2

Process bus

Intelligent Intelligent
CT/VT CT/VT
Switchgear Switchgear
Bay Relay RelayRelay Bay Relay Relay
Relay
Controller X1 X2 Y Controller X1 X2 Y

Figure 5. SAS architecture with a process bus connection

Switchgear with CT/VT Switchgear with CT/VT
Switchgear with CT/VT Switchgear with CT/VT
Breaker IED Breaker IED
IEC 61850 supports the possibility, that these connections can IED withwith
IED IED IED with
with IED
IED
be realised using a communication network. Switchgear may
be connected with distributed remote I/O units that are Figure 7. Possible architecture with process bus connection
60

A communication-based process connection provides –
besides the reduction of copper wires – some additional
advantages. The wiring between the IEDs interfacing the
process (merging units, remote I/Os) and the process may be
done and tested during manufacturing. In other words, instead
of handling many wired connections during engineering,
commissioning and testing, only a few communication
connections could be tested at a factory.
Different topologies for process bus are evaluated in [11].
Two examples of a topology are shown in Figure 8 and
Figure 9. The topology of the communication network
follows typically the topology of the substation. That means
that devices that belong to one bay typically are grouped
together.

Further, with the serial communication defined in IEC 61850 BP

a stream of digitized samples replaces the analogue signal.
The representation of the values is independent from any
specific properties of the instrument transformer. The values
are transmitted with the data type integer or real. In the case
of the real representation, the actual process value from the P P P P P P
power system without any scaling or rating is transmitted. In
C C C
the case of the integer representation, the value is scaled; the
scaling parameters are included in the data model so that any MU MU MU MU MU MU
Bay 1 Bay 2 Bay 3
receiver can calculate the actual process value. As a
consequence, the bay level equipment does not need to handle
issues like the rated values and the ratio of the transformers. Figure 8. Topology with a non-redundant star network

Challenges for implementing an IEC 61850 process bus are
covered in [2]. These include the real time behaviour,
reliability, transient behaviour of a merging unit and time
synchronization. The time synchronization issue is described
in [2] as follows.
- Time Synchronization: One of the challenging topics is
the accurate synchronization of the data sources
providing the sampled values. The communication
network used to transmit the sampled values does not
support a constant or predictable transmission delay.
Therefore, the mechanism chosen assumes a
synchronized sampling at the source. All devices sample
at the same time and add a reference to the sampling
time. With the help of that reference, the application
using the sampled data can correlate the samples received
from multiple sources. Depending on the application and
the requirements concerning acceptable phase errors, the
required synchronization accuracy may be as accurate as
1 microsecond.
In Figure 8, a topology without redundant network on station
bus level is shown. In order to have the main 1 and main 2
protection functions independent, there is a main 1 system
shown in blue consisting of a merging unit (MU), a protection
relay (P) and a switch, and an independent main 2 system
shown in red with its own protection relay, merging unit and
switch. The bay controller (C) is connected to one of the two
switches. The station level connection is non-redundant star
network.
Figure 9 is a topology that includes a redundant
communication network for the station bus. While the
protection system with main 1 and main 2, consisting of the
merging unit and the relay, is redundant as such, the bay
controller is now connected to two independent ring
networks.
BP

Once the profile for time synchronization is completed, most

of these challenges will be resolved. What remains is the
verification of the transient behaviour of a merging unit. This
is discussed within IEC TC38 WG37, which is developing the P C P P C P P C P
IEC 61869-9.
MU Bay 1 MU MU Bay 2 MU MU Bay 3 MU
When using a process bus connection supported with time
synchronization over the network, more flexibility with
regard to the allocation of functions into devices and the Figure 9. Topology with a redundant ring network
network topology will be available that fulfil the different
requirements of the application. Requirements of smaller An important aspect to be considered in a substation
substations with a single protection system will be different to automation system is as well the degradation of the system in
the requirements of a larger substation with a main 1 and case of failures. Not all of the functions are mission critical –
main 2 protection systems. some however must always be available as well in a degraded
system. This aspect needs to be considered when introducing

time synchronization. Figure 10 is an example that illustrates
the different requirements for time synchronization.
Feeder 1 Feeder 2 Feeder 3
zone1 zone1 zone1
MU1b MU2b MU3b
zone3
C1 P1 C2 P2 C3 P3
MU1a MU2a MU3a
zone2
Busbar
PB
Figure 10. Synchronization requirements
- The line protection function P1 needs information from
MU1a. The only synchronization requirement that exists
here is internal to the IED MU1a – the samples from the
three currents and three voltages need to be synchronized
to each other.
- The control function C1 needs information from MU1a
and MU1b to perform the synchrocheck functionality and
the two merging units need to be synchronized. This is
indicated as synchronization Zone 1.
- For the example, it is assumed that Feeder 3 and Feeder 4
are connected to two power lines that run in parallel. In
that case, the protection IEDs P3 and P4 implement an
additional function called parallel line compensation.
That function needs information from both merging units
MU3a and MU4a. These merging units need to be
synchronized which is indicated as Zone 2.
- The differential busbar protection function PB needs
information from all four merging units MU1b to MU4b.
These merging units need to be synchronized which is
indicated as Zone 3.
These specific requirements were addressed by the developers
of the PC37.238 draft standard. The proposed mechanism
provides time distribution with time accuracy in the range of
+/- 1 microsecond. The same communication network
(Ethernet) now can be used for data exchange and time
synchronization. Redundancy of time source and the ability to
form operational islands upon grandmaster change, fault or
maintenance have also been considered.
One of the interesting subjects to be verified is transient
behaviour of the Merging units, in particular with respect to
sample synchronization. This topic requires more analysis,
and recommendations. As the transient behaviour of merging
units is in the scope of IEC TC37 WG37 group, it would be
beneficial to discuss it in that community. Possible ideas
evolve around frequency synchronization by slewing, re-
establishing a common time for sample number 0, etc.
PC37.238 draft standard also specifies a mechanism to
distribute grandmaster identification along with the time
quality information. If grandmaster time is locked to UTC
with a given accuracy, grandmaster identification adds little
X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 61
value. During transient states while distributed time is not
traceable to the absolute time, the knowledge of which
grandmaster a merging unit is synchronized to can help to
Feeder 4
minimize the effect of such change on synchronization and
zone1
operation. For example, decisions that are based on sampled
MU4b
values coming from different merging units, if these are
C4 P4 synchronized to different grandmasters not traceable to
absolute time, could be avoided or delayed until
synchronization to the same grandmaster or a traceable time
MU4a is re-established.
PC37.238 draft standard provides mechanisms for mapping
its local time accuracy and traceability to the global time
source into IEC 61850 parameters, such as timestamps for all
IED data, SmpSync attribute for global / local time synch, etc.
This is described in PC37.238 / D5.5 Annex C [5].
3.2 IEEE C37.238 Synchrophasors
Synchrophasor measurements provide instantaneous
magnitude and angle information relative to UTC for power
system currents and voltages. Measurements for zero
sequence, positive and negative sequence values can be
supported, if desired. Synchrophasor data is assembled by
PMU devices into a defined frame format, and transmitted to
control centre possible via PDC(s) for further forwarding,
archiving or control actions.
An overview of a phasor measurement system is shown in
Figure 11. It shows multiple PMUs and PDCs, and a control
centre that receives synchrophasor data and uses it for real
time monitoring, control and archiving.
GPS Satellite Time Synchronization
Power System
PMU PMU
Streaming synchrophasor
data on the network to the
PDC for archiving . . .
PMU
PMU PMU
ETHERNET
APPLICATIONS
. . . data display
PDC/Server PDC/Server and real time
control actions
Figure 11. Overview of phasor measurement system
Since time synchronization to absolute time, UTC, plays such
an important role, it is discussed in more detail below.
Time synchronization via Global Positioning System (GPS)
as depicted, is a very common mechanism. There has been
much discussion on the reliability of GPS satellites and the
need to improve it [13].
62
Another time distribution mechanism typically used in power
substations is IRIG-B and 1 PPS signals. These require
separate wiring for synchronization purposes only, which
could be avoided if time is distributed on the same network
with synchrophasor data.
Figure 12 shows a typical PMU with embedded GPS receiver,
like those depicted in Figure 13. This device has an Ethernet
link for transmitting synchrophasor data (but not for time
synchronization). A serial interface may also be used, but its
use is deprecated, and considered a legacy. The device can
also have an IRIG-B input for synchronizing to an external
source.
GPS Antenna
Binary I/O
• Alarms to SCADA
• Local Control
Ethernet Link
Analog Inputs
• 2 – 3Φ Voltage
• 4 – 3Φ Currents
IRIG-B
• Data Concentrator • Input
• Remote access • Time Sync
Figure 12. Typical standalone PMU device
Let’s now consider time distribution over Ethernet link, as
specified in the PC37.238 draft standard, to explain what
value it will bring for the PMU devices.
There has been a lot of confusion in the synchrophasor
community regarding the PC37.238 functionality. A common
understanding is essential for the acceptance and adoption of
the new time distribution mechanism. The authors would like
to clarify the PC37.238 functionality and attempt to explain it
in terms familiar to the PMU experts to bridge any
interpretation gaps there might be.
First, it should be explained that the PC37.238 draft standard
only specifies time distribution mechanism. It is similar to
IRIG-B, which provides time codes, GPS and Network Time
Protocol (NTP) in this respect.
IRIG-B time codes are functionally similar to the PC37.238
timestamps. The term “timestamp” should be explained
separately as it seems to cause a lot of confusion.
In test and measurement world, and for PMU devices, the
timestamp contains time of the actual measurement that is
taken at PMU’s analogue input. This is well understood, as
well as the fact that this time has nothing in common with the
time of the data transmission for this measurement (except
that data transmission is delayed from time of the
measurement by the measurement latency).
For the PC37.238 time distribution, timestamp as well is the
time of an event (the measurement for the PMU case), but the
event is the message transmission. Thus, timestamp of the
PC37.238 Sync message is the time when this message was
sent by the grandmaster, i.e. grandmaster time at that
moment. In PC37.238 world timestamps contain the actual
absolute time, which is conceptually different, but similar to
the test and measurement approach, if one considers
transmission of a synchronization message as an
event/measurement point.
Further clarifications should be given to other function
potentially expected from the PC37.238 draft standard, such
as how frequently (multiple times in a given power cycle) and
when synchrophasor measurements shall be taken, etc. The
PC37.238 draft standard does not provide any specification
for these; neither do IRIG-B and GPS. The scope of these
protocols is limited to time distribution function.
One more remark should be made in respect to PC37.238
functionality and 1 PPS signal. This signal may be provided
for testing purposes, but vendors are not required to support
it. It is possible, however, to generate such signal internally
based on precise time provided by the PC37.238 protocol
using the UTC second rollover. Vendors adding PC37.238
support to their PMU devices should be aware of this.
Once a common understanding on functionality is reached per
the initial comments above, technical details on the usage of
PC37.238 for the PMUs are discussed below.
PC37.238 specifies a time distribution mechanism over
Ethernet links that can deliver time with ± 1 microsecond
time accuracy to the inputs of end devices. This means that
the same Ethernet network can be used for synchrophasor
data and time distribution. As the result, IRIG-B wiring /
interfaces become unnecessary, as shown in Figure 13.
GPS Antenna
Binary I/O
• Alarms to SCADA
• Local Control
Ethernet Link
Analog Inputs
• Data Concentrator • 2 – 3Φ Voltage
• 4 – 3Φ Currents
• Remote access
IRIG-B
• Time Sync • Input
• Time Sync
Figure 13. Replacement of IRIG-B interface
As well, the need for a GPS receiver in each PMU unit can be
eliminated, thus reducing the cost, implementation and
commissioning efforts, as illustrated in Figure 14.

It should be noted that for achieving a time accuracy in the
range of 1 microsecond, all devices in the network
infrastructure that interconnect time source(s) and PMUs have
to support the PC37.238 draft standard. This means that
Ethernet switches must be peer-to-peer transparent clocks that
measure their residence time and cable delay. For a direct
connection to a time source over short distances cable delay
measurements and corrections are optional, this simplifies
implementations of slave-only devices, but also limits their
usability.
GPS Satellite Time Synchronization
Power System
GM
PMU
IEEE 1588 Time
PMU
Synchronization
PMU
PMU
ETHERNET
PDC/Server PDC/Server
Figure 14. Replacement of GPS modules
For the timing parameters required for encapsulation into
IEEE C37.118 data frames, the PC37.238 mechanism provide
all of these. The PC37.238/D5.5 Annex C describes the
mapping between PC37.238 local clock accuracy and IEEE
C37.118 time quality nibble [5]. Most importantly PMUs
need to know if the time received is in fact traceable to UTC.
This information is provided in PC37.238 TimeTraceable
flag. Note that physical connection (Ethernet) may still be
functional, but grandmaster (time source) may not be
traceable to UTC anymore. In this case no valid
synchrophasor measurements can be performed. Bit 13 in
synchrophasor STAT word indicates whether PMU time is
locked to UTC (as opposed to just having valid timing input
or not).
Although traceability to UTC time is the most important for
PMU applications, the knowledge of grandmaster
identification, provided by the PC37.238 mechanism can be
useful during transient states as well.
The local time information provided by the PC37.238
protocol can be useful for applications requiring local time,
including day-light savings adjustments.
4 Conclusion
IEEE 1588 Power profile is specified in the PC37.238 draft
standard as a mechanism for precise time distribution. While
approval of this standard is under way an effort should be
made to understand better how it can be applied to specific
X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
PMU
Streaming synchrophasor
data on the network to the
PDC for archiving . . .
APPLICATIONS
. . . data display
and real time
control actions
63
power applications, e.g. synchrophasors and IEC 61850
process bus. Reaching such understanding will raise industry
awareness on the applicability of the profile; it will speed up
and widen its acceptance and adoption.
References
[1] Ch. Brunner “The impact of IEC 61850 on protection”,
IET DPSP Glasgow, April 2008.
[2] Ch. Brunner "Will IEEE 1588 Finally Leverage the
IEC 61850 Process Bus”, IET DPSP, Manchester,
March 30, 2010.
[3] G. Antonova “First IEEE PSRC WG H7 C7 Plug-fest”,
PAC World Magazine, March 2010,
http://www.pacw.org/fileadmin/doc/MarchIssue2010/in
dustry_reports_march_2010.pdf.
[4] IEEE Std. 1588-2008 IEEE Standard for Precision
Clock Synchronization Protocol for Networked
Measurement and Control Systems, August 24, 2008.
[5] PC37.238/D5.5 Draft Standard for Use of IEEE 1588
Precision Time Protocol in Power System Applications,
September 15, 2010.
[6] IEEE Std. C37.118 – 2005 IEEE Standard for
Synchrophasors for Power Systems.
[7] “Implementation guideline for digital interface to
instrument transformers using IEC 61850-9-2”, UCA
international users group (www.ucainternational.org).
[8] ITU-T G.8621 Timing and synchronization aspects in
packet networks
http://www.itu.int/rec/T-REC-G/recommendation.asp?lang=e
n&parent=T-REC-G.8261
[9] J. Amelot, et al “An IEEE 1588 Time Synchronization
Testbed for Assessing Power Distribution Requirements,
ISPCS 2010 New Hampshire, September 2010.
[10] J.C. Tournier, K. Weber “Differences and Similarities
between the Audio Video Bridges and Power System
Profiles for IEEE 1588”, ISPCS 2010, New Hampshire,
September 2010.
[11] J.C. Tournier, T. Werner “A Quantitative Evaluation of
IEC 61850 Process Bus Architectures”.
[12] LXI IEEE 1588 Profile, Draft 2, April 26, 2008.
[13] K. Fodero, etc “Secure, Wide-Area Time
Synchronization”, WPRC 2009, Spokane, October 2009.
[14] M. Goraj, R. Moore “First Practical Experience with
IEEE 1588 High Precision Time Synchronization In
High Voltage Substation with IEC 61850 Process Bus”,
PAC World Conference, Dublin, June 2010.
[15] P802.1AS/D7.2 Draft Standard for Local and
Metropolitan area Networks – Timing and
Synchronization for Time-Sensitive Applications in
Bridged Local Area Networks, August 18, 2010.
[16] TICTOC Transporting PTP messages (1588) over
MPLS networks, September 22, 2010
http://www.ietf.org/id/draft-davari-tictoc-
1588overmpls-00.txt
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
An Experimental Analysis of High-Speed-Distance
Protection
Cezary Dzienis Matthias Kereit
E D EA D E D EA D
Siemens AG Siemens AG
Berlin, Germany Berlin, Germany
cezary.dzienis@siemens.com matthias.kereit@siemens.com
Abstract—In this paper the High-Speed-Distance Protection
algorithm will be discussed. The principle of this novel function is
based on the so called dynamic delta-quantities resulting from
splitting the electrical circuit after a short circuit event by means
of the superposition technique. Due to fact that the estimation of
the fault localization takes place in a time domain, the algorithm
presents a good alternative to the conventional distance
protection, where response speed of the relay plays an important
role (high voltage network). The test results of the High-Speed-
Distance will be discussed and application areas of this function
presented.
Keywords: Protection System, High-Speed-Distance, Loop
Selector, Directional Element, Distance Element, Testing the
Distance Protection
I. INTRODUCTION
The electrical short circuit is one of the most dangerous
phenomena in power systems. It results in system instability
and, in the worst case, outage. Therefore, it is imperative that
short circuits be selectively eliminated from the system
operation. This task is taken over by the protection devices [1]-
[2]. Especially in high voltage networks, stability must be
ensured by fast protection systems, which allow for detection
and clearance of the faults in as short a time as possible. Since
the high voltage networks often distribute energy over
hundreds of kilometers, the most popular fault detection
instrument for this voltage level is the so called distance
protection. Amongst others, its significant advantage is that it
can be installed at any network node and can work
autonomously to protect or be responsible for a specific
network line. The communication with other devices is not
necessary here. The principle of the common distance
protection is to measure short circuit impedance in the
electrical loop containing the fault and, based on this
information, the protection device should decide if the short
circuit is localized in so called protected zones applied with the
settings [1]-[3]. The impedance measurement process has a
negative influence on the tripping time and thereby on the
system operation [4]. In this article a further method for the
short circuit distance estimation will be presented. The
estimation of the fault location takes place by analyzing the
samples. As a result of the estimative nature with regard to the
fault location contained in the described method, it can attain
significantly faster operation compared with conventional
distance protection. Therefore, this computation technique is
65
Jörg Blumschein Michael Claus
E D EA D E D EA PRO
Siemens AG Siemens AG
Berlin, Germany Nürnberg, Germany
joerg.blumschein@siemens.com michael.claus@siemens.com
called High-Speed-Distance (HSD). The procedure is based on
delta-quantities which reflect the electrical state change in the
network due to the short circuit event [5]. The paper will
discuss the theoretical background of the method. Detailed
discussion of the sensitivity of the function will be carried out
as well. The method was successfully implemented in a
prototype device, and the response of the algorithm to different
fault types will be presented. The test version of the algorithm
was successfully tested on a complex hardware network model.
Thus, the acquired results of the algorithm responses are
representative and comparable with a practical application.
II. BACKGROUND OF THE METHOD
A. Theoretical Consideration
As already mentioned, the electrical fault at the line is a
dangerous event that results from rapid, undesirable change in
the physical network structure with a significant impact on the
energy transmission in entire system. Because of that, such
phenomenon should be recognized as quickly as possible and
then the faulty network area eliminated from system operation.
To do so the conventional distance protection has been
designed, the task of which is to measure the impedance of the
fault contained in the electrical loop and to compare it with the
impedance settings that define the operation range of the
protection. In order to calculate impedance the following
simplified expression, based on Figure 1, can be applied:
f cos [u A ] + j ⋅ f sin [u A ] ,
aZ L =
f cos [i A ] + j ⋅ f sin [i A ] (1)
where aZL is the fault impedance fcos, and fsin are the cos and
sin filters respectively. The response time of the (1) depends on
the filter length. Therefore, time performance of this method is
limited mostly by the applied filter length. Normally the
calculation technique to determine the short circuit impedance
can require more samples than those contained in the half cycle
interval of the fundamental period of the system. Depending on
the short circuit location and conditions before short circuit
occurred, the number of samples required is sometimes more
than those in one or two fundamental cycles [4]. Therefore,
another method with better performance can be applied.
Namely, by analyzing the situation before and after the fault
event, the exact fault localization can be carried out without
needing to determine the short circuit impedance. This method
is based on so called delta-quantities. The principle of the
66

method can be considered based on the simplified single phase

system given in Figure 1. In this case the fault is represented as
the closing of the switch K. The protected line exists between
points A and B with line impedance of ZL. The system is
supplied by two sources ES,A and ES,B with certain short circuit
power characterized by the source impedances ZS,A and ZS,B.
The fault is placed in such a way that the calculated impedance
from measurement point A equals aZL and from point B equals
(1-a)ZL. Since the distance protection is an autonomous device
that does not need communication with another end, the
consideration can be performed from just one point, e.g. A
only. From the superposition principle applied to linear
networks, the electrical system from Figure 1 after closing
switch K, can be split into two networks as given in Figure 2
and Figure 3. This is possible if the closing of switch K is
modeled as a series connection of two equivalent voltage
sources uK with opposite sign and if the magnitude equals the
voltage at the location of the short circuit before the short
circuit occurred [5]-[6].
Figure 1. Schematical representation of electrical circuit structure change
after fault on the line
Figure 3. Schematical representation of the electrical circuit in the form of
the delta-quantities
The curves of the calculated currents and voltages for given
circuits are presented in Figure 4. The pre-fault quantities
up,A(t), ip,A(t) are still constant in reference to magnitude and
phase. However delta-quantities reflect the voltage and current
changes in comparison to the previous network state. Therefore
strong changes can be observed. Additionally, an angle of
about 90° (short circuit impedance angle) between these
quantities appears. This is due to the fact that the short circuit
loop has a strong inductive component. As can be seen from
the equivalent circuit in Figure 2, the localization of the fault
can not be carried out. The gathered quantities from this circuit
do not generate a new contribution to the network operation.
They still carry information according to the load conditions.
On the other hand, the delta-quantities resulting from the
equivalent circuit in Figure 3 deliver completely new
information resulting from the short circuit impact. They can
be used for detection of the fault position in the network [6]-
[7].

The network in Figure 2 reflects system operation as if short

circuit didn’t happen. It defines the so called pre-fault
conditions. The equivalent voltage uK doesn’t influence the
voltages and currents in this network because this equivalent
voltage equals the potential at the location of the short circuit.
Figure 3 represents the system operation accompanying the
short circuit event. In this case the equivalent voltage source uK
has significant influence on the voltages and currents in the
considered system (fault condition). Since other active
elements are short circuited and applied in the previous
network system (Figure 2) only this source produces the
currents and voltages. Appling the superposition method to
these two circuits the following equation can be arranged [7]:
u A (t ) = u p , A (t ) + Δu f , A (t ) Figure 4. Electrical quantites during fault event: a) voltage and current
, before and after short circuit; b) voltage and current without short circuit; c)
i A (t ) = i p , A (t ) + Δi f , A (t ) (2) delta-voltage and delta-current.

where uA(t), iA(t) are measured voltage and current at point A.
The up,A(t) and ip,A(t) are quantities which reflect pre-fault
conditions. The Δuf,A(t) , Δif,A(t) delta-quantities represent fault
conditions with decoupling of the load flow.
The calculation of the pre-fault currents ip,A(t), pre-fault
voltages up,A(t), delta-voltages Δuf,A(t) and delta-currents
Δif,A(t) can be carried out if the network structure and its
parameters are known. In a normal case, only the measured
values after and before short circuit are available. Therefore the
relevant delta-quantities can be approximated based on fault
and pre-fault values of iA(t) and uA(t) as expressed in (3):
Δu f , A (t ) ≈ u A (t ) − u A (t − T ) ,
Δi f , A (t ) ≈ i A (t ) − i A (t − T ) (3)

where T is the fundamental period. The uA(t-T) and iA(t-T) are

Figure 2. Schematical representation of an electrical circuit in the form of close to pre-fault quantities up,A(t) and ip,A(t). It is assumed here
the pre-fault quantities that behavior of the network, before short circuit happened, is
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
not disturbed. Otherwise this approximation is afflicted with
relatively significant errors. To avoid the error for frequency
deviation a more complicated approximation technique for
creating delta-quantities must be realized.
B. High-Speed-Distance Function
Three particular modules create the High-Speed-Distance
protection function. These are: loop selector, directional
element and distance element. In addition, the other functions
must be implemented that have the task to stabilize the function
due to disturbances or diverse non-conformed network states.
The cases where the stabilization rules are needed and also
justified will be presented in section III.
Based on the delta-quantities the faulty loop can be selected.
Particularly, such selection takes place by analyzing the delta-
currents and delta-voltages. Investigations of the delta-phase-
to-phase quantities for both current and voltage are very useful
for this task as well. For example, based on the delta-phase-to-
phase currents it can be very easily differentiated between
single and multiple pole faults, and the phases in which the
fault occurred can be detected. The following statement results
from such consideration: for a single-phase fault the delta-
current for healthy phases is close to zero; if the double phase
fault appears then the delta-phase-to-phase current in unhealthy
phases is much larger than other phase-to-phase delta-currents;
in case of the three phase fault all phase-to-phase delta currents
are equal to each other. Since the loop selector must decide in a
short time which loop is defective, the confirmation of the
faulty loop must be carried out. This can be realized by means
of the delta-voltages, graphically shown in Figure 5.
Figure 5. Loop selection based on the delta-voltage
For example, if a single pole fault appears the voltage breaks
down in the defective phase and the ratio between the delta-
voltage in the defective phase to the delta-phase-to-phase
voltage in the healthy phases is high. In the case of a
recognized two-pole short circuit the ratio between the
unhealthy phase-to-phase and the non-defective phase is
investigated. If this ratio is high the phase-to-phase loop is
confirmed. For three-phase faults the ratios between phase-to-
phase voltages are equal to each other. Such a two step loop
selector creates a very robust element. Analyzing delta-
quantities, it can be concluded that the calculated impedance
according to (1) allows for detection of the direction of the
fault:
67
[ ] [ ]
f cos Δu A, f + j ⋅ f sin Δu A, f ,
ΔZ =
[ ] [ ]
f cos Δi A, f + j ⋅ f sin Δi A, f (4)
where ΔZ is the impedance acquired from delta-quantities. ΔZ
equals to –ZS,A (negative impedance) if the fault is in a forward
direction or ZS,B+ZL (positive impedance) if the fault is
localized in a reverse direction. Since the calculation of the
impedance ΔZ doesn’t bring any advantages in comparison to
the method of the conventional distance protection (the
response speed depends still on the filter length), only
impedance sign must be predicted. The prediction is based on
the assumption of the so called replica impedance ZR [8]-[9]
that should be close to the back source impedance ZS,A.
Equation (5) shows the acquiring process of the delta-voltages
in a schematic way:
Δu f = Δu f , A
,
Δu R = Z R D Δi f , A (5)
where ZR is assumed replica impedance. A denotes the
measurement point in the system (station A). The expression is
valid for single phase representation of the system given in
Figure 1. The calculation should take place in the time domain
in which the ZR is replaced by the differential operator in form
of (LRd/dt + RR). The acquired quantities create the trajectories,
which for forward fault are placed in quadrants II and IV of the
ΔuR, Δuf plane. Also, the trajectory circulates in a positive
clockwise direction. For reverse fault, the trajectory is localized
in quadrants I and III. It rotates in a negative clockwise
direction. The usage of the replica impedance ZR allows for
transformation of the delta-current Δif in such way that
occurred trajectories from both delta-voltages ΔuR, Δuf are very
well defined. If only delta-voltage Δuf with delta-current Δif are
taken into consideration, the computed trajectory is placed in
all quadrants and the fault direction can be detected based on
recognizing the rotation sense and position trajectory in the
initial condition (fault begin). Since the fault quantities include
the distortion or, e.g., DC components this is definitely not a
secure instrument [7].
Figure 6. Trajectories and curves of the directional element based on delta-
quanities
The possible trajectories for these two fault directions are
presented in Figure 6. Generally they have approximately an
elliptical shape that results from differences between assumed
replica impedance ZR and real appeared impedance calculated
in the form of delta-impedance ΔZ. If the angle of the assumed
replica impedance ∠ZR is close to the angle of delta impedance
∠ΔZ the ellipse form aspires to be a straight line. If additional
magnitudes of both impedances are the same, a straight line
68
with the slope of 45° appears. Since in the range of small delta-
quantities, high uncertainty appears, (e.g. mostly measurement
error) the so called dead-zone must be introduced. It
contributes certain limitation to the method that can be
annulled if the delta-voltages are simply filtered by an integral
function in the following way [8]-[10]:
τ
F (τ ) = ∫ Δu f (t ) ⋅ ΔuR (t ) dt .
0 phase/phases. As a result, a directional element is activated for
If the function F(τ) is negative (for τ=τ1) then forward fault
will be concluded. For the reverse fault a positive integral
function F(τ) (for τ=τ2) should appear. The uncertainty of the
method is then reflected by the non directional area as
presented in Figure 6. Depending on the availability of the fault
quantities after short circuit, the given threshold for reverse or
forward fault recognition can vary. The replica impedance ZR
is the main factor that has impact on the sensitivity of this
direction estimation method. Thereby two components play
central role: there is magnitude and angle of the replica
impedance ZR. The magnitude can contribute so called under-
function only, e.g. if the chosen magnitude for replica
impedance is too low. However, significant deviation of the
angle from the real delta-impedance ΔZ can introduce over-
function. The influence is presented in Figure 7 and Figure 8.
Integral [-]
0
Source replica impedance angle 86°
-0.2
-0.4
-0.6
-0.8
-1
0 0
4 20
8 30
12
60
Sample number [-] 16
80
70
Source impedance angle [°]
20 90
Figure 7. Sensitivity of the direction method for a different replica
impedance angle
Integral [-]
2
Source replica impedance angle 86°
1.5
1 of the reference voltage uref in the time domain. The index A
0.5
0
-0.5
-1
0 -24
4 -18
8
Sample number [-] 12 Source impedance angle [°]
16 -6
20 0
Figure 8. Sensitivity of the direction method for a different replica
impedance angle
It was assumed here that the replica impedance ZR has a
constant angle of 86°. Also, the forward fault was simulated in
a network with a different angle of reverse source impedance.
The angle of this impedance varied between 90° (inductive)
and -30° (capacitive). It can be noted that even if angle
deviation is big, the integral curve has correct tendency. Only
the oscillated form can be observed. If the source impedance is
capacitive then the curve develops in the defective opposite
direction. This results from the fact that, the assumed model for
delta-voltage calculation is not matched with the physical
condition in the network. Since the source impedance is
inductive and the line impedance, also by series compensated
line, is inductive as well, the direction calculation can be
interpreted as a stable method. In the normal case (three phase
(6)
system) delta-voltages are calculated for the defective
the faulty loop only. The six loops are not calculated in
parallel. Which loop is activated, depends on the signal from
the loop selector, shortly described above. For the phase-to-
phase defective loops the delta-voltages uf and uR are created
from phase-to-phase voltages and currents respectively. For the
phase-earth loops, the delta-voltages uf and uR are built from
phase-to-earth voltages and phase currents respectively. Since
the loop in the phase-earth mesh is inductive, the earth current
with appropriate influence of the replica zero-sequence
impedance ΔZ0 is not taken into account because it
insignificantly influences the directional element algorithm
(see general consideration in Figure 7 and Figure 8). Besides
this replica zero-sequence impedance ΔZ0 is unknown and
similar to the common line replica impedance ZR must be
assumed. It can be an additional error source. The mutual
coupling also has low influence on the directional element,
because during this effect the loop character is still inductive
and consideration from Figure 7 and 8 can be applied.
The calculation of the fault localization is based on the
comparable analysis of the equivalent voltage uK at the location
of the short circuit. Since this location is unknown, the
reference voltage uref must be calculated. This voltage results
from the theoretical voltage at the end of the protected zone as
10 if the fault would appear exactly at end of the protected zone.
50
40 Hereby, the protected zone is characterized by high-speed-
distance impedance (HSD) ZHSD, particularly lower then the
line impedance ZL. Applying Kirchhoff’s laws for equivalent
circuit resulting from the superposition principle (Figure 2), the
reference voltage uref can be expressed as follows (7):
u ref = u p , A − Z HSD D i p , A , (7)
where ZHSD is the differential operator in form of (LHSDd/dt +
RHSD). Usage of this differential operator allows for calculation
denotes the measurement point in the network (in this case the
-30
station A). It must be noted here that all computations must be
preformed in the time domain. In the normal case, formula 7
-12
must be fitted into an appropriate defective electrical loop that
is annunciated by the phase selector. Hereby the differences in
equation between single-pole short circuit to earth and multiple
pole phase faults appear. Expression 7 presents the
simplification that is valid for the single-phase system if,
additionally, the earth line impedance is equal to zero. For
double-phase and three-phase faults the phase-to-phase loop is
selected and calculation takes place on phase-to-phase
quantities for voltage and current respectively. When the
phase-earth loop is selected, the appropriate phase-to-earth
voltage with suitable phase current and earth current is
involved in the calculations. In such case the expressions from
7 and 8 must be completed with the zero-sequence

compensation factor that reflects non-homogeneity between
phase and earth impedances. Since the selector chooses one
loop only, only one loop is calculated by the distance element.
The equivalent voltage at the end of the high-speed-distance
zone can be also calculated from delta-quantities (see section
II) like (8):
u HSD = − Δu f , A + Z HSD D Δi f , A .
In order to detect where the fault is localized, comparison
between these two voltages, uref and uHSD, must be carried out.
Thereby, the following conclusions can be made:
- if uHSD > uref then the equivalent source uK must be
situated in the zone ZHSD, (internal fault)
- if uHSD < uref then the equivalent source uK must be
situated outside zone ZHSD (external, forward or
reverse fault)
- if uHSD = uref then the equivalent source uK is situated
exactly at the zone ZHSD (fault at zone limit)
In order to achieve good stability of the distance element with
good time performance the calculation of the voltages uHSD and
uref is realized by means to of the average rectified values.
These values are obtained in the moved constant window with
the length of the half of fundamental period. Since the building
of the average rectified value for the voltage ukHSD is not
always possible, the missing samples of the voltage ukHSD are
replaced by samples of the uref. This happens only for the short
time after fault. Additionally, the dynamic pick-up
characteristic for distance element is applied, that introduces
stabilization in case of disturbances which can be amplified by
the derivation element in expression 7 and 8.
The sensitivity investigation of the distance prediction method
based on delta-quantities can be performed using the complex
static values. At first the situation of the fault in forward
direction will be considered. From a measurement point of
view the complex equivalent voltage source can be computed
as given in (9). For this task the equivalent circuit from Figure
3 should be applied:
U K = −ΔU f , A + Δ I f , A ⋅ Z f = Δ I f , A ⋅ ( Z S , A + Z f )
, .
U HSD = −ΔU f , A + Δ I f , A ⋅ Z HSD = Δ I f , A ⋅ ( Z S , A + Z HSD )
where Zf is the short circuit impedance. Since this short circuit
impedance is unknown, the theoretical equivalent voltage UHSD
can be computed like in (8), but on complex values. As can be
easy noted, if the fault impedance is lower then HSD
impedance (fault in the zone), the voltage UHSD is higher than
voltage UK. On the other hand, voltage UK is approximately
equal to the Uref computed from the equivalent circuit in Figure
2. Additionally, usage of voltages UK and UHSD allows for
computation of the HSD-characteristic (10):
U ref ≈ U K ≤ U HSD
, HSD-function is guaranteed. Since the HSD protection method
2 (10)
( R S , A + R f ) 2 + ( X S , A + X f ) 2 ≤ ( Z S , A + Z HSD )
where RS,A and XS,A are parameters of the source impedance. Rf
and Xf are resistance and reactance measured at the short
circuit location. A circle results from this equation, see Figure
9 for which the following interpretation can be applied: the
HSD-pick up characteristic is a circle with the center created
X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 69
by the source impedance in reverse direction ZS,A and radius as
the sum of the source impedance ZS,A and HSD-impedance
ZHSD. Since the fault impedance Zf is placed in quadrant I of
the complex plane, only the characteristic in quadrant I is
applicable. Depending on the source impedance ZS,A, the
distance characteristic can vary in R direction (real axis). The
reach in direction of the line impedance ZL is always constant.
(8) Therefore, the best performance of the HSD-algorithm is
obtained if a metal short circuit without transition resistance
appears.
Figure 9. Distance characteristic for the forward fault (HSD-method)
Similar considerations can be carried out for the reverse faults.
The mathematical equation for description of the voltage in
form of UK and UHSD quantities is given in (11):
U K = −ΔU f , A − Δ I f , A ⋅ Z f
.
= −Δ I f , A ⋅ ( Z f + Z L + Z S , B )
U HSD = − ΔU f , A + Δ I f , A ⋅ Z HSD (11)
= Δ I f , A ⋅ ( Z L + Z S , B − Z HSD )
If we compare the magnitudes of both voltages UK and UHSD
with each other and assume that the line impedance ZL is
approximately equal to HSD-impedance ZHSD the following
expression is can be written:
U ref ≈ U K ≥ U HSD
(9) 2 (12)
( RS , B + R L + R f ) + ( X S , B + X L + X f ) ≥ Z S , B
2 2
Also in this case the circle characteristic is obtained. The radius
of the circle equals the magnitude of the source impedance ZS,B
and the circle is shifted as a sum of the line and source
impedance. From this characteristic it can be observed that the
fault impedance Zf cannot reach the inside of the circle (fault in
HSD zone), so that no over-function for reverse fault should
appear. It can be expected that that HSD algorithm behaves
very stable for these fault types. The next conclusion is that the
distance element includes the properties of the directional
element. As a result of that the more stable directionality of the
is based on the time domain calculation, its distance estimation
algorithm can exhibit some uncertainties in reference to the
static characteristics from Figure 9 and Figure 10. It can
especially occur, if the comparison is carried out with a low
number of samples. The potential problems are discussed in
next section.
70
Figure 10. Distance characteristic for the reverse fault (HSD-method)
III. TEST SYSTEM AND TEST RESULTS
According to the considerations from section II, for the
loop selector, directional element, distance element and other
supported methods, the discrete simulation models in
Matlab/Simulink were developed. Based on simulation results
acquired from the real time hardware model, the expanded
models were appropriately optimized so that the desired
balance between performance and stability of the entire HSD-
function was achieved. These models were implemented into a
digital protection device and suitably fitted at its structure. The
function was tested according to its behavior for different
network states (mostly faults) and according to its integration
level in a digital device. The test results are described in this
section. The model on which the tests were performed is a
scaled physical network of 500kV high voltage system. Scaling
of the parameters took place with the commonly used
secondary values. The advantage of such a scaled network
against the RTDS network model, is that many of the different
electro-magnetic effects can be included. In order to guarantee
the real time simulation for the big network structure, an RTDS
system uses strongly simplified models. Thereby some possible
interesting phenomena (mostly with non-linear background) for
protection tests can not be taken into account.
)
XL
54
(66
kM
kM
(6
92
4X
L)
Figure 11. Test system for protection systems
Tested devices were placed at the line D12, and the following
tests were performed:
• performance test
• over-function test
• evolving faults test
• power swing test
• special practice relevant tests
Figure 12. Results of performance test
In Figure 12 the trip times of the HSD-algorithm, for different
fault types and fault localizations on the line, were presented. It
can be observed that the tripping times, measured with an
electronic device relay, are between 5ms and 20ms. These
tripping times reflects the real response time of the algorithm
with measurement uncertainty of about 1ms. For the initiated
faults under 30% of the line length, the HSD-algorithm needs
significantly lower than half of the fundamental period to trip.
The faults from 50% up to 100% of the line length can be
tripped also in fast time, namely under single fundamental
period. Deviation of the tripping times for the same fault types
results particularly from the fault initiation angle. The best
performance of the algorithm can be achieved if this angle
equals the line angle. Besides if disturbances appear, the
tripping time is much higher. Analyzing the tripping times it
can be noted that the HSD-function offers definitely better
performance according to tripping time than a conventional
distance element based on impedance calculation. Hereby it
must be said that HSD is designed to trip clear interpretable
faults on the line. Therefore, this function should operate in
cooperation with conventional distance elements that allow for
tripping of the faults with high complexity. The usage of the
HSD-function can be definitely confirmed if it responds only
when forward fault appears on the protected line. Therefore the
algorithm was tested according to potential over-functions.
Hereby, the following faults outside the protected zone were
initiated on the model: reverse faults, faults on the limit of the
HSD zone, faults on a parallel line, faults during week in-feed,
external faults during frequency deviations as well as external
faults with significant disturbance content. In order to show the
intensity of the tests according to the over-function two cases
were graphically presented. In Figure 13 the forward fault
outside the zone with significant disturbance content was
plotted. For the fast distance function (HSD) based on time
domain calculation, the appeared disturbances during short
circuit event can be especially dangerous, because the signal
processing is limited to a very short pre-processing filter,
mostly anti-aliasing filter only. Since for the distance
estimation procedure a derivative operator is applied, the
disturbances can be amplified and therefore an overreach of the
distance function can appear. This eventually effect can be
recognized in Figure 13. The voltage curve uHSD acquired from
delta-quantities exhibits strong deviation from the fundamental
component of 50Hz sinus wave. For some samples this curve
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
exceeds the reference voltage uref. Hereby it can be suggested
that the fault is inside the HSD-zone and a defective trip can
happen. Based on this example it can be concluded that
prevention of the overreach effect during the disturbances is a
big challenge for the HSD-function.
Figure 13. External Fault at >100% line length with significant disturbances
content
Figure 14. Reverse fault with significant disurbance content
In Figure 14 disturbance problems for the reverse fault are
shown. This is a more complex situation than the overreach
effect in the case of the forward fault. Because of significant
disturbances defective direction of the short circuit can be
detected. Since for reverse fault close to the measurement point
the voltage breaks completely down (significant delta voltage
quantities) and additionally the short circuit current is strongly
disturbed, the directionality of the distance element can not be
guaranteed any more. Then the mis-operation of the HSD-
function can happen. Such potential effect of the disturbances
impact is presented in Figure 14. The reference voltage uref is
exceeded repeatedly by the voltage uHSD. In order to stabilize
the response of the algorithm in case of the disturbances, a
special method was developed that detects the distortions in a
short time and either blocks the HSD-function completely or
introduces an appropriate stabilization threshold. This depends
on the force of the disturbances. The method is based on the
investigation of the function monotony in an available
calculation window. The disturbance detection method
combined with the main HSD function creates a very stable
high speed distance protection. Therefore, usage of the HSD is
also confirmed in a network with significant disturbance
content. Since in the high voltage networks, so called evolving
71
faults can happen, the HSD-function was programmed in such
way that its operation during these fault types is possible as
well. The performance of the method according to the tripping
time is also very good. This is shown in Figure 15. The
transferring of the reverse to forward fault was simulated. In
most cases the tripping time took place in less than half of the
fundamental period. The only time the HSD-algorithm didn’t
trip after 20ms was for the evolving fault BCN. This is due to
the occurred condition according to the earth current.
Moreover, the other fault transferring possibilities were tested
like: internal to internal and internal to external fault with
different delay times. The HSD-function operates very stable
and no over-function is detected.
Figure 15. Results of the evolving fault tests
The interesting case of the fault transferring from external to
internal during the same phase is presented in Figure 16. At
first the reverse fault was initiated and after 20ms the forward
fault was started. After reverse fault, the directional element
recognized reverse fault and the distance element was not
activated. Nevertheless the distance element exhibits also
correct directionality. The uHSD curve is lower than reference
voltage uref. After forward fault with a delay of 20ms the
voltage curve uHSD exceeds the reference voltage uref and a
single pole trip happens. From this consideration it can be
concluded that the HSD-function can be used for clearance of
evolving faults as well. Excellent performance according to the
tripping time and stability of the HSD-function enlarges its
application area to these fault types. This is very important
because generally the tripping time of the conventional
distance protection is a little bit higher for the evolving faults
than for the simple faults. The power swing phenomenon is
relative frequently occurring state in the power system.
Detailed tests were performed for this phenomenon as well.
Three important reactions must be expected from HSD to make
it applicable as distance protection: no over-function during a
pure power swing, no over-function during an external fault
that is combined with a power swing and, a reproducible trip
for an internal fault during a power swing. After numerous
tests it could be observed that HSD-function didn’t exhibit
over-function also during extremely high power swing
frequency, over 10Hz. The appropriate reaction of HSD during
power swing for internal fault is presented in Figure 17. Before
short circuit happened the significant delta-quantities for
voltage and current already appear, which result from power
swing phenomenon. This has naturally a negative influence on
72
HSD-function operation, because this method is based on
delta-quantities. However, during the strong short circuit more
affected delta-quantities are created that contribute to detection
of the fault direction and its localization. This is shown in
Figure 17. After fault the voltage uHSD is higher than uref.
Therefore, it is correctly concluded as an internal fault. In order
to avoid the mis-operation of HSD-function during power
swing some stabilization rules were designed for this network
state as well. It should be noted here that effects can appear that
are similar to those of faults during disturbances. Since lower
frequency components appear during a power swing, the
stabilization algorithm has a more static form. Moreover the
relevant states of different networks were simulated and the
reaction of the HSD-algorithm was investigated. These
network states include switching on fault, external and internal
fault clearance, frequency deviation, weak in-feed, etc.
Figure 16. Evolving fault during the same phase AN reverse to AN forward
Figure 17. Internal fault during power swing
Each of these mentioned effects can have a negative influence
on HSD-function. Therefore during implementation numerous
stabilization and plausibility rules for detection of non
conformed network operation were designed and successfully
tested. The importance of such tests can be considered based
on the analysis of external fault clearance process. After
external fault the HSD-algorithm is picked up. Because the
fault is not in the protected zone no reaction is expected.
Through the external fault clearance in a short amount of time,
from HSD point of view, conditions similar to those of a
normal fault occur. The voltage will be recovered and current
will drop rapidly down, so that big delta-quantities for these
values appear. Additionally during fault clearance, relative big
disturbances can be recognized, which can contribute to an
over-function of the HSD-method. Therefore switching
operation in the network can be dangerous for time based
distance protection. Many such diverse dynamic and static
network states were simulated and no over-function of HSD-
algorithm was detected. Therefore, this function can be
confirmed as stable. Disturbed operation of the measurement
system, in addition to an undesired network state, can influence
negatively the HSD-algorithm and contribute to mis-operation.
These disturbances regarding measurement problem include
CT saturation, CT broken or fuse failure. Extensive tests were
carried out in the range of these phenomena and proper HSD-
reactions were recognized.
IV. SUMMARY
In this paper the High-Speed-Distance protection function was
discussed in detail. The theoretical background of the method,
its sensitivity investigation as well as test results were
presented. As can be concluded, this time based distance
function is characterized by very good time performance and
robustness. Also, the application range of the method is huge.
It spreads from the simple faults through evolving faults to
more complex faults during, e.g., power swing. Nevertheless,
the HSD-algorithm should operate in combination with
conventional distance protection because its under-function
reaction can not be excluded. The HSD-algorithm exhibits
under-function reaction particularly when very complex faults
appear or if there is a strong deviation from the conformed
network operation before the fault occurred. Therefore, HSD
can not completely replace conventional distance protection,
but it can operate as secure support for conventional distance
protection.
REFERENCES
[1] H. Ungrad, W. Winkler and A. Wiszniewski, “Protection techniques in
electrical energy systems,” Marcel Dekker, INC New York 1995.
[2] A. G. Phadke and S. H. Horowitz, “Power System Relaying,” Wiley
Inc., 1995.
[3] G. Ziegler, “Numerical distance protection: principles and applications”,
Publicis Corporate Publishing, 3rd edition 2008.
[4] H.-J. Herrmann, “Digitale Schutztechnik: Grundlagen, Software,
Ausführungsbeispiele,“ VDE-Verlag GMBH, Berlin, Offenbach 1997.
[5] M. Vitins, “A fundamental concept for high speed relaying,” ,” IEEE
Transactions on Power Apparatus and Systems, vol. 100, No. 1. pp. 163-
173, January 1981.
[6] M. Chamia and S. Liberman, “Ultra high speed realy for EHV/UHV
transmission lines – development, design and application,” ,” IEEE
Transactions on Power Apparatus and Systems, vol. 97, No. 6. pp. 2104-
2112, Nov/Dec 1978.
[7] F. Engler, O.E Lanz, M. Hanggli and G. Bacchini, “Transient signals
and their processing in an ultra high-speed-directional relay for
EHV/UHV transmission line protection,” IEEE Transactions on Power
Apparatus and Systems, vol. 104, No. 6. pp. 1463-1473, June 1985.
[8] G. Benmouyal and S. Chano, “Characterization of the phase and
amplitude comparators in UHS directional relays,” IEEE Transactions
on Power Systems, vol. 12, No. 2, pp. 646-653, May 1997.
[9] G. Benmouyal, “Amplitude-independent comparators for UHS
directional relays,” Developments in Power System Protection,
Conference Pubplication IEE, no. 434, pp. 78-82, 25-27th March 1997.
[10] K. S. Prakash, O.P Malik and G.S Hope, “Amplitude comparator based
algortihm for directional comparison protection of transmission lines,”
IEEE Transactions on Power Delivery, vol. 4, No. 4, pp. 2032-2041,
October 1989.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 73

Medición de la Impedancia de Línea y Tierra para una Estimación Correcta del

Factor de Tierra (K0) en Líneas de Transmisión y Mejora de la Confiabilidad de la
Protección de Distancia

Miguel Gutiérrez, OMICRON Electronics, USA

Michael Krugger, OMICRON Electronics Gmbh, Austria

RESUMEN

Dentro de los factores de

incertidumbre que afectan la
localización de la falla y por
consiguiente la confiabilidad de la
operación de la protección de
distancia se encuentra el factor de
tierra. Aunque hay varios programas
de cálculo bastante sofisticados para
la estimación de los parámetros de
línea, la gran incertidumbre que
introduce principalmente la
impedancia de tierra conllevan en
algunos casos a imprecisas
estimaciones del factor K0. Una mala
estimación de K0 puede conducir a la
protección de distancia a problemas
de sobrealcance o subalcance.
En este trabajo se analiza plenamente
un nuevo método para medir las
impedancias de línea y tierra que
determina K0 por medio de inyección
de corrientes bajas a una frecuencia
fuera de la nominal. Los valores de
impedancia de línea y tierra son
interpolados de las mediciones
realizadas en la línea a frecuencia no
nominales. Esto garantiza que el
ruido del sistema no va a afectar la
estimación de los parámetros.
relevadores de protección. Cuando
ocurre una falla en la línea, ya sea
entre fases o a tierra, tiene que ser
despejada en forma selectiva, rápida
y precisa. La selectividad significa que
sólo será desconectada la línea donde
ocurra realmente la falla.
Existen dos métodos básicos para
conseguir selectividad en las líneas de
transmisión la protección diferencial
de línea y la de distancia. El primero
es el mejor, pero requiere de un
gasto mayor, ya que los relevadores
en ambos extremos deben poder
comunicarse uno con otro. Debido a
estos costos en muchas líneas se
usan los relevadores de protección de
distancia.
Uno de los más importantes ajustes
de un relevador de protección de
distancia es la Impedancia de
Secuencia Positiva, la cual es la mitad
de la impedancia compleja de los
circuitos de fase a fase (Figura 1).

I NTRODUCCI ÓN
Para proteger líneas aéreas o cables
de potencia son necesarios los
74
Figura 1. Circuito de Impedancia
entre dos fases
Cuando una falla ocurre los
relevadores de distancia en ambos
extremos miden la impedancia. Si la
impedancia es (típicamente) menor al
80% ó 90% de la impedancia total de
la línea entonces abrirán la línea lo
más pronto posible (Zona 1), porque
es seguro que la falla es interna. Pero
si la impedancia es mayor los
relevadores actuarán con un retardo
(Zona 2), para dar la oportunidad a
otro relevador que se encuentre más
cerca de la falla para despejarla
antes.
En fallas de una ó más fases a tierra,
la impedancia del circuito de falla es
diferente (Figura 2). Porque la
impedancia de la red de tierra o más
bien del circuito a tierra, es diferente,
un factor de ajuste en los relevadores
da la relación entre la impedancia de
la línea y la de tierra. Este factor es
llamado Factor de compensación a
tierra o simplemente factor K, como
ya se ha mencionado antes.
Figura 2. Circuito de Impedancia de
una falla monofásica a tierra
Si los ajustes del relevador son
hechos correctamente un cliente que
está siendo alimentado desde los dos
extremos (Figura 3) continua
recibiendo energía de una línea si los
relevadores 3 y 4 disparan.
Figura 3. Relevadores con alcance
óptimo de zona 1
Si los alcances de impedancia o los
factores K de un relevador no son
hechos correctamente, ocurrirán
subalcances o sobrealcances de zona
Figura 4.
Figura 4. Relevadores con
sobrealcance de zona 1
En el ejemplo anterior tres
relevadores ven la falla en zona 1 y
disparan, una segunda línea está
muerta. El cliente es dejado sin
suministro de energía sin razón.
Además del daño al cliente que se
queda sin energía, el riesgo de perder
la estabilidad del sistema comienza
también a incrementarse por estos
disparos en falso.
DESARROLLO:
DETERMINACION DEL FACTOR K
Desafortunadamente el factor K no
existe. Hay varias formas de
obtenerlo, aquí discutiremos las tres
formas más comunes para calcularlo.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Para todos los casos hay que
considerarlos como constantes de la
línea, en general independiente de la
longitud. Estos expresan la relación
de la impedancia del circuito fase a
fase y de las tres fases a tierra. La
mitad de un circuito fase a fase (por
ejemplo la impedancia de una línea)
es referido como la Impedancia de
Secuencia Positiva Z1; tres veces la
impedancia de un circuito de las tres
fases a tierra es referido como la
Impedancia de Secuencia Cero Z0.
Una forma común es la relación
compleja de la Impedancia de
Secuencia Cero y la Impedancia de
Secuencia Positiva.
Como Z1; es la impedancia de una
línea es llamada también ZL.
comúnmente.
La impedancia de tierra ZE; puede ser
calculada de la I mpedancia de
Secuencia Cero como sigue:
Otra forma de expresar está relación
sería:
Si convertimos las impedancias
complejas ZE y ZL en sus partes real e
75
imaginaria y redefinimos la relación
obtenemos la tercera forma:
El ángulo de la línea es obtenido de la
relación XL/ RL como sigue:
Los relevadores de protección utilizan
algoritmos que hacen uso de estos
diferentes factores K para convertir
todas las fallas de fase a tierra, de tal
manera que estas puedan ser
evaluadas como fallas de fase a fase.
CALCULO DE LOS FACTORES K
Hasta ahora los intentos para medir
las impedancias de línea y los factores
K habían sido muy difíciles. Para
obtener estos datos se han estado
haciendo cálculos manualmente, o
por medio de herramientas de
software como PowerFactory de
DIgSILENT, PSS de Shaw PTI or CAPE
de Electrocon.
Los parámetros necesarios para
calcular la impedancia de la línea son
muchos.
Se requiere la configuración
geométrica (Figura 5):
• Altura sobre la tierra y la distancia
horizontal de cada conductor de fase
y el cable de tierra
• Valor promedio de las catenarias a
lo largo de la línea
Se deben conocer también varios
parámetros eléctricos
76
‡5HVLVWLYLGDGGHOWHUUHQR
• Resistencia de todos los
conductores
Figura 5. Geometría de las líneas
aéreas
• Construcción de los conductores
• Diámetro de los conductores
Pudiera darse el caso de que se
reemplazó en una porción de la línea
un tramo de los conductores por otro
de diferente material o diámetro o
sólo fue reemplazado en una de las
fases y no fue documentado. La
influencia de la resistividad del
terreno y la medición exacta de la
distancia entre los cables y la tierra
son muy difíciles de determinar a lo
largo de toda la línea (especialmente
en líneas largas donde la geografía es
muy variada).
MEDICION DEL FACTOR K
Comparado con los cálculos, la
medición de los parámetros de la
línea incuyendo el factor K es
relativamente simple hoy en día.
La medición es realizada con
corrientes de entre 1 y 100 A.
dependiendo de la longitud de la
línea. El usar una frecuencia selectiva
durante la prueba nos permite
realizar la medición con una corriente
de una fracción de la corriente
nominal. Se han llevado a cabo
mediciones en líneas de hasta 270 km
de longitud. En la figura 6 se muestra
una foto de un equipo disponible en
el mercado para realizar estas
mediciones.
Se realizan siete mediciones por
sistema, tres para cada combinación
de circuitos entre fases, tres para
cada fase a tierra y una para las tres
fases a tierra. Aunque existe
redundancia en las mediciones, esto
nos da confiabilidad por que nos
permite hacer comprobaciones
cruzadas de los resultados y el cálculo
de los factores K para cada fase. A
primera vista parece extraño, pero
especialmente en líneas cortas no
siempre se tiene cuidado de que la
línea sea simétrica, lo que posibilita
diferentes valores para cada una de
las fases. Conociendo este problema
podemos reajustar los factores K para
evitar sobre-alcances de zonas.
Figura 6. Equipo para medición de la
impedancia de Línea.
Los resultados pueden ser exportados
a Microsoft Excel lo que permite un
fácil reprocesamiento de los
resultados, además los resultados son
presentados en un formato mediante
el cual pueden ser usados
directamente para ajustar el relevador
(Figura 7).
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 77

corriente de prueba seleccionada fue

de 10 A.

En la figura 9 se muestran los

resultados de la impedancia medida a
frecuencias diferentes de la nominal.

Figura 7. Resultados finales de las

mediciones

CASO REAL

Una medición en una línea con una

longitud total de 22 km fue
desarrollada en junio de 2004 en una Figura 9. Mediciones de impedancia
empresa de energía que vende más contra frecuencia
de 180,000 GWh de energía eléctrica
por año a sus clientes. Lo más interesante fue la
comparación de los resultados
medidos contra los ajustes del
relevador.

Figura 10. Ajustes del relevador

comparados con los resultados de la
medición.

Figura 8. Primera torre de los 22 km Las Impedancias de Secuencia

de línea en 400 kV Positiva son casi idénticas, pero la
Impedancia de Secuencia Cero
El motivo de la prueba fue un disparo muestra una desviación de 48%. Otra
en falso que estaba sujeto a reevaluación de los cálculos
investigación. demostró, que el tipo de cables de
tierra usados, no fue introducido
La medición completa fue hecha en correctamente en el sistema, una vez
menos de una hora porque la la línea corregido los cálculos fueron muy
no podía estar fuera de servicio por similares comparados con las
un largo período. Las mediciones mediciones de la secuencia cero.
fueron realizadas sin problemas; la
78
El cálculo incorrecto no había sido
cuestionado antes de que ocurriera el
disparo en falso, hasta que el error
fue encontrado. La compañía se
convenció, que si realizaba
mediciones en diferentes líneas de su
sistema, podrían corregir errores en
los ajustes de sus protecciones de
distancia y así evitar disparos
innecesarios, lo cual redundaría en la
confiabilidad de su sistema y
obviamente en grandes beneficios
para la empresa y mejora en la
calidad del servicio prestado a sus
clientes.
CONCLUSI ONES
En nuestros días debido a la alta
carga a las que son sometidas las
líneas de transmisión la medición del
factor K puede significar la diferencia
entre una correcta operación de la
protección o la salida de varias líneas
y todo lo que esto representa. Las
mediciones que se han realizado
muestran que por varias razones los
cálculos presentan resultados
erróneos. Además las mediciones y
los cálculos pueden ser comparados
para mejorar los ajustes y asegurar la
selectividad y la rapidez al despejar
una falla. La medición de la
impedancia de línea y el factor K son
muy importantes para una correcta
operación del relevador de protección
de distancia.
REFERENCI AS:
1- L. Barrios, A. Jarquin, "Medición de
las Impedancias de Secuencia Positiva
y Cero en Líneas de Alta Tensión",
IEEE Vigésima reunión de Verano de
Potencia, RVP 2007. México.
2-Dierks, 2004, “Different
Representation of the earth
Impedance Matching in Distance
Protection Relays Performance”,
Proceedings Southern Africam Power
System Protection Conference 2004,
Eskom Enterprises.
3- Th. Hensler, R Kaufmann, U.
Klapper, M. Krüger, Z Schreiner, 2003
“Portable Testing Device”
4- S. Kaiser, 2004, "Different
Representation of the Earth
Impedance Matching in Distance
Protection Relays". Proceedings
OMICRON User Conference Germany
2004"
CURRI CULUM VI TAE:
Miguel Gutiérrez Recibió su
bachillerato en electrónica y su
Licenciatura en sistemas de potencia
de la Universidad de Costa Rica en
1985 y 1988 respectivamente. El
tiene más de 25 años de experiencia
en el área de protección de sistemas
eléctricos, control y comunicación.
Desde 1999 trabaja con OMICRON
Electronics (E.E.U.U.) como ingeniero
aplicación para el área de América
Latina. Él es miembro del IEEE.
Michael Krüger es manager de la
línea de productos de prueba primaria
con OMICRON electronics GmbH,
Austria. Estudió ingeniería eléctrica
en la Universidad de Aachen (RWTH)
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 79

y la Universidad de Kaiserslautern
donde se graduó en el año 1976. En
1990 recibió doctorado en la
Universidad de Viena, Austria.
Michael Krugger tiene más de 28 años
de experiencia en ingeniería de
potencia especialmente en ingeniería
de alta tensión y diagnóstico de
aislamiento. El es miembro del VDE
e IEEE.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Empleo de Nomogramas para Localización de
Fallas en Redes de Distribución
Abel González Gómez Alexis Martínez del Sol
Resumen – En el p resente t rabajo s e p ropone el e mpleo
de nomogramas para la localización de fallas en redes de
distribución u tilizando v alores d e d istancia a la f alla,
calculados mediante e l al goritmo d e T akagi y val ores d e
distancia a la falla, simulados para tres clases de redes de
igual topología pero de diferente co nstitución. Se muestra
cómo es posible mediante el e mpleo de estos nomogramas
obtener una estimación adecuada de la distancia a l a falla
a pesar de que el algoritmo de Takagi da valores erróneos
de localización.
Palabras clave – Localización de fallas, redes de distribución.
I. INTRODUCCIÓN
El empleo de técnicas de localización de fallas en centros de
gestión de la distribución es una necesidad impuesta en la
actualidad por el desarrollo de las clases de estrategias de
gestión utilizadas para el desarrollo de las llamadas redes
inteligentes o Smart Grid. Términos como continuidad del
servicio, calidad del servicio, fiabilidad etc juegan un
importante rol en la interpretación que se hace hoy del papel
de los sistemas de provisión de energía eléctrica lo que
implica mejorar la precisión de los métodos de localización de
fallas aplicados en redes de distribución ya que hace más fácil
el trabajo de detección y eliminación de las fallas al disminuir
el tiempo empleado para estas tareas. [1]
Numerosas técnicas han sido propuestas en la literatura a
lo largo de los años para resolver la problemática de la
localización de fallas en sistemas eléctricos de potencia.[2] [3-
8]La mayoría de estas técnicas se enfocan en la localización
de fallas para líneas de transmisión de energía. Esto se debe a
que los grandes niveles de potencia transportados por estas
líneas hacen que las fallas en las mimas tengan consecuencias
graves sobre la estabilidad del sistema y por lo general son
capaces de afectar a grandes grupos de consumidores. Por otro
lado las distancias a recorrer en líneas de transmisión para
Gómez González Abel, labora en Arteche Medición y Tecnología, S.A.
de C.V. Industria Mecánica México (e-mail: abgonzalez@arteche.com.mx)
Martínez del Sol Alexis labora en Arteche Centro de Tecnología, A.I.E
Derio Bidea, 2, 48100 Mungia, (Bizkaia), España (e-mail: amds@arteche.es)
Mejía Preciado Roberto, labora en Arteche Medición y Tecnología, S.A.
de C.V. Industria Mecánica 2173, Zapopan, 45132, Jal. México (e-mail:
rmejia@arteche.com.mx)
Mier García Jorge Javier, labora en Arteche Medición y Tecnología, S.A.
de C.V. Industria Mecánica (e-mail: jmier@arteche.com.mx)
81
Roberto Mejía Preciado Jorge Javier Mier García
localizar una falla son por lo general mucho mayores que para
líneas de distribución.[5, 9-10]
Además de esto, estas líneas tienen características cómo
homogeneidad, topología, etc. que las hacen susceptibles de
soluciones relativamente sencillas y económicas en términos
de recursos de medición y de capacidad de cómputo. En esta
clase entran los algoritmos de reactancia como el de Takagi[2]
con los cuales es posible obtener precisiones por debajo del
2% de error siempre que se realice una adecuada estimación
de los fasores de tensión y corriente involucrados en la falla.
Otros algoritmos de reactancia que emplean medición en
ambos extremos de la línea, sincronizadas o no, proveen
mayor nivel de precisión siempre que se cuente con la
posibilidad de utilizarlos.
Adicionalmente en este entorno se han empleado otras
clases de métodos tales como:
1. Redes neuronales artificiales.
2. Métodos de onda viajera.
3. Métodos de componentes de alta frecuencia.
4. Algoritmos Genéticos.
Para los sistemas de distribución sin embargo no se han
encontrado todavía soluciones suficientemente generales que
permitan de manera fácil e intuitiva obtener la localización de
la falla. Esto se debe a múltiples razones como: [9, 11-12]
1. La falta de homogeneidad de las líneas de
distribución.
2. La clase de topología de las redes de distribución que
provoca en muchos casos la existencia de lazos los
cuales crean conocidos problemas para localizadores
que empleen información de un solo terminal de la
línea.
3. La existencia de cargas variables entre los
localizadores de fallas y el lugar de ocurrencia de las
mismas.
4. El hecho de que los localizadores no estén ubicados
directamente sobre las líneas sino comúnmente sobre
alimentadores lo cual dificulta la evaluación de las
contribuciones de corriente de falla.
5. La relativamente alta influencia que ejerce la
resistencia de falla y la clase de aterramiento en este
tipo de redes sobre los valores de corrientes y tensión
de cortocircuito.
82

6. La imposibilidad de disponer en gran cantidad de
casos de valores de corriente directamente de la rama
fallada, etc.
7. En contraste con los sistemas de transmisión las líneas
de distribución alimentan cargas relativamente
pequeñas por lo que disponen de poca capacidad de
medición y/o instrumentación limitada en la mayoría
de los casos a medidores de tensión y corriente,
relevadores de protección de sobrecorriente de bajas
prestaciones y otros como restauradores,
seccionalizadores y fusibles conectados aguas abajo.
Este perfil de reactancia se obtuvo mediante la realización
de un estudio paramétrico de simulación de tres circuitos de
igual topología pero de diferentes valores de impedancia las
cuales se muestran en la TABLA 1.
En el estudio se simularon, utilizando SimPowerSystems
cortocircuitos, monofásicos, bifásicos, bifásicos a tierra y
trifásicos cada 100 metros a lo largo de cada una de las líneas.
En total durante el desarrollo del trabajo se generaron y
analizaron 18000 casos de estudio diferentes.

Esto ha provocado que se empleen técnicas disimiles de Xvs Distancia Linea Mixta CC3F
relativamente poca complejidad y efectividad tales como:

a. Restauración por conmutación.

b. Restauración usando operaciones de restauradores
c. Indicación usando fusibles y operación de
indicadores de paso de falla.
d. Revisión de cables caídos, llamadas de clientes,
mapas
e. Indicaciones de relevadores
f. Oliendo cables quemados

Sin embargo la desregulación de muchos mercados

energéticos que implica la competencia de las diferentes
empresas eléctricas por incrementar la disponibilidad del
Figura 2 Reactancia vs Distancia Línea Mixta
suministro hace que se le preste mayor atención a este
problema especialmente en la última década.
Como se puede ver en esta clase de circuitos existen
valores de impedancia varios para distancias iguales respecto
Por otro lado las características de los sistemas de
al origen del circuito de distribución bajo estudio, o valores de
distribución, mencionadas más arriba, dificultan la unicidad de
reactancia diferentes para la misma distancia. Esta es la razón
enfoque para la solución de este problema. Por esto es que se
por la cual se obtienen múltiples valores de distancia cuando
encuentran en la literatura gran cantidad de métodos diferentes
se emplean métodos de reactancia como el de Takagi para
para la localización de fallas en líneas de distribución. Estos
obtener la localización de las fallas en esta clase de circuitos.
pueden ser clasificados en los siguientes grupos:
Adicionalmente la forma de los diagramas de reactancia
cambia ante cambios de la carga, inserción en ciertos puntos
1. Algoritmos basados en el cálculo de la impedancia del
de la red unidades de generación distribuida o para diferentes
lazo de falla.
sistemas de aterramiento de las redes.
2. Aplicación de métodos de inteligencia artificial tales
como redes neuronales, sistemas neuro-fuzzy y
Un circuito topológicamente idéntico al anterior pero con
algoritmos genéticos.
diferencias en las características de los cables o líneas
3. Métodos estadísticos.
conectadas presentará una diferente gráfica de reactancia
4. Métodos híbridos.
contra distancia como se muestra en las Figura 3 y 4:

II. GRAFICAS DE REACTANCIA Título del gráfico

Para una red radial como la presentada en la FIGURA 1 la
distribución de reactancia vista desde el punto 0 para
cortocircuitos a lo largo de toda la línea es la mostrada en la
FIGURA 2.

Figura 1 Red de Distribución Radial Figura 3 Reactancia vs Distancia Línea Aérea

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 83

Título del gráfico

Distancia Real
Takagi

Figura 4 Reactancia vs Distancia Línea Subterránea Figura 6 Distancia a la Falla y Takagi CC3F. Línea Mixta

Las razones para las diferencias entre las formas de las

curvas de reactancia contra distancia observadas en las figuras
anteriores pueden ser atribuidas a los datos de impedancia y
tamaño de las líneas que se muestran en la tabla siguiente en la
que se muestran las resistencias y reactancias por kilómetro de
las líneas y el tamaño de los tramos en kilómetros.

Tabla 1 Tamaño de tramo e Impedancia por kilómetro Distancia Real

Línea Aérea Línea Mixta Línea Subterránea Takagi

De A R X d R X d R X d
0 1 10,00 0,20 0,36 6,00 0,12 0,40 2,00 0,06 0,11
1 1A 4,50 0,42 0,39 2,00 0,13 0,11 0,30 0,13 0,11
1A 1A1 1,00 0,42 0,39 2,00 0,21 0,12 0,20 0,21 0,12
1A 1A2 2,00 0,61 0,39 3,00 0,61 0,39 0,30 0,13 0,11
Figura 7 Distancia a la Falla y Takagi. CC3F Línea
Subterránea
1 2 5,00 0,20 0,36 4,50 0,08 0,10 2,00 0,08 0,10
2 2A 3,50 0,61 0,39 3,50 0,21 0,12 0,40 0,21 0,12 Los errores cometidos por la aplicación del método de
2 3 8,00 0,12 0,31 6,00 0,20 0,36 2,00 0,08 0,10 Takagi a estos casos se muestran en la Figura 8
3 3A 0,70 0,21 0,12 1,00 0,21 0,12 0,20 0,21 0,12
3A 3B 3,00 0,61 0,39 3,00 0,21 0,12 0,30 0,21 0,12
3A 3C 1,30 0,35 0,38 1,30 0,25 0,12 0,20 0,25 0,12 Error vs Distancia Linea Aerea
Error vs Distancia Línea Mixta
3 4 5,00 0,61 0,39 4,50 0,13 0,11 2,00 0,13 0,11
Error vs Distancia Línea Subterranea

III. APLICACIÓN DEL ALGORITMO DE TAKAGI

La aplicación del algoritmo de Takagi permite obtener
valores de distancias a la falla que se muestran en las Figuras
5, 6 y 7:

Figura 8 Error vs. Distancia. Algoritmo de Takagi

En general se puede ver con la aplicación del algoritmo de

Takagi a este tipo de líneas se obtienen altos niveles de error
el cual de manera general incrementa al incrementar la
Distancia Real
distancia a la cual sucede la falla. Una mayor homogeneidad
Takagi de la línea trae como consecuencia una curva de error más
suave lo cual indica que una mejor parametrización del
algoritmo pudiera permitir la obtención de niveles de exactitud
mayores.

Es posible utilizar los resultados incorrectos del algoritmo

Figura 5 Distancia a la Falla y Takagi CC3F. Línea Aérea
de Takagi incluso cuando los niveles de error son altos, para la
84
localización de fallas siempre y cuando los niveles de
precisión de la aplicación del algoritmo se mantengan
constantes. Esto se verá en la próxima sección.
IV. USO DE NOMOGRAMAS
Aplicando el algoritmo de Takagi, para obtener la distancia
a la falla en los circuitos de distribución mostrados
anteriormente es posible obtener las gráficas de distancia
calculada contra distancia real a la falla mostradas en las
Figuras 9, 10 y 11.
Distancia Calculada vs Distancia Real CC3F
Linea Aérea
Figura 9 Nomograma Línea Aérea
Distancia Real vs Calculada Línea Mixta CC3F
Figura 10 Nomograma Línea Mixta
Distancia Calculada vs Distancia Real CC
Trifásico Línea Subterránea
Figura 11 Nomograma Línea Mixta
El empleo de estas gráficas permite, a pesar de los errores
propios del algoritmo de Takagi vistos en el epígrafe anterior,
obtener estimaciones adecuadas de la distancia a la falla.
V. CONCLUSIONES
La aplicabilidad de los algoritmos de reactancia a redes de
distribución se restringe a sistemas de baja complejidad.
La múltiple localización o múltiple estimación de lugar de
falla es un problema de los métodos de impedancia o distancia
con medición desde una sola punta aplicados a redes de
distribución y no puede ser eliminado a no ser que se disponga
de información adicional.
El empleo de nomogramas permite obtener localización de
fallas en redes de distribución con niveles de exactitud
aceptables a pesar de los errores de los algoritmos de
reactancia a partir de los cuales se generan dichos gráficos
siempre y cuando la precisión del .del algoritmo de reactancia
se mantenga constante.
VI. BIBLIOGRAFÍA
[1] Y. Zhang and M. Prica, "Toward Smarter Current Relays for Power
Grids," in IEEE PES General Meetin, Montreal, Canada, 2006, p. 8.
[2] T. Takagi, et al., "Development of a New Type Fault Locator Using
the One Terminal Fault-Voltage and Current Data," IEEE
Proceedings on Power apparatus and Systems, vol. PAS 101, p. 7,
August, 1982 1982.
[3] D. Novosel, et al., "Unsynchronized two-terminal fault location
estimation," IEEE Trans. on Power Delivery, pp. 130-138, 1996.
[4] M. M. Saha, et al., "Review of fault location techniques for
distribution systems," presented at the Power Systems and
Communications Infrastructures for the Future, Beijing, 2002.
[5] T. Bouthiba. FAULT LOCATION IN EHV TRANSMISSION
LINES USING ARTIFICIAL NEURAL NETWORKS.
[6] A. S. Bretas, et al., "A BP neural network based technique for HIF
detection and location on distribution systems with distributed
generation," presented at the Proceedings of the 2006 international
conference on Intelligent computing: Part II, Kunming, China, 2006.
[7] M. Kezunović, et al., "A Novel Method for Transmission Network
Fault Location Using Genetic Algorithms and Sparse Field
Recordings," in IEEE PES Summer Meeting, ed, 2002.
[8] S. Vasilic and M. Kezunovic, "New Design of a Neural Network
Algorithm for Detecting and Classifying Transmission Line Faults,"
in IEEE PES Transmission and Distribution Conf, ed, 2001.
[9] M. M. Saha, et al., Fault Location on Power Networks, 1st ed.
London: Springer, 2010.
[10] A. Gopalakrishnan, et al., "Fault Location Using Distributed
Parameter Transmission Line Model," IEEE Trans. on Power
Delivery, pp. 1169-1174, 2000.
[11] IEEE, "C37.114-2004 - IEEE Guide for Determining Fault Location
on AC Transmission and Distribution Lines," ed, 2004.
[12] J. J. Mora Flores, "Localizacion de faltas en sistemas de distribución
de energía eléctrica usando metodos basados en el modelo y métodos
basados en el conocimiento," PhD Doctoral, Escola Politécnica
Superior, Universitat de Girona, Girona, 2006.
VII. BIOGRAFÍAS
Abel González Gómez, nació en Santa Clara, Cuba.
Ingeniero Electricista, egresado de la Universidad
Central de Las Villas, Cuba en 1996. Master en
Sistemas Eléctricos en 2000 por la Universidad
Central de Las Villas. Laboró como profesor de la
Facultad de Ingeniería Eléctrica de dicha Universidad
y como Ingeniero de Control en la Empresa de
Telecomunicaciones de Cuba.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Desde el año 2010 se desempeña como Ingeniero de Diseño en la empresa
Arteche Medición y Tecnología S.A. de C.V. Su área de investigación está
relacionada con el análisis de la operación, control y protección de sistemas
eléctricos de potencia y el desarrollo de software de aplicación
Alexis Martínez del Sol nació en Cienfuegos Cuba
en 1964. Recibió el grado de Doctor en Ciencias
Técnicas en la Universidad Central de Las Villas,
Cuba en 1997. En el período de 1987 -1999 trabajó
en la Facultad de Ingeniería Eléctrica de la
Universidad Central de Las Villas donde fue profesor
Asistente y Jefe del departamento de Electroenergé-
tica.
De 1999- 200 fue Profesor Investigador Titular C del Departamento de
Ingeniería Mecánica Eléctrica de la Universidad de Guadalajara. A partir del
año 2005 se vinculo al desarrollo de investigaciones científicas en la empresa
Arteche Medición y Tecnología S.A. de C.V. y durante los años 2007-2010
fue su Director General. Actualmente es el Director de Protección, Control y
Medición del Grupo Arteche. Su área de investigación está relacionada con el
análisis de la operación, control y protección de sistemas eléctricos de
potencia y el desarrollo de software de aplicación. Fue Miembro del Sistema
Nacional de Investigadores (Nivel I) del 2004-2007
85
Roberto M ejía P reciado nació en Guadalajara,
México en 1961. Recibió el grado de ingeniero
electrónico en la Universidad ITESO de Guadalajara
en 1982. Entre 1982 y 1989 trabajó en diferentes
empresas relacionadas al ramo de computación. A
partir de 1989 se ha dedicado al desarrollo de equipos
de medición y control para subestaciones.
Desde el año de 2005 labora en Arteche Medición y Tecnología, S.A. de C.V.
y es integrante del grupo de investigación, desarrollo e innovación de la
empresa. Su área de investigación está relacionada con el diseño electrónico,
análisis de la operación, control y protección de sistemas eléctricos de
potencia y el desarrollo de software de aplicación
Jorge Javier Mier G arcía nació en Báez, Placetas,
Cuba. Recibió el grado de Doctor en Ciencias
Técnicas en la Universidad Central de Las Villas,
Cuba en 2002. En el período de 1988 – 2005 trabajó
en la Facultad de Ingeniería Eléctrica de la
Universidad Central de Las Villas donde fue profesor
Asistente y Miembro del Tribunal Nacional de
Grados Científico.
Desde el año 2006 labora en la empresa Arteche, Medición y Tecnología S.A.
de C.V. y actualmente es Gerente de Diseño. Su área de investigación esta
relacionada con el análisis de la operación, control y protección de sistemas
eléctricos de potencia y el desarrollo de software de aplicación.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Real-Time Multiterminal Fault Location
System for Transmission Networks
Yanfeng Gong, Mangapathirao Mynam, Armando Guzmán, and Gabriel Benmouyal,
Schweitzer Engineering Laboratories, Inc.
Boris Shulim, Orange and Rockland Utilities, Inc.
Abstract—Fault location information is critical for operating
and maintaining transmission networks. Some of the challenges
in calculating accurate fault location include fault resistance,
zero-sequence mutual coupling, load, system nonhomogeneity,
and transmission lines composed of multiple sections with
considerably different characteristics. This paper presents a fully
automated real-time fault location system that provides accurate
fault location information for multiple transmission lines and
makes the results available to system operators and maintenance
personnel. The system is capable of the following:
• Retrieves all triggered event reports from relays and
digital fault recorders (DFRs) automatically. With these
reports, the system identifies the faulted transmission
line(s) within the transmission network.
• Determines the fault type and calculates the fault
location and fault resistance using multiterminal fault
location methods for different types of power lines,
including overhead lines, underground cables, and
composite lines that include both overhead line and
underground cable sections.
• Uses fault location methods that are accurate under fault
resistance, parallel-line mutual coupling, system
nonhomogeneity, and load.
• Supports event reports with a fixed sampling rate or a
variable sampling rate that is a multiple of the power
system operating frequency.
• Supports event reports generated by relays and DFRs
connected to an IRIG-B time source.
This paper compares fault location results calculated from
field event reports and reports that are obtained from models of
real power systems with actual fault locations.
I. INTRODUCTION
Fault location in protective relays has been available for
over 20 years. These relays use impedance-based fault
location algorithms, typically from one terminal of the
transmission line [1] [2]. While these relays have been very
useful in locating the majority of faults, the following
conditions can affect their fault location accuracy:
• High-resistance faults
• Heavy load conditions
• Zero-sequence mutual coupling
87
Two-end fault location methods minimize these errors.
Reference [3] describes a two-end method based on negative-
sequence quantities that can be implemented within line
protection relays using the relay programming abilities.
Reference [4] describes an implementation of this algorithm
using two line relays with relay-to-relay communication to
calculate the fault location in real time.
Variable fault resistance introduces errors in fault location
estimation; accurate time stamping of the measurements aids
in minimizing these errors when using two-end methods.
Synchrophasor technology is promising; however,
available message rates make using this technology
challenging, especially in applications with fast fault-clearing
times.
Alternative traveling wave fault location technology is
available and provides good results, but deployment has been
limited to extra-high-voltage applications because of its high
cost. One of the objectives of the system proposed in this
paper is to provide an economical solution using relays that
perform line protection functions.
While present two-end fault location methods provide good
results in applications where the X/R ratio of the line is
constant, they are not suitable for lines with multiple sections.
These sections have different impedance characteristics,
especially for applications that combine overhead lines and
underground cables. This paper presents a method that is
based on the negative-sequence voltage profile along the line
and is suitable for composite lines.
We use field and laboratory cases where the fault location
is known to determine the accuracy of the proposed method.
The field case shows that the fault location estimation is off by
475 feet in a 26.3-mile composite line application.
We introduce a real-time fault location system that uses
line protection relays and displays fault location information at
the control center within 1 minute after the fault occurrence
when using Ethernet-based communication. The fault location
system automatically retrieves event reports from relays after
a fault occurs. The system accommodates event reports with
both fixed and variable sampling rates.
Orange and Rockland Utilities, Inc. (ORU) plans to deploy
this system in its transmission network and communicate the
fault location results to the ORU Energy Control Center.
88

II. EXISTING MULTITERMINAL FAULT LOCATION METHODS IX IY

VX VTAP VY
Section X Section Y
Two-end-based algorithms offer significant advantages in
fault location accuracy compared with single-end fault
location algorithms. These algorithms are not affected by fault
Relay X Relay Y
resistance, load, and zero-sequence mutual coupling. Fig. 1 Section Z
shows the circuit model of a two-bus system with a fault at
Relay Z
m distance from Terminal X.
IZ

VZ

Fig. 2. Three-terminal line with a fault on Section X.

Fig. 1. Circuit model of a two-bus system with a fault at m distance from
Terminal X.
The two-end impedance-based fault location algorithms
discussed above assume that power lines are homogeneous
Some two-end impedance-based fault location algorithms with a constant X/R ratio. This assumption introduces errors
use voltage and current measurements from both terminals of for nonhomogeneous lines with overhead sections having
the transmission line to estimate the distance to the fault, m, in different line impedances or a combination of underground
per unit, according to (1). cable and overhead line sections. This paper describes a new
VX − VY + I Y • ZL fault location method for homogeneous and nonhomogeneous
m= (1)
( IX + I Y ) • ZL lines.

VX and IX correspond to the voltage and current phasors at III. CONSIDERATIONS FOR VARIABLE FAULT RESISTANCE
Terminal X. VY and IY correspond to the phasors at
Variable fault resistance affects the voltages and currents
Terminal Y. These phasors can be obtained from symmetrical
that the relays at each terminal measure. Fig. 3 shows the sum
or Clarke components or phase quantities. ZL is the line
of the local and remote residual currents (total residual
impedance. We can compensate ZL in (1) according to the
current) and the estimated fault resistance for a B-phase-to-
distributed parameter line model to accommodate long
ground fault. The total residual current is approximately equal
transmission lines [1].
to the total fault current at the fault location. The fault
Alignment of local and remote measurements is critical for
occurred on a 400 kV transmission line with a line length of
obtaining accurate fault location using two-end-based
225 kilometers; wildfires close to the transmission line caused
algorithms for faults with variable resistance. Methods based
the fault condition. Observe that the fault current starts with a
on (1) are sensitive to the errors in measurement alignment.
peak value lower than 828 A that increases to a peak value of
Reference [3] describes a two-end method that uses only
3,860 A in 4 cycles at a rate of 758 A per cycle. The
negative-sequence quantities and solves a second-order
corresponding resistance that we calculate according to (2)
polynomial equation to estimate the distance to the fault.
changes at the beginning of the fault and settles to
These quantities include local current and voltage
approximately 3 Ω after 4 cycles [1].
measurements, as well as remote source impedance and
current magnitudes. Reference [5] presents a method that uses ⎧⎪ V I + VYB IXBk0 − ZL1I YBk0 IXBk0 ⎫⎪
an iterative process to estimate a factor that compensates the R F = real ⎨ XB YBk0 ⎬ (2)
error in alignment and then solves the distance to the fault. ⎩⎪ ( IXBk0 + IYBk0 )2 ⎭⎪
Reference [6] uses time-synchronized phasor where:
measurements for multiterminal-based fault location.
I XBk0 = IXB + k 0 I XR
However, most phasor measurement units (PMUs) offer a
maximum data transfer rate of 60 messages per second. Fast I YBk0 = IYB + k 0 I YR
clearing times reduce the available fault data, making the ZL0 − ZL1
k0 =
synchrophasor-based fault location calculation challenging. 3ZL1
We can use two-end methods to calculate fault location on
VXB is the B-phase voltage at Terminal X.
three-terminal lines by reducing the three-terminal network to
a two-terminal network [3]. Fig. 2 shows a typical VYB is the B-phase voltage at Terminal Y.
three-terminal line. For a fault on Section X, the voltages at IXB is the B-phase current at Terminal X.
the tap (VTAP) calculated from Terminal Y and Terminal Z are IYB is the B-phase current at Terminal Y.
equal, and the voltage calculated from Terminal X is different IXR is the residual current at Terminal X.
from the voltages calculated from the other two terminals. IYR is the residual current at Terminal Y.
Based on this concept, Section X can be identified. ZL1 is the positive-sequence line impedance.
ZL0 is the zero-sequence line impedance.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
A
Ω
Fig. 3. Fault current and resistance for a phase-to-ground fault on a
225-kilometer, 400 kV line.
Without proper event report alignment, the rapid change of
fault resistance in Fig. 3 at the start of the fault can cause
additional errors when estimating fault location in
transmission line applications with fast fault-clearing times.
Accurate time stamping of the measurements improves event
report alignment and minimizes errors caused by variable fault
resistance.
IV. NEW FAULT LOCATION METHOD
Most faults on power transmission lines are unbalanced
faults. The common characteristic of unbalanced faults is that
there are negative-sequence currents and voltages available for
the fault location calculation. This method uses an existing
algorithm to distinguish between unbalanced and balanced
faults [7]. Fig. 4b illustrates the negative-sequence network of
the transmission line with the three line sections shown in
Fig. 4a. The fault is on Section 2; m is the distance to the fault
in per unit of the section length, as measured from Junction D.
Fig. 4. (a) Transmission line with three line sections. (b) Negative-sequence
network of the transmission line for a fault on Section 2.
89
The estimated fault location based on the negative-
sequence network is not affected by zero-sequence mutual
coupling and errors in zero-sequence line impedance.
Accurate zero-sequence line impedances are often difficult to
obtain. For underground cables in particular, the actual zero-
sequence impedance depends on the bonding method of the
shield and the cable configuration [8].
Another benefit of using negative-sequence quantities is
that the line-charging currents have negligible impact on the
accuracy of the fault location estimation because the negative-
sequence voltage is significantly lower than the positive-
sequence voltage along the transmission line.
The proposed method uses the profile of the estimated
negative-sequence voltage magnitude along the transmission
line to determine the fault location. This method identifies the
faulted line section and estimates the distance to the fault in
this section.
A. Identification of the Faulted Line Section
We estimate negative-sequence voltage at each junction
between line sections twice, starting from each line terminal,
in order to construct two negative-sequence voltage profiles.
Therefore, for each line section, there are two calculated
voltages for the left junction and two calculated voltages for
the right junction. The intersection point of the two voltage
magnitude profiles calculated from each terminal is the
negative-sequence voltage magnitude at the fault location.
For a power transmission line that consists of
N homogeneous line sections, the negative-sequence voltages
at the junctions of line section k can be calculated using (3)
through (6). The left terminal of the line is denoted as
Terminal X, and the right terminal of the line is denoted as
Terminal Y.
k −1
V2k _ L _ X = V2meas _ X − ∑ i =1 Z1i • I2meas _ X (3)
V2k _ R _ X = V2meas_ X − ∑ i =1 Z1i • I2meas _ X
k
(4)
V2k _ L _ Y = V2meas _ Y − ∑ i = k Z1i • I2meas _ Y
N
(5)
V2k _ R _ Y = V2meas _ Y − ∑ i = k +1 Z1i • I2meas _ Y
N
(6)
where:
V2meas_X, I2meas_X, V2meas_Y, and I2meas_Y are the negative-
sequence voltage and current measurements from the
Terminal X and Terminal Y relays, respectively.
Z1i is the positive-sequence impedance of line section i,
which is the same as the negative-sequence impedance.
V2k_L_X is the negative-sequence voltage at the left
junction of line section k that is calculated from
Terminal X.
V2k_R_X is the negative-sequence voltage at the right
junction of line section k that is calculated from
Terminal X.
V2k_L_Y is the negative-sequence voltage at the left
junction of line section k that is calculated from
Terminal Y.
90

V2k_R_Y is the negative-sequence voltage at the right C. Fault Location for Three-Phase Faults
junction of line section k that is calculated from For three-phase faults, we calculate the positive-sequence
Terminal Y. impedance to the fault, Ztotal, from either terminal using
Fig. 5 illustrates the negative-sequence voltage magnitude positive-sequence voltage, V1, and current, I1, according
profiles from Terminal X and Terminal Y of the transmission to (10). The algorithm identifies the faulted line section, k,
line shown in Fig. 4. Because the two voltage profiles using (11). Equation (12) determines the total distance to the
intersect at the fault location, the negative-sequence voltages fault, L, where Lengthi is the length of line section i.
at the junctions of the faulted line section meet the two criteria
V1
stated in (7) and (8). Ztotal = (10)
I1
V2k _ L _ X ≤ V2k _ L _ Y (7) k −1
∑ i =1 Z1i ≤ Ztotal < ∑ i =1 Z1i
k
(11)
V2k _ R _ Y < V2k _ R _ X (8) k −1
k −1 Ztotal − ∑ i =1 Z1i
L = ∑ i =1 Length i + • Length k (12)
Z1k

V. TEST RESULTS AND FIELD EVENT CASE STUDY

This section discusses the test results and accuracy of the
fault location estimation on a 26.3-mile, 230 kV transmission
Magnitude

line. As shown in Fig. 7, this composite, nonhomogeneous

line has two overhead sections and two underground cable
sections. Table I lists the parameters of each line section. We
used fault event reports from simulations and a field event to
demonstrate the accuracy of the proposed fault location
algorithm.
Fig. 5. Negative-sequence voltage magnitude profile along the transmission
line for a fault on Section 2.

B. Estimation of the Distance to the Fault on the Faulted

Line Section for Unbalanced Faults
After the algorithm identifies the faulted line section, the Fig. 7. Nonhomogeneous 230 kV transmission line.
algorithm estimates the distance to the fault on the
homogeneous line section. Fig. 6 shows the equivalent circuit TABLE I
TRANSMISSION LINE SECTION PARAMETERS
of the faulted line section. The equivalent negative-sequence
voltage sources, V2k_L_X and V2k_R_Y, are the voltages Line Positive-Sequence Zero-Sequence Length
calculated using (3) and (6) for the faulted line section. Section Impedance (Ω) Impedance (Ω) (miles)
Section 1 0.9522 + j10.5536 8.4111 + j38.4107 19.0
Section 2 0.0291 + j0.4973 0.4840 + j2.6186 2.9
Section 3 0.1957 + j1.5024 1.5235 + j6.7395 4.0
Section 4 0.0026 + j0.0635 0.0661 + j0.2883 0.4

A. Test System Setup and Simulation Results

Fig. 6. Equivalent negative-sequence network for a fault on the faulted
line section.
The voltage magnitudes at the fault location calculated
from the two junctions should be equal to each other,
according to (9). The algorithm solves (9) for the
distance-to-fault value, m. This approach minimizes errors
because of data misalignment.
V2k _ L _ X − m • Z1k • I2X = V2k _ R _ Y − (1 − m ) • Z1k • I2Y
We modeled the power system that includes the
transmission line shown in Fig. 7 in the Real Time Digital
Simulator (RTDS®) that runs an Electromagnetic Transients
Program (EMTP) simulation. Two digital line protection
relays measure voltages and currents at both line terminals via
the RTDS analog interface. These relays are configured to
protect the transmission line using a pilot protection scheme.
The relays use a demodulated IRIG-B signal as a time source.
(9)
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 91

The breaker statuses and trip signals are exchanged among 0.025
the RTDS and relays via the digital I/O interface, as Fig. 8
illustrates. The system setup emulates the real-time, closed- LG Fault
0.02
loop controlled power system. LL Fault

Absolute Error in Miles

LLG Fault
0.015
3P Fault

0.01

0.005

0
5.7 13.3 20.16

Fault Location in Miles

Fig. 9. Fault location errors at different locations with RF = 0.

Fig. 8. Closed-loop test system includes RTDS and the two line relays.

We simulated all fault types (line-to-ground, line-to-line,

line-to-line-to-ground, and three-phase) at 5.7, 13.3, and
20.16 miles from Terminal X to verify the accuracy of the

Absolute Error in Miles

proposed algorithm. The first two locations are on the first
overhead section (Section 1), and the third location is on the
first underground cable section (Section 2). For each fault, we
collected three types of event reports from both terminals:
relay-generated event reports with fixed sampling rates of
8,000 samples per second, relay-generated event reports with
operating frequency-based sampling rates of 8 samples per
power system cycle, and RTDS-generated event reports with
fixed sampling rates of 4,000 samples per second. Fault
location estimation using RTDS-generated event reports
illustrates the theoretical performance of the algorithm. We
use relay event reports to estimate the fault location to include
Fig. 10. Fault location errors with different fault resistances.
the effects of relay measurement errors.
Fig. 9 shows the absolute error in fault location estimation
using RTDS-generated event reports for each fault type with
Absolute Error in Miles

zero fault resistance at different locations. The largest error is

0.02 miles (106 feet), which is 0.07 percent of the total line
length. Fig. 10 shows the absolute error in fault location
estimation using the RTDS-generated event reports for line-to-
ground faults with fault resistance varying from 0 to 100 Ω at
the fault locations discussed above. The results show that the
largest error is 0.04 miles (212 feet), and the fault resistance
has minimal impact on the fault location estimation.
Fig. 11 shows the performance of the proposed method
compared with the single-end fault location method and the
Fig. 11. Fault location error of different methods using relay event reports
existing two-end fault location method using relay-generated with fixed sampling rates for line-to-ground faults at 20.16 miles from
fixed sampling event reports. The single-end fault location Terminal X.
method and existing two-end fault location method use only
For the single-end method, relays at both terminals
the total impedance of the composite line to estimate fault
reported a fault location value. We selected the value with the
location. We obtained these event reports from simulated line-
lowest error for comparison. Fig. 11 shows that the proposed
to-ground faults with fault resistance varying from 0 to 100 Ω
method consistently provides better fault location estimation
at 20.16 miles from Terminal X.
than the traditional two-terminal and single-end methods on
composite transmission lines.
92

Fig. 12 compares fault location errors using event reports

from relays with fixed and power system operating frequency-
dependent sampling rates. The results show that these two

A
sampling methods have similar performance.
Absolute Error in Miles

kV
Fig. 14. Current and voltage measurements of the relay at Terminal Y.

Fig. 15 shows the negative-sequence voltage profiles

calculated from both line terminals. The intersection of the
two voltage profiles indicates the location of the fault. The
Fig. 12. Fault location errors using fixed sampling and operating frequency- new method provides a more accurate fault location than the
based sampling event reports for line-to-ground faults at 20.16 miles. existing methods (see Table II).
B. Field Event Case Study 70
The transmission line described in Section V had an actual 60
Negative-Sequence Voltage Voltage Profile From
single-line-to-ground fault. Fig. 13 and Fig. 14 show the relay
Terminal X
recorded current and voltage measurements for the B-phase- Voltage Profile From
Magnitude in kV

to-ground fault. The proposed method estimated the fault 40 Terminal Y

location at 18.91 miles from Terminal X. The actual fault
location from line inspection was at the junction point
between the first overhead line section and the first cable 20
section, which is 19.0 miles from Terminal Y.
Table II lists the absolute errors in fault location estimation
reported by the proposed method, line relays at both ends, and 0
Miles 19.0 21.9 Terminal Y
the traditional two-end method. Terminal X (0.0)
(26.3)

TABLE II Fig. 15. Negative-sequence voltage profiles for the B-phase-to-ground fault.
ABSOLUTE ERRORS OF THE ESTIMATED FAULT LOCATION IN MILES

New Traditional Two- Relay at Left Relay at Right VI. AUTOMATED FAULT LOCATION SYSTEM
Method Terminal Method Terminal Terminal
Most of the existing multiterminal fault location methods
0.09 2.18 3.91 4.25 are mainly for post-event analysis. To perform this analysis,
we need to collect event reports from all line terminals, align
these data, and estimate the fault location, which is time-
consuming and inefficient. The automated fault location
system (AFLS) that we present in this paper can monitor
A

hundreds of transmission lines in the utility and provide fault

location information in real time without human intervention.
The AFLS includes protective relays connected to an IRIG-B
time source and fault location software running on a computer.
This software automatically retrieves event reports from the
relays, calculates the fault location, and presents the results to
kV

the user. Fig. 16 shows the AFLS architecture to monitor three

lines of a power system. This architecture uses Ethernet-based
communication between the protective relays and the
computer running the fault location software. The software
accommodates serial and Ethernet communications.
Fig. 13. Current and voltage measurements of the relay at Terminal X.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Fig. 16. AFLS monitoring three transmission lines.
A. System Configuration
The initial system configuration requires specifying the
number of terminals of the monitored line, the number of
sections, the section impedances, the communications
parameters of the relays, and the association of the relay with
the corresponding line terminal. After the system is
configured, it does not require additional user intervention.
The software also accommodates line terminals with dual
breakers and compensates for shunt reactors based on their
impedances.
B. Fault Record Retrieval and Archiving
The user can configure the fault location software to
retrieve event reports from the protective relays via event-
driven or polling mechanisms. When using the event-driven
mechanism, the software initiates the retrieval process
immediately upon receiving a message from the relay every
time a new event is triggered. When using the polling
mechanism, the software polls the relays for new events
periodically at user-defined intervals and retrieves the new
event reports. Compared to the polling mechanism, the event-
driven mechanism offers lower latency and minimum
communications traffic. The software stores the retrieved
event reports with filenames consisting of the device name
and event trigger time stamp.
C. Fault Location Calculation and Results Display
After successfully retrieving all of the event reports for a
particular fault, the fault location software automatically
indentifies the faulted line, estimates the fault location, and
displays the results. Fig. 17 shows the display of the fault
location information. This display can be available at the
control center. The reported fault location information
includes the line name, reference terminal, distance to the
fault, fault type, and time of the fault. The software can also
send the fault location information to maintenance personnel
via email.
93
Fig. 17. Fault location information display.
D. System Latency
The AFLS latency depends on the event retrieval time and
fault location computation time. The event retrieval time
depends on the communications medium between the
computer and the relays. For serial communication, the event
retrieval time is a function of the data transfer rate and the file
size. Typically, event retrieval times are in the order of 1 to
3 minutes. For Ethernet-based communication, event retrieval
times are in the order of seconds. The computer performs fault
location calculations in less than 1 second. For systems with
Ethernet communication, the overall system latency is less
than 1 minute.
VII. AFLS DEPLOYMENT
Orange and Rockland Utilities, Inc. (ORU) proposed the
installation of fault-locating equipment at transmission line
terminals in order to implement the two-end fault location
system as part of the ORU Smart Grid Project. The majority
of ORU 345/138 kV transmission lines are overhead
conductors, and some of them are underground cables. These
transmission lines are relatively short, approximately 9 to
10 miles; however, the lines are not easily accessible for
inspection during faults due to difficult terrain. Therefore, a
tool to precisely calculate the fault location is important for
quick restoration of the faulted transmission line.
The algorithm presented in this paper for fault location on
transmission lines using a two-ended technique, as per ORU
design requirements, was tested, and the results are within
1.5 percent accuracy. Based on successful testing results,
ORU allocated funding to implement this fault location
technique on critical transmission lines.
ORU is planning to use relays with fixed sampling rates
and accurate time stamping for fault locating. These relays can
have up to six sets of three-phase current and two sets of
three-phase voltage inputs. Thus, the relays can monitor
multiple lines within a substation.
ORU is in the process of implementing its first double-
ended fault location system. This system will include relays at
each terminal of the transmission line and will communicate
with a workstation located at the ORU Energy Control Center
via serial communication over optical fiber. The workstation
will process the fault data recorded by relays. The results,
including calculated fault location, will be emailed to the
system operator.
94
VIII. CONCLUSION
This paper presents an automated fault location system for
transmission networks. The system uses a new multi-end fault
location algorithm that is suitable for composite transmission
lines. The fault location algorithm and the automated system
have the following characteristics:
• The algorithm uses the negative-sequence voltage
profile along the transmission line to identify the
faulted section, makes a network reduction, and
estimates the fault location.
• A field case validates the accuracy of the algorithm for
a phase-to-ground fault on a 26.3-mile 230 kV
composite line. In this case, the fault location
estimation is off by 475 feet.
• After the user configures the system, the system
reports fault location information in less than 1 minute
when using Ethernet-based communication without
human intervention.
• The system works with existing protective relays to
provide an economical real-time fault location
solution.
• Event reports with accurate time stamps improve
multiterminal fault location accuracy, particularly for
faults with varying fault resistance.
IX. ACKNOWLEDGMENT
The authors would like to thank Tariq Rahman of San
Diego Gas and Electric for providing the line parameters and
fault event reports of the field case presented in this paper.
X. REFERENCES
[1] E. O. Schweitzer, III, “Evaluation and Development of Transmission
Line Fault-Locating Techniques Which Use Sinusoidal Steady-State
Information,” proceedings of the 9th Annual Western Protective Relay
Conference, Spokane, WA, October 1982.
[2] T. Takagi, Y. Yamakoshi, M. Yamaura, R. Kondow, and
T. Matsushima, “Development of a New Type Fault Locator Using the
One-Terminal Voltage and Current Data,” IEEE Transactions on Power
Apparatus and Systems, Vol. PAS-101, Issue 8, August 1982,
pp. 2892–2898.
[3] D. A. Tziouvaras, J. Roberts, and G. Benmouyal, “New Multi-Ended
Fault Location Design for Two- or Three-Terminal Lines,” proceedings
of the 7th International Conference on Developments in Power System
Protection, Amsterdam, Netherlands, April 2001.
[4] K. Zimmerman and D. Costello, “Impedance-Based Fault Location
Experience, ” proceedings of the 31st Annual Western Protective Relay
Conference, Spokane, WA, October 2004.
[5] D. Novosel, D. G. Hart, E. Udren, and J. Garitty, “Unsynchronized
Two-Terminal Fault Location Estimation,” IEEE Transactions on
Power Delivery, Vol. 11, Issue 1, January 1996.
[6] S. Lopez, J. Gomez, R. Cimadevilla, and O. Bolado, “Synchrophasor
Applications of the National Electric System Operator of Spain,”
proceedings of the 34th Annual Western Protective Relay Conference,
Spokane, WA, October 2007.
[7] A. Guzmán, V. Mynam, and G. Zweigle, “Backup Transmission Line
Protection for Ground Faults and Power Swing Detection Using
Synchrophasors,” proceedings of the 34th Annual Western Protective
Relay Conference, Spokane, WA, October 2007.
[8] J. Vargas, A. Guzmán, and J. Robles, “Underground/Submarine Cable
Protection Using a Negative-Sequence Directional Comparison
Scheme,” proceedings of the 26th Annual Western Protective Relay
Conference, Spokane, WA, October 1999.
XI. BIOGRAPHIES
Yanfeng Gong received his BSEE from Wuhan University, China, in 1998,
his MSEE from Michigan Technological University in 2002, and his PhD in
electrical engineering from Mississippi State University in 2005. He is
currently working as a research engineer at Schweitzer Engineering
Laboratories, Inc., in Pullman, Washington. He is a member of IEEE.
Mangapathirao Mynam received his MSEE from the University of Idaho in
2003 and his BE in electrical and electronics engineering from Andhra
University College of Engineering, India, in 2000. He joined Schweitzer
Engineering Laboratories, Inc. (SEL) in 2003 as an associate protection
engineer in the engineering services division. He is presently working as a
lead research engineer in SEL research and development. He was selected to
participate in the U.S. National Academy of Engineering (NAE) 15th Annual
U.S. Frontiers of Engineering Symposium. He is a member of IEEE.
Armando Guzmán received his BSEE with honors from Guadalajara
Autonomous University (UAG), Mexico. He received a diploma in fiber-
optics engineering from Monterrey Institute of Technology and Advanced
Studies (ITESM), Mexico, and his MSEE from the University of Idaho, USA.
He served as regional supervisor of the Protection Department in the Western
Transmission Region of the Federal Electricity Commission (the Mexican
electrical utility company) in Guadalajara, Mexico, for 13 years. He lectured
at UAG and the University of Idaho in power system protection and power
system stability. Since 1993, he has been with Schweitzer Engineering
Laboratories, Inc., in Pullman, Washington, where he is a research
engineering manager. He holds numerous patents in power system protection
and metering. He is a senior member of IEEE.
Gabriel Benmouyal, P.E., received his BASc in electrical engineering and
his MASc in control engineering from Ecole Polytechnique, Université de
Montréal, Canada, in 1968 and 1970. In 1969, he joined Hydro-Québec as an
instrumentation and control specialist. He worked on different projects in the
fields of substation control systems and dispatching centers. In 1978, he
joined IREQ, where his main fields of activity were the application of
microprocessors and digital techniques for substations and generating station
control and protection systems. In 1997, he joined Schweitzer Engineering
Laboratories, Inc., as a principal research engineer. Gabriel is an IEEE senior
member and a registered professional engineer in the Province of Québec and
has served on the Power System Relaying Committee since May 1989. He
holds over six patents and is the author or coauthor of several papers in the
fields of signal processing and power network protection and control.
Boris Shulim is a principal transmission relay protection and system planning
engineer for Orange and Rockland Utilities, Inc. (ORU) in Spring Valley,
New York. Boris is a graduate of Polytechnic Institute of Kishinev, Russia.
Since 1985, he has been employed by ORU. His responsibilities have
included the application and design of protective relay systems, coordination
and settings studies, and the analysis of relay performance during system
disturbances. Boris has extensive experience with field troubleshooting and
relay and control testing. He is a professional engineer in the state of
New York.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Protección de Líneas de Transmisión de doble Circuito con
Relevadores de Sobre-Corriente
Adriel Nájera Guevara1, David Sebastián Baltazar1
adrielng@hotmail.com
1
Instituto Politécnico Nacional, SEPI ESIME ZAC Programa de Posgrado en Ingeniería Eléctrica
Teléfono (55) 5234-5678 Fax (55)5 345-6789
Resumen –– En e ste a rtículo s e muestran el ef ecto q ue
presentan e l a coplamiento mutuo de l as l íneas de
transmisión de do ble c ircuito c on bus es c omunes e n l a
protección d e s obre co rriente. T ambién s e p resenta u n
ejemplo del ajuste y coordinación de dicha protección en
líneas acopladas magnéticamente.
Palabras Clave – Línea d e Tr ansmisión, P rotección,
Relevador d e s obre-corriente, p rotección d ireccional,
acoplamiento mutuo, i mpedancia d e s ecuencia cer o
mutua.
Abstract –– In this article they are the effect that presents
the mutual c oupling of t he l ines of transmission of
double circuit with buses common in the protection of on
current. Also one appears an example of the adjustment
and c oordination o f t his p rotection i n c onnected l ines
magnetically.
Keywords –– Transmission L ine, P rotection, rel ay o n-
current, directional pr otection, m utual coupling,
impedance of mutual sequence zero.
I. INTRODUCCIÓN
Las líneas de transmisión constituyen el elemento
principal que sirven para transportar la energía eléctrica de
un lugar a otro. Éstas se pueden clasificar según su longitud
en líneas largas, medianas o cortas o también por su tipo de
configuración en líneas radiales, anilladas o paralelas. En la
actualidad en lo que respecta a la configuración de las líneas
en paralelo es más común utilizar este tipo de configuración
ya que se refuerzan los enlaces y se proporciona un
aumento en la cantidad de energía eléctrica que se transmite
y al mismo tiempo brinda una mayor confiabilidad al
sistema eléctrico.
Para proteger las líneas de transmisión se deben de tomar
en cuenta las características de las líneas como son su
longitud y su configuración, ya que el cálculo de los ajustes
y la coordinación de las protecciones varía entre una
configuración y otra, a medida que varios enlaces se unen
para formar una red. Se debe de tener en cuenta que en caso
de falla existirán un mayor número de aportaciones que
deben ser consideradas en los ajustes de los relevadores para
95
dsebasti@ipn.mx
que garanticen la dependabilidad y la seguridad del sistema
de protección [1].
En el presente artículo se analiza el efecto que tiene el
acoplamiento mutuo en los relevadores de sobre-corriente.
En las líneas de transmisión de doble circuito y que
además converjan en las mismas instalaciones, deben de
hacerse consideraciones especiales como es el caso de los
acoplamientos mutuos de secuencia cero, así como las
inversiones de las corrientes en caso de falla. La reactancia
mutua provoca una variación en la reactancia de secuencia
positiva entre un 3% a un 5%, así mismo los acoplamientos
mutuos provocan sobre corrientes de secuencia cero y la
aparición de impedancias mutuas de secuencia cero que
llegan a afectar entre un 50% y un 70% de la impedancia de
secuencia cero de la línea [2]. Entonces él no tomar en
cuenta estos factores ha ocasionado un sin número de
disparos incorrectos, por esta razón es necesario realizar un
análisis que conlleve a proponer ajustes que garanticen la
operación correcta de los relevadores instalados en las líneas
de transmisión en paralelo.
II. RELEVADOR DIRECCIONAL DE SOBRE-
CORRIENTE
Para que un relevador direccional de sobrecorriente
pueda discriminar la dirección de la falla requiere valores de
corriente y voltaje, esta característica de direccionalidad
normalmente se emplea en líneas anilladas o paralelas.
En los sistemas radiales la utilización de los elementos
direccionales de sobrecorriente es prácticamente
innecesaria, esto porque la aportación de corriente de falla
de la carga hacia el punto de falla en el alimentador es
prácticamente despreciable.
En general se sabe que un elemento direccional clásico
responde al desfasamiento de la corriente y el voltaje de fase
que entra al relevador, para las fallas que se presentan sobre
la línea protegida para “fallas hacia delante” la corriente se
atrasa con respecto al voltaje, es decir que el ángulo entre la
tensión y la corriente corresponden al ángulo de impedancia
del lazo de la falla.
96
Si una falla llega a presentarse en la línea adyacente (fallas
hacia atrás) el ángulo del voltaje permanece casi sin cambios
y el ángulo de la corriente cambia alrededor de 180°, esta
información es utilizada por el elemento direccional para
discriminar entre fallas hacia delante y hacia atrás como se
muestra en la figura 1. La entrada de voltaje actúa como
referencia del ángulo de desfasamiento y es denominado
como voltaje de polarización del relevador, así mismo la
señal de corriente contiene la información para determinar la
ubicación de la falla y es denominada como cantidad de
operación del relevador.
Figura 1 Fasores de Voltaje y Corriente para determinar la dirección de la
falla
En este artículo se da un énfasis especial en la protección
direccional de sobrecorriente para fallas a tierra (67N), en
este tipo de fallas todas las cantidades de secuencia están
presentes, por lo tanto pueden utilizarse estas componentes
para la polarización del elemento direccional. En algunos
casos se utilizan las cantidades de secuencia cero debido a
que son relativamente fáciles de obtener, la corriente
residual que se obtiene de tres transformadores de corriente
conectados en estrella representa la corriente de secuencia
cero.
Así mismo la tensión que se obtiene de tres transformadores
de potencial con el primario conectado en estrella aterrizada
y el secundario en conexión delta abierta representa una
tensión de secuencia cero. Tanto la corriente de secuencia
cero como la tensión de secuencia cero, pueden ser
utilizadas como cantidad de polarización para un elemento
direccional de secuencia cero.
Como se mencionó anteriormente el comportamiento de las
fallas depende de la configuración que tengan las líneas, por
ejemplo para líneas paralelas en un doble circuito, los
efectos de acoplamientos mutuos con una alta impedancia
de secuencia cero y que además tengan fuentes de secuencia
aislada como es el caso de líneas con un solo bus en común,
afectan directamente a los elementos direccionales de
secuencia cero y por lo tanto el relevador puede dar una
mala operación, en las siguientes secciones se hará una
mención puntual sobre el efecto de los acoplamientos
mutuos.
Sin embargo, cuando se habla de líneas paralelas que poseen
buses comunes en ambos extremos de la línea como es el
caso del articulo, se ha demostrado que los elementos
direccionales polarizados con secuencia cero no producen
una declaración incorrecta de la dirección de la falla para
éste tipo de configuraciones, esto suponiendo que el umbral
del par mínimo es alcanzado para todas las fallas, pero
llevado a la práctica en el momento en el que alguno de los
interruptores asociado al bus común abre, se presenta el
efecto como si se tuviera una fuente de secuencia cero
aislada afectando los elementos direccionales de secuencia
cero. Él problema anterior puede resolverse utilizando los
elementos de secuencia negativa, los cuales tienen una
aplicación común en las instalaciones donde se tienen
transformadores de potencial conectados en delta abierta o
bien donde se tienen transformadores de potencial
disponibles en el lado opuesto de un banco de
transformación con conexión estrella-delta.
La componente de secuencia negativa se caracteriza por su
disponibilidad y confiabilidad en la cantidad de
polarización, así que mientras los voltajes y corrientes de
secuencia cero no siempre están disponibles, los de
secuencia negativa siempre lo están.
Se puede decir que es indistintamente del tipo de elemento
direccional que se use y de su conexión, el elemento de
sobrecorriente siempre responde a la corriente residual 3I0
[1 ].
Actualmente las protecciones digitales integran la
información del voltaje y la corriente, realizando los
cálculos de los elementos direccionales y evaluando los
datos, a fin de adoptar la decisión direccional correcta, es
decir que los relevadores modernos permiten utilizar
múltiples elementos direccionales, según las condiciones del
sistema.
Estas razones conducen a elegir elementos de secuencia
negativa para la polarización del elemento direccional
debido a que dichos elementos no son sensibles a la
impedancia de acoplamiento mutuo de secuencia cero para
topologías de líneas paralelas, inclusive con aquellas que
poseen fuentes de impedancia de secuencia cero aisladas.
El elemento que actúan para fallas a tierra es el 51N que
corresponde al elemento residual de sobre-corriente y se
define con las siguientes características:
51NC.- Tipo de curva del elemento residual de sobre-
corriente, en la figura 2 se muestra el tipo de curva utilizada
en los ajustes de las líneas ANG1 y ANG2.
51NTD.- Tiempo de dial del elemento de sobrecorriente.
51NP.- “Pickup” del elemento de sobre-corriente.
Entre otros datos de ajuste del relevador se deben colocar
los valores de relación de transformación de corriente
(CTR), relación de transformación de voltaje (PTR). Para
ajustar los elementos que brindan la direccionalidad del
relevador se deben ingresar los datos de: ángulo de torque
máximo de la línea (MTA) y la sensibilidad con un valor
proporcional al “pickup” del elemento sobre-corriente 51P
siendo 4 A<51pp<12.6 A
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 97

El caso anterior no es el único que se presenta ya que

existen diversos y estos se muestran en las figuras
siguientes:

Figura 4 Acoplamiento mutuo con un solo extremo común

Figura 5 Acoplamiento mutuo sin ningún extremo común

Figura 2 Curva Inversa Relevador SEL 1 [7]

III.- LINEAS CON ACOPLAMIENTO MUTUO

La Impedancia de secuencia cero mutua se presenta

cuando existen dos o más líneas en paralelo, ya sea en la Figura 6 Acoplamiento mutuo y apertura inicial en un extremo con el otro
misma torre o en adyacentes. Esta impedancia permite cerrado.
calcular la caída de voltaje de secuencia cero que se
presentaría en una línea, debido a la circulación de corriente
de secuencia cero por la otra línea [3]. La expresión de
cálculo es:

=
V01 Z 01 I 01 + Z 0 m I 02 (1)

Donde V01, Z01 e I01 son el voltaje, impedancia y

corriente de secuencia cero respectivamente, en el circuito 1
donde ocurre la falla, Zom es la impedancia de secuencia
cero mutua e I02 es la corriente de secuencia cero por el
circuito 2. En la figura 3 se muestra el caso donde se tienen
dos líneas acopladas con dos buses en común.
Figura 3 Acoplamiento mutuo con extremos comunes
Figura 7 Múltiples acoplamientos
En la figura 4 se tienen el acoplamiento mutuo en líneas en
paralelo pero en las cuales únicamente se tienen un bus en
común, en la figura 5 se tienen el acoplamiento mutuo pero
en este caso no se tienen ningún bus en común, para este
caso las líneas pueden estar al mismo nivel de tensión o a
diferente nivel de tensión. En la figura 6 se tienen el
acoplamiento mutuo pero un extremo de la línea se
encuentra abierta mientras que el otro extremo se encuentra
cerrado.
98
En el último caso (figura 7) cada línea de transmisión estará
acoplada con las otras dos en parte de su trayecto y no en
todo. Estos acoplamientos también se pueden presentar entre
líneas de diferente nivel de tensión, como es el caso de
líneas en paralelo, en un camino común de líneas, o el poco
frecuente caso de circuitos con diferente nivel de tensión en
una misma torre. En los dos últimos casos hay una dificultad
práctica para llevar las señales necesarias que permitan
incorporar las corrientes de secuencia cero mutuas a los
relés.
IV EFECTO DEL ACOPLAMIENTO MUTUO DE
SECUENCIA CERO EN LÍNEAS DE TRANSMISIÓN DE
DOBLE CIRCUITO.
El comportamiento de los relevadores de protecciones se
analiza en términos de las componentes de secuencia.
Figura 8 Acoplamiento mutuo en líneas de doble circuito [8]
Los efectos de acoplamientos entre alimentadores paralelos
en las componentes de secuencia positiva y negativa,
usualmente son muy pequeños y pueden despreciarse.
Sin embargo el efecto del acoplamiento mutuo en la
secuencia cero, es considerable llegando a afectar entre un
50% y un 70% de la impedancia de secuencia cero de la
línea. El resultado de los acoplamientos mutuos trae consigo
sobre-corrientes de secuencia cero y esto trae consigo la
aparición de impedancias mutuas de secuencia cero [4]. En
algunas ocasiones esta impedancia de secuencia cero puede
resultar en una operación incorrecta de la direccionalidad
del relevador para fallas a tierra. Cuando se presenta una
falla de fase a tierra en algunas ocasiones sucede que para
líneas del mismo nivel de voltaje donde existan impedancias
mutuas de secuencia cero provoquen una operación errónea
del relevador de fallas a tierra.
Cuando ocurre una falla a tierra en un circuito, el voltaje
aplicado al otro circuito incluye un voltaje inducido
proporcional a la corriente de secuencia cero en el otro
circuito sano. Cuando los parámetros de las líneas paralelas
son similares para una falla externa provocará que los
efectos de acoplamientos mutuos se cancelen unos a otros
compartiendo las corrientes de secuencia cero para fallas
remotas.
Otra situación que requiere se tome en cuenta son los
efectos de acoplamientos mutuos cuando una de las líneas
paralelas esta fuera de servicio y aterrizada en ambos
extremos, en caso de existir una falla a tierra en la línea que
está en servicio esta puede inducir una corriente en el
circuito aterrizado del alimentador, causando una señal de
compensación mutua engañosa.
La corriente producida por el acoplamiento mutuo varía
inversamente con la distancia de separación de las líneas y
directamente con el voltaje que llevan las mismas. [5]
V. EFECTO DEL ACOPLAMIENTO MUTUO EN LA
PROTECCIÓN DIRECCIONAL DE SOBRECORRIENTE.
A menudo este tipo de relevadores se emplean como
protección de respaldo en las líneas de transmisión y ofrecen
una buena cobertura en fallas de alta impedancia cuando se
aplican como esquema de comparación direccional. Para
fallas hacia el extremo remoto de la línea el acoplamiento
mutuo puede causar una reducción en la corriente disponible
de secuencia cero en ambas líneas. Esto tendrá un efecto en
la sensibilidad del elemento de tierra y el ajuste del pickup
debe ser modificado. Por otro lado, un disparo secuencial en
una línea fallada puede causar la operación de la protección
de la línea sana. Para una falla de fase a tierra cerca de un
extremo de la línea, dicho extremo abrirá primero y el
extremo remoto posteriormente. Esto puede inducir una gran
cantidad de corriente de secuencia cero en la línea adyacente
sana durante el periodo de transición donde sólo un extremo
ha abierto y se mantiene alimentando la corriente de falla a
través del extremo remoto. Esto puede provocar que un
elemento de tierra opere, entonces el ajuste del elemento
sobrecorriente debe considerar las situaciones mencionadas
de manera cuidadosa.
El efecto de un circuito paralelo abierto y aterrizado en
ambos extremos tendrá como consecuencia una gran
corriente para fallas en el extremo remoto y necesita
considerarse cuando se determina el ajuste de la protección
del relevador de sobrecorriente.
Efecto d e l os disparos s ecuenciales e i nversión d e
corriente
Cuando una falla es librada secuencialmente en uno de los
circuitos de una línea de circuitos dobles con fuentes de
generación en ambos extremos, la corriente en la línea sana
puede invertirse en un periodo de tiempo y esto puede
provocar disparos por permisivos de sobrealcance o
bloqueos de comunicación que se utilizan en los esquemas
de distancia. Bajo condiciones de carga (el mismo caso que
en condiciones de fallas externas) la diferencia de corrientes
entre ambos circuitos es mínima, mientras que después de
que ocurra una falla ocurra en una de las líneas, la diferencia
de corriente es significativamente grande en la línea sana y
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 99

la línea fallada. Cuando existe una diferencia grande en las En las siguientes figuras se muestra las mismas condiciones
fuentes de ambos extremos, la magnitud de la corriente en que la simulación anterior solo que ahora se considera el
ambas líneas del lado de la fuente débil es similar. Los acoplamiento mutuo.
relevadores de sobrecorriente de tierra se ve afectado por los
acoplamientos con condiciones de secuencia cero y los
ajustes de “Pick up” deben calcularse bajo condiciones de
operación diferentes [6].

V SIMULACIÓN Y PRUEBAS

A continuación se muestran las simulaciones hechas en el

programa ASPEN ON LINER, partiendo de un estado en el
que no existe el efecto de acoplamientos mutuos y
posteriormente se harán con efecto de dichos acoplamientos.
Figura 11 Tiempos de operación de los relevadores direccionales de
La impedancia de secuencia cero mutua considerado en este sobrecorriente con efecto mutuo
tiene el valor siguiente en PU:

Z0M=0.11074+j0.27506

Primer caso de acoplamientos mutuos “Fallas al extremo

remoto”.

Para este caso se simulara falla monofásica a tierra al 2% del

extremo remoto de las líneas ANG1 y ANG2, estas
simulaciones se harán sin considerar el efecto del
acoplamiento mutuo y posteriormente considerando el
acoplamiento mutuo. Las figuras que se mostraran a
continuación representan una red parcial.
Figura 12 Aportación de corrientes de secuencia cero con acoplamiento
mutuo

De las figuras 9 y 11 se observa que cuando se toma en

Figura 9 Tiempos de operación de los relevadores direccionales de
sobrecorriente sin efecto mutuo
cuenta el acoplamiento mutuo de las líneas se tienen un
aumento en el tiempo de operación de las protecciones, esto
se debe a que se tiene una reducción de la corriente de
secuencia cero como se observa al compara las figuras 10 y
12.
Segundo caso de acoplamientos mutuos “Falla interna con
enlaces cerrados”
Para este caso se simulo una falla monofásica al 2% de la
línea ANG2 con ambas líneas en servicio. Las primeras
simulaciones se harán sin considerar el acoplamiento mutuo
y posteriormente se harán considerando el acoplamiento
mutuo.

Figura 13 Tiempos de operación de los relevadores direccionales de

Figura 10 Aportación de corrientes de secuencia cero sin acoplamiento sobrecorriente sin efecto mutuo
mutuo
100
Figura 14 Aportación de corrientes de secuencia cero sin acoplamiento
mutuo
A continuación se muestran los tiempos y corrientes de
secuencia cero pero tomando en cuenta el acoplamiento
mutuo.
Figura 15 Tiempos de operación de los relevadores direccionales de
sobrecorriente con efecto mutuo
Figura 16 Aportación de corrientes de secuencia cero sin acoplamiento
mutuo
Para este segundo caso puede observarse que el efecto del
acoplamiento mutuo al presentarse una falla interna se
refleja en un incremento de los tiempos de operación de los
relevadores de sobrecorriente, teniendo un aumento de 60ms
en la línea ANG2 y un aumento de 80ms en la línea ANG1.
VI. CRITERIOS DE AJUSTE Y COORDINACION DE LA
PROTECCION DIRECCIONAL DE SOBRECORRIENTE
Para seleccionar el ajuste y la coordinación del relevador
direccional de sobrecorriente es necesario seguir el
procedimiento de los criterios de ajuste que se mencionan a
continuación.
Para el cálculo del 67N debe tenerse en cuenta el
compromiso del esquema que es operar para fallas dentro de
la propia línea y respaldar hasta donde sea posible, esto será
inherente al cálculo y deberá quedar definido en las
conclusiones del propio estudio.
El ajuste del pickup se obtiene de calcular fallas a tierra en
los extremos de las líneas adyacentes al bus delantero con el
interruptor abierto en el bus remoto, con el objeto de
conocer en cada caso la aportación de la corriente de 3I0 que
pasa por el relevador y así seleccionar el pickup de corriente
que garantice una sensibilidad adecuada para la detección de
las fallas a respaldar. El valor de ajuste de pickup será a
criterio entre el 30% y 50% del valor mínimo encontrado en
el cálculo. En algunos casos esto no será posible por lo que
se deberá mencionar en el estudio. Para el cálculo de la
palanca se simula la falla a tierra en el bus adyacente para
seleccionar el valor de la palanca que proporcione un tiempo
de operación entre 0.4-0.5 seg. aproximadamente, suficiente
para coordinar con el relevador 67N delantero que deberá
operar en un tiempo de 0.1 a 0.2 seg para fallas en
terminales con máxima aportación, así como con la zona 1
del relevador de distancia a tierra. El cálculo deberá
efectuarse a la salida de las líneas adyacentes con interruptor
abierto en el otro extremo, considerando inclusive la
contingencia más probable y seleccionando la máxima
aportación.
Para realizar el ajuste del relevador direccional de
sobrecorriente se calculan las aportaciones obtenidas del
programa ASPEN ON LINER simulando fallas a tierra de
las líneas LAT, QMD y MZL y se selecciona la línea que
tenga menor aportación de corriente 3I0 para garantizar la
sensibilidad de la protección.
En la siguiente figura se muestra las líneas adyacentes a las
líneas ANG2 e ANG1.
Figura 17 líneas adyacentes a las líneas ANG2 e ANG1.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 101

En la tabla siguiente se muestran las aportaciones de IV. CONCLUSIONES

corriente.
La configuración de las líneas en paralelo que tienen el
mismo derecho de vía se ven afectadas por el fenómeno de
ANG1 ANG2 Linea Adyacente acoplamiento electromagnético, esto se refleja en un
Amperes Amperes Amperes incremento en la impedancia de secuencia cero y en una
172 172 LAT disminución de la corriente de secuencia cero, lo que puede
171 171 QMD ocasionar que el relevador direccional de sobrecorriente
170 170 MZL tenga una operación incorrecta en la direccionalidad para
fallas a tierra. El acoplamiento de secuencia positiva y
Para el cálculo del pickup de la línea se usa la corriente más negativa son valores pequeños por lo que estas no influyen
pequeña que aportan las líneas adyacentes. en el ajuste y coordinación de las protecciones.

I=170 (0.3) =51 amp Para que se tenga un funcionamiento adecuado de la

Ipickup= I/RTC= 51amp / 160= 0.318 amp protección se debe de tomar en cuenta el acoplamiento
mutuo de las líneas, ya que si no se toma en cuenta se
tendrían operaciones incorrectas en dichas protecciones ya
que el acoplamiento mutuo genera un incremento en el
tiempo de operación de los relevadores direccionales de
sobrecorriente [ver figuras 9,11,13 y 17], por lo anterior se
hace necesario considerar el acoplamiento mutuo al
momento de realizar el ajuste y coordinación de las
protecciones.

Por otra parte, para seleccionar el pickup adecuado se hace

necesario conocer la aportación de 3I0 de las líneas
adyacentes, esto con la finalidad de garantizar la
sensibilidad de la protección para que opere correctamente.

REFERENCIAS
[1] P. M. Anderson “Power System Protection” John Wiley, 1998.

[2] Stanley H. Horowitz “Power System Relaying”. John Wiley, 1991.

[3] Russell Mason “The Artic and Science of Protection Relaying”.

Jhon Wiley 1956
[4] AIEE Committee Report, “Protection of Multiterminal and Tapped
Lines” Trans. AIEE,80 April 1961, pp. 55-65
[5] Wheeler, S.A., “Influence of Mutual Coupling Between Parallel
Circuits on the Setting of Distance Protection,” Proc. IEE, 117, (2),
1970, pp. 439-444
[6] Popovic L.M “ A practical method for evaluation of ground current
distribution on double circuit parallel lines” IEE Transactions on
Volume 10, Issue 1, Jan 1995 page 187-193
[7] Y.G. Paithankan “Transmission Network Protection Theory and
Practice”. Marcel Dekkel Inc.
[8] Alstom “Protective Relaying Application Guide”
Figura 18 Tiempos de operación y coordinación de los elementos
direccionales de sobrecorriente y protección temporizada de sobrecorriente
de neutro del banco con simulación de falla en el bus adyacente ala L.T.´s
IP-PC e ITP93060PIC.

En la figura 18 se observa que para una falla en el bus

adyacente de las líneas analizadas los relevadores
direccionales de sobrecorriente operan en un tiempo de
550ms. Así mismo el elemento temporizado de
sobrecorriente de neutro respaldara para esta falla en un
tiempo de 1.26s garantizando con estos tiempos que para esa
falla operen las protecciones primarias y las de respaldo de
la propia línea.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
METODOLOGÍA PARA EL ANÁLISIS DE PERTURBACIONES EN SISTEMAS DE POTENCIA MEDIANTE
ATPDRAW – CASO DE PUENTE ABIERTO EN LA LÍNEA ESMERALDA – YUMBO 2 A 230kV
Diego Alejandro Tejada Arango
Analista Análisis Operativo
Análisis Operativo
Dirección Gestión de la Operación
ISA-Colombia
dtejada@isa.com.co
RESUMEN. Este documento presenta el análisis,
la metodología, simulación y experiencia de
análisis postoperativo de Interconexión Eléctrica
S.A. (ISA) asociada a una falla en un circuito a 230
kV compensado serie por un puente abierto que
ocasionó indisponibilidad del circuito fallado por
casi 24 horas. Este artículo hace especial énfasis
en la localización de la falla y metodología de
simulación para la caracterización desde el punto
de vista eléctrico del evento.
PALABRAS CLAVES. Interconexión Eléctrica S.A.
(ISA), Sistema de transmisión Nacional (STN),
ATPDraw, fallas longitudinales, recierre,
registrador de fallas, compensación serie, Sistema
de Información de Descargas (SID), análisis de
fallas.
INTRODUCCIÓN.
Las redes de transmisión de un sistema de
potencia, están constantemente expuestos a fallas
ocasionadas por descargas atmosféricas,
vegetación, acercamientos con circuitos de otros
niveles de tensión, entre otros. Por su parte, las
fallas más comunes en las líneas de transmisión
son los cortos circuitos a tierra y entre fases, como
lo son fallas monofásicas, trifásicas, bifásicas a
tierra y bifásicas aisladas. Sin embargo, ante el
envejecimiento de las redes de transmisión más al
incremento del flujo de potencia a través de ellas,
comienzan a ser más comunes fallas del tipo
longitudinal, como conductores rotos (que
tradicionalmente han sido más comunes en niveles
de distribución). En la Tabla 1se presenta una
estadística de las fallas en circuitos de transmisión
presentadas en el período 2001 - 2010 en la red de
ISA.
Causa Porcentaje
Descarga Atmosférica 47.2%
Derribamiento de Torres 11.8%
Quema o Fuego Bajo la Línea 1.6%
Daño en Conductor - Bajante 0.6%
103
Sadul Segundo Urbaez Gonzalez
Especialista Operación Red
Análisis Operativo
Dirección Gestión de la Operación
ISA-Colombia
surbaez@isa.com.co
Causa Porcentaje
Acercamiento Conductores 0.7%
Alta Vegetación 2.1%
Caída de Árbol 0.3%
No Establecida 33.3%
Otros 2.5%
Total -
Fuente: Elaboración Propia
Nota: La causa no establecida, representa fallas que son
de naturaleza fugaz y con características iguales a las
causadas por descarga atmosférica,
Tabla 1. Estadística de fallas 2001 - 2010
Como se aprecia en la Tabla 1, las fallas
asociadas a ruptura de conductores y daños en
bajantes han representado del orden de 0,6% del
total de fallas. El análisis, caracterización y
simulación de este tipo de fallas es de especial
interés y para ello en este artículo se describe y
analiza el evento del 26/12/2010 en el circuito
Esmeralda – Yumbo 2 a 230kV, el cual tuvo como
causa de la falla, un puente abierto en la fase C de
la torre 321.
1. DESCRIPCIÓN DEL EVENTO
El 26/12/2010 a las 18:33 se presentó el disparo
del circuito Esmeralda - Yumbo 2 a 230 kV con 130
MW ante falla en el circuito y operación de sus
protecciones principales, ver Figura 1. La falla se
localizó en la torre 321 a 16.51 km desde Yumbo
por un puente abierto en la fase C del circuito, en
la Figura 2 se presenta el registro fotográfico de la
falla mediante la inspección aérea realizada por
ISA.La línea tiene una compensación serie a 6 km
de la subestación Yumbo, en las instalaciones de
la subestación San Marcos 230kV, la cual quedó
por by-pass a raíz de la falla en el circuito.
104
Fuente: ISA-Sistema de Transmisión Nacional
Figura 1. Diagrama unifilar
Fuente: Inspección de la línea realizada por ISA
Figura 2. Fotografía de la causa de la falla
2. ANALISIS DEL EVENTO ESMERALDA –
YUMBO 2
Como se mencionó en numerales anteriores, la
falla se produjo por un puente abierto en la fase C
de la torre 321. En la torre fallada la fase C está
en la parte superior y la fase B en la intermedia
(ver Figura 2), al abrirse el puente de la fase C, se
hizo contacto con la fase B y produjo la falla entre
las fases del circuito. Se produjo el disparo de
ambos extremos por recierre en falla en Esmeralda
y falla en tiempo de reclamo en Yumbo. La
compensación serie se fue a by-pass (cierre del
interruptor 2CS10), luego se realizó intento
decierre desde Yumbo el cual fue exitoso, luego se
realizó intento de cierre manual desde Esmeralda
el cual fue no exitoso por falla en las fases B y C,
cuando se realizó estecierre desde Esmeralda, en
Yumbo se detectó la falla sólo en la fase B y la
bahía realizó recierre monofásico exitoso en
Yumbo. Ambos cierres manuales se realizaron
con la compensación serie en by-pass. Debido a
este rechazo en la energización, se procedió a
revisar la localización de la falla e inspección del
circuito.
2.1 RELES INSTALADOS PARA PROTECCION DEL
CIRCUITO
Para la protección del circuito se tienen
instalados en ambos extremos de la línea, relés
PXLN y LFDC ambos del fabricante Alsthom. El
relé PXLN es un relé con principio de operación
distancia y como particularidad para la protección
de esta línea su zona 1 fue temporizada 120ms,
con el fin de prevenir disparos indeseados ante
fallas externas, que presenten incursiones
transitorias en la zona 1 del relé. El relé LFDC es
de un principio de operación por cantidades
superimpuestas (∆V, ∆I).
2.2 ANÁLISIS DE LOS REGISTROS DE FALLA EN
ESMERALDA
En la bahía de línea de Esmeralda 230 kV campo
a Yumbo 2 se detectó Falla en la fase C en el relé
LFDC (no se detectó falla en la fase B, la tensión
de fase no disminuyó y la corriente no aumentó lo
suficiente como para que los deltas generaran
disparo de la fase B) el relé PXLN presentó
omisión de actuación en la primera falla lo cual es
normal porque la zona 1 está temporizada 120ms.
La falla fue aclarada por el interruptorL250, con
apertura de la fase C a los 77 ms y con recierre a
los 802 ms (tiempo muerto: 725 ms, una vez
realizado el recierre se presentó nuevamente falla
en la fase C a los 805 ms, aclarada por el
interruptor L250, con apertura de las fases ABC a
los 872 ms por orden del relé LFDC y del PXLN
por su función de cierre en falla, ver Figura 3.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Fuente: Registrador de Fallas en Esmeralda
Figura 3. Oscilografía del evento en Esmeralda
De este registro es importante destacar dos
aspectos para detectar la apertura de la fase C en
algún punto de la línea. Primero, la corriente de
prefalla de la fase C es capacitiva, lo que indica
que ya no había flujo de potencia por esta fase en
el circuito, ver Figura 4. Segundo, la tensión de la
fase C durante el ciclo de recierre monofásico es
igual al de la fase B, lo cual indica la unión de
ambas fases, ver Figura 5.
Fuente: Registrador de Fallas en Esmeralda
Figura 4. Análisis de la corriente de prefalla
Fuente: Registrador de Fallas en Esmeralda
105
Figura 5. Análisis de la tensión durante el ciclo de
recierre monofásico
2.3 ANÁLISIS DE LOS REGISTROS DE FALLA EN
YUMBO
En la bahía de Yumbo 230 kV campo a Esmeralda
2 se detectó falla en las fases B y C en el relé
LFDC, el relé PXLN presentó omisión de actuación
en la primera falla lo cual es normal porque la zona
1 está temporizada 120ms. La falla fue aclarada
por el interruptor L190, con apertura de las fases
ABC a los 70 ms y con recierre trifásico Barra Viva
– Línea Muerta a los 663 ms (tiempo muerto: 593
ms). Una vez realizado el recierre se presentó
nuevamente falla en la fase B a los 805 ms
(después del recierre del otro extremo) el relé
LFDC disparó nuevamente, la segunda falla fue
aclarada por el interruptor L190, con apertura de
las fases ABC a los 867 ms, ver Figura 6.
Fuente: Registrador de Fallas en Yumbo
Figura 6. Oscilografía del evento en Yumbo
Al igual que para el extremo de Esmeralda, para el
extremo de Yumbo también vale la pena destacar
de los registros de la falla la corriente de prefalla
en la fase C, la cual también indicada apertura del
conductor antes de la falla, ver Figura 4.
Fuente: Registrador de Fallas en Yumbo
Figura 7. Análisis de la corriente de prefalla
2.4 ANÁLISIS DEL TIPO DE FALLA
Se analizó la falla desde ambos extremos,
mostrando que por la característica de la falla, ésta
fue bifásica en un extremo y monofásica en el
otro. Ahora bien, una pregunta que surge en el
momento del análisis es si la falla fue bifásica a
106
tierra o bifásica aislada. Una duda que se
presenta al observar que las corrientes en las
fases falladas son de magnitudes muy diferentes y
ángulos de fase también diferentes de 180º.
Con el fin de entender un poco mejor el tipo de
falla se procedió a realizar una sincronización de
los registros de ambos extremos y mediante el
análisis fasorial, en el momento de la falla, poder
determinar el tipo de falla presentada. En la Figura
8, se muestra como los fasores de corriente de las
fases B en ambos extremos de la línea suman el
valor negativo del fasor de corriente de la fase C
en el extremo donde la falla se presentó de
manera bifásica. Esto nos lleva a lanzar la
hipótesis de que la falla fue bifásica aislada de
tierra, esta hipótesis será validad mediante la
simulación del evento en el ATPDraw en el
numeral 4 del presente artículo.
2.5 ANÁLISIS DE LA LOCALIZACIÓN DE LA
FALLA
La longitud total del circuito es de 193.3 km, la
localización de la falla desde Esmeralda dada por
la protección principal de la línea fue 100.4km.
Fuente: Elaboración Propia
Figura 8. Análisis fasorial de las corrientes de falla
Se procedió a validar la localización de la falla con
el registro de la corriente capacitiva desde el
extremo de Yumbo. Este método de localización
está basado en que para una línea de transmisión
en estado estable a frecuencia industrial la
magnitud de la corriente capacitiva de la línea es
directamente proporcional a la longitud de la línea.
Para el caso de una fase abierta la corriente
capacitiva corresponderá a los kilómetros desde la
subestación hasta el punto donde se encuentra
abierta la fase.
Para nuestro caso particular del circuito Esmeralda
– Yumbo 2, al energizar desde Yumbo el registro
mostró que las fases sanas (A y B) tenía una
corriente de 88.9Amp capacitivos y que la fase
afectada (C) tenía una corriente de 8.74Amp
capacitivos, ver Figura 9. Haciendo una regla de
tres (si para 193.3km son 88.9km, para 8.74Amp
serán 20km) se determinó que la localización de la
falla era alrededor de 20km desde Yumbo. El área
de mantenimiento determinó que la falla estaba en
la torre 321 a 16.51km desde Yumbo (error de
3.49km en la localización mediante este método).
A pesar de ser una falla sin resistencia de falla, por
sus características eléctricas, puede verse que se
produce un alto error en el cálculo de la
localización si empleamos métodos de localización
basados en la impedancia de la línea.
En Yumbo la localización con el Registrador de
Fallas fue un valor entre 29.9km y 42km desde
esta subestación utilizando el loop fase B a tierra.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Fuente: Elaboración Propia
Figura 9. Magnitudes de la corriente capacitiva
3. MODELO DE SIMULACIONEN ATPDRAW
Con el análisis del evento presentado en los
numerales anteriores, se procedió a corroborar el
comportamiento de la falla mediante una
simulación en ATPDraw. Para el modelado de
cualquier sistema eléctrico existen procedimientos
dentro del proceso de generación de un caso de
simulación y que en todos los casos en que se
quiera reproducir un evento deben ser aplicados
de una u otra forma. Por esta razón, se proponen
los siguientes puntos importantes para todo
proceso de modeladode fallas en sistemas de
potencias:
1. Se debe visualizar el punto donde ocurrió la
falla y a partir de eso determinar que partes de
todo el sistema se modelaran de forma
detallada y cuáles de manera simplificada.
2. Obtener los datos necesarios para la
representación en ATPDraw de los dispositivos
que serán modelados. En caso de no
conseguir estos datos, utilizar datos típicos o
datos de un dispositivo similar.
3. Por medio de un Software de simulación de
sistemas de potencia (para el caso de ISA
Power Factory DIgSILENT) obtener los datos
de los equivalentes de Thevenin (R0, X0, R1 y
107
X1) de las zonas del sistema que no se
modelaran en detalle.
4. Al tener todos los datos necesarios, se
representa en el software ATPDraw cada uno
de los dispositivos (Líneas, Transformadores,
Reactores, Condensadores, etc.) que se
encuentran dentro de la zona donde ocurrió la
falla y son relevantes en el análisis del evento.
5. Al tener modelado el sistema (sin la falla), se
debe sintonizar el modelo con los valores de
prefalla (Tensiones y Corrientes) entregados
por los registros de los Registradores de falla y
relés.
6. Al tener sintonizado el sistema seleccionamos
el punto real donde ocurrió la falla en el
modelo, en caso de ser en una línea de
transmisión, utilizar la distancia entregada por
el localizador de fallas como una primera
aproximación.
7. Por medio de los registros se debe visualizar el
tipo de falla que ocurrió en el evento. En caso
de no estar muy claro en el registro, se debe
realizar una sensibilización, aplicando
diferentes fallas al modelo hasta encontrar la
que represente las formas de ondas del
registro durante la falla (la falla se ubicará de
acuerdo al ítem anterior).
8. Al tener el tipo de falla, por medio de los
interruptores del sistema modelado
representaremos la secuencia de apertura y
cierre de cada uno de los interruptores que
intervienen en el evento.
9. Luego se simula en el programa y se observa
que los aportes de corrientes sean
consecuentes con los aportes vistos en el
registro, en caso contrario, mover el punto de
falla.
10. Dentro del proceso de sintonización de la falla,
existen ciertos parámetros que pueden ser
variados, entre los cuales están la ubicación
de la falla, impedancia de la falla y los valores
de secuencia cero.
De acuerdo a la metodología expuesta, a
continuación se presentan los parámetros de
modelado para el caso del evento de la línea
Esmeralda – Yumbo 2.
108

3.1 EQUIVALENTES DE CORTO CIRCUITO 11la configuración del modelo creado en ATPDraw
Para los equivalentes de corto circuito se para la simulación.
emplearon los elementos LINESY_3 (para modelo
de las redes de secuencia) y ACSOURCE (para la Parámetro Línea Esmeralda – Yumbo 2
fuente de tensión trifásica), ver Figura 10. Longitud [km] 193.3
Año de entrada 1971
R0 [ohm/km] 0.4013
X0 [ohm/km] 1.3275
B0 [µS/km] 2.1592
R1 [ohm/km] 0.0547
X1 [ohm/km] 0.4618
B1 [µS/km] 3.4880
Fuente: XM – Parámetros Técnicos (PARATEC)

Tabla 3. Parámetros de la línea de transmisión.

Fuente: Elaboración Propia

Figura 10. Elementos usados en los equivalentes

Los parámetros de secuencia (R0, X0, R1, X1)

empleados para la simulación se obtuvieron a
partir de los niveles de corto circuito trifásico y
monofásico de cada subestación. La tensión de
prefalla línea-neutro se obtuvo de los registros del
evento. Los parámetros obtenidos se presentan
en la Tabla 2.
Fuente: Elaboración Propia
Parámetro S/E Esmeralda S/E Yumbo
Figura 11. Modelo de línea usado en ATPDraw
Icc 3f [kA] 13.48 10.39 3.3 COMPENSACIÓN SERIE
Icc 1f [kA] 13.15 11.48 Para la simulación de la compensación serie de la
R0 [ohm] 1.306 0.979 línea, se empleó un modelo simplificado
X0 [ohm] 9.863 8.048 condensador más interruptor de by-pass, ver
R1 [ohm] 0.817 0.902 Figura 12.
X1 [ohm] 9.279 12.044
Vprefalla [kV] 131 130
Fuente: Elaboración Propia

Tabla 2. Parámetros para los equivalentes.

3.2 LÍNEA DE TRANSMISIÓN

La línea de transmisión Esmeralda – Yumbo 2 a
230kV está en una estructura decircuito doble con
un diseño de torre Normalizada Clase II, la
configuración esvertical, con un conductor de fase
AAAC 1200 61/0 y un cable de guarda:
ALUMOWELDAW 7 NUM9 7/0. En la Tabla 3 se
presentan los parámetros de la línea y en la Figura
Fuente: Elaboración Propia
Figura 12. Modelo empleado para la compensación
serie
Este condensador serie, compensa la línea en un
50%, tiene una capacidad de 3x22Mvar, tensión
nominal de 31.5kV y tiene para su protección un
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 109

spark gap, un varistor y un interruptor. Para esta

simulación sólo se modeló el interruptor debido a
que producto de la energía producida en el varistor
hizo que la compensación rápidamente se fuera a
by-pass (cierre del interruptor) y la compensación
no tuviera más efecto en la falla.

3.4 MODELO PARA LA SIMULACION

Con todos los parámetros de entrada descritos en
los numerales anteriores y los datos obtenidos del
análisis de la falla, se creó el modelo de simulación
en ATPDraw, presentado en la Figura 13:

Fuente: Elaboración Propia

Figura 14. Tensiones en el extremo de Esmeralda

Fuente: Elaboración Propia

Figura 13. Modelo realizado en ATPDraw

4. SIMULACIONES DEL EVENTO CON

ATPDRAW
Con los elementos del numeral 3 y la información
de los tiempos de operación reales del evento del
numeral 2, se procedieron a realizar las
simulaciones con el ATPDraw. Para confirmar la
hipótesis de una falla bifásica aislada de tierra, se
procedió a realizar este tipo de falla en el modelo y
a validar los valores de simulación respecto a los
valores reales de los registros de falla. A
continuación en la Figura 14, Figura 15, Figura 16
y Figura 17, se presenta una comparación de los
resultados y algunos valores de especial interés
para realizar una comparación, la primera gráfica
siempre será asociada al registro real y la segunda Fuente: Elaboración Propia
al resultado de la simulación.
Figura 15. Tensiones en el extremo de Yumbo
110
Fuente: Elaboración Propia
Figura 16. Corrientes en el extremo de Esmeralda
Fuente: Elaboración Propia
Figura 17. Corrientes en el extremo de Yumbo
En la Figura 18, se presenta una comparación de
las corrientes de falla punto a punto entre la
simulación y el registro real de falla.
Fuente: Elaboración Propia
Figura 18. Comparación de las corrientes de falla
De acuerdo con los resultados anteriores, una
comparación entre los resultados entregados por la
simulación y los valores reales dados por el
registrador de fallas, se puede concluir que
definitivamente la falla que ocurrió durante el
evento fue bifásica aislada de tierra, debido a que
representa con fidelidad las formas de ondas y
valores de tensión y corriente en la simulación
realizada en ATPDraw.
5. CONCLUSIONES Y RECOMENDACIONES
 Las fallas asociadas a apertura de puentes o
ruptura de conductores representan un gran
reto para los ingenieros de operación, debido a
que para este tipo de fallas los métodos de
localización por impedancia pueden generar
errores que afectan la disponibilidad de la
línea, en este artículo además de presentar
una caracterización de la falla ocurrida en el
circuito Esmeralda – Yumbo 2, se realizó
énfasis en la localización de falla mediante
corriente capacitiva, el cual para el caso de
ISA ha generado inspecciones en la línea con
este tipo de fallas más efectivas que los
métodos de localización por impedancia.
 En los sistemas de potencia ocurren diferente
tipo de fallas a diario, sin embargo fallas como
la presentada en este artículo son poco
comunes, aún cuando el envejecimiento de las
redes y el incremento del flujo flujos de
potencia, hacen más propenso los circuitos a
este tipo de falla. Con lo cual las simulaciones
para su caracterización cobra relevancia en la
operación del sistema de potencia.
 En este artículo hemos planteado una
metodología para la caracterización de la falla
mediante el programa de simulación ATPDraw
con resultados satisfactorios.
 La caracterización de la falla mediante el
ATPDraw comprobó que se trataba deuna falla
bifásica aislada, en la cual las corrientes de
falla no tuvieron el comportamiento típico de
este tipo de fallas (igual magnitud, desfase de
180º), lo cual representa un reto para las
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 111

protecciones eléctricas.Por esta razón se Santander- 1998. Empleado de Interconexión

recomienda incluir este tipo fallas dentro de las Eléctrica S.A E.S.P desde 1997, donde ha
pruebas a relés de protección. desempeñado los cargos de Analista de Operación
y actualmente desempeña el cargo de Especialista
6. REFERENCIAS de Operación de la Dirección Gestión de la
operación de la Subgerencia Técnica de
[1] DOMMEL H. W., "Emtp Rule Book", Can/Am Interconexión Eléctrica S.A E.S.P
EMTP User Group, 1995
[2] DOMMEL H. W., "Electromagnetic
Transients Program Manual (EMTP Theory
Book)", Portland, OR: Bonneville Power
Administration, 1986
[3] FURNAS, "Transitorios Eléctricos E
Coordenação De Isolamento", 1987
[4] HØIDALEN H. K., "Graphical Preprocessor
To ATP", 2008
[5] MARTÍNEZ J. A., GUSTAVSEN B.,
DURBAK D., "Parameter Determination For
Modeling System Transients—Part I:
Overhead Lines", IEEE Transactions On
Power Delivery, VOL. 20, NO. 3, JULY 2005,
pp. 2038-2044
[6] ISA, “Registro De Falla De La Subestación
Esmeralda”. 26/12/2010.
[7] ISA, “Registro De Falla De La Subestación
Yumbo”. 26/12/2010.
[8] ISA, “Metodologíapara el mejoramiento del
proceso análisis de perturbaciones mediante
la incorporación de herramientas de
simulación de sistemas de potencia para el
equipo de análisis operativo de ISA”. 2007.

Diego Alejandro Tejada Arango

Nació en Medellín, Colombia. Se graduó de

Ingeniería Eléctrica en la Universidad Nacional de
Colombia sede Medellín y como Especialista en
Transmisión y Distribución de Energía Eléctrica en
la Universidad Pontificia Bolivariana.
Ha trabajado en la Universidad de Antioquia como
profesor en simulación de sistemas de potencia, en
Ingeniería Especializada S.A. como ingeniero
consultor y actualmente se desempeña como
Analista Operativo en Interconexión Eléctrica S.A.
Sus campos de trabajo incluyen los estudios
eléctricos en sistemas de potencia y su planeación.

Sadul Segundo Urbaez Gonzalez

Nació en Fonseca, Colombia. Se graduó de

Ingeniería Eléctrica en la Universidad Industrial de
Santander (UIS) -Bucaramanga, Colombia, en
1994, y terminó sus estudios de maestría en
Potencia Eléctrica en La Universidad Industrial de
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 1113

Protecting Distribution Feeders

for Simultaneous Faults
Jorge Betanzos Manuel and Héctor E. Lemus Zavala, Comisión Federal de Electricidad, Mexico
Eliseo Alcázar Ramírez, David Sánchez Escobedo, and Héctor J. Altuve, Schweitzer Engineering Laboratories, Inc.

Abstract—Overhead distribution systems may experience
faults involving more than one feeder. During simultaneous
faults, the transformer low-voltage-side overcurrent relay
measures a current greater than the current measured by faulted
feeder relays. Therefore, the transformer relay may trip faster
than faulted feeder relays. Transformer relay misoperation
affects service availability in circuits not involved with the fault.
In this paper, we describe the causes of simultaneous faults on
distribution feeders and discuss overcurrent protection
coordination problems caused by these faults. We then propose
low-cost protection schemes using multifunction relays with
communications and logic programming abilities. We summarize
the operation experience of 19 simultaneous fault protection
schemes installed in several substations of two Comisión Federal
de Electricidad (CFE) distribution divisions in Mexico. Finally,
we analyze the operation of a protection scheme for an actual
simultaneous fault.
I. INTRODUCTION
In radial distribution substations, feeder relays typically
include instantaneous and inverse-time overcurrent elements.
The transformer low-voltage-side relay provides backup for
feeder faults and typically includes inverse-time overcurrent
elements. Utilities normally use automatic reclosing of
overhead feeder breakers. The transformer low-voltage-side
breaker lacks automatic reclosing.
analyze the operation of a protection scheme for an actual
fault involving two feeders of the Oaxaca Uno Substation
located in Oaxaca de Juárez, Oaxaca, Mexico.
II. SIMULTANEOUS FAULTS
The need to improve service availability has increased the
complexity of distribution network topology. Disconnect
switches allow transferring loads to alternate sources under
emergency conditions. Limitations on the rights of way make
it necessary to use multicircuit overhead lines or single-circuit
lines that run close to each other. As a result, simultaneous
faults involving more than one circuit are becoming quite
common. Typical causes of simultaneous faults include:
• Multicircuit lines or lines sharing the same right of
way
• Switching operations
• Thunderstorms
Fig. 1 shows a distribution system with four feeders and
normally open (NO) disconnect switches between adjacent
feeders.
B1
Feeder 1
T
S1 (NC)
S12 (NO)

T
For feeder faults, the faulted feeder relay and transformer B2
Feeder 2
T
relay measure practically the same current. The feeder relays BT
are set to operate faster than the transformer low-voltage-side S2 (NC)
T

B3 S23 (NO)
relay to trip only the faulted feeder. However, for Feeder 3
T
simultaneous feeder faults, the current measured by the
S3 (NC)
transformer relay is greater than the current measured by each
T

B4 S34 (NO)
faulted feeder relay. The transformer relay may operate faster Feeder 4
T
than the faulted feeder relay and undesirably trip the
S4 (NC)
transformer low-voltage-side breaker. Transformer breaker
misoperation affects service to the loads of healthy feeders. Fig. 1. Operation of tie disconnect switches may cause simultaneous faults
Given the growing incidence of simultaneous faults, in distribution systems.
Comisión Federal de Electricidad (CFE), the Mexican national
Under normal conditions, the system shown in Fig. 1
electric utility, decided to apply simultaneous fault protection
operates radially. For a permanent fault on Feeder 2 between
schemes in distribution substations several years ago. For
Breaker B2 and the normally closed (NC) Disconnect
example, the CFE Southeastern Distribution Division has
Switch S2, Breaker B2 trips and recloses to lockout. Operation
18 schemes in operation, and the CFE Jalisco Distribution
personnel open Disconnect Switch S2 to isolate the fault and
Division recently commissioned one scheme.
then close either Disconnect Switch S12 or Disconnect Switch
In this paper, we discuss the overcurrent protection
S23 to restore service to the Feeder 2 load connected beyond
coordination problems caused by simultaneous faults. We
Disconnect Switch S2. When remote access to motor-operated
describe two types of simultaneous fault protection schemes
disconnect (MOD) switches is available, the system operator
for distribution substations and summarize the operation
can send control commands to MOD switches from the
experience of the schemes installed in several substations of
distribution system dispatch center, in which case, service
the CFE distribution divisions mentioned above. Finally, we
114

restoration may take minutes. When remote access is not
available, the system operator dispatches field personnel to
manually perform the switching operations, in which case,
service restoration may take hours.
Because switching operations are infrequent events, it is
necessary to periodically close and open the NO disconnect
switches as a preventive maintenance operation. This
operation may cause a simultaneous fault in two ways:
• The disconnect switch fails during the test.
• A feeder fault occurs while the disconnect switch is
closed.
III. RELAY COORDINATION PROBLEMS
For a feeder fault in a radial distribution system, the
transformer low-voltage-side overcurrent relay and the faulted
feeder overcurrent relay measure practically the same current
(see Fig. 2).
plus load currents from unfaulted feeders. The relay of each
faulted feeder measures only the feeder fault current. Hence,
the transformer relay inverse-time overcurrent element may
trip faster than or simultaneously with the feeder relay
inverse-time overcurrent element. Transformer relay
misoperation for simultaneous feeder faults disconnects the
faulted and healthy feeders. All of the loads fed by the
transformer lose service for permanent or temporary faults,
because the transformer low-voltage-side breaker lacks
automatic reclosing. Operation personnel must travel to the
substation to manually reclose the transformer breaker, in
which case, service restoration may take hours, even for a
temporary fault.

Fig. 2. For a feeder fault, the transformer relay and faulted feeder relay
measure practically the same current.

Fig. 3 shows an example of coordination between the

inverse-time overcurrent elements of the transformer and
feeder relays. For feeder faults, the inverse-time overcurrent
elements must coordinate for all possible fault current values.
The typical coordination time interval (CTI) is 0.2 to
0.4 seconds. When both elements have the same type of time- Fig. 3. Coordination of inverse-time overcurrent elements for feeder faults.
current curve, the minimum separation between the curves
occurs for the maximum fault current value. In the
coordination example shown in Fig. 3, the curve separation
equals a CTI of 0.3 seconds for a maximum fault current of
6,750 A. The transformer relay overcurrent element must also
protect the transformer against through faults. Hence, the
time-current curve of the transformer relay overcurrent
element must be located between the feeder relay overcurrent
element curve and the transformer through-fault capability
curve, as shown in Fig. 3. The transformer relay actually
measures the sum of all feeder currents, including load. If load
is high enough, the transformer relay settings (pickup or time
dial) must be increased. The examples shown in Fig. 3, Fig. 5,
and Fig. 6 assume light load conditions.
For a simultaneous fault involving two or more feeders (see
Fig. 4. For a simultaneous fault, the transformer relay measures a current
Fig. 4), the transformer low-voltage-side relay measures the greater than the current measured by each faulted feeder relay.
total fault current (sum of the currents on all faulted feeders)
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Fig. 5 illustrates a protection coordination problem for the
simultaneous fault shown in Fig. 4. We assume that the feeder
inverse-time overcurrent elements have identical time-current
curves, and we disregard load currents for simplicity. For a
6,750 A simultaneous fault causing equal currents in both
feeders, IT = 6,750 A and I1 = I2 = 3,375 A. The operating time
of each feeder inverse-time overcurrent element is
0.42 seconds. The operating time of the transformer inverse-
time overcurrent element is 0.6 seconds. The operating time
margin is 0.6 – 0.42 = 0.18 seconds (smaller than the CTI).
The transformer relay may misoperate for this fault.
Fig. 5. The transformer relay inverse-time overcurrent element misoperates
for a simultaneous fault involving two feeders.
Feeder currents can be different for a simultaneous fault.
For example, a fault may result in IT = 6,750 A,
I1 = 4,850 A, and I2 = 1,900 A. For this fault, the operating
times are 0.35 seconds for the Feeder 1 relay, 0.6 seconds for
the Feeder 2 relay, and 0.6 seconds for the transformer relay
(see Fig. 5). Operating time margins are
0.6 – 0.35 = 0.25 seconds for the Feeder 1 relay and
0.6 – 0.6 = 0 seconds for the Feeder 2 relay. The transformer
relay may misoperate for this fault. In this example, the
Feeder 1 breaker trips first, causing the Feeder 2 current to
increase (this feeder now carries the total fault current). The
Feeder 2 relay will actually trip in less than 0.6 seconds, but
not fast enough to prevent transformer breaker misoperation.
Feeder overcurrent relays typically include instantaneous
overcurrent elements. The transformer and feeder overcurrent
relays coordinate well for simultaneous faults that cause
feeder currents greater than the pickup setting of the feeder
instantaneous overcurrent elements. In Fig. 6, for example, the
115
feeder instantaneous overcurrent elements are set to 3,000 A.
For a 6,750 A simultaneous fault causing currents of 3,375 A
in both faulted feeders, the feeder relays trip instantaneously,
and the transformer inverse-time overcurrent element does not
operate.
Fig. 6. Feeder instantaneous overcurrent elements ensure coordination only
for simultaneous faults that they can detect.
However, for line-end faults or resistive faults, feeder
currents can be smaller than the pickup setting of the feeder
instantaneous overcurrent elements. A coordination problem
may occur, because only the feeder inverse-time elements
detect the fault. In the Fig. 6 example, for a 5,000 A
simultaneous fault causing currents of 2,500 A in both faulted
feeders, the operating time of the feeder inverse-time
overcurrent elements is 0.49 seconds. The operating time of
the transformer inverse-time overcurrent element is
0.62 seconds. The operating time margin is
0.62 – 0.49 = 0.13 seconds. The transformer relay may
misoperate for this fault.
IV. PROTECTION SCHEMES FOR SIMULTANEOUS FAULTS
A solution to the coordination problem for simultaneous
feeder faults is to detect the simultaneous fault condition and
accelerate tripping of the faulted feeder relays to preserve
coordination. A small time delay (typically 3 to 6 cycles)
provides security for inrush feeder currents caused by cold-
load restoration. Two types of schemes are possible:
• Distributed
• Centralized
116

In distributed schemes, the simultaneous fault protection
logic resides in the faulted feeder relays. In centralized
schemes, the logic may reside in the transformer relay or logic
processor. In any scheme, the devices must have
communications and logic programming abilities.
Simultaneous fault protection schemes require
communication between the devices. Fig. 7 shows two
methods of communication between the devices used in the
scheme. The method shown in Fig. 7 (a) consists of wiring a
relay output contact to a logic input of a relay or logic
processor. An advantage of this method is that relays from
different manufacturers can be used in the scheme with no
additional equipment. The other method, shown in Fig. 7 (b),
uses direct digital communication between devices over
copper wire or fiber-optic cable. An advantage of this method
is that the relays and logic processor may continuously
monitor the communications channel condition and issue an
alarm in case of problems. This method can be applied with
relays from different manufacturers by adding remote I/O
modules to the scheme.
OUT101 IN101
Transmit
detectors) to identify the faulted feeders. Pickup current
settings of the 51P and 51G fault detectors should be equal to
the settings of the phase and ground inverse-time overcurrent
elements of the corresponding feeders. For a simultaneous
fault involving Feeder 1 and Feeder 2 (see Fig. 4), the 51P
and/or 51G fault detectors of the Feeder 1 relay and Feeder 2
relay operate, and OR Gate 1 asserts. Each relay sends the OR
gate output bit to one of the adjacent relays, and this
information is sequentially communicated to all of the feeder
relays. In the Feeder 1 and Feeder 2 relays, the OR Gate 2 and
AND gate assert to declare a simultaneous fault (bit SV1
asserts). After a security delay (TPU), the timer asserts bit
SV1T. A typical TPU setting is 3 to 6 cycles. The timer reset
time (TDO) must be greater than the total fault-clearing time.
A typical TDO setting is 9 cycles. The bit SV1T assertion
initiates the feeder breaker tripping. The almost instantaneous
breaker operation at the faulted feeders guarantees
coordination with the transformer low-voltage-side relay.
The faulted feeder relays also initiate reclosing of the
faulted feeder breakers. The reclosing times of breakers in
double-circuit lines should be different so that they reclose
sequentially. If the first feeder breaker recloses successfully,
the second feeder breaker is allowed to reclose. For permanent
faults, the first feeder breaker recloses and trips again, and the

Receive

Copper
first feeder relay issues a reclosing-blocking signal to the
Relay Output Relay Logic second feeder relay to prevent reclosing of the second feeder
Contact Input
breaker onto a fault.
Relay 1 Relay 2 The simultaneous fault protection scheme clears faults in
(a) 3 to 6 cycles plus the breaker operating time. The fault-
clearing time is comparable to instantaneous tripping, even for
faults that fall out of the reach of the feeder relay
instantaneous overcurrent elements. In addition, this logic
discriminates between single and simultaneous faults and
TMB1A RMB1A issues an alarm for simultaneous faults (not shown in Fig. 8).
Transmit

Receive

Copper
or The alarm helps operation personnel to quickly and safely
Relay Relay
Transmitted Bit Optical Received Bit restore service to the loads.
Fiber
The logic of the distributed scheme using hardwired
Relay 1 Relay 2
connections is similar to the logic shown in Fig. 8. This
(b)
scheme requires wiring the output contact of each feeder relay
Fig. 7. Two methods of communication between the devices include to the logic inputs of all other feeder relays. The resulting
(a) wiring a relay output contact to a logic input of another device and (b) scheme is more complex and less reliable than the scheme in
direct digital communication between the devices over copper wire or fiber-
optic cable. Fig. 8. In addition, the relays cannot supervise the condition of
the copper wires.
Fig. 8 depicts the logic diagram of a distributed scheme for The distributed simultaneous fault protection scheme is
four feeders using direct digital communication. Feeder relays easy to apply as an enhancement to existing installations.
communicate via copper wires or fiber-optic cable in a looped When existing relays have logic programming abilities,
scheme, and each feeder relay communicates with two implementing the scheme requires only wiring and relay
adjacent feeder relays. The scheme uses phase (51P) and programming.
ground (51G) instantaneous overcurrent elements (fault
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 117

A B C D

RMB1A

RMB2A

RMB3A

RMB4A
Phase
Fault 51P Reclose
79RI
Detector Initiate
1 TPU
SV1 SV1T
Ground TDO
Fault 51G 2 To Tripping
TR
Detector Logic

TMB1A

TMB2A

TMB3A

TMB4A
Feeder 1 Relay

RMB1A

RMB2A

RMB3A

RMB4A
Phase
Fault 51P Reclose
79RI
Detector Initiate
1 TPU
SV1 SV1T
Ground TDO
Fault 51G 2 To Tripping
TR
Detector Logic
TMB1A

TMB2A

TMB3A

TMB4A

Feeder 2 Relay
RMB1A

RMB2A

RMB3A

RMB4A

Phase
Fault 51P Reclose
79RI
Detector Initiate
1 TPU
SV1 SV1T
Ground TDO
Fault 51G 2 To Tripping
TR
Detector Logic
TMB1A

TMB2A

TMB3A

TMB4A

Feeder 3 Relay
RMB1A

RMB2A

RMB3A

RMB4A

Phase
Fault 51P Reclose
79RI
Detector Initiate
TPU
1 SV1 SV1T
Ground TDO
Fault 51G 2
TR
To Tripping
Detector Logic
TMB1A

TMB2A

TMB3A

TMB4A

Feeder 4 Relay

A B C D

Fig. 8. Logic diagram of a distributed simultaneous fault protection scheme using direct digital relay-to-relay communication.
118

Phase
Timer 1
Fault 51P 79RI Reclose
Detector Initiate
TPU
TMB1A R1P1 SV1 SV1T T1P1 RMB1A
1
Ground TDO
Fault 51G 1 To Tripping
TR
Detector Logic

Feeder 1 Relay Feeder 1 Relay

Phase
Fault 51P Timer 2 Reclose
79RI
Detector Initiate
TPU
TMB1A R1P2 SV2 SV2T T1P2 RMB1A
2
Ground TDO
Fault 51G 2 To Tripping
TR
Detector Logic

Feeder 2 Relay Feeder 2 Relay

Phase
Timer 3
Fault 51P 79RI Reclose
Detector Initiate
TPU
TMB1A R1P3 SV3 SV3T T1P3 RMB1A
3
Ground TDO
Fault 51G 3 To Tripping
TR
Detector Logic

Feeder 3 Relay Feeder 3 Relay

Phase
Timer 4 Reclose
Fault 51P 79RI
Detector Initiate
TPU
TMB1A R1P4 SV4 SV4T T1P4 RMB1A
4
Ground TDO
Fault 51G 4 To Tripping
TR
Detector Logic

Feeder 4 Relay Transformer Relay or Logic Processor Feeder 4 Relay

Fig. 9. Logic diagram of a centralized simultaneous fault protection scheme using direct digital communication between the devices.

Fig. 9 depicts the logic diagram of a centralized scheme for The centralized simultaneous fault protection scheme
four feeders using direct digital communication between the concentrates all fault information in one device. This device
devices. Feeder relays communicate radially with the may provide sequential event reporting, which facilitates fault
transformer relay or logic processor via copper wires or fiber- analysis. When the scheme uses a logic processor, the
optic cables. For a simultaneous fault involving Feeder 1 and processor can provide additional functions, such as fast bus
Feeder 2 (see Fig. 4), the fault detectors 51P and/or 51G and tripping, breaker failure protection, and automatic restoration
the OR gate of these relays assert. Each relay transmits bit of unfaulted transformers [1].
TMB1A to the transformer relay or logic processor, where
received bits R1P1 and R1P2 assert. As a result, OR Gate 1, V. FIELD OPERATION EXPERIENCE
OR Gate 2, AND Gate 1, and AND Gate 2 assert in the Table I summarizes the simultaneous fault protection
transformer relay or logic processor. After a security delay, schemes operating in the CFE Southeastern Distribution
Timer 1 asserts bit SV1T, and Timer 2 asserts bit SV2T. The Division and the CFE Jalisco Distribution Division. The first
transformer relay or logic processor then transmits bit T1P1 to scheme was commissioned in 2003.
the Feeder 1 relay and bit T1P2 to the Feeder 2 relay to initiate
TABLE I
tripping and sequential reclosing of the faulted feeder SIMULTANEOUS FAULT PROTECTION SCHEMES
breakers. OPERATING IN TWO CFE DISTRIBUTION DIVISIONS
The logic of a centralized scheme using hardwired
Southeastern Jalisco
connections is similar to the scheme shown in Fig. 9. Scheme Type Distribution Distribution Total
However, the scheme is more complex and less reliable, Division Division
because two copper wires run between each feeder relay and
Distributed 17 0 17
the transformer relay or logic processor. In addition, the relays
cannot supervise the condition of the copper wires. Centralized 1 1 2
Total 18 1 19
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Fig. 10. Simplified one-line diagram of the Oaxaca Uno Substation.
These schemes have operated correctly for all
46 simultaneous faults that have occurred on feeders of both
distribution divisions. The causes of the faults are as follows:
• Faults in double-circuit lines: 20
• Thunderstorms: 19
• Faults during circuit-looped operation: 7
No scheme misoperations have occurred for faults
involving only one feeder, cold load pickup conditions, or
other abnormal conditions.
VI. EXAMPLE OF SCHEME OPERATION FOR AN ACTUAL FAULT
A. Scheme Operation Analysis
The Oaxaca Uno Substation (see Fig. 10), located in the
city of Oaxaca de Juárez, state of Oaxaca, Mexico, has a
12/16/20 MVA, 115/13.8 kV transformer and an
18/24/30 MVA, 115/13.8 kV transformer. Each transformer
feeds four radial feeders. This substation has a centralized
simultaneous fault protection scheme using a protection
processor and direct digital communication between the
devices.
During a severe thunderstorm on May 18, 2009, a
temporary phase-to-phase simultaneous fault occurred on
119
distribution feeders OAX-4010 and OAX-4020. The fault
current contributions were 2,001 A on the OAX-4010 feeder
and 823 A on the OAX-4020 feeder. The prefault demands
were 5.9 MVA on the OAX-4010 feeder, 6.8 MVA on the
OAX-4020 feeder, and 28.9 MVA on the transformer.
From the sequential event report of the OAX-4010 feeder
relay (see Fig. 11), we conclude the following:
• The phase fault detector 51P operated at
20:29:16.918 hours.
• The feeder relay transmitted fault detection
information (bit TMB1A asserted) to the logic
processor at 20:29:16.918 hours.
• The feeder relay received a tripping command (bit
RMB1A asserted) from the logic processor at
20:29:17.051 hours.
• The feeder relay sent a trip signal to the breaker (bit
TRIP asserted) at 20:29:17.055 hours.
• The breaker opened (bit 52A deasserted) at
20:29:17.130 hours. Breaker operating time is
75 milliseconds (4.5 cycles).
The sequential event report of the OAX-4020 feeder relay
is almost identical to the report shown in Fig. 11.
120
Fig. 11. Sequential event report of the OAX-4010 feeder relay.
From the sequential event report of the logic processor
(Fig. 12), we conclude that the processor:
• Received fault detection information from the
OAX-4010 feeder relay (bit R1P1 asserted) at
20:29:16.941 hours.
• Received fault detection information from the
OAX-4020 feeder relay (bit R1P2 asserted) at
20:29:16.953 hours.
• Declared a simultaneous fault involving the
OAX-4010 feeder (bit SV5 asserted) at
20:29:16.953 hours.
Fig. 13. Oscillogram recorded by the OAX-4010 feeder relay.
• Declared a simultaneous fault involving the
OAX-4020 feeder (bit SV6 asserted) at
20:29:16.953 hours.
• Sent a tripping signal to the OAX-4010 feeder relay
(bit T1P1 asserted) at 20:29:17.053 hours when the
timer expired (bit SV5T asserted). Timer pickup
setting is 100 milliseconds (6 cycles).
• Sent a tripping signal to the OAX-4020 feeder relay
(bit T1P2 asserted) at 20:29:17.053 hours when the
timer expired (bit SV6T asserted). Timer pickup
setting is 100 milliseconds (6 cycles).
Fig. 12. Sequential event report of the logic processor.
From the oscillogram recorded by the OAX-4010 feeder
relay (Fig. 13), we conclude the following:
• The fault started on Cycle 3.0.
• The phase fault detector 51P operated on Cycle 3.75.
• The feeder relay closed its contact (OUT 12 asserted)
on Cycle 12.
• The scheme operating time was 9 cycles.
• The total fault-clearing time was 13.5 cycles (given a
4.5-cycle breaker operating time).
The oscillogram recorded by the OAX-4020 feeder relay is
almost identical to the oscillogram shown in Fig. 13.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
B. Cost Analysis
For the actual temporary fault discussed previously, the
simultaneous fault protection scheme avoided transformer
breaker misoperation, and the faulted feeder breakers
successfully reclosed.
This fault could have caused the transformer low-voltage-
side breaker to misoperate if the Oaxaca Uno Substation
lacked a simultaneous fault protection scheme. Transformer
breaker misoperation would have caused all transformer loads
to lose service. Operation personnel would have had to travel
to the substation to manually reclose the transformer breaker.
Assuming a transformer breaker misoperation for this fault,
we can determine the cost of nonserved energy. When the
fault occurred, the transformer load was 28.9 MVA, or
27.455 MW at a 0.95 power factor. If service restoration time
equals 1 hour, the amount of nonserved energy is 27,455 kWh.
Assuming an energy price of Mex$1.09/kWh, the cost of
nonserved energy is Mex$29,926. If the cost of personnel
travelling to the substation equals Mex$2,600, the total
economic loss resulting from one fault that causes transformer
breaker misoperation is Mex$32,526 (equivalent to
US$2,502). The actual economic loss may be higher if the
utility has to pay penalties for service interruption.
VII. CONCLUSIONS
From the results presented in this paper, we conclude:
• Causes of simultaneous faults involving two or more
overhead distribution feeders include multicircuit lines
or lines sharing the same right of way, switching
operations, and thunderstorms.
• Simultaneous faults may cause misoperation of the
time-delayed overcurrent elements of the transformer
low-voltage-side relay.
• Simultaneous fault protection schemes prevent
transformer low-voltage-side breaker misoperations.
These schemes improve service quality by avoiding
unnecessary service interruptions to the unfaulted
feeders.
• In distributed schemes, the simultaneous fault
protection logic resides in the feeder relays; in
centralized schemes, the logic resides in the
transformer low-voltage-side relay or logic processor.
• The CFE Southeastern Distribution Division and the
CFE Jalisco Distribution Division have
19 simultaneous fault protection schemes in operation;
these schemes have correctly cleared all
46 simultaneous faults. No scheme misoperations have
occurred so far.
VIII. REFERENCE
[1] D. Sánchez, E. Alcázar, O. Márquez, H. Altuve, and A. Martínez,
“Multifunction Relays and Protection Logic Processors in Distribution
Substation Applications.” Available: http://www.selinc.com.
121
IX. BIOGRAPHIES
Jorge Betanzos Manuel received his BSEE degree in Electrical and
Industrial Engineering from the Oaxaca Technological Institute in 1983. He
joined Comisión Federal de Electricidad (CFE) in 1984 as a construction
supervisor in the Southeastern Distribution Division (SDD). From 1986 until
1993, he was a protection, control, and metering engineer at the Istmo and
Malpaso Transmission Zones of the CFE Southeastern Transmission Region.
From 1993 until 1997, Mr. Betanzos was head of the Protection Office of the
Oaxaca Distribution Zone. From 1997 until 2000, he was head of the Office
for High-Voltage Network Studies of the SDD Planning Department. Since
2000, Mr. Betanzos has worked as head of the Protection Department of the
CFE SDD. He leads projects on power quality monitoring and modernizing
protection, control, and metering panels in the CFE SDD.
Héctor E. Lemus Zavala received his BSEE degree in Electromechanical
Engineering from the University of Guadalajara, Mexico. In 1999, he joined
Comisión Federal de Electricidad (CFE), where he was supervisor of the
Jalisco Distribution Division (JDD). Since 2007, he has worked as head of the
Protection Office of the CFE JDD in Guadalajara, Jalisco, Mexico. His
activities include supervision, maintenance, improvement, and commissioning
of protection and control systems for distribution substations. He was a
member of the Committee of Distribution Protection Specialists and the
Permanent Protection Committee (CFE Transmission). Mr. Lemus is currently
a member of the Regional Protection Committee of the CFE Operation
Division.
Eliseo Alcázar Ramírez received his BSEE degree from the Oaxaca
Technological Institute in 1998. From 1999 until 2001, he was head of the
Protection, Control, and Metering Department in the Southeastern
Distribution Division (SDD) of Comisión Federal de Electricidad (CFE) in
Tehuantepec, Mexico. From 2001 until 2004, he was head of the Protection
Office of the CFE SDD. During this time, he was engaged in activities of
supervision, maintenance, improvement, and commissioning of protection,
control, and metering systems. His expertise includes fault analysis, short-
circuit studies, protection coordination, and protection system design. In April
2004, Mr. Alcázar joined Schweitzer Engineering Laboratories, Inc. (SEL),
where he is currently a protection engineering supervisor in San Luis Potosí,
Mexico. His activities include protection, control, and metering system design
and commissioning, as well as technical support and training on SEL products
for engineers from utilities and industrial plants.
David Sánchez Escobedo received his BSEE degree in 1994 from the
University of Guanajuato, Mexico, and his M.Sc. degree in 2005 from the
University of Guadalajara, Mexico. From 1994 until 1998, he was head of the
Protection and Metering Office in the Western Transmission Area of
Comisión Federal de Electricidad (CFE) in Guadalajara, Jalisco, Mexico.
Mr. Sánchez served on the faculty of the Autonomous University of
Guadalajara in 1998. From 1998 until 2000, he worked for INELAP-PQE in
Guadalajara, Mexico, as a protection system design engineer. In September
2000, Mr. Sánchez joined Schweitzer Engineering Laboratories, Inc., where
he is currently the electrical engineering manager in San Luis Potosí, Mexico.
He has authored and coauthored several technical papers.
Héctor J. Altuve received his BSEE degree in 1969 from the Central
University of Las Villas in Santa Clara, Cuba, and his Ph.D. in 1981 from
Kiev Polytechnic Institute in Kiev, Ukraine. From 1969 until 1993, Dr. Altuve
served on the faculty of the Electrical Engineering School at the Central
University of Las Villas. From 1993 to 2000, he served as professor of the
Graduate Doctoral Program in the Mechanical and Electrical Engineering
School at the Autonomous University of Nuevo León in Monterrey, Mexico.
In 1999 through 2000, he was the Schweitzer Visiting Professor in the
Department of Electrical Engineering at Washington State University.
Dr. Altuve joined Schweitzer Engineering Laboratories, Inc. in January 2001,
where he is currently a distinguished engineer and director of technology for
Latin America. He has authored and coauthored more than 100 technical
papers and holds three patents. His main research interests are in power
system protection, control, and monitoring. Dr. Altuve is an IEEE senior
member and an IEEE Power Engineering Society distinguished lecturer.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 123

Communications for the Smart Grid

Mark Adamiak Peter Sanza
GE Digital Energy GE Energy
Multilin Nuclear

Abstract

As the smart grid drives into the main stream of the utility enterprise, it becomes incumbent on the industry to iden-
tify an architecture based on what is the smart grid, what are the communication pieces involved, and how do they
fit together. The “pieces” are the existing communication standards. The choice of a standard, however, is not a
random process. There is an engineering process for the selection of relevant standards and subsequent migration to
an Architecture. Such a process was funded by the Electric Power Research Institute and the output of this process
is the IntelliGrid architecture [1]. This paper examines the architecture process of identifying the system require-
ments and the subsequent process of linking the requirements with candidate standards. Finally, the standards al-
ready chosen by the NIST as Smart Grid standards are presented.

IntelliGrid Enterprise Activities

In all cases, an architecture must be based on the functions it is required to perform. In order to identify these func-
tions, a mechanism known as a Use Case was employed. A use case starts with a narrative that describes a specific
smart function in the environment of interest. Distillation of the use case identifies data items and their movement
through the environment under study. In the energy environment, there are multiple areas of interest. In order to
facilitate categorization of use cases, the energy environment was broken down into 6 primary business functions,
namely:: Market Operations, Transmission Operations, Distribution Operations, Primary Generation, Distributed
Energy Resources, and Customer Services. Over 400 potential use cases were identified and the most demanding of
there were elaborated in additional detail to construct a complete high-level set of requirements for the communica-
tions infrastructure. The requirements were further categorized as follows:

• Communication configuration requirements, such as one-to-many, mobile, WAN, LAN, etc.

• Quality of service and performance requirements, such as availability, response timing, data accuracy, etc.
• Security requirements, such as authentication, access control, data integrity, confidentiality, non-
repudiation, etc.
• Data management requirements, such as large databases, many databases particularly across organizational
boundaries, frequent updates, etc.
• Constraints and concerns related to technologies, such as media bandwidth, address space, system compute
constraints, legacy interface, etc.
• Network management requirements, such as health and diagnostics of infrastructure and equipment, remote
configuration, monitoring and control, etc.

As an example in this paper, the Demand Response use case is reviewed. The IntelliGrid Architecture considered
the Demand Response system as part of the Customer Services functional area. While it is clear that Demand Re-
sponse functionality operates within this domain, it is important to note that Demand Response is not an isolated
island of functionality. The entire premise of the IntelliGrid Architecture is that each of these envisioned applica-
tions must interact with other domains and functional areas within the Energy industry. Interoperability between and
among other Demand Response systems and other Energy industry applications can be seen as one of the key driv-
ers behind the IntelliGrid Architecture.

Given that a communication channel will exist into the home, commercial, or industrial electrical grid, the Intelli-
Grid Architecture identified a number of applications that directly touch the Demand Response system. The com-
124

plete list can be found on the IntelliGrid Architecture website, but Customer Domain specific functions are listed
here as follows[2]:

1. Automatic meter reading (AMR)

• Sub-metering
• Load Monitoring
• Sub-contracted metering
• Energy usage display
• Demand profiles
• Auto-pay / Pre-pay metering
• Outage detection and isolation
• Remote connect/disconnect
• Measurement of customer outage minutes/hours
2. Customer trouble call management
3. Real-time Pricing (RTP)
• Day ahead schedule
• Hour ahead emergency condition
• Automatic in-home load curtailment
• Available by-pass mode
4. Load management
• Direct Load Control under emergency conditions
• DER Watt/VAR dispatch
5. Building/Home Energy Management Services
• Building management
• Building security
• Customer remote access
• Customer energy bidding
• Load analysis
• Equipment monitoring (e.g. clogged air filters, failed water heater element, etc.)
• Occupancy based heating and lighting controls
• Home insulation level analysis
6. Electric Car as Generation Source
7. Weather
• In-home weather forecasts
• In-home lightning and severe weather alert
• Lightning location report

In addition, the customer communications infrastructure will enable other IntelliGrid “cross domain” activities such
as:

• Feeder Voltage Optimization

• Downed conductor detection
• Faulted feeder isolation / feeder re-deployment
• Distributed Energy control and isolation
• Distribution based VAR support to transmission
• Distribution SCADA
• Micro-grid establishment / control

IntelliGrid Demand Response Environments

Each of the myriad interrelated functions defines its own set of detailed functional and non-functional requirements.
An architecture is not, however, intended to simply fulfill a patchwork of requirements. The architecture is not simp-
ly the union of the lists of detailed requirements for each function. Functions often have conflicting requirements
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 125

and a good architecture must be flexible enough to accommodate such incongruous anomalies. To realize this, the
IntelliGrid Architecture invented what were called “Environments”.

An IntelliGrid Architecture Environment is defined as an information environment, where the information exchang-
es of power system functions have essentially similar architectural requirements, including their configuration re-
quirements, quality of service requirements, security requirements, and data management requirements. These Envi-
ronments reflect the requirements of the information exchanges, not necessarily the location of the applications or
databases (although these may affect the information exchanges and therefore the environment). Since functions can
have multiple types of information exchanges, these functions often operate across multiple Environments.

The IntelliGrid Architecture defined twenty-one Environments that completely describe the communication re-
quirements for the information exchanges as shown in the following figure[3]:

Figure 1: IntelliGrid Environments

The IntelliGrid Architecture defines 21 Environments that span the entire Electric Energy Enterprise

Demand Response and all of the ancillary services it provides, enables, or directly touches, operates in several of
these environments. A brief synopsis of the relevant environments and typical applications follows[3]:
126

Environment 9: Control Centers to ESPs embodies the requirements for communica-

tions between utility control centers and Energy Service Providers or “aggregators”.
Typical applications in this environment include Real-time pricing negotiations, ag-
gregated customer metering and settlements as well as data mining for later use in
market operations. This is characterized as a business-to-business (B2B) environment
ESP
having very strict requirements for security and a contractual level of timeliness and
record keeping (and audit trails).

Environment 11: Control Center to Customers encompasses the requirements

for what has traditionally been termed commercial or industrial metering and
includes the requirements for any data exchange that goes directly between
control centers and customer sites. Typical applications include metering of
large customers, control of distributed energy resources, and load curtailment.
This is characterized as a B2B environment but involving operational rather
than financial data.

Environment 15: DER Monitoring and Control encompasses the communication re-
quirements between distributed energy resources and the organizations that must moni-
tor and operate them. Typical applications include an ESP aggregating data from or
operating multiple small generators including renewable power, small hydro, co-
generation or any other mini-grid. This may also include demand response and adjust-
ment of power quality. This environment is characterized by the fact that it is vital that
these message exchanges are not tampered with, monitored, or interfered with by unau-
thorized persons.

Environment 16: Intra-Customer Site defines requirements for communications that are local
to customer sites (residential, commercial, or industrial). Typical applications include a cus-
tomer printing processes online or offline in response to RTP, a customer locally managing a
distributed energy resources in response to external factors (environment, price, regulations,
etc.). Additionally this also includes building and home automation systems. This environ-
ment is characterized by critical data with local scope and limited impact on the overall power
grid. Data is real-time potentially peer-to-peer with application specific security.

Environment 17: Inter-Customer Sites captures requirements for communications

between customer sites (residential, commercial, or industrial). This is a relatively
new environment that is not widely deployed yet but would include typical appli-
cations such as coordination of customer loads (e.g. A/C or refrigeration compres-
sor startup) and mini-grid management. It is characterized by high security re-
quirements because data crosses organizational boundaries.

Environment 18: Customer to ESP. This environment encompasses the communication, dis-
tributed processing, security, and data management requirements for applications that exist
between ESPs, typically the distribution level energy utility, and the end-use point-of-sale
energy customers like residential homeowners and apartment dwellers. Typical applications
include automatic (or manual) meter reading, monitoring and control of DER, demand re-
ESP sponse and RTP. It is characterized by extremely large volumes of data with frequent config-
uration and topology changes to the communications network. Commands are often broadcast
due to the large number of end users but address space must be massively scalable.

IntelliGrid Requirements for Demand Response Implementations

Based upon the above Environments, the IntelliGrid Architecture proposes a base set of high level requirements for
Demand Response Systems that will also aid in achieving interoperability with other systems sharing the same infra-
structure. These requirements are categorized as follows [3]:
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 127

Configuration Requirements
• Support interactions between a few "clients" and many "servers"
• Support peer to peer interactions
• Support interactions across widely distributed sites
• Support the frequent change of configuration and/or location of end devices or sites
• Support multi-cast or broadcast capabilities
• Support interactions within a contained environment (e.g. substation or control center)

Quality of Service Requirements

• Provide medium speed messaging on the order of 10 seconds
• Support contractual timeliness (data must be available at a specific time or within a specific window of
time)
• Support medium availability of information flows of 99.0+% (~3.5 days/year outage)

Security Requirements
• Provide Identity Establishment Service (you are who you say you are)
• Provide Authorization Service for Access Control (resolving a policy-based access control decision to en-
sure authorized entities have appropriate access rights and authorized access is not denied)
• Provide Information Integrity Service (data has not been subject to unauthorized changes or these unauthor-
ized changes are detected)
• Provide Confidentiality Service (only authorized access to information, protection against eavesdropping)
• Provide Inter-Domain Security Service (support security requirements across organizational boundaries)
• Provide Non-repudiation Service (cannot deny that interaction took place)
• Provide Security Assurance Service (determine the level of security provided by another environment)
• Provide Audit Service (responsible for producing records, which track security relevant events)
• Provide Security Policy Service (concerned with the management of security policies)
• Provide Path and Routing Quality of Security (being able to determine a secure communication path)
• Provide Firewall Transversal
• Provide Privacy Service (the ability to ensure person information is not disclosed)
• Provide User Profile and User Management (combination of several other security services)
• Provide Security Protocol mapping (the ability to convert from one protocol to another)
• Provide Security Discovery (the ability to determine what security services are available for use)

Network and System Management Requirements

• Provide Network Management (management of media, transport, and communication nodes)
• Provide System Management (management of end devices and applications)
• Support extensive data validation procedures

Data Management Requirements

• Support the management of large volumes of data flows
• Support extensive data validation procedures
• Support keeping data consistent and synchronized across systems and/or databases
• Support timely access to data by multiple different users
• Support frequent changes in types of data exchanged
• Support management of data whose types can vary significantly in different implementations
• Support specific standardized or de facto object models of data
• Provide discovery service (discovering available services and their characteristics)
• Provide conversion and protocol mapping
• Support the management of data across organizational boundaries
128

IntelliGrid Design Principles [4]

In order to design an architecture, one must have “guiding principles” as to how to identify the pieces of the archi-
tecture and how they are to be put together. The IntelliGrid architecture identifies several such principles described
below.

One of the most important system integration principles in IntelliGrid is the concept of Technology Independent
Architecture (TIA). TIA is technology neutral or technology agnostic. It can insure successful integration of the var-
ious utility enterprise applications without requiring changes to the application’s internal operation. It can also
achieve high level of interoperability and interworkability with the built-in intelligence of auto-configuration and
self-discovery. Figure 2 illustrates the TIA framework.

Composite Data Mining

Portals Applications and Analysis

Utility
Local Legacy Field
Applications Devices
Common Services/Interfaces *

Discoverable
Information
Models

Wholesale and Networking

Retail Market and
Operations Computer
Hardware
* Includes: security,
object naming, ** Includes: domain
platform services Databases, Directories, Web Pages & Other File Types objects, security
(transactions, time, and Registries Documents (e.g. email, etc.) objects, managed
etc.) device objects, etc.

Figure 2 Technology Independent Architecture

The guiding principle of the IntelliGrid Architecture shows that Common Services, Common Information Models, and Generic
Interfaces enable scalable interoperability in a heterogeneous technology environment.

Three key information-modeling elements in TIA framework are,

• Common Services – These are atomic building blocks frequently required by the utility applications. Intel-
liGrid further breaks the common services down to four categories, namely “system and network manage-
ment services”, “data management and exchange services”, “common platform services”, “common securi-
ty services”.
• Common Information Models – These are common data that are exchanged between services and applica-
tions. This includes the suggested data format, data attributes and their relationships.
• Generic Interfaces – Generic Interfaces are used as the mechanism for exchanging Common Information
Model data between services. Generic Interfaces correspond to how data is exchanged. It specifies a set of
standard verbs such as “get”, “set”, “report”, which allows different applications to communicate with each
other.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 129

These common information-modeling elements are the key to achieving higher-level interoperability of power sys-
tem distributed information systems.

Common Services
Common Services are commonly defined functionality derived by identifying the crosscutting distributed infor-
mation requirements. Common Services allow components to be treated as black boxes. This facilitates greater flex-
ibility, as components are less dependent on how each works internally.

However, the use of Common Services does not by itself substantially reduce the complexity of dealing with differ-
ent platforms such as Java, .Net or Web Services. Also, Common Services do not necessarily deal with the disconti-
nuity of the meaning of data. Lastly, Common Services do not deal with the discontinuity caused by different data
access mechanisms such as “read/write data” or “subscribe to data”.

To overcome semantic heterogeneity a common information model is used as the common language that all services
use to communicate. To overcome platform heterogeneity, the generic interface is required. The generic interface
can be implemented on any platform. While the different implementations of the generic interface are not interoper-
able, “off the shelf”, the mapping from one platform specific implementation to another is simple and well known.

Common Information Models

In order to precisely describe the meaning of a set of terms, engineers often create an information model. An infor-
mation model describes a collection of related real world objects. An information model describes objects in terms
of classes, attributes and relationships and provides unique names and definitions to each object.

The EPRI/IEC Common Information Model (CIM) describes data typically used in the power system. The CIM con-
tains object types such as substations, breakers, and work orders as well as other data typically found in an EMS,
SCADA, DMS, or work, and asset management system. More recently, the CIM is being extended to include trans-
mission reservation and energy scheduling information. In general, the benefit of creating an information model
include:

• Models give context to data improving understanding and productivity.

• Models enable automation of setup and maintenance tasks.

Generic Interfaces
The mechanism used to exchange data is determined by an application’s interface. However, the native interface
provided by an application is typically limited. For example, typically existing interfaces:

• Do not expose data within the context of a common inter-application data model.
• Do not provide a means to discover what business object instances are serviced by a particular component
instance other than a rudimentary listing of legacy IDs (tags) that cannot be viewed within the context of an
inter-application data model such as a power system network model.

Without a means to discover what data an application processes, plug and play is nearly impossible to achieve. To
address these impediments to plug and play and the need for a common exchange mechanism, “Generic Interface” is
introduced to specify how data are exchanged. The phrase “Generic Interface” is an umbrella term for four interfac-
es types:

• An interface for mapping names to ID’s and visa versa.

• A request/reply oriented interface that supports browsing and querying randomly associated structured data
– including schema (class) and instance information.
• A publish/subscribe oriented interface that supports hierarchical browsing of schema and instance infor-
mation. This interface would typically be used as an API for publishing/subscribing to XML formatted
messages.
130

• Applications use the generic interfaces to connect to each other directly or to an integration framework
such as a message bus or data warehouse. A technology neutral interface allows applications to be designed
independently of the capabilities of the underlying infrastructure.

Generic interfaces provide the following key functionality required for creation of a plug and play infrastructure:

• Interfaces are generic and are independent of any application category and integration technology. This fa-
cilitates reusability of applications supporting these interfaces.
• Interfaces support schema announcement/discovery – The schemas are discoverable so that component
configuration can be done programmatically at run time. Programmatically exposing the schema of applica-
tion data eliminates a great deal of manual configuration.
• Interfaces support business object namespace presentation – Each component describes the business object
instances that it supports within the context of a common namespace shared among all applications such as
a power system network model like the EPRI Common Information Model (CIM). It is not enough to mere-
ly expose the application data schema, one must also expose what specific breakers, transformers, etc., that
an application operates on. This also eliminates manual configuration as well as provides a means for a
power system engineer to understand how enterprise data is organized and accessed.

IntelliGrid Recommended Implementation Technologies:

There are too many recommendations to capture in this project summary, but common themes can be identified as
follows[1]:

• Harmonize the existing common services, information models, and interfaces, as well as create new stand-
ards where they are needed, so the power industry speaks a common communications language of ‘nouns’
and ‘verbs’ that can be translated into different technologies. This is a key requirement for the higher levels
of system integration now emerging across the energy industry
• Integrate security, systems, network management, and technical development (i.e. new technologies),
which have too often been considered separate tasks.
• Unify technologies between power system automation networks, corporate networks, and inter-business
networks, again by linking them to common information models, services, and interfaces.
• Remember that developing an industry-level architecture is a process – not an end point. Requirements and
enabling technologies are constantly changing. Although the guiding principles should remain constant, in-
dividual solutions will change over time.

Based on the identified design principles, IntelliGrid makes a link from design guidelines to recommended technol-
ogies that best embody the stated principles and meet the identified requirements. IntelliGrid makes a point of rec-
ognizing that many needed technologies may not exist and encourages the identification and subsequent standardiza-
tion of such technologies. The list below is a first level summary of the “primary” recommended technologies for
the identified environments. The list is organized by functional layer. For the complete list of applicable technolo-
gies, please refer to the IntelliGrid.info website[3]:

Data Exchange:
• IEC61850 – Communication Networks and Systems in Substations
o Data Models
o Abstract Services
o Substation Configuration Language
• ANSI C12.19 Metering Tables
• AEIC Guidelines for Implementation of ANSI C12.19
• IEC61970 Part 3 Common Information Model (CIM)
• IEC61970 Part 4 Generic Interface Definition
• IEC61968 SIDM System Interfaces for Distribution Management
• IEC60870-6 Inter Control Center Protocol
• IEC62325 on Framework for Energy Market Communications
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 131

• NERC e-tagging
• NAESB OASIS for Market Transactions
• IEC62056 – Data Exchange for Meter Reading, Tariff, and Load Control
• Universal Description, Discovery, and Integration (UDDI)
• Simple Object Access Protocol (SOAP)
• EbXML
• XML Metadata Interchange (XMI)
• Meta Object Facility (MOF)
• Globally Unique Identifiers (GUID)
• S/NTP (Simple/Network Time Protocol)
• ANSI/ISO/IEC 9075 – Structured Query Language (SQL)

Security
• ISO/IEC 10164-8:1993 Security Audit Trail Function - Information technology - Open Systems Intercon-
nection - Systems Management - Security,
• ISO/IEC 18014-1:2002 Time-Stamping Services - Information technology - Security Techniques - Part 1:
Framework - Security, Data Management
• ISO/IEC 10181-7:1996 Security Audit and Alarms Framework - Information technology - Open Systems
Interconnection -- Security Frameworks for Open Systems - Security,
• FIPS PUB 112 Password Usage - Security,
• FIPS PUB 113 Computer Data Authentication - Security,
• RFC 1510 The Kerberos Network Authentication Service (v5)
• RFC 2196 Site Security Handbook - Security,
• RFC 2401 Security Architecture for the Internet Protocol - Security,
• RFC 2527 Internet X.509 Public Key Infrastructure Certificate Policy and Certification Practices Frame-
work - Security,

Transport:
• TCP / Internet Protocol IPV4 / IPV6

Network Management:
• Simple Network Management Protocol (SNMP)

Physical/Data Link:
• IEEE 802.x (LAN, WAN, WiFi, WiMax, Ethernet)
• SONET
• ATM

NIST Selected Smart Grid Standards – Rev 1.0

As part of the Energy Independence and Security Act of 2007, the North American Institute of Standards and Tech-
nology (NIST) was mandated by Congress to coordinate a “framework of protocols and model standards to achieve
interoperability of the Smart Grid”. As part of this mandate, NIST has recently released the first set of “accepted”
standards for use in Smart Grid communications [5]. It is to be noted that this is a work in progress and is not exclu-
sionary. The list of these standards follows closely to the recommendations made by the IntelliGrid document. The
list of selected standards is as follows:

•AMI-SEC System Security Requirements

•ANSI C12.19/MC1219 – Revenue Metering
•BACnet ANSI ASHRAE 135-2008/ISO 16484-5 – Building Automation
•DNP3 - Substation and feeder device automation
•IEC 60870-6 / TASE.2 - Inter-control center communications
132

•IEC 61850 - Utility automation and protection

•IEC 61968/61970 - Application level energy management system interfaces
•IEC 62351 Parts 1-8 - Information security for power system control operations
•IEEE C37.118 - Phasor measurement unit (PMU)communications
•IEEE 1547 - Physical and electrical interconnections between utility and distributed generation (DG)
•IEEE 1686-2007 - Security for intelligent electronic devices (IEDs)
•NERC CIP 002-009 - Cyber security standards for the bulk power system
• NIST Special Publication (SP) 800-53, NIST SP 800-82 - Cyber security standards and guidelines for federal in-
formation systems, including those for the bulk power system
• Open Automated Demand Response (Open ADR) - Price responsive and direct load control
• OpenHAN - Home Area Network device communication, measurement, and control
• ZigBee/HomePlug Smart Energy Profile - Home Area Network (HAN) Device Communications and Information
Model

This list will continue to grow as new standards are identified and as new standards are developed to meet the identi-
fied gaps in the existing standards.

Conclusion
The IntelliGrid Architecture provides a foundation for the operation of the Smart Grid and offers an optimized ap-
proach to build future visions. There are many long-term benefits to the energy industry that will be realized
through implementation of the IntelliGrid principles and recommended technologies. Clearly the IntelliGrid Archi-
tecture has profound ramifications for a broad range of advanced power systems applications. Careful planning of an
open and standards-based system designs will support integration of advanced systems thus realizing the IntelliGrid
vision for the Smart Grid of the future.

References
[1] Peter Sanza, Joe Hughes, et. al.; “IntelliGrid Architecture Volume I ‘User Guidelines and Recommendations’
Final Report”, 2004, Electricity Innovation Institute (E2I) Consortium for Electric Infrastructure to Support a
Digital Society (CEIDS); www.IntelliGrid.info.
[2] Peter Sanza, Joe Hughes, et. al.; “IntelliGrid Architecture Volume II Appendix F ‘Task 1 Enterprise Activities’
Final Report”, 2004, Electricity Innovation Institute (E2I) Consortium for Electric Infrastructure to Support a
Digital Society (CEIDS); www.IntelliGrid.info.
[3] Peter Sanza, Joe Hughes, et. al.; “IntelliGrid Architecture Volume IV Appendix E ‘Environments’ Final Re-
port”, 2004, Electricity Innovation Institute (E2I) Consortium for Electric Infrastructure to Support a Digital
Society (CEIDS); www.IntelliGrid.info.
[4] Peter Sanza, Joe Hughes, et. al.; “IntelliGrid Architecture Volume IV ‘Technical Analysis’ Final Report”, 2004,
Electricity Innovation Institute (E2I) Consortium for Electric Infrastructure to Support a Digital Society
(CEIDS); www.IntelliGrid.info.
[5] NIST Recognized Standards for inclusion in the Smart Grid Interoperability Standards Framework – Release
1.0; www.nist.gov/smartgrid/standards.html
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 133

Wide Area Monitoring with Phasor Measurement Units

M. WACHE 1, H.-J. HERRMANN 2,
Siemens AG 1(GER), Siemens AG 2(GER),

(PMU). The integration of PMU functionality in protection

Abstract-- The stability of electric transmission networks is and bay control devices is discussed.
getting more attention in the last years all over the world.
Following several wide area supply interruptions for example in II. SYNCHROPHASOR MEASUREMENTS
Europe and North America, it is widely accepted that the electric
transmission networks have to be improved in capacity and Network stability can be endangered by various impacts, as
stability. shown in Figure 1. Heavy loads in weak networks, combined
The building of new lines is mostly not possible in a fast manner, with switching or loss of generation, may lead to inter-area
so measures to enhance the stability of existing networks are of oscillations. Overloaded transmission corridors can lead to
great interest. One well accepted method is the use of phasor voltage collapse. Overloaded meshed networks can end in
measurement data from widely spread positions in the electric cascading outages. All these events should be avoided
networks. Synchrophasors are vector measurements sent by whenever possible. A valuable support is given by the use of
phasor measurement units (PMUs) with a standardized protocol.
synchrophasor measurements, which extend significantly the
With a suitable software solution, the information coming from
the Phasor Measurement Units (PMU) does help the control
situational awareness in the control center compared to the
center engineers in getting awareness of the stability situation in much slower RMS values. In the following, an example for a
the whole network. This helps them to make the right decisions phasor data processing software is shown which helps the
even in critical situations. SIGUARD Phasor Data Processing control center operator in understanding the information of the
System is such a system which is presented and discussed. synchrophasor data measurements.
With SIGUARD PDP-System, it is possible to visualize the power
swings in a clear way so that the damping can be evaluated easily.
The geographic screen shows at a glance where the problems have
occurred. The SIGUARD PDP System is both an online tool
which continuously evaluates the PMU measurements and an
offline tool (selectable mode). In offline mode, critical events can
be evaluated with archive data without using a separate tool.
The phasor data values are collected from the PMUs using the
standard protocol IEEE C37.118 which is available in several
standard products on the market. SIGUARD PDP can be
connected to single PMUs and to Phasor Data Concentrators as
well. The application of the new system is shown with an
application example in the network of a major german
transmission utility.
The paper closes with an outlook to the status of integration of
PMU functions into protection and bay control devices. In this
chapter, the possible new functions are discussed which result
from higher integration in the devices, combined with new Figure 1: Application of Phasor Measurements
powerful communication

KEYWORDS-- Phasor Measurement Unit (PMU), III. PHASOR MEASUREMENT UNIT

synchrophasors, wide area monitoring, stability analysis,
integration into bay level
I. INTRODUCTION
The use of synchrophasors to monitor and improve the
stability of electrical power systems is becoming more
important in the last years. The aim is to monitor the system
state, to sharpen awareness for the system stability and to use
the load limits of the lines correctly within the given system
limits, without extending the network. This paper presents the
application of a software for monitoring the synchrophasor
measurements coming from the Phasor Measurement Units
Phasor Measurements are well known in power systems; this
type of measurement is used for example in distance
protection devices since a long time for decision of trip signal.
The quite new idea is to collect Phasor Measurements from
substations widely spread in the transmission network and
monitor them on one single point to gain an overview about
overall network stability situation. The devices which compute
the Phasor measurements and then send them via a
standardized interface (IEEE C37.118) are called Phasor
Measurement Units (PMU). The PMU function can also be
included in already existing hardware, as for example in
protection devices.
The use of Phasor Measurement Units (PMU) for Wide Area
Monitoring Application is widely discussed in conferences, see
134
for example [1]. The following applications have been
discussed previously:
- Verification of dynamic model: Comparison of simulated
dynamic network data with measurements from PMUs (for
example described in [5])
- Detection of subsynchronous oscillations and analysis of
their damping
- Combining slow EMS data with synchro phasor
measurements on control center level to identify significant
dynamic performance issues
- Improvement of state estimation with PMU data
- Increasing of Overhead Line Capacity
- Calculation of Voltage Stability Index
Application of PMU is mainly discussed in transmission
systems up to know, but the interest of the operators of
distribution systems is rapidly growing.
PMUs have to be provided with a highly accurate time
synchronization on GPS basis. Only with a precise time base,
the synchrophasor data from a large area is comparable and
valid for information about the system state. The precision of a
PMU is expressed with the Total Vector Error TVE which is a
measurement for the combined error of measurement and time
precision of the device. To be compliant with the IEE C37.118
standard, a PMU may have a maximum TVE of 1%.
With the use of the highly precise PMU measurements which
are transmitted at a rate of 10 to 50 Hz (for a 50 Hz system),
not only slow load changes can be monitored but also fast and
transient power swings. The area supervised by PMU
measurements can be one single transmission line, but also a
large geographic area like the UCTE network.
IV. PHASOR DATA PROCESSING SYSTEM
This chapter is dealing with a synchrophasor based monitoring
software which gives support to the system operator for getting
an overview on network stability situation.
The phasor data processing software provides additional
awareness for the wide area dynamic conditions in the
network. The gap between fast but local acting protection
devices and slow acting control center applications as shown
in Fig. 2 is closed by the phasor data processing systems.
Figure 2 Wide Area Monitoring System in comparison to protection and
SCADA
Figure 3 shows the structure of a Phasor Data Processing
System. The Phasor Data Concentrator (PDC) is the central
part which collects the phasor data from the PMUs or from a
sub-PDC. The PMUs are widely spread in the power system to
get a complete picture of the system state. The PDC controls
an Archive for storage of data. It contains a system monitor for
control of the communication connections, internal data flow
and valid PMU data. The HMI is working with online data
directly from the PMUs or for analysis of disturbances or other
events from the past also with offline data from archive.
Figure 3: Structure of a Phasor Data Processing System
The Phasor Data Processing System may have interfaces to the
following components:
SCADA System. A SCADA System may improve its state
estimation with the additional use of phasor measurements.
State estimation may become faster and more precise and
convergence of calculations is improved.
Wide Area Control System. This summarizes different types of
systems with automated response to disturbances which are
detected by using wide area measurements, for example
synchrophasor measurements. For example capacitor banks are
controlled on base of results from PMU measurements.
Wide Area Protection System. Under this headline the
following applications are described:
Voltage Stability protection
Power swing detection and starting of countermeasures
Supervision of coupling between large area networks
These applications are in pilot phase and currently research
work is going on.
Additionally, there may be an exchange of PMU data with
other PDP systems to share synchrophasor data with
neighboured transmission system operators. This would be
done also via IEEE C37.118 protocol.
In the following, the HMI part of the Phasor Data Processing
System is described. The HMI has to support the control
center operator to gain awareness about critical situations and
to find the reason for problems. So it has to be intuitive and
simple. It has to fulfill the following requirements:
General indication about power system stability (ok / critical).
Online Configuration of the measurement view with respect to
selection of measurements, phasor or timebased view and
setting of the limits to be supervised
Easy change between online and offline mode to be able for
quick analysis of archive data
Geographical representation of the network for quick detection
of problem zones
Data Export for customer specific analysis and reports
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
The following screenshots show how these requirements have
been fulfilled by the Phasor Data Processing System
“SIGUARD” (all using simulated process data). Figure 4
shows the general layout of the Main Screen of the System. It
is divided into four parts:
Power System Status. Here a summary value for the status of
the supervised power system is shown. The nearer the curve is
coming to the limit, the more critical is the system stability
status.
Geographical View. This view contains the overview over the
whole monitored network with all substations and lines. The
operator can see at once in which region the critical incidents
occur since the colour of substations and lines is changing
depending on voltage (for substation colour) and current (for
line colour).
Data Area. This area in the center of the HMI is reserved for
Measurement charts (time based view or phasor view). The
measurements are selected via drag and drop from the
configuration area. So this area is not a static predefined
selection, it can be adapted at any time to the needs of the
situation.
Configuration Area. It contains the list of all measurements
sent by the PMUs to the system. There are two types of
measurements: Analogs (for example frequency, active and
reactive power) and Phasors (Voltage, Current). If necessary
also nose curves can be defined and offered for selection here.
Figure 4: General Layout of HMI of SIGUARD Phasor Data Processing
System
The complete measurement data of selected time slices can
also be exported (csv format) and so the user is able to
perform analysis in specific tools and to create reports in
customer specific templates.
V. APPLICATION EXAMPLE
The first application of the SIGUARD system is running since
three months at a german TSO with 7 PMUs on 400 kV level,
distributed in the whole geographic area of the TSO. The
system is configured for supervision of voltage angles between
north and south area and for frequency supervision. It can be
135
easily reconfigured online if other measurements shall be
supervised. The archive is configured to one week so that
critical occurences can be analyzed in offline mode. The
practical experiences will grow over the time, especially
waiting for ocurrances in the network to be analyzed. In ETG
conference “intelligent networks” in October 2009,
Düsseldorf, Germany, a first summary of experiences will be
presented.
VI. INTEGRATION TRENDS IN BAY LEVEL
Intelligent electronic devices (IED) for protection, control and
measurement are characterized by a modularity and flexibility
in hard- and firmware. The microprocessor allows multitasking
operation in real time for the different functions. The hardware
is freely scalable and can be adapted regarding the application
requirements. All this leads to a reduction of devices in the
field and reduces costs of the device hardware, in the cubicle
manufacturing and finally in the wiring. The functionality is
activated via parameterization. This means the configuration
of the device (masking of the hardware regarding the
application, selection of the needed functions) and the setting
of the different functions. All these tasks are supported by
powerful engineering tools.
Figure 5 shows an example for a feeder. The IED has multiple
current inputs for connecting to protection and measurement
current transformers. The voltage should be connected on a
measuring voltage transformer with the required accuracy. Via
the engineering tool the device is completely parameterized.
This means the routing of the analog inputs and the masking of
functions to these inputs. Figure 5 illustrates two containers,
which include the necessary functions. All protection
functions, which are required by the application, are in the
container “Protection”. The container “Measurements” covers
the measurement functions and additionally the function
“Phasor Measurement – PMU”. The IED is synchronized via
GPS or an equivalent synchronization method, e.g. according
IEEE 1588. The available serial interfaces allow a flexible
communication. A separate port can be used for
communication with a phasor data concentrator via IEEE
C37.118.
Figure 5: Integration of PMU functionality into a multifunctional IED
136

[3] A.G. Phadke, J.S. Thorp: Synchronized Phasor Measurements and their
Applications, Springer Verlag 2008
A further progress is the local operation of PMU data. In
[4] C. Rehtanz, K. v. Sengbusch, T. Sezi, R. Simon.: Schutz- und
modern schemes, the IEDs are linked together and can Überwachungskonzepte auf Basis zeitsynchroner Messungen,
exchange information in the substation or between substations. ETG/BDEW-Tutorial „Schutz- und Leittechnik“, Fulda, 11.-12. Nov.
The IEC61850 supports this via GOOSE mechanism. On the 2008 (in German)
other hand modern line differential relays have a powerful [5] CIGRE Session 2008 Paper C2-101: B. Ayuev, P. Erokhine, Y.
Kulikov: PMU Application for IPS/UPS Dynamic Performance
communication between the devices. The extended bandwidth Monitoring and Study
allows the transmission of additional information. In a cost [6] CIRED 2009 Prague 8-11 June 2009, Round Table 3b “PMU and Wide
effective way, phasors can be transmitted from the remote line Area Measurements in Distribution Systems”, Prof. Styczynski, Prof.
end via the protection data interface. Figure 6 illustrates the Sauvain, Dr. Buchholz, Dr. Wache
new possibilities.
As a decentralized solution (locally in a device), in a first step
the realization of monitoring functions can be implemented,
e.g. load angle supervision. An alarm can be given if the
stability limit will be reached. Another application is the
detection of a power swing and the realization of interrupting
the power system on predefined points or intelligent control
solutions. Especially the control solutions require interaction
between different parties. For example the access to the
control equipment of a power plant must be possible. Research
work is necessary particularly in the field of the intelligent
control with PMU data to find the right balance between what
is theoretically possible and practically feasible.

Figure 6: Local operation PMU data

VII. CONCLUSION
The phasor measurements are currently on their way from
theory to practical applications. Monitoring Software helps the
transmission system operators and the manufacturers to
understand dynamic behaviour of the transmission networks
and to create further applications with the goal to build wide
area control and wide area protection systems. The technical
evolution in device and communication technology gives a
good base for powerful support by the bay devices for the new
applications.

BIBLIOGRAPHY

[1] CIGRE Session 2008, Special Report for Group C2 “System Operation
and Control”, by Special Reporters Paulo Gomes, Gerhard Krost, Rui
Pestana
[2] CIGRE Session 2008 Paper C2-112: T. Sezi, J. Warichet, B. Genet, J.-
C. Maun: Bringing New Vizualization Tools for the Detection and
Mitigation of Dynamic Phenomena in the Transmission System
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Implementación de la Smartgrid en Redes de
Distribución Existentes
Aitor Arzuaga
ZIV R&D
Zamudio, Spain
a.arzuaga@ziv.es
La llegada de las Smartgrids a las redes de distribución de energía
eléctrica está dotando a las infraestructuras de un nivel de
tecnología y modernización desconocido hasta el momento. Todo
ello viene fruto de la necesidad de modernizar y dotar de
inteligencia a la infraestructura para lograr las mejoras de
eficiencia y reducción de emisiones de CO2, integración de
renovables, mejora de la calidad del suministro y gestión de
perfiles de carga.
Todas estas innovaciones se están poniendo en práctica en las
nuevas secciones de redes de distribución eléctrica en construcción
actualmente, donde toda la instalación es nueva. Sin embargo, la
mayoría de la red de distribución ya está ya desplegada, y si es
preciso lograr los objetivos mencionados anteriormente, va a ser
preciso modernizarla completamente. Esto plantea un reto
superlativo, debido a que estas infraestructuras son muy diversas y
de diferentes épocas, no fueron concebidas para ser modernizadas
en un futuro, han recibido históricamente poca inversión, en
muchos casos cuentan con un mantenimiento mínimo, y están en
operación.
Este trabajo resume los aspectos más importantes a tener en
cuenta al acometer un proyecto de modernización de red de
distribución existente con las funciones avanzadas más habituales
en un sistema de Smartgrid, como son telegestión de contadores,
supervisión de red, automatización de media tensión y
comunicaciones.
I. INTRODUCCIÓN
Estamos en el inicio de una nueva etapa en el sector de la
distribución de energía eléctrica. En paralelo se están
acometiendo despliegues de nuevas tecnologías y funciones en
lo que se ha venido a llamar la “Smartgrid”. Además el sector
se encuentra en un escenario convulso y de innovación
acelerada, fruto de los objetivos de reducción de emisiones de
CO2, la integración de las fuentes de generación distribuida,
los requisitos crecientes de calidad de suministro, los
despliegues de contadores de energía inteligentes que
permiten gestionar la demanda, la llegada del vehículo
eléctrico… todos estos factores no hacen sino acelerar la
adopción de la Smartgrid.
137
Rafael Quintanilla
ZIV Grid Automation
Zamudio, Spain
r.quintanilla@ziv.es
Este cambio de paradigma conlleva cambios que no son
sencillos, y se basa en la utilización masiva de las tecnologías
de información y comunicación ya en uso desde hace tiempo
en otros ámbitos. Y la parte de distribución de media y baja
tensión es la que más inversión precisa para cumplir los
objetivos.
La introducción de la Smartgrid en la red de distribución
de media y baja tensión y centros de transformación, con el
objetivo de automatizar y supervisar la red, requiere la adición
de funciones adicionales a la infraestructura existente. Tal es
el caso de los concentradores de datos de medida, funciones
de automatización, supervisión de MT/BT, comunicaciones,
sensores, sistemas de alimentación de respaldo… Todas estas
funciones son innovadoras, y están en muchos casos en su
etapa de introducción al mercado, o incluso en fase de
desarrollo.
Este artículo analiza el problema de la integración de las
funciones de Smartgrid en los centros de transformación de la
red de distribución de MT/BT existentes, que son la gran
mayoría de las redes y que están actualmente en explotación, y
que por tanto son los que presentan el principal problema de
integración por su casuística y diversidad.
Los problemas que aparecen son completamente distintos
a los que pueden ocurrir en un despliegue de un segmento de
red completamente nuevo, que serían básicamente
tecnológicos. En este caso, estamos combinando el problema
de la novedad tecnológica con el problema logístico de una
red que está en operación, es diversa, no está actualizada, no
cuenta con supervisión (más que por parte de los propios
clientes), es una especie de “museo viviente” de la historia de
la empresa eléctrica, y además tiene un nivel de
mantenimiento mínimo para seguir operando. Por todo ello, el
tratamiento que requiere el problema es específico.
II. OBJETIVOS DE LA SMARTGRID
El objetivo de la implantación de la Smartgrid en la red de
distribución de MT y BT consiste en proporcionar a la
empresa eléctrica nuevos servicios y funciones de valor
añadido que le permitan mejorar la operatividad de su red y
138
básicamente gestionar su negocio de forma más eficiente [1].
A continuación se destacan algunos de ellos:
• Implementar un sistema de telegestión de contadores
de clientes que permita una comunicación
bidireccional y en tiempo real con el cliente, y por
tanto establecer modelos de gestión de demanda.
• Localizar de forma precisa las faltas producidas en la
red de MT y analizar la respuesta y diagnóstico que se
le pueda dar.
• Reducir los tiempos de respuesta a incidencias y por
tanto mejorar los ratios de prestación de servicio a los
clientes.
• Mejorar la gestión de las instalaciones de distribución,
homogeneizando soluciones y actualizando
inventarios.
• Prolongar la vida de las instalaciones actuales en
servicio por medio de sistemas de supervisión y
monitorización avanzados.
Estos resultados por si solos justifican sin duda el reto
planteado. Sin embargo, la magnitud de la tarea que es
necesario acometer es titánica. Se trata de redes que pueden
consistir de millones de clientes, con decenas de miles de
centros de transformación en operación, y con un nivel de
diversidad muy elevado.
A continuación se detallan los aspectos funcionales
contemplados en el despliegue de Smartgrid, centrándolo en
las tres áreas de aplicación: baja tensión, media tensión y
comunicaciones.
III. FUNCIONALIDAD EN BAJA TENSIÓN
En baja tensión se concentran los puntos de suministro de
los clientes finales de la compañía eléctrica, y es este hecho el
que está sirviendo de detonante para lanzar los proyectos de
Smartgrid y las inversiones que llevan asociadas. En efecto, en
numerosos países se están poniendo en práctica medidas para
fomentar el despliegue de contadores inteligentes, que
permiten discriminar los patrones de consumo en distintos
horarios y tarifas, transmitiendo señales de precio a los
clientes, y por tanto dando un primer paso necesario para
implementar sistemas activos de gestión de la demanda.
Sin embargo, en una Smartgrid en una red de distribución
existente se pueden implementar múltiples funciones
adicionales a los contadores inteligentes, que van a ser tratadas
a continuación.
A. Smart Meters
Los contadores que permiten las funcionalidades
avanzadas de gestión de tarifas, cargas y clientes (también
conocidos como contadores inteligentes o smartmeters) son
actualmente el núcleo de todos los proyectos de Smartgrid. Es
a través del contador que el cliente adquiere consciencia del
precio de la energía, y es a través de la variación dinámica del
precio que el cliente será consciente de la variación temporal
de los costes de la energía, lo que permitirá establecer por
tanto mecanismos de reacción de los usuarios a los precios, y
por tanto una gestión real de la curva agregada de demanda,
aplanando los picos de consumo que tan costosos son para el
sistema, y por tanto aumentando la eficiencia del sistema y
reduciendo las emisiones de CO2 (por permitir un mejor
aprovechamiento de las energías renovables) [2].
La clave de los contadores inteligentes se basa en que
incorporan tecnologías de comunicaciones para un
intercambio de datos bidireccional con el sistema central de
medida, permitiendo por tanto recoger remotamente las
lecturas de consumo, pero también por ejemplo efectuar una
orden de desconexión remota del cliente. La tecnología de
comunicación empleada para conectar con los contadores
inteligentes puede ser diversa dependiendo de la topología de
la red de distribución y la densidad de clientes por
transformador. Las tecnologías utilizadas de forma más usual
son las siguientes:
• Tecnologías PLC. Se utilizan principalmente en
entornos urbanos, con contadores ubicados tanto en el
interior como en el exterior de las viviendas de los
clientes, y en los que la concentración de clientes por
centro de transformación es elevada. Esto es bastante
común en muchos países europeos. Se distinguen las
tecnologías de primera generación, de baja capacidad
(SFSK y similares), y las de segunda generación
(PRIME, G3, Meters&More), que permiten tasas de
transmisión mucho más elevadas al utilizar
modulaciones más complejas. Además, algunas como
PRIME [3] incorporan mecanismos de autodetección
de la mejor topología de la red para que la misma se
establezca de forma plug&play.
• Radio mesh. Se utiliza en entornos residenciales de
densidad media o baja, cuando los contadores se
encuentran generalmente fuera de las viviendas de los
clientes. Este es el caso en algunos países de América.
• Tecnología celular (GPRS), de aplicación en entornos
rurales de baja densidad de clientes, donde las otras
tecnologías no logran una cobertura suficiente.
Clientes – Smart meter
PLC – BT
Centro Transformación – Concentrador
GPRS
PLC - MT
Fibra óptica
Oficinas Centrales - Sistema
Figure 1. Comunicaciones como clave de los Smartmeters
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Actualmente, tal como se puede ver en la figura 1, el
consenso de la mayoría de los proyectos de despliegue es que
será necesaria una solución mixta (PLC, radio, fibra, GPRS)
para llegar a un 100% de cobertura, dada la diversidad
existente en las redes de distribución (y por tanto en las áreas
geográficas de prestación de servicio de la empresa eléctrica).
Por tanto, en función de la arquitectura de comunicaciones
utilizada, será necesario un equipo especial en el centro de
transformación, denominado concentrador de datos, que
recoge (vía PLC o radio) los datos de consumos y curvas de
carga de cada uno de los contadores, y por otro lado recibe las
órdenes desde el sistema central y las reenvía hacia cada
contador. Este elemento físico no se utiliza si los contadores
tienen acceso IP desde el sistema de medida (por ejemplo con
comunicaciones GPRS), aunque si puede existir un elemento
lógico que realice la misma función de agregación en el
sistema (concentrador virtual). La información entre el
concentrador y el sistema central se puede intercambiar en
múltiples formatos, siendo uno comúnmente utilizado el
formato XML, utilizando protocolos basados en IP para la
transmisión (FTP,. HTTP, web services).
B. Otras funcionalidades avanzadas en baja tensión
Además de los contadores inteligentes, existen múltiples
funcionalidades adicionales que se pueden añadir en baja
tensión, incluso en redes existentes con las dificultades que
eso conlleva. A continuación se detallan las más
representativas:
• Monitorización de la red de BT. Al disponer de
elementos de monitorización en la salida de baja
tensión del transformador, se puede monitorizar la
tensión y corriente en la línea de baja, y por tanto
vigilar la calidad de onda, sobretensiones y huecos,
armónicos y todos los parámetros significativos de la
onda. Además esta monitorización abre la puerta a la
integración de generación distribuida (renovable), ya
que se establece el punto de control necesario en la
salida del transformador (y por tanto entrada en la red
de distribución).
• Detección de desequilibrios entre fases. Al
monitorizar de manera independiente las tres fases del
suministro, resulta muy sencillo detectar el equilibrio
o desequilibrio de la carga de las fases y así poder
tomar medidas correctoras que lleven a la
compensación y equilibrado del transformador, para
reducir su stress y pérdidas.
• Detección de averías. El elemento gestor de baja
tensión (concentrador de medidas), ubicado en el
centro de transformación, posee un enlace de datos
activo con cada uno de los contadores de cliente que
cuelgan de dicho CT. Por tanto, un suceso como una
pérdida de conexión con un contador o grupo de
contadores permitirá lanzar una alarma de posible
avería, antes incluso de que los clientes afectados lo
notifiquen.
• Detección de pérdidas y fraude por balance de
energías. El equipo concentrador residente en el
139
centro de transformación posee la información de la
energía suministrada por ese centro en un intervalo de
tiempo determinado. Como a su vez va a recoger la
información de consumo de cada uno de los
contadores de cliente que se conectan a dicho centro,
esto permite detectar las pérdidas o posible fraude de
forma muy sencilla. En efecto, realizando el
sumatorio de las energías consumidas por todos los
clientes, esto debe ser igual (con una mínima
incertidumbre dada por la precisión de los equipos de
medida y las pérdidas de la red de distribución) a la
energía suministrada por el centro. En el caso de que
la diferencia supere un valor umbral, se pueden
establecer alarmas.
• Identificación de línea y fase a la que pertenece cada
contador. Uno de los problemas más comunes que se
encuentran las compañías eléctricas en las redes de
distribución es que no tienen ningún registro de en
qué línea de salida o incluso fase está conectado cada
uno de los clientes, e incluso qué edificios se
alimentan a través de un centro de transformación
determinado. Una forma de solventarlo es provocar un
corte de tensión a cada una de las líneas y fases, de
forma que se sepa qué abonados quedan sin
suministro. Sin embargo, en los nuevos sistemas de
smartmetering, al disponer de equipamiento
inteligente en ambos extremos del enlace (centro y
abonado) se pueden establecer mecanismos que
permiten identificar la fase, e incluso la línea, a la que
pertenece cada abonado. Esto es debido a que las
señales de comunicación por Powerline
Communications (PLC) que intercambian pueden
referenciarse temporalmente a los pasos por cero de
cada una de las fases, o de la fase a la que está
conectado el contador mediante un mecanismo
denominado “zero crossing detection” [4]
IV. FUNCIONALIDAD EN MEDIA TENSIÓN
Una vez de lanzado el despliegue, fruto de la necesidad de
instalar contadores inteligentes, y dado que será preciso
instalar equipamiento en el centro de transformación, con las
dificultades que ello conlleva, la pregunta natural que surge
es: ¿qué más funcionalidades se pueden aportar para la
empresa eléctrica, una vez de instalar equipamiento en el
centro? La respuesta evidente consiste en dotar al centro de
equipamiento que permita supervisar las magnitudes de media
tensión, y ejecutar acciones, integrado como un equipo de
remota o telecontrol más del SCADA de la compañía.
Para poder supervisar las magnitudes de media tensión es
preciso poder captar las tensiones y corrientes presentes en las
líneas de MT, respetando los aislamientos de seguridad. Por
ello se precisan sensores específicos, que pueden ser distintos
para cada tipo de celda, en función del tamaño y modo de
conexión que permita [5]. Este es uno de los mayores retos a
la hora de implementar esta función en centros existentes. Las
tensiones se pueden recoger con divisores, bien resistivos o
capacitivos, que adapten la señal de MT. Las corrientes se
capturan con transformadores toroidales o bobinas de
Rogowski, en función de la precisión requerida o el nivel de
140
dificultad que sea asumible en el montaje. Con divisores y
toroides es posible llegar a niveles de precisión del 1% ó
mejor en tensión, y similar en corriente.
Las remotas o equipos que supervisan y actúan sobre los
elementos de media tensión pueden ser capaces de comandar
los interruptores que conectan el centro, y por tanto de quitar y
reponer el servicio. Además, ante el evento de una falta, tienen
que ser capaces de funcionar con normalidad, por lo que en
estos casos la alimentación tiene que estar respaldada por un
sistema de batería que garantice un periodo de funcionamiento
durante un tiempo suficiente (varias horas) en ausencia de
alimentación primaria.
Las funcionalidades que se pueden implementar en media
tensión para una red de distribución en funcionamiento son las
siguientes:
• Supervisión de corrientes y tensiones de MT.
Evidentemente se pueden comprobar los niveles de
distorsión, armónicos, huecos, sobretensiones, etc. que
se producen en la línea de media tensión. La
aplicación más inmediata es el control del flujo de
potencia en la línea, de cara a controlar su capacidad,
la del transformador del centro, y a optimizar el
funcionamiento del sistema.
• Detección de paso de falta. Al controlar en tiempo real
el estado de las tensiones y corrientes, es posible
realizar una detección de un paso de falta de forma
direccional (si ha pasado una falta por la línea, y con
qué sentido). Así se puede saber en qué lugar está la
falta relativo a la posición del centro de
transformación en la línea de MT. Si en el SCADA
central se recoge la información agregada para todos
los centros, se puede determinar de forma muy
sencilla la ubicación física de la falta, entre dos
centros.
Figure 2. Mecanismo de detección paso de falta (DPF)
• Automatización de la red. Un paso que se puede dar,
una vez que se tiene la información de corrientes y
tensiones en media tensión, pero a la vez se puede
disponer de salidas controladas con las que ejecutar
mandos, es el de automatizar completamente el centro
de transformación y los elementos de conexión
(interruptores). Un ejemplo de mecanismo sencillo de
automatización consiste en un aislamiento automático
de una falta detectada, una vez detectado el paso de la
falta y su direccionalidad. Otro mecanismo de
automatización más complejo puede ser una
reposición automática del servicio tras el aislamiento
de la falta.
• Funciones de protección. Teniendo la información en
tiempo real de la forma de onda de tensiones en las
tres fases, y corrientes en las fases y neutro, se pueden
aplicar los algoritmos y funciones de protección que
son de aplicación en las líneas de salida de la
subestación, por parte del relé que controla la
cabecera.
V. COMUNICACIONES
A. Comunicaciones desde el CT hacia el despacho
Uno de los elementos catalizadores de la revolución en las
redes de distribución de energía eléctrica son las
comunicaciones. Para llevar la Smartgrid a todos los rincones
de la red de distribución, es preciso ser capaz de dotar de
comunicaciones a los centros de transformación existentes. En
su gran mayoría, dichos centros no están comunicados
actualmente, ya que las empresas eléctricas han restringido el
acceso de comunicaciones a ciertos centros de reparto más
importantes en la red de media tensión. Por el contrario, todas
las subestaciones de media y alta tensión están comunicadas y
telecontroladas.
Figure 3. Arquitectura de red de comunicaciones
La red utilizada se basa en tecnología IP, de forma que en
todos los casos se va a llegar a los centros de transformación
(y a los equipos allí instalados) desde cualquier punto de la red
de la empresa eléctrica, por sus direcciones IP. Esto permite
tener gran flexibilidad a la hora de utilizar equipamientos y
soluciones estándar, y a la vez garantizar estabilidad,
escalabilidad y seguridad en el sistema.
El aspecto más importante en el caso de la red de
distribución es que se debe llegar a un gran número de
instalaciones, y esto hace prácticamente imposible llegar con
medios propios, tanto por dimensión (en muchos casos se trata
de decenas de miles de centros), como por cuestiones
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 141

económicas. Por tanto esto supone una ruptura con la forma presta un operador de telecomunicaciones). En el caso
tradicional de comunicar con los nodos clave de la red y de las redes de distribución ya existentes, puede
subestaciones, que en muchos casos se realiza con medios ocurrir que los centros de transformación sean
propios para asegurar la disponibilidad. En el caso de la subterráneos. En ese caso la cobertura celular puede
Smartgrid, es necesario recurrir a medios externos para ser limitada, aun en entornos urbanos, y complicada
proporcionar el enlace de red de área extensa (WAN). Esto de mejorar.
amplía las posibilidades de utilización disponibles.
• ADSL. Sobre todo en los entornos urbanos, debido a
La siguiente tabla resume los requisitos de comunicación la penetración masiva de la banda ancha xDSL en los
requeridos para aplicaciones de Smartgrid, para diversas hogares, es posible que sea viable la instalación de
aplicaciones: puntos de conexión ADSL en los centros de
Capacidad Latencia Disponibilidad
transformación, contratados a una compañía de
Tipo telecomunicaciones. La disponibilidad y capacidad es
Requerida Requerida Requerida
óptima (velocidad de varios Mbps y latencias de
Vertical
Telemedida de (uplink más Medio
pocos ms), a costa de unos costes de operación
<5s Media-alta elevados. Además la disponibilidad de equipamiento
contadores ancho (100Kbps)
banda) xDSL adecuado a las condiciones de utilización en
Bajo centros de transformación es limitada. Esto hace que
Supervisión y Vertical, su utilización sea limitada, salvo que la criticidad de
(<50Kbps <1 s Alta
control de MT simétrico
) algún servicio lo justifique.
Automatización
Horizontal y
Alto • Powerline Communications (PLC) sobre líneas de
vertical, <50ms Muy alta
de Distribución
simétrica.
(>1Mbps) media tensión. Esta es una posibilidad muy interesante
para las compañías eléctricas ya que posibilita la
Vertical comunicación a través de medios propios (las líneas
Medio
Gestión de (uplink más
infraestructura ancho
(>200Kbp <1s Media de MT), de forma que los costes de operación son
s) mínimos. Esta tecnología se basa en la utilización de
banda)
modulaciones apropiadas a las características del
Tabla 1. Requisitos técnicos de comunicaciones para diversos medio, como por ejemplo OFDM o Spread Spectrum
servicios en Smartgrid [6] [8], obteniendo velocidades dependientes del medio,
Como se puede observar en la tabla 1, a partir de los pero que pueden superar 1Mbps, con latencias en el
servicios que se implementen en la Smartgrid, las necesidades orden de 50ms. Los centros de transformación que se
de comunicaciones serán distintas, y por tanto se requerirán agrupan a lo largo de la misma línea de media tensión
arquitecturas de comunicaciones de distinto tipo, y con unos se convierten en celdas de una misma red de
costes de inversión y mantenimiento muy diferentes. Por otro comunicaciones, que comparte un medio común. En
lado, como en todo proyecto real, un requisito básico es que un punto de la celda se añade una salida WAN hacia
los costes de inversión de despliegue y mantenimiento sean lo el exterior (GPRS, fibra, ADSL…) para completar la
más reducidos posible. Ello hace que se tengan que conectividad. Las señales de comunicación se acoplan
seleccionar entre las diversas tecnologías disponibles, y que no a las líneas de MT a través de unos dispositivos de
haya una única válida para todas las ocasiones, sino que se acoplamiento que permiten salvaguardar el
emplean distintas soluciones tecnológicas para cada tipo de aislamiento galvánico. La utilización de PLC sobre
centro de subestación, en función de su relevancia en la red, MT en centros de transformación existentes es
ubicación, antigüedad y servicios soportados en el mismo. compleja, ya que requiere de soluciones de
acoplamiento específicas en función de que las celdas
Las tecnologías que se emplean más usualmente son las del centro sean de mampostería, de aire o de gas SF6
siguientes: [9].
• Tecnología celular (GPRS). La penetración masiva de • Satélite. Cuando no existe otro modo de
los teléfonos móviles en la sociedad, con los servicios comunicación, siempre se puede recurrir a servicios
avanzados de datos que llevan aparejados, ha de comunicación de datos por satélite (como VSAT),
significado que los operadores han desplegado una red que permiten establecer enlaces con cobertura global
de comunicaciones prácticamente ubicua en todas las y una tasa de transmisión dependiente del contrato
zonas habitadas [7]. Al querer aumentar sus líneas de solicitado, pero que puede alcanzar los centenares de
negocio, proporcionan una oportunidad para Kbps Sus desventajas son que requieren la instalación
aprovechar dicha capacidad para otras aplicaciones a de una infraestructura considerable (antenas), y los
las empresas, que puede ser aprovechada por la costes de operación son elevados. Como las redes de
eléctrica. En concreto, la presencia casi universal de la distribución eléctrica van, usualmente, ligadas a
tecnología GPRS, que permite velocidades de hasta centros de población humana, no suele ser preciso
80Kbps en el enlace descendente y 20Kbps en el recurrir a comunicaciones por satélite para
ascendente, proporciona un canal de comunicaciones comunicaciones en Smartgrid.
adecuado (tal vez algo escaso para algunas
aplicaciones) y razonablemente barato (el servicio lo
142
• Radio. Algunas empresas eléctricas poseen redes de
radio en bandas de frecuencia asignadas (canales VHF
o similar), para aplicaciones de telecontrol y
monitorización. Dichas bandas se pueden utilizar
también para las comunicaciones derivadas de la
Smartgrid, porque además las condiciones de
propagación y cobertura en esas partes del espectro es
muy buena. Sin embargo, requieren la construcción y
mantenimiento de la infraestructura por la propia
empresa eléctrica, y la capacidad obtenida es baja
(pocos Kbps).
B. Comunicaciones dentro del CT
Otra parte clave del sistema de comunicaciones consiste en
la interconexión de todos los elementos inteligentes que son
instalados en el centro de transformación. Los distintos
servicios requeridos (telegestión de contadores, supervisión y
automatización de MT, servicios auxiliares…) deben ser
concentrados en un equipo de comunicaciones que permita
enrutar el tráfico hacia el despacho de control.
En la práctica, todos los elementos presentes en el centro
se conectan utilizando una red Ethernet en el propio centro.
Las razones para hacerlo son múltiples:
• Es una tecnología madura, probada y barata, pero aun
con una larga vida por delante.
• Sus prestaciones (100mbps) cubren de sobra todas las
aplicaciones que puedan venir en un futuro a un
centro de transformación.
• Se comporta perfectamente en entornos adversos
como puede ser un centro de transformación.
• Existe una multitud de equipamiento disponible de
muchos proveedores distintos.
• Permite segmentar los distintos servicios de forma
muy sencilla utilizando VLANs.
Para salir del centro hacia el despacho, se puede construir
una red a nivel 3 (con lo que haría falta un router, por ejemplo
con tecnología GPRS), o se puede construir una red a nivel 2
(con lo que haría falta un bridge, por ejemplo con tecnología
PLC de MT). Esto depende de la arquitectura elegida a nivel
de celdas de CT, ya que, en último caso, todas las
comunicaciones hacia el despacho se realizarán a nivel IP y
por tanto habrá un router de comunicación, bien sea a nivel de
centro o de cabecera.
VI. INSTALACIÓN E INTERACCIÓN CON EL CENTRO DE
TRANSFORMACIÓN
A la hora de acometer el despliegue de un sistema de
Smartgrid en una red de distribución existente y en operación,
en muchas ocasiones se pone todo el foco del proyecto en los
aspectos tecnológicos del mismo, y sin embargo toda la parte
de despliegue es un problema logístico que no hay que
olvidar, y bastante complejo como se va a analizar a
continuación.
La red de distribución se puede considerar como la
“hermana pobre” del sistema eléctrico, ya que durante muchos
años no ha recibido el mismo nivel de inversiones que las
redes de transporte y distribución en alta tensión, al ser su
criticidad para el sistema menor. Adicionalmente, las
instalaciones no están telegestionadas ni supervisadas, ni se
visitan cada poco tiempo como es el caso de las subestaciones.
Por otro lado, en casi todos los países las compañías
eléctricas han ido creciendo por fusiones o adquisiciones, de
forma que la red de distribución resultante es bastante
heterogénea en cuanto a topologías y equipamientos. No solo
por provenir de organizaciones distintas, sino por provenir de
la misma organización pero en distinta época. Ello hace que la
variabilidad que se encuentra y para la que hay que estar
preparado sea enorme. Un primer paso inicial del proyecto
consiste en la selección previa de qué instalaciones o centros
son susceptibles de ser actualizados con la nueva
funcionalidad, y cuales son simplemente demasiado antiguos y
precisan ser sustituidos completamente o dejados a un lado.
La conclusión a todo esto es que en muchos casos, cuando
se plantea el despliegue de la Smartgrid, la información que se
verifica al realizar un replanteo en cada uno de los centros
difiere de la que figuraba en los archivos de la compañía
eléctrica. Por ello aparecen errores, desajustes,
modificaciones, sorpresas… que pueden complicar
sobremanera el despliegue, pero con las que hay que convivir.
Es muy común que celdas de gas se conviertan en aire y
viceversa, lo que puede requerir el empleo de sensores y
acoplamientos distintos a los inicialmente acopiados. Siempre
merece la pena un replanteo previo, su coste queda
amortizado.
Otro punto importante es el descargo en el que se realiza la
actualización del centro de transformación con los nuevos
equipamientos. Habitualmente se pone todo el foco en el coste
del equipamiento, pero el coste del descargo supone una parte
importante del total, por lo que es preciso tener en cuenta su
duración. Por tanto la solución de Smartgrid a utilizar
(equipamientos, armarios) tiene que ser lo más integrada
posible [10], de forma que se facilite el tiempo y la mecánica
de la instalación. Asimismo, todo en el despliegue tiene que
estar organizado de forma que se produzca una instalación y
validación rápida del centro.
Adicionalmente, las dificultades que se encuentran en la
instalación son notables, ya que las infraestructuras de los
centros no se diseñaron para que nuevos equipamientos
pudieran ser añadidos a posteriori. Por tanto los espacios
disponibles son en muchas ocasiones mínimos, y es preciso
desarrollar equipos y sensores que se adapten específicamente
a la diversidad de celdas y centros existentes en la red de
distribución, que pueden ser bastantes.
Por último, no debe olvidarse que la red de distribución es
muy distinta de una subestación, en la que todos los elementos
están probados y en perfecto orden de revista. En un centro de
transformación, es probable que por ejemplo un interruptor de
un transformador no haya sido maniobrado en más de diez
años, y no existe ningún mecanismo que permita verificar su
estado. Por tanto, pueden ocurrir situaciones como que el
interruptor falle a la hora de reponer el servicio tras el
descargo. Este tipo de situaciones deben estar previstas. No
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
hay que perder de vista que las instalaciones pueden tener
bastantes años a sus espaldas.
VII. CONCLUSIONES
A modo de resumen, para afrontar con garantías la
implementación de funcionalidades avanzadas de Smartgrid
en una red de distribución es útil tener en cuenta las
siguientes conclusiones:
• La solución de arquitectura de la Smartgrid debe ser
sencilla, al menos al comienzo.
• Hay que poner unos requisitos comunes al proyecto
que permitan actualizar de una misma forma el 90 ó
95% de las instalaciones, para evitar caer en
personalizaciones excesivas para cada tipo de
infraestructura de centro de transformación.
• Los requisitos funcionales y operacionales deben
identificarse claramente antes del comienzo de los
trabajos. Es imprescindible trazar una raya de
separación entre los requisitos necesarios y los
deseables.
• El foco del coste debe contemplar el sistema
completo, teniendo en cuenta los costes de
instalación, validación y puesta en servicio,
adicionalmente a los costes de los equipamientos, ya
que pueden ser una parte importante del total.
• Los servicios que se quieren implementar son los que
determinan de forma última los requisitos de
capacidades de comunicaciones requeridos, y por
tanto la arquitectura que se necesita.
• Existe un compromiso necesario entre prestaciones
del sistema y coste de la solución, que debe de
afrontarse al inicio del sistema de forma que se
establezcan objetivos realistas (de presupuestos y
plazos) al proyecto.
143
• Es necesario planificar visitas a las instalaciones,
revisiones de los planes, actualizaciones, replanteos,
etc… ya que a largo plazo son rentables.
• Los gestores del proyecto y la dirección de la
empresa debe prepararse para la variabilidad e
incidencias que pueden ocurrir en una red de
distribución con decenas de años de servicio.
• Diferentes escenarios (por ejemplo zonas urbanas,
rurales, etc) requerirán de diferentes soluciones. En
las redes existentes no hay una solución homogénea
que sirva para todo un despliegue.
REFERENCIAS
[1] R. Quintanilla, T. Yarza, “Making the Smartgrid Real: a case study”,
CIGRE SEAPAC Conference, 2011.
[2] ESMIG, “A guide to Smart Metering”, Brussels, 2009.
[3] A. Arzuaga, I. Berganza, A. Sendin, M. Sharma, V. Varadarajan,
“PRIME interoperability tests and results from field”, IEEE Smartgrid
Conference 2010.
[4] http://www.prime-alliance.org/
[5] J.A. Moreno, A. Arzuaga, C. Coca, “Advanced Sensors for the
Smartgrid”, CIRED Conference 2011, unpublished.
[6] A. Arzuaga, T. Arzuaga, J. Salat, “Enabling Smartgrid
Communications over MV Lines”, CIGRE SEAPAC Conference, 2011
[7] A. Arzuaga, T. Arzuaga, M. Zamalloa, “Implementation of Telecontrol
Applications over GPRS Networks”, PAC World Magazine, winter
2010
[8] A. Arzuaga, T. Arzuaga, J. Salat, “Enabling Smartgrid
Communications over MV Lines”, CIGRE SEAPAC Conference, 2011
[9] A. Arzuaga, T. Arzuaga, J. Salat, “Enabling Smartgrid
Communications over MV Lines”, CIGRE SEAPAC Conference, 2011
[10] A. Arzuaga, J. Arriola, Z. Ojinaga, T. Arzuaga, M. Zamalloa,
“Integrated solution for the Smartgrid”, PAC World Conference 2011,
unpublished..
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 145

ESQUEMA DE PROTECCION Y CONTROL PARA LA INTERCONEXION en

13.8 KV, DEL PERMISIONARIO MEXICO-CARBON

Marco Antonio Ortiz Hernández Luis Del Ángel Diego

Pedro Martínez López

Comisión Federal de Electricidad

División Golfo Centro

Resumen: •evita contaminación por fuga de gas

En la actualidad se ha presentado un
incremento en la interconexión de generación
de baja escala con diferentes esquemas, o
también conocida como generación distribuida.
SF6, aceite, etc.)
Todo lo anterior se refleja en la misión y en la
competitividad de la empresa.

Los generadores pueden establecerse con Se conecta al sistema de Distribución

diferentes esquemas de acuerdo al permiso mediante el circuito de media tensión PUO-
que le solicitan a la Secretaria de Energía: 4115, de la SE Puerto.

• Cogenerador La S.E. Puerto pertenece ala

• Auto abastecimiento
• Auto abastecimiento con venta
de excedentes.
La instalación de generación en redes que han
sido diseñadas solo para abastecer clientes,
provocando dificultades en la correcta
operación de las protecciones instaladas. Esto
conduce a la adecuación de los esquemas de
protección en los puntos de interconexión y en
las redes donde se conectan los nuevos
generadores.
Zona Tampico y se encuentra ubicada en el
corredor industrial de la Cd. de Altamira, con
tres líneas de transmisión con Voltaje Nominal
de 115 kv, un transformador de Potencia con
voltaje primario 115 kv voltaje secundario de
13.8 kv. con capacidad de 20 MVA y 4
alimentadores de media tensión. La figura 1
Muestra el diagrama simplificado de la S.E.
Puerto.
PUO-73420 PUO-73140 PUO-73150

PUO-73428 PUO-73148 PUO-73158 PUO-77010

I. INTRODUCCION
PUO-72010 PUO-72018
DIAGRAMA UNIFILAR SIMPLIFICADO
Dentro de la División Golfo Centro se tienen 30 MVA
115/13.8 KV
S.E. PUERTO
interconectados 23 permisionarios, con las
diferentes modalidades. PUO-42015 PUO-42018

Uno de los permisionarios es MEXICO

PUO-4118 PUO-4128 PUO-4138 PUO-4148

CARBON una empresa de la industria PUO-4115 PUO-4125 PUO-4135 PUO-4145

petroquímica, que como consecuencia de sus

procesos producen vapor, el cual utilizan para
Figura 1
generar energía para su auto abastecimiento y
venta a sus socios a través de la red de CFE,
lo que se conoce como porteo de energía, por
lo que requiere sincronizar sus generadores
con el sistema y permanecer conectado.
II. DESARROLLO
146
Para la atención de la Solicitud de conexión e
interconexión del servicio, se revisan las
características de los equipos, se determinan
los requerimientos para la interconexión de
acuerdo a la Ley del Servicio Público de
Energía Eléctrica y su Reglamento.
Requerimientos principales:
• Agregar a los Sistemas SCADA
los generadores y su
trayectoria hasta el punto de
interconexión.
• Implementación de los
esquemas de Protección en
para la correcta operación del
sistema eléctrico.
• Adecuación a los equipos de
medición.
Agregar a los Sistemas SCADA los
generadores y su trayectoria hasta el
punto de interconexión.
La instalación de equipo de control que
cumpla con las características del equipo
instalado en el área de control (CENACE) o en
el centro de control mas cercano, (Centro de
Control de Distribución).
Figura 2
EL sistema de control remoto, debe
monitorear los siguientes parámetros:
Por cada interruptor de maquina
Señales digitales
• Posición de interruptor de unidad (
abierto/cerrado )
Señales analógicas
• Potencia activa instantánea trifásica de
salida (MW)
• Potencia reactiva instantánea trifásica
de salida (MVAR)
• Energía activa de salida (MWH)
• Energía reactiva de salida (MVARH)
• Corriente por fase
• Voltaje trifásico (promedio)
• Frecuencia
En el punto de interconexión (para cada
interruptor línea de llegada)
Salidas de control
• Mandos de interruptor (
apertura / cierre )
Señales digitales
• Posición de interruptor de línea
( abierto/cerrado )
• Posición de cuchillas de línea (
abierta/cerrada )
Señales analógicas
• Potencia activa instantánea
trifásica de salida ( MW )
• Potencia reactiva instantánea
trifásica de salida ( MVARr )
• Corriente de línea
• Voltaje trifásico ( promedio )
Señales digitales (alarmas de apoyo para
supervisión)
• Falla mecanismo
• Baja presión SF6
• Mecanismo descargado
• Mecanismo bloqueado
• Opero protección primaria
• Opero protección de respaldo
• Falta vcd circuito cierre
• Falta vcd circuido disparo 1y 2
Implementación de los esquemas de
Protección en para la correcta operación
del sistema eléctrico.
Implementación de los esquemas de
Protección en para la correcta operación del
sistema eléctrico, tanto en el punto de
interconexión como en el punto con esquema
de protección mas cercano sobre la red de
conexión.
En este caso el punto mas cercano al punto de
interconexión es el interruptor PUO-4115. El
esquema asociado a este interruptor esta
preparado con protecciones de sobrecorriente

las cuales son adecuadas solo para líneas con
operación radial, es decir donde el flujo de
corriente en condiciones de carga normal y en
condiciones de falla siempre es en el mismo
sentido. Figura 3
SE Puerto
PUO-4115
50/51 79
Flujo de corriente
La configuración inicial del
circuito esta prevista para una
operación radial.
Figura 3
El equipo instalado en las instalaciones del
permisionario esta enfocado en controlar y
proteger sus equipos y sus generadores,
además no se tienen equipos que no están
aprobados por el LAPEM en sus funciones de
protección.
Los equipos no están enfocados a la
protección del sistema al que se
interconectarán (SEN). Figura 4
SE Puerto
PUO-4115
50/51 79
G Permisionario México Carbón
Figura 4
La operación del circuito cambia por lo que
será necesaria la adecuación de los esquemas
de protección del punto de interconexión, para
salvaguardar la integridad del personal que
labore en la operación y mantenimiento de la
línea de distribución, la integridad del sistema
eléctrico y de los demás usuarios conectados
X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 147
en la línea, ya que al estar sincronizados los
generadores del permisionario con el Sistema
se debe asegurar el disparo del punto de
interconexión, con tiempos de operación
adecuados.
La primera adecuación es en el Interruptor
PUO-4115 en S.E. Puerto. Con un esquema
de Protección de alimentador (PP), con
funciones de Protección de Sobrecorriente con
unidad direccional de secuencia positiva,
negativa y cero, (67F/67N), Equipado con
funciones de bajo voltaje, alto voltaje,
Funciones de sincronismo , Alta Y Baja
Frecuencia, recierre automático, localización
de fallas, Oscilografía, Registro secuencial.
Figura 5.
SE Puerto
PUO-4115
67/67N
59/27/25
T1 6 MVA
13.8/ 4.16kv
D-Y
G
Permisionario México Carbón
Figura 5
La segunda adecuación es en el Punto de
Punto de Interconexión
El permisionario no cuenta con
interconexión POSCO MEXICO, con la
equipo de protección para el instalación de un Interruptor de Potencia con
sistema donde se interconecta
Línea Subterránea
Corriente Interruptiva 31 KA y Esquema de
Protección equipado con las siguientes
T1 6 MVA funciones de protección hacia la carga:
13.8/ 4.16kv
D-Y
Protección de Sobrecorriente con unidad
direccional de secuencia positiva, negativa y
cero, (67F/67N), Equipado con funciones de
bajo voltaje, alto voltaje, Potencia inversa,
Funciones de sincronismo Alta Y Baja
Frecuencia, Oscilación de Potencia, recierre
automático, localización de fallas, Oscilografía,
Registro secuencial. Esquema de Protección
Equipado con las siguientes funciones hacia la
línea: Protección de Sobrecorriente con unidad
direccional de secuencia positiva, negativa y
cero, (67F/67N), Equipado con funciones de
bajo voltaje, alto voltaje, Potencia inversa,
localización de fallas, Oscilografía, Registro
secuencial. Figura 6.
148

interconexión, se firma el contrato de

interconexión, y se le autoriza la entrada en
operación comercial, el permisionario comenzó
PUO-4115
su operación comercial en diciembre del
SE Puerto
67/67N 2008.
59/27/2 67/67N
5 59/27/25
67/67N
59/27/25 Ya estando en operación, en S.E. México-
Carbón, el esquema de protección del lado
T1 6 MVA
primario del transformador, presentó una falla
13.8/ 4.16kv
D-Y
y el permisionario no dio la importancia
requerida para restablecer las funciones de
G
protección, y el día 26 de febrero del 2009 se
Permisionario México Carbón
presentó una falla en una terminal del cable
subterráneo, esta evento nos dio la evidencia
del desplazamiento vectorial que se tiene en
una falla de fase a tierra en un sistema no
En tren el punto de interconexión del y el aterrizado. La conexión del transformador en
interruptor de lado primario existe una línea su lado primario es en DELTA y este se
subterránea de aproximadamente 350 metros comporta como un filtro de las corrientes de
por lo tanto adicionalmente se instaló en el secuencia cero. Se provocaron sobre voltajes
lado primario del transformador un esquema en las fases no falladas, provocando daños en
de Sobrecorriente direccional de secuencia la instalación del permisionario y la operación
positiva, negativa y cero, y funciones de alto de las protecciones del lado secundario del
voltaje. En la figura 7 se muestran los transformador una vez que la falla evolucionó
esquemas de protección. y se convirtió en falla entre fases. A
continuación se describe la secuencia del
evento y los efectos en el voltaje. Figura 8.

PUO-4115
SE Puerto

67/67N
59/27/25 67/67N PUO-4115
67/67N 59/27/25
59/27/25 SE Puerto 2.- Opera el punto de interconexión

67/67N 1.- Falla de fase a tierra en línea

59/27/25
Oscilo grafía del relevador 3.- No dispara el interruptor del lado
T1 6 MVA del punto de interconexión primario y permanece alimentando
13.8/ 4.16kv la falla
D-Y
T1 6 MVA
13.8/ 4.16kv
D-Y

Permisionario México Carbón G 4.- No operan los controles con protecci

Incluida en el Generador

Figura 7 G Permisionario México Carbón

Figura 8
Las figuras 9 y 10, muestran los valores de
Adecuación a los equipos de medición. voltaje alcanzados mientras se mantuvo la
aportación a la falla de los generadores del
Las adecuaciones a los equipos de medición permisionario.
consisten en la instalación de un medidor
principal y un medidor de respaldo en el punto
de interconexión, el medidor debe tener la
capacidad de medir energía activa y reactiva
en los cuatro cuadrantes, para determinar el
consumo y la aportación del permisionario a la
red de CFE.

Después de que el permisionario realiza las

adecuaciones y cumple con los requisitos de
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 149

PUO-73420 PUO-73140 PUO-73150

PUO-73428 PUO-73148 PUO-73158 PUO-77010

Oscilografía del relevador del punto de interconexión,
no existe corriente al estar abierto el equipo solo se
tienen las señales de voltaje
PUO-72010 PUO-72018

Concentrar estados de equipos involucrados

en la lógica de monitoreo para determinar la desconexión al
sistema de CFE.

PUO-42015 PUO-42018

PUO-4118 PUO-4128 PUO-4138 PUO-4148

PUO-4115 PUO-4125 PUO-4135 PUO-4145

Figura 11
Figura 9

PUO-72018
Fibra Óptica
PUO-73148 (Canal principal)
Lógica
PUO-73158
de
PUO-73428 Procesador Procesador relé de
Lógico Radio Espectro Disperso Lógico
S.E. PUO S.E. MXC protección
PUO- 42018 (Canal respaldo) Cogenerador
PUO- 4118
S.E. MXC
PUO- 42015 Cogenerador

PUO- 4115

Monitores de estado de
interruptores a través de
protocolo.
PUO-73140
PUO-73150 Enlace de comunicación Disparo de interruptor de llegada a S.E. MXC
PUO-73420 por puertos seriales
PUO-72010
Sobre voltajes detectados por los relevadores del punto de interconexión, PUO-4115
el voltaje nominal de fase a tierra es de 7.9 KV, los voltajes detectados
de fase a tierra hasta 14.9 KV.

Como consecuencia daño de apartarrayos, cargadores de baterías y un

transformador de servicios propios en la instalación del permisionario.

Figura 12

Figura 10
PUO-73420 PUO-73140 PUO-73150

Posterior al evento de disparo, el permisionario PUO-73428 PUO-73148 PUO-73158 PUO-77010

solicita apoyo y asesoría a CFE para habilitar

funciones adicionales sobre los equipos e PUO-72010 PUO-72018 Procesador
Lógico
PUO

implementar un esquema que los conmute a

modo isla, al perderse la interconexión con el PUO-42015
PUO-42018
Procesador
Lógico

sistema de CFE.
MXC

PUO-4115 PUO-4118

67F, 67N, 59N, 25,

27, 32

Se evalúan todos los posibles escenarios bajo MXC-4115

los cuales debe de enviarse la cogeneración a Punto Interconexión CFE - MXC

modo isla y se propone concentrar los estados

de los interruptores y cuchillas involucrados en
un procesador lógico para transmitirlo al punto
de interconexión y posteriormente a la planta Figura 13
de cogeneración para, mediante lógicas de
control, abrir el interruptor de llegada a la S.E.
MXC al momento de perderse el enlace CFE -
MXC. El arreglo se muestra en las figuras 11,
12 y 13. En la figura 14 se ilustra la operación de la
transferencia de la señal al momento que se
150

pierde la interconexión del permisionario a V. BIBLIOGRAFIAS

través de la apertura del interruptor PUO-
72010, que corresponde al lado primario del ING. MARCO A. ORTIZ
Transformador de la S.E. Puerto. HERNANDEZ

PUO-73420 PUO-73140 PUO-73150

PUO-73428 PUO-73148 PUO-73158 PUO-77010

Egresado del Instituto Tecnológico de San Luis
Potosí de la carrera de Ingeniería Industrial en
PUO-72010 PUO-72018 Procesador
Lógico
Eléctrica.
PUO
Ingreso a C.F.E. en 1996 al Departamento
PUO-42015
PUO-42018
Procesador
Divisional de Proyectos y Construcción de la
Lógico
MXC División Golfo Centro, con el puesto de
PUO-4115 PUO-4118

67F, 67N, 59N, 25,

Supervisor de Construcción “C”.
27, 32

MXC-4115
En 1997 se incorpora al Departamento
Punto Interconexión CFE - MXC
G
Divisional de Subestaciones y Líneas, como
profesionista en Entrenamiento de la Oficina
Figura 14 de Protecciones.
Con este arreglo complementario el En 1997 se le asigna la Jefatura de la Oficina
permisionario controla su generación al de Protecciones de la Zona Victoria.
momento de perderse la interconexión, A partir de Septiembre del año 2005, se
evitando problemáticas en sus generadores desempeña como Jefe de Oficina de
como rechazos de carga y fluctuaciones en el Protecciones, en el Departamento Divisional
voltaje y la frecuencia. de Subesbestaciones Y Líneas de la División
Golfo Centro.
III. CONCLUSIONES
Teléfono: (01-833) 226-2932 E-Mail
Es muy importante establecer y mantener los : marco .ortiz01@cfe.gob.mx
esquemas de protección y control de manera
conjunta con los permisionarios, para
salvaguardar la integridad del personal, las
instalaciones y la estabilidad del sistema ING. PEDRO MARTINEZ
eléctrico. LOPEZ
A partir del 14 de Julio 2010 se cuenta con el
Procedimiento de Atención a las Solicitudes de
Interconexión de Permisionarios a la Red del
Servicio Público de Energía Eléctrica en el Egresado del Instituto Tecnológico de Cd.
cual se establecen los requerimientos mínimos Madero en la carrera de Ingeniería Eléctrica en
para la correcta operación de los Potencia.
permisionarios y generación distribuida. Ingreso a C.F.E. en 1992 al Depto. Divisional
de Proyectos y Construcción con el puesto de
IV. REFERENCIAS Supervisor de Construcción “C”.
En 1993 se incorpora al Depto. Divisional de
[1] “Protective Realying. Principles and Subestaciones y Líneas, específicamente a la
Aplications. Third Editiion. J. Lewis Ofna. de Protecciones, fungiendo como
Blackburn. Ingeniero Sustituto.
En 1997 se le asigna la Jefatura de la Ofna. de
[2] ”Moderm Solutions for Protección. Protecciones de la Zona Tampico, puesto que
Control, and Monitoring of Electric Power desempeña actualmente.
Systems.Hector J. Altuve Ferrer, Edmund
O. Schweitzer, III. Teléfono: (01-833) 226-09-24
E-Mail : pedro.martinez02@cfe.gob.mx
[3] “PAC Word Magazine, Summer 2008.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 151

ING. LUIS DEL ANGEL DIEGO

Egresado del Instituto Tecnológico de Cd.

Madero en la carrera de Ingeniería Electrónica
Ingreso a C.F.E. en 1998 al Depto. Divisional
de Comunicaciones y Control puesto de
Supervisor de Construcción “C”.
En 2001 se le asigna la Jefatura de la Ofna. de
Control al Depto. Divisional de
Comunicaciones y Control puesto que
desempeña actualmente.

Teléfono: (01-833) 241-33-26

E-Mail : luis.delangel@cfe.gob.mx
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
EL MANTENIMIENTO CENTRADO EN CONFIABILIDAD CON EL APOYO DE
LOS SISTEMAS DE PROTECCIONES DE LAS SUBESTACIONES DE
DISTRIBUCIÓN.
Jorge Betanzos Manuel
Comisión Federal de Electricidad
División Sureste- Zona Huajuapan.
Resumen:
A nivel mundial, la industria está
focalizada en aumentar la productividad sin
incrementar sus costos, lo que obliga a
muchas compañías a luchar por optimizar
sus niveles de producción para permanecer
competitivas. En este sentido, se requiere
un sistema de monitoreo de tiempo real
para realizar los estudios que nos permitan
una mejor planeación de los trabajos de
mantenimiento y mejoras a las
instalaciones, de tal forma que se
minimicen las interrupciones y las pérdidas
de energía, para mantener el suministro
dentro de los parámetros determinados en
los contratos de suministro establecidos
con los clientes y en los compromisos de
servicio.
Palabras Claves: MBC, Tiempo real.
I. INTRODUCCION
El Mantenimiento Basado en la confiabilidad
(MBC) es el mantenimiento que se debe hacer
para que las instalaciones hagan lo que la
empresa desea que hagan, para lo cual se
implementó un sistema de información que
permite definir los indicadores del desempeño
de las subestaciones, considerando la filosofía
del Mantenimiento Centrado en Confiabilidad,
a través de los datos proporcionados por los
Dispositivos de Protección, Control y Medición
instalados en las subestaciones de
Distribución.
Algunos de los datos que nos proporcionan
son los siguientes:
• Monitoreo de estados y alarmas de
interruptores, Tableros PCM
• Monitoreo en tiempo real de
mediciones por fase de voltajes,
corrientes, potencias y frecuencia
153
José Carlos López Valencia
• Monitoreo de Calidad de energía:
Sag/Swell, Armónicas
• Monitoreo de alarmas y temperaturas
de los transformadores de potencia
tanto en tiempo real como históricos
• Monitoreo de Tensión de servicios
propios, Tensión del banco de baterías,
temperatura de casetas y tableros PCM
• Reportes de arranques, disparos,
apertura/cierre de interruptores y
corriente de fallas.
Con esta información determinar cuáles son
las tareas de mantenimiento adecuadas para
cualquier activo físico, con el objetivo principal
de reducir el costo de mantenimiento, para
enfocarse en las funciones más importantes
de los sistemas, y evitando o quitando
acciones de mantenimiento que no son
estrictamente necesarias, algunos de los
beneficios obtenidos son:
• Reducción del tiempo del
restablecimiento del suministro de
energía eléctrica
• Seguridad del Personal (al evitar
traslados a las subestaciones.)
• Reducción de la energía dejada de
vender.
• Mayor productividad de los
colaboradores de la empresa.
• Incremento del conocimiento de la
operación del sistema eléctrico.
• Impacto ambiental. Reducción de la
contaminación (al evitar el traslado, se
reduce el consumo de combustible. Se
evita contaminación por fuga de gas
SF6, aceite, etc.)
Todo lo anterior se refleja en la misión y en la
competitividad de la empresa.
154

II. DESARROLLO Figura 1.0 Diagrama general del monitoreo en

El sistema de monitoreo en Tiempo real consta
de cuatro módulos los cuales tienen las
siguientes funciones:
Fuente de la información: la cual se obtiene de
los relevadores de protección, Medidores de
calidad de energía y cuadros de alarmas:
Enlace de I nformación: Para poder tener las
alarmas y mediciones se cuenta con la
aplicación “Syscom” la cual interroga a los
DEI’s por medio del protocolo DNP el cual
coloca un servidor OPC.
Procesamiento de la información: Una vez que
se tiene la información ya sea en el servidor
OPC o integrada al ION ENTERPRISE es
guardada en una base de datos de MySQL.
Despliegue de la información: El despliegue de
la información se realiza a través de una
página WEB la cual puede ser consultada en
la intranet de CFE.
tiempo real
Como en cada subestación se tiene instalado
un concentrador se crea un canal a nivel
superior el cual envía las mediciones de
bancos y líneas a un concentrador zonal en el
cual podemos ver el comportamiento de las
cargas de todos los transformadores y el flujo
de energías de las líneas de transmisión.
A. Monitoreo de la Subestación Eléctrica
Los datos monitoreados en cada una de las
subestaciones son estados, alarmas,
mediciones de líneas, bancos, circuitos,
bancos de capacitores, bancos de baterías,
temperaturas de los transformadores,
corrientes y distancias de falla; estas variables
son desplegadas en un unifilar (Figura 1.1) de
la subestación el cual puede ser visto a través
de una página web.

En la figura 1.0 se muestra la implementación

general del monitoreo en tiempo real en la cual
la fuente de información son todos los DEI’s ya
sean Relevadores, cuadros de alarmas,
medidores, sensores de temperatura, UTR’s
etc. Los cuales cuenten con un protocolo de
comunicación estándar en este caso DNP,
Modbus. En la siguiente fase del concentrador
es donde se instalan las aplicaciones que
Figura 1.1 Diagrama Unifilar de la Subestación
realizan la función de maestra para interrogar
NCH
a los DEI’s, esta aplicación (SYSCOM) realiza
la función de maestra, coloca los datos en un
servidor OPC, y guarda todos los registros en
una base de datos MySQL, es en esta
aplicación donde se realiza también la
conversión hacia un canal con protocolo ION
para ser integrados al ION ENTERPRICE y de
esta aplicación es publicada a una pagina
WEB por medio de la herramienta VISTA
también de ION.

Figura 1.2 Diagrama de monitoreo de alarmas

Los reportes generados de las alarmas son

de gran utilidad para identificar la secuencia de
eventos, estos reportes incluyen el nombre del
dispositivo, fecha, hora del evento, nombre de
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 155

la alarma generada y valor o cambio de

estado. Figura 1.3 En la gráfica 1.0 se observa el comportamiento
de arranques de las protecciones de un
esquema de línea, se muestra el numero de
disparos y algunas de las alarmas como son
alto voltaje, bajo voltaje, baja presión de gas
SF6, con estas alarmas podemos a parte de
programar el patrullaje de la línea verificar la
periodicidad de alarmas de interruptores como
de los niveles de tensión con lo cual identificar
fallas frecuentes en interruptores.

Figura 1.3 Histórico de alarmas

Con toda esta información se puede realizar

un filtrado y conteo de cuantos arranques y
disparos se generan por cada circuito, que tan
repetitivas son algunas alarmas para identificar
ya sea fugas de gas, incrementos constantes
de temperatura etc. Teniendo mejor medidas Grafica 2.0 Comportamiento de protecciones
estos parámetros podemos diseñar un de líneas
programa de Mantenimiento de forma mas
eficiente, es importante tener claro que cada En la gráfica 3.0 se observa el
cliente y cada subestación tiene una comportamiento de arranques y disparos del
necesidad particular,. Por esto, es importante esquema de la protección del banco, alarmas
definir la criticidad de los equipos que forman configurables para identificar la carga en OA y
parte de la subestación, de los procesos y de en FA, para dar un seguimiento puntual a la
los modos de falla; esto último de acuerdo a sobrecarga de transformadores.
ciertas metodologías de confiabilidad.

De los reportes generados de las bases de

datos con el historial de fallas (Tablas 1.1-1.4)
se pueden realizar graficas para ver de una
forma más visual el comportamiento de
arranques, disparos y alarmas.

En la gráfica 1.0 se observa el

comportamiento de arranques de fase a tierra,
arranques de fallas entres fases, el número de
disparos y las veces en operó la función del re- Grafica 3.0 Comportamiento de protecciones
cierre. Con esta información se puede realizar de Bancos
un programa de inspección a los circuitos el
cual sea más puntual a los circuitos críticos. B. Monitoreo del transformador de potencia

El transformador es considerado el equipo

más importante de una subestación, lo que
hace suponer que es en él donde se debe
invertir la mayor cantidad del presupuesto de
mantenimiento [2]. Durante su operación, los
transformadores de potencia están sujetos a
esfuerzos que degradan su sistema de
aislamiento. Las causas principales de
degradación son: temperatura excesiva,
Grafica 1.0 Comportamiento de protecciones presencia de oxígeno y humedad que
en circuitos combinadas con los esfuerzos eléctricos
156

aceleran el proceso. Es por esta causa que se

requiere de monitorear constantemente los
parámetros de temperatura de aceite,
temperatura de devanados, Nivel de líquidos,
tensión en el lado de alta, tensión de salida y
carga en demanda máxima; teniendo estos
datos en tiempo real e históricos nos permite
operarlos con un mayor grado de seguridad,
permitiendo minimizar o postergar los costos
asociados a la reparación o substitución de los
mismos.

En la figura 1.4 se muestra la instalación de

los sensores de temperatura de líquidos y de Figura 1.5 Monitoreo del Transformador de
aceite Potencia

C. Sincronización en tiempo Real

Para realizar un análisis correcto de de los

eventos ocurridos en una subestación y
compararlos con los de otra es necesario que
todos los relevadores estén sincronizados en
tiempo, es por esto que se empleó el protocolo
NTP para realizar la sincronía de la maquina
instalada en la subestación y esta a su vez
sincronice a los demás equipos de la
subestación.

• Los servidores NTP distribuyen el Tiempo

Universal Coordinado (UTC).

Figura 1.4 Instrumentos del Transformador de

Potencia

Los beneficios de los sistemas de monitoreo

en línea son los siguientes:

• Proporcionan información sobre la

condición operativa de los
transformadores.
Figura 1.6 Sistema de sincronización de
• Crean un historial de datos.
tiempo real
• Permiten en algunos casos, sobrecargar
los transformadores sin reducir su vida útil.
• Cambian de mantenimientos periódicos a
mantenimientos basados en la condición
real del equipo.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
III. CONCLUSIONES
Podemos resumir que la confiabilidad es una
herramienta poderosa para suministrar
ventajas competitivas, que pueden
incrementar la rentabilidad, la seguridad y
eficacia de los equipos en una subestación
eléctrica al mejorar sus programas de
mantenimiento y por ende tener la satisfacción
de los clientes y usuarios.
Es necesario establecer sistemas de
monitoreo que permitan adquirir y administrar
los datos para realizar el mantenimiento
basada en la confiabilidad de los equipos.
Para esto se puede apoyar de los sistemas de
Protección, Control y Medición
Se requiere la participación de diferentes
disciplinas o áreas del conocimiento para una
mejor implementación de un sistema de
mantenimiento basado en la confiabilidad
operativa de los equipos
IV. REFERENCIAS
[1] “Analyze relay fault data to improve service
realiability”, Roy Moxley, Schweitzer Engineer
Laboratories, Inc. Pullman, WA USA
[2] ”Using information from relays to improve
the power system”, David J. Dolezilek and
Dean A. Klas, Schweitzer Engineer
Laboratories, Inc. Pullman, WA USA
[3] “Implementation of a New Subestation
Restoration System Using Protective Relays
and a Logic Processor”, Bernard S. Tatera, Jr.
Pacific Gas & Electric Company San
Francisco, California Spokane, Washington.
2001
157
V. BIBLIOGRAFIAS
Jorge Betanzos Manuel
Se graduó en 1983 de Ingeniero Industrial Eléctrico
en el Instituto Tecnológico de Oaxaca (ITO). En
1984 inició sus actividades en la CFE-División
Sureste como supervisor de construcción. De 1986
a 1993 trabajó en la CFE-Región de Transmisión
Sureste como Ingeniero de Protección, Control y
Medición de la zona Itsmo y zona Malpaso. De
1993 a 1997 se desempeñó como Jefe de la
Oficina de Protecciones de la Zona de Distribución
Oaxaca. De 1997 a 2000 se desempeñó como Jefe
de Estudios de Redes de Alta Tensión del
Departamento de Planeación en la CFE-División
Sureste. Del 2000 a la fecha se ha desempeñado
como Jefe del Departamento de protecciones de la
División Sureste. Está a cargo del proyecto de
monitoreo de calidad de energía de la División
Sureste.
José Carlos López Valencia Obtuvo el Título de
Ingeniero en Electrónica en la Universidad
Tecnológica de la Mixteca, en Huajuapan de León,
Oaxaca. México (2004). En el 2005 ingresó a CFE-
Distribución, se ha desempeñado como Jefe de
Oficina de Control Tuxtla (2006), a partir del 2007
se desempeña como Jefe de la Oficina de
protecciones de la Zona Huajuapan En el 2008
realizó un Diplomado en sistemas eléctricos de
Potencia. Su trabajo incluye, diseño de sistemas de
protección, puestas en servicio.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Using Fault Tree Analysis to Evaluate
Protection Scheme Redundancy
Ramón Sandoval and César A. Ventura Santana, Comisión Federal de Electricidad, Mexico
Héctor J. Altuve Ferrer, Ronald A. Schwartz, David A. Costello, Demetrios A. Tziouvaras,
and David Sánchez Escobedo, Schweitzer Engineering Laboratories, Inc.
Abstract—In this paper, we apply fault tree analysis to
compare the dependability and security of transformer and
generator protection schemes with different degrees of
redundancy. We also compare the scheme costs. For each
scheme, we use a single protection scheme as the reference. We
then evaluate schemes with dual redundancy and two-out-of-
three voting schemes. We also evaluate the effect of
comprehensive commissioning testing, hidden failures, and
common-mode failures, as well as using relays from the same or
different manufacturers in redundant schemes.
I. INTRODUCTION
In a redundant protection scheme, all of the components
except the breaker are redundant. Breaker failure protection
provides a functional duplication of the breaker. Redundant
scheme design must consider both aspects of reliability:
dependability and security. Dual-redundant schemes enhance
dependability because two independent schemes operating in
parallel are less likely to fail to clear a fault. However, dual-
redundant schemes can also reduce security because there are
two schemes that could operate for an out-of-zone fault. A
fully redundant two-out-of-three voting scheme enhances
security without impairing dependability. However, such a
scheme would be very expensive, considering the cost of dc
power systems, channels, and instrument transformers.
Sharing these scheme components affects the voting scheme
dependability and security. In addition, voting schemes are
more complex than dual-redundant schemes and require
redundant relays to have similar sensitivities to ensure the
operation of at least two of the three relays.
In the past, protection schemes were composed of several
discrete relays, with each relay performing a single function.
These traditional multirelay schemes provided no redundancy.
Today, microprocessor-based relays provide many protection
functions. One relay can replace a whole scheme of discrete
relays at a much lower cost. Adding a multifunction relay
provides redundancy, without significantly increasing cost.
Fault tree analysis is a practical tool for system reliability
evaluation. Engineers can use fault tree analysis to compare
the relative reliability of proposed protection schemes.
Analyzing protection scheme dependability and security
requires different fault trees. When constructing each tree, the
protection engineer identifies which component failure causes
a failure to trip or an undesired trip. This analysis leads to
different tree topologies and different unavailabilities or
failure rates.
159
In this paper, we compare the dependability and security of
transformer and generator protection schemes with different
degrees of redundancy. We also compare the scheme costs.
We start from a single protection scheme and add equipment
to create dual- and triple-redundant (with two-out-of-three
voting logic) schemes. We also evaluate how comprehensive
commissioning testing, hidden failures, common-mode
failures, and the use of relays from the same or different
manufacturers influence scheme reliability.
II. RELIABILITY CONCEPTS
Reliability is the ability of an item to perform a required
function under stated conditions for a stated period of time.
Reliability and related variables are time-dependent
probability quantities. In many applications, reliability
analysis using time-independent quantities provides results
that are approximate but still of practical value. Table I
defines the measures often used to describe product reliability
performance, assuming constant failure and repair rates [1].
TABLE I
COMMONLY USED RELIABILITY MEASURES
Measure Definition
Termination of the ability of an item to perform
Failure
its required or specified function.
Total number of failures divided by total unit
Failure rate (λ) operating time or uptime. Data are collected
from field observations or tests.
Total number of repairs divided by total unit
Repair rate (μ)
operating time or uptime.
Average time between start of operation or
Mean time to failure
(MTTF)
return after repair and failure. For a constant
failure rate, MTTF = λ–1.
Average time to correct a failure and restore a
Mean time to repair unit to operating condition. Includes
(MTTR) preparation, active maintenance, and logistics
time. For a constant repair rate, MTTR = μ–1.
Mean time between Average time between failures for units
failures (MTBF) repaired and returned to use.
MTBF is the sum of MTTF and MTTR. Because MTTR is
usually small compared to MTTF, we assume that MTBF is
approximately equal to MTTF and that MTBF = λ–1.
Protective relays and protection systems are designed to be
repairable. Therefore, measures of reliability should include
the possibility of failure and repair. Availability is a measure
that considers repeated cycles of failure and repair.
160
Availability is the probability or fraction of time that a
device or system is able to operate. Equation (1) defines
availability A for constant failure and repair rates.
μ MTTF MTBF
A= = ≈ (1)
λ + μ MTTF + MTTR MTBF + MTTR
Relay users are often concerned with the amount of annual
downtime that may occur in a protection system.
Unavailability is the probability or fraction of time a device or
system is unable to perform its intended function. Equation (2)
defines unavailability U for constant failure and repair rates.
λ MTTR
U = 1− A = = ≈ λ MTTR (2)
λ + μ MTBF
From (2), observe that we can lower unavailability by
decreasing the MTTR (monitor the self-testing of
microprocessor-based relays, and keep spares in stock). We
can also lower unavailability by increasing the MTBF (use
equipment with low failure rates and robust designs).
As probabilities, availability and unavailability are
dimensionless numbers from 0 to 1. However, we can convert
them to minutes or seconds per year by multiplying by the
appropriate factors.
III. PROTECTION SCHEME REDUNDANCY
Protection systems consist of devices that detect faults on
the power system (protective relays) and apparatus that
interrupt fault current (circuit breakers) [1] [2]. In some cases
(fuses and automatic circuit reclosers), both functions are
combined.
The protection system design philosophy for responding to
the failure to detect faults and the failure to interrupt faults is
generally in one of two categories [1]:
• Redundant systems.
• Overlapping relays tripping different interruption
devices (remote backup protection).
Redundant protection systems use redundant components
to eliminate single points of failure for detecting faults.
Redundant systems are typical in transmission lines of
networked systems, because relays are less able to detect
faults in adjacent zones and because the result of delayed
tripping is more severe Redundant protection system are
increasingly used in large transformers and generators.
A redundant protection system may include the following:
• Redundant primary relays.
• Redundant communications channels.
• Redundant instrument transformers or separate voltage
secondary circuits for each set of primary relays.
• Redundant dc control power systems.
• Breakers with redundant trip coils.
A breaker failure protection scheme covers failure of the
breaker to interrupt the fault.
The application of redundant systems is now economical
at all voltage levels because of the low cost of modern
multifunction relays and the elimination of most other
equipment by using the ancillary features of these relays.
Redundant system architecture actually reduces the
complexity of many tasks, such as coordination and designing
to eliminate single points of failure. It also enables the design
of continuous self-testing features that reduce the chance of
hidden failures and eliminate most periodic maintenance and
inspection [2].
Redundant protection systems may include dual or triple
sets of relays. Dual-redundant schemes typically use OR
tripping logic to ensure fault clearing. This bias toward
dependability comes from the assumption that delayed fault
clearing may be more dangerous to the power system than
tripping healthy power system elements. However, modern
power systems operate close to their security limits. For
example, an undesired trip of a heavily loaded transmission
line, or a large generating unit or transformer, may cause
transient stability problems or trigger a cascading breaker-
tripping event. Several large power system blackouts have
been triggered or compounded by undesired line and/or
generator tripping. In protection systems with three sets of
relays, two-out-of-three voting logic is an alternative to
improve security. Some wide-area protection schemes, where
an undesired trip may have devastating consequences for the
power system, use two-out-of-three voting schemes.
Redundant protection systems may use identical or
different relays. Some engineers consider that using relays
with different operation principles and hardware platforms
reduces the risk of incorrect operations caused by common-
mode and hidden failures [3]. As a consequence, some of them
recommend using relays from different manufacturers.
However, modern multifunction relays allow the application
of different protection principles even with identical relays. In
addition, the same manufacturer usually provides similar
protection functions in different relays (different hardware
platforms). Some utilities use two different relays from the
same manufacturer in redundant systems. Furthermore, relay
manufacturers use many common types of components from
the same suppliers. Finally, the probability for the same
component to fail at the same time in two identical (or
different) relays is very low.
Many industries requiring high reliability use dual-
redundant systems with identical components. The aviation
industry is one example [4].
Increasingly, utilities and other electric power users are
adopting the redundant system approach of the aviation
industry. According to an independent survey, 55 percent of
utilities in the United States and Canada use the same
manufacturer in dual primary systems for high-reliability
protection designs [5].
Historically, utilities provided dual-redundant primary
protection by applying two electromechanical relays with
different operating principles for protection either by zone or
by phase. Today, each primary system can include different
principles of operation: line differential and directional
comparison primary protection, for example, complemented
by breaker failure protection and distance and/or directional
overcurrent backup protection.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Using identical relays in a dual primary protection system
provides the following advantages [1]:
• Two identical systems allow engineers to design one
system and use it twice—lower settings labor, higher
settings reliability, and lower incidence of human
error.
• Common designs, algorithms, and settings ensure
optimum protection coordination.
• Common automation and integration simplify
architecture and reduce cost.
• A common operator interface makes system operators
more comfortable.
• Personnel can analyze data with the same skills and
tools.
• Personnel can train in depth on one relay instead of
having to learn how to use two relays for the same
purpose.
• Troubleshooting is simpler because it is easier for
users to compare the reports of two identical relays for
the same fault.
IV. FAULT TREE ANALYSIS
Fault tree analysis is a practical tool for evaluating how a
component failure contributes to a specific failure event [1]
[6]. Fault tree analysis is useful for comparing the relative
reliability of proposed protection schemes. It is appropriate for
considering the top-down reliability performance of a system
for specific failure events.
The failure event of interest is called the top event. A
system may have more than one top event that merits
investigation. The failure rate for the top event is a
combination of the failure rates of the basic events (the tree
roots) that contribute to the top event. Basic events are
individual component failures with identified failure rates. We
use AND, OR, and other gates to represent combinations of
failure rates. OR gates express the idea that any of several
failures can cause the protection system to fail. The OR gate
output is the sum of the failure rates of the input events. AND
gates express the idea that failures must occur simultaneously
to cause the protection system to fail. The AND gate output is
the product of the failure rates of the input events. We can also
use availability, unavailability, or MTBF figures instead of
failure rates in fault tree analysis.
The power system performance requirements (preserving
transient stability, for example) determine the top event of the
fault tree. If, for example, the power system requires high-
speed fault clearing to preserve transient stability, the top
event should only consider high-speed protection. However, if
the power system remains stable after a breaker failure
protection operation, the top event should also consider
breaker failure protection.
Fault tree analysis helps in analyzing the security or
dependability of a protection system. Security is the ability of
a system to never trip for an out-of-zone fault or when no fault
is present. Dependability is the ability of a system to never fail
to clear an in-zone fault.
161
Analyzing the dependability and the security of a
protection system requires different fault trees. For
constructing each tree, we should identify which component
failures may cause a failure to trip (a dependability problem)
or an undesired trip (a security problem). This analysis leads
to different tree topologies and different failure rate (or
unavailability) values. For example, any relay failure could
cause a failure to trip if a fault occurs during the relay
downtime. However, not all relay failures cause an undesired
trip. Hence the relay failure rate or unavailability value to use
for dependability analysis is higher than the value to use for
security analysis.
In this paper, we use unavailability for dependability fault
trees because failures to clear faults depend on component
downtime per failure. We use failure rate for security fault
trees because undesired trips typically occur at the instant a
component fails [7].
Fault trees allow comparing the relative unavailability of
various protection schemes. By keeping the fault trees simple
and making simplifying assumptions, engineers can analyze
the fault trees easily with hand calculations. The advantages of
fault tree analysis include the following [1]:
• While the failure rate or unavailability data of
individual components are approximate, some are
substantiated by field measurements, so fault trees
give useful order-of-magnitude results.
• With different top events and fault trees, engineers can
easily evaluate dependability-related failures versus
security-related failures.
• Fault tree analysis is a critical step in ensuring the best
application of limited engineering resources.
V. TRANSFORMER PROTECTION EVALUATION
A. Transformer Protection Schemes
In this paper, we compare the reliability of three schemes
for protecting a delta-wye transformer with single breakers at
both sides. All schemes use multifunction transformer relays
with only current inputs.
Fig. 1a shows the single protection scheme, which includes
one relay, one set of current transformers (CTs) on the
transformer high-voltage (HV) and low-voltage (LV) sides,
one dc power system, and breakers with single trip coils.
Fig. 1b shows the dual-redundant protection scheme, which
includes two relays, two sets of CTs on each side of the
transformer, two dc power systems, and breakers with
redundant trip coils. The scheme has only one CT on the
transformer neutral grounding conductor, but we also evaluate
the effect of adding another CT to the neutral circuit. To create
a two-out-of-three voting scheme, we add a third relay to the
Fig. 1b scheme, connected to the same CTs and the same dc
power system as one of the other two relays. In redundant
schemes, we assume all the redundant components are of
similar quality. We assume relays have the same reliability
indices, sensitivities, and speeds of operation.
162

HV Bus LV Bus
52-HV 52-LV
HV Bus LV Bus
52-HV 52-LV

125 Vdc-1
Main 1 Transformer
Protection Relay

Transformer Protection
Relay

125 Vdc-2
Main 2 Transformer
Protection Relay
125 Vdc

(a) (b)

Fig. 1. Single and dual-redundant transformer protection schemes using multifunction relays.

B. Fault Tree Analysis relay application or settings errors, breaker failures, dc power
We created 24 fault trees (12 for dependability analysis and system failures, CT failures, dc system and CT wiring errors,
12 for security analysis) for the following combination of and hidden failures. We assign to these events the
transformer protection schemes and other factors: unavailability values shown in Table VI in the appendix. In
• Single scheme. this fault tree, the OR gate reflects the fact that, in a single
• Dual-redundant scheme with two neutral CTs. protection scheme, the failure of any component causes a
• Dual-redundant scheme with one neutral CT. scheme failure to clear a fault. We can modify the fault tree as
required to consider other scheme configurations, to include
• Dual-redundant scheme with relays from different
other events of interest, or to use other unavailability values.
manufacturers.
• Dual-redundant scheme that experiences common-
mode failures.
• Redundant two-out-of-three voting scheme.
For each of these schemes, we created a fault tree that
considers the normal process of commissioning testing of the
protection scheme and another fault tree that reflects the effect
of comprehensive commissioning testing. Reference [8]
describes a process with a checklist for consistent and
thorough commissioning tests. Reference [9] reviews best
practices and provides a list of lessons learned from
commissioning protective relay systems.
Table VI in the appendix shows the reliability indices that
we used in the fault trees and includes an explanation of the
method we followed to determine each value.
We describe several fault trees in this section and then
summarize the results obtained from all the fault trees in Fig. 2. Dependability fault tree for the single transformer protection scheme.
Table II in the next section. According to [9], we include in a separate fault tree the
1) Single Schemes effect of comprehensive commissioning testing by modifying,
Fig. 2 shows the dependability fault tree for the single as explained in the appendix, the unavailabilities
scheme (see Fig. 1a). The top event is “protection fails to clear corresponding to the following:
an in-zone fault,” which means that this fault tree considers • Relay application or settings errors.
only protection for faults inside the differential zone defined • DC power system failures.
by the CT location. We assume the relay provides differential • DC wiring errors.
and restricted earth fault (REF) protection functions [1]. The • CT wiring errors.
basic events considered in this fault tree are: relay failures, • Hidden failures.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 163

Fig. 3 shows that comprehensive commissioning testing In the fault tree shown in Fig. 5, we assume the breakers to
reduces the single scheme unavailability from 1,807 • 10–6 to have redundant trip coils, so we split the breakers into two
821 • 10–6, a significant improvement. parts. We represent breaker trip coil failures or dc circuit fuse
821
operations at the basic level (under OR Gate 1). Their
contribution to a failure to clear the fault is practically
Note: Numbers shown are Protection Fails to Clear eliminated by the AND gate. If the trip coils operate correctly,
unavailabilities • 106 an In-Zone Fault
a breaker failure to interrupt current (a stuck contact
mechanism or a failure of the contacts to extinguish the arc)
will cause a failure to clear the fault, no matter the redundancy
of the scheme. Hence we represent breaker failures to interrupt
current above the AND gate in Fig. 5 as inputs to OR Gate 2.
Because the other input to this OR gate has a very low
unavailability value (because of redundancy), the breaker
failures to interrupt current become the dominant factor in the
scheme dependability. This fact emphasizes the importance of
Relay Relay HV LV DC CT DC CT Hidden
Fails App. or Breaker Breaker System Fails Wiring Wiring Failures
good breaker maintenance and monitoring and the need for
137 Settings Fails Fails Fails 7•9 Errors Errors 5 breaker failure protection. A more expensive solution would
Errors 200 200 6 = 63 10 0
200
be to install two breakers in series in a critical system.
Fig. 3. Effect of comprehensive commissioning testing on the dependability
fault tree for the single transformer protection scheme.

Fig. 4 shows the security fault tree for the single scheme.
The top event is “protection produces an undesired trip.” This
security fault tree considers the same events as the
corresponding dependability fault tree shown in Fig. 2 but
uses the security failure rates shown in Table VI in the
appendix.
4981

Note: Numbers shown Protection Produces an

are failure rates • 106 Undesired Trip

Relay Relay HV LV DC CT DC CT Hidden

Fails App. or Breaker Breaker System Fails Wiring Wiring Failures
333 Settings Fails Fails Fails 7 • 157 Errors Errors 50
Errors 333 333 1000 = 1099 250 583
1000

Fig. 4. Security fault tree for the single transformer protection scheme.

2) Redundant Schemes
Fig. 5 depicts the fault tree for the dual-redundant
transformer protection scheme, which results from adding a
Fig. 5. Dependability fault tree for the dual-redundant transformer
second neutral CT to the scheme shown in Fig. 1b. The AND protection scheme, using relays from the same manufacturer.
gate reflects the fact that the failure of any redundant
component does not cause a failure to clear a fault. The inputs Fig. 6 depicts the fault tree for the dual-redundant
to this gate have slightly different values because, in a scheme transformer protection scheme with only one neutral CT (see
with relays from the same manufacturer, we assign slightly Fig. 1b), which is a typical scheme. All the scheme
different unavailabilities to relay application and settings components are redundant, except the neutral CT, which
errors, as explained in the appendix. The multiplication of constitutes a single point of failure for the scheme. Neutral CT
unavailabilities reduces the output of the AND gate to a value failures affect REF protection, but not differential protection.
close to zero. We can modify this fault tree as required to
represent systems with lower redundancy (single neutral CT or
dc power system or breaker trip coil, for example).
164

The right side of the Fig. 6 fault tree represents the
differential protection. The output of OR Gate 2 is the
unavailability value resulting from differential protection
failing to operate. We multiply this value by 0.9 under the
assumption that differential protection detects 90 percent of all
internal faults (AND Gate 2). The output of AND Gate 2 is an
input to AND Gate 5, which represents differential protection
redundancy. AND Gate 5 practically eliminates the
contribution of differential protection to the top event.
The left side of the Fig. 6 fault tree represents the REF
protection. We represent neutral CT failures and neutral CT
wiring errors as inputs to OR Gate 3. We multiply the value of
the OR Gate 3 output by 0.1 (AND Gate 3) under the
assumption that ground low-current faults (detected only by
REF protection) represent 10 percent of all transformer
internal faults. The output of OR Gate 1 represents all the
other events that affect REF protection. This output does not
include the contribution of the phase CTs connected on the
transformer delta side because REF protection does not use
currents from the delta side. We use the output of OR Gate 1
(represented by the triangle symbol) multiplied by 0.1 in AND
Gate 1 as an input to AND Gate 4. This AND gate, which
represents protection redundancy, practically eliminates the
contribution of this part of REF protection to the top event.
The single neutral CT represents the greatest contribution of
REF protection to the top event. Adding a second neutral CT
moves the failure events related to this CT to the lowest level
in the fault tree, as shown in Fig. 5.

165

Note: Numbers shown

Protection Fails to Clear
are unavailabilities • 106
an In-Zone Fault

3 0 2

3 4 5

HV Breaker LV Breaker
Fails to Fails to
26 154 180 1459 1684
Interrupt Interrupt
Current Current
3 Main 1 REF Main 2 REF Main 1 Differential Main 2 Differential
80 80
Protection Fails Protection Fails Protection Fails Protection Fails
Fault
Detected
by REF 1459
0.1
1 2
Neutral Neutral
CT Fails CT Wiring
9 Errors
17
1621

Fault 2
Fault
Detected 1544 Detected
by REF by Differential
0.1 0.9

1544

CT CT
Fails Wiring
3•9 Errors
1 = 27 50

Relay Relay HV LV DC CT DC CT Hidden

Fails App. or Breaker Breaker System Fails Wiring Wiring Failures
137 Settings Trip Coil Trip Coil Fails 3•9 Errors Errors 10
Errors Fails Fails 30 = 27 50 50
1000 120 120

Fig. 6. Dependability fault tree for the dual-redundant transformer protection scheme with one neutral CT.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 165

The topology of the fault tree for the dual-redundant
scheme using relays from different manufacturers is identical
to that of Fig. 5. In this fault tree, we use a higher
unavailability value for relay application and settings errors
than for the case of relays from the same manufacturer, as
explained in the appendix.
Fig. 7 represents the effect of common-mode failures on
the dependability of the dual-redundant transformer protection
scheme. To create this fault tree, we started from the Fig. 5
fault tree and added the common-mode failures at the same
level as breaker failures to interrupt current. We split
common-mode failures into two types: failures that result
from the hardware or firmware of two devices failing
simultaneously and those that result from common errors in
device settings or in system design.

Fig. 7. Dependability fault tree for the dual-redundant transformer protection scheme, considering common-mode failures.
166

Fig. 8 shows the security fault tree for the dual-redundant undesired trip. The result is lower security (a higher failure
transformer protection scheme. OR Gate 2 reflects the effect rate) than that of the single scheme (see Fig. 4).
of redundancy: any of the two schemes may cause an

Fig. 8. Security fault tree for the dual-redundant transformer protection scheme.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 167

Fig. 9 shows the dependability fault tree for the two-out-of-
three voting transformer protection scheme. The scheme has
three multifunction relays. Tripping occurs when at least two
of the relays operate. The effect of the voting logic is that the
output of AND Gate 3 is practically zero. A fully redundant
voting scheme (having three sets of each scheme component)
would have a very high dependability, only limited by the
breaker failures to interrupt current [10]. However, in Fig. 9,
we assume that two of the schemes (referred to as Main 2 and
Main 3) share the dc power system and the CTs. The shared
components become single points of failure for the Main 2
and Main 3 schemes. When one of these components fails,
both schemes fail simultaneously and the voting scheme fails
to clear the fault. For this reason, we represent dc power
system failures, CT failures, and CT wiring errors at the same
level as breaker failures to interrupt current, as inputs to OR
Gate 3. The result is lower dependability (a higher failure rate)
than in a fully redundant voting scheme. We keep dc system
wiring errors at the lowest level in the fault tree because we
assume that dc circuits for the Main 2 and Main 3 schemes are
independent even with a common battery. We keep breaker
trip coil failures at the lowest level in the fault tree because we
assume that the voting scheme is arranged to energize both
breaker trip coils [10]. In this analysis, we assume the three
schemes have the same sensitivity. If the schemes had
different sensitivities (because of different settings, principles
of operation, or manufacturers) and if two of the schemes did
not detect a high-resistance in-section fault, the two-out-of-
three voting scheme would fail to clear the fault. For this
reason, we recommend that voting schemes use relays with the
same sensitivity.

370

Note: Numbers shown Protection Fails to Clear

are unavailabilities • 106 an In-Zone Fault

0
HV Breaker LV Breaker DC CT CT
Fails to Fails to System Fails Wiring
Interrupt Interrupt 3 Fails 7•9 Errors
Current Current 30 = 63 117
80 80

3 3 3

Main 1 and Main 2 Main 1 and Main 3 Main 2 and Main 3

Transformer Transformer Transformer
Protection Fails Protection Fails Protection Fails

1 2

1647 1897 1687 1687

Main 1 Transformer Main 2 Transformer Main 2 Transformer Main 3 Transformer

Protection Fails Protection Fails Protection Fails Protection Fails

Same as Main 3
Transformer Protection

1 2

Relay Relay HV LV DC CT DC CT Hidden Relay Relay HV LV Hidden DC

Fails App. or Breaker Breaker System Fails Wiring Wiring Failures Fails App. or Breaker Breaker Failures Wiring
137 Settings Trip Coil Trip Coil Fails 7•9 Errors Errors 10 137 Settings Trip Coil Trip Coil 10 Errors
Errors Fails Fails 30 = 63 50 117 Errors Fails Fails 50
1000 120 120 1250 120 120

Fig. 9. Dependability fault tree for the two-out-of-three voting transformer protection scheme.
168

Fig. 10. Security fault tree for the two-out-of-three voting transformer protection scheme.

Fig. 10 shows the security fault tree for the two-out-of-
three voting transformer protection scheme. AND Gates 1
and 2 reflect the fact that two schemes need to misoperate to
cause an undesired trip. The result is high security (a low
failure rate). However, the shared dc system and CTs, which
we represent as inputs to OR Gate 3, impair the scheme
security.
C. Protection Scheme Reliability Comparison
Table II summarizes the results obtained from the
24 transformer protection fault trees.
From Table II, we conclude the following:
• Comprehensive commissioning testing improves the
single scheme dependability 2.2 times.
• In dual-redundant schemes, the effect of breaker
failures to interrupt current significantly reduces the
impact of comprehensive commissioning testing on
dependability. When considering common-mode
failures, comprehensive commissioning testing
improves the dual-redundant scheme dependability
3.1 times. It also improves the voting scheme
dependability 1.6 times.
• Comprehensive commissioning testing improves
security between 1.9 and 2.2 times.
• The dependability of the dual-redundant scheme is
1,807/163 = 11.1 times that of the single scheme.
• In a dual-redundant scheme, adding a second neutral
CT does not significantly improve dependability. If
we assume that breaker failure protection meets the
performance requirements of the power system, the
second neutral CT improves the scheme dependability
5/3 = 1.7 times. We calculated these values by
considering the unavailability value for breaker
failures to interrupt current to be zero.
• If we assume that breaker failure protection meets the
performance requirements of the power system, the
dependability of the dual-redundant scheme is
6/3 = 2 times higher when using relays from the same
manufacturer than when using relays from different
manufacturers. We calculated these values by
considering the unavailability value for breaker
failures to interrupt current to be zero.
• The dependability of the voting scheme is
1,807/370 = 4.9 times that of the single scheme.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 169

TABLE II
TRANSFORMER PROTECTION RELIABILITY COMPARISON

Dependability (Unavailability • 106) Security (Failure Rate • 106)

Protection Scheme Normal Comprehensive Normal Comprehensive

Commissioning Commissioning Commissioning Commissioning
Testing Testing Testing Testing
Single 1,807 821 (2.2 times) 4,981 2,573 (1.9 times)
Dual redundant 163 160 (1.0 times) 9,946 4,930 (2.0 times)
Dual redundant with one neutral CT 165 163 (1.0 times) 10,020 5,087 (2.0 times)
Dual redundant with relays from
166 161 (1.0 times) 11,196 5,180 (2.2 times)
different manufacturers
Dual redundant with
668 213 (3.1 times) 10,471 4,955 (1.9 times)
common-mode failures
Redundant two-out-of-three voting 370 229 (1.6 times) 2,999 1,576 (1.9 times)

Note: The numbers in parentheses represent the effect of comprehensive commissioning testing. These numbers are the ratios of the unavailabilities or failure
rates with normal testing to the unavailabilities or failure rates with comprehensive testing.

• The dependability of the dual-redundant scheme is Table III summarizes the cost estimation results. It shows
370/163 = 2.3 times that of the voting scheme. Sharing that, for this example, converting the single scheme into a
the dc power system and the CTs affects the voting dual-redundant scheme costs $13,480 and converting the
scheme dependability. single scheme into a two-out-of-three voting scheme costs
• Common-mode failures impair the dependability of $21,110. This is a low price to pay for the protection scheme
the dual-redundant scheme 668/163 = 4.1 times. reliability improvement provided by redundancy, given the
Comprehensive commissioning testing and detailed high costs of transformer outages and repairs. However, if we
setting and design reviews reduce the dependability require the addition of a dc power system or a set of CTs to
impairment to 229/160 = 1.4 times. achieve full redundancy, we must consider their cost in the
• The security of the single scheme is comparison.
9,946/4,981 = 2.0 times that of the dual-redundant TABLE III
scheme. COST COMPARISON OF TRANSFORMER PROTECTION SCHEMES
• The security of the voting scheme is Protection Scheme
4,981/2,999 = 1.7 times that of the single scheme. Item Dual
Sharing the dc power system and the CTs affects the Basic Voting
Redundant
voting scheme security.
• The security of the voting scheme is Relays $5,860 $11,720 $17,580
9,946/2,999 = 3.3 times that of the dual-redundant Engineering $4,000 $5,000 $6,000
scheme. Wiring and testing $1,790 $2,510 $3,280
• The security of the dual-redundant scheme is
Field wiring $9,300 $15,200 $15,200
11,196/9,946 = 1.1 times higher when using relays
from the same manufacturer than when using relays Total cost $20,950 $34,430 $42,060
from different manufacturers. Incremental cost – $13,480 $21,110
D. Cost Comparison
We evaluated the costs resulting from adding redundancy VI. GENERATOR PROTECTION EVALUATION
to the single transformer protection scheme. We used a A. Generator Protection Schemes
computer program for protection system cost estimation to We compare the reliability of three schemes for protecting
determine the cost of the basic, dual-redundant, and voting a high-resistance-grounded generator connected in a unit
schemes. Our cost evaluation includes the following: arrangement with the step-up transformer. The unit has a
• Relays. single breaker on the transformer HV side and no generator
• Engineering (relay programming and panel wiring breaker. All the schemes use multifunction relays that provide
design). generator protection and include the step-up transformer in the
• Panel wiring and testing. unit differential zone. The relays do not provide dedicated
• Field wiring, including cable and labor costs transformer protection.
(assuming the distances from the instrument Fig. 11a shows the single protection scheme, which
transformers and breakers to the relays to be includes one relay, single sets of CTs and voltage transformers
300 meters). (VTs), one dc power system, and one breaker with a single
170
trip coil. Fig. 11b shows the dual-redundant protection
scheme, which includes two relays, two sets of instrument
transformers, two dc power systems, and a breaker with
redundant trip coils. To create a two-out-of-three voting
scheme, we add a third relay to the Fig. 11b scheme,
connected to the same instrument transformers and the same
dc power system as one of the other two relays. In redundant
schemes, we assume all the redundant components are of
similar quality. We also assume relays have the same
reliability indices, sensitivities, and operation speeds.
Fig. 11. Single and dual-redundant generator protection schemes using
multifunction relays.
B. Fault Tree Analysis
We created 20 fault trees (10 for dependability analysis and
10 for security analysis) for the following combinations of
protection schemes and other factors:
• Single scheme.
• Dual-redundant scheme.
• Dual-redundant scheme with relays from different
manufacturers.
• Dual-redundant scheme that experiences common-
mode failures.
• Redundant two-out-of-three voting scheme.
Table VI in the appendix shows the reliability indices that
we used in the fault trees.
Fig. 12 shows the dependability fault tree for the single
generator protection scheme (see Fig. 11a). The top event is
“protection fails to trip unit for in-zone faults or abnormal
conditions.” The fault tree includes the basic events
considered for transformer protection (see Fig. 2) and the
failures of the generator grounding system because this system
provides a voltage input signal to the relay. This fault tree
considers all the generator protection functions that the
multifunction relay can provide using the current and voltage
inputs shown in Fig. 11a. The fault tree does not include other
possible scheme modules, such as resistance temperature
detector (RTD) modules for thermal protection, ground
modules for field ground fault protection, or signal injection
modules for stator ground fault protection. In this fault tree,
we use the unavailability values shown in Table VI in the
appendix. We can modify the fault tree as required to consider
other scheme configurations, to include other events of
interest, or to use other unavailability values. The other fault
trees for the single generator protection scheme (dependability
fault tree considering the effect of comprehensive
commissioning testing and security fault trees) have the same
topology as that of Fig. 12 but different reliability indices.
Fig. 13 depicts the dependability fault tree for the dual-
redundant generator protection scheme shown in Fig. 11b. All
the scheme components are redundant, except the generator
grounding system, composed of a transformer with a resistor
connected to its secondary. The transformer provides a voltage
input signal to the relays. The relays use this signal and the
zero-sequence voltage measured at the generator terminals to
provide 100 percent stator ground fault protection. This
protection combines a neutral overvoltage element with a
third-harmonic voltage differential element [1]. The generator
grounding system is a single point of failure for stator ground
fault protection. As a result, the fault tree topology is similar
to that of the dual-redundant transformer protection scheme
with one neutral CT (Fig. 6).
The left side of the Fig. 13 fault tree represents the stator
ground fault protection. We assume stator ground faults to be
30 percent of all generator internal faults and abnormal
operating conditions, so we use a 0.3 multiplier as an input to
AND Gates 1 and 3.
The right side of the Fig. 13 fault tree represents all the
other generator protection functions. The output of OR Gate 2
is the unavailability value resulting from these other protection
functions failing to operate. We use a 0.7 multiplier as an
input to AND Gate 2 in this case. AND Gates 4 and 5
practically eliminate the contribution of the redundant
protection scheme to the top event. Hence the breaker failures
to interrupt current and the generator grounding system
failures determine the protection scheme dependability.
We created the other fault trees for the dual-redundant
generator protection scheme (security fault trees and
dependability fault trees considering the effect of
comprehensive commissioning testing and common-mode
failures) using the same methodology as for dual-redundant
transformer protection schemes.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 171

Fig. 12. Dependability fault tree for the single generator protection scheme.
87

Protection Fails to Trip

Note: Numbers shown
Unit for In-Zone Faults or
are unavailabilities • 106
Abnormal Conditions

5 0 2

3 4 5

Breaker
Fails to
432 508 Interrupt 1225 1400
Current
Main 1 Stator Ground Main 2 Stator Ground 80 Main 1 Generator Main 2 Generator
Fault Protection Fails Fault Protection Fails Protection Fails Protection Fails

Fault Neutral
Detected Grounding
by Ground System
Fault 1 1225
Fails
Protection 15
0.3 2

1750

Fault 1442 Fault Detected 2

Detected by Generator
by Ground Protection
Fault 0.7
Protection
0.3
1442

CT CT
Fails Wiring
4 • 27 Errors
1442 = 108 4 • 50
= 200

Relay Relay Breaker DC VT VT DC Hidden

Fails App. or Trip Coil System Fails Wiring Wiring Failures
137 Settings Fails Fails 3 • 15 Errors Errors 10
Errors 120 30 = 45 50 50
1000

Fig. 13. Dependability fault tree for the dual-redundant generator protection scheme using relays from the same manufacturer.
172

Fig. 14 shows the dependability fault tree for the two-out-
of-three voting generator protection scheme. The scheme has
three multifunction relays. Tripping occurs when at least two
of the relays operate. We assume that two of the schemes
(referred to as Main 2 and Main 3 in Fig. 14) share the dc
power system and instrument transformers, which become
single points of failure for the voting scheme. We represent dc
power system failures, instrument transformer failures, and
wiring errors at the same level as breaker failures to interrupt
current, as inputs to OR Gate 2. We represent the neutral
grounding system, another single point of failure of the voting
scheme, as another input to OR Gate 2. Because of the scheme
redundancy, the outputs of AND Gates 2 and 5 are zero.
Hence the shared component, the breaker failures to interrupt
current, and the grounding system failures determine the
scheme dependability. In this analysis, we assume that the
three schemes detect the same faults and abnormal conditions.
If the schemes had different fault or abnormal condition
coverage (because of different settings, principles of
operation, or manufacturers) and if two of the schemes did not
detect an event, the two-out-of-three voting scheme would fail
to trip the generator. For this reason, we recommend that
voting schemes use relays with the same fault and abnormal
condition coverage.

518

Protection Fails to Trip

Note: Numbers shown
Unit for In-Zone Faults or
are unavailabilities • 106
Abnormal Conditions

Breaker DC CT CT VT VT
Fails to System Fails Wiring Fails Wiring 0
Interrupt Fails 4 • 27 Errors 3 • 15 Errors
Current 30 = 108 200 = 45 50 5
80

Fault Neutral
Detected Grounding 2 2 1
by Ground System
Fault Fails Main 1 and Main 2 Main 1 and Main 3 Main 2 and Main 3
Protection 15 Generator Protection Fails Generator Protection Fails Generator Protection Fails
0.3 0

Output of
2 AND Gate 5 in Fig. 13 3

0 0 0 1097 1097

Main 1 and Main 2 Stator Main 1 and Main 3 Stator Main 2 and Main 3 Stator Main 2 Generator Main 3 Generator
Ground Fault Protection Fails Ground Fault Protection Fails Ground Fault Protection Fails Protection Fails Protection Fails

Output of
AND Gate 4 in Fig. 13

1567

Fault 1
Detected
by Generator
Protection
0.7

Relay Relay Breaker DC Hidden

Fails App. or Trip Coil Wiring Failures
137 Settings Fails Errors 10
Errors 120 50
1250

Fig. 14. Dependability fault tree for the two-out-of-three voting generator protection scheme.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 173

TABLE IV
GENERATOR PROTECTION RELIABILITY COMPARISON

Dependability (Unavailability • 106) Security (Failure Rate • 106)

Protection Scheme Normal Comprehensive Normal Comprehensive

Commissioning Commissioning Commissioning Commissioning
Testing Testing Testing Testing
Single 1,845 726 (2.5 times) 7,152 4,077 (1.8 times)
Dual redundant 87 85 (1.0 times) 14,143 7,793 (1.8 times)
Dual redundant with relays from
88 85 (1.0 times) 15,393 8,043 (1.9 times)
different manufacturers
Dual redundant with
592 138 (4.3 times) 14,668 7,693 (1.9 times)
common-mode failures
Redundant two-out-of-three voting 518 244 (2.1 times) 5,568 3,481 (1.6 times)

Note: The numbers in parentheses represent the effect of comprehensive commissioning testing. These numbers are the ratios of the unavailabilities or failure
rates with normal testing to the unavailabilities or failure rates with comprehensive testing.

C. Protection Scheme Reliability Comparison • The security of the voting scheme is

Table IV summarizes the results obtained from the 7,152/5,568 = 1.3 times that of the single scheme.
20 generator protection fault trees. Sharing the dc power system and the instrument
From Table IV, we conclude the following: transformers affects the voting scheme security.
• Comprehensive commissioning testing improves the • The security of the voting scheme is
single scheme dependability 2.5 times. 14,143/5,568 = 2.5 times that of the dual-redundant
• In dual-redundant schemes, the effect of breaker scheme.
failures to interrupt current and the common use of the • The security of the dual-redundant scheme is
voltage signal from the grounding transformer 15,393/14,143 = 1.1 times higher when using relays
significantly reduce the impact of comprehensive from the same manufacturer than when using relays
commissioning testing on dependability. When from different manufacturers.
considering common-mode failures, comprehensive D. Cost Comparison
commissioning testing improves the dual-redundant We evaluated the costs resulting from adding redundancy
scheme dependability 4.3 times. It also improves the to a generator protection scheme. Our cost evaluation
voting scheme dependability 2.1 times. includes:
• Comprehensive commissioning testing improves • Relays.
security between 1.6 and 1.9 times. • Engineering (relay programming and panel wiring
• The dependability of the dual-redundant scheme is design).
1,845/87 = 21.2 times that of the single scheme. • Panel wiring and testing.
• The dependability of the voting scheme is • Field wiring, including cable and labor costs
1,845/518 = 3.6 times that of the single scheme. (assuming the distances from the instrument
• The dependability of the dual-redundant scheme is transformers and breakers to the relays to be
518/87 = 6.0 times that of the voting scheme. Sharing 150 meters).
the dc power system and the instrument transformers Table V summarizes the cost estimation results.
affects the voting scheme dependability. TABLE V
• The dual-redundant scheme has the same COST COMPARISON OF GENERATOR PROTECTION SCHEMES
dependability when using relays from the same
Protection Scheme
manufacturer as when using relays from different
manufacturers. Item Dual
Basic Voting
• Common-mode failures impair dependability of the Redundant
dual-redundant scheme 592/87 = 6.8 times. Relays $5,860 $11,720 $17,580
Comprehensive commissioning testing and detailed Engineering $4,000 $5,000 $6,000
setting and design reviews reduce the dependability
Wiring and testing $1,790 $2,510 $3,240
impairment to 138/85 = 1.6 times.
• The security of the single scheme is Field wiring $6,880 $12,540 $13,540
14,143/7,152 = 2.0 times that of the dual-redundant Total cost $18,530 $31,770 $40,360
scheme.
Incremental cost – $13,240 $21,830
174
Table V shows that, for this example, converting the single
scheme into a dual-redundant scheme costs $13,240 and
converting the single scheme into a two-out-of-three voting
scheme costs $21,830. This is a low price to pay for the
protection scheme reliability improvement provided by
redundancy, given the high costs of generator outages and
repairs. However, if we require the addition of a dc power
system or a set of instrument transformers to achieve full
redundancy, we must consider their cost in the comparison.
VII. CONCLUSIONS
From the transformer and generator protection schemes
studied in this paper, we conclude the following:
• Fault tree analysis is an excellent tool to compare the
relative reliability of protection schemes. The
topologies and reliability indices of fault trees used for
dependability analysis are different from those used
for security analysis.
• This paper shows dependability and security fault
trees for typical single, dual-redundant, and voting
protection schemes for transformers and generators.
We also show how to consider the effect of
comprehensive commissioning testing, hidden
failures, common-mode failures, and the use of relays
from the same or different manufacturers in redundant
schemes.
• The paper provides a table of reliability indices for use
in fault trees. We calculated some indices from
measured field data. However, other indices come
from technical literature or were estimated based on
experience. Utility engineers can refine these indices
by using field data from their power systems.
• Comprehensive commissioning testing improves the
dependability of single and voting protection schemes.
When considering common-mode failures, this testing
also improves the redundant scheme dependability.
• Comprehensive commissioning testing improves the
security of all transformer and generator protection
schemes.
• Dual-redundant transformer and generator protection
schemes have higher dependability and lower security
than single schemes.
• In transformer and generator protection schemes, the
voting scheme has the highest security, and the dual-
redundant scheme has the lowest security.
• Adding a second neutral CT improves the
dependability of redundant transformer protection
schemes when breaker failure protection meets the
power system requirements.
• Breaker failures to interrupt current have a significant
impact on the dependability of redundant schemes. A
breaker failure to interrupt current causes a failure to
trip the transformer or generator, no matter the
redundancy of the rest of the scheme. To improve the
redundant scheme dependability, we recommend
providing good breaker maintenance and applying
breaker failure protection.
• In generator protection schemes, the grounding system
represents a single point of failure that limits the effect
of redundancy on the scheme dependability.
• Using relays from the same manufacturer improves
the dependability and security of redundant protection
schemes.
• Hidden and common-mode failures do not
significantly affect the reliability of redundant
protection schemes.
• Common-mode failures affect the reliability of
redundant protection schemes. Comprehensive
commissioning testing and detailed setting and design
reviews significantly reduce the dependability
impairment.
• The costs of converting a single transformer or
generator protection scheme into a dual-redundant
scheme or a voting scheme are relatively low, unless
this conversion requires adding a dc power system or
instrument transformers.
VIII. APPENDIX
This appendix shows the reliability indices that we used in
the dependability and security fault trees in this paper. We
also explain how we calculated or estimated these indices. We
have confidence in the relay failure rates because we have
measured them for many years. We estimated other indices
based on our experience and the information available in
technical literature.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 175

A. Reliability Indices Used in Fault Trees 45 percent of the incorrect operations, while relay failures
Table VI shows the unavailability values that we used for caused only 4.5 percent of the incorrect operations. Using this
dependability fault trees and the failure rate values that we information, we assume U = 1,000 • 10–6 for relay application
used for security fault trees. We also show the MTBF values or settings errors.
that we used to calculate the failure rates. We assume this value falls 80 percent (U = 200 • 10–6) with
comprehensive commissioning testing and by analyzing relay
TABLE VI
RELIABILITY INDICES USED IN FAULT TREES event reports to find application or settings errors.
For two identical relays, we use U = 1,000 • 10–6 for one
Dependability Security relay and U = 1,250 • 10–6 for the other relay to account for
Event MTBF Failure possible additional errors when manually applying settings to
Unavailability • 106
(Years) Rate • 106 this other relay. For two relays from different manufacturers,
Relay fails 137 3,000 333 we assume the resulting unavailability to be close to the sum
of the relay unavailabilities because of the differences in
Relay application
or settings errors
1,000 1,000 1,000 application considerations and settings rules. Hence we use
U = 1,750 • 10–6 for each relay. We assume these values fall
Breaker fails 200 3,000 333
80 percent (U = 200 • 10–6, U = 250 • 10–6, and U = 350 • 10–6,
Breaker fails to
80 – – respectively) with comprehensive commissioning testing and
interrupt current by analyzing relay event reports to find application or settings
DC power
30 1,000 1,000
errors.
system fails
3) Breaker Fails
CT fails 9 6,370 157
References [13] and [14] provide utility breaker failure data
VT fails 15 3,600 278 collected in a CIGRÉ survey for the 1988 to 1991 period for
Generator breakers between 62.5 kV and greater than 700 kV. The
grounding 15 3,600 278 reported failure rate for all the breakers is λ = 6,720 • 10–6,
system fails which gives MTBF = 149 years. Assuming that half of these
DC system
50 4,000 250
failures are failures to open (a dependability problem), we can
wiring errors use MTBF = 300 years for dependability analysis.
CT or VT Hence, for breakers with one tripping coil, we use MTBF =
50 4,000 250
wiring errors 300 years and calculate the unavailability assuming the
Hidden failures 10 20,000 50 following [7]:
Common-mode
• Ninety percent of failures are detected by the usual
failures (hardware 5 40,000 25 monitors in the breaker and in some relays (breaker
or firmware) monitoring, event reporting, trip and close circuit
Common-mode monitoring) and other devices.
failures (settings 500 2,000 500 • Another 5 percent of failures are detected by visual
or design errors) inspections every two months.
B. Comments on the Reliability Indices for Dependability • The remaining 5 percent of failures are detected by
Fault Trees (Unavailabilities) maintenance every two years.

1) Relay Fails
Our calculation using observed field failure data gives
MTBF = 100 years (λ = 10,000 • 10–6) for dependability
analysis. This MTBF value includes hardware and firmware
failures and the effect of taking the relay out of service for
corrective actions derived from service bulletins.
Reference [11] gives an interval from 30 minutes to 2 weeks
for MTTR. Assuming an average value of MTTR = 5 days, we
have:
U = λ • MTTR = (0.01 failures/year) (120 hours) (1/8,760
hours/year) = 137 • 10–6
2) Relay Application or Settings Errors
Experience shows that relay application and settings errors
cause more protection dependability problems than relay
failures. For example, [12] analyzes incorrect protection
operations in a utility during an 18-month period and
concludes that settings errors and other human errors caused
U=
1 ⎛ 0.90 • 2 days 0.05 •1 month ⎞
•⎜ + + 0.05 •1 year ⎟
300 years ⎝ 365 days/year 12 months/year ⎠
= 197 •10 –6
Hence, for breakers with one trip coil, we use U = 200 • 10–6.
For breakers with redundant trip coils, we use U = 80 • 10–6 to
account for the increased reliability resulting from trip coil
redundancy and from the lower impact of blown fuses in the
dc power circuits.
4) Breaker Fails to Interrupt Current
In redundant schemes, a breaker failure to interrupt current
causes a failure to clear the fault, no matter the redundancy of
the rest of the scheme. For this reason, in dependability fault
trees, we represent breaker failures to interrupt current
separately from trip coil failures and blown fuses in the dc
tripping circuits. According to [14] and [15], which report
176
breaker failure data collected in a CIGRÉ survey for the 1974
to 1977 period, mechanical failures are around 70 percent of
all breaker failures. Assuming that half the mechanical failures
are caused by stuck trip coils, we can estimate that breaker
failures to interrupt current after the trip coil operates
represent around 40 percent of all breaker failures. Hence we
use U = 0.4 • 200 • 10–6 = 80 • 10–6 for breaker failures to
interrupt current and U = (200 – 80) • 10–6 = 120 • 10–6 for all
the other breaker failures.
5) DC Power System Fails
We use U = 30 • 10–6 according to [16]. We assume this
value falls 80 percent (U = 6 • 10–6) when we provide proper
battery maintenance, monitor the system voltage and the
battery charger, and use efficient ground detection systems.
We consider redundant dc power systems to have redundant
batteries, battery chargers, and wiring.
6) CT Fails
Reference [14] provides instrument transformer failure data
collected in a CIGRÉ survey for the 1985 to 1995 period. The
reported failure rate for all CT failures that result in a CT
outage is λ = 1,570 • 10–6, which gives MTBF = 637 years.
For an MTTR = 2 days, we get U = 8.6 • 10–6 and will use
U = 9 • 10–6 per CT.
7) VT Fails
According to [14], the failure rate for all VT failures that
result in a VT outage is around λ = 2,800 • 10–6. Hence MTBF =
360 years. For an MTTR = 2 days, we get U = 15.2 • 10–6 and
will use U = 15 • 10–6 per VT.
8) Generator Grounding System Fails
The generator grounding system consists of a transformer
with its primary connected between the generator neutral and
ground and with a resistor connected to its secondary.
Generator stator ground fault protection receives voltage
information from the secondary of the generator grounding
transformer. A transformer or resistor failure could cause the
stator ground fault protection to fail to operate. Assuming that
the frequency of generator grounding transformer or resistor
failures is comparable to that of VT failures, we use
MTBF = 360 years, which gives U = 15 • 10–6 for MTTR =
2 days.
9) DC System Wiring Errors
Experience shows that dc system wiring errors cause more
protection reliability problems than dc power system failures.
We assume the unavailability caused by dc system wiring
errors to be U = 50 • 10–6. We assume this value falls
80 percent (U = 10 • 10–6) with comprehensive commissioning
testing and by analyzing relay event reports to find dc system
wiring errors.
10) CT or VT Wiring Errors
We assume the unavailability caused by CT or VT wiring
errors to be equal to that caused by dc system wiring errors.
Hence we use U = 50 • 10–6 per CT or VT three-phase circuit.
We assume this value falls to zero (U = 0) with
comprehensive commissioning testing (using the advanced
commissioning features available in modern relays) and by
analyzing relay event reports to find CT or VT wiring errors.
11) Hidden Failures
Hidden failures are very infrequent events. We assume the
unavailability caused by hidden failures is less than 10 percent
of that caused by a relay failure. This is based on experience
and the assumption that hidden failure unavailability must be
less than known and measured data. Hence we use
U = 10 • 10–6 for hidden failures. We assume this value falls to
U = 5 • 10–6 with comprehensive commissioning testing and
by analyzing relay event reports.
12) Common-Mode Failures
Common-mode failures may result from the hardware or
firmware of two devices failing simultaneously or from
common errors in device settings or in system design.
We assume the common-mode failures caused by hardware
or firmware problems to be even less frequent than hidden
failures. For example, the probability of a relay component
failing at the same time in two redundant relays is very low,
even if this component has an abnormally high failure rate.
Hence we use U = 5 • 10–6 for these common-mode failures.
We assume this value falls to U = 3 • 10–6 with comprehensive
commissioning testing and by analyzing relay event reports.
We assume the common-mode failures caused by settings
or design errors to be around half the failures caused by relay
application and settings errors. Hence we use U = 500 • 10–6
for these common-mode failures. We assume this value falls
90 percent (U = 50 • 10–6) by carefully reviewing settings and
designs and analyzing relay event reports.
C. Comments on the Reliability Indices for Security Fault
Trees (Failure Rates)
1) Relay Fails
Relays are typically designed to fail in a safe mode, not to
trip. Our calculation using observed field failure data gives
MTBF = 3,000 years (λ = 333 • 10–6) for security analysis.
2) Relay Application or Settings Errors
Experience shows that relay application and settings errors
cause more protection security problems than relay failures.
Hence we assume MTBF = 1,000 • 10–6 (λ = 1,000 • 10–6) for
relay application or settings errors. We assume this value falls
to λ = 200 • 10–6 with comprehensive commissioning testing
and by analyzing relay event reports to find application or
settings errors.
3) Breaker Fails
Assuming that the breaker failures that cause undesired
closures are around ten times less likely than the breaker
failures that cause failures to open, we define MTBF =
10 • 300 = 3,000 years (λ = 333 • 10–6) for breakers with one
trip coil. For breakers with redundant trip coils, we define
MTBF = 3,000 / 0.4 = 7,500 years (λ = 133 • 10–6).
4) DC Power System Fails
The U = 30 • 10–6 value that we adopted for dependability
analysis represents MTBF = 100 years for an MTTR of
one day, which is typical for battery systems. Assuming that
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
the dc power system failures that cause undesired trips are
around ten times less likely than the dc power system failures
that cause failures to trip (a conservative assumption), we
define MTBF = 1,000 years (λ = 1,000 • 10–6) for security
analysis. We assume this value falls 80 percent (λ = 200 • 10–6)
when we provide proper battery maintenance, monitor the
system voltage and battery charger, and use efficient ground
detection systems.
5) CT Fails
Assuming that the CT failures that cause undesired trips are
around ten times less likely than the CT failures that cause
failures to trip, we define MTBF = 6,370 years (λ = 157 • 10–6)
for security analysis.
6) VT Fails
Assuming that the VT failures that cause undesired trips
are around ten times less likely than the VT failures that cause
failures to trip, we define MTBF = 3,600 years (λ = 278 • 10–6)
for security analysis.
7) Generator Grounding System Fails
Assuming that the generator grounding transformer or
resistor failures that cause undesired trips are around ten times
less likely than those failures that cause failures to trip, we
define MTBF = 10 • 360 = 3,600 years (λ = 278 • 10–6) for
security analysis.
8) DC System Wiring Errors
Experience shows that relay application and settings errors
cause more protection security problems than dc system
wiring errors. Hence we assume MTBF = 4,000 years
(λ = 250 • 10–6) for dc system wiring errors. We assume this
value falls 80 percent (λ = 50 • 10–6) with comprehensive
commissioning testing and by analyzing relay event reports to
find dc system wiring errors.
9) CT or VT Wiring Errors
Experience shows that relay application and settings errors
cause more protection security problems than CT or VT
wiring errors, which are comparable with dc system wiring
errors. Hence we assume MTBF = 4,000 years (λ = 250 • 10–6)
for CT or VT wiring errors. We assume this value falls to zero
(λ = 0) with comprehensive commissioning testing (using the
advanced commissioning features available in modern relays)
and by analyzing relay event reports to find CT or VT wiring
errors.
10) Hidden Failures
Assuming it takes around six months to detect a hidden
failure (MTTR = 0.5 years), the U = 10 • 10–6 value that we
adopted for dependability analysis represents MTBF =
50,000 years. Assuming that hidden failures have the same
likelihood of causing failures to trip as causing undesired
trips, we assume a conservative value of MTBF =
20,000 years (λ = 50 • 10–6) for security analysis. We assume
this value falls to λ = 25 • 10–6 with comprehensive
commissioning testing and by analyzing relay event reports.
177
11) Common-Mode Failures
Assuming it takes around six months to detect a common-
mode failure (MTTR = 0.5 years), the U = 5 • 10–6 and
U = 500 • 10–6 values that we adopted for dependability
analysis represents MTBF = 100,000 and MTBF = 1,000 years,
respectively. We assume that common-mode failures have the
same likelihood of causing failures to clear faults as causing
undesired trips. Hence, for security analysis, we assume a
conservative value of MTBF = 40,000 years (λ = 25 • 10–6) for
failures caused by hardware or firmware problems and
MTBF = 2,000 years (λ = 500 • 10–6) for failures caused by
settings or design errors. We assume these values fall to
λ = 15 • 10–6 and λ = 50 • 10–6, respectively, with
comprehensive commissioning testing and by analyzing relay
event reports.
IX. REFERENCES
[1] H. J. Altuve Ferrer and E. O. Schweitzer, III (eds.), Modern Solutions
for Protection, Control, and Monitoring of Electric Power Systems.
Schweitzer Engineering Laboratories, Inc., Pullman, WA, 2010.
[2] M. J. Thompson, “The Power of Modern Relays Enables Fundamental
Changes in Protection and Control System Design,” proceedings of the
60th Annual Conference for Protective Relay Engineers, College
Station, TX, March 2007.
[3] J. Sykes, V. Madani, J. Burger, M. Adamiak, and W. Premerlani,
“Reliability of Protection Systems – What Are the Real Concerns?,”
proceedings of the 63rd Annual Conference for Protective Relay
Engineers, College Station, TX, March 2010.
[4] D. Costello, “Fly Safe and Level: Customer Examples in Implementing
Dual Primary Protection Systems.” Available: http://www.selinc.com.
[5] Newton-Evans Research Company, Worldwide Study of the Protective
Relay Marketplace in Electric Utilities, 2006–2008, Volume 1, North
American Market, 2008.
[6] P. M. Anderson, Power System Protection. New York: IEEE
Press/McGraw-Hill, 1999.
[7] E. O. Schweitzer, III, B. Fleming, T. J. Lee, and P. M. Anderson,
“Reliability Analysis of Transmission Protection Using Fault-Tree
Methods,” proceedings of the 24th Annual Western Protective Relay
Conference, Spokane, WA, October 1997.
[8] K. Zimmerman, “Commissioning of Protective Relay Systems,”
proceedings of the 34th Annual Western Protective Relay Conference,
Spokane, WA, October 2007.
[9] K. Zimmerman and D. Costello, “Lessons Learned From
Commissioning Protective Relay Systems,” proceedings of the 62nd
Annual Conference for Protective Relay Engineers, College Station, TX,
March 2009.
[10] E. O. Schweitzer, III, D. Whitehead, H. J. Altuve Ferrer,
D. A. Tziouvaras, D. A. Costello, and D. Sánchez Escobedo, “Line
Protection: Redundancy, Reliability, and Affordability,” proceedings of
the 37th Annual Western Protective Relay Conference, Spokane, WA,
October 2010.
[11] IEEE PSRC Working Group I-19, “Redundancy Considerations for
Protective Relaying Systems,” A Report to the IEEE Power System
Relaying Committee, 2010. Available: http://www.pes-psrc.org/.
[12] R. Moxley, “Analyze Relay Fault Data to Improve Service Reliability,”
proceedings of the 30th Annual Western Protective Relay Conference,
Spokane, WA, October 2003.
[13] Study Committee 13, Working Group 13.06, “Final Report of the
Second International Enquiry on High Voltage Circuit-Breaker Failures
and Defects in Service,” CIGRÉ Technical Brochure No. 83, 1994.
[14] C. E. Sölver, “Past Cigré Surveys on Reliability of HV Equipment,”
CIGRÉ presentation, June 2006. Available: http://www.mtec2000.com/
cigre_a3_06/Rio/past.pdf.
178

[15] G. Mazza and R. Michaca, “The First International Enquiry on Circuit-
Breaker Failures and Defects in Service,” Electra (CIGRÉ), No. 79,
pp. 21–91, December 1981.
[16] R. Sandoval and J. Leon, “Evaluation of Methods for Breaker Flashover
Protection,” proceedings of the 31st Annual Western Protective Relay
Conference, Spokane, WA, October 2004.
X. BIOGRAPHIES
Ramón Sandoval is a protection engineer for Comisión Federal de
Electricidad (CFE) at the Topolobampo Thermal Power Station. He has
worked for CFE since 1992 in the electrical maintenance of power and
industrial equipment such as induction motors, synchronous generators,
breakers, automatic voltage regulators (AVRs), and step-up transformers. For
the last five years, Mr. Sandoval has been a power station protection engineer,
installing, testing, and applying different types of protective equipment
commonly used in industrial plants and power systems. This includes a
variety of electromechanical, static, and digital multifunction relays. He
received training in power system modeling and simulation from LAPEM
using ATP and has worked on developing field procedures for protective relay
testing using power system simulators and transient simulation software.
César A. Ventura Santana received a BSEE degree and an M.Sc. degree in
Electrical Engineering from the Guadalajara Autonomous University in 1981
and 1997, respectively. He joined Comisión Federal de Electricidad (CFE) in
1983. Mr. Ventura served as a construction supervisor in the Nayarit
Transmission Zone until 1984. From 1984 until 1997, he was a protection,
control, and metering engineer at the Jalisco Transmission Zone of the CFE
Western Transmission Region. From 1997 until 1998, Mr. Ventura was in the
extra-high-voltage and high-voltage network studies department of the
protection and metering division. Since 1998, he has been head of the
metering department of the protection and metering division. Since 2006,
Mr. Ventura has worked as head of the Western Metrology Laboratory of the
CFE Western Transmission Region. He leads projects on power quality
monitoring, measurement and instrumentation, and metering connectivity in
the CFE Western Transmission Region.
David A. Costello graduated from Texas A&M University in 1991 with a
BSEE. He worked as a system protection engineer at Central Power and Light
and Central and Southwest Services in Texas and Oklahoma. He has served
on the System Protection Task Force for ERCOT. In 1996, Mr. Costello
joined Schweitzer Engineering Laboratories, Inc., where he has served as a
field application engineer and regional service manager. He presently holds
the title of senior application engineer and works in Boerne, Texas. He is a
senior member of IEEE and a member of the planning committee for the
Conference for Protective Relay Engineers at Texas A&M University.
Demetrios A. Tziouvaras received his BSEE from the University of New
Mexico and MSEE from Santa Clara University. He is an IEEE senior
member and a member of the Power System Relaying Committee (PSRC) and
CIGRÉ. Mr. Tziouvaras previously worked at Pacific Gas and Electric
Company, where he held various protection engineering positions, including
principal protection engineer for 18 years. In 1998, he joined Schweitzer
Engineering Laboratories, Inc., where he currently holds the position of senior
research engineer. Mr. Tziouvaras holds four patents and has authored and
coauthored more than 50 technical papers. He served as the convener of
CIGRÉ working group B5.15 on “Modern Distance Protection Functions and
Applications” and is a member of several IEEE PSRC and CIGRÉ working
groups.
David Sánchez Escobedo received his BSEE degree in 1994 from the
University of Guanajuato, Mexico, and his M.Sc. degree in 2005 from the
University of Guadalajara, Mexico. From 1994 until 1998, he was head of the
Protection and Metering Office in the Western Transmission Area of
Comisión Federal de Electricidad (CFE) in Guadalajara, Jalisco, Mexico.
Mr. Sánchez served on the faculty of the Autonomous University of
Guadalajara in 1998. From 1998 until 2000, he worked for INELAP-PQE in
Guadalajara, Mexico, as a protection system design engineer. In 2000,
Mr. Sánchez joined Schweitzer Engineering Laboratories, Inc., where he is
currently the electrical engineering manager in San Luis Potosí, Mexico. He
has authored and coauthored several technical papers.

Héctor J. Altuve Ferrer received his BSEE in 1969 from the Central
University of Las Villas, Santa Clara, Cuba, and his Ph.D. in 1981 from Kiev
Polytechnic Institute, Kiev, Ukraine. From 1969 until 1993, he served on the
faculty of the Electrical Engineering School at the Central University of Las
Villas. He served as a professor of the Graduate Doctoral Program in the
Mechanical and Electrical Engineering School at the Autonomous University
of Nuevo León, Monterrey, Mexico, from 1993 to 2000. From 1999 to 2000,
he was the Schweitzer Visiting Professor at Washington State University’s
Department of Electrical Engineering. In January 2001, Dr. Altuve joined
Schweitzer Engineering Laboratories, Inc., where he is currently a
distinguished engineer and director of technology for Latin America. He has
authored and coauthored several books and more than 100 technical papers
and holds four patents. His main research interests are in power system
protection, control, and monitoring. Dr. Altuve is an IEEE senior member.

Ronald A. Schwartz earned a BSEE from Ohio State University in 1968 and
an M.Sc. in Electrical Engineering from the University of Maryland in 1970.
He has served in the Oregon Quality Award Program as Senior Examiner. In
addition, Mr. Schwartz joined Schweitzer Engineering Laboratories, Inc. in
1998 and has served on the board of directors since February 1994. He is
currently a senior vice president for quality. Mr. Schwartz founded and served
as principal for International Quality Associates, Inc. of Beaverton, Oregon, a
consulting and training firm helping companies develop and implement
effective management systems. Prior to founding International Quality
Associates, he was employed for eight years by Sequent Computer Systems,
also of Beaverton, as component engineering manager as well as a reliability
engineer.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 179

EXPERIENCIAS OPERATIVAS Y CRITERIOS PARA DETERMINAR FALLAS DE

ALTA IMPEDANCIA EN LINEAS DE TRANSMISIÓN DE ENERGÍA A ALTO VOLTAJE

Sadul S. Urbaez Gonzalez Jhon Albeiro Calderon Serna Luis Everley Llano Zuleta
Especialista Operación Red Especialista Operación Red Director Gestión de la Operación
surbaez@isa.com.co jacalderon@isa.com.co lellano@isa.com.co

DIRECCIÓN GESTIÓN DE LA OPERACIÓN

Gerencia Transporte de Energía
Interconexión Eléctrica S.A. E.S.P

RESUMEN. Las dificultades anteriores se tornan críticas

En el análisis de perturbaciones de una red
de transmisión se tienen múltiples tipos de
fallas, algunas de las cuales son de alta
impedancia (FAI) y pueden no ser
detectadas por las funciones de distancia de
las protecciones principales del circuito y,
por lo tanto, su tiempo de aclaración por
funciones de respaldo (sobrecorriente
direccional de tierra temporizada) puede
comprometer la seguridad del sistema ante
el disparo no selectivo de circuitos
adyacentes, debido a la dificultad de lograr
una adecuada coordinación de operación de
protecciones en este tipo de fallas.
Si bien la ocurrencia de este tipo de fallas es
de baja probabilidad, éstas revisten un alto
impacto para el sistema de potencia, toda
vez que, normalmente llevan a la
indisponibilidad permanente de la línea con
la consecuente necesidad de intervención.
Para este tipo de fallas, se dificulta también
su localización, debido a que normalmente,
se presenta una leve variación tanto de la
magnitud del voltaje como de la diferencia
angular entre el voltaje y la corriente vista
desde las bahías asociadas con el elemento
fallado, lo cual causa una alta incertidumbre
a los algoritmos convencionales para estimar
una adecuada localización. En efecto, en
determinados tipos de fallas de alta
impedancia (donde previamente hay claridad
que la falla es interna) el valor de
localización calculado mediante algoritmos
convencionales arroja resultados
inaceptables como por ejemplo que la falla
se localiza a una distancia muy superior al
100% de la línea.
para la operación teniendo en cuenta que
por la naturaleza de este tipo de fallas se
aumenta la probabilidad de que sea
recurrente y por tanto se debe normalmente
actuar de manera rápida en su búsqueda
debido al alto impacto que ello implica para
la seguridad del sistema eléctrico de
potencia. La incertidumbre en la
localización, se torna entonces una variable
fundamental para esta labor de búsqueda de
la ubicación de la falla y ello representa en
diversas ocasiones un esfuerzo logístico
importante con las implicaciones económicas
asociadas.
El artículo pretende dejar plasmado una
serie de criterios tanto automáticos como
prácticos para los análisis de fallas en líneas
de transmisión, con el fin de poder
determinar a partir de la información
contenida en los registros de fallas reales la
causa de la falla y tener algunos elementos
que ayuden a poder ubicar el sitio de la falla.
Se destaca en este análisis el contenido
teórico de una investigación que se ha
venido adelantando en ISA en los últimos
años [2] a [11], algunos lineamientos
prácticos que se han derivado de la
experiencia operativa y del análisis de
registros de ISA de fallas de alta impedancia
[12] a [14] y una tesis de maestría de la
Universidad Nacional sobre localización de
fallas utilizando técnicas de Inteligencia
Artificial [15]. Todos los anteriores
conceptos convergen en una propuesta de
desarrollo informático integrado [16] que
pretende suministrar a los analistas de
Operación y Mantenimiento, la información
tanto de la causa como la localización de la
falla con alto valor agregado para la toma de
decisiones en tiempo real.
180
PALABRAS CLAVES.
Análisis de Perturbaciones, Alta Impedancia
Registros de fallas en líneas de transmisión,
Descubrimiento de Conocimiento desde
Datos (Knowledge Discovery from Data:
KDD), Minería de Datos, Inteligencia
Artificial, Algoritmos de Localización de
Fallas, Experiencias Operativas.
INTRODUCCIÓN.
En el análisis de perturbaciones de una red
de transmisión se tienen múltiples tipos de
fallas entre las cuales se tienen las fallas
francas, monofásicas, bifásicas y trifásicas
así como unas fallas de muy poca
probabilidad pero de alto impacto para el
sistema de potencia, las cuales son las fallas
de alta impedancia.
Un reto que normalmente se debe afrontar
en la operación en tiempo real es determinar
la causa de la falla y su ubicación con el fin
de tomar acciones inmediatas para
restablecer el servicio y prevenir una
recurrencia. Particularmente, las fallas de
alta impedancia revisten una especial
dificultad en lo relacionado con la
localización acertada de la ubicación de la
falla por lo cual se hace cada vez mas
indispensable disponer de técnicas que
ayuden en su identificación efectiva y
oportuna.
Durante muchos años se ha tenido un gran
dilema en poder determinar si una falla en
una línea de transmisión obedeció a un
problema de alta vegetación. Inclusive se
vuelve un reto para los analistas de
perturbaciones y el personal de líneas, dado
que las fallas de alta impedancia por sus
características, generalmente no son
detectadas por las funciones de distancia de
los relés de protección. Los datos de
localización para estas fallas por ser de alta
impedancia no tiene un nivel de certeza alto
y por el contrario los datos de localización
entregados, por ejemplo, por los algoritmos
de localización de los relés de protección, a
veces son superiores a la misma longitud de
la líneas donde se presentó dicha falla.
Este artículo tiene como finalidad dejar unos
criterios ó pasos que se deben tener en
cuenta en los análisis de registros de fallas
para poder determinar si una falla se puede
considerar de comportamiento de alta
impedancia y de esta manera tratar de
obtener un dato posible de la localización del
sitio de la falla.
Inicialmente, se presentará el marco teórico
y conceptual que se ha venido trabajando en
los últimos años comenzando desde los
trabajos previos de investigación y
desarrollo [2] a [8], las metodologías para el
procesamiento de los registros de fallas
(extracción de características principales)
[9], Metodologías KDD (Knowledge
Discovery from Data) y Minería de Datos
(para la identificación de la causa de las
fallas) [10] y [11], algunos criterios
prácticos para localización de fallas de alta
impedancia a partir de registros de
oscilografía [12] a [14], Algoritmos para
localización de fallas basados en técnicas de
Inteligencia Artificial [15] y finalmente el
Análisis, Diseño, Construcción y Validación
de un Sistema Informático Integrado para la
localización de fallas en sistemas eléctricos
de transmisión de potencia [16] con una
propuesta de integración al Sistema de
Información de Descargas (SID), Sistema de
Información Geográfico (SIG), Bases de
datos de tablas de torres con despliegue de
la configuración de conductores con
disposición física de las fases, alturas de
conductores, resistencia de puesta a tierra,
resistividad del terreno y la posibilidad
futura de interacción con un sistema de
fotografía digital (tipo Google earth) con
información precisa de imágenes del sitio de
falla.
PROYECTOS DE INVESTIGACIÓN Y DESARROLLO.
Desde hace aproximadamente 15 años, en
ISA se ha venido trabajando en el desarrollo
de competencias para el análisis y el
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
diagnóstico de las fallas que se presentan en
el Sistema de Transmisión Colombiano.
Dentro de todo este contexto se han logrado
avances importantes en algunos proyectos
de investigación y desarrollo los cuales se
mencionan a continuación:
“Sistema de Información de la Gestión Operativa
(SIGO)” [2].
Esta es una aplicación con la que se ha
logrado elaborar una base de datos con
información especializada asociada con el
diagnóstico de fallas de todas las
perturbaciones que se presentan en la red
de ISA. Es decir, en la actualidad hay una
minería de datos importante del sistema de
transmisión Colombiano que se encuentra
disponible para ser utilizada en diversos
proyectos de investigación.
“Formulación plan piloto para diagnóstico
automático de eventos” [3].
ISA presentó en Julio de 2004 un trabajo en
el cual se obtuvo la formulación para el
desarrollo de un sistema automático de
análisis de eventos a nivel de subestación
para la subestación Ancón Sur de ISA,
basado en información de señales digitales
de la RTU (Unidad Terminal Remota) y de
los relés de protección SEL-321. En este
trabajo se realizó la modelación del esquema
de protección y se definieron reglas de
diagnóstico para el análisis de los eventos
mediante Redes de Petri Coloreadas. Este
proyecto se desarrolló también con la
participación de la Universidad Nacional de
Colombia, sede Medellín.
“Herramientas informáticas para el diagnóstico
automático de eventos en líneas de transmisión
de energía eléctrica” [4].
Este proyecto se formuló ante COLCIENCIAS
por parte de Interconexión Eléctrica S.A.
E.S.P. como entidad beneficiaria y ejecutora
y la Universidad Nacional de Colombia sede
Medellín como entidad ejecutora. El
proyecto finalizó con éxito en el año 2008.
181
A partir del proyecto anterior se generó un
prototipo que fue aplicado por ISA para el
desarrollo del programa DAE (Diagnóstico
Automático de Eventos) y el Diagnóstico
Distribuído de Eventos (DDE).
Programa para Diagnóstico Automático de
Eventos (DAE) [5].
El programa DAE actualmente se encuentra
en el proceso final de pruebas para su
puesta en servicio en el CSM y en el análisis
postoperativo en la Dirección Gestión de la
Operación de ISA.
“Diagnóstico Distribuído de Eventos (DDE)” [6].
El DDE es un proyecto de investigación que
se viene adelantando entre ISA, Colciencias
y la Universidad Nacional. Este proyecto
pretende automatizar el proceso de análisis
de eventos mediante el procesamiento
distribuido y en tiempo real de las diversas
fuentes de información disponible en la red
de ISA (SOE, registradores de fallas,
registros de relés, Sistemas de
Automatización de Subestaciones (SAS)
entre otras).
Modelo Adaptativo de Inteligencia Artificial
(MAIA) [7].
Como parte del proyecto de investigación [4]
se definieron los lineamientos generales para
procesar los registros de oscilografía con el
fin de determinar los tópicos más relevantes
para un diagnóstico automático de eventos.
Para ello se Analizó, Diseñó, Construyó y
Validó un Modelo Adaptativo de Inteligencia
Artificial (MAIA) [7] el cual es un modelo
inspirado en las capacidades de hardware
(conocimiento conexionista de bajo nivel:
redes neuronales) y software del cerebro
humano (conocimiento intuitivo de alto
nivel: intuición y experticia).
En la Figura 1 se ilustra un diagrama
conceptual del procedimiento para
determinar los tópicos fundamentales del
diagnóstico automático de fallas.
Inicialmente los registros de
osciloperturbografía ingresan a un módulo
182

de preparación de datos (Interfaz entre el

mundo exterior y el modelo neuronal) que
realiza las siguiente funciones: lee la
información de los registros en el formato
estandarizado IEEE COMTRADE (Common
Format For Transient Data Exchange),
realiza el filtrado analógico y digital de las
señales de tensiones y corrientes y
preprocesa (normaliza) la información con el
fin de facilitar el entrenamiento de los
algoritmos de Inteligencia Artificial.

Posteriormente, la información ingresa al Figura 1: Mapa conceptual inspirado en el cerebro

Modelo Adaptativo de Inteligencia Artificial
(MAIA) que se encarga, mediante la
utilización de técnicas de inteligencia
artificial, de determinar los tópicos
fundamentales del diagnóstico: detección de
la falla, direccionalidad de la falla, la
clasificación de la falla, identificación del
componente fallado, la localización de la falla
y la identificación de la causa de la falla.
Este modelo MAIA es entrenado mediante
una Metodología de Entrenamiento y
Validación Adaptativa (MEVA) descrita en [7]
la cual permite mediante técnicas de
Aprendizaje de Máquina lograr
entrenamiento en línea y de esta forma
aprender de la experiencia operativa y
adaptarse dinámicamente a los cambios que
se van presentando en el sistema eléctrico
de potencia. El alcance del trabajo de
investigación [7] cubrió los tópicos de:
detección de la falla, la direccionalidad de la
falla y la clasificación de la falla. La
identificación del tramo fallado se abordó en
[5] y en [8]. Finalmente, la identificación de
la causa de la falla y la localización de la
falla se abordaron en [9] a [11].
humano: Percepción del mundo, conocimiento
conexionista (sistema neuronal), conocimiento
intuitivo (Experticia humana) y Aprendizaje de
Máquina (conocimiento Adaptativo).
PROCESAMIENTO DE REGISTROS DE FALLA [9].
Con base en los lineamientos definidos en
[7], en el año 2007 se desarrolló e
implementó una metodología para el
procesamiento de los registros de falla
ocurridos en la red de ISA entre los años
2001 al 2006 la cual se encuentra
documentada en [9]. Con base en la
experiencia operativa de ISA en el análisis
de perturbaciones se definió un vector
(patrón de entrenamiento) con las
características principales que se pueden
extraer de un registro de oscilografía en
formato COMTRADE (Standard Common
Format for Transient Data Exchange). En la
Figura 2 se ilustra el vector característico
considerado.

Finalmente, el modelo MAIA se integra en un

Sistema Basado en Conocimiento describo Figura 2: Vector característico utilizado.
en [7] para determinar automáticamente el
diagnóstico de fallas. Las magnitudes a extraer del registro de
fallas mostradas en esta Figura
corresponden con:

o Tensiones RMS promedio.

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 183

o Corrientes RMS promedio.

o Ángulos de tensiones y corrientes.

o Deltas o variaciones de tensión y

corriente de acuerdo con dos criterios:
cantidades super impuestas (diferencia
entre el valor actual y el valor 1 ciclo
anterior) y comparación de promedios
rms (Vfalla/Vss). Siendo “ss” un sufijo
que indica valor de estado estable.

o Tensión y corriente residual. Figura 4: Resultado del procesamiento del registro

para la ocurrencia SIGO 2052005.
o Ángulos de la tensión y corriente
residual.

En la Figura 3 se ilustran los punteros inicial Luego de efectuado un procedimiento

y final entre los que se realizó el
procesamiento digital de señal para la
extracción del vector característico.
riguroso de depuración de los registros de
falla mediante un software implementado en
MATLAB [9] se obtuvieron un total de 3556
patrones de entrenamiento útiles de eventos
reales de ISA con las características
descritas en la Figura 2 y resultados
similares a los mostrados en la Figura 4.

Una vez obtenidos estos patrones

Figura 3: Rango de extracción del vector característico.
característicos, se procedió a realizar un
agrupamiento de la información de acuerdo
con la causa de falla establecida en el SIGO
por los analistas de perturbaciones de ISA.
Con base en este criterio se determinaron
las 6 causas de falla más representativas tal
como se describe en la Figura 5.

En la Figura 4 se ilustra el resultado del

procesamiento del registro del evento del
programa SIGO con consecutivo de
ocuurencia 2052005 para el caso de una
falla a tierra en la fase B en el circuito
Guatiguará-Primavera a 230 kV.
184

Figura 5: Diagrama circular de las causas de fallas

agrupadas para los 3556 patrones procesados.

Finalmente, se realizó un análisis estadístico

de la información anterior calculando la
media y la desviación estándar de las
diferentes variables características
encontrándose que los ángulos de las
tensiones y corrientes individualmente, eran Figura 6: Resultados del análisis estadístico para las
datos poco significativos con relación a la fallas de alta impedancia.
identificación de la causa de la falla por lo
que se decidió eliminar estas variables del Para las otras cinco causas de falla
vector característico e incluir en su lugar una analizadas se realizaron análisis estadísticos
variable correspondiente con la diferencia similares conformándose de esta manera
entre los ángulos de la tensión y la corriente una matriz de 3556 registros estructurados
de cada fase, la cual condujo a resultados con información de valor agregado para el
más representativos en lo relacionado con la entrenamiento de un sistema inteligente
identificación de la causa de la falla. En la para la identificación de la causa de la falla a
Figura 6 se ilustran los resultados del partir de registros de oscilografía. En la
análisis estadístico para las características Figura 7 se resume todo el contexto utilizado
encontradas para las fallas de alta en el procedimiento anterior.
impedancia. Los resultados obtenidos, como
era de esperarse, muestran una excelente
coherencia con los criterios prácticos que se
han detectado de la experiencia operativa
los cuales serán descritos con más detalle en
el numeral CRITERIOS Y LINEAMIENTOS
PRACTICOS de este artículo. Nótese que la
media en la relación Vf/Vss es de 0.989, es
decir, la magnitud del voltaje varía muy
poco (menos del 2%) y la media de la
diferencia angular entre el Vf (Voltaje de
falla) y la If (Corriente de falla) es del orden
de los 38 grados.

Figura 7: Contexto utilizado para el procesamiento de

los registros.

SISTEMA PARA LA IDENTIFICACIÓN DE LA CAUSA

DE LA FALLA (SICAF) [10] y [11].
Una vez se desarrolló la fase del
procesamiento de los registros reales de los
eventos de ISA y se extractaron las
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
características principales orientadas a la
identificación de la causa de la falla el
siguiente paso consistió en aplicar
metodologías de Descubrimiento de
Conocimiento desde Datos (KDD: Knowledge
Discovery from Data) y Minería de Datos
[10] y [11]. En el proceso KDD se realizó la
limpieza, reducción, y transformación, de
variables y registros de falla. En la Minería
de Datos, se clasificó un conjunto de
registros de falla, a través de ciertas
características eléctricas. A partir del
análisis anterior se determinó el modelo más
apropiado a utilizar para la identificación de
la causa de falla a partir de registros
oscilográficos (algoritmo K*) y se aplicó a un
prototipo informático en lenguaje java
denominado “SICAF” (Sistema de
Identificación de la Causa de la Falla).
En la Figura 8 se resumen los grupos o
clusters que se pueden clasificar a partir de
los 3556 patrones determinados en el
trabajo previo descrito en [9].
Figura 8: Clusters que se pueden formar a partir de los
3556 patrones analizados.
Partiendo del conjunto de datos ilustrado en
la Figura 8 se llevó a cabo el siguiente
procedimiento de KDD y Minería de Datos
cuyo detalle se encuentra descrito en [10] y
en [11]:
185
1. Preprocesamiento de datos: se extrajeron
las variables características más relevantes
de acuerdo a las sugerencias recogidas de
las experiencias del analista de
perturbaciones. Esas variables son
integradas a un formato de registro
transformado de menor magnitud que
representará cada uno de los registros de
falla.
2. Exploración de datos: se realizó un
examen gráfico de la naturaleza de las
variables características individuales por
cada tipo de causa de falla y un análisis
descriptivo numérico que permita cuantificar
algunos aspectos gráficos de los datos.
3. Análisis outlier: Se identificaron y
eliminaron los posibles casos atípicos
(outliers) de acuerdo a la técnica estadística
del recorrido intercuartílico.
4. Reducción de atributos: Se identificaron
las variables características más
representativas y suficientes para predecir
las causas de falla, a partir de un registro
COMTRADE.
5. Análisis cluster: Se identificaron posibles
similitudes entre grupos de registros. Estas
técnicas permitieron observar solapamientos
y organización de los datos gráficamente.
6. Clasificación y predicción: Se
seleccionaron los algoritmos más apropiados
y se obtuvieron los resultados más
representativos de clasificación de datos
orientados a la identificación de la causa de
falla mediante los registros de falla.
7. Combinación de algoritmos: Se evaluaron
los resultados de mezclar los mejores
algoritmos identificados en el punto anterior,
para así dar la posibilidad de mejorar la
exactitud de predicción.
8. Variación de parámetros: Se efectuó un
análisis de sensibilidad de los resultados
186

obtenidos de las variaciones realizadas a los Etapa de Conocimiento. Para el registro

parámetros del algoritmo de clasificación particular analizado el sistema determina el
seleccionado de acuerdo a los resultados y vector característico y luego aplicando el
conclusiones de la actividad anterior. algoritmo K* identifica que se trata de un
atentado a torre definitiva.

Figura 11: GUI para el programa de aplicación SICAF

(Etapa de Conocimiento).

CRITERIOS Y LINEAMIENTOS PRACTICOS.

Figura 10: Procedimiento KDD utilizado en el análisis

de identificación de la causa de la falla. Con el fin de entender la dificultad en
determinar el tipo de falla se mostrará una
En la Figura 10 se resume el procedimiento serie de registros reales donde se ha podido
KDD utilizado y las diversas etapas que se comprobar que la causa de la falla fue una
abordaron en este análisis: Etapa de Datos alta vegetación.
(numerales 1 a 5), Etapa de información
(numerales 6 y 7: Minería de Datos y En la Figura 11 se muestra un
numeral 8 Medición de la exactitud de comportamiento típico de una falla de alta
algoritmos, con la cual se determinó que el impedancia en la fase C en la línea Guavio –
algoritmo adecuado para la identificación de Torca 2. El comportamiento de una falla de
la falla es el K* [10] y [11]) y Etapa de muy alta impedancia y cerca de
Conocimiento que incorpora la información subestaciones de fuente fuerte no produce
extraída de la etapa de información, a un una mayor caída de tensión y la corriente de
sistema con una GUI (Graphical User falla normalmente no llega a valores
Interface), de modo, que esa información superiores a la capacidad de transmisión del
genere valor a futuras acciones. A partir del circuito. Si el circuito no tiene
análisis anterior se construyó la aplicación implementado el 67N en comparación
SICAF para la identificación de la causa de direccional la falla puede despejarse en
fallas a partir de registros COMTRADE. tiempos del orden de segundos. En la Figura
12 se muestra una falla de alta impedancia
En la Figura 11 se ilustra la interfaz gráfica en la fase A originada por una quema cerca
del programa SICAF implementada para la
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
de la línea causando acercamiento con
vegetación en la línea Los Palos – Ocaña.
Figura 11. Comportamiento típico de una falla de alta
impedancia en la línea LT Guavio – Torca 2 el 16 de
septiembre del 2003 a las 21:36 horas
Figura 12. Comportamiento de una falla de alta
impedancia en la línea Los Palos – Ocaña el día 2 de
abril de 2007 a las 17:22 horas.
En la Figura 12 se puede ver que la falla
presentó una caída de tensión del 86% y
una corriente de falla de 6.5 kA rms
originado por una falla de alta impedancia
muy cerca a la subestación Los Palos.
En la Figura 13 se puede apreciar una falla
en la fase C que se auto extinguió al ser de
muy alta impedancia y no alcanzo a operar
la función de 67N dado que la corriente de
falla se extinguió antes que se cumpliera el
tiempo de la función 67N.
187
Figura 13. Falla de alta impedancia en el circuito Torca
– Guavio 2 del 21 de septiembre de 2004 a las 05:27
horas.
Estos tres ejemplos muestran que
determinar si una falla fue de alta
impedancia no es una tarea fácil para los
analistas de fallas, dado que no basta con
evaluar magnitudes de corriente y voltaje
para llegar a conclusiones respecto a su
característica.
CRITERIOS PARA DETERMINAR QUE UNA FALLA
ES DE ALTA IMPEDANCIA
Criterio de ángulo de falla
En una línea de transmisión de alta tensión (mayor
a 230 kV) la impedancia de la línea esta
conformada prácticamente por la componente de
inductancia de la línea, pues la resistencia es
despreciable. Con este concepto se tiene entonces
que el ángulo de la línea está comprendido entre:
83° < θ < 90°
Por lo tanto, si se presenta una falla en algún
punto de la línea y esta es de baja impedancia (RF
→ 0) implica entonces que el ángulo entre el
voltaje y la corriente durante la falla, será
prácticamente el ángulo de la línea.
Ahora si la falla que se presenta en la línea es de
alta impedancia o sea una RF muy grande y
188
comparable con la inductancia de la línea entonces
el ángulo entre el voltaje y la corriente de falla
estará relativamente cerca a cero:
0° < θ <=45°
Con este criterio podemos analizar las tres
fallas mostradas inicialmente desde el punto
de vista del ángulo de falla. En las figuras
14, 15 y 16 se muestra la variación del
ángulo de falla para cada una de las fallas
mostradas en las figuras 11, 12 y 13.
Figura 14. Angulo entre el voltaje y corriente de falla
en la línea LT Torca – Guavio 2 del evento del 16 de
septiembre de 2003 a las 21:36 horas.
Figura 15. Angulo entre el voltaje y corriente de falla
en la línea Los Palos – Ocaña del evento 02 de abril del
2007 a las 17:22 horas.
Figura 16. Angulo entre el voltaje y corriente de falla
en la línea Torca – Guavio 2 del evento del 21 de
septiembre de 2004 a las 05:27 horas.
Las tres fallas tienen en común en que el
ángulo entre voltaje y la corriente de falla es
menor a los 45 grados, lo que significa de
acuerdo al criterio que es una falla de alta
impedancia.
Con este criterio hay que tener cuidado
dependiendo del circuito y punto donde se
presentó la falla y sobre todo de cual
subestación se tiene el registro de falla
donde se mide el ángulo donde se podría
llegar a concluir que una falla de alta
impedancia sea vista como una falla de baja
impedancia.
En la figura 17 se ilustra la falla en la línea
Los Palos – Ocaña del 2 de abril de 2007 a
las 17:22 horas con el registro visto desde la
subestación Ocaña.
Figura 17. Angulo entre el voltaje y corriente de falla
en la línea Los Palos – Ocaña del evento 02 de abril del
2007 a las 17:22 horas.
Se puede observar que el ángulo de falla
esta del orden de los 74 grados, la falla es
despejada en menos de 60ms y los deltas
de corriente y tensión son altos e indicaría a
simple vista que la falla fue de
relativamente baja impedancia. Sin
embargo, si se hace un análisis del registro
de falla en Los Palos simultáneo con el
registro capturado en Ocaña, se puede ver
que la falla fue más cerca de la subestación
Los palos, debido a los aportes de corriente
y el delta en la caída de tensión. Si se le
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 189

agrega inductancia de la línea para una de alta impedancia, pero no es información

longitud del circuito que son 160 km, se suficiente para concluir que la falla
puede deducir que la falla vista desde Ocaña presentada sea de alta impedancia.
con ángulo superior a los 45 grados también
corresponde a una falla de alta impedancia.
Esto debido a que el valor de la inductancia CONOCIMIENTO DE CARACTERISTICAS DE
de la línea sigue siendo grande con relación LONGITUD Y CONFIGURACIÓN DEL CIRCUITO
a la resistencia de falla para el registro visto
desde Ocaña, lo que esta causando que el En un análisis de comportamiento de una
ángulo sea grande. falla es importante conocer la longitud del
circuito. Eso nos determina que tanta
Por el contrario, en el registro de fallas visto
inductancia se tiene para entrar a evaluar el
desde la subestación Los Palos donde la falla
valor de la resistencia de falla visto con el
es muy cerca de dicha subestación se
comportamiento del ángulo de falla. Así
obtiene un ángulo del orden de los 20
mismo se debe saber la configuración del
grados, lo que indica que la resistencia de
circuito con relación a la ubicación física de
falla es de un valor muy superior a la
las fases para ver si es posible que la alta
inductancia de la línea al punto de falla. Si la
vegetación sí sea la causa posible de la falla.
falla es de baja impedancia simplemente la
Por lo tanto, una línea muy corta con falla de
inductancia al punto de falla seria muy alto
alta impedancia tendrá un comportamiento
en relación a la resistencia de la línea al
muy diferente a una línea muy larga con
punto de falla, lo que equivale a un ángulo
falla de alta impedancia.
de falla muy alto.

En resumen con el criterio de ángulo de falla

con registro de un solo extremo de la línea TOPOLOGÍA DEL TERRENO POR DONDE PASA EL
no habría información suficiente para CRICUITO DE TRANSMISIÓN.
determinar si la falla fue o no de alta
impedancia.
Un aporte muy grande para el análisis del
Criterio por cambio ó delta de voltaje. comportamiento de una falla y determinar
el sitio de falla, es considerar la topografía
del terreno por donde fue trazada la línea,
Si se observa las Figuras 11 y 13 podemos
pues de acuerdo cómo es la zona, por
ver que la caída de tensión durante el
ejemplo, alta vegetación a lo largo de
tiempo de la falla no es grande, lo que nos
servidumbre, o zona de vegetación baja, o
hace deducir que la falla debe ser de alta
zona de desierto, o zona costera, son
impedancia, sin embargo, si se observa la
ingredientes que ayudan a ubicar el sitio
Figura 12 donde la caída de voltaje es
posible de la falla de alta impedancia. Esa
considerable, ya el concepto de que las fallas
información es necesaria para poder
de alta impedancia producen poca caída de
correlacionar los datos de localización que se
voltaje ya no sería aplicable para decir que
están calculando con las diferentes
la falla ilustrada en la Figura 12 es de alta
herramientas que entregan un dato
impedancia.
aproximado de la localización.
En resumen, el criterio de cambio de voltaje
pequeño en la fase fallada es típico de falla
190

LOCALIZACIÓN DE FALLAS EN SISTEMAS DE subestación, el error es grande con un

TRANSMISIÓN MEDIANTE TÉCNICAS DE crecimiento exponencial. Del estudio
INTELIGENCIA ARTIFICIAL [15].
realizado en [1], se concluyó que el más
preciso de los algoritmos evaluados fue el
En la mayoría de los esquemas de Ericsson.
localización de falla suelen ser usados
Una característica importante asociada a las
algoritmos para intentar encontrar el punto
fallas de alta impedancia es su no linealidad,
donde ha ocurrido una falla en líneas de
ó sea una corriente de falla producida puede
transmisión [1], usando la información
no ser puramente sinusoidal, siendo una
almacenada mediante mediciones durante la
composición de diferentes frecuencias
falla (registros de fallas). Los métodos más
armónicas a la fundamental. Una no
comunes usan la información de tensiones y
linealidad está generalmente asociada a un
corrientes de un extremo de la línea y,
arco eléctrico que puede estar asociado a su
considerando que podría haber una
vez al punto de contacto del conductor con
resistencia de falla, intentan evaluar la
la tierra u otro objeto durante la ocurrencia
distancia a la cual ocurrió la falla. Sin
de una falla de alta impedancia. Debido a
embargo en los últimos años se han venido
sus características, la presencia de fallas de
desarrollando modelos de localización del
alta impedancia puede no ser detectada en
punto de la falla con la ayuda de inteligencia
las señales de tensión y corriente utilizado
artificial [15], lo que difiere de los algoritmos
por los dispositivos de protección y
clásicos que se basan en cálculos
localización (por ejemplo, una variación de
determinísticos y donde es necesario un
amplitud de la corriente generada por la falla
modelo matemáticamente bien definido.
de alta impedancia, puede ser confundida
Los algoritmos para la localización de fallas con un crecimiento de carga). Como
de mayor aplicación como: Reactancia, resultado los sistemas de protección y
Takagi, Schweitzer y Ericksson, hacen uso localización pueden no actuar correctamente
de la información de las variables que en la localización y eliminación de este tipo
almacenan los relés de protección. Estos de falla, resultando en riesgos para la
algoritmos presentan gran certeza en la población y daños en materiales,
localización de fallas de baja impedancia y a consumidores y en el sistema.
corta distancia del relé, sin embargo en
Teniendo en cuenta la revisión del estado del
éstos se realizan suposiciones sobre el
arte en [15] y los análisis previamente
comportamiento y estructura del sistema de
realizados indican que los métodos
potencia, limitando su alcance real. Además
tradicionales con los que actualmente se
para fallas de alta impedancia presenta en
está abordando el problema de localización
sus pronósticos graves deficiencias [1]. En
no son los más adecuados debido a los
[1] se analizaron los cuatro algoritmos
grandes errores en los que incurren. Lo
localizadores de fallas mencionados, en una
anterior motiva la búsqueda de nuevas
línea de transmisión, aplicados en el sistema
metodologías para abordar el problema de la
de transmisión colombiano y se determinó
localización como lo son las de técnicas de
cual es más preciso y eficiente por medio de
inteligencia artificial.
simulaciones en EMTP/ATP. Estos
algoritmos son muy precisos ante fallas En [15] luego de realizada una labor de
francas, sin embargo para fallas de alta búsqueda del estado del arte y con base en
impedancia y sobre todo lejos de la los patrones de falla obtenidos desde
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
simulaciones con el programa ATP se
proponen dos técnicas de inteligencia
artificial para la localización de fallas:
Método de Red Neuronal Artificial y el
Método de análisis adimensional combinado
con un sistema ANFIS (Sistema de
Inferencia Neuro-Difuso Adaptativo).
Para el método de Red Neuronal la investigación
de [15] encontró que para la localización de las
fallas de alta impedancia resulta apropiado
seleccionar como variables de entrada las
cantidades superimpuestas de la magnitud del 1er
armónico de voltaje y corriente, |ΔVA1φ|, |ΔIA1φ|,
de la magnitud del 3er armónico de corriente
|ΔIA3φ| y del cociente entre la magnitud de los
fasores de voltaje y corriente de 1er armónico
|ΔVA1φ| / |ΔIA1φ|.
Estas variables se obtienen directamente del
programa de simulación ATP mediante la
utilización de un “MODEL”, en el que se realiza un
“script” con la transformada de Fourier para el 1er
y 3er armónico, previa obtención de la variable
superimpuesta, realizando la diferencia de un ciclo
de la señal de interés en prefalla y postfalla.
En la Figura 18, se aprecia el proceso para la
localización de la falla con el método de Red
Neuronal Artificial (RNA).
Figura 1: Proceso de localización de la falla mediante
el uso de RNA.
191
Para el método de análisis adimensional en la
investigación de [15] se obtuvieron curvas con
parámetros adimensionales PI 1 y PI 3 obtenidos
con la exploración de fallas en el programa ATP
para diversas resistencias de falla y localización de
la falla.
El número adimensional PI 3 está relacionado con
la variable de localización de la falla, por tanto es
posible a partir de PI 1 obtener una primera
aproximación al punto falla. Sin embargo la
gráfica presenta una pequeña inclinación en la
primera mitad, por lo que para un valor de PI 1,
existen un amplio rango de valores de PI 3, y por
consiguiente del punto de incidencia de la falla.
Para la segunda mitad de la curva la inclinación
aumenta lo que disminuye la franca de posibles
valores para PI 3, aun así todavía es amplia, siendo
necesario la utilización de técnicas de Inteligencia
Artificial (IA) para lograr disminuir la
incertidumbre asociada con estas franjas.
Figura 19: Números adimensionales PI 3 vs PI 1 grupo
2 para todo el espectro de Resistencia de falla Rf.
Luego de efectuado un análisis de diversas
técnicas de IA en [15] que la técnica que más se
ajusta a la solución del problema anterior es un
sistema ANFIS. Las variables que se seleccionaron
como entrada para el sistema ANFIS fueron las
cantidades superimpuestas de la magnitud de
corriente y el voltaje del primer armónico
192
(|ΔIA1φ| y |ΔVA1φ|) debido a que, después de
hacer un análisis con las combinaciones entre
todas las variables que componen el sistema, se
encontró que estas dos tienen un buen
comportamiento, es decir, que puede ser
fácilmente ajustado y los datos no contienen
mucho ruido, además la relación entre las dos
variables es bastante lineal, y es posible sólo con
estas dos variables determinar el valor de la
resistencia de falla, que es la variable de salida.
La metodología con el sistema difuso ANFIS,
permite la obtención por inferencia del valor de la
resistencia de falla para los valores de entrada
(|ΔIA1φ| y |ΔVA1φ|) calculados a partir de los
registros de osciloperturbografía. A continuación,
con este valor se selecciona la correspondiente
curva de aproximación de la Figura 19 y con ello se
determina la distancia a la cual se ubica la falla. En
la Figura 20 se describe este proceso.
Figura 20: Números adimensionales PI 3 vs PI 1 grupo
2 para todo el espectro de Resistencia de falla Rf.
Las conclusiones más relevantes de la
investigación realizada en [15] son las siguientes:
La investigación se orientó hacia el
acondicionamiento de los datos y estudio de las
tendencias de las variables eléctricas ante una falla
monofásica a tierra, la falla más comúnmente
presente en los SEP, para presentar una solución al
problema de localización de FAI.
Se pudo concluir que el acondicionamiento de las
señales eléctricas, realizando el proceso de
superimposición, consistente en la diferencia en
un ciclo de la variable en prefalla y postfalla, y
posteriormente la obtención del fasor de los
armónicos 1 y 3, permite describir mejor el
comportamiento de las variables ante falla en el
sistema de transmisión en comparación con los
acondicionamientos tradicionalmente realizados
en estas variables.
El análisis adimensional suministró una práctica
herramienta para el estudio de las tendencias de
las variables, señalando el camino para la
extracción de la máxima cantidad de información
ante una falla en el SEP. De tal manera que en la
Figura 19 de PI 1 vs PI 3, se aprecia una relación
que le permite al analista de perturbaciones tener
cierta idea de la localización del punto de falla.
La utilización de herramientas difusas permitió
muy buena aproximación a la solución del
problema, teniendo en cuenta la relación
presentada en la Figura 19, permite inducir una
solución de este tipo. Al evaluar tres líneas de
transmisión con diferentes ángulos de
transferencia, longitud de las líneas y de
impedancias de los generadores, los mejores
resultados en la predicción de la resistencia de
falla y localización de la falla se obtuvieron al
entrenar el modelo ANFIS con los datos de la línea
3 subestación B, presentando en el caso particular
un error promedio 3.75% en la localización.
Al resolver el problema con redes neuronales se
encontró, que la arquitectura (4,9,1), cuatro
entradas, nueves neuronas ocultas y una salida
presenta mejores resultados que el sistema difuso,
con un error promedio del 0.482% y un error
máximo del 5.08% al evaluar todas las líneas aquí
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 193

planteadas. Las variables de entrada son la estadísticas de densidad de descargas

corriente y voltaje superimpuesto del primer
armónico ó fundamental, el ángulo del tercer
armónico; y el cociente entre el voltaje y corriente
del primer armónico.
Las cuatro variables utilizadas como entrada en la
red neuronal, y la relación presentada en la Figura
19 PI 1 vs PI 3, permite generalizar en cierto grado
la utilización de estas herramientas para líneas de
transmisión a un nivel de tensión 230kV y longitud
mayor de 100km, caso de estudio en esta
investigación.
atmosféricas en la zona, estadísticas de
causas de falla agrupadas a lo largo de
largo de las líneas ó información de
interés que haya sido recolectada por el
personal de mantenimiento de líneas
como por ejemplo la existencia de
árboles, cruces con líneas de
distribución, torres en madera, entre
otras.
o El analista identifica las coordenadas
(longitud y latitud) de las torres
adyacentes a la falla.

o El analista ingresa al Sistema de

Información de Descargas (SID) y
verifica la existencia de descargas
atmosféricas en la zona.

o El analista ingresa al Sistema de

SISTEMA INFORMÁTICO INTEGRADO PARA
LOCALIZACIÓN DE FALLAS (SILF) [16]. Información Geográfica (SIG) y obtiene
una imagen geográfica de la zona donde
fue localizada la falla y trata de
El proceso que se lleva a cabo hoy en día correlacionar esta información con la
para localizar fallas en el Sistema Eléctrico mayor cantidad posible de la información
de Potencia (SEP) de Interconexión Eléctrica suministrada por el personal de
S.A. (ISA) es realizado por el Analista de mantenimiento de líneas.
Análisis Operativo, así:
Teniendo en cuenta lo anterior, se
o El analista carga un archivo en formato vislumbran las bondades de un sistema
COMTRADE, en un software especializado informático integrado que utilice la
que le permite obtener una localización información disponible en la actualidad
aproximada de la falla mediante desde varias fuentes (tablas de torres,
algoritmos convencionales [1]. información de archivos planos, SID, SIG,
relés de protección, SICAF, criterios
o Basado en la información anterior el
prácticos de los expertos en análisis de fallas
analista identifica entre cuales torres se
entre otras) y despliegue de manera ágil y
presentó la falla, ayudado por archivos
oportuna de toda la información de valor
en formato .XLS denominados Tablas de
agregado que le permita tanto al analista del
Torres. Adicionalmente, mediante
CSM como al analista de perturbaciones y de
información contenida en archivos planos
mantenimiento identificar tanto la causa de
puede determinar la disposición física de
la falla como su localización,
los conductores de las torres asociadas
complementada con información de valor
con la localización así como información
agregado del sitio de falla que le permita
sobre resistividad del terreno,
definir con claridad y oportunidad las
resistencias de puesta a tierra,
194
acciones a seguir para minimizar
probabilidad de recurrencia de la falla.
En el trabajo documentado en [16] se
proporciona el análisis y el diseño detallado
de la manera como podría implementarse un
sistema informático integrado para la
localización de fallas (Figura
considerando además la posibilidad de
interactuar con imágenes satelitales que
ilustren de manera precisa las condiciones
reales asociadas con el punto de falla tal
como se ilustra en la Figura 22.
Figura 21: Sistema Informático Integrado para la
localización de fallas.
la
21)
Figura 22: Vista de la interfaz del sistema informático
integrado mostrando una posibilidad de interactuar
con imágenes satelitales tipo Google earth.
CONCLUSIONES Y RECOMENDACIONES
La investigación que se ha venido
adelantando en ISA en los últimos años en
torno de la identificación de la causa de la
falla y la localización de fallas en sistemas de
transmisión ([2] a [11]), indica que
mediante la utilización de técnicas de
inteligencia artificial es posible lograr
resultados altamente satisfactorios en torno
de la búsqueda de la solución a este
problema que sin lugar a dudas representa
un reto para las empresas de electricidad en
el mundo.
Los resultados que se han obtenido de los
análisis realizados hasta la fecha se pueden
resumir en las siguientes conclusiones:
El procesamiento digital de señal propuesto
en [7] utilizado en la extracción del vector
característico orientado a identificación de la
causa de la falla [9] y posteriormente
aplicado a técnicas de KDD y Minería de
Datos [10] y [11] arrojó excelentes
resultados con relación a la clasificación
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
requerida para la identificación de la causa
de la falla. La metodología propuesta, es
adaptativa, es decir permite incorporarle
aprendizaje de máquina de tal manera que
se vaya adaptando a las condiciones
requeridas por cada sistema de potencia en
particular y de esta forma con el paso del
tiempo lograr almacenar un conocimiento
valiosísimo para la memoria corporativa de
la empresa.
Con relación a la localización de la falla la
investigación indica que se obtienen muy
buenos resultados cuando se combina tanto
el conocimiento conexionista almacenado en
forma de redes neuronales como de
conocimiento intuitivo basado en reglas y
experticia humana. En [15] se presenta
una propuesta interesante encaminada en
estos dos aspectos con resultados de
validación muy valiosos en lo relacionado
con la localización de fallas, especialmente
de alta impedancia donde el estado del arte
actual encuentra aún inmensas dificultades
para su solución.
Con relación al conocimiento intuitivo en
torno de la localización de fallas de alta
impedancia, en ISA se ha avanzado en la
formulación de criterios prácticos para el
análisis de los registros asociados
mejorándose con ello la efectividad en la
búsqueda de este tipo de fallas. Así
entonces, a partir del análisis del
comportamiento de un registro de falla
desde el punto de vista de la variación del
voltaje y de la corriente desde una
subestación cualquiera, se ha podido
determinar si la falla es de alta impedancia y
con ello establecer con cierto grado de
certeza un dato posible de localización. Lo
anterior se logra, realizando un análisis
integral considerando diversos aspectos
tales como: ángulo de fallas, deltas de
voltajes y corrientes, conocimiento de la
configuración y longitudes del circuito fallado
y por último de la topografía por donde
cruza dicha línea.
Finalmente, como resultado de todo el
contexto anterior se ha logrado analizar,
diseñar, construir y validar un prototipo para
un sistema informático integrado [16] con el
195
cual se pretende consolidar todos los
resultados obtenidos e integrarlo con las
base de datos del SID, SIG, tabla de torres,
resistividad del terreno, resistencia de
puesta a tierra de las torres, disposición de
conductores, detalles de configuraciones y
transposiciones, cruces con líneas de baja
tensión, información estadística de la causa
de falla discriminada acorde con la
ubicación, estadísticas de densidad de
descargas atmosféricas e incluso la
posibilidad de interactuar con imágenes
satelitales detalladas (estilo google earth)
asociadas con el punto de falla, con lo cual
se brindará un soporte valioso para ayudar a
los operadores del CSM en la toma de
decisiones en tiempo real, a los analistas de
operación para determinar la localización
adecuada del punto de falla y al personal de
mantenimiento para definir con antelación
las acciones correctivas a realizar y definir la
logística requerida acorde el tipo de causa
de falla encontrado.
REFERENCIAS
[1] Idárraga, G.; Valencia, R.; Hernandez, J.
“Algoritmos para Localización de Fallas en
Líneas de Transmisión, Análisis para el Caso
Colombiano”. Dyna, noviembre, año/vol. 71,
número 143 Universidad Nacional de
Colombia Medellín, Colombia pp. 95-100.
[2] Llano, Luis Everley. “Sistema de
Información de la Gestión Operativa
(SIGO)”. Metodología para el Análisis de
Perturbaciones. Interconexión Eléctrica S.A.
Medellín. 2000.
[3] ISA-UNAL. “Formulación plan piloto para
diagnóstico automático de eventos”.
Interconexión Eléctrica S.A. Medellín. 2004.
[4] ISA-COLCIENCIAS-UNAL. “Herramientas
informáticas para el diagnóstico automático
de eventos en líneas de transmisión de
energía eléctrica”. 2005.
[5] Arias, Francisco; Hoyos, Andrés y
Calderón, Jhon Albeiro. “Construcción,
Validación y Evaluación de desempeño del
Programa DAE (Diagnóstico Automático de
Eventos)”. Interconexión Eléctrica S.A.
2009-2010.
196
[6] ISA-COLCIENCIAS-UNAL. “Diagnóstico
Distribuído de Eventos (DDE)”. 2010.
[7] Calderón, Jhon Albeiro. “Modelo
Adaptativo de Inteligencia Artificial para el
Diagnóstico Automático de Fallas a partir de
Registros de Osciloperturbografía”. Tesis de
Maestría en Ingeniería de Sistemas.
Universidad Nacional de Colombia
(Medellín). 2007.
[8] Llano, Luis Everley. “Diagnóstico
automático de eventos en tiempo real en un
sistema de transporte de energía a través
del SOE y SCADA usando técnicas de
inteligencia artificial”. Tesis de Maestría en
Ingeniería de Sistemas. Universidad
Nacional de Colombia (Medellín). 2007.
[9] Ballesteros, David Nereo; Calderón Jhon
Albeiro. “Clasificación y Procesamiento de los
registros de perturbaciones de la red de
transmisión de ISA”. Interconexión Eléctrica
S.A. 2007.
[10] Gómez, Carlos Andrés; Calderón Jhon
Albeiro. “Análisis de Eventos de Líneas de
Transmisión [2001-2006] para la
Identificación Automática de la Causa de
Falla Aplicando Técnicas de Minería de
Datos”. Interconexión Eléctrica S.A. 2007
[11] Gómez, Carlos Andrés; Lobo, Oscar
Ortega; Calderón Jhon Albeiro.
“Identificación Automática de la Causa de
Falla de Líneas de Transmisión usando un
proceso KDD”. Interconexión Eléctrica S.A-
Universidad de Antioquia. 2008.
[12] Análisis de perturbación de la
ocurrencia 6092003 en el circuito LT Torca –
Guavio 2 del 16 de septiembre de 2003 a las
21:36 horas. Interconexión Eléctrica S.A.
2003.
[13] Análisis de perturbación de la
ocurrencia 1532007 en el Los Palos – Ocaña
del evento 02 de abril del 2007 a las 17:22
horas. Interconexión Eléctrica S.A. 2007.
[14] Análisis de perturbación de la
ocurrencia 5682004 en el circuito LT Torca –
Guavio 2 del 21 de septiembre de 2004 a las
05:27 horas. Interconexión Eléctrica S.A.
2004.
[15] Quintero Crespo, Erwin. “Desarrollo de
un Modelo para la Localización de Fallas en
Sistemas de Transmisión de Energía
Eléctrica Utilizando Técnicas de Inteligencia
Artificial”. Tesis de Maestría. Universidad
Nacional de Colombia. Medellín. 2010.
[16] Mendoza, José D.; Calderón Jhon
Albeiro. “Sistema Informático para la
Localización Semiautomática de fallas en el
Sistema Eléctrico de Potencia Colombiano
Usando Técnicas de Inteligencia Artificial”.
Interconexión Eléctrica S.A. 2008.
Sadul Segundo Urbaez González
Ingeniero electricista de la Universidad
Industrial de Santander – Bucaramanga,
Colombia, en 1994, y terminó sus estudios
de maestría en sistemas de potencia
eléctrica en La Universidad Industrial de
Santander-. Empleado de Interconexión
Electrica S.A E.S.P desde 1997, donde ha
desempeñado los cargos de Analista de
Operación y actualmente desempeña el
cargo de Especialista de Operación de la
Dirección Gestión de la operación de la
Subgerencia Técnica de Interconexión
Eléctrica S.A E.S.P.
Jhon Albeiro Calderón Serna
Ingeniero Electricista de la Universidad
Nacional de Colombia en el año 1988 con
Maestría en Ingeniería de Sistemas de la
Universidad Nacional de Colombia en el año
2007. Entre 1989-1993 trabajó como
Ingeniero de diseño en la firma consultora
Ingeniería Especializada Blandón S.A. Desde
1993 se encuentra vinculado a Interconexión
Eléctrica S.A. donde se ha desempeñado
como Ingeniero del Centro Nacional de
Despacho (CND), Analista Postoperativo,
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 197

Ingeniero de Evaluación Integral de la

inversión y Especialista de Operación.
Actualmente, es coordinador del equipo
Estudios de Protecciones de la Dirección
Gestión de la Operación de ISA y profesor de
cátedra del postgrado de transmisión y
distribución de la Universidad Pontificia
Bolivariana (UPB) en subestaciones y
protecciones.

Luis Everley Llano Zuleta.

Ingeniero Electricista de la Universidad

Nacional de Colombia (1997), con un
diplomado en Ingeniería del Mantenimiento
en la Universidad de Antioquia (2004), y
Master en Ingeniería de la Universidad
Nacional de Colombia (2007). Analista de
Operación de 1997 a 2005, Ingeniero de
Operación hasta el 2008, Especialista de
Operación hasta el 2009 y desde noviembre
de 2009 como Director de Gestión de la
Operación en Interconexión Eléctrica S.A.
E.S.P. El mayor campo de estudio ha la
evaluación de la operación por el diagnóstico
de eventos y desempeño de los sistemas de
protección, así como las técnicas de
Inteligencia Artificial y Sistemas de
Información aplicados a estos tópicos.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
ABATIMIENTOS DE VOLTAJE PROLONGADOS EN LA RED DE MEDIA
TENSIÓN ANTE FALLAS EN LA RED DE 230 Y 115 KV EN ZONA DE
DISTRIBUCIÓN CULIACÁN DIVISIÓN NOROESTE
Eduardo Acosta Félix
Comisión Federal de Electricidad
División Noroeste
Resumen:
En verano del año 2009 Y 2010 se
presenta falla en el bus de 230 kV de
Subestación Mazatlán Dos, operando
correctamente en tiempo la 87B de MZD.
Aunque la falla fue liberada en tiempo, se
presentan disparos en respaldos de media
tensión de varias subestaciones de las
Zonas de Distribución Culiacán y Mazatlán,
de igual manera se queman más de 70
fusibles de transformadores de distribución
(sectores). Se tienen registros de
relevadores de protección así como de
medidores de calidad de energía en las
Subestaciones de las Zonas Culiacán y
Mazatlán que confirman un abatimiento de
Voltaje con una duración aproximada de 5
segundos,
El abatimiento de Voltaje en la Red de
distribución se debió a la falta de
compensación reactiva en el nivel de 115
kV al momento de la falla en S.E. MZD.
Como medida preventiva a corto plazo se
implementaron disparos de alimentadores
de distribución seleccionados en zona
Culiacán que quedan dentro mediante la
función de recierre para evitar la operación
de los respaldos de media tensión.
I. INTRODUCCION
El día 5 de Agosto del 2009 se presenta falla
en TC fase “C” de Interruptor 91030 de S.E.
MZD, se disparan los respaldos de media
tensión de las SE’s CUC, CUS, QLA y MZN,
así como también se queman 88 fusibles de
los transformadores de Distribución (sectores)
de las ciudades de Culiacán y Mazatlán.
• Se presenta falla en transformador de
corriente de fase “C” en bahía 91030
causando su explosión, operando la
199
Iván Soto Rios
diferencial de Bus 1 (87B1-230 KV),
liberando la falla al abrir los
interruptores asociados a este bus,
También, se aprecia la operación de la
diferencial de grupo de la U3 liberando
interruptor 90030 de Bus 2. Adicional a
esta falla, se presenta una segunda
falla, abatiendo más el voltaje en
ciertos nodos del Área Noroeste.
El día 22 de Septiembre del 2009 se presenta
falla en TC fase “B” de Interruptor 73950 de
S.E. CUC, se disparan los respaldos de media
tensión de las SE’s SPE, NTO, CUS y CMI, así
como también se queman 70 fusibles de los
transformadores de Distribución (sectores) de
la ciudad de Culiacán.
• El día 22 de Septiembre se presenta
disturbio en Zona Culiacán provocado
por falla en TC fase B de la línea CUC-
73950-CSA en la S.E. Culiacán Cuatro
(CUC), operando protección diferencial
de barras de 115 kV (87B) librándose
la falla en 160 mseg, afectándose 199
MW y 125,585 usuarios.
• Al despejarse la falla en la S.E.
Culiacán Cuatro se presenta un
abatimiento de Voltaje en la Zona, el
cuál tarda aproximadamente 7
segundos en recuperarse provocando
que los bancos de las Subestaciones
Pericos (SPE T-2), Navolato (NTO T-
1), Culiacán Sur (CUS T-1) y Culiacán
Milenium (CMI T-1) se disparen por
operación de sus relevadores de
sobrecorriente de Neutro de respaldo
de baja tensión (51N).
• Adicionalmente a esta carga, operan
fusibles en 70 transformadores de
distribución (sectores) en Zona
Culiacán derivado del abatimiento de
voltaje, así mismo se dispararon
200

algunos elementos fusibles en Capacidad 40 MVA 115/13.8 KV

arrancadas de ramales de circuitos de I nominal 1673.47 A
distribución y algunos fusibles de Ajuste de pickup Fase 5 x 400 = 2000 A (200%
subestaciones particulares del sistema capacidad OA)
de agua potable en la Ciudad de Ajuste de pickup Neutro 0.9 x 400=360 A (35%
Culiacán. capacidad OA)

Se encuentra ajustado en ecuación de disparo

El día 4 de Agosto del 2010 se presenta falla
en TPC de fase B de línea MZD-93800-HBL,
en S.E. MZD, disparando Líneas MZD-93800-
HBL y MZD-93860-HBL, y operando también
87B del Bus 1 230 kV por contaminación del
área abriendo todos los interruptores
asociados al mismo, así mismo se dispara el
respaldo de media tensión de las S.E. MZO y
en Zona Culiacán operan los esquemas de
bajo Voltaje (27) quedando dentro con la
función de recierre.
solamente elementos de sobrecorriente de
tiempo inverso de fase y neutro de acuerdo a
los criterios normalizados para Subestaciones
de Distribución.
Aquí se observa el abatimiento de voltaje al
momento de ocurrir la falla hasta en un 50%
del valor nominal

• A las 00:46:47 hrs. explota TPC de fase

B de línea MZD-93800-HBL, en S.E.
MZD, disparando Líneas MZD-93800-
HBL y MZD-93860-HBL, y operando
también 87B del Bus 1 230 kV por
contaminación del área abriendo todos
los interruptores asociados al mismo.

II. DESARROLLO
En esta grafica se observa que al momento del
FALLA TC 91030 MZD disparo no se ha recuperado el voltaje.
5 DE AGOSTO DEL 2009

Al ocurrir la falla por explosión del

transformador de corriente, opera
correctamente la diferencial de Bus 1,
liberando los interruptores de acuerdo a la
siguiente secuencia y observándose su
operación en la Figura 1 en un tiempo menor a
0.6 ciclos (8.91 ms). El Anexo 1 muestra la
localización de la falla.

Valores antes de la falla

V= 7.9 KV
I = 1364.8 A
P= 30.33 MW
Q= 5.78 MVAR
fp = 0.98

Valores al momento del disparo

V= 6.1 KV
Análisis de operación de protecciones de I = 2519.2 A
sobrecorriente de transformadores de P= 36.95 MW
potencia ante falla ocurrida el día 5 de Q= 20.465 MVAR
Agosto de 2009 fp = 0.87
Transformador 1 de S.E. CUC
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 201

En Zona Culiacán se presento la salida de Conclusiones (falla 5 Agosto 2009):

88 sectores por operación del fusible
derivado del abatimiento de voltaje Se tienen registros de relevadores de
protección así como de medidores de calidad
Tr de distribución de energía en las Subestaciones de las Zonas
75 KVA 13,200/220 Volts Culiacán y Mazatlán que confirman un
I nominal = 3.3 Amp abatimiento de Voltaje con una duración
Voltaje de operación durante la falla 4.1 kV aproximada de 5 segundos para Zona
fase a tierra. Culiacán y 2.8 segundos en Zona Mazatlán,
Corriente de operación durante la falla 6.6 ocasionando la operación de 88 sectores de
Amp en Media Tensión Media Tensión y 4 bancos de transformación
por sobrecorriente.
1 2 3 4 5 7 10 2 3 4 5 7 100 2 3 4 5 7 1000 2 3 45 7
Se concluye que la operación de los
1000 1000
700 700 relevadores de protección y los fusibles de los
500
400
500
400 transformadores de Distribución en la red de
300 1 300
200 200 media tensión fue correcta ya que operaron de
100 100
acuerdo a su curva tiempo-corriente los cuales
70
50
70
50
están ajustados de acuerdo a los criterios y
40
30 1. Fus 75 KVA CHANCE-K002
40
30
normativa de CFE del proceso de Distribución.
20 Total clear. 20
Ic= 6.6A T= 4.84s

S10 FAULT DESCRIPTION: 10

E7 Close-In Fault on: 0 8a 13.8kV - 0 BUS1 0.22kV 1T 3LG 7R=120
C
O5 5 FALLA TC 73950 CUC
N4 4
D3 3 22 DE SEPTIEMBRE DE 2009
S
2 2

1
.7
1
.7
• El día 22 de Septiembre se presenta
.5
.4
.5
.4
disturbio en Zona Culiacán provocado
.3
.2
.3
.2
por falla en TC fase B de la línea CUC-
73950-CSA en la S.E. Culiacán Cuatro
.1 .1
.07 .07 (CUC), operando protección diferencial
.05 .05
.04
.03
.04
.03
de barras de 115 kV (87B) librándose
.02 .02 la falla en 160 mseg, afectándose 199
.01
Fault I=6.6 A
.01
MW y 125,585 usuarios.
1 2 3 4 5 7 10 2 3 4 5 7 100 2 3 4 5 7 1000 2 3 45 7
CURRENT (A) • Al despejarse la falla en la S.E.
For
TIME-CURRENT CURVES @Voltage By
No.
Culiacán Cuatro se presenta un
abatimiento de Voltaje en la Zona, el
cuál tarda aproximadamente 7
Registro de medidor ION de 52030 CUC segundos en recuperarse provocando
ajustado a + - 10 % del Voltaje nominal que los bancos de las Subestaciones
Pericos (SPE T-2), Navolato (NTO T-
1), Culiacán Sur (CUS T-1) y Culiacán
Milenium (CMI T-1) se disparen por
operación de sus relevadores de
sobrecorriente de Neutro de respaldo
de baja tensión (51N).
• Adicionalmente a esta carga, operan
fusibles en 70 transformadores de
distribución (sectores) en Zona
Culiacán derivado del abatimiento de
voltaje, así mismo se dispararon
algunos elementos fusibles en
Se observa que en el registro del medidor del arrancadas de ramales de circuitos de
TR1 de S.E. CUC se detecta un disturbio de distribución y algunos fusibles de
duración 5 seg subestaciones particulares del sistema
de agua potable en la Ciudad de
Culiacán.
202
T1 CUS
Magnitudes de voltaje en el T1 CUS al
momento de ocurrir la falla
Magnitudes de voltaje en el T1 CUS al
momento del disparo, se observa que el
voltaje no se ha recuperado al 100 %
Con el propósito de analizar el abatimiento de
Voltaje y encontrar las causas que lo
generaron se programo una reunión el 25 de
Septiembre del 2009 con el Área de Control
Noroeste (ACNO), la Gerencia de Transmisión
y Transformación Noroeste (GTTNO) y la
Gerencia Regional de Producción Noroeste
(GRPNO), acordando en dicha reunión que se
solicitaría el apoyo de la Gerencia de
Ingeniería Especializada de CFE (GIE)
El 28 de Enero del 2010, se acuerda en el
comité regional de Protecciones Noroeste
como medida preventiva a corto plazo la
Implementación de esquemas de protección
por bajo Voltaje para evitar el disparo del
respaldo de banco de distribución y la
operación de fusibles en la red.
Criterios (27):
• Se deberán de configurar 2 tiros de
carga (equivalente al 50 % de la carga
promedio del transformador de
potencia), el primero operaria sobre 1
circuito a un nivel del 80% del Voltaje
nominal de Fase a tierra con una
duración de 1 segundo y un nivel de
reset (drop out) del 90% del valor
Nominal de Voltaje.
• El segundo tiro de carga operaria sobre
1 circuito a un nivel del 80% del Voltaje
nominal de Fase a tierra con una
duración de 1.5 segundos y un nivel de
reset (drop out) del 90% del valor
Nominal de Voltaje.
• Quedando dentro la carga de 2 o 3
circuitos de acuerdo a su importancia.
• Los disparos se implementarán a
través de los relevadores de
sobrecorriente de los alimentadores
realizando el recierre correspondiente a
los 15 segundos.
• Se deberá de habilitar una alarma a
control supervisorio por la operación de
cada uno de los pasos de tiro de carga.
• Se deberá habilitar un bloqueo del
esquema a través de un conmutador
para evitar disparos en falso, durante
trabajos de mantenimiento.
• Se deberá de habilitar una lógica para
bloqueo por ausencia de potencial en
media tensión, para detectar la
operación de fusibles de los TP’s del
bus de media tensión y evitar disparos
erróneos.
• El voltaje deberá ser monitoreado por
todos los relevadores de media
tensión, para asegurarse que existe un
abatimiento real del mismo.
FALLA TPC 93800 MZD
4 AGOSTO DEL 2010
A las 00:46:47 hrs. explota TPC de fase B de
línea MZD-93800-HBL, en S.E. MZD,
disparando Líneas MZD-93800-HBL y MZD-
93860-HBL, y operando también 87B del Bus
1 230 kV por contaminación del área abriendo
todos los interruptores asociados al mismo.
Derivado de la falla en subestación Mazatlán
Dos de 230 kV el día 04 de Agosto del 2010 a
las 00:46:47, operó el interruptor 42010-MZO
de subestación Mazatlán Oriente y en Zona
Culiacán operaron los esquemas de bajo
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 203

voltaje (27’s) realizando un recierre de 15 De acuerdo a la GIE existen dos soluciones

segundos de acuerdo a la relación anexa. para esta problemática de abatimiento de
Voltaje:
PROTECCIONES OPERADAS: Esquema 27
de bajo voltaje pasos 1 y 2 en Zona Culiacán y 1. Tiros de carga.
51N de interruptor 42010-MZO de Zona 2. Compensación dinámica reactiva.
Mazatlán por abatimiento de voltaje.

CARGA AFECTADA: Se afectan 146 MW IV. BIBLIOGRAFIAS

(recierre) en Zona Culiacán y 36 MW en Zona
Mazatlán por disparo de MZO-42010 en S.E.
Mazatlán Oriente.
Podemos observar el comportamiento del
voltaje al ocurrir la falla.
Eduardo Acosta Félix
Egresó del Instituto Tecnológico de Hermosillo
(ITH) obteniendo el título de Ingeniero
Eléctrico en el año de 1995. En 1996 inició sus
actividades en la Comisión Federal de
Electricidad División Noroeste como supervisor
de protecciones en la Zona Caborca. Del 2000
al 2002 se desempeñó como Jefe de Oficina
de Subestaciones en la Zona Caborca. Del
2002 al 2003 se desempeñó como Jefe de
Departamento de Distribución en la Zona
Caborca. Del 2003 a la fecha se ha
desempeñado como Jefe de Oficina Divisional
de protecciones de la División Noroeste.

Iván Soto Rios

Egresó del Instituto Tecnológico de Hermosillo
obteniendo el título de Ingeniero Eléctrico en el
año 2004. En este mismo año ingresó a la
Comisión Federal de Electricidad ocupando el
puesto de supervisor de construcción en el
Departamento Divisional de Proyectos y
Construcción de la División de Distribución
Noroeste. Desde el año 2005 se encuentra
laborando como supervisor Divisional de
III. CONCLUSIONES
Protecciones.
Los esquemas de bajo Voltaje (27’s)
implementados en Zona Culiacán operaron de
forma correcta evitando un abatimiento de
Voltaje prolongado de 4 segundos o más, ya
que para estas falla el abatimiento de Voltaje
en la Red de Media Tensión duro 1 segundo
por la implementación de los 27’s

El comité regional de Protecciones Noroeste

solicito a la Gerencia de Ingeniería
Especializada de CFE que realizara el estudio
para determinar la causa de los abatimientos
de voltaje prolongado.

De acuerdo al estudio realizado por la GIE el

abatimiento de Voltaje en la Red de las Zonas
Culiacán y Mazatlán se debió a la falta de
compensación reactiva en el nivel de 115 kV
debido a falla evolutiva en DPC de la S.E.
MZD en 230 kV.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Energy Production Using the Co-Generation of
Methane Gas Connected to Copel’s Distribution
Grid – Tests and Procedures
Eloi J. Rufato
Department of Protection System
COPEL - University of Sao Paulo- USP-POLI – UTFPR
Curitiba, Paraná, Brasil
eloi.ruffato@copel.com
ABSTRACT
This is a groundbreaking project in Brazil to generate
electric power in biodigestors using pig slurry. The purpose
of the experience is to study the technical and economic
feasibility of setting up biodigestors in rural properties
dedicated to pig-raising, in order to generate power with
methane gas yielded by the decomposition of collected
organic matter that will be consumed by the livestock
business, the surplus of which may be sold to the local
service distributor.
It is the generation of pig waste-based power using
biodigestors.
To enable the connection of this type of distributed power to
COPEL’s grid there are some special measures that must be
taken in order to protect the electric system and ensure
safety, as they are low-power operating systems, and until
now it was not feasible to equip them with the protection
systems in place on larger generators.
The project has been developed and monitored for nearly a
year in a small rural property in the Foz do Iguaçu region,
and its success could lead to a triple benefit. “For pig-
raisers, generation of power in biodigestors would be a
source of additional income from their activity, while
COPEL would strengthen its energy availability to supply
the market”, while greatly helping environmental balance,
as by using pig slurry in the biodigestor chambers it
prevents the waste from running off into rivers, lakes and
reservoirs and from affecting the properties of water that
facilitate the proliferation of algae and other harmful
microorganisms to human health.
The main environmental problem ensuing from pig-raising
activities is the phosphorus-rich pig slurry that is the main
nutrient of what is called “blue algae”, a type of organism
that contaminates and deteriorates water, and when
swallowed or even if it touches the skin can cause serious
health problems to human beings and animals alike.
In the case of the property where the tests were run, electric
capacity for power generation is 50 kilowatts – enough to
generate power for 100 homes with average consumption
standard.
205
Carlos M. V. Tahan and Carlos C.B. de Oliveira
Department of Energy and Automation
University of Sao Paulo-USP-POLI
São Paulo, SP, Brasil
barioni@pea.usp.br and cmvtahan@pea.usp.br
In the past, decentralized generation projects in rural
environments would usually be stalled by a serious
operation and safety problem, e.g. generation in parallel
with the distribution grid of distribution companies. It would
jeopardize the electricians who would be exposed to serious
accidents, as the company would loose all its control of the
flow of electric power in their facilities.
That is the main work of COPEL’s team of technicians and
engineers that have joined the project: to develop a low-cost
system capable of generating electric power safely for rural
producers and also for electricians and other operators who
must operate and maintain distribution grids.
In order to be connected to the distributing company’s grid,
those units of distributed generation, usually at low power,
require certain special cares in terms of electric protection
and safety – similar to systems and devices installed on
large generation plants.
To manufacture the prototype installed at the Granja
Colombari rural property in the Foz do Iguaçu region,
COPEL made investments on field tests, sophisticated lab
and technology on electric studies to ensure that the
solution proposed for connecting micro-generators to the
company’s grid will not lead to problems, but rather, to the
solution. The operation of the generator that is run on pig
slurry yielded by 3 thousand pigs is monitored by
oscillographs.
Field tests are being run on the facilities for short-circuit
and operations on the Distribution Lines by COPEL’s
technicians, and results have been quite satisfactory.
The purpose of this study is to show technical solutions
adopted by Protection and connection systems for those
small generators on distribution lines, and to show
procedures carried out during the Short Circuit Tests for the
System Protection Tests.
Index T erms – Protection, Production of Energy, Methane
Gas, Distributed Generation.
I INTRODUCTION
One of the main functions of the distributed generator
protection systems connected to subtransmission grids or
those that distribute electric power is to immediately
interrupt the connection whenever an islanding occurrence
206

is detected. Islanding occurs when part of the electrical grid
is disconnected from the rest of the distribution company’s
system, but is still energized by one or more distributed
generators connected to it, forming an isolated subsystem of
the distribution company. Such occurrence should be
avoided as it is a safety hazard to both people and
equipment, as well as it can deteriorate the quality of power
delivered by the distribution company to their customers.
This study analyses and presents the main adjustments made
and tests run to assess and adjust anti-islanding protection
relays for synchronous generators.
The main relays used for this type of generation connection
are small-sized, comprising phase-shift relays or “jump
relay”, conventional sub/over-frequency relays and
sub/over-voltage relays.
III VECTOR JUMP RELAY
The vector jump relay or voltage jump relay is installed to
operate the generator’s circuit-breaker coupling to the
distribution system, the operation principle of which is
based on angle displacement of the phase between the
generator’s internal voltage “Eg” and the voltage at the
generator’s terminals “Vtg”.
As shown in Figure 1, under the permanent regime there is
dVi voltage drop between the generator’s terminal voltage
“Vtgi” and the generator’s internal voltage “Egi”, due to
voltage drop caused by the generator’s current during its Xd
synchronized reactance. There is “θ” angular discrepancy
between the generator’s terminal voltage and the generator’s
internal voltage, Figure 1a.

II PROBLEMS ENSUING FROM GENERATORS

OPERATING UNDER ISLANDING

Problems ensuing from small generators operating under the

islanding mode, which are connected to electrical systems of
power distribution companies, are related to operational
safety and personal safety, affecting distribution companies
and third parties, and causing technical and commercial
problems in terms of nominal value variance in power
supply (voltage and frequency) and legal problems, which
are listed below:[2] Figure 1 – Vector Jump Relay.

• Due to the new configuration that will result from Manufacturers of those relays recommend values within 4-
generator islanding, after the circuit-breaker that 12 degrees range, depending on the type of generator and
services the feed circuit substation is opened, it is grid to which they will be connected.
possible that protection equipment that still remain
in the system will suffer decrease of sensitivity and IV CONFIGURATION UNDER STUDY
incoordination, among which: fuse links, repeater
fuse keys, reclosers and circuit-breakers with The 34.5kV distribution system that was tested is located in
protection relays. the Foz do Iguaçu region, as shown in the single-wire
• The new resulting configuration may cause diagram in Figure 2, comprising the following elements.
variation of value for the supplying of voltage and
frequency for other consumers that continue to be
fed by the islanded generator, whereby those values
could be outside limits set forth by the regulating
agency, either in excess or in the lack of.
• As the distribution company’s protection
equipment has recloser relays that enable the circuit
to be reactivated should short-circuit situations
occur up to 3 times, it is necessary to install devices
to enable generators to reconnect only after they
have been checked for synchronism and voltage in
the generator’s circuit-breaker connection.
• If generators continue to feed part of the
distribution circuit after a failure has occurred and
if the protection located at the point where the
generator connects to the distribution system fails Figure 2 – Configuration of the System under Study – São
to lockout, safety problems may occur involving Vicente Feeder.
technical personnel from power companies and
outsourced personnel, as there will be energized V PROTECTION SYSTEM TO CONNECT
cables on the ground lacking activated support to
GENERATORS TO 34.5 KV FEEDERS
eliminate this type of fault.
The diagram presented in Figure 3 shows the minimum
protection system with trip points for coupling small-size
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
generators to COPEL’s Distribution System 34.5 kV
feeder.[3]
NOTE – Generator protections have not been addressed,
which will be assessed and installed by the autoproducer.
Protection Functions that should be incorporated to the
system in Figure 3:
• Phase and neutral current overcurrent,
instantaneous and timed – 50/51 – 50/51N
• Over-voltage (3 phases) – 59
• Under-voltage ( 3 phases) - 27
• 81 O/U Over and Under Frequency
• Time Relay - 62
• Synchronism Check Relay – 25
• Vector Jump – 78
Figure 3 – Minimum Protection for connection to
Autoproducer 34.5 kV feeder.
VI LAB TESTS
Tests were carried out using ATP-developed cases, whereby
operating conditions and scheduled short-circuit tests we
simulated for field conditions, as per Table 1.[1]
207
Table 1 – Tests Performed
CASE 1
Simulation 1 - XG2 vector jump relay
TEST TYPE – Phase-Ground Short-Circuit at the São
Vicente Feeder
ANTICIPATED RESULT: Generator’s circuit-breaker
protections should be triggered, turning it off before it is
automatically reconnected to the feeder, as shown in Figure
4.
TEST RESULT - KG_52a circuit-breaker was opened when
the vector jump relay was triggered, with 89 ms.
Figure 4 – Current, voltage and vector jump relay
oscillography - Lab Test.
VII FIELD TESTS
After lab test results were approved, a panel with protection
equipment was installed at Granja Colombari and short-
circuit and anomaly tests and real simulations were run on
COPEL’s distribution system at the point of connection, as
shown in Figure 5. Ten field tests were scheduled and run.
Figure 5 – Phase-Ground short-circuit test performed at
the São Vicente Feeder.
208
VIII TEST RESULTS
Test 4 consists in mounting the generating unit in parallel
with COPEL’s distribution grid.
ACTION: To cause a single-phase short-circuit on the 34.5
kV distribution grid, São Vicente circuit, at the test point, as
per the single-wire diagram in Figure 2, next to the circuit
source.
ANTICIPATED RESULT – Protections on the generator’s
circuit-breaker are triggered causing it to halt operation
before the feeder is automatically started up again.
TEST RESULT – Circuit Breaker was opened when the
vector jump relay was triggered, with 61 ms operating time
as shown by oscillography in Figure 6, when the generator’s
current was halted.
Figure 6 – Current, voltage and vector jump relay
oscillography – Field Test.
IX SUMMARIZED RESULTS TABLE
Table 2 shows time of operation of protection, circuit-
breaker opening time and which protection was triggered
when field tests were run.[1]
Table 2 – Opening Time Protection
X CONCLUSIONS
Based on the study of the oscillograms and measurement
results, and data ensuing from field tests, it is shown that
90% of the tests were run under the protection of “Vector
Jump” relay triggering for different types of simulations and
tests, among which are phase-ground short-circuits next to
the source, phase-ground short-circuits next to the load,
phase opening at the feed circuit, phase-phase-ground short-
circuits, which caused the removal of the motor-generator
unit operating in parallel with the distribution system,
whereby time of operation of protection units were 60-80
milliseconds.
Considering that is a low-cost type of protection compared
to a complete protection system for generator connection
operating in parallel with distribution systems, and based on
results observed during simulations and field tests, we deem
that the protection system in place using a jump vector relay
performed as anticipated; operation involved a motor-
generator unit run on methane gas in parallel with COPEL’s
distribution circuits for the Granja Colombari pilot project.
The system is still operating in parallel, and quantities and
occurrences in the generator and feeder are being monitored
via test, so that throughout a one-year period it will be able
to define if the system is performing satisfactorily for all
occurrences observed, thus making it possible to define that
system as a standard to create a Standard Norm for this type
of connection.
Acknowledgments
The authors acknowledge the support and facilities provided
by Copel, Paraná, Brazil.
REFERENCES
[1]MARQUEZ, Diógenes, OLIVEIRA, Francisco,
MOTOIE, Eduardo, 2008, Relatório de Ensaios no Gerador
Sincrono de Pequeno Porte em Paralelo com sistema de BT
Granja Colombari. COPEL, Curitiba, Brasil.
[2] VIEIRA, José Carlos M. Jr, 2006, “Metodologia para
Ajustes e Avaliação do Desempenho dos Relés de Proteção
Anti-Ilhamento de Geradores Síncronos Distribuídos”
FEEC/UNICANP, Campinas, São Paulo.
[3]RUFATO, Eloi J, 2006, “Viabilidade Técnica e
Econômica da Modernização do Sistema de Proteção da
Distribuição”, POLI-USP, São Paulo, Brasil.
[4]OLIVEIRA, Carlos B, RUFATO, E, J, 2005, “Short
Circuit Tests for Detection of High Impedance Faults”,
CIRED-2005, Turin, Italy.
[5]IEEE, “IEEE-1547 Standart for Interconnecting
Distributed Resources with Electrical Power Systems”,
2003.
[6]ANEEL, 2008 “Procedimentos de Distribuição –
Módulo 3 – Acesso aos Sistemas de Distribuição”, Brasil.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 209

Optimal Energy Resource Allocation in a Microgrid Environment

A. A. PALIZBAN S. ALLAN M. ADAMIAK M.ZADEH

HAJIMIRAGHA
GE Digital Powertech Powertech GE Digital GE Digital
Energy Labs Labs Energy Energy
U.S.A. Canada Canada U.S.A. Canada

SUMMARY

The DOE’s vision of the future electric power infrastructure, GRID 2030, identifies microgrids as one
of three major technical cornerstones for a more reliable and congestion-free energy delivery system,
and describes distributed intelligence and clean power as key technologies needing development[1].
Microgrids are envisioned as local power networks that utilize DER and manage the local energy
supply and demand. The DERs are relatively small in size, and consequently, there is a trend towards
their integration at the distribution level. As such, microgrids increase the viability of DER in the
distribution grid by aggregating these resources into clusters that have better grid stability properties
than a collection of standalone generators. While microgrids would typically operate connected to the
national bulk power transmission and distribution system, they would have the ability to disconnect
from the grid and function in “island mode” when necessary. This would increase customer reliability
by reducing their susceptibility to grid disturbances, while offering desirable security features. In the
long term, a move towards a more decentralized power infrastructure has the potential to guarantee a
more resilient and survivable grid where faults and/or infrastructure damage remain localized.

Historically, a distribution network consisted passive loads dispersed along radial feeders originating
from a substation with one or more connections to the bulk electricity system. A typical feeder would
have reclosers or sectionalizers spaced down its length with manually operated tie switches to other
feeders. Automatic control was limited to basic reclosing operations. SCADA control and monitoring
was minimal. Communications, if available, was limited to low bandwidth radio or dial-up circuits.
Local distributed generation was virtually nonexistent. A microgrid is essentially a distribution
network wherein local distributed energy resources (DERs) are utilized to supply some portion of the
electric power and/or local heating demands and can be operated in an autonomous mode. Microgrids
will often have no pre-existing communications network or control system. Microgrids may not have
dedicated operating personnel. A microgrid could be formed from a utility distribution network, an
industrial complex, a university campus, or a remote community.

Recently, new forms of renewable generation technologies utilizing non-traditional energy sources
such as wind, solar, biomass, or waste heat have been developed. As well, energy storage
technologies such as electrolyzers and hydrogen generators (e.g., fuel cells) have emerged and have
been coupled to the renewable generation as a way to mitigate renewable intermittency. In addition to
210

renewables, conventional (dispatchable) generation may also be integrated into the distribution
system. The resulting evolution in the makeup of the distribution network leads to the concept of a
microgrid.

Widely recognized technology gaps discourage the use of microgrids. In particular, their low inertia
environment is not presently conducive to the interconnection of inverter-based assets typically used
by renewable or clean power/energy sources, and existing protection systems are inadequate.
Furthermore, the limited generation capacity in these grids, coupled with the intermittent nature of
renewable energy sources, creates the need for more intelligently coordinated operation of generation
and loads to ensure a more stable and effective use of the available energy. Finally, controls and
protection to allow more autonomous separation and re-integration with the bulk grid have been
developed, but need to receive regulatory and end user acceptance and be fielded and commercialized
more rapidly.

KEYWORDS

Microgrid, Dispatch, DER, Hydrogen Storage, HARP

BENEFITS

At a high level, the interest in microgrid power systems is driven by a growing desire to locate DER
closer to load centers. This interest in DER is being fuelled by a number of factors, including:

• Transmission congestion, and logistical problems with siting new transmission lines make it
appealing to site DER in distribution systems to cope with new loads.
• Utilization of DER can help utilities defer investments in generation and transmission capacity.
• DER have the potential to offer increased total energy efficiency when used with combined heat
and power (CHP) or combined cooling heat and power, and can therefore reduce energy costs.
• When appropriately integrated, DER have the potential to improve power availability and quality.
• Distributed systems offer potential security advantages over centralized systems.
• DER promote fuel diversity (e.g., biomass, landfill gas, flare gas, wind, solar) and therefore reduce
overall energy price volatility.
• Renewable DER such as wind and solar photovoltaics provide emissions-free energy.
• DER offer a quicker solution with regards to installation, lead time and siting relative to
centralized generation.

While all of these benefits make DER attractive, the primary concern at the utility level is the system
operation and protection issues associated with the existence of a large number of independent power
producing assets operating without coordination. Microgrids offer a framework that resolves this
concern through the aggregation of DER into well-behaved entities that can be dispatched by the
utilities.

ENERGY STORAGE

The previously mentioned benefits are augmented through the incorporation of energy storage. With
proper control, energy storage can offset the intermittency/variability inherent in renewables -
achieving a higher utilization of these resources; allow dispatchable generators to operate closer to
their maximum theoretical efficiencies; and improve microgrid transient stability by responding
quickly to power imbalances.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 211

OPTIMAL DISPATCH

Economic Dispatch is the process of allocating the required load demand between the available
generation units such that the cost of operation is minimized. Cost of operation is typically defined by
fuel cost but can include maintenance or other costs. Furthermore there may be additional constraints
imposed such as the need to minimize the production of greenhouse gas or the need to accommodate
CHP applications. While considerable work has been done on similar problems in the power system
community[2],[3], the important contributions of this work are:

1. The ability to include future predicted values of loads, renewable generation, and fuel and
electricity prices
2. The ability to In addition to dispatch power while simultaneously dispatching heat, optimally.
3. The ability to commit and de-commit various assets such as diesel gensets, hydrogen (fuel cells
plus electrolyzers), solar, wind, and hydro renewable generation, combined heat and power (CHP)
units, and electrical and thermal storage.

The microgrid optimal dispatch algorithm uses a technique known as model predictive control. This
technique makes use of historical data in the form of daily, weekly, and annual load profiles; hydro,
wind, and solar forecasts; and fuel or electricity market pricing information (if a bulk grid connection
exists). Given these predictions the algorithm determines the cost of operation for a fixed time period
into the future, typically 24 hours. Then, it optimizes the objective function of an optimization
problem that models this cost of operation in order to determine the required control actions. These
control actions include selecting the machines to be operated at any given time, selecting a machine
for isochronous control and selecting the remaining machines for droop control. The algorithm
determines when energy should be stored and when it should be returned to the system. It determines
the best power reference point for each droop machine and for each storage device.

Forecasts

Inputs 80 Optimization Outputs

Xm
0

Model

Figure 1 – Model Predictive Control

The dispatch algorithm includes an advisory mode whereby the algorithm recommends the optimal
operating point for each resource to the HMI but does not automatically carry out control actions.

OPTIMIZATION

The cost function data required to compute the costs associated with a particular dispatch profile
includes the following:

1. Predicted fuel costs over the prediction horizon.

2. Fuel consumption curves of the various generators. These can be arbitrary functions of both the
power and heat generated by the units.
3. Cost of importing and exporting from the grid.
4. Load shedding costs. The cost of load shedding is assumed to be proportional to the amount of
shortfall between the load demanded and supplied.
212

5. Startup costs for each of the three different types of units (pure electric generation, CHP, and pure
heat generation) depends on the unit and the time it has been off. Note that the startup cost is only
applied at the first time step at which the generator is turned on.

The total predicted cost for the system, neglecting the startup costs of the units, is given by

N + t0
∑ (∑ C f (i, t )) + (∑ Cl (iel , t )) + C g (t ) (1)
t =t0 i iel

where, Cf (i, t) represents the fuel consumed by unit i at time t, Cl(iel, t) represents load shedding cost
at load iel at time t, and Cg(t) represents the cost of importing from the grid (or the price of exporting to
the grid).

This cost must be minimized subject to various (linear) constraints such as upper and lower bounds on
power generation from various machines, ramp rate constraints. The following assumptions are made
to the problem to allow it to be formulated in the manner described:

1. Fuel curves are approximated by affine or convex quadratic functions. In many important cases,
the fuel consumption is in fact an affine function of power generated. This is also true in all cases
where the efficiency is constant over the operating range of the unit.
2. It is assumed that cost of import form the grid is always greater than or equal to price of export.
This is true in practice.
3. For CHPs, the relation between heat and power is assumed to be linear.
4. Startup costs are taken into account outside the convex optimization routine.

The overall optimal dispatch problem can be formulated as a convex optimization problem in dual
form by converting each of the constraints and cost functions into convex constraints.

For a background in convex optimization theory sufficient for the discussion in this section, see [4].
Assume that u(i, t) are given over the prediction horizon {t ∈ Z : t0 ≤ t ≤ t0 + N}. A convex
optimization problem is described by:

max(b′y ) (2)

subject to

A′y ≤ K ⋅ c (3)

The vector y of decision variables is as follows:

y = [Ye; Yc; Yt; Yel; Ytl; Yes; Yts; Yaux ] (4)

where

Ye is the dispatchable generation from the pure electric generators

Yc is the dispatchable generation from the CHP generators
Yt is the dispatchable generation from the pure heat generators
Yel is the power supplied to the dispatchable electric loads
Ytl is the power supplied to the dispatchable thermal loads
Yes is the vector of energy stored in the batteries
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 213

Yts is the temperature of the thermal storage units

Yaux is a set of auxiliary variables used to make the problem convex.

Note that each of these variables is defined at every time step. For example, if there are Ne pure
electric units, all of which are dispatchable, and we are using a prediction horizon of N steps, then Ye
is of dimension Ne × N.

In order to solve an optimal dispatch problem for a specific microgrid using the optimization code
developed by the GE Global Research Center, only a few high-level details of the assets in the
problem need to be specified. The algorithm generates the appropriate matrices and vectors of the
large-scale optimization problem, and calls an open-source solver CLP (Coin-or-Linear Programming)
solver distributed under the common public license (CPL) [4]to solve it. The solution provides
supervisory control set points for all dispatchable generations, loads, and grid generation as a function
of time.

THE MICROGRID CONTROL SYSTEM

Each DER will include local control loops that regulate real and reactive power, or alternately,
frequency and voltage at given reference points. The central controller provides a supervisory control
layer for the microgrid. The central controller monitors the state of all resources in the system. The
dispatch algorithm is processed, and resulting control actions are sent back to each DER. Inputs to the
system include the power generated by each DER and the power consumed by each load Control
actions include starting/stopping a DER, Changing the power reference point of a DER, changing the
control mode of a DER (isochronous or droop), opening/closing a breaker, or changing an IED setting.
The microgrid control system (MCS) is composed of a central controller, a communications network,
and a local HMI for configuration and monitoring.

Setup Tool

Loads

Power consumed, Operational Parameters

On-line status (settings)

Power generated,
Renewable On-line status
generators Availability Microgrid
Controller

Power generated
Power consumed Power generated, Start/Stop Start/Stop
State of charge On-line status Isoc/Droop Input Pref
Online status Availability Output Pref Output Pref
Availability
Dispatchable
generators

Storage
Devices

Figure 2 – Microgrid Control System

The dispatch algorithm is implemented on an embedded processing platform. The main processor is a
Freescale PowerPC 8270 running the VxWorks operating system with 64 MB of nonvolatile RAM
(disk on chip). The processor has a native floating-point engine. A separate processor handles
communications. The Modbus RTU protocol is chosen in order to support a wide range of devices
including PLCs, smart meters and protective relays. Both Ethernet and serial media are available. The
214

initial design is simplified by utilizing communication protocol converters wherever necessary to

connect various PLCs to the MCS network.

Since DERs and loads are likely to be dispersed across the microgrid, a communications network is
required to transport data between the central controller and remote devices. The communications
network should be well suited to application in the distribution system. It should support both
centralized and distributed control schemes. Ethernet is envisioned as the best choice in terms of
topography, reliability, cost, and ease of installation and maintenance.

Ethernet provides a flat architecture that is easily extendable and supports multiple protocols in order
to accommodate a broad range of devices. Ethernet provides encapsulation over IP (tunneling) for
serial asynchronous multi-drop protocols including Modbus. For ranges of 100-2000 metres, multi-
mode fiber-optic cable is recommended. This medium is immune to ground potential differences and
transients generated by faults or switching events. For longer distances (up to 10 km) Ethernet radios
are used to extend the communications network. These devices employ frequency hopping, spread
spectrum radio. This transmission method is highly immune to interference. The topology is point-to-
multipoint with a provision for multiple access points for further reliability. Security features include
encryption using RC4-128 with automatic key rotation and authentication using 802.1x, and CHAP.

THE BELLA COOLA HARP PROJECT

The Bella Coola project is an initiative to install a hydrogen assisted renewable power (HARP) system
in community of Bella Coola which is part of BC Hydro’s non-integrated service area. The Hydrogen
storage system has been designed by Powertech Labs and the microgrid control system by General
Electric with key components supplied by Hydrogenics Inc. and Dantherm Power.

The Bella Coola power system is comprised of two generating plants and two communities. The Ah
Sin Heek (ASK) generating station has six diesel generators ranging in size from 300 to 2500 kVA.
Clayton Falls is a run-of-river, hydro-electric plant with two generators having ratings of 700 and 1400
kVA. A single 25 kV feeder connects the two generating stations with the community of Bella Coola
tapped from this circuit. The community of Hagensborg is tapped from a second feeder which is fed
from the ASK station. Bella Coola and Hagensborg have summer/winter peaks of 2.1/1.5 MW and
2.6/1.7 MW respectively.

Figure 3 – Bella Coola Site Map

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 215

There is no interconnection with the bulk power system. A hydrogen energy storage system has been
installed at the ASK station. This system is made up of a 320 kW Hydrogenics electrolyzer capable of
converting electricity to hydrogen at a production rate of 60m3/h, a 100 kW fuel cell which uses
Ballard Power’s air cooled stacks that have been packaged by Dantherm Power, and a hydrogen
storage system. The storage system receives hydrogen from the electrolyzer at low pressure (150 psi).
A 50 horsepower compressor compresses the hydrogen to high pressure (500 – 2850psi) and feeds a
storage system with a capacity of 13920 litres.

Figure 4 Hydrogen Storage System Setup at Powertech Labs

The compression of hydrogen requires energy, however a much larger footprint and larger number of
storage tanks are required if hydrogen is stored with no additional compression.All of the hydrogen
storage components have been containerized and factory tested in order to simplify installation and
commissioning at site. The hydrogen storage system also provides energy for a hydrogen-powered
service vehicle to be used by the operating staff.

The aim of this project is to increase the utilization of the Clayton Falls hydro plant by storing energy
as hydrogen when appropriate and returning this energy to the system at optimal periods in order to
reduce the consumption of diesel fuel and the corresponding emission of green house gas.
216

Figure 5 – Bella Coola Microgrid Control System

SIMULATION RESULTS

The dispatch algorithm has been tested using Matlab. The input to the simulation is the system loading
and renewable contribution captured by the Bella Coola logging system. The response of the diesels
and storage is modeled within Matlab. The plot of Figure 6 shows the results over 1 day period. Note
that the electrolyzer operates during periods where the load is at a minimum. The fuel cell provides
power to the system during the period of peak loading. The overall action of the storage system serves
to produce a relatively flat output from the diesels.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 217

Figure 6 – Matlab Simulation of a Representative Highly Penetrated Hydro Usage Scenario

CONCLUSIONS

This paper described a Microgrid Control System (system architecture, control algorithms and
hardware) that offers: optimal dispatch of distributed energy resources (DER), intentional islanding
capability, and electrical load / energy management. The MCS technology addresses the complexity of
electrical demand, heat and power generation, and power distribution challenges. It also addresses the
need for energy security, continuous operation, and environmental concerns by reducing carbon
footprint. DERs within the microgrid are equipped with communications and local controllers that
respond to supervisory commands to regulate real power, reactive power, frequency, and/or voltage.
Intelligent electric devices located elsewhere in the microgrid provide system loading, voltage, and
frequency information and carry out switching operations. The MGS implements a centralized,
supervisory control layer. It polls all resources, executes central control algorithms, and sends
resulting control commands back to each resource. The core of the MCS technology is currently under
development by GE Digital Energy on an MCS product platform that enables:

• Aggregation and algorithms to enable dispatch-ability

• Flexibility to integrate a diverse set of controllable DER assets
• Integration of renewable energy asset
• Improved energy efficiency and optimal energy utilization
• Reduction in cost of energy and total cost of ownership
• Improved power quality and availability in islanded operation

BIBLIOGRAPHY

[1] GRID 2030—A National Vision for Electricity’s Second 100 Years”, United States Department of
Energy, Office of Electric Transmission and Distribution, July 2003.
218

[2] Electric Power Systems Applications of Optimization. James A Momoh, Marcel-Dekker, 2001
[3] MatPower software, Ray Zimmerman et al., http://www.pserc.cornell.edu/matpower/
[4] Lectures on Modern Convex Optimization. Aharon Ben-Tal and Arkadi Nemirovski, MPS-SIAM
Series on Optimization, 2001.
[5] COIN-OR LP, a simplex solver; https://projects.coin-or.org/Clp
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 219

CALBECK: UN SOFTWARE PARA CALIBRACIONES TOTALMENTE

AUTOMATIZADAS DE RELEVADORES DE PROTECCIÓN EN CENTRALES
GENERADORAS
Marco Vinicio González Gómez, Arturo Ahumada Zúñiga, Carlos Moreno Fierro, José Lizárraga González
Comisión Federal de Electricidad. Gerencia Regional de Producción Noroeste
Ángel Ernesto Gómez Sánchez, Daniel Iván Zamorano Acosta, Juan Alberto Castro Araujo
Instituto de Investigaciones Eléctricas. Gerencia de Supervisión de Procesos

RESUMEN

Derivado a los avances tecnológicos actuales que tienen
los equipos relevadores de protección multifunción
digitales así como de los modernos equipos de prueba
para comunicarse y ser controlados por medio de un
protocolo definido, es posible diseñar con las funciones
necesarias para crear una aplicación que se comunique a
los equipos para modificar la configuración, enviar
comandos y extraer datos de todos los dispositivos que
intervienen en el proceso de una evaluación operativa que
se le realiza a un relevador de protección. Este
documento presenta las características del software de
calibración automatizada de relevadores de protección
denominado CalBeck, el cual fue diseñado en lenguaje
grafico LabVIEW con el fin de optimizar las tareas y
reducir tiempos de ejecución en las evaluaciones
operativas a los relevadores de protección y utilizando las
macros de un sistemas de poder DOBLE F6150 con el
que se generaran las condiciones de falla para las
protecciones con las que cuentan estos equipos, limitando
los errores humanos en la captura de información y en la
programación de los equipos.
INTRODUCCION
La correcta operación de las funciones de protección
dentro del proceso de generación en la CFE, requiere de
equipos que se encuentren en óptimo estado de
operación y dentro de los rangos de exactitud
establecidos por el fabricante del equipo de protección
digital multifunción. En el método tradicional de
calibración, la captura de la información se hace en forma
manual, lo que lo hace una tarea meticulosa y expuesta a
errores por parte del personal que realiza la calibración.
Con la verificación de operación automatizada de los
relevadores de protección multifunción se aprovecha la
tecnología con que cuentan los equipos digitales para
realizar los comandos y extraer la información relativa al
sistema de protección.
El sistema de apoyo para la calibración de instrumentos de
protección se desarrolló para los relevadores de protección
de generadores y considera la verificación contra los errores
definidos por el fabricante del equipo. Utilizando como
plataforma de desarrollo LabVIEW se diseñaron las
funciones de calibración automatizada para los relevadores
de protección Beckwith M3425, Beckwith M3425A, Sel300G
y Sel387, el sistema se denomina “CalBeck” y se emplea
para obtener información de las lecturas de los diversos
parámetros a tomar en cuenta para calcular el porcentaje de
error de los equipos mencionados anteriormente, también
obtiene y almacena información de los relevadores en un
informe en formato Excel. El sistema configura la fuente de
alimentación para aplicarlo al relevador de protección
seleccionado, respalda la configuración actual del relevador,
configura al equipo para su correcta operación
(dependiendo de la función a verificar), aplica el
procedimiento de calibración automáticamente, se obtienen
las lecturas y finalmente genera un informe de calibración
en los formatos establecidos.
DESARROLLO
El sistema de calibración automatizada de relevadores de
protección CalBeck surge de la necesidad de facilitar el
trabajo rutinario del personal de protecciones, entre otras:
Automatizar las pruebas, eliminando en gran parte el error
humano que se presentaba en calibraciones manuales.
Permitir a los usuarios realizar otras actividades sin
necesidad de estar sujetos a estar realizando mediciones
cada cierto tiempo (durante la elaboración del presente
documento, un relevador de protección Beckwith M3425
se encontraba en proceso de calibración).
El software de calibración automatizada de los
relevadores de protección, es un sistema informático que
sustituye funcionalmente las tareas que se realizan de
manera manual, actividad en que se empleaban hasta dos
días por cada relevador. Actualmente con este desarrollo,
un relevador de protección se calibra en todas sus
funciones en aproximadamente 12 hrs (dependiendo de
las funciones existentes en el equipo).

El sistema está constituido por los siguientes elementos:

• Software de calibración.
• Equipo Doble F6150
• Relevador de Protección Beckwith o Schweitzer (SEL)

Figura 1. Esquema de conexión y arquitectura para el proceso de calibración automático.

220
Los componentes del sistema de calibración pueden ser
descritos como sigue:
La PC es en donde se encuentra el software desarrollado
para la calibración de los relevadores de protección
Beckwith ó SEL soportados; debe enviar de acuerdo a los
procedimientos de prueba del relé, datos específicos al
equipo de referencia (DOBLE F6150) para aplicar señales
analógicas de corriente y voltaje al equipo en prueba ó
sometido a calibración (SEL ó Beckwith). Al mismo tiempo
debe establecer comunicación vía RS232 con los equipos
bajo prueba para leer, modificar y restablecer la
configuración y parámetros de cada una de las funciones
que se calibran en el equipo. La PC es la encargada de
calcular el error en la prueba y de generar el reporte
correspondiente.
El equipo de referencia (DOBLE F6150), es el encargado
de generar voltajes y corrientes con las características de
amplitud, frecuencia y fase, necesarios para originar las
condiciones de operación del equipo en prueba, todos
estos datos son enviados por medio de un puerto RS232
y utilizando las DLL´s proporcionadas por el fabricante
para establecer los mecanismos de comunicación. El
equipo F6150 tiene además integrada una variedad de
algoritmos especializados para realizar pruebas sobre los
equipos, dichas algoritmos son llamados MACROS. Con
las MACROS se pueden hacer pruebas para determinar
el tiempo de operación (DELAY), el valor en el que entra
en operación la función (PICKUP).
Para los fines de calibración se aplican las siguientes
macros:
BSRHOI = BUSQUEDA BINARIA DE CORRIENTE
BSRHOV = BUSQUEDA BINARIA DE VOLTAJE
TIMEI = TIEMPO DE CORRIENTE
TIMEV = TIEMPO DE VOLTAJE
PHROTI = RAMPA DE FASE
SSIMULT= PRUEBA DE TRES ESTADOS
LRAMPF = RAMPA DE FRECUENCIA
Los equipos en prueba (SEL300G, SEL387, Beckwith
M3425 ó M3425A), son los que estarán sometidos al
proceso de calibración, a éstos se les respalda toda su
configuración interna en un archivo con extensión “CSV”.
La aplicación tiene la capacidad de leer y modificar los
parámetros de cada función y así calcular en función de
ellos los valores de las señales aplicadas por el equipo de
referencia para tener los puntos de operación programadas.
Se establece la comunicación a través de un puerto RS232
utilizando el protocolo propietario. Al terminar las pruebas
de calibración se les restablece la configuración original.
CARACTERISTICAS
Las principales características del software CalBeck se
describen a continuación.
Durante el proceso de calibración es necesario enviar datos
e instrucciones necesarias para controlar la operación de
cada uno de los equipos tanto para el equipo de referencia
como el equipo en prueba. Los mecanismos de
comunicación para cada equipo son diferentes, funciona de
manera distinta por lo que se realizan procesos separados y
diferentes para su comunicación.
El equipo DOBLE F6150 es en este caso el equipo de
referencia, es un simulador de sistemas de potencia que
cuenta con alta potencia, flexibilidad y un software para
correr pruebas de simulación sobre esquemas de protección
y relevadores tanto electromecánicos como los
microprocesados.
Los mecanismos de comunicación para el equipo Doble
F6150 se realiza a través de una DLL. La DLL recibe
información acerca del equipo y el puerto de
comunicaciones que se utilizará y con esta información
realiza el trámite con el sistema operativo para solicitar los
servicios de comunicación. Además también se le envía
datos de configuración como son el número, tipo, rango,
nombre de las fuentes.
Los relevadores de protección del Generador M-3425 y M-
3425A, SEL300G y de grupo SEL387 son unidades
microprocesadas que usa tecnología de procesamiento

digital de señales para proporcionar más de veinte
funciones de protección de generadores.
La nomenclatura sigue los estándares de ANSI/IEEE
C37.2-1991, Estándar de Números de Función de
Dispositivos de Sistema de Potencia Eléctrica. Los seis
contactos de entrada pueden ser programados para
bloquear cualquier función de relevador y/o disparar el
registrador de Oscilografía. Cualquiera de las funciones
o los contactos de entrada pueden ser programados
individualmente para activar uno o más de los ocho
contactos de salida programables.
El mecanismo de comunicación para estos equipos es a
través de un protocolo de comunicación documentado e
implementado por el fabricante, el protocolo es el BECO
2200 para los relés Beckwith y ASCII SEL para los relés
SEL con el cual se pueden ejecutar comandos tanto de
lectura como de escritura hacia el equipo. Para la
implementación de este mecanismo no se requiere el
uso de alguna DLL, por lo que el trámite de solicitud de
recursos con el sistema operativo para el manejo de un
puerto de comunicación se realiza directamente desde
la aplicación desarrollada.
Para la calibración se realizan una serie de pasos con los
que se pretende seguir correctamente los procedimientos
de prueba y calibración de CFE. En esencia estos pasos
pueden dividirse en 4 grupos de tareas
CONFIGURACIÓN DE EQUIPOS.
En esta parte del algoritmo se realiza la identificación de
los equipos tanto de referencia como del equipo en
prueba, estableciendo los mecanismos
comunicaciones necesarias. Durante esta etapa se realiza
una búsqueda del equipo de referencia y de prueba de
dos modos en forma manual y en forma automática. En la
forma automática, el usuario le indica al software en que
puerto se encuentra cada equipo para que realice la
búsqueda en esos puertos, en dado caso en que el
usuario seleccione el modo automático el software
realizara la búsqueda de un equipo en todos los puertos
disponibles hasta encontrarlo, si encuentra el equipo en
algún puerto quitara este puerto de la lista de búsqueda
para el siguiente equipo. En dado caso que no encuentre
el equipo informará que no se encuentra disponible y
terminará el programa.
RESPALDO DE PARÁMETROS DEL EQUIPO EN
PRUEBA.
Al encontrar los equipos en la búsqueda se establecen
los mecanismos de comunicación con el equipo a través
del sistema operativo y ya se está en posibilidad de
enviar la configuración al equipo de referencia. También
en este punto ya se tiene la comunicación con el equipo
en prueba lo que permite empezar a generar una
lectura de todos los parámetros del equipo y guardarlos
en un archivo de respaldo con extensión “CSV”. La ruta
del archivo se mantiene durante toda la ejecución del
X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 221
programa y al final es este mismo archivo que se utiliza para
restablecer la configuración del equipo. En dado caso que
exista un error en la ejecución del programa y no se
restablezca correctamente el equipo de prueba, existe la
opción de manera automática que se detecte que no se ha
restablecido correctamente y no se realiza un nuevo
respaldo, se solicita que se realice una restauración manual
solicitando el nombre y ubicación del archivo de respaldo al
terminar esto se sale del programa. También en esta parte
si la comunicación y el archivo de respaldo es correcto, se
modifica la configuración para apagar todas las funciones
disponibles ajenas a la función a probar en el equipo para
que en el proceso de calibración no influyan con disparo
erróneo. Solo se enciende la función a probar.
EJECUCIÓN DE LA PRUEBA DE CALIBRACIÓN.
Aquí se selecciona la función a calibrar del equipo Beckwith,
de esta se leen los parámetros de configuración. Con los
parámetros de configuración se calculan los valores de
prueba que serán aplicados por el equipo de referencia
F6150, se enciende la función a probar y se envía la
información al equipo de referencia de la macro a utilizar para
activar la macro de prueba, se espera el resultado de la
prueba y se calcula el error según criterios del fabricante para
determinar si el equipo pasa o no la prueba. Una vez
ejecutada la prueba y calculado el error se genera el reporte
de salida y luego se apaga la función seleccionada a fin de
que esta no opere en alguna otra prueba y garantizar que la
prueba que se realizo es únicamente de esta función.
Estos pasos se realizan en cada función a calibrar. Se pueden
enumerar de la siguiente forma:
 Selección de la función a calibrar.
de
 Lectura de los parámetros del equipo para la función
seleccionada.
 Se selecciona el tipo de prueba, PICKUP o DELAY.
 Se calculan los valores de prueba según los parámetros
de la función.
 Se envía estos valores de prueba a la macro a utilizar en
la calibración, esto es si la prueba es de PICKUP O
DELAY.
 Se enciende la funciona a calibrar y se hacer correr la
macro de prueba.
 Se calcula el error según el resultado obtenido y la
exactitud del fabricante.
 Se genera el reporte de la calibración y se apaga la
función seleccionada.
Existen 5 modos de ejecutar las pruebas, en estos modos
básicamente la diferencia es la ejecución de manera
automática. Los modos son los siguientes:
 Calibrar prueba: en este modo de operación solo se
ejecuta una de las pruebas seleccionadas en la
función que se está trabajando al colocarse sobre la
macro a utilizar, si no hay macro seleccionada no se
ejecutará la prueba y marcara un error en el modo.
Solo se puede ejecutar una prueba al dar inicio, no se
podrán correr más de una de manera consecutiva.
222

 Calibrar pruebas seleccionadas: en este modo

de operación solo se ejecutan todas las pruebas
seleccionadas de la función con la que se está
trabajando de manera consecutiva al dar una sola
vez inicio. Al iniciar la prueba se irán ejecutando de
una a una de manera consecutiva cada una de las
pruebas seleccionadas hasta concluir todas.
 Calibrar función: al seleccionar este modo de
Aquí se realiza el respaldo de la configuración completa del
operación de manera automática se cargaran
equipo en prueba, tiene una barra progresiva que muestra
todas las pruebas que la función con la que se
el avance en el respaldo. Un indicador del estado de
esté trabajando tenga y al dar inicio se ejecutara
conexión del equipo y otro para cuando se termino por
cada una de ellas de manera consecutiva hasta
completo el respaldo.
completar todas las pruebas.

 Calibrar funciones activas: al deshabilitar todas

las funciones se hace un monitoreo y se genera
una lista de las funciones que se encontraban
encendidas en el equipo y es esta lista la que se
ejecuta de manera consecutiva. Con este modo de
operación se puede calibrar un equipo con
solamente las funciones que el equipo está
utilizando. Aquí se realiza el apagado de cada una de las funciones
que están disponibles en el equipo, al mismo tiempo se
 Calibrar todas las funciones: en este modo de genera una lista de las funciones que la versión del equipo
operación se calibran en su totalidad el equipo, se tiene disponible, esta lista es utilizada para permitir
va probando cada una de las funciones con todas solamente la calibración de estas funciones. Tiene un
las pruebas que tenga de manera consecutiva indicador del progreso de la aplicación.
hasta terminar todas, no importa si se encontraba
encendida o apagada se calibrara indistintamente. INTERFAZ DE LA APLICACIÓN DE CALIBRACIÓN

RESTABLECER LA CONFIGURACIÓN DEL EQUIPO.

Al finalizar las pruebas de calibración y al salir de la

aplicación de calibración se restablece la configuración
completa del equipo esto es leyendo el archivo que se
creó al inicio con extensión “CSV” y escribiendo estos
datos en el equipo. El proceso de restauración de la
configuración también queda indicado en el LOG lo que
permite conocer si se restauro correctamente el equipo la
última vez que se genero el respaldo. En caso que el
programa haya sido terminado de forma abrupta o no se
haya restaurado en el momento que al iniciar el programa
se revisa y se pide al usuario que realice la restauración
solicitando que se le dé un archivo de restauración, el
software restaura el programa y sale para iniciar
normalmente.
INTERFAZ DE USUARIO
Se desarrolló una interfaz hombre máquina (IHM) para el
usuario a fin de facilitar la tarea de calibración, esta
interfaz se describe a continuación.
RESPALDAR CONFIGURACIÓN DEL BECKWITH Ó
SEL.
Este es el aspecto final de la internas del software de
calibración del cual se pueden distinguir diferentes regiones.
Del lado derecho de la pantalla se encuentra una serie de
indicadores que nos muestran en qué función se encuentra
en la interfaz del equipo en prueba, es decir se encuentra
de color verde el indicador correspondiente a la función
seleccionada. En la parte superior de estos indicadores se
encuentra un control que permite seleccionar una función al
usuario.
En la parte inferior del lado derecho se encuentra
indicadores de los puertos de comunicación utilizados por
cada uno de los equipos y un indicador luminoso que estará
en verde cuando la comunicación se encuentre concretada.
También hay una serie de controles que permiten
seleccionar el modo de operación de la aplicación, teniendo
5 opciones de selección.

En la parte inferior del lado derecho se encuentra tres

botones con los indicadores de INICIO, SALIR, ABORTAR
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
PRUEBA. Estos botones operan de manera general en
toda la aplicación. El botón “Inicio” inicia propiamente
dicho el proceso de calibración según el modo de
operación, habilita la función a calibrar y enciende la
macro de la prueba, si se desea abortar la prueba se
oprime el botón “Abortar Prueba” que estará habilitado
al momento de iniciar una prueba. El botón “Salir” es
para abandonar la aplicación y automáticamente se
inicia el proceso de restauración de la configuración del
equipo en prueba.
Del lado izquierdo se encuentran tres pestañas. La
primera pestaña muestra la interfaz para el equipo en
prueba Beckwith, la segunda pestaña muestra datos de
la configuración, la tercera muestra el Log de la
aplicación.
INTERFAZ DE PARÁMETROS PARA EL EQUIPO
BECKWITH.
.
Del lado derecho se muestra la interfaz de usuario del
equipo Beckwith, en la que se distingue cada una de los
parámetros que la función seleccionada tiene y una
pestaña para los resultados de las pruebas. Estos
indicadores son muy particulares de la función. Del lado
derecho de la interfaz del Beckwith se muestran un botón
con el numero “1” que habilita todos los controles de
parámetros para poder modificar y cambiar su valor, al
habilitar los controles de parámetros del equipo también
se habilita una serie de controles para la selección de las
pruebas a realizar, por cada prueba se presenta una
nueva pestaña con la macro a utilizar en la prueba
seleccionada. Esta interfaz muestra un indicador luminoso
en la parte superior izquierda que se pone de color verde
en cuanto termina de leer todo los parámetros de la
función. En la selección de las pruebas a realizar en la
calibración de la función se muestra una pestaña con los
parámetros de la macro a utilizar en la prueba
seleccionada, como se puede observar en la figura
siguiente.
En la pestaña de resultados se muestran las tablas con
los datos y resultados de cada una de las pruebas
realizadas, esta una vista preliminar de los datos que se
223
envían al reporte final. Cada tabla es diferente y dependen de
la función seleccionada.
RESTAURACIÓN DE CONFIGURACIÓN.
Cuando se termina de realizar todas la pruebas de
calibración y se desea salir del programa, se puede oprimir
el botón salir y automáticamente se restaura el equipo.
INSTALACIÓN DEL SISTEMA
El instalador del sistema CalBeck, se encuentra en un
paquete de instalación dentro del cual el ejecutable principal
es “Setup.exe” e incluye los archivos de soporte necesarios
para que el software funcione de manera correcta bajo las
plataformas Windows 2000, XP y Vista. Para Windos 7 se
requiere instalar un “parche” de National Instruments para
que reconozca los puertos seriales. Al ejecutar el archivo
“setup”, este a su vez ejecutará un asistente para la
instalación del software, el cual se encarga de crear las
carpetas de soporte y extraer los archivos ejecutables de
Calibra, así como los formatos de los informes de
calibración. Una vez finalizado el asistente de instalación del
software CalBeck, automáticamente iniciará también la
instalación del controlador de la llave electrónica
programable.
COMUNICACIONES Y PROTOCOLOS
La comunicación entre la computadora de aplicación y los
equipos digitales se realiza en protocolo propietario, por lo
que CFE tuvo que firmar documentos de no divulgación
de la información confidencial con los fabricantes.
CFE agradece a los fabricantes de los equipos
mencionados en este documento, así como el fabricante
del software de desarrollo (National Instruments) por el
apoyo para la realización del proyecto.
La comunicación se realiza a través de puertos seriales.
La integridad de los datos es verificada con cálculos
frecuentes de códigos de seguridad CRC´s.
CONCLUSIONES
224

egresado del ITESM campus

El proyecto “CalBeck”, ha demostrado que es posible Monterrey en 1983.
automatizar procesos de calibración que normalmente
son manuales, aprovechando la tecnología de
procesamiento y comunicaciones de los equipos
patrones, simuladores y relevadores de protección
digitales.

El trabajo de calibración completo, funciones básicas

activas que normalmente son 1 a 2 días se han
reducido a 4 horas (las funciones mas tardadas son
las de operación lenta y reset definido) además el
personal puede ocupar gran parte de esas 4 horas en
otras actividades. El resultado de las calibraciones
está libre de errores humanos al ser totalmente
automatizado y con rutinas de verificación que checan
que el relevador de protección es entregado en su
programación original. José Lizárraga González.
Trabaja para la Comisión Federal
No menos importante es la labor de equipo de trabajo de Electricidad desde el 08 de
de los autores del proyecto, tanto por parte de CFE- Julio de 2002. Es profesionista del
Generación como del IIE, que salvamos obstáculos departamento de protecciones de
técnicos importantes. la Central Hidroeléctrica Humaya
desde 2004. Apoya, supervisa y
REFERENCIAS realiza pruebas en campo de los
proyectos de Innovación
[1] Instruction Book M-3425 Generator Protection, Tecnológica de la especialidad.
Beckwith Electric Co. Inc. and BECO2200- M-3425 Es egresado con honores del
Communication Data Base for M-3425 Integrated Instituto Tecnológico de Culiacán
Protection System, Device I.D. = 45. en 2002.
[2] Serial Communication with Beckwith Electric’s
Integrated Protection System Relays.
[3] F6x Protocol Specification, Doble Eng. Company.
[4] FICDUseandDeployment.doc, F Series Daniel Iván Zamorano Acosta.
Components Detailed Design Specification. M.C. en Ingeniería Mecatrónica
Comunications DLL’s For Developers Use & egresado del CENIDET e
Deployment, Doble Engineering Company. Ingeniero en Electrónica, con
[5] FInstInter.chm, F Instrument Interface DLLs especialidad sistemas Digitales
Documentation, Doble Engineering Company. por el Instituto Tecnológico de Los
March 2007. Mochis. Trabajo en el Instituto de
[6] Instructivos de Operación Schweitzer (SEL300G, Investigaciones Eléctricas
SEL387), Doble Engineering Company. March desarrollando interfaces en
2007. LabVIEW de las macros (DLLs)
del equipo Doble F6150 para su
CURRICULUM integración al sistema de
calibración del equipo Beckwith
Marco V. González Gómez. M-3425 Y SEL300G.
Trabaja para la Comisión Federal
de Electricidad desde el 01 de
Enero de 1984 como ingeniero
especialista en protecciones,
medición y automatización de
Centrales Generadoras y
Subestaciones. Ha sido
encargado del proceso de
modernización y automatización
de protecciones y medición de
Centrales Hidroeléctricas de CFE
en el Noroeste. Coordina,
especifica y administra proyectos
de innovación tecnológica para el
Área de Protección y Medición.
Es ingeniero electricista
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Increased Sensitivity of Loss of Field Protection based on
Admittance Measurement
Hans-Joachim Herrmann, Siemens AG, E D EA PRO LM1, Nürnberg
Andre Smit, Siemens Energy, Wendell NC
Abstract-- This contribution focuses on loss of field protection
and introduces the admittance measuring technique. The
theoretical background is covered extensively to facilitate a
clear description of the factors associated with loss of field
(under-excitation). Apart from deriving the stability limits, the
transformation is described of the capability diagram to the
admittance plane and consequently the admittance and
impedance measuring techniques are compared. After
introducing the „Admittance measuring technique“ and typical
protection characteristics some practical applications are
described. These are primarily focused on the different
methods for deriving the setting values. The simplicity of
converting impedance setting values of existing relays to
admittance settings is also illustrated. Setting and test
recommendations are given. With examples from transient
tests with a RTDS system and real loss of field faults the
transient behavior and the reaction of the protection is
demonstrated.
KEYWORDS-- Generator protection, numerical protection,
loss of field, under-excitation, static stability, impedance
measurement, admittance measurement
I. INTRODUCTION
Under-excitation of a generator or a total loss of excitation can
result from a short circuit or open circuit in the excitation
circuit, a mal-operation of the automatic voltage regulator,
incorrect control of generators and transformers, or in the
event of a generator connected to a system with excessive
capacitive load. In this context under-excitation means that the
excitation of the synchronous machine is less than required for
stable operation at a particular power level. This excitation
limit determines the steady state stability characteristic of the
generator. If the excitation is not sufficient to provide the
power demanded of the generator, then this stability limit is
exceeded. The machine will slip and thereby obtain the
required excitation from the connected three phase system.
Depending on the construction of the generator, nature of the
excitation circuit, system conditions, the amount of supplied
power as well as the influence of voltage and power regulators
(AGC), rotor acceleration may result in local overheating in
rotor and stator, over-voltages on the rotor, mechanical impact
on the generator mountings and power swings in the three
225
phase system. In addition the large reactive power
consumption can trigger an wide area voltage collapse. To
prevent, or at least limit the duration of these harmful effects
loss of field protection (Siemens call it under-excitation
protection) is required to detect this under-excited condition,
and initiate timely disconnection of the machine.
The protection function may be implemented in different ways.
The impedance measurement [1, 2], is a widely applied
measurement principle. Amongst other reasons this technique
was chosen due to the widespread and proven use of
impedance measuring elements in electro-mechanical relays.
Approximately 40 years ago, Siemens however adopted a
different route [3, 4]. A solution was found, which on the one
hand may be directly derived from the capability diagram of
the generator and on the other hand was immune to
fluctuations of the generator voltage. This required a
transformation of the generator diagram to the admittance
plane and the processing of admittance measured values.
II. THE CAPABILITY DIAGRAM OF GENERATORS
To aid the following dissertation, the definition of the
fundamental electrical quantities is provided hereunder:
Sign convention: exported power (P, Q >0) is positive
(generator reference arrow-
system is used)
Apparent power: described by the symbol S and has the
dimension VA (kVA, MVA)
Active power: described by the symbol P and the dimension
W (kW, MW)
Reactive power: described by the symbol Q and the
dimension Var (kVar, MVar)
When per unit (p.u.) values are used, the generator nominal
values such as the nominal apparent power SN, nominal
voltage VN, and nominal current IN are used for the
conversion.
According to the Cartesian co-ordinate system definition (x-
axis = real component and y-axis = imaginary component), the
first quadrant defines the operating range (P > 0 and Q > 0) of
the generator. In the event of under-excitation (loss of field),
operation is in the 4th quadrant (P > 0 and Q < 0). Graphic
representation of the los of field protection takes place in the
4th quadrant. To avoid “neck strain”, the diagram is often
rotated to the left and mirrored in text books (refer to figure1).
The author adopts this form of representation.
226

In the case of an ideal turbo generator, the theoretical stability

limit is ϑ = 90°. Accordingly the limit value in the derived
representation is given by the direct axis reactance xd. For the
salient pole generator, this limit is dependent on the
reactance’s xd and xq, as well as excitation and the terminal
voltage. The base point on theoretical limit on the Q-axis is
determined by the quadrature axis reactance xq. The permitted
rotor angle ϑ (theoretical stability limit) is less than 90°. These
limits are graphically shown in figure 2 and 3 by means of the
Figure 1: Alternative representation of the capability diagram voltage and current vectors, as well as the capability diagram
for both machine types.
The stability limit is derived from the equations for the active
and reactive power of the machine [5, 6]. It is assumed that the
generator operates on an infinite network. The terminal voltage
of generator is equal the network voltage. Equation (1) and
equation (2) are the general defining equations and may be
used directly for the salient pole generator, which has different
direct axis and quadrature axis reactance. Due to the difference
in xd and xq a reluctance response circle with the diameter
xd - xq
V2 Figure 2: Vector and capability diagram of the turbo generator with xd =
xd xq xq (Eexc excitation voltage; I stator current).
( )
results. This circle indicates the steady state power that the
generator can produce with zero excitation (E = 0).

EV V2 xd - xq
P =3 sin ϑ + 3 sin2ϑ
xd 2 xd xq
(1)

EV V2 xd - xq
Q =3 cosϑ − 3 (1 + sin 2ϑ )
xd xd xq
(2)
Figure 3: Vector and capability diagram of the salient pole generator (xd ≠
with: E rotor voltage (field e.m.f.) as a phase to star point xq)
voltage in p.u.
V terminal voltage (a phase to star point voltage) of the From these diagrams it is apparent that the operating range of
generator in p.u. the generator is limited:
xd synchronous direct axis reactance in p.u.
xq synchronous quadrature axis reactance in p.u. In the over-excited range:
ϑ rotor angle (angular displacement) by the power supplied by the turbine and the excitation (rotor
values) and in the under-excited range:
Note: If the phase-to-phase voltage representation is used in by the power supplied by the turbine, the stator limits or the
equation (1) to (4) the factor 3 must be removed. stability limit.
For operation on an interconnected system, the actual
In the case of turbo generators the equations (1), (2) are (practical) stability limits apply. These take into consideration
simplified as the direct axis (xd) and quadrature axis (xq) the superseding reactances (e.g. unit transformer) which are
reactance are approximately the same. always present, and a security margin (approximately 10%
reserve at changing load). The actual admissible value for
stability is therefore smaller than the theoretical value. The
EV
P=3 sin ϑ manufacturer of the generator specifies the limits that must be
xd (3)
adhered to with the capability diagram. As in figure 1, various
representations and scales of the axes can be found. These may
have dimensions (MW, Mvar) or be dimensionless or per unit
EV V2 values (p.u.). The latter representation is preferred. If the
Q=3 cosϑ − 3 values in a dimensioned characteristic are divided by the
xd xd (4) nominal apparent power, the result is the p.u. representation.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 227

The following figures 4 and 5 provide an example of a turbo
generator and salient pole generator capability diagram.
Figure 4: Capability diagram of a turbo generator
Compared to the value at nominal voltage, the stability limit is
shifted to the right during under-voltage conditions and further
limits the amount of reactive power that may be imported. The
influence is proportional to the square of the voltage. The
over-voltage conditions are not critical as the stability limit is
shifted to the left in this case.
The foregoing explanations apply to slow variations of system
conditions. During sudden changes of load or system
conditions, transient quantities apply, and a transient response
will occur. Therefore a dynamic stability limit also exists. To
reach a simplified approximation, the transient values (x’d, x’q
and E‘) are applied to equation (1) and (2) [5]. In figure 6 the
basic result is shown. For this purpose it was assumed that the
steady state and transient quadrature axis reactance is the
same. From the diagram it is apparent that the machine may
even remain stable in the “dynamic” condition with a rotor
angle >90°. An analogy with the turbo generator can also be
found. In this case the dynamic stability limit is determined by
the transient direct axis reactance. In practice the limit is also
greater than 90° and is in the range between 110 and 120°.

Figure 5: capability diagram of a salient pole generator

In the case of the turbo generator the stability curve is shifted

to the right due to the superseding reactances, when compared Figure 6: Dynamic stability limit (salient pole generator)
to the theoretical curve. On the other hand, in the case of the
In figure 7 the statements regarding the limits during loss of
salient pole generator, the theoretical stability characteristic is
filed (under-excitation) are summarized. They are:
shifted to the right as a whole. The intersection with the
The practical (steady state) stability limit is to the right of the
reactive power axis is approximately at the centre of the
theoretical value and is given by the capability diagram of the
reluctance power circle, as the theoretical intersection (-V2/xq)
generator. It applies at nominal voltage.
is instable as a result of the missing directional force [5].
If the generator is operated with a voltage V < VN, the limit is
The representation in the capability diagram is applicable with
shifted to the right.
nominal voltage and current (VN, IN). Constant voltage may
To consider „dynamic conditions“, a dynamic stability limit is
however not always be assumed. The following computation
introduced. If it is exceeded, the machine must be
example illustrates the influence on the stability limits by
disconnected from the system immediately, as a pole slip will
variation of the voltage. The theoretical stability limit of the
most likely take place.
turbo generator (refer to figure 2) is used to illustrate the
influence of a 10% change in the voltage [7]. stability
If the excitation is equal to 0 the rotor voltage E=0. The theorectical
dynamic
practical
stability
limit at
V < VN
maximum reactive power that can be imported is Q = -V2/xd, stability
limit
limit
P [p.u.]
and reaches the following values: maximum
value of
theorectical
steady-state

V2 0,9 2 0.81
theorectical stability

At : V = 0.9 : Q = − =− =−
dynamic limit
stability

xd xd xd limit
under over
excited excited
2 2
V 1,1 1.21 Q [p.u.]
At : V = 1.1 : Q = − =− =−
xd xd xd V2
xd

V2
x’d
Figure 7: Summary of the statements regarding stability limits
228

For both generator types - salient pole and turbo generator - limit is exceeded. As a result of the slip an additional flux
the stability limits (steady-state and dynamic) are calculated in appears in the excitation circuit or an additional induced rotor
an example by using equation 1 and 2. voltage appears which attempts to maintain the armature
The following data are used: reaction of the machine at a constant level. This is apparent
from figure 10 where in the range between 90° and 180° the
Salient pole generator: excitation current increases significantly again. Only shortly
Generator terminal voltage V =1.0; before 180° is the rotor accelerated towards the stator pole
Rotor voltage E = in steps (0; 0.25; 0.5; 0.75; 1.0; 1.68) (zero load state with inverted pole). This large acceleration
Reactances: xd = 1.0; xq = 0.6; xd’ = 0.3; causes large slip and thus increased influence by the damper
windings. When ϑ = 180° the flux change and therefore the
Figure 8 shows the calculation results. The “half” circles (no longer measurable) rotor voltage becomes equal to zero.
represent the P= f (Q) for different angles (from 0° to 180°) As this takes place in an inductive circuit, the zero crossing of
and the selected rotor voltage E. The stability is reached at the the excitation current and the measured rotor voltage is
maximum of the active power (dP/dϑ = 0). The practical delayed. The rotor is now decelerated until it almost reaches
steady-state stability is calculated with a safety margin of 9 %. synchronous speed as the synchronising torque shortly after ϑ
For the dynamic stability limit the same calculation was done = 180° becomes very large. The result of this is a high torque
with the transient direct axis reactance. impulse that is also noticeable as a significant real power
Turbo generator: impulse (refer to Figure 10a). The mechanical power driving
Generator terminal voltage V =1.0; the machine is however too large to allow a recovery and the
Rotor voltage E = in steps (0; 0.4; 0.8; 1; 1.5; 2.51) machine will continue slipping. Between ϑ = 180° and 360°
Reactances: xd = 1.919; xq = 1.885 xd’ = 0.35; and also between n 180° and (n+1) 180° this sequence is
repeated. Some deviations apply during the transient state
Figure 9 shows the calculation results for turbo generator. before the steady state slip condition is reached. The reactive
power minimum following the first torque impulse therefore
has a different value compared to the following swing cycles.
Dynamic Theoretical Practical
stability limit steady-state steady-state
2.0
stability limit stability limit
2
The swing and transient conditions are particularly severe in
P[p.u]
P1
i
the salient pole machine due to the difference in direct axis and
P3
i
1.5
quadrature axis reactance. The slip changes dramatically
during one cycle.
P5
i
E=1,68
P6
i
1
P dy 1 E=1,0
i E’=0
P dy 2
i
E=0,5
Sthy 0.5
a

Spry
a

Stdy
a
E=0 Q[p.u]
3.5 3 2.5 2 1.5 1 0.5 0 0.5 1
cosφ 1/xd
E’=0,25
1/xq
0.5 0.5
1/x’d
3.5 Q 1 , Q 3 , Q 5 , Q 6 , Q dy 1 , Q dy2 , Sthx , Sprx , Stdyx , Q N 1
i i i i i i a a a

Figure 8: Calculated stability limit of a salient pole generator.

2
2.0

Dynamic Theoretical P[p.u]

stability limit steady-state
Practical
stability limit
steady-state
1.5 stability limit
P1
i

P3
i E=2,51
P6
E’=0
i

a)
1
P dy1
i

P dy2
i

Sthy
a

0.5
Spry
a

Stdy E=0,8
a

cosφ

3 2.5 2 1.5 1 0.5 0 0.5 1

E’=0,4
1/xd Q[p.u]
1/x’d
0.5 0.5

Figure 9: Calculated stability limit of a turbo generator.

The loss of synchronism by a salient pole generator is

illustrated in figure 10 [3]. In the diagram the increase of the
rotor angle following loss of excitation can be seen. Due to the
constant turbine power, the real power does not change, the
imported reactive power increases and the steady state stability
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 229

P Q
G= B=−
V2 V2 (8)

The values from the axis in the generator capability diagram

must simply be divided by the square of the voltage. If
subsequently the sign of the reactive component is inverted,
the transformation is complete. When V = VN = 1, the per unit
numerical values in the capability diagram are identical with
those in the admittance diagram (refer to figure 11). From the
per unit capability diagram it is therefore possible to directly
derive the setting values of the loss of field protection.

b)
Figure 10: Loss of synchronism condition on a 30-MVA-salient pole
generator (caused by rapid loss of excitation) with P= 0.8, ie =1.3*ie0 [3]
[a) course of power flow; b) excitation signals (ie =excitation current, ie0=
zero load nominal excitation current, ve = excitation voltage, ve0 = zero load
excitation nominal voltage]
Figure 11: Capability diagram of the generator and admittance diagram

III. ADMITTANCE MEASURING PRINCIPLE The protection measuring algorithms are based on the
equations (5) and (8). The vector signals are derived from the
As mentioned in the introduction, the transformation of the sampled instantaneous values in the 3 phase to ground voltages
capability diagram into the admittance plane has the distinct and the 3 phase currents. The positive sequence components
advantage that when using the p.u. representation, a direct are calculated from these vectors. According to the definition
reference to the generator capability diagram is provided, that in equation (5), the positive sequence voltage and current
is independent of the actual generator voltage. components are employed to calculate the active and reactive
The transformation is explained by the following equation: power. Division by the positive sequence voltage V1
according to equation (8), results in the transformation from
The equations (5) and (6) describe the fundamental definition the power plane into the admittance plane. To avoid over
of the complex power and the complex admittance. functions in the case short circuits close to the generator
terminals an undervoltage blocking is implemented. The
S = V ∗ I* S = P + jQ (5) threshold is fixed at V1 = 25 % of the nominal voltage.
Figure 7 can be used to extract the characteristics required by
the protection functions. The given static stability limit must
I be monitored. Generally 2 lines are sufficient for this purpose.
Y= Y =G + jB
V An additional threshold value which depends on the dynamic
(6)
with: Y admittance stability limit applies. From the area of extreme under-
G conductance (real component of the admittance) excitation (on left of char. 3), it is highly unlikely that the
B Susceptance (reactive component of the admittance) machine will recover to the stable operating range. Fast
tripping is therefore required in this case. This is different if
The relationship for the transformation can be derived by the static stability limit is exceeded (char. 1 and 2 in figure
multiplying in equation (6) with the conjugated complex 12). In this case, if the excitation voltage is still sufficiently
large, a recovery by the machine to the stable operating range
voltage.
is not inconceivable. The monitoring of the excitation voltage
(Vexc <) is therefore introduced as an additional criterion.
I ⋅ V* S* P − j Q P Q This criterion controls the tripping time of char. 1 and 2. In
Y= = = = 2 −j 2
V⋅V *
V 2
V 2
V V this way, over functions as result of transient transgression of
(7)
the static stability limit due to dynamic impulses that are
followed by a recovery to statically stable conditions, are
Comparing the coefficient in equation (6) and (7) it results prevented.
in the definition of the admittance values. To set these characteristics, the setting parameters that consist
of the intersection with the B-axis as base point of the line, and
the inclination, are applied. The angle of inclination may be
derived from the generator capability diagram (refer also to
230
section 5). The setting should be such that it is close to the
given stability characteristic. The excitation circuit controller
characteristic supersedes this.
The setting threshold of excitation voltage supervision is
approximately 50% of the no-load excitation voltage (Vexc0).
Figure 12: Characteristic of the admittance protection (turbo generator)
[8](At salient pole generators char.1 is approximately 1/xd +1/2(1/xq -1/xd)
and char.2 is approximately 1/xd with α2 = 100°)
The protection response resulting from the characteristics in
figure 12 is the following. To avoid over functions by the loss
of field protection due to transient phenomena (e.g. out of step
conditions) a time delay is recommend.
characteristic 1, 2 exceeded, excitation voltage monitoring
(Vexc <) not picked up
This condition must be alarmed, and when configured,
tripping with long time delay (approx. after 10 s) is carried
out.
characteristic 1, 2 exceeded, excitation voltage monitoring
(Vexc <) picked up
For this condition, tripping with short time delay (approx.
0.5 s to 1.5s) is required.
characteristic 3 exceeded
For this condition, tripping with short (0.3 s), or no time
delay (after transient studies) is required.
Figure 12 shows additional the characteristic of the
minimum excitation limiter (MEL), which is a control
function. The MEL acts to limit the reactive power (Var) flow
into the generator. When the reactive power flow into the
generator excesses the MEL setpoint, the MEL becomes active
to increase terminal voltage which reduces the reactive power
in-flow. The voltage increase continues until the reactive
power flow is reduced below the MEL setting [9]. The loss of
field protection must be coordinated with the MEL. The MEL
realization should also consider the voltage behavior due to a
reduced terminal voltage (see figure 7).
IV. COMPARISON WITH THE IMPEDANCE MEASURING
As stated in the introduction, the impedance measuring
principle is widely applied [1, 2]. To obtain the setting
parameters for the impedance principle, the generator
capability diagram must be transformed (mapped) into the
impedance plane. This transformation mathematically
corresponds to the mathematical inversion of a locus.
Consequently, the direct reference to the capability diagram is
lost. In accordance with the locus theory, (a line that just
misses the origin becomes when inverted a circle that touches
the origin), the transformation of the theoretical stability
characteristic to the impedance plane results in the section of a
circle. This transformation is illustrated in Figure 13. All
points to the left of the stability limit in the capability diagram
are located inside the semi-circle (shaded area) in the
impedance plane.
Figure 13: Transformation in the impedance plane
Conversely, the characteristics in accordance with IEEE [1]
can be converted to the per unit generator capability diagram.
The following rule must be observed here: circles which do
not pass through the origin, will again give circles when
inverted. Figure 14 shows the transformation. It is apparent
that in comparison to the admittance principle (refer to figure
12) this provides a much rougher approximation of the
stability characteristic. This measuring principle cannot detect
if the stability limit provided by the generator manufacturer is
continuously exceeded by a small amount. The control system
(MEL) or the operating personnel must be relied upon to
detect such failures. Alternatively, additional monitoring (e.g.
additional impedance circle) must be provided.
The greater margin is however of advantage during dynamic
situations (transient transgression of the stability
characteristic). Such incidences do not result in pick up, or
only result in transient pick up, by the protection. Additionally,
the significant points, as well as the rules for the
transformation are indicated in figure 14.
Figure 13: Transformation in the impedance plane
In figure 15 a summarized comparison of the admittance and
impedance measuring principle is shown for a turbo generator.
The typical setting recommendations were considered for this
purpose (refer to figures 12 and 14). For example, in the case
of under-excitation, the admittance measurement can, as a
result of it’s more accurate match to the static stability
characteristic, provide an early alarm. This may be seen from
the indicated trajectory in the event of under-excitation.
Furthermore figure 15 shows that the two measuring
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 231

techniques are largely similar. The significant differences may

be found in the thresholds. In figure 16 a practical example of the parameter conversion
is shown. The left hand section of the diagram indicates the
Impedance principle Admittance principle previous characteristic and possible setting values of the
Admittance Plane
protection that is to be replaced. The right hand section shows
4
the setting table of the numerical protection with the converted
parameters. With the conversion equation (9), the reactances
2
are converted to per unit values. Furthermore, the excitation
voltage monitoring is not used. The three characteristics of
Figure 12 are therefore applied separately. Characteristic 1 is
0 used for alarm purposes. With the angle of 80°, a good
estimate of the stability limit is achieved. The time delay for
the alarm was set to 10 s. Additional an alarm signal is
automatically generated if the admittance is over the
2

characteristics (OR logic) and stored in the puffer named “trip

4
8 7 6 5 4 3 2 1 0 1
log”. Characteristic 2 provides the replica of the larger
impedance circle of the previous protection and trips with
short time delay. The converted susceptance value for
2/X’d = 7,4 1/Xd = 0,55

characteristic 2 is 0.51 and a time delay of 1 s was selected. To

Figure 15: Comparison of the impedance and admittance measurement loci in
achieve a better match to the circle, the characteristic is
per unit – capability diagram (xd = 1.81; x’d = 0.27)
slightly tilted to the left (refer to figure 15), and a setting of
100° was chosen.
With these setting parameters, characteristic 3 corresponds to
V. APPLICATION the inner circle.
In this section, the following question is addressed: how are
the setting values derived?

The question is essentially answered in sections 2 to 4.

Depending on the available information, the different methods
for obtaining the setting parameters are employed:

a) If the generator capability diagram is available, the

setting parameters may be derived directly there from
(per unit representation required). Settings for
characteristic 3 are derived according to figure 12. Figure 16: Conversion example: impedance to admittance values

b) If the direct axis reactance of the generator is known,

the setting values may be directly obtained from the VI. TESTING OF THE CHARACTERISTIC
recommendations given in figure 12. For the slopes,
the indicated angles must be used. This The main objective of the test is the verification of the settings.
As the characteristics are made up of straight lines, two test
recommendation is in accordance with the IEEE
points result: the base point (BP) of the lines (I) and a further
recommendation. [1]. point on the line (II). The test is carried out with the nominal
voltage. As test variable, the current is changed in amplitude
c) If the protection is replaced in the course of a and phase.
protection refurbishment, the previously used During the test to establish the base point, the current leads the
impedance settings can naturally be converted to voltage by 90° (capacitive current). The value of the current is
admittance settings. Assuming secondary setting derived by multiplying the set value with the nominal
values, the following equation provides the secondary current (example: Char. 1: 0.56 * 5 A = 2.8 A). The
conversion to secondary per unit admittance values. injected test current is therefore 2.8 A with a phase angle of +
90°.
1 VN, Sec The test of the second point (II) is preferably carried out at an
= angle of + 45°. The injected test current is derived from the
x d,sec 3 I N, Sec ⋅ X old, Setting intersection of the tested characteristic with the line along
(9) which is tested. The test current is calculated by defining the
equation of the two lines, and equating the intersection.
with: VN,Sec secondary nominal Alternatively, the intersection can be derived graphically.
voltage (e.g. 120V) Figure 16 shows the results in a graphic. The line equations for
IN,Sec secondary nominal current (e.g. computation of the intersection are also shown. On the right
5A) side of the figure the test voltage and current are shown for the
Xold,Setting „old“ previous setting in Ohm two test points (I, II). Drawn is only one phase of the three
232

phase system. The current is leading (capacitive). For testing
the behaviour during an undervoltage situation (shift of the
characteristic to right (see figure 7)) the test voltage and
current must be reduced by the same ratio. Shall be the test
done with 90% of the rated voltage, than the voltage and
current for the 100% case must be multiplied by the factor 0.9
(e.g. test point (I): V = 0.9 * 120V/√3 = 108V/√3 and I = 0.9
* 2.8 A = 2.52 A).
Figure 17: Testing of the loss of field characteristic
To get from the users practical loss of field failures is very
seldom. Figure 20 shows such a record and the analysis with
the graphic tool SIGRA. During the commissioning of a pump
storage station it was not possible to open the generator circuit
breaker. The generator was in the pump mode (motor
operation). By the operator the field breaker was opened. On
the left side of the figure the phase voltage and current is
shown in RMS values. Below are the binary traces. The pickup
and the trip event are recorded. The event “Exc<U< Trip”
means that the trajectory is over characteristic 1 and additional
the field voltage supervision (Vexc<) picked up. The time
delay was set 1 sec. The loss of excitation voltage can be seen
by trace TD3. The right side of the figure shows the calculated
impedance and power. The cross in the P, Q plane marks the
trip by loss of field protection. While the generator CB was out
of service the breaker failure protection becomes active and
tripped the high voltage CB after a delay of 0.5 sec. The open
high voltage CB can be seen by the interrupted phase current.

VII. TRANSIENT BEHAVIOR OF LOSS OF FIELD PROTECTION

The transient behavior of the protection can only be shown
with real failures or with dynamic simulations. For dynamic
simulation two methods are used:

• Real time digital simulator (RTDS)

The generator and a network is simulated. All
numerical calculations are done in real time. Via
amplifiers the transient data can be injected into the
relay directly. Closed loop test are possible.

• Transient network calculations program

The generator and a network is simulated. The Figure 18: Transient test with RTDS (char 1: 0.55∠80° 10s; char 2: 0.51∠90°
transient calculations are done offline. The result of 10s; char. 1.1∠110° 0s)
the simulation can be stored in a comtrade file. With
this file the protection function can be test via digital 150
150

test equipment. In this case only an open loop test is

possible.
100

A 200 MVA generator with a step-up transformer and network

G1
i

Ch1 ( l )
Load
was simulated with a RTDS system. The protection measures Ch2 ( l )
point
the phase current from the star point side and the phase to
Ch3 ( m )

50

ground voltage from the terminal of the generator. A loss of

field situation was initiated by a voltage regulator failure
(voltage jump from V = 1.08 to 0.8). The generator operated 0 0
250 200 150 100 50 0 50

under full load condition (P = 160 MW, Q = 25 MVar; field 237.814 B1 , l , l , m

i
12.87

current if = 1.87 If0). The loss of field protection tripped with

Scaling in percent - related to primary values X calculation every 50ms

characteristic 3, because for the characteristic 1 and 2 a time Figure 19: Calculated admittances
delay of 10 s was set. Figure 18 shows the stored fault record
from the device. To get a better overview only one phase (B =
L2) plus the important binary traces are drawn. A reduction in
the voltage and increased phase current can be seen. The
binary traces show the pickup (trajectory is over char. 1) and
the trip via char. 3 (Exc<3 TRIP). In addition figure 19 shows
the calculated admittance and the used characteristic 1 and 3.
The time between the crosses is 50ms. The longer distance
indicates an increasing of the slip, which can be clear seen if
the trajectory crosses char. 3 (left straight line).
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 233

[6] Weßnigk, K.-D.: Kraftwerkselektrotechnik (power

X, R station electronics*). vde verlag gmbh, Berlin, Offenbach
1993, ISBN 3-8007-1724-7
[7] Born,E.; Fischer, A.:Elektronischer
Untererregungsschutz (electronic under-excitation
protection*). Siemens-magazine. (1972) paper.12; p. 912
–915
P, Q [8] Multifunction Generator, Motor and Transformer
Protection Relay 7UM62. (2008) Siemens
0,5 s by 50 BF
Manual, order No. C53000-G1176-C149-6
[9] Reimert, D.: Protective Relaying for Power
Generation Systems. Taylor @Francis, ISBN 0-8247-
0700-1, 2006

Figure 20: Trajectory during a real loss of field failure.

VIII. CONCLUSION
The static stability limit is given by the generator capability
diagram. This limit is closely matched by the admittance
measurement. It correctly takes into consideration the
influence of the generator voltage on the capability diagram.
The loss of field of both basic generator types, salient pole and
turbo generator, was explained at length. The fundamentals of
the stability limit and the signals that influence it were
discussed. Subsequently the transformation from the capability
diagram to the admittance plane, admittance measuring
technique and implementation of the under-excitation
protection function, was looked at. Furthermore the differences
relative to the impedance measurement and the transformation
of the capability diagram to the impedance plane were dealt
with. The section “Applications” was dedicated to the
calculation of the setting values and showed the different
solutions. The scope of testing is reduced to a minimum by
employing numerical technology. A method for the
verification of characteristic setting values was shown.
Additionally the transient performance of the loss field
protection was briefly demonstrated with two examples.

BIBLIOGRAPHY

[1] IEEE Guide for AC Generator Protection. IEEE Std.

C37.102 – 1995,
Approved 12 December 1995, ISBN 1-55937-711-9
[2] IEEE Tutorial on the Protection of Synchronous
Generators. (1995) IEEE Catalogue Number:
95 TP 102
[3] Fischer, A., Zurowski, E: „Neuartiger
Untererregungsschutz“ (New type of under-excitation
protection*)Siemens magazine , (1966) paper.8, p. 634 –
640
[4] Untererregungsschutz (under-excitation protection*)
RG66, product pamphlet of Siemens, 1967
[5] Bonfert,K : Betriebsverhalten der Synchronmaschine
(synchronous machine operational response*). Berlin,
Göttingen, Heidelberg, Springer-Verlag 1962
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 235

NEW ALGORITHM FOR TRANSFORMER DIFERENTIAL PROTECTION THAT WILL

DETECT, TURN TO TURN FAULTS IN WINDINGS TRANSFORMERS IN A SECURE
MANNER AND FAULTS WITHIN THE DIFERENTIAL PROTECTION ZONE IN
TRANSFORMERS AND GENERATORS

Rafael Alberto Cordova Cruz, Comision Federal de Electricidad (CFE),

Northwest Transmission Region rafaelcordovacruz@gmail.com

Summary: Currently the detection of
turn to turn faults in Transformers are
hard to detect because transformer
differential relays are set to detect faults
in the Differential zone using positive
sequence components. If the negative
sequence component is added to the
differential protection, turn to turn faults
can be detected in a secure way. This
report introduces a new algorithm for
transformer differential protection,
which will guarantee the detection of
both, turn to turn faults and faults within
the zone of protection. This algorithm
works in transformers and generators.
The oil samples taken from the top and
bottom of the tank during testing were not
conclusive to assure that this unit had any
internal problems since at the moment of
the trip, the unit only had 20 MW. In
addition to the low load condition, the
ambient temperature was mild and this did
not allow the oil to be uniform inside the
tank.
And in this way, with the analysis of the
event records by protection personnel, and
tests performed by substation personnel
and chemical engineer, it was not clear that
the A phase unit was damaged, so the
question was still in the air…..

On August 28 at 13:39 hrs (2008), a fault Is the equipment in good working

occurred on the high side winding of phase conditions to be energized?
A of the 230/115kV 100 MVA
Autotransformer located at Hermosillo
Cinco station (HLI). Transformer
differential relays operated (87T), tripping
breakers HLI-92020 and HLI-72020. CFE
Protection and Substation personnel went
to the station to restore service and assess
the damage.

When analyzing the event records from

Figure 1 Current phasors from the fault
87T relay, an internal fault was not clearly
defined in the oscillography because the
How can we validate that there is really
phase angle between the high side and low
a fault in the differential zone with
side phasors only changed a few degrees
phasor information that is not very
from what the phasors would look like
convincing?
under normal operation of the
autotransformer. So it was left to the
The analysis of the negative sequence is a
substation personnel to test the
good tool to determine without doubt that
autotransformer and determine if there was
there is really a problem within the
an internal fault in the unit.
differential zone of protection.
236

Negative sequence current will only be

present during fault conditions or in small
quantities during normal operation
conditions, and delta – wye transformer
faults can be analyzed regardless of there
configuration. For example, if a fault occurs
on the low side winding of a distribution
transformer which is connected delta on
the high side and wye on the low side, the Figure 2 High side winding currents from 87T relay
negative sequence current per unit (p.u.) is
the same on the high and low side
windings. This is not true for the zero
sequence currents, which will only be
present on the low side windings.

In order to do an analysis with negative

sequence currents of the autotransformer
fault from HLI station, it was necessary to
create a protection model so that with the
event records from the transformer
differential relay and/or comtrade files from
a digital fault recorder, it could be
determined that there was a fault inside the
differential zone.
The protection model was done with
Mathcad software, and was validated with
software from the transformer differential
relay vendor.
Figure 3 High side winding currents from model and
relay
Data points from both, the relay and
Mathcad, coincide. In Figure 3, the series
of data that show on the chart with dotted
points is taken from the relay. The series of
data in the smooth line is from Mathcad
model. The information taken from the low
side current transformers of the
Autotransformer also match the protection
model.

The analysis of the validation of the The current angle of the negative
protection model, which will be sequence, before and during the fault, is
demonstrated below, will focus only on the shown on Figure 4.
magnitudes and angles of the negative
sequence currents before, during and after
the fault (validation of protection model was
done with all three sequence components).

The initial validation of the model was to

ensure that the event records from 87T
relays match the protection model as
figures 2 & 3 show.

The sample rate of the oscillography is 240 Figure 4 IW12 and IW22 angles values a) before
Hz or 4 samples per cycle. fault b) during fault

Figures 5, 6, 7 and 8 show the negative

sequence current angles from the High and
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Low side windings from the Mathcad
model.
Figure 5 High side negative sequence current
angles vs. High side current angle
Figure 6 Low side negative sequence current
angles vs. Low side current angle
Figure 7 High side negative sequence current
angles vs. High side current angle
Figure 8 Low side negative sequence current
angles vs. Low side current angle
237
With the above charts, the protection model
is validated.
What algorithm model will be used to
detect turn to turn faults in the
differential zone?
Protection Engineers do not have a clear
way to identify a turn to turn fault because
of the way the phasor currents move during
these types of faults. The model that was
developed includes the negative sequence
currents according to Eq.1 and Eq.2.
IA _ H 2
I 22 = Eq. 1
IA _ X 2
IA _ H 1
I11 = Eq. 2
IA _ X 1
Where:
I22 is the result obtained from the relation
of two negative sequence phasors
IA_H2 is the negative sequence phasor of
high side winding
IA_X2 is the negative sequence phasor of
low side winding
I11 is the result obtained from the relation
of two positive sequence phasors
IA_H1 is the positive negative sequence
phasor of high side winding
IA_X1 is the positive negative sequence
phasor of low side winding
When a fault outside the differential zone is
simulated, the high side and low side
negative sequence phasors are equal and
in the opposite direction (same magnitude,
180 degrees apart), so the relation of the
negative sequence currents (I22) is 1 for
external faults and different than 1 for
internal faults. I11 is 1 just before the fault
is present, because there is only positive
sequence under normal operating
conditions.
238

In reference to the above, Figure 9 shows

phasor I22 during the fault.

Figure 10 Positive and Negative Sequence relation

for turn to turn fault
Figure 9 Phasor I22 magnitude for turn to turn fault
An equation for a negative sequence
differential trip could be obtained
In Figure 9, you can see that the relation
considering the following:
between the high side and low side
negative sequence currents is close to 1
before the fault, but the value changes to a • Adequate level of negative
magnitude higher than 1 when the negative sequence differential current
sequence phasors drift apart because of • Relation of High/Low positive
the internal turn to turn fault. sequence currents
• Level of high side negative
Before doing the relation of the negative sequence currents
sequence currents, it’s important to do the • Level of low side negative sequence
current match between the high and low currents
side of the autotransformer; this will help to
do the adjustments for the settings of the With the above, the trip of the relay model
negative sequence differential. is shown in Figure 11.

The autotransformer CT ratios for the

differential protection are: High side –
400/5, Low side – 600/5. Because the
Mathcad and Matlab/Simulink model was
done using high side values, a multiplier of
0.75 was used for the low side values to
match both currents.

The relation between the high and low side

positive sequence currents (I11) also have
a value of 1 before the fault, because
during this time, the autotransformer only
had positive sequence current flowing
through him. Figure 10 shows the positive
and negative sequence relations.
Figure 11 Trip for negative sequence differential
In Figure 9 the fault was detected in 5.25
cycles, so we only need to wait for the
other requirements to come true to allow
the trip, this happens in cycle 6, which
guarantees a fast trip for a turn to turn fault
on an autotransformer.

It is true that a small current difference

between high side current phasor IAW1
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 239

and low side current phasor IAW2 can be

seen in Figure 1, but a greater magnitude
difference is more obvious in the negative
sequence phasors as shown in Figure 12,
this is also characteristic of the behavior of
the negative sequence differential.

Figure 13 Phase A currents from simulation

Figure 14 shows phasor I22 values. This

figure demonstrates that for through faults,
I22 has a value of 1 and a value different
than 1 when the fault is in the differential
zone.
Figure 12 Negative Sequence current phasors
Note that at the start and end of the
A model was generated in ATPDraw that
simulation (blue rectangles) there are
simulates the conditions when the fault
instable values for I22; this is because at
occurred, to verify the effectiveness of the
the start of the fault, the electrical grid is
algorithm for faults within the differential
balanced so the negative sequence value
zone and for through faults.
is very small. And at the end of the
simulation the instability is present because
the opening of the high and low side
The ATPDraw model simulated bolted
breakers was not simulated.
through faults and faults in the differential
zones on phase A in the following
sequence:

• Initial fault, through fault on low side

• 20ms later, low side fault in
differential zone
• 20ms later, through fault on high
side
• 20ms later, high side fault in Figure 14 Phasor I22 magnitudes
differential zone
IAW1 and IAW2 have angles close to 0 and
Once the simulation was completed, the 180 degrees for through faults as shown in
results from ATPDraw were reprocessed at Fig 15.
Mathcad model. In Mathcad the information
was filtered by a cosine complete cycle
algorithm to eliminate signals that were of
no interest for the analysis (signals greater
than 60 Hz).

Figure 13 shows fault currents on phase A.

Figure 15 Phase angles of IAW1 and IAW2
currents
240
The angle that results from the difference
of the High side and Low side negative
sequence phasors show a similar behavior.
This is shown in Fig 16.
Figure 16 High side and Low side Negative
Sequence angle difference
Primary current values of the negative
sequence differential are shown in Fig 17.
The blue rectangular boxes represent
through fault simulations.
Figure 17 Negative sequence differential current in
primary values
When through faults were simulated, the
negative sequence phasor magnitudes are
equal and in opposite direction, thus the
negative sequence differential does not
measure any difference between them as
shown in Figure 18.
Figure 18 Negative sequence differential phasors
during through fault
Figure 19 shows phasor I22 with a value of
1 during an external fault.
Figure 19 Phasor I22 during an external fault
How works this algorithm with a real
insulation generator fault?
On July 20, 2008 a winding failure occurred
on the 375 MVA Generator 2 of
Hydroelectric generating plant ‘El Cajon’,
which is located in Nayarit, Mexico; this
failure caused the unit to trip offline [4].
Figure 1 show the generator terminal
voltages, in which the voltage sag on the
failed phase and the voltage swell on the
other two phases can be seen, due to the
fact that the generator neutral is connected
thru a 200 KVA, 17/0.24 kV transformer
with 48 mΩ resistive value on its secondary
winding. During the fault the neutral voltage
appears. Also, it is shown that it took 325.6
ms to trip the generator by the 64G
protection scheme.
Figure 20 Voltage chart at generator terminals
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Figure 21 shows all six currents which do
not seem to register any change due to this
failure.
Figure 21 Generator’s current chart
Figure 22 shows the relationship between
the positive sequence current phasors of
both current transformer sets for the fault,
this value is close to 1.
Figura 22 Relationship between positive sequence
currents
By zooming into Figure 22, the variation
between the positive sequence phasor
currents can be appreciated.
Figura 23 Relationship of positive sequence phasor
currents shown at a different scale (zoom in)
241
As it can be seen in Figures 22 & 23, the
positive sequence relationship value has
little change during this type of fault. Is this
small change sufficient to detect the fault
by the relay?
Figure 24 depicts the relationship between
the negative sequence current phasors.
Figura 24 Relationship between negative sequence
phasor currents
In Figure 24, the relationship is held to a
value close to 1 before the fault occurred,
which indicates that there was negative
sequence present at both sets of the
generator’s current transformers.
Figure 25, shows the angular movement of
the relationship between the negative
sequence currents. In sample 182 the
angle value is 180 deg and in sample 183,
the value is -179.6 deg.
Figura 25 Angular movement of the relationship
between negative sequence phasor currents
242

The sudden change in angle shown in Figure 27 shows an amplified view of

Figure 25, is an indicative that there is Figure 26.
some fault present in the generator
windings, because according to Figures 15,
16 & 19, when a fault is outside of the
differential zone there is no angular
movement between the relationship of
negative sequence current phasors, unless
the fault is inside the protected zone.

Also, as it can be seen, the currents in

Figure 21 do not show an increase in value
which is an indicative that this is a high
impedance fault in the windings.

Figure 26 shows an amplified view of the

angular movement so it can be compared
in the same chart with the generator neutral Figura 27 Amplified view in which the angular
movement can be seen to obtain the trip signal for
voltage. the negative sequence protection

Fault in Generator 2 of Hydroelectric power

plant ‘El Cajon’, can be seen in Annex 2
[4]. The failure was located in slot 438
(without bracing) of branch 3B4.

In this work was used In Mathcad and

Figura 26 Angular movement of the relationship
between negative sequence phasor currents and
Vng
Matlab/Simulink model the information
filtered by a cosine complete cycle
algorithm but can be used the half cosine
cycle algorithm to detect the fault faster
than cosine complete cycle algorithm in
winding generator, because this a fault with
high resistance, and the half cosine cycle
algorithm is stable with this condition.

In Figure 26 it can be seen that the Figure 28 shows current fault information
algorithm has detected and is ready to with low resistance and Figure 29 shows
send a trip signal in less than a 1 cycle the same fault with high resistance with
from when the fault occurred and it does cosine complete cycle algorithm and half
not have to wait 325.6 ms that it takes with cosine cycle algorithm outputs.
the traditional protection method that is
used in almost all power plants for
insulation failures. It fits to mention that the
differential relays did not operate for this
fault, due that they only work with positive
sequence phasors.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 243

The Annex 1 shows the pictures after

inspection of the transformer.

The Annex 2 shows the pictures after

inspection of the generator.

This algorithm was realized one day after

the event of autotransformer (August
29/2008).

Figura 28 Full cosine algorithm output and half CONCLUSIONS:

cosine algorithm output with low resistance fault
• The proposed algorithm will detect in
a secure manner turn to turn faults
in transformers and it also provides
an excellent way to detect faults in
the transformer differential zone.

• The negative sequence information

is an excellent tool for analyzing
faults because it will only be present
during disturbances in the electrical
grid.

• If the Protection Engineer uses

information from the negative
Figura 29 Full cosine algorithm output and half sequence phasors when analyzing
cosine algorithm output with high resistance fault transformer faults, he can determine
without doubt that the event was
caused by either a turn to turn,
A great Thank You to Ing. Pedro Alonso internal differential or through fault.
Verdugo Ibarra from Comisión Federal de Thus he can make a quick and
Electricidad for his contribution to this sound decision to replace the failed
paper with the ATPDraw software unit, normalizing the operation of the
simulation (autotransformer). station and maintaining the reliable
and safe electrical grid.
And to Ing. Eduardo Guerrero Flores from
Comisión Federal de Electricidad for the • This algorithm also applies to protect
pictures of generator´s damage. generators in a secure manner with
a fast trip because sensitivity and
And to Ing. Gerardo R Trevino from speed is gained, when compared
American Electric Power for translating this with the current traditional
paper. protections methods used in power
plants.
244

• Traditional positive sequence Biography

differential relays do not detect this
type of faults, thus it is necessary to
wait till the traditional power plant
64G protection operates, and in this
case it takes 325.6 ms.

Rafael Cordova received his BSEE from

Instituto Tecnológico de Hermosillo, México
in 1990 and received his Masters degree
with honors in 2008 from Instituto
Tecnológico de Aguascalientes, México.
Rafael has been with CFE since 1990 and
has great interest in studying the behavior
of protection relays and power systems.
REFERENCES:

[1] E. O. Schweitzer III and J. Roberts,

“Distance Relay Element Design”
Proceedings of the 19th Annual Western
Protective Relay Conference,
Spokane, WA, October 1992.

[2] Liancheng Wang, “High Speed

Microprocessor Distance Relaying for
Transmission Lines”, ABB Power T&D
Company, Inc.

[3] Rafael A. Córdova Cruz, “Simulación de

una Protección de Distancia Multifunción
que incluye una Protección Direccional de
Sobrecorriente, Protección de Falla de
Interruptor y Protección de Potencia
Utilizando el Programa Simulink”, Masters
Thesis.

[4] Eduardo Humberto Guerrero Flores,

“Análisis de falla a tierra en Generador
Eléctrico de 375 Mw, 17 Kv con
enfriamiento en Aire”, Acapulco IEEE,
Summer 2009.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 245

Annex 1

The fault was found on jumper that goes to

High Side Bushing.

Another view of damaged winding.

Jumper to high side winding.

Picture of fractured winding.

View of the winding after removing cover.

Another view of the damage found.

Pictures shows carbon residue in the rest

of the winding.

Residue of burnt paper at the bottom of the

winding.
246

Annex 2

The failure was located in slot 438 (without

bracing) of branch 3B4 [4]

Detailed view of the stator bar damage

Faulted stator bar in its slot

Section of the faulted bar showing a

detailed view of the damaged insulation

Faulted bar (frontal view). It clearly shows

the damaged part of the bar (Where the
wedge filling was cut).

How the damage could be done

(assumption) during wedge installation
procedure
Detailed view of the stator bar damage
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 247

WHY WE SHOULD MEASURE LINE IMPEDANCE?

Benton Vandiver, Technical Director, OMICRON electronics, USA
Dr. Alexander Apostolov, Principal Engineer, OMICRON electronics, USA

INTRODUCTION
The performance of transmission line protection relays when a fault occurs in the system is
important for improvements in the stability of the system and reduction of their effect on
sensitive loads. Reducing the fault clearing time for more possible fault conditions is one of the
main goals in the development, application and setting of such relays.
The operating time of a transmission line protection relay is a function of many different factors.
Some of them are related to the operating principle and the design of the relay itself. The paper
analyzes the impact of errors in the line impedance parameters on the accuracy of the short
circuit currents and voltages calculation, the settings of the distance and overcurrent relays and
the fault clearing times for different line lengths and fault locations. The accuracy of the fault
location calculation is also affected. This paper explains the difficulty of k-Factor settings and
points out cost effective solutions for preventing incorrect behaviour of distance protection
schemes.
The inaccurate values of the mutual coupling of parallel transmission lines are another important
factor that may affect the operation of the relays for faults involving ground. This is also
discussed in the paper.
The transmission line impedances used for short circuit currents calculation and the setting of
distance relays are normally derived from the results of a line constants program calculation or
systems studies. Due to the large number of influencing factors (e.g. wire types, spiraling and
average sag of the wires, shield handling on cables, specific soil resistivity) these calculations
can be prone to error.
Actual measurement of the fault-loop impedance is the best way to ensure that the distance and
overcurrent relay settings are correct. The second part of the paper describes an advanced
method for these measurements and calculations that provide the impedance data for the
different applications that use it. Comparisons of estimated and measured line impedances are
presented at the end of the paper.
Measuring mutual coupling between power lines can be done using a similar method.

IMPORTANCE OF K-FACTORS
To protect an overhead line or a power cable protective relays are needed. When a fault occurs
on the line, such as an arc between phases or against ground, it has to be cleared safe, selective
and fast. Selectivity means that the line is only switched off, if the fault is really on this very line
[1].
There are two basic methods to obtain selectivity on power lines, differential protection or
distance protection. The better principle is the first one, but there is by far more effort involved,
because the relays on both ends of the line need to communicate with each other. This paper does
248

not further discuss this method. For cost reasons on most power lines distance protection relays
are used.
One of the most important settings of a distance protection relay is the Positive Sequence
Impedance, which is half of the complex impedance of the phase to phase loops (Figure 1).

Fig. 1 – Impedance loop between two phases

When a fault occurs the distance relays on both ends measure the impedance. If the impedance is
(typically) below 80% or 90% of the line impedance they switch off as fast as possible (zone 1),
because it is for sure that the fault is on this very line. If the impedance is higher the relay
switches off delayed (≥ zone 2), to give another relay that might be closer to the fault the chance
to clear it first.
On faults of one or more phases against ground, the impedance of the fault loop is different
(Figure 2). Because the impedance of the ground path, or to be more precise, of this ground loop,
is different, a factor within the relay gives the relation between the line and the ground
impedance. This factor is called ground impedance matching factor or simply k-factor, as it is
often referred to.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 249

Fig. 2 – Impedance loop on a single phase ground fault

If the relay settings are done properly a customer that is supplied from two ends (Figure 3)
continues to receive energy from one line if the other trips.

Fig. 3 – Relays with optimum zone 1 reach

If the impedances or k-factors of a relay are not set properly, zone over- or under-reaches will
occur (Figure 4).

Fig. 4 – Relays with zone 1 over-reach

In the example above three relays instead of two see the fault in zone 1 and trip, a second power
line is dead. The customer lost power for no reason. Besides the damage of customers having no
power, the risk of loosing system stability becomes also higher by such false trips.

DIFFERENT K-FACTOR FORMATS

Unfortunately the k-factor does not exist. There are various formats out there; the three major
types are discussed here. For all types it is to say that they are constants of the line, in general
independent from the length. They express the relationship of the impedance of a phase to phase
loop and a three phase to ground loop. Half of a phase to phase loop (i.e. the impedance of one
line) is referred to as Positive Sequence Impedance Z1; three times the impedance of a three
phase to ground loop is referred to as Zero Sequence Impedance Z0.
One common format is the complex ratio of the Zero Sequence Impedance and the Positive
Sequence Impedance.
250

Z0
k0 = (1)
Z1
Because Z1 is the impedance of one line it is also referred to as ZL quite often.
ZL = Z1 (2)
The ground (or British "earth") impedance ZE can be calculated from the Zero Sequence
Impedance as follows:
Z0 − ZL
ZE = (3)
3
Defining the ground impedance this way, obviously leads to strange results with a negative
inductive component in ZE, as soon as the three-phase to ground inductance is much smaller than
the inductance between two phases. This is the case on some power cables when the shield is
close to the conductors but the conductors are relatively far from each other. This fact is of no
further concern; it is just good to know that it can happen.
Another possibility to express the relationship is the ratio of ground to line impedance.
ZE
kL = (4)
ZL
kE or sometimes referred to as k0 are other common names for this definition. One has to be
careful how a k-factor is defined before using it.
Splitting the complex impedances ZE and ZL into their real and imaginary parts R and X defines
real ratios, this leads us to the third commonly used definition.
RE X
and E (5, 6)
RL XL
Conversions between the different k-factor formats are possible.
k 0 = 1 + 3k L (7)
For converting from the format (5) and (6) to the other formats the other line constants (or at
least the line angle) have to be known.
RE / RL XE / XL
kL = + (8)
1 + jX L / R L 1 − jR L / X L
The line angle can be used to obtain the ratio XL / RL that is needed for the conversion in (8).
tan(ϕL ) = X L / R L (9)
Distance protection relays use algorithms that make use of these different k-factors to convert all
phase to ground faults, so they can be assessed as if they were phase to phase faults. This allows
using the same zone polygons independent from the line geometry. Because different relays can
use different algorithms, identically measured voltages and currents may lead to different
calculated impedances depending on the algorithm used.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 251

Details of these algorithms [2] are not further discussed in this paper; it is only to mention that
the entry format of the k-factor does not allow deducing which algorithm is used by the relay.

CALCUALTION OF K-FACTORS
Up to now the effort to measure line impedances and k-factors was so great that it has hardly
been done. To obtain the this data it had been calculated manually using physical constants, or by
using appropriate software tools [3] like PowerFactory from DIgSILENT, PSS from Shaw PTI or
CAPE from Electrocon, to name a few. The parameters needed to calculate the line impedance
are many.
The geometrical configuration is needed (Figure 5):
• height above ground and horizontal distance for each phase conductor and each ground
wire
• average sag of the line and ground wires at mid-span

Fig. 5 – Overhead line geometry

Several electrical parameters have to be known:
• ground/soil resistivity
• DC resistance of all conductors
• spiralling construction of the conductors
• geometrical mean radius of the conductors
• overall diameter of the conductors
Similar parameters are needed for calculating line impedances of power cables, on a first glance
they might seem even simpler, but as this may be the case for new cables it might be the opposite
252

for old installations where often a mixture of different cable types is used – and not documented
too well either.
In general it can be said that the calculation of the Positive Sequence Impedance works quite
well and in general sufficient for the Zero Sequence Impedance as long as the ground or ground
wire is a consistent good one. When the ground wire or shield is not a very good conductor and a
large component of the fault current is flowing back through the soil, things tend to become
complicated. The influence of the ground/soil resistivity, pipes, other buried metal structures, and
the accurate distance of the wires above ground, make it very difficult to determine the
impedance along the whole length of the line (especially in complicated landscape geometry and
multiple infrastructure crossings).
Another cause for concern is that a huge number of parameters are involved in the calculation of
line parameters. If one parameter is wrong this might cause a substantial error. In the Positive
Sequence Impedance there are several, but even more prone to error is the Zero Sequence
Impedance or k-factor, because they need accurate parameters for their calculation.
On several occasions when our team found incorrect relay settings it was the Zero Sequence
Impedance or the k-factor that was in error. But we also had the situation that two similar lines
were just mixed up.

MEASUREMENT OF K-FACTORS
Compared to the effort for accurate calculations, the actual measurement of line parameters
including the k-factor is today relatively
simple.
The measurement of the line impedance
requires the use of specialized equipment
that includes several components:
• test set comprising a frequency
variable amplifier
• signal coupling unit
• ground protection device Fig. 6 – Test equipment for line Z measurement
The test set used for the line impedance measurements is multi functional, frequency variable
device for various tests on primary equipment. It may be required to generate currents up to
800A or voltages up to 2000V. Support for various automated tests on CTs, VTs, power
transformers or other primary equipment is necessary to improve the efficiency of the primary
testing process.
In the application of line impedance measurement it is used as a frequency variable power
generator, measurement tool and analyzer. Due to the variable frequency generation it is possible
to generate signals first below then above mains frequency. Using a digital filter algorithm
allows measuring frequency selective at the frequency that is currently generated, this means all
other frequencies but the generated one are filtered out. Any disturbances at the mains frequency
from nearby equipment or lines are therefore ignored during the testing.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 253

The coupling unit is used for galvanic decoupling of the generated signals in the output direction
and analyzed signals in the input direction. The decoupling is needed mainly for safety reasons.
For optimization of the performance it is an advantage to have a range selector switch and a built
in voltmeter for a quick check of any induced voltages or high burdens.
The protection device is a safety tool for easy connection to the overhead line or power cable.
Existing grounding sets of the substation may be used. In case of unexpected high voltage on the
power line due to faults on a parallel system, lightning discharges or transients due to switching
operations, the protection device should be capable of discharging short transients or
permanently shorting fault currents of up to 30kA for at least 100ms. These safety features are
necessary to allow the user safe operation even in critical situations.
The measurement is performed with currents between 1 and 100A depending on the line length.
Using frequency selective measurement allows using injected currents a fraction of the size of
the nominal currents. To ensure high accuracy of the measurement the highest current range for
the given line length is chosen. Measurements on lines up to 270km (123 miles) have been
performed so far.
Overall seven measurements per system are made, three for each combination of phase to phase
loops, three for each phase against ground and one for all three phases against ground. There is
some redundancy in these measurements, allowing reliability crosschecks and calculation of
individual k-factors for each phase. The latter seems strange at a first glance, but especially for
short lines having a symmetrical line is not a priority, leading to very different values for the
phases. This results in smaller k-factors and avoids zone overreaches in most cases.
The actual measurement results can be loaded into Microsoft Excel allowing easy post
processing; the results are displayed in a format for direct usage in relay settings (Fig. 7).

Fig. 7 – Major measurement results

CASE STUDY
This US utility had experienced some unexpected trips of unfaulted line sections on their sub-
transmission network. Investigations had lead to a suspicion of incorrect relay settings leading to
zone over reaching, but the reason was not evident. Utilizing this method of directly measuring
the line parameters, they were able to isolate the cause of the over reaching problem.
So far, 16 lines have been tested and documented with actual measurements. A review of the
results show 15 of 16 lines with consistent higher values for the calculated zero sequence
impedance as compared to the measured zero sequence impedance. In fact, the average
percentage error was 51%. (with a range of 10% to 107% error)
254

Results are shown in the graph of Figure 9, and it is interesting to note that the positive sequence
impedance measured values matched the calculated values within 3.5% on average. This
validated the overall measurement results in the mind of the utility.
Comparison of Z Meas to Calc

16.000

14.000

12.000

10.000
Ohms

8.000 Z1 Meas
Z1 Calc
6.000 Z0 Meas
Z0 Calc
4.000

2.000

0.000 Z0 Meas
B- U- O- D- Z1 Meas
V- L- O-93 G-
262 151 301 212 Z- U- A-27 O-67 I-269 Z-26
568 272 163 182 411 A- Q-
573 225
Line Name

FIGURE 9 – Measured impedance versus calculated impedance per line tested

To put this in relaying settings perspective, the graph of Figure 10 shows the comparison of the
k0 values of each line using the existing calculated results and those based on the measured
results. It easily shows that 12 lines are exposed to serious overreaching conditions and 3 lines to
minor underreaching conditions. The average error is 59% with a range of -15% to 147% error.
So the overall effect on the relay settings was dramatic and points to the need for performing
further testing.
The utility has since implemented a program for testing all of the sub-transmission system and
making the necessary settings changes based on the measured results. Murphy has not yet
provided any tests of these new settings on the lines tested, but that’s just the way he works.
Comparison of k0 Meas to Calc

2.50

2.00

1.50
Mag

1.00 k0 Meas
k0 Calc

0.50

0.00
B- U- O- D- k0 Meas
V- L- O-93 G-
262 151 301 Z- U- A-27 O-67 I-269
212 568 272 Z-26 A- Q-
163 182 411
573 225
Line Name

FIGURE 10 –Comparison of k0, measured versus calculated per line tested

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 255

In the analysis of the results the utility wanted to arrive at a conclusion as to why the traditional
method of calculated zero sequence was so far off. It was previously thought that the physical
data was accurate and sufficient for good results. The conclusions drawn were:
1) The ground resistivity "assumption" of 100 ohm-meters may be in error.
2) How often the cable circuits are actually bonded to ground and where.
3) Neutral wires are not run on overhead construction; however no consideration is given
to under-built distribution (4 or 13kV with a neutral cable).
4) Is there a big water pipe, gas pipe, railway, or other infrastructure in the ROW?
If these variables can be accounted for then the calculated values may become closer to the
measured values results.
CONCLUSION
Today the costs and effort for Line Impedance and k-Factor measurements are a fraction of what
they used to be. Measurements showed that for several reasons calculations often gave wrong
results. Therefore, both measurement and calculation will be done in future. Safe, selective and
fast failure clearance is only possible, if all relay parameters are set properly. Line impedance
and k-Factor are of highest importance for a fully operational distance protection relay.
REFERENCES
[1] A. Apostolov, B. Vandiver, "WhyShould We Measure Line Impeadence?", Proceedings
Georgia Tech Relay Protection Conference 2007, OMICRON electronics Corp, USA.
[2] W. Doemeland, Handbuch Schutztechnik, Huss-Medien GmbH, Berlin, Germany, 48-49
[3] S. Kaiser, 2004, "Different Representation of the Earth Impedance Matching in Distance
Protection Relays", Proceedings OMICRON User Conference in Germany 2004, OMICRON
electronics GmbH, 11.1-11.5
[4] A. Dierks, 2004, "Accurate Calculation and Physical Measurement of Transmission Line
Parameters to Improve Impedance Relay Performance", Proceedings Southern African Power
System Protection Conference 2004, Eskom Enterprises, 143-149
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 257

EVENT-DRIVEN LOAD SHED SCHEME AND OPERATION TESTS

Mark Adamiak, PE Michael Schiefen Gary Schauerman Bernard Cable

Fellow IEEE BASF BASF GE Energy
GE Digital Energy Freeport, TX Port Arthur, TX Schenectady, NY
Wayne, PA michael.schiefen@basf.com gary.schauerman@basf.com bernard.cable@ge.com
Mark.Adamiak@ge.com

the 14 can subsequently be shed as needed upon detection of

Abstract - Many large industrials operate internal co-generation
systems to provide a substantial portion of their own power as
well as to use the resulting output heat as part of their industrial
process. Depending on the amount of co-generation, on loss of
connection to the main grid or instability of the grid, these
industrials may have the option of supplying all internal electrical
loads from the co-generation or rather, shed a portion of the
load in order to maintain plant stability. In order to maintain and
maximize plant operations during either an islanded condition, a
severe (and dropping) underfrequency condition, or a prolonged
under voltage condition, a load shed scheme needs to
dynamically monitor the load-generation unbalance in the plant
and, upon detecting or determining any of the above conditions,
needs to quickly initiate shedding of the optimal amount of load
– based on a user-determined priority schedule. Examples of
an actual load shed test are shown and analyzed.
subsequent under-frequency or under-voltage conditions.
II. LOAD SHED SCHEME DESIGN
A load-shed scheme must be able to quickly and reliably
detect a plant island condition or, after some time delay, an
under-frequency condition, or a prolonged under voltage. The
first step is the detection of an island condition. From the
diagram in Fig. 1, it can be seen that the “primary” island
detection is accomplished by determining the Open/Close status
of external source breakers – S1 through Sn. If all the source
breakers are “opened” or if several of the source breakers are
opened and the remaining source(s) are thermally limited, load
shed action must be initiated (Note: thermal limitations allow for
a time delayed shedding of load)

Keywords: Load Shed, IEC 61850, GOOSE, Load Priority G2

G1 Gn
I. INTRODUCTION
G2
Load Shed has been a tool to maintaining power system
stability for many years [1]. In many of the large industrial G1 Gn
complexes where co-generation exists, the economics of being
able to survive a separation from the electric power grid have L3
become a major driving force for change. The economic
L1 Industrial
evaluation includes cost of lost product-in-process, the cost of Power
lost production (including down time resulting from process
upset), the cost of equipment failure resulting form unexpected L2
Complex Ln
shutdown, and the cost of any environmental impact. A tool that
has been used for many years to address the stability of electric S1
grids is Load Shed [1].
Sn
Many of these industrial-manufacturing campuses are often S2
arranged to produce numerous products – each in its own
manufacturing facility within the campus. At a given instant in
time, the product produced from each facility can be assigned a
priority and a related value. In addition, the electric power feed
to each sub-manufacturing facility is typically individually
controllable. Source 1 Source n
The combination of priority and control enables a load shed Source 2
scheme to be configured to “prioritize” which load gets shed and Fig. 1 Plant-Source Load Model
in what order. Certain loads may get marked as “non-
sheddable” and make up part of the base load. In addition to a “detected” island, the scheme must be able to
For example, in a campus where there are 14 sub- “force” an island. Specifically, if an underfrequency or
manufacturing centers, each center can be assigned a “shed” undervoltage condition is detected (as a result of instability of
priority from 1 to 14 and the available load from each facility can the external source), the scheme must trip all source breakers
be measured. When a shed criterion is identified, loads, as to purposefully separate from the main grid.
needed, from 1 to N (N ≤ 14) are then shed. Un-shed load from
258
When in island mode, if one of the internal generating sources
trips off-line, the load-shed scheme must compensate for the
lost internal generation.
A. Decision Criteria
The decision as to whether to shed load and how much load
to shed is based on the measurement of the dynamic load-
generation balance. The internal plant load is calculated by
summing the power flows on feeds into the plant, specifically:
Load = PG1 + PG 2 + ... + PGn + Ps1 + PS 2 + ... + PSn (1)
The “positive” direction of power flow is set to be into the plant
such that if there is excess generation from the internal
generators, the power flow through source breakers S1 through
Sn becomes negative and the internal Load is still calculated
correctly.
The total internal generation is then calculated as:
Generation = PG1 + PG 2 + ... + PGn (2)
The balance point is then computed as:
Balance = Load − (Generation + Spinningreserve) (3)
where
SpinningReserve The amount of load that the
generators can assume
without going unstable;
Typically, this is at least 5% of the present machine loading with
higher values being machine-dependent.
A positive value for the Load-Generation Balance indicates
that the load is greater than the available generation and
spinning reserve and that, upon detection of an island condition,
a load-shed is required in order to maintain the plants electrical
system stability. A negative value indicates that there is excess
generation and that a generator may need to be shed upon
detection of an island condition.
In computing the required load to shed in a load shed
scenario, each sheddable load group will be summed until the
summed value is equal to the “balance” point. Specifically, the
controller will sum until:
Shed _ Amount = ∑i =1 LoadGroupi ≥ Balance
N controller is achieved through the use of the IEC GOOSE. The
(4)
where
LoadGroupi is priority “i” out of N. A logical
convention is that Priority 1 loads
sheds first, priority 2 loads shed
second, etc.
Once islanded, there is still a chance that events in the plant
could start to take down the local island. To address the
“sinking island” scenario, and given that there is additional load
to shed, the controllers can implement the secondary shed
criteria based on under frequency and under voltage. Typical
settings for these quantities would be:
If: Frequency < 58.8 Hz for 1.0 Second or
Voltage < 0.85 pu for 1.5 seconds
Then shed the next priority of load.
Note that it is necessary to coordinate the underfrequency
elements on the generator protection relays with the islanded
underfrequency load-shed values. As is shown in Figure 5 in
the test results, the setting engineer also needs to take into
account frequency undershoot under certain control situations.
B. Implementation Architecture
To reiterate, the architectural requirements for the system are
that it be fast, reliable, and secure. In order to implement these
requirements, a distributed architecture connected by IEC-
based Generic Object Oriented Substation Event (GOOSE)[2]
over Ethernet communications is proposed as it has the point-
to-multipoint performance available. The components of a
priority-based load shed scheme include measuring/monitoring
relays, a communication network, controllers, and mitigation
devices.
Since load shed schemes are critical systems in a plant,
redundancy in the scheme is very much desired. In the utility
world, this is typically achieved through complete duplication of
all components. In the plant environment, a good approximation
of redundancy can be achieved with some component
redundancy and through the use of the failure properties of the
IEC GOOSE.
With the logic of the load-shed scheme residing in the
controllers, they become the first point of redundancy. The
controllers are connected to a redundant Ethernet
communication system such that no single failure of either
controller or a component of the Ethernet network affects the
operation of the scheme. Each controller receives all
measurements from the system and each controller sends a
load-shed command when required. In order to operate the
Ethernet network in a redundant ring, the IEEE Rapid Spanning
Tree Protocol (RSTP) should be configured in the Ethernet
switches. This algorithm detects rings and ring breaks and
dynamically re-connects the communication system to re-route
messages as needed.
The ability of each controller to receive load/generation
information and the ability to send shed messages from either
IEC GOOSE is a multicast Ethernet data frame that contains
packet description information and a user-defined dataset. The
GOOSE dataset can contain information such as remote
measurements of Watts, Vars, and breaker status and can carry
load-shed commands to the mitigation devices in the scheme.
An overview of this architecture is shown in Fig. 2.
As a Multicast Ethernet message, the GOOSE message
can be sent to multiple locations simultaneously. As such, when
a metering / monitoring device sends the Watt flow value from a
location, it is received by both controllers simultaneously.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Controller 1 Controller 2
Ethernet Switch 1 Ethernet Switch 2
Measurement/
Measurement/
Mitigation IEDs
Measurement/
Mitigation IEDs
Mitigation IEDs
Fig. 2 Redundant Controller Architecture
Similarly, when a shed command is sent from either controller,
all the mitigation devices in the system receive the command at
essentially the same time as the time on the wire of a small
GOOSE message operating on a 100MB Ethernet network is
about 20 µsec.
It should be noted that, for data security, all Ethernet physical
interfaces should be implemented over fiber optic media. This
implementation has a second advantage, which is that the
Ethernet signal can travel up to 2km in a single span of
multimode fiber. In large industrial facilities, this becomes
necessary. Distances longer than 2km can be achieved through
the use of an Ethernet switch as a repeater or using single
mode fiber/switch ports which can then operate up to 40 km.
The GOOSE message does not have an IP address and does
not make use of either the Internet protocols of TCP or UDP.
To achieve reliability, the message is sent several times in quick
succession such that if one packet of information is missed or
lost due to a communication error, a repeat of the packet is
quickly re-sent. In the case of a binary-only GOOSE message
(as would be used in sending the load-shed commands), the
first repeat time can be as little as 4ms.
Transmission of the GOOSE is triggered by either a data
change of an item in the dataset being sent or by a periodic time
delay. Data change in a binary value is obviously the change of
the value either from 0 to 1 or 1 to 0. In the case of an analog
value, data change is calculated as a user-defined percent
change in the analog value from a given base. For example, if
the base value for Watts was 100MW and the percent change
detect was set at 1%, a 1MW change in the load from the
previous value would result in the transmission of the GOOSE
containing the updated value of this data item in its dataset.
If, however, the data in the GOOSE dataset is not changing,
the subscribing device cannot detect the integrity of the sending
device. Detection of remote device integrity is accomplished
through a data item in the header of the GOOSE message
known as Time Allowed to Live or TAL. The TAL value sent
from the publisher tells all the subscribers receiving the
published GOOSE that the next GOOSE transmission from the
publisher will occur in XX time where XX is typically in the 1 to
60 second range. If the receiving device fails to receive a new
GOOSE from the publisher in the stated time period, the
259
Subscriber can declare the sending device as “failed” and use
alternate values for those in the expected GOOSE.
It is the ability of setting alternative values in the subscribing
device for those in a missing GOOSE that enables this
architecture to operate without complete redundancy of the
metering and mitigation devices. For example, if the metering
device measuring the load in a facility fails, the subscriber of this
load information can choose to substitute either the last known
value of load or it can select a worst-case load value that was
pre-set by the designing engineer (user selectable option). In
this manner, the scheme can continue to effectively operate.
Similarly, if a mitigation relay fails to report to the controllers, the
control algorithm can assume that the load controlled by the
failed relay is un-available and proceed to select an alternate
load to shed.
C. Time Synchronization
The scheme architecture should include a GPS-based clock
to time-synchronize events and data (as needed) in each of the
installed controllers, measurement relays, and mitigation relays.
The GPS clock provides two different time synchronization
interfaces and related levels of accuracy over which time
synchronization can be achieved, namely, IRIG-B and SNTP.
The measurement relays may have synchrophasor recording
capability, which is used as a long-time and dynamic trend
recorder in the operation of the plant electrical system.
Recording is initiated via a number of different triggers such as
over/under frequency, over/under voltage, Rate of Change of
Frequency (ROCOF), over current, etc. Time-synchronization
of each of these synchrophasor relays requires 1µsec absolute
time accuracy. This level of accuracy can be achieved through
the use of the DC Level Shift IRIG-B time synchronization
protocol via direct wiring from the GPS clock to each relay.
As the load controllers may be located a significant distance
from the GPS clock and as they do not need the same level of
accuracy, time synchronization can be achieved over the
Ethernet communication channel using the Simple Network
Time Protocol – SNTP. Time synchronization via SNTP is
typically able to achieve 1ms (worst case – 10ms) time accuracy
which is adequate for Sequence Of Events (SOE) time
stamping.
In the near future and where measurement relays may be
remotely located, the 1µsec time accuracy will be obtainable
through the implementation of the IEEE 1588 high-accuracy
time sync protocol over the Ethernet network [3]. An
utility/industrial profile is in the process of being developed.
With all the devices time synchronized, an integrated SOE
report can be created that interleaves all the events from all the
devices into one common report.
D. Line Outage / Island Detection
A line outage can be detected either through breaker position
or line current measurement or a combination of both. The
obvious mechanism for breaker position is the use of either “a”
or “b” contacts or both. A single contact is functional and if
used, it is recommended that the “b” switch be used so that a
positive signal is seen when the breaker is actually operated.
Additional security can be achieved through the use of double-
point breaker status as this provides a mechanism for detecting
260
failures in the breaker monitoring system and/or failure in the
break racking system. In the double-point implementation, in
order for a breaker to be declared “open”, the “a” contact has to
be open and the “b” contact has to be closed. If both the “a”
and “b” contacts report the same value, a ”Breaker a/b
Mismatch” alarm can be issued. For dependability, a critical
breaker status can be monitored by two different relays and an
“OR” used to detect a breaker open.
When detecting an open condition, it is desirable to include
debounce timing to the contact to provide a better level of
security that the open condition was not a result of a transient
coupled voltage in a parallel conductor. For shorter lengths of
wire (less than 50m), a 8ms debounce time is adequate. For
longer runs, debounce times as long as 20ms may be required.
Debounce time can be shortened through the use of an active
impedance on the digital input. When a change of voltage is
detected on such an input, the impedance of the input takes on
a low value to more quickly drain any capacitive charge. After
some time, the impedance is raised resulting in a current draw
in the 1 to 2ma range.
Additionally, line outage detection can be validated by looking
at the line current. Given that the measured line current is less
than a set value AND the fact that a breaker open was detected,
a line outage can be more securely declared. Note that a value
of “Zero” for the current is not used as current from a CT can
often have a residual tail that would significantly delay the
detection of zero current. With the tail described, a zero current
detection could take 10 to 12 ms – adding security and only
slightly slowing down the overall declaration of an island
condition.
E. Load Shed Activation
By having the shed solution pre-computed and the sheddable
loads pre-armed, the scheme can be ready to operate as soon
as an island condition is detected. Subsequently, the
appropriate loads can be shed with no appreciable time delay.
In an overall scheme, the time delays come from 4 primary
sources: breaker status indication, logic processing, output
contact operation, and breaker operation. On the breaker status
indication, all breaker status inputs are typically denounced for
8ms before becoming “valid” to use in the logic. Once validated,
the breaker status values are fed into a logic engine (typical
execution time – 1-2 ms). Any decisions to shed load are then
communicated via GOOSE to the appropriate remotely located
controllers. Note that the communication time is not noted as a
primary source of delay as the time on the wire of the
communication message is less than 300µsec. In the receiving
relay, there is typically a 1ms delay for logic processing, a 3ms
time delay in the operation of the output contact (solid state
outputs take this down to 100 µsec), and finally, there is a 32-48
ms breaker operate time. In actual timing tests, an average time
of 13ms was measured from initiation of the island condition to
energization of the trip coils on the load-shed breakers or
lockouts. Note that lockouts may be needed on some loads to
prevent automatic or operator re-starts.
F. User Interface
Monitoring and control of the system can be achieved either
through a PC-based HMI and/or through LEDs and pushbuttons
as are typically available on the system IEDs. Examples of the
type of information that can be mapped to an LED include
system status, breaker status, scheme operational information,
device status/alarms, and communication system alarms.
Pushbuttons on the controllers can be used to:
• Enable/Disable the system
• Reset latched LEDs/Alarms
• Reset events
• Change operating breakers
In addition to status and alarm information, high and low-
speed oscillography can be provided with high sample rates up
to 15,360 samples per second and low-speed rates (based on
Synchrophasors) of up to 60 samples per second. The high-
speed oscillography is designed to capture faults whereby the
low-speed oscillography can provide a view into the dynamics of
a plant. The various oscillographies can be triggered by
functions such as over or under frequency, over or under
voltage, and rate of change of frequency. A single
synchrophasor record can span over 40 minutes in length.
Fifteen-minute phase current, MW, and MVar demand
information can typically be logged and retrieved for load
analysis.
G. Live Islanding Tests
Below are results from a load-shed system that was tested in
two actual plant-grid separation situations, namely: - Scenario 1,
G1 230 kV 230 kV G2
Feed 1 Feed 2
G1 G2
GA Positive MA MB GB
Power
Positive
Power
TIE
Load
Figure 3
Load
34.5kV Bus Sources
where a load shed was not required and scenario 2 where a
load shed was required. Figure 3 shows a simple 1-line of the
plant electrical system. As the plant was not operating at full
load and as the available sheddable load was only 3.5MW, the
load-shed levels in the controller were temporality modified for a
smaller shed range. The plant-grid separation was created by
manually tripping the external feeds from the utility into the plant
(breakers MA and MB in Figure 3). As a precaution in both
scenarios, the trip signals were removed from all but the
available sheddable load (a 3.5MW pump motor).
In scenario 1, the gas turbines were set to deliver the entire
internal plant load plus an additional 9.4 MW, which was
effectively exported to the local utility. Prior to the separation, it
was noted (as expected) that none of the load shed tiers were
“armed” or expected to operate. Upon creation of the island, as
expected, there was an immediate but small voltage increase
(0.27%) in the positive sequence voltage (V1), however, within
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 261

16ms, the voltage regulator initiated a 0.87% drop in voltage

(see Fig. 4). The turbines did start to accelerate but 1.78
seconds after islanding, the turbine controls had started the
slow-down process of the turbine. The resulting positive
sequence Synchrophasor voltage angle for the island is shown
in Fig. 5.
+ 180°

Generator
Slowing
∆t = 1.66 sec
Island Created

Nominal - 180°
Voltage ∆t = 1.78 sec
0.87% Fig. 5 V1 Synchrophasor Angle Response to Over-powered Island
Voltage Drop Creation

Fig. 4 V1 Magnitude Synchrophasor Response to Over-powered

Island Creation Again, opening the main breakers created the island. The
island was detected in 8ms (the debounce time on the breaker
contacts) and a 3-Tier Load Shed command was immediately
issued. The 3.5 MW motor was off-line (including breaker
The absolute time reference of the Synchrophasor angle plot operation time) in 50ms from the detection of the island. Figure
allows one to calculate the dynamic frequency of the system. 7 shows the response of the positive sequence voltage – V1.
Specifically, the actual system frequency – factual can be Upon creation of the island, the voltage immediately drops only
calculated from the nominal frequency – fnominal – (60 Hz in this 0.21% and then starts to recover. When the motor load is shed,
installation) and the change in the angle over time (θ in degrees): the voltage overshoots 0.66% - a very nominal amount.
∆θ 1
f actual = f no min al +  Overpow ered Island System Frequency
∆t 360 (5)

By applying equation 5 to the angle plot of Fig. 4, the resulting 60.50

dynamic frequency can be computed as shown in Fig. 6.
60.00
In scenario 2, the turbines were programmed to output 4.55
Frequency

MW less than the internal plant load. This scenario was 59.50
designed to execute the “shed” commands. As mentioned
59.00
earlier, the arming levels were temporarily lowered to force the
arming of the respective load shed tiers. Prior to islanding, it was 58.50
noted that all 3 Tiers were “armed” and ready to operate upon
58.00
detection of the island condition. The recovery from the 4.55 MW
deficit was to be made up of two sources, namely, a 3.5 MW 0.00 10.00 20.00 30.00 40.00
motor shed and spinning reserve (chosen at 5%) of the gas Tim e (sec)
turbines.

Fig. 6 Dynamic Frequency of Gas Turbines

262

petrochemical industry at one of the world largest naphtha

Voltage overshoot steam crackers. His jobs have included various engineering,
maintenance and supervisory positions. Gary is a Registered
on load shed – Professional Engineer in the State of Texas.
0.66% Michael J. Schiefen, received the AAS from Texas State
Technical College, in Waco, Texas, 1978 and BSEE from Texas
A&M University, College station, 1987. He is currently with
Nominal BASF Chemical Corporation, Freeport, Texas, where he
Voltage provides technical support for the major electrical projects for
North American and Operations support for the Freeport Site.
∆t = 1.5 sec His experience as both a electrical technician and engineer,
has encompassed conceptual design, detailed engineering,
acceptance testing, maintenance and operations for Utilities,
3.5MW Load shed Electrical Cooperatives, and Major Petrochemical facilities. His
specialty is power systems engineering, maintenance and
operations. He is a Registered Professional Engineer in the
Voltage drop on State of Texas.
islanding – 0.21% Mark Adamiak received his Bachelor of Science and Master
of Engineering degrees from Cornell University in Electrical
Fig. 7 V1 Synchrophasor Magnitude Response to Underpower Engineering and an MS-EE degree from the Polytechnic
Island Creation Institute of New York. Mark started his career with American
Electric Power (AEP) in the System Protection and Control. In
1990, Mark joined General Electric where his activities have
ranged from advanced development, product planning,
application engineering, and system integration. Mr. Adamiak
III. CONCLUSIONS has been involved in the development of both the UCA and
IEC61850 communication protocols. He was also the Principle
Today’s manufacturing facilities require a higher degree of Investigator on the EPRI IntelliGrid Architecture project. Mark
availability of electrical energy than in the past. Although load is a Fellow of the IEEE, the US Regular Member for the CIGRE
shed as a reliability mechanism is not a new concept, the design Protection & Control study committee, a registered Professional
of a system capable of real-time balancing of generation and Engineer in the State of Ohio and a GE Edison award winner for
load provides many advantages in terms of performance and 2008.
flexibility as demonstrated through the inclusion of actual test Bernard Cable received his Associate in Applied Science
results. It was demonstrated that such distributed generation – (Electrical Technology) and Associate in Science (Engineering
load matching schemes can be successfully implemented using Science) degrees from Mohawk Valley Community College and
modern microprocessor-based relays, Ethernet communication a Bachelor of Science (Electrical Engineering) and a Master of
and IEC 61850 GOOSE. Science (Engineering) degrees from Union College. At General
Electric, Bernie’s activities have included performing studies and
IV. REFERENCES providing design engineering services for industrial and
municipal utility electrical distribution systems and providing
consulting engineering services to industrial and municipal utility
[1] Northwest Power Pool Transient Stability and Load customers. Bernie’s study/design engineering experience
Shedding Controls for Generation - Load Imbalances, includes short-circuit analysis, protective device coordination
IEEE Transactions on Power Apparatus and Systems, and protection scheme design, generator, motor and power
Vol. PAS 100, No. 7 July 1981. plant auxiliary protection, power flow analysis, transient stability
[2] Communication Networks and Systems for Power Utility analysis, large motor starting analysis, harmonic filter design
Automation, IEC 61850 – 2004. and power system conceptual design. Recent activities have
[3] IEEE 1588-2008 A Precision Clock Synchronization also included integration of modern digital multi-function
Protocol for Networked Measurements and Control protective relays into overall plant electrical distribution systems
Systems; and interconnection of these relays to create high-level control
http://standards.ieee.org/board/nes/projects/1588- schemes.
01.pdf.

V. Vita

Gary Schauerman received a Bachelor of Science degree in

Electrical Engineering from Lamar University in Beaumont,
Texas. Since 1978, he has worked at a medical teaching
university in Texas, an electric utility serving several southern
states, an oil refining facility and is currently employed in the
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 263

Protection of Battery Energy Storage Systems

Roger Hedding – ABB Inc. Pat Hayes – ABB Inc.

In today’s power systems, growing demand, aging infrastructure, system constraints as well as the
increasing renewable energy portfolio have increased the need for utilities to find new ways to
manage their system and increase reliability. One area that is surfacing as a potential solution to
this is what is commonly referred as the “holy grail” of the industry or energy storage. The utility
industry is the only industry that does not have a common warehouse or inventory of the product
they produce. When a customer turns on a light switch, the power is generated immediately. To
store this power in a warehouse has not been done economically in the past.

That all being said, energy storage is not a new market as we have seen many types of systems
over the years, such as compressed air, pumped hydro, flywheel, Ultra-capacitors and
Uninterruptible Power Supply (UPS) systems. They all have various levels of storage capability
and have a purpose where they can add value to a utility grid.

The applications for energy storage has been well documented by a report written by Sandia
National Laboratories1 and has broken it down to the following areas: Electric Supply, Ancillary
Services, Grid Systems, End User/Utility Customer and Renewables Integration. Some of the
largest benefits to the utility market have been discussed to be in the areas of ancillary services,
capital deferrals, load shifting, peak shaving and renewable energy.

Energy Storage has two components that are the basis of the design. Each Energy Storage
System (ESS) needs to have a power component (kW rating) and an energy component (kWh
rating). Another way to look at it is how much power do you need and for how long do you need
it? A 1MW system for 15 minutes of storage would have a 250kWh energy rating.

The development of batteries has helped grow momentum in the energy storage market. A
typical Battery Energy Storage System (BESS) has three main areas: 1) Battery and Battery
Management System; 2) Power Conditioning System (PCS) and 3) AC Grid. The type of
battery used in the system defines the amount of power and energy ratings of the BESS system.
A typical system can look like the Figure 1 below.

Figure 1 - Typical BESS components

264

The inverter lineups are sized to handle the power rating of the battery string. The DC battery
connections are bussed to the individual inverter modules where it is converted to a regulated AC
voltage. The AC output from the inverter modules is connected to a common AC bus and then to
the low voltage AC circuit breaker where it is available as a three phase AC output. The line side
of the AC breaker can be connected to the secondary side of the step-up transformer. Each
inverter lineup is connected to its own secondary winding on the transformer which acts as an
isolation.

The power conversion from DC power to AC power in the PCS is performed by using IGBT-
based inverter modules. The module is a four-quadrant switching-mode converter, i.e. the power
flow through the converter is reversible. Shown in Figure 2 is an inverter module with the side
cover removed. The AC power output from the inverter needs a combination of reactors and
capacitors for filtering to minimize harmonics. See schematic of a three-phase inverter bridge
and filter scheme in Figure 3.

Figure 2. Inverter Module (front view).

Typically the individual inverter modules can be hard paralleled to add up to the kW output
needed for the system. The inverters are sized for the power, temperature, altitude, additional
reactive power support as needed for the application.

RFI

Figure 3. Inverter Module topology.

In case of an internal failure, the faulty module can be automatically disconnected and the
remaining system can continue operation at slightly reduced power. If the inverter modules are
complete three phase units, it is also possible to remove complete modules within a PCS lineup
and continue operation at proportionately reduced power.

The PCS system allows control of both real power (P) and reactive power (Q) based on the
system requirements. Advanced control features in the “Generator Emulation” mode of operation
make the PCS look like a traditional synchronous machine to the power system. In addition the
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 265

The PCS can have Statcom capability (VAr control). Additional control modes could include
high and low voltage ride through, auto-island functionality and black start capability.

There are two areas of concern when discussing the protection of the Battery Energy Storage
System. First, protection of the BESS system itself, second, removing the BESS from the system
for external faults.

Internal Fault protection

Figure 5 shows a typical 2 MW BESS system.

Figure 5
Typical 2MW BESS System

Each of the batteries is controlled by a Control unit (Battery Management System or BMS). The
Control unit is in turn controlled by the Power Conditioning system Master Controller. The PCS
Master Controller determines if the batteries are needed to supply real power (discharging), or if
not fully charged, to charge the batteries. The Master Controller with BMS, based on the system
needs and past history, can determine which battery to use. The battery modules are protected by
a DC breaker, contactor, or a fuse.

Using a combination of high speed semiconductor fuses, and dc and ac breakers, the BESS
system is covered for most conceivable internal fault contingencies.

External Fault Performance

Before discussing external fault performance of the BESS system let’s examine the overload
capability.
266

Time % of Rated Power

10 min 120
30 sec 150
3 sec 175
3 sec 200 @ 75% preload

Table 1
BESS Overload Capability

From table 1 it can be seen that the output of the BESS is limited to 200% rated power for 3
seconds. A 2 MW BESS can supply 4 MVA short circuit power for 3 seconds before shutting
down.

Islanding Detection

One of the criteria in IEEE 1547 is to detect an Islanding condition in 2 – 4 seconds. Often times
decaying frequency has been used as a measurement of the islanding condition. In contrast to a
synchronous generator whose frequency will decrease if the electrical power needed exceeds the
available mechanical power. A BESS maintains constant frequency under overload conditions.
Classical means of detecting an islanding condition where the load is much greater than the
generation is to use under frequency relays. This approach isn’t viable in BESS applications.

It’s possible through the Master Controller to impress a small high frequency ripple on the 60 Hz.
Under normal non-islanding conditions, the output of the BESS is just a small portion of the total
generation, so the percent of this high frequency current in relation to the 60 Hz current in small.
However, during islanding conditions, when the BESS is separated from the system, even though
the amount doesn’t change, the percent becomes much higher, and is easily detected by frequency
relays.

Removal of the BESS During External Fault

During normal operating conditions a BESS can either supply MW into the utility system
(discharge) or it can absorb watts from the system to charge the batteries,

Figure 6
Charging/Discharging Characteristic of BESS

Figure 6 shows the typical charging and discharging currents for a Battery Energy Storage
System. Note the currents are mostly resistive. Applying a directional overcurrent relay to detect
a fault condition doesn’t look promising at first glance. Since overhead lines, cables, and
transformers are mostly inductive in nature, we know that fault current in a given phase lags the
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 267

phase voltage by 75º to 85º. A directional relay can be used to detect external faults as shown in
figure 7.

Fig. 7
Directional Overcurrent Relay Setting

The directionality of modern microprocessor relays can be defined by setting the Relay Operate
Angle, and Maximum Torque Angle to achieve the desired response. The fault current is limited
to twice normal current, and the duration fault current is supplied from the BESS is limited to 3
seconds. These factors must be taken into account when setting the directional overcurrent relay.

Other relays can be used for the detection of faults and isolation of the BESS from the system.
For instance, an impedance relay can be used looking into the system from the BESS system.
Although feasible, it’s not economically desirable on small 2MW systems.

Conclusion

With the advent of more and more wind generators, and solar projects being placed on the utility
grid, Battery Energy Storage Systems will find there way to level out the peaks and valleys these
devices generate. It’s a prudent protection engineer that understands these new concepts before
they are placed on their system. This paper introduced a typical BESS, and discussed its
construction, operation, limitations, and finally, protection.

Since these are solid state devices, not rotating machines, some rethinking of the normal
protection philosophy needs to be done to assure the BESS gets disconnected in a timely manner.

References

1) “Energy Storage for the Electricity Grid: Benefits and Market Potential Assessment Guide”,
Sandia National Laboratories Report # SAND2010-0815, February 2010. Jim Eyer and Garth
Corey.

2) “Application of Battery Energy Storage In Power Systems” W.R Lachs, D. Sutano , IEEE
Catalog Number 95TH8025

3) “Value Analysis of Battery Energy Storage Applications in Power Systems” A. Oudalov,

D. Chartouni, C. Ohler, G. Linhofer PSCE 2006 1-4244-0178-X/06
268

4) PCS100 ESS Grid Connect Interface for Energy Storage Systems User Manual,
2UCD19000E001

Biographies

Roger Hedding graduated from Marquette University and joined Westinghouse Electric Corp. After
receiving a Masters degree in Electrical Engineering from the University of Pittsburgh, Roger became a
District Engineer, and eventually moved to Milwaukee where he currently resides. As a Relay
Development Consultant he guides the development of relay products for the North American market.
Roger is a IEEE senior member, and Secretary of the IEEE Power Systems Relay Committee. Roger has
authored or co authored many papers in power systems protection. His hobbies include playing golf,
traveling with his wife, and playing with his grandchildren.

Pat Hayes is currently an Account Manager for Energy Storage Systems for ABB Inc. Pat has been in this
position for the past eight (8) months and is responsible for Sales and Marketing of Energy Storage
Systems in the North American market. Prior to this, Pat was an Account Manager for ABB Inc, where he
had been employed for the past thirteen (13) years. At this role, Pat was responsible for managing sales,
developing commercial contracts, supporting ABB and ABB legacy products from specification through
retirement and providing technical assistance for projects associated with large Investor-Owned Utilities
(IOU's). Prior to ABB, Pat worked with WESCO for five (5) years in a few different roles. The last role
was a Sales Engineer supporting utilities in the Minnesota and Wisconsin areas, with main responsibilities
including sales and technical support with primary interest on metering systems. Pat has taught at metering
schools held at the Chippewa Valley Technical College in Eau Claire, WI. Pat has been an IEEE member
for 20 years and served on the Planning Committee for the Minnesota Power Systems Conference for the
past 10 years. Pat has a BS in Electrical Engineering from the University of North Dakota.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Power System Protection Testing - Exploring the IEEE C37.233 Guide
V. Madani, H. DoCarmo, M. Adamiak, R. Beresh, D. Novosel
P. McLaren; J. Verzosa, M. Kezunovic, A. Apostolov, J. Ariza; E. Udren, G. Henneberg
Abstract— This paper s ummarizes t he co ntent o f th e I EEE
C37.233 Guide f or P ower S ystem P rotection T esting. T he
guide pr esents t est a pproaches, de finitions, a nd pr ocedures
both for the components and the overall protection and control
systems. I t co vers s uggested req uirements f or P ower S ystem
Protection Scheme Testing, S ystem A pplication T ests
(including the scope and level of tests) , and benefits of testing
the overall p rotective s chemes. T he G uide d escribes t he
methods, extent, a nd ty pes o f s ystem te sts fo r p rotection
applications at var ious vol tage l evels. C ontrol f unctions
inherent to the protective systems a re i ncluded. D esign t ests,
commissioning tests, r outine m aintenance t ests, a nd o ngoing
performance a ssessment t ests a re d escribed w ith d iscussion o f
what each of t hese t est categories aims to accomplish. T est of
equipment u sed i n t he s ystem p rotection s cheme, associated
communications equipment, auxiliary po wer s upplies, a nd t he
control of power apparatus are also addressed.
Index Terms—protection system t esting, p rotective rel aying
schemes, System Integrity Protection Schemes, certification tests,
maintenance tests, protection system communications testing, IEC
61850, unconventional instrument transformers.
I. INTRODUCTION
The IEEE C37.233 guide covers suggested test requirements
for Power System Protection Scheme Tests, System
Application Tests, and benefits of the Overall Protective
Schemes Tests. The guide encompasses overall system testing
procedures (generators, line, line reactors, transformer,
capacitors, special protection schemes, end-to-end testing,
distributed application within substation, etc.) and data
collection requirements as well as the test procedure
definitions. Methods, extent, and types of system tests for
protection applications at various voltage levels are also
covered. Control functions inherent to the protective systems
are included. Importance of line testing, indirect trip
applications, open / closed loop tests, and dynamic / non-linear
tests are also covered.
General approach and specific procedures for testing
protective relaying systems that include multiple interacting
relay components, auxiliary devices, and power apparatus are
covered. The procedures focus separately on design tests,
commissioning tests, routine maintenance tests, and ongoing
performance assessment tests with discussion of what each of
these test categories aims to accomplish. The Guide has a
comprehensive reference list of type tests for protective
devices as well as overall protection scheme performance tests
for various types of protection schemes.
269
II. BENEFITS & JUSTIFICATION FOR DIFFERENT TYPES OF
TESTING
Traditionally, the protection and control systems are designed
to identify local faults based on voltage, current, and frequency
determined at a specific point on the utility system. The
protection and control system is designed to isolate the faulted
area. Changes in the operation of the power system, in terms of
location and connection of generation sources, transmission
capacity, and load demand, and changes in power system
equipment such as the growing availability of reliable wide
area communications result in changes in some basic
philosophies of the protection and control system. For
example, wide-area protection schemes and system integrity
protection schemes (SIPS) are applied more frequently to
prevent failure of the utility system during widespread events,
or to prevent local faults from cascading into adjacent
equipment or power systems. Application of SIPS coupled by
the narrower operating margins of the power system is
commonplace today and requires a comprehensive look at
traditional localized practices and focus more on testing of the
protection and control schemes as a system, as opposed to
individual components.
The goal of testing the entire protection and control system is
to validate the performance of the entire scheme, including the
individual components to backup, redundant, and auxiliary
components. The test methods should include verifying the
designed interactions between components work correctly in
terms of maintaining the dependability and security of the
scheme at a desirable level of performance. The benefit of
testing individual components is known from years of
experience: if the individual parts work and the scheme is
designed and installed correctly, there is a high probability the
system will work correctly. However, protection system testing
verifies that the scheme is correct in both concept and
application settings. An example is end-to-end testing of
transmission line protection, using test conditions based on
fault contingency studies. The different test cases, run
simultaneously at each end of the line, confirm that all relays,
communications equipment, and auxiliary equipment work for
commonly anticipated events and worst-case scenarios. It is
even possible to simulate typical modes of failure of the
individual components to ensure the protection and control
system will still identify and manage fault events.
The key benefit of the system tests includes verification of the
protection and control scheme in its entirety. A proper test
documents the performance of the protection and control
system and identifies limitations of performance. The test
270
documentation also serves as a baseline for evaluating future
performance during actual events versus the expected
performance. Baseline performance is used to compare
differences between expected and actual events. Testing the
performance of the overall protection system, especially wide-
area protection schemes or SIPS, is dependent on static and
dynamic system models. As these schemes and the system
behavior during events the schemes are designed to protect are
very complex, testing can provide the engineer a better
understanding of the protection system, the limits of
performance, and the possible modes of failure. One type of
testing for system SIPS is to install the system, while blocking
any control actions, to analyze the performance of this
protection system during actual system events. Such actual test
data can be used to improve the static and dynamic models of
the power system. A further benefit is actually illustrating or
understanding the protection and control system performance
during certain types of events and where and why failures may
occur. The importance of written test procedures and
documentation are described in the “Considerations for
Testing and Self-Monitoring” section of this guide.
The design of the protection and control system requires
careful consideration of the performance requirements for the
specific location, analysis of power system behavior, and an
understanding of the actual performance limits of components
available for use in this application. Testing for the protection
and control system at a specific location should be performed
to prove the assumptions made during the design phase,
starting with the individual components and working towards
testing total system performance.
III. TYPES OF TESTS
Different types of tests are described in the Guide. Some of
the tests are device specific and others are application
oriented.
Certification tests
Certification tests, also referred to as Acceptance Tests or
Type Tests, are performed by a certification organization or
application tests on demand of a user. Certification tests are
objective and can be accepted by a wide range of users.
Certification tests are normalized tests under normalized
procedures and with normalized equipment. For numerical
protection devices, type tests cover hardware as well as
software. Each hardware or software version can be a
different type. The results of these tests are acceptable to a
wide range of users, independent of the application.
Certification tests consist of conformance tests and
performance tests.
Conformance tests
The goal of conformance tests is to verify the performance of
the protective device or protection system against a set of pre-
defined specifications.
 Functional conformance tests: Functional conformance
tests verify if the functionality of a protective function is as
expected. Tests are focused on verifying the general
characteristics against specification by means of signals
without transients and DC-components. Functional
conformance tests are generally steady state accuracy tests.
 Technological conformance tests: Technological
conformance tests verify the response of the protective
function to external disturbances and internal failures.
Technological tests cover the verification of the hardware
quality, a reliability assessment and an evaluation of the self-
supervision. The verification of the hardware quality includes
insulation properties and environmental conditions
(electromagnetic compatibility, mechanical and climatic).
Performance tests
A performance test (also referred to as acceptance or type test)
is a type of certification test that describes the limits of
performance of protective devices used in a specific
application. The test is a generic way to rate the performance
of a specific device. The test results clearly describe the ability
of the device to perform according to its specifications and the
performance of the relay during certain types of common
performance requirements and fault conditions. A simple
power system model is generally used to simulate various
conditions to derive the test quantities. These types of tests are
typically performed when evaluating a new relay for use on a
specific power system, or, in some cases, when evaluating new
firmware. See IEEE C37.231 for additional information
regarding firmware revision control. The severity of firmware
changes can be used to determine whether new suites of tests
are needed. The general results from certification tests include
operating speed, consistency and repeatability of performance,
boundary conditions for optimal performance, understanding
of settings methodology, and the suitability of this relay for
typical applications. As an example, certification tests of a
distance relay will illustrate operating speed of the distance
element, and the reliability of the distance element for zone
boundary faults and during CCVT transients. These tests,
therefore, determine the general suitability of a distance relay
for EHV and HV protection applications.
The goal of performance tests is to verify the behavior of the
protection under realistic network conditions. It includes
aspects like dc offset, source-to-impedance ratio (SIR), fault
resistance, various types of faults, etc. In contrast to
conformance tests, performance tests do not focus on how a
certain function is performed within the protection, rather on
what is to be expected from a function under certain power
system conditions. In case of multi-function protection,
distinction can be made between function tests and scheme
tests.
 Functional performance tests: During functional
performance testing, each function is tested and evaluated
separately. The performance of an individual protective
function is examined in detail.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
 Scheme performance tests: Scheme performance tests
verify the performance of a protective function under
various conditions. Because numerical protection contains
a large number of functions and combinations, it is rather
impractical to test them all. Scheme performance tests are
therefore limited to a number of practical protection
applications, with the protection schemes configured as
applicable.
Software (firmware) upgrade of a protective function should
be treated as testing a new protection type. Due to the interval
of software upgrades, full replication of all tests may not be
cost effective. Therefore, a limited test program can be
performed if, in an earlier stage, a complete type test is
performed. The limited program should consist of tests to
verify
 That the new or changed function works correctly
 That no other functions are declining. This is more
complicated because of the number of functions and
possible combinations.
Some probabilistic analysis may help to determine which
elements or hardware components need retesting.
Application tests
Application tests are specific tests to determine the suitability
of a relay for a specific protection system design application or
location. Application tests are mainly driven by the user.
Application tests are based on a detailed model of the power
system and include performance testing against a wide variety
of possible fault conditions. This type of test typically uses
transient simulation to better replicate the behavior of the
power system. The goal is to ensure a specific relay will
perform for a specific application or location before actual
installation. While certification tests focus on specific relays,
which are the basic building blocks, application tests may also
be used to verify the entire protection scheme. Application
tests are typically performed during the evaluation and design
phases.
Application tests have many benefits. Application tests
document that a specific relay, algorithm, or protection scheme
is the correct choice for a certain set of power system
conditions or criteria. It is typical, for example, to use a
transient model of the specific power system, using multiple
fault conditions, to determine the suitability of a distance relay.
System integrity protection schemes require significant
application tests as various contingency transient system
modeling is inherently used to design the scheme and specify
the scheme response to power system events.
Another benefit of application tests can be determination and
documentation of appropriate settings. Application tests verify
the performance of a protection system in total response and
will highlight the failure or overlap of specific elements of the
271
scheme. The operating settings of these elements can then be
modified based on the results of the application tests.
Application tests are based on the use of transients for testing
protective relays in order to simulate the behavior of the
network during faulted conditions, e.g.:
 Fault resistance
 Current transformer saturation
 Voltage transformer transients
Commonly practiced methods of creating test files for transient
tests and use of transient signals obtained from fault recorders
are described in the Guide and references to other IEEE
Standards are provided.
Commissioning tests
Commissioning tests are intended to ensure that the protection
system will operate as designed after field installation. These
tests verify the individual components, interactions between
components, device set points, communications system, and
scheme redundancy along with wiring and installation. These
are basic tests that must be performed for any new installation
or significant modification to an existing installation and are
typically combinations of certification and application tests.
Certification tests performed during commissioning verify that
elements perform in a field installation in a manner consistent
with laboratory results. This step is very important for
equipment with performance that varies due to environmental
conditions or age of components such as electro-mechanical
relays. Application tests performed during commissioning
verify that the total installed scheme works as designed and
intended, especially in regard to installation accuracy and
interaction between various protection and control system
components.
Commissioning tests are vitally important as these are the tests
that prove that the actual installed system will work as
designed. Commissioning tests can become very advanced.
End-to-end tests ensure that all terminals and the overall
communication system function properly. The most important
benefit of commissioning tests is verifying that the protection
and control system works in the field as it was designed. A
second benefit is documentation of field performance as a
benchmark for verifying performance for future tests.
For the elements or features that have been set on a device, the
tests normally follow the functional hierarchy of the protection
system in the following order:
 Check the available system parameters and measurements
and make sure that they meet the technical specifications.
 Test the enabled and configured protection elements.
 Test the built-in protection schemes.
 Test the multi-relay protection schemes.
Commissioning tests of an entire system are also intended to
verify the hard-wired or communication-based interfaces
272
between the multiple microprocessor-based devices included
in a protection scheme, or other distributed applications. The
interface with the auxiliary contacts of the breaker and its trip
coils are generally included as part of the tests.
Commissioning tests may require the use of multiple
synchronized test devices in order to verify the performance of
protection schemes or other distributed applications. Proper
simulation of abnormal conditions for each of the devices in a
scheme is essential for the testing.
In summary, commissioning tests are performed to make sure
the protection system is properly installed and working as
expected in a substation based on the design.
Maintenance tests
Maintenance tests are specific tests to validate that the
protection and control system is operating correctly after a
period of time of field installation. Some of these tests are
calibration tests to confirm that individual components are still
operating within desirable performance parameters. This type
of test is especially vital for the components susceptible to
degraded or changing characteristic due to aging and wear.
Application tests may be performed during maintenance
testing to verify that the total protection system works, from
measurement of power system values, to properly identifying
fault characteristics, to the operation of the interrupting device
and associated interlocking with automatic or manual
restoration devices and schemes. One form of maintenance
testing involves performing forced outage of protective
equipment when performance of the respective scheme is
identified as questionable and immediate attention to servicing
the relay or the scheme is needed.
There are different philosophies concerning when it is
appropriate to perform maintenance tests in an attempt to best
balance assurance that the protection system will perform
correctly versus the costs involved in maintaining the system.
Aside from forced outage testing, there are two basic
methodologies in use which are time-based (periodic)
maintenance and condition-based maintenance practices.
Periodic maintenance testing is performed to ensure that the
protection and control system is operating correctly after a
period of time of field installation. These tests may include
calibration tests as described earlier. Full system performance
tests may be used to confirm that the total protection system
functions from measurement of power system values, to
properly identifying fault characteristics, to the operation of
the interrupting device and associated interlocking with
automatic or manual restoration devices and schemes.
Time-based maintenance is based on scheduled intervals for
review and maintenance of the system, to demonstrate that
routine testing of the system and system components are
performed. Test schedules are generally determined by
operating experience with a system or device, manufacturer’s
recommendations, and the availability of resources to perform
the testing. Periodic testing may include review of recent
power system activities on the particular terminal and whether
the entire protective system has operated correctly since last
scheduled interval. If determined that a protection system and
scheme has performed correctly, the interval may be adjusted
based on most recent operational experience on the respective
terminal. Similar to performance-based maintenance,
condition-based maintenance includes examination of the relay
system, history of specific systems and devices, and operating
experience to perform maintenance only when necessary to
ensure adequate performance of the protection system.
Intelligent electronic devices such as microprocessor based
protective relays add an additional form of maintenance
testing, which is the ability to perform continuous self-
monitoring for correct performance of the device. Self-
monitoring capabilities vary between devices and may include
the ability to monitor the incoming analog measuring circuits,
the device output contacts, time synchronization signals,
communications signals, internal hardware (including battery
status and system board voltages), non-volatile EEPROM
memory, and the internal software algorithms. Failure of a
self-test routine typically generates an alarm available through
SCADA communications and output contacts. Certain self-test
alarms may disable the functioning of the device.
Troubleshooting operating problems, from a protection system
testing perspective, is generally a combination of
commissioning tests and calibration tests, designed to identify
specific components or specific parts of the protection system
design that do not provide the desired operation results.
Troubleshooting always involves good engineering practices
and experience to identify the cause of operating problems.
IV. MAINTENANCE PRACTICES
It is very important to the reliability of the electric system that
protection and control systems function properly.
Malfunctioning systems can contribute to major power system
outages and widespread cascading events. Maintenance
programs have demonstrated benefits in uncovering problems
and allowing the problems to be addressed under controlled
power system environment. Maintenance and test programs
often incorporate the following types of maintenance practices:
1. Time-base maintenance (TBM) – These test intervals
are applied to components or groups of components.
The intervals may have been developed from prior
experience or manufacturers’ recommendations. The
TBM verification interval is based on a variety of
factors, including experience of the particular asset
owner, collective experiences of several asset owners
who are members of a country or regional council,
etc. The maintenance intervals are fixed, and may
range in number of months or in years..
TBM can include review of recent power system
events near the particular terminal. Operating records
may prove that some portion of the protection system
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 273

has operated correctly since the last tests were  Region 2: The TBM intervals that are adjusted up or
performed. If specific protection scheme components down based on results of analysis of maintenance history
have demonstrated correct performance within of statistically significant population of similar products
specifications, the maintenance test time clock may that have been subject to TBM.
be reset for those components.  Region 3: Optimal TBM intervals based on regions 1 and
2.
2. Performance-based maintenance (PBM) –
Maintenance intervals are established based on
analytical or historical results of TBM failure rates on TBM
a statistically significant population of similar
components. Some level of TBM is generally
followed to justify continued use of PBM-developed
extended intervals for low occurrence of test failures
or in-service failures.
1 2
3. Condition-based maintenance (CBM)– Continuously
3
or frequently reported results from non-disruptive self
monitoring of components demonstrate operational
status as those components remain in service.
CBM PBM
Whatever is verified by CBM does not require
manual testing, but taking advantage of this requires
precise technical focus on exactly what parts are
included as part of the self diagnostics.
Microprocessor-based protective relays perform
continuous self-monitoring to verify correct operation
Figure 1 - Relationship of time based maintenance types
of most components within the device. Self-
monitoring capabilities may include the ac signal
Regulatory Considerations
inputs, analog measuring circuits, processors and
memory for measurement, protection, and data
The effectiveness of a maintenance program is often tracked
communications, the trip circuit monitoring, and
and determined by the asset owner. These tracking
signals for protection and data communications. For
mechanisms allow the owner to optimize the program based on
those conditions, failure of a self-monitoring routine
the application. However, when the performance of a
generates an alarm and may inhibit operation to avoid
protective device or system has a greater impact to the overall
false trips. When internal components, such as
power system, the proper performance of the protective system
critical output relay contacts, are not equipped with
will benefit the entire electrical grid which may be composed
self-monitoring, they can be manually tested. Method
of several interconnected power systems, or in some cases
of testing may be local or remote, or through
possibly part of a regulatory controlled electric system. In
inherent performance of the scheme during a system
some countries or electrical grid systems, the regulatory
event.
agency may require evidence that the asset owners each have
established a prudent maintenance program that meets a
The TBM is the overarching maintenance process of which the
minimum level of standards and practices established by the
other types are subsets. Unlike TBM, PBM intervals are
regulatory body. Furthermore, the regulatory agency may
adjusted based on good or bad experiences. The CBM
require evidence that the owner is performing maintenance
verification intervals can be hours or even milliseconds
based on the program. Depending on the level of regulatory
between non-disruptive self-monitoring checks within or
oversight, the asset owners may be asked periodically to either
around components as they remain in service.
voluntarily, or through a cohesive and collective established
audit process, or both, demonstrate that they have met their
TBM, PBM, and CBM can be combined for individual
obligations to maintain the protection and control system.
components, or within a complete protection system. Figure 1
illustrates the relationship between various types of
In a regulatory system, the owner may be asked to certify that
maintenance practices described in this section. In the Venn
they can:
diagram, the overlapping regions show the relationship of
TBM with PBM historical information and the inherent
 Demonstrate and submit detailed maintenance records and
continuous monitoring offered through CBM. This figure
that asset owners are cognizant that regulatory agencies
shows:
may require verification of the maintenance program
including evidence that protection systems and
 Region 1: The TBM intervals that are increased based on
components are being maintained and tested per the
known reported operational condition of individual
owners program.
components that are monitoring themselves.
274
 Submit a program documenting the methodology or
philosophy behind the owner’s program.
 Show a tracking mechanism and archiving system for the
maintenance records related to the protection and control
equipment and systems.
V. TEST REQUIREMENTS FOR PROTECTION SCHEMES AND
SYSTEMS
A. Line protection testing
Relay protection schemes for transmission lines can be
generalized into nonpilot and pilot protection schemes. The
nonpilot relaying system is used on radial transmission lines or
other lines where high-speed tripping is not required. It is also
used in conjunction with a pilot relaying scheme as a backup
line protection. The most commonly used relays in nonpilot
schemes are phase and ground overcurrent and stepped
distance relays. Refer to the normative references in IEEE Std
C37.113-1999 for further details.
The types of line protection covered under this section include:
 Non-pilot protection schemes
o Overcurrent relaying
o Distance relaying
 Pilot Protection schemes
 Directional comparison blocking and unblocking
schemes
 Direct underreaching transfer trip scheme
 Permissive overreaching transfer trip scheme
 Permissive underreaching transfer trip scheme
 Phase comparison relaying scheme
 Line current differential scheme
 Direct Transfer trip
 Automatic reclosing schemes
Throughout the life cycle of a relay, the relay is subjected to a
certification or factory acceptance test, application test,
commissioning test, and preventive maintenance test.
Certification test
Certification tests may be conducted in the relay supplier
factory or at another independent facility to verify the
performance of the relay against established parameters and
specifications.
Application test
The application test typically involves bench testing of the
relay to confirm that its elements are in working order for a
given application before the relay gets installed at a substation.
Relay schemes and logic can also be tested in the laboratory
prior to field implementation. Transient tests using digital
simulators have been developed mainly to emulate the
traditional model power system concept in order to evaluate
protective relay response to power system transients. Nonpilot
as well as communication-based transmission-line relaying
schemes (i.e., current differential, DCB, etc.) can be tested in a
laboratory by injecting simulated voltage and current
waveforms into relays under test. Relay input signals are
generally derived from transient simulation software tools.
Transient simulation techniques also provide tools for
evaluating the overall performance of protective relaying
schemes because they assist in testing the hardware, relay
algorithms, settings, configuration, speed of operation, and
transient performance of the scheme. Transmission-line relay
evaluation is based on line topology and length—short-line,
medium-line, and long-line models. All three transmission-line
models may have a similar topology and may consist of
parallel lines to introduce mutual coupling between them.
Simulation of a strong source at one end with a relatively weak
source at the remote end will assist in comparing relay
performance under different conditions. Relays are required to
be tested for internal and external faults during a maximum
power flow situation. Faults at different inception angles will
demonstrate relay performance under different dc offset
conditions. Likewise, varying fault resistances will also
validate relay response. Relays are also to be tested for
evolving faults and current reversal conditions. It is desirable
to evaluate relay performance under stressed conditions, such
as CT saturation and CCVT transients.
Relays designed for series-compensated line are often tested
with system models incorporating a series compensated line
with the metal-oxide varistor (MOV) and bypass breaker. It is
desirable to verify the performance of communication-
dependent relay systems with degraded or corrupted
communication networks. Some of the commonly practiced
methods include the injection of noise in audio-tone
communication systems or introduction of bit-error rates in
digital communications networks.
Real-time or model-based power system testing provides a
measure for evaluating the overall dependability and security
of the scheme. In the case of transmission-line protection
performance evaluation, these tests can be used to validate
performance on heavily loaded long lines, lines with series
capacitors and shunt reactors, especially for bulk transmission
applications. Model-based power system testing is practiced
where the response of the schemes cannot be evaluated
analytically or by conventional test methods due to the
complex interaction of various power system components
during faults and the high-speed communications schemes
required. In addition, model-based power system testing
provides a means of thoroughly investigating the transient
performance of the relay system without subjecting the system
to primary fault condition (such as stage tests).
It is important that the instrument transformers used in the
model have accuracy class such that the relay system burdens
do not cause errors in the magnitudes or distortion of
waveforms of the currents and voltages of the model during
testing. This does not preclude changes to permit investigation
of performance on saturated waveforms.
Transient simulation tools such as the Electromagnetic
Transient Program (EMTP), the Alternative Transient Program
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
(ATP), or the EMTDC allow the user to prepare relay signal
playback via power amplifiers in automated fashion repeatedly
and with varying system conditions (i.e., fault resistance, fault
inception angle, etc.) if desired. Development of an adequate
transient power system model is required to generate signals
for different applications. Transient tests are sometimes
referred to as dynamic or application tests.
For more thorough or complete application tests, “closed-
loop” digital simulators can be used. Closed-loop tests require
transient model and power system simulation where output
signals from the simulation are used as inputs to the relaying
system under test. Relay outputs responding to these signals
can then be fed back to the simulated power system, in turn
changing the model configuration. This mechanism provides
the means for testing both the relay system’s response to
power system disturbances as well as the modeled power
system’s response to the relay operation. Therefore, closed-
loop tests can provide a realistic testing environment for a
relaying system but require hardware to handle exchange of
many input and output signals during a simulation. Closed-
loop simulator software tools must also provide adequate
means for accurate transient modeling, simulation control, and
reporting results.
B. Transformer protection schemes
For transformer protection, several types of protective schemes
are commonly applied, including current differential, time
overcurrent, overexcitation, and sudden pressure. The
transformer protective relay location and technology applied
may also influence the type of protection design and testing.
For example, consider a two-windings transformer when some
level of protection for each high-voltage winding is located in
the respective control building, and the buildings are large
distances apart. Fiber optics may be used as a means of
interface between protective devices in the different buildings
to communicate decisions or information between the devices.
There are also transformers that are part of a transmission line,
and testing may involve communication equipment; refer to
the Guide for additional information. When redundant
equipment of communication interfaces is designed, testing
often is conducted with one complete system at a time.
Transformer differential protection testing
During commission testing, company personnel conduct visual
inspections of the relays, wire connections, and design
schematics. There are also several different current
transformer tests performed in order to verify proper CT
condition and connections. These tests are done when the
transformer is out of service. By applying voltage and current,
several tests are performed including ratio, polarity, and
saturation. A ground test is performed to check for undesired
grounds.
Relay testing is done by applying settings and verifying the
desired outcome. Typical tests include differential/slope,
harmonic, minimum pickup, voltage protection, and sudden
pressure. For the differential/slope test, two current sources are
used to inject current into single-phase relays or three-phase
275
relays with no vector compensation. It is suggested that for
three-phase numerical transformer differential relays that have
vector compensation settings, especially those that involve
zero-sequence removal, three phase current injection into both
two relay windings be performed. The restraint coil is tested
by applying a low mismatch current to see whether the
restraint coil blocks the relay from tripping. A high-mismatch
current is then applied to the operate coil to verify it trips the
external devices. A harmonic test includes simulating a second
and a fifth harmonic current to verify the relay(s) will not
operate for overexcitation or inrush current. Relays that
employ waveform recognition, for example, to block the
differential function from operating during inrush conditions,
can be tested by use of COMTRADE files or real-time
simulation techniques. Similar types of test file cases could
generally be applied to devices that do not have harmonic
waveform recognition.
For relays with internal compensation, three-phase testing will
simplify the evaluation of the relay performance.
For minimum pickup tests, a current source is used to inject
current into the overcurrent relay(s) to verify the threshold for
tripping. The timing of the overcurrent relay in a transformer
protection scheme is typically slower and is used as a backup
protection to the current differential relay.
Voltage tests verify operation for voltage-dependent elements.
Where overexcitation is applied, system testing may involve
the testing of threshold set points under steady-state conditions
as an initial set of tests followed by simulated volts/Hertz
(V/Hz) conditions using transient simulation tools or use of
COMTRADE files from system events to verify the
overexcitation performance for a generator step-up transformer
or for a true power system condition with high magnitudes of
volts/Hertz. The sudden pressure relay is tested by applying
pressure to the pressure switch and observing the alarm and
trip contacts for proper operation. A manual trip is initiated to
test the auxiliary and lockout relays for proper operation of the
above relay schemes. Also, SCADA tests are applied to ensure
all alarms and targets are operating properly.
Maintenance tests for transformer protection may need to be
conducted with the transformer energized. Therefore, the
proper isolation of the elements under test is critical. It is
important to keep some levels of overlapping protection in
service when maintenance tests are performed with the
transformer energized. The tests involve periodic checking of
the transformer protective scheme. The tests may not be as
comprehensive as the commissioning test; however, they
cover checking out all critical functions of the protection
system. Steady-state tests are widely used when applicable for
relay periodic maintenance and involve the use of current and
voltage elements. Output contacts are also verified for proper
functioning.
276
C. Distribution protection
Similar to transmission protection, testing of modern
distribution devices and schemes requires the availability of a
set of tools that will simplify the testing process, and at the
same time, it will prove the high quality of the testing process.
The functionality of modern state-of-the-art distribution feeder
protection relays includes the following:
 Basic fault protection
 Advanced protection schemes
 Abnormal system conditions protection
 Load-shedding
 Automation
 Monitoring
 Recording
 Analysis
The testing of distribution protection schemes is intended to
evaluate their performance under different fault, system, and
breaker conditions.
Different tests are designed to monitor the relay operation for
fault conditions, such as follows:
 Faults on the protected feeder
 Faults on an adjacent feeder
 Distribution bus faults
 No fault
Because the distribution protection logic and schemes are
tested in this condition, the relay reaction to the receiving of
correct control signals under the above-listed fault conditions
is tested as well. These test values can be used as part of both
commission and routine maintenance tests.
D. Shunt capacitor protection
Testing of relays for shunt capacitor banks is not much
different than testing normal overcurrent relays. Some smaller
banks are fused for fault protection. Depending on the size and
configuration of the bank, there may be an electronic
controller or separate relays installed to accomplish some of
the following functions:
 Bank unbalance detection due to failed capacitors
 Bank bus fault protection
 Under/over voltage protection to automatically switch the
bank on and off
 Capacitor discharge timer
The testing of these functions for a new installation or for
periodic maintenance is quite similar.
E. Bus protection
By virtue of the application, bus differential protection can be
tested at a single location. A bus differential protection scheme
can be represented as summation differential using simple
nondirectional electromechanical induction disk-phase time
overcurrent relays (instantaneous elements are generally not
used). Another scheme includes the solid-state high-impedance
differential. The most recent practice has been to use
microprocessor-based multifunction relays, enabling the phase
time overcurrent (TOC) elements emulating an
electromechanical induction disk relay.
The tests performed on the phase TOC installations are as
follows:
 Physical examination of each individual relay’s
components (control spring, contact condition).
 Calibration tests including minimum pickup and timing
tests.
 Functional tests, manually initiating trip contacts to
operate the auxiliary lockout relay.
 Phasing tests to verify the current transformer
connections, including a check to make sure the current in
each phase is zero with all feeder contributions from the
bus included; then checking each feeder’s contribution by
removing its contribution, one at a time, and verifying that
its contribution shows up in each phase.
For the high-impedance bus differential, the tests are as
follows:
 Physical examination.
 Voltage pickup, taking care to avoid leaving voltage
applied to the relay for longer than 10 s.
 High-set overcurrent unit, when applicable.
 Functional tests by manually firing the internal SCR to
operate the auxiliary lockout relay.
 Shorted bus differential CTs circuits. The bus differential
scheme design may have provisions to disable the bus
differential protection on short-circuit detection.
The following tests may also be necessary:
 Accuracy pick-up current (certification test, acceptance
test, and commissioning test).
 Accuracy operating characteristic (certification test and
acceptance test).
 Influence frequency and harmonics on accuracy pickup
current and operating characteristic (certification test).
 Operating times (certification test, acceptance test,
commissioning test, and application tests).
 Stability for load current and through current faults
(certification test, acceptance test, commissioning test,
and application test).
 Influence CT saturation with and without remanence on
operating time, including different types of transformers in
the bays (certification tests and application tests).
 Influence CT saturation with and without remanence on
stability for through current faults, including different
types of transformers in the bays (certification tests and
application tests).
 Influence evolving faults and evolving external to internal
faults on stability and operating times (certification test
and acceptance test).
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
 Verifying correct operation of breaker failure protection
(certification test, acceptance test, and commissioning
test).
 Selectivity bus section operation (certification test,
acceptance test, and commissioning test).
Other types of bus differential protection schemes, such as
those that use paralleled CT connections as input to the bus
differential, and most low impedance schemes that use
individual restraint inputs, can also be verified for open and
shorted CT connections.
Additional commissioning and maintenance testing
considerations are included in IEEE Std C37.103TM-
2004.
F. Breaker failure protection and control
Circuit breakers can fail to trip for various reasons. When this
happens, breaker failure protection detects the continued
presence of the fault and trips adjacent breakers to isolate the
fault after a predetermined length of time. Breaker failure
protection is used in transmission and distribution systems
where delayed backup clearing can threaten system stability or
cause outage of too many circuits because of remote backup.
Fast fault clearance and minimization of outages also improves
power quality.
Breaker failure protection, when properly designed, applied,
tested, commissioned, and maintained, enhances power system
reliability and stability. However, when it misoperates, it can
easily have disastrous consequences; hence, the security of the
breaker failure protection system is of paramount importance,
and comprehensive testing is required for any breaker failure
protection system. Refer to IEEE Std C37.119TM.
The testing of any protection function should be performed in
a way that matches as closely as possible real-life power
system conditions. The sequence of steps in a test is a function
of the requirements for the specific scheme and system
condition. This is especially important in the case of breaker
failure protection because of the importance of this function as
well as because it can be implemented in many different ways.
Breaker failure protection testing requires a good
understanding of the operating principles that are used in
developing the function implemented in the protection under
test.
G. Reactor protection
Reactor protection has generally consisted of one or more
protection functions. The protection type chosen is generally a
selection based on the size of the reactor and importance of the
reactor to the power system. Typical protection schemes
include high-impedance current differential, current
differential, phase overcurrent, negative sequence overcurrent,
ground overcurrent, voltage unbalance, overvoltage,
Volts/Hertz, and distance relays. In addition to specific
277
functions provided by the relays, some reactors have a pole
disagreement scheme to ensure all three phases are energized.
Pole disagreement schemes can be provided through a
combination of overcurrent elements and breaker contact
logic. Reactor protection and control schemes may be
integrated with an automated trip and insertion scheme to aid
system operators with the burden of managing system voltage.
Refer to IEEE Std C37.109TM-2006 or shunt reactor
protection guide.
Testing of the reactor protection will be similar regardless
whether it is a voltage control, current limiting, or a line
reactor. The most thorough testing is completed at the time of
energizing new equipment. The testing should include all the
equipment that the relay is wired to or can be expected to
impact. Because the relay action is based on currents and
voltages, it is imperative that the relay input sources are tested
and the performance characteristics of the sources (i.e., current
transformers and potential transformers or CCVTs) are known.
A wide range of simulation tests covering common to
uncommon faults as well as various operating conditions
should be performed. If the power system configuration or
conditions will allow, all breakers energizing the reactor
should be tripped at least once to verify the continuity of the
trip path.
H. Generator protection
Generators are a key part of a power system. It is most critical
that generator protection systems operate properly to clear
faults and trip the machine for abnormal operating conditions.
Refer to IEEE Std C37.102TM-2006 for an ac generator
protection guide for further details.
Wide area schemes that employ generator tripping should
include that function in the testing of the overall scheme,
preferably at times when the generator is offline.
The following elements will require phase checks with the
machine at some load level to allow checks of the phase angle
between voltage and current: 21 V, 32 V, 40 V, and 51 V. It
should be noted that many microprocessor-based generator
protection relays will allow this check to be done using a
computer connected to the relay, with the appropriate
software, to allow the user to observe what quantities the relay
is measuring. This precludes the need to use external
equipment (phase angle meter, etc.) connected to test facilities
(switches, blocks, etc.) to verify phasing.
In addition to static tests (minimum pickup, slope, etc.),
differential elements should be tested according to IEEE Std
C37.103TM-2004.
The preferred method to verify performance of the out-of-step
protection function (device 78) is dynamic testing. If studies
are available, it is also preferable to test loss-of-field elements
(device 40) dynamically.
278
Extreme caution should be used when testing generator
protection on a machine that has a generator breaker between
the generator step-up transformer (GSU) and the transmission
system to avoid unnecessarily tripping the GSU by any backup
elements. Considering that there are different methods of
connecting a generator to the power system, different
protection and interlocking strategies are considered. For
example:
 When the generator is unit connected, the GSU is
connected to the generator and is likely shut down when
the associated protection is being tested.
 When there is a generator breaker between the generator
and the GSU, and different zones of protection are
applied.
In any case, the protective scheme test procedures often
incorporate considerations of the functions that overlap with
protection between the GSU and the power system. Some
protection schemes may include breaker failure, bus
differential and generator load reduction logic, and any
interlocking scheme that blocks generator protection from
tripping switchyard breakers when the isolation device is open.
Testing of the lockout relays (LORs) associated with generator
protection is common practice. Initial tests generally include
allowing the LOR to trip all intended devices (generator
breakers, field breakers, turbine stop valves, etc.). Some
generator protective functions may perform some control
action other than tripping the machine (e.g., volts/Hertz
operates voltage regulator to reduce excitation) and should be
allowed to perform this action at least on initial testing.
Reverse power protection can involve interlocks with turbine
stop valve limit switches. Testing this scheme can include
online testing, during which the turbine is tripped just prior to
taking a machine offline, waiting for the stop valves to close
and the machine to motor briefly, and then allowing the
reverse power relay to trip the machine (of course, with an
operator standing by in case the protection does not operate).
This provides a good test of the overall operation of the
reverse power protection as an entire scheme.
I. Trip circuit logic scheme
The trip circuit logic scheme includes the logic and circuits
needed to trip the required breakers in a substation in order to
clear a fault. In some cases, the reach of this scheme is not
confined to the substation but also to the remote substations.
Typically, this scheme can be found in bus and breaker failure
protections. When these two protections exist in the same
substation, it is not unusual that the trip circuit logic scheme is
shared.
J. Communication Testing - Power line carrier testing
Power line carrier equipment is used in many protection and
control applications (refer to IEEE Std 643). The most
common applications include pilot transmission line protection
such as blocking or unblocking schemes, direct transfer
tripping, phase comparison, etc. Other less common
applications include voice and data traffic in addition to the
protection, based on the bandwidth a trap is designed for, or
the type of coupling (single-phase or multi-phase) to the
transmission line. For the purpose of system testing, the
following section covers single-phase coupling. The same
techniques are used in phase-to-phase and three-phase
coupling.
The types of tests include:
 Line Trap
 Carrier Transmit Measurement Terminated and Bridged
 Carrier Receive Measurement Terminated and Bridged
 Measurement specification
 Amplifier impact of power line carrier testing
Line trap
This test requires the transmission or distribution line to be de-
energized. The trap tuning (resonant frequency set point) is
verified by checking impedance versus frequency. The center
frequency is tuned to the maximum impedance, measured
using the impedance meter and signal generator. When
verifying the trap frequency set point or points it is best to use
the highest impedance possible for the used spectrum. Refer
to the IEEE C 37.233 for additional details.
K. System Integrity Protection Scheme (SIPS) Test
Requirements
The SIPS encompasses Special Protection System (SPS),
Remedial Action Schemes (RAS) and schemes such as
Underfrequency (UF), Undervoltage (UV), Out-of-Step
(OOS), etc. These schemes provide reasonable
countermeasures to slow and/or stop cascading outages caused
by extreme contingencies. They stabilize interconnected
power systems by preventing overloading of the lines,
arresting voltage decline, initiating pre-planned separation of
the power system, etc.
The SIPS, armed for pre-defined outages, initiate pre-planned,
automatic, and corrective actions. Their design is based on
studies of pre-defined outages for variety of conditions.
SIPS cover a wide range in the scale of the individual
schemes. For example, a local area protection scheme may
provide thermal overload protection for a single transmission
line. Such local schemes may involve testing requirements and
procedures more analogous to those used for testing equipment
for fault protection in contrast to schemes designed to provide
wide area protection. Such a thermal SIPS may monitor
atmospheric or conductor temperature, wind speed and load
current, provide a thermal model of the protected equipment,
and send a signal to one or more locations to open line
terminals or shed load based on the application. A thorough
test program will challenge the specific functions of the
scheme design, but the limited scheme functionality will
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
naturally also determine the scale and complexity of the testing
necessary.
Within the Guide, there are examples and architectures for
comprehensive monitoring and arming systems. The
complexity is attributed to:
 Selection of various equipment
 Identification of monitoring points
 Types of alarms and priority classification
 Various contingencies associated with equipment
abnormal conditions
 Types and availability of real time data
 Considerations for various categories of input and output
tests
 Development of the test scenarios, coupled by provisions
for automated testing
One form of SIPS is wide-area protection and control type
schemes. Wide-area protection schemes typically cover large
geographic and/or electrical areas to protect against a wide
range of system problems. Test objectives typically include a
broader range and volume of equipment and system models
than for local schemes. For example, SIPS that remediate
transient stability conditions are based on extensive system
modeling, include detailed monitoring of critical facilities,
system load levels or other conditions, and may take a variety
of actions such as system islanding, load shedding, generator
shedding, and shunt capacitor and reactor switching.
Furthermore, such wide area protection schemes may involve
many different entities with different background and
practices. This complexity requires very stringent and detailed
test procedures to confirm both security and dependability of
those schemes.
System variables, measurement inputs, sources of current and
voltage location, status inputs and outputs plus types of control
signals are examined with comprehensive testing solutions to
various types of schemes.
The complexity of the scheme, its purpose, space availability,
and other factors may drive some of the decisions associated
with the scheme applied and the levels of tests to be
performed. Types of tests listed below are explained further in
the Guide:
 Proof-of-concept/lab tests.
 Field commissioning tests.
 Detailed system-wide performance tests.
 Validation through State Estimation.
SIPS performance can be more sensitive to system additions
and modifications than other protection systems. It is critical
that scheme tests include assessment of the system models
upon which the SIPS design is based. These assessments need
to be performed in conjunction with major system additions as
well as periodically, e.g. every five years, to include effects of
more routine system changes and growth.
279
VI. TESTING PROTECTION & CONTROL SYSTEMS WITH
UNCONVENTIONAL VOLTAGE AND CURRENT SENSING INPUTS
A. Testing inputs to relays and microprocessor-based devices
with low-level analog inputs per IEEE C37.92-2005.
The purpose of IEEE C37.92-2005 is to define a low-level
standard interface between relays, and unconventional sensors
which cannot easily generate the high-level signal outputs like
the conventional current and voltage instrument transformers.
The sensors are typically designed with analog electronic
signal generating circuits operating at less than ±15 V and
delivering signals to relatively high input impedances of
electronic circuits in the relays. Other types of sensors, such
as Rogowski coils, have passive low-energy outputs that are
also compatible with these same high-impedance electronic
inputs. This section discusses testing of the relays and their
inputs. Testing the sensor outputs is covered in the section on
testing outputs of unconventional sensors.
Relays and other microprocessor-based devices with low level
analog inputs per IEEE C37.92-2005 are typically
microprocessor based, with analog electronic input circuits
interfaced to analog-to-digital converters. Beyond the low-
energy input circuits for voltage and current signals, these
relays are identical to those designed for conventional
instrument transformer secondary signal inputs of 69 Volts or
120 Volts and 5 Amperes. The overall functional testing or
verification strategy is the same.
Instrument transformer standards IEC 60044-7 and IEC
60044-8 specify similar types of low-level interfaces for
voltage and current signals, although these standards list
several normal values in lieu of a single standard value.
Testing approaches are the same.
The major design difference is that the input isolating
transformers are wound differently (or deleted entirely) so that:
 The relay has a high impedance input (50k ohms typical)
and accepts a low-energy signal of about 200 mV RMS to
represent a primary current corresponding to 1 per unit
primary load. This is scaled so that the instantaneous
peak of a 20 per unit fault current fully offset will be
slightly over 11 volts, and within amplification range of
commonly used operational amplifier circuits.
 The relay has a high impedance input and accepts a low-
energy signal of about 4 V RMS to represent a primary
voltage of 1 per unit.
It is important to note that, with this type of low-level
interface, primary current signals are represented by a voltage
signal. This is fundamentally different from conventional
relays, whose current inputs require a CT secondary current
proportional to primary power system current.
Such a relay can be tested via injection testing methods that
are used for conventional relays, except that the signals must
280
be scaled to these lower standard levels, and the current input
signal must be converted to voltage across a burden resistor. If
the test set is based on microprocessor or computing
technology, the modification may include removing or
bypassing the power amplifier needed to drive conventional
relays needing large signal inputs from the low level electronic
signal sources in the test set.
It is easy to test these relays using a conventional relay test set
with a simple transformer based adaptor. Transformers and
interface circuits similar to or exactly like those installed in the
front end of a conventional microprocessor relay can be used
to build the adaptor. The normal voltage signal of 69 V RMS
is scaled with a small wound transformer or a resistive divider
to the 4 V signal used by the relay with IEEE C37.92 voltage
interface. The 5 A current signal is passed through a small
current transformer with a resistive burden, such that the
voltage developed across the burden is 200 mV. Alternatively
the test set 5 A current output can be directly connected to a 40
milliohm burden resistor, if such an accurate burden resistor is
available.
It is similarly easy to scale signals to several of the many
alternate low-energy signal levels given in IEC 60044-7 or
IEC 60044-8. Test signals for a relay having this interface are
also easy to generate with a personal computer and a low-level
electronic analog output – even a small electronic amplifier
connected to a sound card could produce adequate energy to
simulate faults. Real-time digital simulators can be connected
to these relays without power amplifiers. Some optical current
and voltage sensors with electronic analog outputs offer the
user the convenience of operating in a test mode generate 50
Hz or 60 Hz standard signals that can verify connections from
the sensor to the relay and relay measurement accuracy.
B. Testing inputs to relays with digital interfaces per IEC
61850-9-2 process bus.
This section of the Guide introduces process bus test
architecture using test merging units GPS clock
synchronization.
C. Auxiliary TMU
In this test configuration, a set of spare Merging Units (MU)
located in the control house are connected to the relay either
through an Ethernet switch or direct connected (as dictated by
the implemented process bus architecture). Current, voltage,
and status signals are then injected into the MU in the
appropriate form (standard analog signals or low level analog
signals) and the TMUs then process the inputs and stream
them accordingly to the connected Relay Under Test (RUT).
Note that the TMUs must be appropriately synchronized.
If mapped in the dataset, the Test bit in the Quality flag shall
be set – indicating to the RUT that the data being received is
Test data. If the RUT is also in Test mode, it shall process the
received data and provide protections and measurement values
as if connected to the real system. Outputs to controls are to
be executed as defined in the IEC 61850 standard.
D. PC-based test TMU
- This path uses a standard PC either connected to the Ethernet
switch or directly connected to the RUT. Note that in the
latter configuration, the PC may be required to support
multiple Ethernet ports/data streams to the RUT.
E. Primary/secondary injection
Another possible test mode is primary or secondary signal
injection. In this mode of testing, a signal is injected either
into the inputs of the existing MU or injected at the High-
Voltage interface of the process equipment.
GOOSE testing
It should be noted that although the process bus is designed
primarily for communication from a MU to a relay, there is a
need to send control signals (e.g. – breaker trip and close
signals) from the relay to the MU. In this circumstance, it is
logical that the IEC 61850 GOOSE mechanism be used over
the same physical medium. GOOSE messages would contain
status values that would be mapped into output contacts in the
MU.
In the first mode of operation, an output bit is set in the
GOOSE message and it is observed that the respective output
of the MU operates. Operation may be observed by actions
such as a breaker actually opening or closing or the operation
of an auxiliary relay.
In the second mode of operation, the TEST bit in the GOOSE
is set. In this mode of operation, the TEST GOOSE is sent
and the MU sets a Sequence of Events message to the effect
that a TEST GOOSE was received for a particular output
contact. The contact, however, is not operated. Refer to the
IEC 61850 standard for details.
The Edition 2 of the standards series IEC 61850 has made
some changes. One of the changes in the Edition 2 is related
with the test mode function.
“The test mode function allows the local or remote operator to
check at any time any function of the system using process
signals also but avoiding any impact on the process (blocking
of process outputs).”
To allow detection of simulated data on a low layer of
communication a “simulation-bit” (S) has been introduced into
the Ethernet frame. If this is set, S: when the bit is set, the
GOOSE message has been issued by a publisher located in a
test device and not by the publisher as specified in the
configuration file of the device. Simulation” is a Boolean
value (TRUE/FALSE) in the header of all GOOSE messages,
this parameter shall indicate with the value TRUE that the
message and therefore its value have been issued by a
simulation unit. The GOOSE subscriber will report the value
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
of the simulated message to its application instead of the “real”
message depending on the setting of the receiving IED. Refer
IEC 61850-8-1 Ed. 2, IEC 61850-9-2 Ed. 2 and IEC 61850 –
7-2 Ed 2 for details.
The annex A of the standard covers steps to follow detail for a
certain mode
 Output of status
 Response to controls
 How to process incoming data of different qualities (as
"test", blocked"…)
VII. TESTING OUTPUTS OF UNCONVENTIONAL SENSORS
The standard method for verifying the accuracy of an
unconventional sensor is the same as that used for
conventional CTs and PTs – a primary injection test with
accurate measurement of the output using accurate
instruments, or the relays and meters connected to its outputs.
In general, such injection testing is used only during
commissioning or apparatus-outage maintenance.
For a typical installation of an optical sensor, the optical losses
of all fibers are measured along with the losses of the optical
columns, using, for example, a hand held optical power loss
meter. After connecting the fibers with the optical columns
and the electronics the losses are measured by means of the
electronics itself. If both measurements are within acceptable
levels the unit is commissioned. Calibration is set at the
factory, but a field recheck is conducted with primary injection
as for a conventional CT.
Primary injection with some vendors’ optical current
transformers may be easier than for conventional CTs. The
effective transformation ratio of the optical CT can be changed
on site via software so that a lower primary current injection
can be used to achieve target output signal levels. This allows
testing the protection and instrument transformer together with
more compact primary-source current generators.
VIII. REVIEW OF ANNEXES
The IEEE C37.233 includes a comprehensive review of
supplementary information regarding protection system testing
in five annexes.
Annex A comprises of an informative bibliography of up-to-
date ANSI, IEEE, IEC and Cigré standards and reports,
conference papers and other technical articles, including North
American Electric Reliability Corporation (NERC) documents,
that supplement the normative references found in the main
body of the Guide.
Annex B includes a detailed example of a line current
differential (87L) protection scheme testing procedure, with
guidelines to assess the integrity of communication channels
used in the differential scheme; check for channel asymmetry;
testing of supervising and blocking elements, and stub bus
protection; testing of the differential characteristic, minimum
281
pickup and timing; assessment of 87L through fault security
and dependability; single-pole tripping applications; and 87L
on-load tests.
The impact of high impedance faults to protective relay
performance and system testing is discussed in Annex C. The
main types of power system grounding are discussed, as well
as their impact on the performance evaluation of the fault
location function in line protection relays. A fault-resistance
based classification of faults is presented, followed by a
discussion of the sensitivity limits and coordination
considerations of relevant protection elements, such as
negative and zero-sequence TOC and IOC, as well as ground
quadrilateral elements, to detect various types of resistive
faults. Transient simulation and short circuit programs are
suggested tools for generating test cases for ground distance
elements in particular. Annex C also discusses the efficacy of
some pilot schemes in detecting high resistance faults. For
testing such schemes, the use of GPS-synchronized tests and
playback of pre-recorded resistive fault simulations
(COMTRADE files) is the preferred evaluation technique.
Annex D presents transformer oil and winding temperature
computational methods using temperature profile assumed in
the IEEE transformer loading guide.
Finally, Annex E explains how time delays can be
compensated for when using relay test sets with different
performing characteristics for GPS-synchronized end-to-end
testing.
IX. CONCLUSIONS
The benefit of testing individual components is well known
from years of experience. Different types of schemes, local or
communication assisted, and their intended applications
determine the types and level of testing and verification of
interlocking. As modern power systems operate with narrower
margins and are becoming more complex, it has become more
common to apply SIPS and to consider protection and control
schemes for wide-areas of a power system. Some of these
schemes utilize reliable wide area communications that result
in changes in some basic philosophies of the protection and
control system. As a result, in addition to testing individual
protection and control equipment, it has become necessary to
test the overall protection and control systems.
The goal and the key benefit of testing the entire protection
and control system are to validate the performance of the
entire scheme. These tests are performed starting from the
individual components level to backup, redundant, and
auxiliary component. The purpose is to validate the
interaction between these components work correctly in terms
of maintaining the dependability and security of the scheme at
a desirable level of performance. As described in this paper,
the Guide encompasses overall system testing definitions and
procedures for generators, line, line reactors, transformer,
capacitors, special protection schemes, field commissioning
and end-to-end testing, distributed application within
substation, etc., as well as data collection requirements. It is
also important to include indirect trip applications, open /
closed loop tests, and dynamic / non-linear tests.
282

The presented approach is to verify basic behavior of scheme

components first, followed by testing of interconnected
components in a function-oriented assembly. The complete
approach includes design testing, commissioning testing,
performance assessment testing, and addressing various
maintenance practices such as time based maintenance,
performance based maintenance and condition based
maintenance.
As electric power systems continue to be more complex
(renewable and distributed generation, energy efficiency
requirements, electrical vehicles, etc.), it is expected that even
more focus will need to be put on wide area monitoring,
protection, and control schemes that require system testing.

X. ACKNOWLEDGMENT
This paper is based on the report created by IEEE Power
System Relaying Committee Working Group C 37.233. The
authors acknowledge contributions by the Working Group
members that have made this summary paper possible. IEEE
C 37.233 working group members are:

V.Madani (Chairman), H.DoCarmo (Vice-Chairman),

M.Achterkamp, M.Adamiak, M.Agudo, A.Apostolov, J.Ariza,
M.Begovic, R.Beresh, G.Brunello, D.Bui, S.Chilcote,
G.Colpron, R.Das, K.Gardner, D.Goodrich, A.Giuliante,
G.Henneberg, R.Hunt, B.Kasztenny, T.Kern, M.Kezunovic,
G.Kobet, P.McLaren, M.Mendik, K.Mustaphi, D.Novosel,
F.Rahmatian, A.Santos, T.Seegers, J.Sykes, E.Udren,
A.Vaccaro, B.Vandiver, J.Verzosa, I. Voloh

The authors also acknowledge feedback and suggestions from

the IEEE PSRC System Protection Subcommittee and Main
Committee members.

XI. REFERENCES

[1] IEEE Guide for Protection System Testing, IEEE

Standard C37.233-2009, Dec. 2009.
[2] IEC 61850- Ed.2:
Communication networks and systems for power utility
automation –
Part 7-2: Basic information and communication structure
– Abstract communication service interface (ACSI)
Part 7-4: Basic communication structure – Compatible
logical node classes and data object classes
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 283

IEC 61850 GOOSE APPLICATIONS TO DISTRIBUTION PROTECTION

SCHEMES
Dr. Alexander Apostolov Benton Vandiver

OMICRON electronics

USA
INTRODUCTION
The industry is going through significant changes due to the increased requirements
for improved quality of power supplied by the utility in order to avoid costly
interruptions of manufacturing or other processes caused by voltage sags, swells or
unbalanced conditions when a short circuit fault occurs in the distribution system.
The paper discusses the requirements for reduction in the duration of different short
circuit faults.
The behavior of typical distribution feeder protection or substation protection systems
is analyzed from the perspective of the definitions of voltage related power quality
events.
The use of IEC GOOSE in the development and implementation of distribution
protection schemes is analyzed from the point of view of improvement in the
performance of the distribution protection schemes and the reduction of the overall
fault clearing time.
The effect of changes of the system configuration on the performance of the
protective relays is then analyzed. Multifunctional protective IEDs are then
considered and it is demonstrated that by using all available protection and
programmable logic functions combined with the exchange of GOOSE messages
between different distribution protection IEDs, the user can significantly reduce the
effect of short circuit faults on sensitive loads supplied from the distribution
substation.
Adaptive protection based on detected changes in the system configuration,
combination of instantaneous, definite time and inverse time-delayed phase, ground
and negative sequence elements will result in significant reduction in the duration of
the fault. This will lead to changes in the voltage level/time characteristics of the fault
condition and reduced probability for the costly interruption of voltage sensitive
processes.
The following schemes using GOOSE messages are described:
• Distribution bus protection
• Selective backup tripping
• Sympathetic tripping scheme
• Breaker Failure protection
The requirements and tools for testing of such schemes are described at the end of
the paper.
284

EFFECTS OF SHORT CIRCUIT FAULTS

The prolonged effect of short circuit faults on sensitive equipment supplied by
distribution feeders can lead to their failure and significant losses. This is pushing the
requirements for the performance of distribution protection systems and making
them similar to transmission protection systems requirements.
The improvement of power quality during short circuit faults can be achieved in
several different ways. Like any other problem that has to be solved, we need first to
understand the nature of the problem and its effect on sensitive users. The most
common short circuit faults in the system – single-phase to ground faults – are
characterized by the fact that they introduce a voltage sag in the faulted phase, and
at the same time they result in a voltage swell in the two healthy phases. This is
clearly seen in Figure 1 that shows the recorded waveform and the voltage phasor
diagram for a single-phase to ground fault.

Figure 1 Phase voltages for a single-phase-to-ground fault (Phase A)

The case of two or three-phase faults is quite different. For three-phase faults all
phases experience a voltage sag, while for a two-phase fault - the two faulted
phases will have lower voltages, with the healthy phase without a significant change
compared to the pre-fault levels.

Fig. 2 ITI (CBEMA) curve from a manufacturing plant

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 285

Fig. 2 shows a plot of depth vs. duration of actual cases from a high-volume
manufacturing plant, with some of them resulting in process shutdown due to
variable speed drives and vacuum pumps failures.
There are several factors that determine the voltage level during a short circuit fault
on the transmission or distribution system:
• System configuration
• Fault location
• Fault resistance
The first characteristic of a voltage sag – the depth – is something that we can’t
control, but we have to study in order to be able to predict or estimate the effects of
different faults on the sensitive equipment.
The second characteristic of the voltage sag – the duration – is the parameter that
we can control by properly applying the advanced features of state-of-the-art
multifunctional distribution feeder protection relays. The focus of this paper is the
impact of IEC 61850, and especially the use of GOOSE messages in distributed
protection schemes that can reduce the fault clearing time in distribution substations.

DISTRIBUTED PROTECTION APPLICATIONS

Peer-to-peer communications are used to perform protection, control, monitoring and
recording functions. Any function can be divided into sub-functions and functional
elements.
The functional elements are the smallest parts of a function that can exchange data.
These functional elements in IEC 61850 are called Logical Nodes. When a function
is executed based on the exchange of communications messages between two or
more devices, it is called “distributed function”.
The exchange of data is not only between functional elements, but also between
different levels of the substation functional hierarchy. It should be kept in mind that
functions at different levels of the functional hierarchy can be located in the same
physical device, and at the same time different physical devices can be exchanging
data at the same functional level.
Figure 3 shows Logical Connections (LC) - the communication links between
functional elements - in this case logical nodes of the P and R groups. IEC 61850
also defines interfaces that may use dedicated or shared physical connections - the
communication links between physical devices.
The allocation of functions between different physical devices defines the
requirements for the physical interfaces, and in some cases may be implemented
into more than one physical LANs.
286

Bay
computer P..
IF 8

LC2 R...
IF 8
Protection LC1
IED P...

R...
Protection
P... IED

Distributed
function

Fig. 3 Distributed Function definition in IEC 61850

The functions in the substation can be distributed between IEDs on the same, or on
different levels of the substation functional hierarchy. IEC 61850 defines three such
levels:
• Station
• Bay/Unit
• Process

Fig. 4 Logical interfaces in Substation Automation Systems

 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 287

These levels and the logical interfaces are shown by the logical interpretation of
Figure 4. IEC 61850 focuses on a subset of the interfaces shown in Figure 4 with
Interface 8 (shown in red) being used for high-speed peer-to-peer communications.
The logical interfaces IF8 is defined as direct data exchange between the bays
especially for fast functions like interlocking.

Distribution Bus Protection

The protection and control in substations is distributed in nature by the fact that each
protective relay is designed in general to provide primary protection of individual
substation equipment such as transmission and distribution lines, transformers,
capacitor banks, etc.
The only substation equipment that requires a centralized form of protection in
conventional systems is the busbar. Transmission buses are typically protected by
bus differential protection relays. They require current signals from each primary
equipment connected to the bus to be available at the central location of the bus
differential protection. The scheme becomes much more complicated and expensive
if the current transformer ratios are different. Things get even worse if the bus
differential protection is used in a substation where the bus configuration may
change.
Because of the high cost and the increased requirements for maintenance, in many
cases bus differential protection is not installed on distribution or sub-transmission
buses. As a result, bus faults are cleared by backup relays with longer fault clearing
times caused by the need for time coordination between the distribution feeder
relays and the transformer relays. This becomes a significant power quality problem
because of the increased duration of voltage sags.
Multiple protective IEDs with IEC 61850 GOOSE can be connected to the substation
LAN and used in distributed bus protection applications for distribution systems.
In case of a fault on any of the protected feeders (F1 in Figure 5), the feeder
protection IED will see a fault. The same fault current will be seen by the
transformer protection IED. As soon as the overcurrent protection element of the
feeder relay starts, the IED will send a GOOSE message indicating the detection of a
fault on the feeder.
The transformer protection IED subscribes to GOOSE messages from all feeder
relays. When it receives the message indicating that there is a fault on one of the
feeders, the overcurrent protection element that is used for bus protection is blocked.
If the fault is on the bus (F2 in Figure 5), no feeder IED will see a fault, the
transformer protection IED is not going to receive a GOOSE message indicating a
feeder fault. This indicates a bus fault and the relay is going to trip the transformer
breaker to clear the fault.
The peer-to-peer communications based bus protection requires an operating time
for the fault detection of about one cycle for the relays involved. The addition of 0.25
cycle (4-5 ms) for the communication message and the safety time delay of 0.75
cycle in the transformer protection relay ensures a total operating time of about 2
cycles.
288

Fig. 5 High speed peer-to-peer communications based distributed bus protection

The benefit of this scheme is that instead of clearing the bus fault with the long time
delay of a coordinated backup transformer protection, the only time delay required
will be the longest possible overcurrent element starting time plus a safety margin.
The benefit of the peer-to-peer communications based distributed bus protection is
that it provides fast fault clearance for distribution bus faults without the need for any
additional protection equipment.

Sympathetic Trip Logic

The changes of fault conditions in the distribution system impact not only the
sensitive loads, but also depending on the load may lead to the undesired operation
of protection elements of multifunctional relays on healthy feeders.
Detecting the operation of a relay on adjacent feeder can be use to adjust the
sensitive settings of the relays on the healthy feeders for the duration of an inrush
condition following the clearing of a fault in a distribution system with a significant
number of motor loads.
This is known as a Sympathetic Trip scheme. As soon as a relay detects a fault on
the feeder that it is protecting, it sends a GOOSE message to all other feeder relays
informing them to expect an inrush as a result of the voltage recovery following the
clearing of the fault.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 289

Fig. 6 Sympathetic trip protection

Each of the relays on the healthy feeders subscribes to GOOSE messages from all
adjacent feeder protection IEDs and when it receives a message indicating adjacent
feeder fault, it adapts its settings for the period of time that the expected inrush
condition is going to last. Two options are usually available:
• block the sensitive overcurrent setting
• reduce the sensitivity by increasing the pickup setting for the duration of the
inrush
The benefit of using GOOSE messages in such a scheme is that instead of the large
number of required wires between the binary inputs and relay outputs of all
distribution feeder protection IEDs, the just need to publish and subscribe to GOOSE
messages from the adjacent feeders' IEDs.

Selective Backup Tripping

The common approach that many utilities have taken is to use a single protection
IED on a feeder. In case of failure of this relay, faults on the line are cleared by the
backup overcurrent protection on the transformer or sectionalizing breaker.
The problem with this approach is the long fault clearing time that may affect
sensitive loads fed by the distribution substation. A solution that significantly reduces
the duration of the fault is based on the adjustment that the backup relay can make
in its decision to trip based on the knowledge that a specific IED has failed.
This adaptive form of protection uses the normally closed contacts of the feeder
relays that close when the relay is not healthy. When the transformer of
sectionalizing breaker relay sees a fault and does not get any blocking signal from
any of the feeder relays, it knows that there are two possible cases:
290

• The fault is on the feeder with the failed relay

• The fault is on the distribution bus
Since the probability for a fault on a distribution feeder is much higher that the
probability for a distribution bus fault, the relay first sends a signal (1) to trip the
breaker of the failed relay. If this does not clear the fault, then it is clear that the fault
is on the bus and it is cleared by tripping the source breakers with signals (2).

Fig. 7 Selective backup tripping

The conventional implementation of this scheme is based on the use of the normally
closed contact of an output relay that closes when the relay fails. This
implementation requires hardwiring between all feeder relays and the dedicated opto
inputs of the transformer relay.
The IEC 61850 GOOSE repetition mechanism can be used to eliminate the need of
the above described hard wiring. If the transformer protection IED subscribes to
GOOSE messages from all feeder protection IEDs, within the maximum repetition
time interval it will receive a GOOSE message from all healthy IEDs.
If one of the feeder protection IEDs fails, it will stop sending GOOSE messages. This
will cause the enabling of the selective backup trip logic in the transformer protection
IED.

Breaker Failure Protection

Breaker failure protection is a scheme that is typically used at the transmission level
of the system due to the impact of such event on the stability of the electric power
system. With the availability of built in breaker failure protection function in many
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 291

multifunctional protection IEDs and the increasing requirements for decrease in the
duration of distribution faults.
The distributed breaker protection scheme can be implementation using two different
approaches depending on the location of the breaker failure detection element.

Fig. 8 Breaker failure protection

In the first case the breaker failure protection element is in the multifunctional
transformer protection relay. When the distribution feeder protection relay operates,
it sends a GOOSE message indicating the change of state of any of the protection
functional elements.
The transformer protection relay subscribes to this message, and when it receives
the change of value of a feeder protection functional element Operate data object to
True, initiates the breaker failure protection function. If the breaker fails to trip, the
fault current will keep the level of the current above the pickup setting of the breaker
failure detection element, the timer will time out and the relay will trip the required
breakers to clear the fault as shown in Figure 8.
Another implementation of the scheme is based on a built-in breaker failure
protection in each of the distribution feeder protection IEDs.
In this second when the distribution feeder protection relay operates, it initiates the
built-in breaker failure protection function. If the breaker fails to trip the breaker
failure protection function will operate and send a GOOSE message indicating the
change of state of this protection functional element.
The transformer protection relay subscribes to this message, and when it receives
the change of value of a breaker failure protection function element Operate data
object to True, will trip the required breakers to clear the fault as shown in Figure 8.
292

Functional Testing Of IEC 61850-8-1 And IEC 61850-9-2 Based Bay And
Substation Level Distributed Applications

The testing of distributed protection functions that are based on IEC 61850 GOOSE
are similar functionally to the testing of hardwired schemes. The main difference is
that in this case the test devices need to be able to act as IEC 61850 devices, i.e. to
be able to publish and subscribe to GOOSE messages.
If the distributed scheme includes devices located remotely from each other in the
substation, we may need multiple test devices with virtual simulators or analog
outputs. The simulation of the substation and system environment required for the
functional testing of bay and system level functions will require the simulation of
multiple IEDs.
A test system designed for IEDs or distributed applications based on IEC 61850
have multiple components that are needed for the testing of the individual functions,
as well as a complete application. A simplified block diagram of such a system is
shown in Figure 9.

Fig. 9 Test system/configuration tool, simplified block diagram

The first component of the test system is the test Configuration Tool. It takes
advantage of one of the key components of the IEC 61850 standard – the Substation
Configuration Language. The Configuration Tool is used to create the files required
for configuration of different components of the test system. It imports different
configuration files defined by Part 6 of IEC 61850.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 293

The test system Configuration Tool reads the information regarding all IEDs,
communication configuration and substation description sections. This information is
in a file with .SCD extension (for Substation Configuration Description) and is used to
configure the set of tests to be performed.
The overall functionality of any IEC 61850 compliant device is available in a file that
describes its capabilities. This file has an extension .ICD for IED Capability
Description.
The IED configuration tool sends to the IED information on its instantiation within a
substation automation system (SAS) project. The communication section of the file
contains the current address of the IED. The substation section related to this IED
may be present and then shall have name values assigned according to the project
specific names. This file has an extension .CID (for Configured IED Description).
The second component of such a system is a Simulation Tool that generates the
current and voltage waveforms. The specifics of each simulated test condition are
determined by the complete, as well as the configured functionality of the tested
device or application.
The simulation tool requirements will also be different depending on the type of
function being tested. For example, if the tested function is based on RMS values or
phasor measurements, the simulation tool may include a sequence of steps with the
analog values in each of the steps defined as Phasors with their magnitude and
phase angle. Based on these configuration parameters the simulation tool will
generate the sine waveforms to be applied as analog signals or in a digital format to
the tested components or systems.
If the tested functions are designed to detect transient conditions or operate based
on sub-cycle set of samples from the waveform, an electromagnetic transients
simulation will be more appropriate.
The third component of the test system is the Virtual IED simulator that is used to
represent components of the system that are not available at the time of testing, for
example during factory acceptance testing. During the testing this module send
GOOSE messages that the function or Subfunction under test uses as inputs that
determine its behavior under the test conditions applied.
The fourth component of the test system is the Test Evaluation Tool that includes the
monitoring functions used to evaluate the performance of the tested elements within
a distributed sampled analog value based system. Such evaluation tool requires
multiple evaluation sub-modules that are targeted towards the specifics of the
function being tested. In our case they are based on monitoring the GOOSE
messages from a tested IED.
The fifth component of the test system is the Reporting Tool that will generate the
test reports based on a user defined format and the outputs from the simulation and
evaluation tools.

CONCLUSIONS
The application of IEC 61850 GOOSE messages allows significant improvements in
the protection of distribution substations that reduce fault clearing times and
minimize the effect of short circuit faults on sensitive loads.
294

Using such high-speed messages eliminates the need for multiple hard wired
connections.
In some cases the implementation of a hard-wired distribution protection scheme
(such as sympathetic trip logic) in a large substation requires also that all protection
IEDs have a significant number of binary opto inputs and relay outputs. The
publisher/subscriber mechanism used with GOOSE messages eliminates this
problem.
The testing of distributed applications requires the use of testing equipment that can
simulate and subscribe to GOOSE messages. In some cases multiple test devices
need to be synchronized in order to test a distribution protection scheme.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 295

IEC 61850 - Impact on Protection Systems

M.C. Janssen, UTInnovation, The Netherlands

substation level are described. The distribution of signal and

Abstract – IEC 61850 Communication Networks and Systems data processing functions between the different devices are
for U tility A utomation is a standard f or c ommunications that analyzed.
creates an environment t hat w ill allow significant changes in the Some specific substation applications based on GOOSE
way t he po wer s ystem i s pr otected a nd o perated. This paper messages and sampled analog values are described and
describes the i mpact o f I EC 6 1850 o n th e protection of power demonstrate the advantages of the new technology.
systems.
Merging U nits c ommunicating o ver E thernet w ith multiple
Improvements in functionality, combined with practical
devices, the u se of G OOSE m essages for t rip a nd block elimination of many performance or safety issues are also
functionality o r a central c omputer t hat re ceives a nd processes covered in the paper.
current a nd v oltage samples with a rate of 2 56 samples/cycle
allow t he i mplementation of a ll t ypical protection, automation, II. CONVENTIONAL SUBSTATION DESIGN
control, monitoring and recording functions in an efficient way.
In addition t hese c oncepts can a lso be us ed o utside of t he
Conventional substations are designed using standard
substation allowing the i mplementation of Wide Area Protection design procedures for high voltage switchgear in combination
using standardized communications. with copper cables for all interfaces between primary and
secondary equipment.
Index Terms— IEC 61 850, protection, w ide a rea p rotection, Several different types of circuits are used in the
substation automation systems, process bus, station bus. substation:
• Analog (current and voltage)
I. INTRODUCTION • Binary – protection and control signals
The IEC 61850 standard Communication Networks and • Power supply – DC or AC
Systems for Utility Automation allows the introduction of new
designs for various functions, including protection inside and
outside substations. The levels of functional integration and
flexibility of communications based solutions bring significant
advantages in costs at various levels of the power system. This
integration affects not only the design of the substation but
almost every component and/or system in it such as
protection, monitoring and control by replacing the hardwired
interfaces with communication links. Furthermore, the design
of the high voltage installations and networks can be
reconsidered regarding the number and the location of
switchgear components necessary to perform the primary
function of a substation in a high voltage network. The use of
high-speed peer-to-peer communications using Generic Object
Oriented System Event (GOOSE) messages and sampled
values from Merging Units allows for the introduction of
distributed and wide area applications. In addition, the use of
optical local area networks leads in the direction of copper- Fig. 1: Conventional substation design
less substations.
The paper describes substation communications Depending on the size of the substation the location of the
architectures and different types of devices required for each switchgear components and the complexity of the protection
of the communication architectures are described. The paper and control system there very often are a huge number of
analyzes the main functional modules in a conventional cables with different lengths and sizes that need to be
microprocessor relay and compares it with the implementation designed, installed, commissioned, tested and maintained.
of protection functions in IEC 61850 based systems with A typical conventional substation has multiple instrument
distributed analog values. Merging units that provide the transformers and circuit breakers (Figure 1) associated with
interface between the current and voltage sensors and the the protection, control, monitoring and other devices being
intelligent electronic devices at the equipment, bay or connected from the switch yard to a control house or building
with the individual equipment panels.
Marco C. Janssen is with UTInnovation, Duiven, The Netherlands
(m.c.janssen@utinnovation.com),.
296
These cables are cut to a specific length and bundled,
which makes any required future modification very labor
intensive. This is especially true in the process of refurbishing
old substations where the cables insulation is starting to fail.
The large amount of copper cables and the distances that
they need to cover to provide the interface between the
different devices exposes them to the impact of
electromagnetic transients and the possibility for damages as a
result of equipment failure or other events.
The design of a conventional substation needs to take into
consideration the resistance of the cables in the process of
selecting instrument transformers and protection equipment, as
well as their connection to the instrument transformers and
between themselves. The issues of CT saturation are of special
importance to the operation of protection relays under
maximum fault conditions. Also Ferro resonance in voltage
transformers has to be considered with relation to the correct
operation of the protection and control systems.
Failures in the cables in the substation may lead to
misoperation of protection or other devices and can represent a
safety issue. In addition open CT circuits, especially when it
occurs while the primary winding is energized can cause
severe safety issues as the induced secondary e.m.f. can be
high enough to present a danger to people’s life and
equipment insulation.
The above is definitely not a complete list of all the issues
that need to be taken into consideration in the design of a
conventional substation. It provides some examples that will
help better understand the impact of IEC 61850 in the
substation.
In order to take full advantage of any new technology, it
necessary to understand what it provides. The next part of the
paper gives a short summary of some of the key concepts of
the standard that have the most significant impact on the
substation design.
III. IEC 61850 SUBSTATION HIERARCHY
The development of different solutions in the substation
protection and control system is possible only when there is
good understanding of both the problem domain and the IEC
61850 standard. IEC 61850 describes the functionality of
systems and applications in the substation using an object-
oriented modeling approach. This in combination with
standardized communication services and main stream
communication technologies defining how data is
communicated between functions in the substation.
The modeling approach of IEC 61850 supports different
solutions from centralized to distributed functions. The latter
is one of the key elements of the standard that allows for
utilities to rethink and optimize their substation designs.
A function in an IEC 61850 based integrated protection
and control system can be local to a specific primary device
(distribution feeder, transformer, etc.) or distributed and based
on communications between two or more IEDs over the
substation local area network.
IEC 61850 defines several ways for data exchange
between IEDs that can be used for different forms of
distributed applications. They introduce a new concept that
requires a different approach in order to define the individual
components of the systems in substations.
Considering the requirements for the reliability,
availability and maintainability of functions, it is clear that in
conventional systems numerous primary and backup devices
need to be installed and wired to the substation. The
equipment as well as the equipment that they interface with
must then be tested and maintained.
The interface requirements of many of these devices
differ. As a result specific multi core instrument transformers
were developed that allow for accurate metering of the energy
or other system parameters on the one hand and provide a high
dynamic range used by e.g. protection devices.
With the introduction of IEC 61850 different interfaces
have been defined that can be used bys substation applications
using dedicated or shared physical connections - the
communications links between the physical devices. The
allocation of functions between different physical devices
defines the requirements for the physical interfaces, and in
some cases may be implemented in more than one physical
Local Area Network (LAN) or by applying multiple virtual
network on a physical infrastructure.
The functions in the substation can be distributed between
Intelligent Electronic Devices (IEDs) on the same, or on
different levels of the substation functional hierarchy –
Station, Bay or Process as shown in Figure 2.
Fig. 2: Logical interfaces in IEC 61850
The logical interfaces of specific interest to distributed
applications are defined [1] as:
Interface IF4: CT and VT instantaneous data exchange
(especially samples) between process and bay level
Interface IF8: direct data exchange between the bays
especially for fast functions like interlocking
The first one is used typically for Process bus
applications, and the second for Station bus communications.
A significant improvement in functionality and reduction
of the cost of integrated substation protection and control
systems can be achieved based on the IEC 61850 based
communications as described below.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
IV. IEC 61850 PROCESS BUS
Non-conventional instrument transformers with digital
interface based on IEC 61850-9-2 or the implementation
guideline IEC 61850-9-2 LE [2] result in improvements and
can help eliminate some of the issues related to the conflicting
requirements of protection and metering IEDs as well as
alleviate some of the safety risks associated with current and
voltage transformers .
The interface of the instrument transformers (both
conventional and non-conventional) with different types of
substation protection, control, monitoring and recording
equipment as defined in IEC 61850 is through a device called
a Merging Unit. The definition of a merging unit in IEC 61850
is as follows:
“Merging unit: interface unit that accepts multiple
analogue CT/VT and binary inputs and produces multiple time
synchronized serial unidirectional multi-drop digital point to
point outputs to provide data communication via the logical
interfaces 4 and 5”.
Merging Units can have the following functionality:
• Signal processing of all sensors – conventional or
non-conventional
• Synchronization of all measurements – 3 currents and
3 voltages
• Analogue interface – high and low level signals
• Digital interface – IEC 60044-8 or IEC 61850-9-2
It is important to be able to interface with both
conventional and non-conventional sensors in order to allow
the implementation of the system in existing or new
substations.
The Merging unit has similar elements (as can be seen
from Figure 3) as a typical analog input module of a
conventional protection or multifunctional IED. The
difference is that in this case the substation LAN performs as
the digital data bus between the input module and the
protection or functions in the device. They are located in
different devices, just representing the typical IEC 61850
distributed functionality.
Fig. 3: Merging unit
Depending on the specific requirements of the substation,
different communications architectures can be chosen as
described in the next section of the paper.
297
V. IEC 61850 SUBSTATION ARCHITECTURES
IEC 61850 is being implemented gradually by starting
with adaptation of existing IEDs to support the new
communications standard over the station bus and at the same
time introducing some first process bus based solutions.
Station Bus Based Architecture
The functional hierarchy of a station bus based
architecture is shown in Figure 4. It represents a partial
implementation of IEC 61850 in combination with
conventional techniques and designs and brings some of the
benefits that the IEC 61850 standard offers.
Fig. 4: Station bus functional architecture
The current and voltage inputs of the IEDs (protection,
control, monitoring or recording) at the bottom of the
functional hierarchy are conventional and wired to the
secondary side of the substation instrument transformers using
copper cables.
The above architecture however does offer significant
advantages compared to conventional hardwired systems. It
allows for the design and implementation of different
protection schemes that in a conventional system require
significant number of cross-wired binary inputs and outputs.
This is especially important in large substations with multiple
distribution feeders connected to the same medium voltage
bus where the number of available relay inputs and outputs in
the protection IEDs might be the limiting factor in a protection
scheme application. Some examples of such schemes are a
distribution bus protection based on the overcurrent blocking
principle, breaker failure protection, trip acceleration schemes
or a sympathetic trip protection.
The sympathetic trip protection is of special importance in
case of a fault on a distribution feeder in a substation with
predominantly motor load. The voltage sag caused by the fault
and the following clearing of the fault will result in a load
inrush condition that may lead to the undesired tripping of
unfaulted feeders. In order to prevent this, the protection relay
that detects the fault and issues the trip signal to clear it will
have to apply a signal to the inputs of all relays on adjacent
feeders to block them for a certain time in a manner similar to
the Cold Load Pickup logic of distribution feeder relays.
The number of relay inputs, outputs and connections
between all feeder relays can be quite large. In contrast, such a
scheme is fairly straight forward to implement in a substation
using IEC 61850 communications using GOOSE messages
over the station bus as shown in Figure 5.
298

through a relay output wired into an input. The relay output

Protection Protection
IED IED
GOOSE GOOSE
Fig. 5: Sympathetic Trip protection with GOOSE
The relay that detects the feeder fault sends a GOOSE
message over the station bus to all other relays connected to
the distribution bus indicating that it has issued a trip signal to
clear the fault. This can be considered as a blocking signal for
all other relays on the bus. The only requirement for the
scheme implementation is that the relays connected to feeders
on the same distribution bus have to subscribe to receive the
GOOSE messages from all other IEDs connected to the same
distribution bus.
The reliability of GOOSE based schemes is achieved
through the repetition of the messages with increased time
intervals until a user defined time is reached. The latest state is
then repeated until a new change of state results in sending of
a new GOOSE message. This is shown in Figure 6.
typically has an operating time of 3-4 ms and it is not unusual
that the input may include some filtering in order to prevent an
undesired initiation of this critical function.
As a result in a conventional scheme the time over the
simple hard wired interface, being the transmission time
between the two functions, will be between 0.5 and 0.75
cycles – longer than the required 0.25 cycles defined for
critical protection applications in IEC 61850 based systems.
Protection Another significant advantage of the GOOSE based
IED
solutions is the improved flexibility of the protection and
control schemes. Making changes to conventional wiring is
very labor intensive and time consuming, while changes of the
“virtual wiring” provided by IEC 61850 peer-to-peer
communications requires only changes in the system
configuration using the Substation Configuration Language
(SCL) based engineering tools.
Station and Process Bus Architecture
Full advantage of all the features available in the new
communications standard can be taken if both the station and
process bus are used.
Figure 7 shows the functional hierarchy of such a system.

Fig. 7: Station and Process bus functional architecture

IEC 61850 communications based distributed applications

Fig. 6: GOOSE message repetition mechanism
The repetition mechanism does not only ensure that the
initial signal is not going to be missed by a subscribing relay.
It also provides means for the continuous monitoring of the
virtual wiring between the different relays participating in a
distributed protection application. Any problem in a device or
in the communications will immediately (within the limits of
the maximum repetition time interval) be detected and an
alarm will be generated and/or an action will be initiated to
resolve the problem. This is not possible in conventional hard
wired schemes where problems in the wiring or in relay inputs
and outputs can only be detected through scheduled
maintenance.
One of the key requirements for the application of
distributed functions using GOOSE messages is that the total
scheme operating time is similar to or better than the time of a
hard wired conventional scheme. If the different factors that
determine the operating time of a critical protection scheme
such as Breaker Failure Protection are analyzed, it is clear that
it requires a relay to initiate the breaker failure protection
involve several different devices connected to a substation
local area network as shown in the simplified block diagram in
Figure 8.
Merging Units (MU) will process the sensor inputs,
generate the sampled values for the 3 phase currents and
voltages, format a communications message and multicast it
on the substation LAN so that it can be received and used by
all the IEDs that need it to perform its function. This “one to
many” principle similar to that used to distribute the GOOSE
messages provides significant advantages as it not only
eliminates current and voltage transformer wiring it also the
supports the addition of new ideas and / or applications using
the sampled values in a later stage as these can simply
subscribe to receive the same sample stream.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Substation Substation
HMI Computer
Router WAN
Ethernet Switch Substation Bus
IED IED IED IED SCADA Master
Ethernet Switch Process Bus
MU IO U IO U MU
Fig. 8: Communications architecture for process and station bus
Another device, the IO Unit (IOU) will process the status
inputs, generate status data, format a communications message
and multicast it on the substation LAN using GOOSE
messages.
All multifunctional IEDs will receive the sampled values
messages as well as the binary status messages. The ones that
have subscribed to this data then process the data make a
decision and operate by sending another GOOSE message to
the to trip the breaker or perform any other required action.
Fig. 8 shows the simplified communications architecture
of the complete implementation of IEC 61850. The number of
switches for both the process and substation busses can be
more than one depending on the size of the substation and the
requirements for reliability, availability and maintainability.
Fig. 9 Substation design with process and station bus
Figure 9 is an illustration of how the substation design
changes when the full implementation of IEC 61850 takes
place. All copper cables used for analog and binary signals
exchange between devices are replaced by communication
messages over fiber. If the DC circuits between the substation
battery and the IEDs or breakers are put aside, the “copper-
less” substation is a fact.
The next possible step when using station and process bus
is the optimization of the switchgear. In order for the
protection, control and monitoring functions in a substation to
operate correctly several instrument transformers are placed
299
throughout the high voltage installation. However with the
capability to send voltage and current measurements as
sampled values over a local area network it is possible to
eliminate some of these instrument transformers. One example
is the voltage measurements needed by distance protections.
Traditionally voltage transformers are installed in each
outgoing feeder. However if voltage transformers are installed
on the busbar, the voltage measurements can be transmitted
over the local area network to each function requiring these
measurements. An example of such a solution is shown in
Figure 10. These concepts are not new and have already been
applied in conventional substations. In conventional
substations however it requires large amounts of (long) cables
and several auxiliary relays limiting or even eliminating the
benefit of having less voltage transformers.
I 21 I 21
IU IU U
Ethernet Switch
Fig. 10 Alternative substation design with process bus
Process bus based applications offer important advantages
over conventional hard wired analog circuits. The first very
important one is the significant reduction in the cost of the
system due to the fact that multiple copper cables are replaced
with a small number of fiber optic cables.
Using a process bus also results in the practical
elimination of CT saturation of conventional CTs because of
the elimination of the current leads resistance. As the
impedance of the merging unit current inputs is very small,
this results in the significant reduction in the possibility for CT
saturation and all associated with it protection issues. If non-
conventional instrument transformers can be used in
combination with the Merging Units and process bus the issue
of CT saturation will be eliminated completely as these non-
conventional CTs do not use inductive circuits to transduce the
current.
Process bus based solutions also improve the safety of the
substation by eliminating one of the main safety related
problems - an open current circuit condition. Since the only
current circuit is between the secondary of a current
transformer and the input of the merging unit is located right
next to it, the probability for an open current circuit condition
is very small. It becomes non-existent if optical current
sensors are used.
Last, but not least, the process bus improves the flexibility
of the protection, monitoring and control systems. Since
current circuits can not be easily switched due to open circuit
concerns, the application of bus differential protection, as well
as some backup protection schemes becomes more
complicated. The above is not an issue with process bus,
because any changes will only require modifications in the
300

subscription of the protection IEDs receiving the sampled

analog values over IEC 61850 9-2. VIII. BIOGRAPHIES
Marco C. J anssen received his BS degree in Electrical
VI. CONCLUSIONS Engineering from the Polytechnic in Arnhem, The
Netherlands. He has worked for over 21 years in the field
IEC 61850 is a communications standard that allows the of Smart Grids, Protection, Control, Monitoring, Power
development of new approaches for the design and Quality, Advanced Metering Infrastructures and
refurbishment of substations. A new range of protection and Substation Automation. From 1990-95 he was a
Technical Specialist in the Protection and Automation
control applications results in significant benefits compared to
group at NUON, The Netherlands. 1995-2001 he was a
conventional hard wired solutions. Senior Consultant at KEMA, From 2001 - 2005 he was a
It supports interoperability between devices from different Marketing Manager at Electron Automation, The
manufacturers in the substation which is required in order to Netherlands. Since 2005 he is the president of UTInnovation a Dutch
company providing consulting services for Smart Grids, Substation Automation,
improve the efficiency of microprocessor based relays
Protection, Communication, Power Quality and Advanced Metering
applications and implement new distributed functions. Infrastructures
High speed peer-to-peer communications between IEDs
He is member of IEC TC57 WG 10, 17, 18, 19, the IEEE PES Power System
connected to the substation LAN based on exchange of Relaying Committee and CIGR B5 and D2 working groups. He is editor of the
GOOSE messages can successfully be used to replace Quality Assurance Program for the Testing Subcommittee of the UCA
hardwiring for different protection and control applications. International Users Group, holds one patent is the author of the “I Think”
Sampled analog values communicated from Merging column in the PAC World Magazine and has authored and presented more
than 32 technical papers.
Units to different protection devices connected to the
communication network replace the copper wiring between
the instrument transformers in the substation yard and the
IEDs.
Such systems provide some significant advantages over
conventional protection and control systems used to perform
the same functions in the substations:
• Reduced wiring, installation, maintenance and
commissioning costs
• Optimization possibilities in the design of the high
voltage system in a substation
• Easy adaptation to changing configurations in the
substation
• Practical elimination of CT saturation and open
circuits
• Easier implementation of complex schemes and
solutions as well as easier integration of new applications and
IEDs by using GOOSE messages and sampled values that are
multicasted on the communications network and that the
applications and IEDs can simply subscribe to

VII. REFERENCES
[1] IEC 61850, Communication networks and systems for power utility
automation
[2] IEC 61850-9-2 LE: Implementation Guideline for Digital Interface to
Instrument Transformers Using IEC 61850-9-2, UCA International
Users Group
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Ethernet as a Channel for Protective
Relaying
John Benckenstein
“It was only as recently as 1960 when the
concept of using glass fibers for
communications was first proposed by
researchers at Standard Telecommunication
Laboratories.” In 1995 I wrote a paper called
“System Reliability Improvements through use
of Fiber Optics” and that was the first sentence.
In that paper I suggested that traditional relay
communications, like pilot wire, audio tone,
power-line carrier and direct fiber optics, would
all lose popularity to Time Division Multiplexed
(TDM) fiber optic channels. Subsequently, the
use of TDM has rapidly expanded within nearly
every Electric Utility in the world. Apparently I
was right.
I know there are those who prefer the single
fiber, single function approach for relaying. I
also believe this is the right choice for many
situations, especially when there are multiple
backups using different communication
mediums and/or routes. We were forced away
from dedicated fibers by either lack of enough
fiber, or the desire to get revenue for those
fibers. However, multiplexed channels have
worked out okay and are even an improvement
adding channel redundancy for most failure
modes.
Back in 1995, nobody I knew thought anything
would replace TDM; and especially not for
protective relaying. However, this new and
mysterious Ethernet is offering benefits worth
your consideration for mission critical
applications. Today I am telling you that
Ethernet is ready to take over as your new
favorite communication channel for protective
relaying.
Ethernet is largely reported to have been
invented in the mid 1970s by Robert Metcalfe,
who was a member of the Xerox research staff.
Since its beginnings as a way to connect
computers to copiers, Ethernet has continued to
evolve. By the end of this decade, most of your
substation communications will be done
301
exclusively over Ethernet. The speed at which
Ethernet takes over your relay channels will
depend on you, because the technology for this
is here today. Now let’s talk about the relay
engineer’s unique thought process.
Most of us are Engineers, and we are all
“techies” to some degree. We think with logic
and leave quick uninformed decisions are for
others with less important jobs. We have all
been called cheap or conservative more than
once in our lives and that does not bother us. We
are just being practical by retaining our assets.
We think it natural when our children point out
to us that in two separate vacation photos, taken
10 years apart, we are wearing the same shirt.
Why waste money when it still works?
We will actually read and understand the
specifications. We study numerous reports
before we buy a car, house, mountain bike,
home theater or anything expensive. That is our
cautious and analytical side that we carry
proudly into the office. It is why our companies
hired us in the first place. At the office, we
won’t buy into the latest technology until it is
proven; or until we study it long enough to be
100% sure buying it won’t make us look bad.
We stayed with electromechanical relays well
into the microprocessor age. Only when the cost
and proven benefits outweighed the fear of
change did we accept them. Multiplexing
channels over fiber is another example you did
not quickly accept and I am sure you can think
of others. Maybe relaying over Ethernet will be
next.
Contrary to this thinking, for the really cool and
high tech stuff outside of the office, we will buy
the first version out. The first PDA, smart phone,
IPOD, IPAD, X10–controlled Christmas lighting
or pocket calculators are good examples. We
may not buy the car with the most luxury, but
call them features and we are sold. We will keep
that car for ten years, because it still works. We
will buy the toaster; washer or dryer with the
most digital displays and features, even if it
costs a little more and even if our spouse does
not appreciate it. It has to be better because it
has more gadgets, right? And if it breaks
because of all those gadgets, we can probably fix
302
it from spare parts in our garage. If you are like
me you still have the power adapter from every
device you owned since college somewhere in
that garage.
Today’s Ethernet is similar to that. Not the
frequently breaking part, but the myriad of
gadgets and features to catch our interest part. It
follows then that this is what makes us interested
and wary of Ethernet at the same time. It is
undeniably higher tech than TDM. But this high
tech truly adds value and is mostly automatic; so
you do not need to fully understand it to use it. I
know this goes against our grain; but a lot of
very smart people have spent a lot of time and
money getting Ethernet right over the past 35
years. They have done this so you would feel
comfortable trusting it with your network assets.
It has a specification sheet with more acronyms
than any other industry, including system
protection.
RMON, SNMP, Telnet, VLANs (Port-
based and Tag-based), GVRP, Port-
mirroring, CLI, Port Security, SNTP,
BootP, DHCP, IGMP Snooping, 802.1p,
QoS, RSTP
I am asking you to ignore 80% of those
acronyms because they are not features you need
to understand to apply relaying over Ethernet.
However, some of these items are what will
ultimately provide you with the security to trust
this new technology. Remember RSTP, VLAN
and QoS. (Spanning Tree Protocol, Virtual
Local Area Network and Quality of Service)
These are all part of today’s Ethernet and all
play a role in making it suitable for relaying.
“This is not your Father’s Ethernet.” It is up to
those of us producing this technology to prove to
you that it is worthy of your trust.
The much talked about, desired, and dreaded
“smart grid” requires even more bandwidth from
our networks. Those responsible for these
networks are facing the decision of what will
that network look like. Will it be based on past
proven SONET technology? We feel safe with
SONET. It brings us predictable results and has
a track record that lets us sleep soundly at night.
As is often the case, there is something new,
right here, right now. It is something that better
interfaces with smart grid devices and makes
even better use of fiber than SONET. It is TDM
packets over Ethernet. This can provide you
with the same level of comfort, flexibility,
capacity and performance as SONET. The
question is; can you trust it?
Several utilities are building IEC61850-enabled
substations today. This involves using Ethernet
and fiber to replace copper wiring within the
substation. Clearly this already indicates some
level of acceptance for Ethernet carrying relay
data.
Utility IT departments replace their
infrastructure every 5 to 10 years. As those
systems are being replaced, they are finding less
SONET and more Ethernet systems to choose
from. T1 and SONET will be around for many
years, but clearly Ethernet is beginning to take
over.
Example: Most of you probably have a Voice
over IP phone network (VoIP) in place which
was purchased as a less-costly alternative to
traditional switched telephone services. VoIP
was a technology driven by the telecom
industry. Since utilities are a small segment of
the telecom industry, we are forced to follow
their technology lead or risk losing support
down the road. Do you buy and install new
relays every 5 to 10 years? My guess is the
relays you are replacing today were installed 20
to 25 years ago. But what are you going to do
with your channel equipment when your
communications group starts replacing your
backbone SONET with Ethernet? You will need
a way to interface those relays with Ethernet.
Fortunately you do have some choices today.
Several manufacturers now offer multiplexers
that interface standard communication protocols
like C37.94, RS232 and audio tones to TDM
packets over Ethernet. Even the
electromechanical pilot wire relays can function
properly through an Ethernet channel.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 303

All of these channels, plus telephone and video Routable protocols are more difficult to secure
as direct IP traffic, can be carried on special and must be interfaced to the outside world
broadband packet-based multiplexers designed through firewalls. Layer 2 packets contain
for use in power substations. Unlike a lot of the Media Access Control (MAC) addresses to
telecom-based systems, these multiplexers meet guide the packets from source to destination. A
the strict environmental and EMC standards unique MAC address is assigned by the
specified in IEEE1613. manufacturer and generally cannot be changed
by the user. Layer 2 packets are switched, but
I do not expect to make believers out of you
are not routable so they are inherently more
with just this one paper. I do hope that when this
secure and common for substation to substation
presentation is over you will better understand
communications. We will limit discussion to
what the potential is for today’s TDM-packet
layer 2 packets in this paper, but there is a place
over Ethernet solutions.
for both types within the Utility.
Basics of Ethernet and how it can carry TDM
channels: When any type of channel is
transferred over a digital media, the data must
first be converted into a series of bytes sent
every 125 uSec. Since there are 8 bits in every
byte, this is the equivalent of 64Kbps, which is
known as a DS0. It is the function of channel
modules to convert the actual data into this
series of bytes or samples. This is true for a
simple contact, voice, audio tone, RS232 stream,
or 64kb synchronous channels. TDM and Packet
technologies are the same up to this point. The
difference starts after the DS0 samples/bytes are Several companies make Ethernet switches
created. designed for the harsh substation environment
Similar to DS0 channels that carry data over that meet the requirements of IEEE1613.
TDM, packets carry data over Ethernet. The The transport or backbone
basic Layer 2 or 3 packets are similar. The connections are typically
difference is that layer 3 packets contain IP made through SFP
addresses and as such are said to be routable. transceivers, which stand for
304

Small Form-factor Pluggable. These are

available from a number of sources
with optical ranges from 500 meters to
over 100 km providing an economical
selection for the links between
switches. Bandwidth will range from
100MB up to 10 Gigabyte, but today
the most common bandwidth is 1.0 GB.
The switches include a number of
10/100MB copper (or fiber) ports to
connect local substation devices to each
other or to the backbone. Several
switches connected together with fiber
or other links in a ring or mesh
configuration is called a network.
Now that you have a general idea about
building the hardware side of the
network, let me explain something
about the data itself by comparing it to
source and destination address are embedded
TDM. For simplicity, we will group all TDM
inside the packet along with the data. The
systems together as it matters little for this
Ethernet channel is a sophisticated pipe used to
discussion whether you are using T1, E1,
intelligently transport the packet wherever it
SONET OC-12, SDH STM-3 bandwidth
needs to go. A single packet can also carry
systems. All of them have roughly the same
multiple DSOs (bytes) so it is not sent as often
channel latency, self-healing ring design, basic
as a DS0 sample in a TDM system.
circuitry, and network management philosophy.
The next question is; how does the network
With TDM the DS0 channel is mapped in
know where to send this packet? The first
advance of any data transfer by first selecting its
unique packet sent from source to destination
time slot and then routing that timeslot from end
will be received by each switch in route. Those
to end. This includes mapping any pass-through
switches will retransmit the packet out all ports
connections at in-between nodes. The bandwidth
because they do not yet know how to reach the
that carries this channel is “locked down”. It is
destination. Eventually, in a several dozen
there and available forever and the route from
microseconds, the packet will reach its
end to end is known. Even in redundant loop
destination. From that point on, all switches
systems, the redundant path is known. Channel
have “learned” the proper ports to route that
latency can be calculated with accurate results
packet. Future similar packets will be forwarded
making its performance deterministic, which of
out only the one correct port. This is the self-
course is highly desired for relaying. Not only is
learning bridge feature and another item you will
this DS0 locked down, but the entire bandwidth,
not need to fully understand if you can just
even when not allocated for channels is allocated
accept that it does work.
sending idle code across the fiber for any
unassigned channels. This is not bad, as it To review how to setup a packet connection, you
assures that any channel added will have the must know where the channel cards are installed
same reliable deterministic behavior. So why at both ends, period. The route including all in-
would anybody want to do anything different? between nodes is determined for you,
automatically.
A packet is used in Ethernet to carry the same
DS0, but there are no timeslots or pass-through
connections to set up in advance or ever. The
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
The TDM system uses framers, time slot
aligners, and digital cross-connect switches to
route the channel. You must know the physical
arrangement of the network, and the time slots at
all in-between nodes to configure a channel.
Settings need to be made at every node either
manually or with the help of a good network
manager. In Ethernet we use switches for this.
They have fewer parts, are less costly, and setup
the channel in one step without requiring
complete knowledge of the physical network.
I have explained how the packet is transported
using the switch part of the network. The
multiplexer can be thought of as a separate
functional block that works hand in hand with
the switch network. Whereas the switch network
is a standard product available from many good
vendors, the multiplexer suitable for protective
relaying is a more specialized device. It is the
multiplexer that contains the channel modules
305
that convert information into the
bytes. The multiplexer forms the
packets from these bytes by
adding MAC addresses and other
information to ensure the packet
is reliably received error-free, in
sequence and on time.
I indicated that a packet can carry
more than one DS0. I did not say
why this was important or explain
how one of these magical
multiplexers works. Each node
will have two ports that access
the network. They may ultimately
be combined into one physical
port via a switch, but each serves
a very different purpose.
The DATA port transmits and
receives data packets formed by
the packet processor, which were created from
samples taken by the channel modules.
Additional information is included from settings
made via the admin port that help direct the
packet through the network.
The ADMIN port has an IP address and, along
with your browser, is used to deliver web pages
to your PC for configuring the multiplexer.
Several password authentication types are
available.
• Administrator – Can create user accounts
and set their access levels
• Guest – Can view anything, but cannot
make any changes
• User – Can view and change anything
except authentication level of users
• Superuser – Can view or change anything
including user levels. They are both users
and administrators. (This is the level
most substation devices were set to prior
to the adoption of NERC/CIPS
regulations.)
Configuration of node names, IP addresses,
individual channel card and system clock
settings are performed through the admin port to
all nodes. The admin port is also used for
automatically populating each nodes database
with the contents of every other node on the
network. Let me repeat that part, seconds after
306
you add a node or channel module, every node is
updated with the type and location of every
channel module in every chassis. This is why
configuration of end to end channels is so easy.
Every node is completely aware of the entire
network and any changes as they happen,
automatically.
Settings for channel cards fall into two
categories, application and control.
1. Application settings: These vary with
channel type and are coordinated with the
relay or other end devices passing data
across the channel. They are applied to one
end at a time. These same setting options are
used when channels are part of a TDM-
based system. Examples are:
a. The settings for a four-wire voice card
would be transmit and receive gain and
whether or not signaling is enabled.
b. A transfer trip relay card would have
settings for contact bounce, security, trip
hold delay, addressing and guard before
trip logic.
2. Control Settings: These control how that
data is passed through the network and are
applied to both ends simultaneously. They
are unique to the packet network and a big
reason why Ethernet is now considered
suitable for protective relaying.
a. VLAN: Virtual Local Area Network:
This setting is optional and is used in
conjunction with your switch network.
VLAN tags allow a group of packets to
reside on a common switch network
alongside other packets with different
VLAN tags. The packets appear as if
isolated on their own private network.
Proper use of VLAN tags provides
better control over the route packets take
through the network. They are
mandatory for providing what is
advertized as bumpless or hitless
switching on ring networks and for
when using packet-based multiplexers
as inputs to MPLS networks. MPLS
stands for multi-protocol label switching
and is useful for providing deterministic
and known primary and backup routes
through complex Ethernet networks.
b. QoS: Quality of Service: This is
optional, but very important to ensuring
deterministic delays/latency across a
network for critical channels. Eight QoS
levels are available with the highest
level 7 generally reserved for switch to
switch management messaging. QoS
becomes a factor in latency when a
network is over 65% loaded. Above this
it is likely that several packets will
occasionally want to be multiplexed
onto the transport fibers at nearly the
same time. Quite simply, those packets
with the higher QoS settings are
permitted to pass first, thereby assuring
your relay channels will have the
latency you expect. The preference
given to higher QoS channels is
weighted so even with very high traffic
present, lower priority packets will
eventually pass. If you know your
network loading will always be under
50%, then it is less important to apply
QoS as almost all packets will be passed
without delay.
c. Bandwidth E conomy: This allows the
user to choose between having a lower
latency channel using relatively high
bandwidth, or a longer latency channel
using less bandwidth.
If a single DS0 byte was included in
each packet, the effective bandwidth for
that channel would be about 5.5 Mbps
with a latency of under 1 mS. This is
because you must send each packet
every 125 uSec and each packet
contains about 87 byes. 87 bytes x 8
bits/byte = 696 bits. 696 bits / 125 uSec
= 5.5 Mbps. 87 bytes is the minimum
sized packet that can be considered
standard. By including more bytes in
each packet you can reduce the
frequency that packet is sent thereby
spreading the effective overhead across
more data bytes. When a more typical
setting of 16 bytes (TDM frames) per
packet is selected, the bandwidth is
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 307

reduced by a factor of almost 16

to 392 Kbps. The channel latency
at this setting is about 6.2 mS.
For reference; VoIP systems
generally use a setting of 64
frames per packet with an
inherent delay of 24 mS.
d. Jitter Buffer: Jitter is defined as
the variable delay experienced by
channels traversing a network.
Without buffering, the existence
of jitter means lower quality data.
The jitter buffer allows you to
smooth out these delays by buffering up
packets at the receive side. Think of a
jitter buffer being a bucket of water with
a small hole in the bottom that precisely
regulates water (packets) exiting. A DS0
from within the packet (water) must exit
exactly once every 125 uSec or the
quality of the data is compromised.
All four parameters (VLAN, QoS, Bandwidth
Economy, and Jitter) are automatically copied to
both ends whenever you make a change to either
end via the configuration software. Whenever
this happens, a new connection is established
and labeled with a unique “context” number.
The connection includes information from all
four parameters, plus the channel slot position.
You could say this process sets up the system
from end to end similar to the time slot maps
used with TDM, but without all the steps. The
context number is embedded into each packet so
the destination immediately knows how to
decode the packet by directing the data to the
right jitter buffer which is already tied to the
proper channel module. VLAN and QoS tags
become part of the actual packet as these are
used to pass the packet through the network.

Effect of Jitter on Data

The water level in the bucket is initially
set to be half full. In this way if the
water flow slows down for a while, the
water level in the bucket is lowered, but
it does not affect the output unless it
runs dry. Similarly if packets start
arriving too soon, the water level rises
so no data is lost, unless it overflows the
sides. Buffers add delay to the data, so
they should be sized as small as
practical without risk of over or under Example of Data with No Jitter and Good Clock
flow. A setting of four packets (two over
and two under) is normal for fiber-based The very nature of Ethernet is asynchronous as
systems. Higher settings may be packets are sent “as soon as possible”. However,
required if a non-fiber backbone is to make it suitable for relaying you must make
involved that introduces variable delays. the asynchronous Ethernet become
synchronous. TDM is a synchronous network
308

because all of the framing is set to repeat on 125 etc. This is all fairly well known and was the
uSec boundaries. The timing for this primary subject of my last paper.
synchronization is embedded in the overhead of
It is accepted that TDM systems configured in
the TDM frame sequence. Packet networks do
rings (loops) will effectively reroute channels
not have a common overhead frame to pass
broken by a failed fiber. They do this by
clock between nodes. What they do have is a
remapping the affected channels in the opposite
highly accurate internal oscillator to establish
direction around the ring, thereby avoiding the
clock packets that are multi-cast to all remaining
broken fiber. The engineer will always know
nodes at precise intervals. All “remote” nodes
exactly what path their channels are taking
will use the intervals between clock packets to
before and after the break. In some systems,
calibrate their own internal oscillators to match
manual setup of alternate paths may be required
the master oscillator’s frequency. It is this
for optimum switching performance. Switching
recovered clock frequency that each node uses to
times for TDM systems will range from 5 mS to
take packets out of the bottom of the jitter
250 mS with 50 mS being most common. TDM
buffers. This allows every node to look at the
systems designed with relaying in mind will
same relative data at exactly the same time,
hedge towards the lower settings.
which by definition makes it synchronous.
Ethernet handles redundant path switching very
Even if the clock signal is momentarily lost, the
different from TDM. To understand this, you
local oscillator, which you remember has
need to know Rapid Spanning Tree Protocol
already been offset to match the master
(RSTP). This was one of the three important
frequency, will step in to ensure this timing stays
switch specifications mentioned earlier. Let’s
100% solid. For private networks over fiber, this
consider an Ethernet network configured in a
has been proven to be equal to or better than
ring. In the telecom world, costs are assigned to
TDM timing performance.
each link between switches, which are used to
There were three main obstacles to overcome find the least cost route for any connections. For
before Ethernet could be considered acceptable utility private network applications this cost
for carrying your critical relay channels. factor is not normally important, but the ability
to select a root node is still important.
Ethernet Acceptance Hurdle Number One
Is there a redundant path option when my most When we talk about ring or loop architecture we
direct channel route is broken? are referring to the physical structure and not the
When a main fiber, patch cord or transceiver path data will take within this structure. It is a
requirement of Ethernet that no packet is
fails in a single fiber pair solution, you lose that
allowed to circle endlessly around a physical
relay channel. Of course you will
also have backups so this is not
completely unacceptable. Your
planned redundancy is without
question compromised until
repairs are made. Designing with
alternate routes and self-healing
rings can add redundancy with
minimal effort or cost making it
easier for the engineer to accept
using a single fiber pair for
multiple functions.
TDM systems implement self-
healing rings with labels and
acronyms like, APM, ILS, Mini-
DACS, FFHS, UPSR, BLSR,
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
ring. There must be only one unique route
between any two nodes. This would cause what
is known as a firestorm and will bring the
network to its knees in milliseconds by blocking
all traffic. RSTP is a standard feature of
managed switches and simply enabling it
effectively prevents firestorms even when the
physical configuration contains rings. RSTP
settings allow you to select the one port on the
ring network that will be blocked for all data.
(Multiple ports will be blocked for mesh
networks.) So the “allowed” data path on a ring
looks more like a “C”. Since no data can pass
through the blocked port, some traffic will
necessarily be directed the long way around the
ring to reach its destination.
When a path break is detected by one of the non-
blocked ports, the system will automatically
enable the previously blocked port. This causes
any packet stream previously using the newly
blocked port to find a new route. This is
automatic and requires no preconfiguration or
interaction by the user.
So how long does it take to reroute your critical
circuits? When STP was first introduced, you
could expect switching delays of 20 to 30
seconds. Rapid Spanning Tree Protocol (RSTP)
was adopted a few years later, reducing the
switching time to 1 to 2 seconds. Some of the
switch manufacturers who specialized in Utility
applications developed their own proprietary
versions of RSTP called S-Ring, ERSTP, etc.
These implementations lowered switching times
for rings to less than 5 mS per node. MESH
309
networks take a little longer. Then in 2004, a
new version of RSTP was released that could
meet similar switching times and was an open
standard anyone could adopt. Subsequently, the
use of proprietary protocols diminished allowing
some interoperability between switch vendors.
Now you know that switching times for rings are
comparable between TDM and Packet networks
configured in rings.
• 15 to 50 mS for packet networks of 3 to 10
nodes
• 5 to 150 mS for TDM networks
Either is generally considered acceptable for
most relay applications.
Multiple Spanning Tree Protocol (MSTP) is a
relatively new variation of RSTP. MSTP-
enabled switches have the ability to set different
ports as blocking for different VLANs. By
prudent selection of VLANs for each channel,
you can ensure the primary path will be the
shortest route. If you choose to route primary
and secondary channels opposite directions
around the same ring, then you never have to
wait for rerouting delays on both channels for a
fiber break.
Some vendors are taking advantage of an
additional possibility with MSTP that allows for
near zero switching delay for breaks. The
principle involves sending the same channel
information both directions around a ring in two
nearly identical packet streams. Logic at the
receiving end listens to both directions, but only
accepts data from one of the packet streams.
With this system, it is likely that no data
will be lost and there will be essentially
zero delay to “reroute” to the alternate
path. This performance has been called
bumpless, hitless or zero-delay switching.
The only drawback to this feature is that it
doubles the bandwidth used within the
network.
This is one benefit that TDM simply
cannot match. Packet over Ethernet
systems can give you better channel
availability than TDM right when you need
it; when your fiber breaks concurrent with
a line fault.
310
Ethernet Acceptance Hurdle Number Two
Is channel latency consistent and low enough for
relaying? Is it deterministic?
Latency is defined as the time it takes a channel
input at the originating node to reach the output
at the destination node. This could be a transfer
trip contact, an RS232 data byte, or any piece of
data. Latency in a TDM system is always the
same low value. This can be calculated by
knowing: the fiber distance the data must travel,
the pass-through delay at in-between nodes, and
the internal processing delay of the channel
cards themselves.
Due to the need for the jitter buffers, channels
sent over Ethernet are not as fast as channels
over TDM, but are generally considered
acceptable. The tables here show typical channel
latency for RS232 and PRS (transfer trip)
modules exclusive of any relay processing
delays. Note that because you have control over
how many DS0 samples you put into each
packet (shown in Bandwidth Economy); you
have full control over latency and bandwidth
used for each individual channel. The faster
channel settings will use more bandwidth, while
the slower settings will use less bandwidth.
For medium to large systems, with a lot of nodes
between the ends, Ethernet can actually provide
a faster channel due to the way data is passed-
through at in-between nodes. In TDM, you can
have 125 to 250 uSec of delay to receive,
reframe and transmit the DS0 channel at every
pass-through node. In Ethernet, this function is
handled completely by the switch with pass-
through delays of under 10 uSec. Remember that
in most cases, this delay is tolerable anyway so
either system will still satisfy the time
requirements.
Ethernet Acceptance Hurdle Number Three
What about Bandwidth?
I can understand how you could be somewhat
surprised by the amount of bandwidth used to
send a single DS0 over Ethernet compared to
TDM. But what is the right way to compare
bandwidth?
Only a few relays require latency of less than 1
millisecond and those will use about 5.5 Mbps
of bandwidth, or 0.55% of the GigE channel.
Some older pilot wire or less sophisticated
current differential relays fall into this category
and must be set for 1 frame per packet to
function properly. These relays cannot
compensate for channel latency like modern
current differential relays.
If this same channel were running over an OC-
12 SONET network that DSO will use only 64
Kbps, or 0.01% of the total bandwidth
(622Mbps). On the surface this looks like
Ethernet uses 50 times more bandwidth than
TDM.
But that is not 100% correct.
Rarely are channels required to be under 1 mS in
latency. More than one relay engineer has told
me back to back channel latency of 6 to 10mS,
was completely acceptable for Blocking, DTT or
POTT schemes. From the chart, a latency of 6
mS will require about 392 Kbps of bandwidth
for the RS232 channel, which equates to 0.04%
of the 1.0 Gbps channel.
This is only six times greater bandwidth than for
normal TDM channels. However, most SONET
systems cannot protect individual DS0 channels
for redundant loop switching. Instead they must
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
switch an entire T1, which is 1.544 mbps of
bandwidth, or four times more bandwidth than
Ethernet. Now it is true any additional DS0 used
in that particular T1 you get for free without any
additional bandwidth. If four DS0s are used per
T1 then the bandwidth used for either TDM or
packets is nearly identical.
Even that is not a completely correct.
The subtle point here is that no matter how many
channels you set up over any SONET OC-12
system, the actual bandwidth used on the fiber
stays constant at 622 Mbps, or 100% of the
available bandwidth. With Ethernet, the actual
bandwidth used on the fiber varies with the
number of channels and the number of frames
per packet in each channel. You have complete
control over this.
What am I inferring here besides reciting a lot of
times and percentages? The bottom line is that a
packet multiplexer with a gigabit switch can
have roughly the same effective bandwidth as an
OC-12 SONET multiplexer. The big difference
is that the packet solution will offer a
considerable savings in initial dollars spent,
configuration time, and ease of operation.
311
What about Scalability?
If you had one of these Broadband Multiplexers
today, and found you needed more bandwidth,
you would simply upgrade your switches to 10
GigE or in a few more years to 100 GigE. There
would be no requirement to change the
multiplexer.
If you had an OC-12 system today and found
you needed more bandwidth, you may be able to
upgrade the optical drivers to OC-48 or OC-192;
but you could also need to replace the DACS
and T1 mappers. I expect that the cost to do this
would be relatively high and take longer to
complete than simply replacing the switches in a
packet network. With manufacturers shifting
development and production more towards
packet Ethernet networks, there is no guarantee
an upgrade will be easy to accomplish without a
wholesale replacement of the system.
Packet over Ethernet technology is more
scalable and future–proof than any past or
current system, including all levels of TDM. The
best parts are that this high technology is
automatic, easy to use, reliable, and produces
deterministic results.
312
Ocala Utility Services - Installation and History
In April of 1995, Ocala Utility Services (OUS)
in Ocala, FL installed a FOCUS FFHS (Four
Fiber Hot Standby) scheme between 20 of their
substations. They wanted to improve the
reliability of the 16 HCB pilot wire relay
systems in use on their network. Prior to 1995,
the HCB/PM relay schemes averaged three
correct and 5 incorrect operations each year due
to failing or marginal pilot wires. SCADA
information was carried over a combination of
leased and owned wire circuits. These were not
only costing money to lease, but frequently
received errors requiring retransmission of RTU
data which slowed down the entire SCADA
system. Telephone circuits to the substations
were costing an additional $5,090.00 annually.
The then new fiber multiplexer system
successfully improved relay reliability and
lowered their operational costs by eliminating
leased line charges and reducing maintenance
hours.
Pete Baker is the Relay Supervisor at Ocala
Utility Services in charge of the project. Now 16
years after the initial installation, OUS has
replaced that system with a new packet-based
multiplexer called BB FOCUS. Pete stated that
one of the reasons for Ocala selecting the BB
FOCUS was its ability to reuse all of the channel
cards still in service from the original T1 system.
This new system has been at least partially
operational since June of 2009 with SCADA
being the first to be moved onto the new
platform. All RTUs had previously been
converted from RS232 to Ethernet, so it was a
straightforward process to move them from the
temporary switch network onto the BB
FOCUS GigE switches. The existing
telephone channel modules were
moved into the BB FOCUS next. It
operates a little like a VoIP system,
but only one call is permitted at a time;
and there is just one trunk line to the
external world.
The relays were the last channels to be
transitioned to the BB FOCUS. Six of
the original 16 HCB systems were
converted to a combination of SEL
387 and SEL 311L relays during some
line reconfiguration work. The
channels for these new relays required
6NF optical interface modules meeting
the ANSI C37.94 standard. These
were the first relays placed into service
over BB FOCUS anywhere in the world. When
the HCB channel interface modules were moved
into BB FOCUS we believe it marked the first
time in history that this 60-year old relay design
was operating over a modern packet over
Ethernet multiplexer. If you are familiar with the
HCB relay, then you know this is a 100%
electromechanical design that cannot tolerate
channel latency in excess of about 1.5
milliseconds. The fact that this strict
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 313

specification could be met for a 20-node required working through some challenges
network says a lot about the capabilities of during the installation; but it was all worth it
packet over Ethernet technology for low latency according to Pete. The system currently is
and consistency. providing reliable channels for several HCB,
311L and 387L relays. More channels will be
Pete Baker became skilled at maintaining the
added in the coming months.
previous T1 FOCUS system over the years
including making adjustments to time slot maps.
He will tell you now that the mapping
configuration process for the BB FOCUS is
more intuitive and faster than the previous
system. Being a beta site for this new technology
314
Broadband Power Line Carrier at American Electric Power
Another example of using packets to carry relay
channels comes from a completely different
direction. American Electric Power (AEP) in
Gahanna, OH is upgrading the existing pilot
wire relay protection on some of their 69 kV
lines. Senior Engineer Mark Majka says AEP
prefers to use fiber and current differential
relaying. The problem was finding a channel for
the current differential relays when fiber was not
available. Mark was in charge of the project
during the lab and field trials.
They settled on a system from Amperion Inc.,
based in Lawrence MA that used a highly
specialized broadband power-line carrier
(BPLC) technology to carry Ethernet across a
power line. The connection to the current
differential relays, which require a 64 Kbps
TDM link, is accomplished using TDM to
Ethernet converters. Successful field trials were
completed in 2010 on a 5 mile, 69 kV line in
Newark, Ohio. Five miles point to point is close
to the maximum allowed distance without
exceeding FCC regulations for radiated power.
The BPLC system supports longer distances
using regenerators every four to five miles.
Referring to the block diagram of the system
you will notice four basic parts.
The relays chosen for this trial included GE L90,
SEL 421 and SEL 311L. One TDM to Ethernet
system evaluated used the AMETEK BB
FOCUS multiplexer to convert the 64kb and
RS232 relay channels to Ethernet for the
modem. In the modem a unique modulation
scheme produces 1536 individual carrier
frequencies in the 2 to 34 MHz range. An active
filter module isolates individual frequencies
while providing precise gain control for each.
Finally, the BPLC from the filter is injected onto
the power line through a modified lightning
arrestor developed by Amperion. Line to line
differential coupling was chosen to give the best
bandwidth, lower emissions and greater noise
immunity. This also helps the system ride
through single phase to ground faults.
If some of the frequencies are partially or totally
blocked by line characteristics, the rest of them
are automatically adjusted to compensate. The
immunity to noise was tested with external line
arcing switch operations and cattle prods arcing
to pole grounds with no measureable
degradation of performance. A consistent 15
Mbps of bandwidth was available providing
more than enough for the required 64kb and
RS232 channels. Modifications to the jitter
buffer designs were made during the course of
the project to account for the larger and more
variable latency of packets running over the
BPLC system. This was primarily due to the
half-duplex nature of the modems which caused
data to arrive in spurts. Jitter was further
increased by the higher priority messaging sent
between modems, which are required to keep
throughput at optimum levels.
The relays in the field were subjected to a
comprehensive series of simulated internal and
external faults with the aid of Doble’s satellite
synchronized test sets.
• Relay trip times for faults at 50% of line
were 38 to 42 milliseconds.
• Similar results were achieved for faults at
5% and 95 % of the line.
• No trips for external faults.
• Results confirmed similar tests made in the
laboratory prior to field installation.
• Overall trip times averaged about 20
milliseconds slower than a point to point
fiber channel.
• Reductions in overall trip times are expected
after additional development.
Although trip times were longer than ideal,
analysis showed that they were within safety
margins and acceptable to AEP for 69kV
applications.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 315

References:
1. System Reliability Improvements through use of Fiber optics, Benckenstein, John R., Pulsar
Technical Publication 96-1, Coral Springs, FL, 1996.
2. IEEE Std. 802.1Q-2005, Virtual Bridged Local Area Networks.
3. Pete Baker, Relay Supervisor, Ocala Utility Services, Substation Group
4. Mark Majka, Senior Engineer, Protection and Control, AEP
5. HVBPL (BPLC) Communications over Transmission Lines, EEI TD&M Conf., 10-5-2010
Bruce Renz, President, Renz Consulting LLC and Nachum Sadan, CEO , Amperion, Inc

Biography for John Benckenstein:

John received his BSEE from Purdue University in 1980. Since then he has worked in the
Electric Utility industry with positions in sales, applications, product management and
business development. He has survived three acquisitions, allowing him to maintain
basically the same job with four companies including Westinghouse Electric, ABB Power
T & D, Pulsar Technologies, and AMETEK Power Instruments since 2006. John has
contributed to several IEEE PSCC working groups and is a past author and panel member
for substation automation and multiplexed communications discussions. John is a
member of IEEE and CIGRE professional associations.

John Benckenstein
AMETEK Power Instruments
Pulsar Division
4050 NW 121st Ave.
Coral Springs, FL 33065

Email: john.benckenstein@ametek.com
Telephone: +1.954.344.9822, ext 204
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 317

TESTING ETHERNET NETWORK FOR GOOSE

TRIPPING
Daniel Espinosa- CFE, Rene Aguilar, James Ariza- Megger, Jorge Mendoza- Ruggedcom

Abstract--Comision F ederal de E lectricidad ( CFE) i s t he federal

utility company in Mexico. CFE has started t o w ork on Generic
Oriented O bject S ubstation E vent (GOOSE) messages not o nly
for interlocking functions but for circuit breaker tripping as well.
In o rder t o u se G OOSE m essages f or trip, CFE is using
Input/Output M odules ( MES i s t he S panish a cronym) installed
near t he s witchyard equipment. MES s ubscribe t o t he G OOSE
messages published by the protection devices through an Ethernet
LAN v ia f iber o ptic ca ble. As G OOSE m essages a re t aking
more critical roles in the protection and control task of the
substation, i t i s i mportant fo r the protection en gineer t o
develop new test methods to evaluate the performance of
the Ethernet switches and IT network topologies.

This paper describes a method for testing the performance and

reliability of Ethernet managed switches when GOOSE messages
are u sed t o t rip ci rcuit b reakers. Rapid S panning T ree P rotocol
(RSTP) is tested to verify the communication recovery time on a
switch failure event.

Index Terms—GOOSE Tripping, IEC61850, RSTP,

Managed Switch.

I. INTRODUCTION

A s communications in the electrical substation take on

more critical roles in the protection and control tasks of
the utility, it is important for the protection engineer to
understand the basics of the IT network. Many experienced
protection engineers find discussions of IT network issues to
be dense and perhaps intimidating. They have not faced the
need to understand the behavior and performance
characteristics of IT Networks in the substation. Ethernet
switches and IT network topologies are as important to
understand as protective relays in order to achieve substation
goals related to the availability, dependability, security and
maintainability [1].
In order to achieve redundancy for SCADA and GOOSE
message transmission, CFE has chosen to interconnect LAN
switches in a simple ring configuration. Each protection and
control cabinet has one LAN switch to connect the protection
and control devices, which includes the bay’s associated
MES. To achieve redundancy, these devices are also
connected by using the standby Ethernet port to the LAN
switch located to the nearest cabinet. The network
architecture is shown in Figure 1[2].
Figure 1 Ethernet network architecture for protection trips
using GOOSE messages
This network design reduces the number of LAN switches
required to send a GOOSE message from a protection relay to
a MES in the same bay while at the same time adding
redundancy; however, currently there are neither references
nor procedures to test the performance and reliability of IT
network topology when it is used for GOOSE tripping. Tests
described in this paper have not been performed by CFE in the
past.
II. ETHERNET NETWORK DESIGN CRITERIA
In order to avoid manufacturer dependency, LAN switches
must use RSTP defined by IEEE 802.1D-2004, to allow
communication recovery on network events such as LAN
switch or link failure .[3]
The network configuration must be as simple as possible. The
more switches that are interconnected, the more difficult the
network analysis, due to a fail, will be. In most cases, a simple
network ring is recommended.
318
In practice the maximum number of switches in an optimized
ring occurs when the number of priority bridge levels has been
exhausted. This limits the size of the ring to 31switches. Rings
of more than 31 switches are still possible but will failover and
recover in a slower fashion [4].
There are two types of failures on this ring configuration that
can produce time delays on GOOSE message transmission:
Case 1. Cabinet’s LAN switch fails.
Protection relays are unable to send GOOSE messages through
the LAN switch where they are connected. The communication
is recovered when both relay and MES has used its standby
Ethernet port on the backup LAN switch. The switching time
between hot and standby port was not measured in this test set
up. By specification, the backup switch must start to transmit
as soon as the first packet is sent from the devices; therefore,
the Ethernet link between the switch and device must be
established.
Case 2. Any other LAN switch in the ring fails.
Due to RSTP and switch capabilities, the communication can
be affected in two ways:
a) Communication between switches is truncated, but
communication between local ports continues.
b) Communication between switches and local ports is
truncated.
Communication between local ports is the traffic between
devices connected to the same switch, and for trip purposes is
the most critical one on the network topology shown in Fig 1.
A. LAN Switch Considerations
The following considerations should be taken into account
when designing LANs using GOOSE messages for trips.
a) Local traffic in the LAN switch must continue and no
time delay must be admitted when a RSTP
reconfiguration is in progress. That means, time
delivery must be consistent and just depend on relay
transmission time processing.
b) RSTP reconfiguration due to a failure on one ring’s
switch will affect traffic from one device to others
located in a different switch on the network. Breaker
failure is one example of protection scheme that
requires the transmission of GOOSE messages from
one cabinet to another (inter bay trips). Time delay is
accepted if reconfiguration is in progress and no
transmission is possible.
B. Network Latency
Latency in a communications network is defined as the time it
takes for a message to traverse the network from the
transmitter to the receiver. In Ethernet networks, latency may
come from the following sources:
- Store and Forward Latency, Switch (Lsf)
- Switch Fabric Latency (Lsw)
- Wireline Latency (Lwl)
- Queuing Latency (Lq)
With a 100MB Ethernet connection from the devices to the
switch and a frame length of 137 bytes, the total ideal delay is
approximately 18.96 microseconds (LSF=10.96µs, Lsw=8µs,
Lwl=0 and Lq=0) In other cases, with up to 24 switches, the
total delay could be around 500 microseconds depending on
network architecture and background traffic. During the
following tests background traffic was added to the network
equivalent to Lq = 2 µs, so all calculated results could have an
error of +/- 2 µs.[5]
III. TESTING
A. Measuring the Operation Time of the Protection Relay
In any electrical power system, the less time required to clear a
fault, the less damage will be received by the switchgear. Due
to this, any protection scheme must take into account the
operation time required to energize the circuit breaker’s trip
coil. In some schemes, the protection relays close a built-in
contact to energize the trip coil. In other cases, auxiliary relays
are used for this purpose, which increases the operation time
of the scheme up to 10 ms.
Based on CFEs previous project experiences, tripping by using
GOOSE messages through MES are equivalent to schemes
using auxiliary relays.
To determine the operation time of the protective relay using
GOOSE message, the time is measured from the inception of
the fault up to the moment the test set detects the GOOSE
message.
In order to analyze the network performance, it is necessary to
measure the operation time of the scheme in the best
conditions. To achieve better time delay, due to network
latency, a scheme where the relay and test set are connected to
the same LAN Switch will be used.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia 319

configured with the edge port set to FALSE. The ports used to
connect the IED and test set were set to TRUE. Setting up the
ports with this configuration will avoid the RSTP
reconfiguration process to consider the Edge ports as a
possible path to the network, speeding up the network healing
process upon a link or switch failure. The RSTP Bridge
Priorities of the switches were setup to ensure SW1 to be the
RSTP root, and SW2 to be the Root Bridge backup. Using this
topology the ring was logically opened between SW5 and
SW6.

Figure 2 Test Setup for measuring the operation time of the
protection relay
In this scenario, the test set will measure the time taken from
the inception of the fault to the instant the relay publishes the
trip GOOSE message. This will require that the test set be able
to subscribe to the trip GOOSE message.
To avoid proprietary protocol dependency, only standard
RSTP protocol was considered during this test. RSTP healing
process will get longer delay after the Root switch fails and
before the close of the link between SW5 and SW6 is decided.
In fact, a power failure of the RSTP root switch leads to the
worst case scenario for recovery time and this is what the test
was trying to accomplish.

The protection relay is configured as a impedance element to Due to the test conditions, SW1 starts the test with the power
publish a GOOSE message when a fault is detected on zone 1. source disconnected. Then, in order to get stable conditions on
The relay test set is configured to simulate a fault at 50% of the network, SW1 is energized for one minute before a source
the line length of the distance protection. failure is simulated on the root switch and a zone 1 fault is
applied to the relay. The power source SW1 failure initiates
The operation time of the relay was measured between 13.19 the reconfiguration process while at the same time a trip
and 14.39 ms. Twenty tests were performed which gave an GOOSE message is being published by the protection relay.
average of 13.63 ms, with no background traffic.
If the RSTP reconfiguration process affects any local or
As can be seen above, the protection relay tripped in less than switches intercommunication, a time delay on trip GOOSE
a cycle. This was caused by the sub cycle tripping algorithm in transmission is expected.
the relay.
B. Network Failure Test Setup
In order to verify the performance of the network, nine (9)
LAN switches, one protection relay and one relay test set with
GOOSE support, were used. All Ethernet ports used UTP-5
rated cables at 100 MB/s communication speed.

The protection relay and the relay test set are configured as
described in section A. In the test method described below a
power source failure on the LAN switch was simulated. No
other network failure events were simulated such as link
failure

The power source of the RSTP root switch is connected

through a fast speed digital output (1 ms) of the test set. In this
manner the test set can connect and disconnect the power
source to the switch and simulate a ring fault.

Root Switch
OUT1 Relay Test Set
LAN Switch
Figure 4 Test's network topology and configuration

115 V

Figure 3 Test connection to Switch 1 Power Supply

The switches are connected in a ring configuration as shown in
Figure 4. The ring ports used to interconnect switches were
320
C. Determining Root Switch Fault Instance
One of the biggest challenges before the test was executed;
was to determine the right instance to disconnect the root
switch. The first tests performed did not give accurate results
because the ring did not add any time delay. This was due to
not knowing the correct instance for simulating the root switch
failure.
In order to determine the exact instance when SW1 does not
transmit any package, the link between SW5 and SW6 was
physically disconnected. The protection relay is connected to
SW5 and the test set to SW6, as shown in Figure 5. In this test,
when SW1 fails, there is no communication path between the
relay and the test set. Therefore, no GOOSE message will be
detected by the test set.
Figure 5 Test connections used to determine SW1 fault
instance
The test sequence is as follows:
SW1 is energized by closing output 1 on the test set for
approximately one minute. This allows the network to
stabilize. After one minute, a power source failure is applied at
an instance (t) before a zone 1 fault is applied to the protection
relay which publishes a trip GOOSE message. The test set will
stop the test when the GOOSE message is received. If no
message is detected then the test was stopped by timeout.
It’s necessary to consider that the GOOSE message had a
BOOLEAN and a BITSTRING in the DataSet with an
Ethernet frame length of 137 bytes. The total network latency
is approximately 170.64 microseconds for the nine switches.
Instant (t) was started at 500 milliseconds before fault
inception and was reduced gradually up to a range between
193 and 196 ms. 196 ms before a GOOSE message is
published, the message has a Probability of 1 to be
transmitted. SW1 is able to perform its task even when it is
physically disconnected due to the time required to de-
energize its integrated circuits. After 193 ms, the message has
a probability of 1 to NOT TRANSMIT, because SW1 hasn’t
any power to perform its task.
With this latency and no deterministic processing times on
relay and switches, a ±3 ms window is a huge one because
SW1 can transmit the test message in just 94.8 µs, that is 31
times smaller than the window the fail can happen, if the nine
switches’ latency is taken in account that represents 17 times.
If we consider the probability of an Electrical Power System
and this network topology fault coincidence, including the
switch’s capability to continue working for a long time on
power fail, we will find a very low value. Even that, taken into
account statistics from the Utility power system’s faults and
more accurate samples using this test procedure, it is possible
to know the probabilities and help to adopt Ethernet networks
for circuit breaker trips.
D. Determining Time Delay in network due to RSTP
Two tests were performed in order to determine the time delay
added by RSTP on GOOSE message circuit breaker trips.
Consider that the ring is closed physically but no logical link
exists between SW5 and SW6.
Test 1: Local Communication on Ring Fault
This test will verify that no delay exists for local traffic when a
RSTP reconfiguration is in progress. For this test the test set
and relay are connected to SW5 as shown above in Figure 4
The test sequence is as follows:
SW1 is energized by closing output 1 on the test set for
approximately one minute. This allows the network to
stabilize. After one minute, a power source failure is applied at
an instance (t) before a zone 1 fault is applied to the protection
relay which publishes a trip GOOSE message. The test set will
stop the test when the GOOSE message is received. If no
message is detected then the test was stopped by timeout.
Table 1 Local Communication measures on ring fault
Item Relay SW1 power source
Operation (ms) disconnection
(ms)
1 13,30 193,00
2 12,90 193,00
3 14,10 194,00
4 15,20 194,00
5 13,50 194,00
6 13,70 195,00
7 13,80 195,00
8 13,80 196,00
9 13,29 196,00
As can be seen from table 1, no time delay was added to
protection relay operation time and the tolerance of the relay
operation time is ± 2 ms. The LAN switch meets the design
consideration (a) criteria described in section A.
 X Simposio Iberoamericano Sobre Protección de Sistemas Eléctricos de Potencia
Test 2: Transmission Delay on Ring Fault
In this test, a time delay is expected if GOOSE message
transmission occurs during network reconfiguration as
described in consideration (b) in section A and as shown in
Figure 7. If the GOOSE publication is inside the shaded area, a
delay is expected. There exists natural indeterminism on
processors performance for publication, transmission, and
detection by protection relay, LAN switch and test set,
respectively, this causes different results in the same instant for
power source failure simulation.
The test procedure is similar to test 1. Test set is connected to
SW6 and protection relay is connected to SW5 as shown in
Figure 6.
Figure 6 Test connections used to determine SW1 fault
instance
As can be seen at table 2 bellow a time delay was measure
during the reconfiguration of the network. The time delays
measured were between 31.81 and 31.77 ms. This gives an
average of 3.5 ms delay per switch.
For the majority of the items no time delay was measured, this
was due to the GOOSE messge was published and transmitted
by the network before RSTP initiated. The LAN Switch meets
design consideration (b) described in section A.
Table 2: Time Delays measured on a ring fault
Item Relay SW1 power source
Operation (ms) disconnection
(ms)
1 46,40 194,00
2 12,90 194,00
3 13,50 194,00
321
4 15,10 194,00
5 14,20 195,00
6 13,70 195,00
7 45,00 196,00
8 45,20 196,00
Figure 7 Graphical results for item 1, 7 and 8(Table 2)
IV. CONCLUSIONS
Circuit breaker tripping using GOOSE messages can be an
option for Utilities. Network topology and switch constraints,
including a hardened hardware, must be considered in order to
ensure availability, dependability, security and maintainability
goals of the Power System and primary equipment. Consider
to use hardwired if network performance, including switches
transmission capabilities and devices fail over, don’t meet
your requirements.
When LAN tripping is used for busbar and breaker failure
protection schemes time delays due to the reconfiguration
process on a faulted ring could be a big issue, even when the
probability of a network and Power System's fault take place at
the same time is very low. Unless it is not an issue for Power
System stability or primary equipment damage, in a given
installation.
Power loss on switches is not an issue for GOOSE tripping if
the LAN switch can perform its task for a time enough, as
found in the switches used, because a low probability of
coincidence with a power system fault.
Statistical analysis is required to investigate the probability of
coincidence of a power system’s fault and network fault.
Consider the inherent non-deterministic response of computer
based devices, and circuit integrated behavior on power source
disconnection or other type of fails. For accurate results the
number of measures must be incremented up to 200 or at least
150 for each test.
322
For large network topologies, with more switches, time delay
due to ring reconfiguration, must be considered.
Reducing the number of switches on the network could be a
good solution, but no UTP could be used due to the distance
between IEDs and switches; use of fiber optic cable is
recommended. CFE has considered the use of a LAN Switch
per cabinet due to modularity and easy test at factory.
It’s important to reproduce these tests using fiber optic for
interconnection of switches and devices, in order to verify its
behavior.
A fail over test must be performed in the relay in order to
know its performance when the primary link to the main LAN
switch is lost – refer to section I and Figure 1. Fail over and
device performance to transmit GOOSE packages could be an
issue if it takes to much time. If this is the case, consider using
hard wire for trip and breaker failure initiation (50BF).
V. REFERENCES
[1]International Standard IEC 61850 – Comunications
networks and systems in substations, First Edition 2003.
[2]General Characteristics for Substation Automation Systems
based on IEC61850 Standard, Revision 2, Comision
Federal de Electricidad, January, 2009
[3]IEEE 802.1d-2004 IEEE Standard for local and
metropolitan networks - media access control (MAC) –
bridges, Jun 2004.
[4]Rapid Spanning Tree in Industrial Networks, Michael
Galea, RuggedCom Inc., 2006
[5]Latency on a Switched Ethernet Network , Application
Note 8,RuggedCom Inc,April 2008
[6]NRF-041-CFE – Normalized Protection Schemes for
Transmission Lines, first edition, Comision Federal de
Electricidad, 2005.
VI. BIOGRAPHIES
Daniel Espinosa received his B.S. in Electrical
Engineering from Instituto Politécnico Nacional at
Mexico City. He is a protection and automation engineer
at Comision Federal de Electricidad since 1998.. He has
started CFE's requirements on LAN Switches and LAN
architecture when GOOSE messages where considered
for tripping. He has been involved on engineering,
commissioning, testing and problems solutions at
substations using GOOSE tripping. He has been nominated to be CIGRÉ
member as a representative of Mexico at SC B5 Protection and Automation
and is working on CFE and National standardization committees for
protection and automation, mainly on IEC 61850 standard adoptions. He is
developing test scenarios related to GOOSE messages for protection trips and
improving documentation requirements for SAS in CFE.
Rene Aguilar received his B.S. in Electrical Engineering
from the University of Texas at Austin. He worked on an
APPDS (Automatic Protection Device Detection System)
used for detecting coordinating issues between devices in a
distributed generation system.
In 2006, he joined Megger as an application engineer in
the technical support group. He is in charge of developing
automatic testing for numerical relays as well as the
implementation of IEC 61850 on the Megger products. Rene has extensive
experience in the testing and commissioning of electrical schemes and multi-
vendor device applications of IEC 61850. He is a member of the IEEE and an
active member of Power System Relaying Committee PSRC.
James A riza ( M’03) received his B.S. in Electrical
Engineering from Universidad del Valle, Cali, Colombia.
He has extensive experience in the testing and
commissioning of electrical schemes, performing power
system studies and design, and electrical system fieldwork
supervision. Between 2000 and 2005 he worked as field
protection engineer and then as project manager at GERS
USA. He has previously worked with EPSA an electric
utility in Colombia and Fraunhofer IMBT an R&D
technology centre in Florida, US. In 2005, James joined Megger, where he
has served as an application engineer. He presently holds the title of senior
application engineer and he is in charge of relay OEMs accounts
management. He is instructor of hands on seminars covering theory and
practical application of IEC 61850. He is a member of the IEEE and an active
member of Power System Relaying Committee PSRC.
Jorge M endoza is a Field Application Engineer
working for RuggedCom, Inc. since November 2004
providing world wide post and pre-sales customer
support for Ethernet switches, routers and serial
servers.. Jorge has over 13 years of experience in
designing, implementing and troubleshooting complex
telecomunication networks. Prior to joining
RuggedCom Jorge was a Network Engineer for IBM Corporation and
Avborne, Inc. Jorge has also 5 year of experience working as Control Systems
and Instrumentation Engineer for complex Engineering projects in the oil
industry. Jorge graduated from the National Polytechnic University of
Venezuela (UNEXPO) with a Bachelor of Science in Systems Engineering.
Jorge also holds the Postgraduate Degree of Specialist in Telecom Systems
and Data Networks from the Electrical Engineering School of the Central
University of Venezuela (UCV)
Índice de Autores

Los números en negrita indican los artículos en los que el autor aparece como primer autor.
A Hedding, R. 263 Sanza, P. 123
Acosta, E. 199 Henneberg, G. 269, 317 Schauerman, G. 257
Adamiak, M. 123, 257, 209, 269 Herrmann, H.-J. 225, 133 Schiefen, M. 257
Aguilar, R. 317 Hosseini, S. 1 Schwartz, R. A. 159
Ahumada, A. 219 Sebastián, D. 95
Alcázar, E. 113 J Shulim, B. 87
Allan, S. 209 Janssen, M. C. 295 Smit, A. 225
Altuve, H. J. 113, 159 Soto, I. 199
Antonova, G. S. 55 K
Apostolov, A. 283, 247, 263 Kasztenny, B. 29 T
Ariza, J. 269, 317 Kereit, M. 65 Tahan, C. M. V. 205
Arzuaga, A. 137 Kezunovic, M. 269, 317 Tejada, D. A. 103
Krugger, M. 73 Tziouvaras, D. A. 159
B
Benckenstein, J. 295 L U
Benmouyal, G. 87 Lemus, H. E. 113 Udren, E. 269, 317
Beresh, R. 269 Liu, K. 1, 9 Urbaez, C. S. 179, 103
Betanzos, J. 113, 153 Lizárraga, J. 219
Blumschein, J. 65 Llano, L. E. 179 V
Brunner, C. 55 Lloyd, G. J. 1, 9 Vandiver, B. 247, 283
Burger, J. 29 López, J. C. 153 Ventura, C. A. 159
Verzosa, J. 269
C M
Cable, B. 257 Madani, V. 269 W
Calderon, J. A. 179 Martínez, A. 81 Wache, M. 133
Castro, J. A. 219 Martínez, P. 145 Wang, Z. 9
Claus, M. 65 McLaren, P. 269
Cordova, R. A. 235 Mejía, R. 81 Y
Costello, D. A. 159 Mendoza, J. 317 Yip, H. T. 1, 9
Mier, J. J. 81
D Miller, H. 29 Z
Davila, H. 17 Moreno, C. 219 Zadeh, M. 209
De Oliveira, C. C. B. 205 Moscoso, M. 1, 9 Zamorano, D. I. 219
Diego, L. del Á. 137 Mynam, M. 87
DoCarmo, H. 269
Dzienis, C. 65 N
Nájera, A. 95
E Novosel, D. 269
Espinosa, D. 317
O
F Ortiz, M. A. 145
Fischer, N. 29
P
G Palizban, A. 209
Gómez, Á. E. 219
Gong, Y. 87 Q
González, A. 81 Quintanilla, R. 137
González, M. V. 219
Gutiérrez, M. 73 R
Guzmán, A. 87 Rufato, E. J. 205

H S
Hajimiragha, A. 209 Sánchez, D. 113, 159
Hayes, P. 263 Sandoval, R. 159

323