Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Acta Wexionensia
No 33/2004
Terotecnología
Imad Alsyouf
En esta tesis se describe la función de costo de mantenimiento eficaz para lograr ventajas competitivas. Se
explora por medio de una encuesta que se utilizan prácticas de mantenimiento, y cómo las políticas de
mantenimiento se seleccionan en industrias suecas. Además, se sugiere un modelo para la selección de la
política de mantenimiento más rentable, y cómo mejorar la eficacia de las condiciones de mantenimiento de
toma de decisiones basada. Finalmente se discute la forma de evaluar el impacto de las prácticas de
mantenimiento en los objetivos estratégicos del negocio.
Los principales resultados obtenidos en la tesis son: 1) Una mejor comprensión de la organización de
mantenimiento, gestión, sistemas y el estado de mantenimiento en la industria sueca. Por ejemplo, se encontró que
alrededor del 70% de las empresas suecas siguen considerando el mantenimiento como un centro de costos.
También se hace hincapié en los enfoques de mantenimiento preventivo y predictivo. 2) La mayoría de las empresas
suecas, es decir, alrededor del 81%, utilizar el conocimiento acumulado y la experiencia dentro de la empresa como
un método para la selección de mantenimiento. Además, aproximadamente el 31% usa un método basado en el
modelado del tiempo hasta el fracaso y la optimización. Alrededor del 10% de efecto modo de fallo uso y análisis de
criticidad (FMECA) y árboles de decisión y sólo el 2% de uso de criterio múltiple toma de decisiones (MCDM).
Para lograr ventajas competitivas, las empresas deben hacer lo correcto, por ejemplo, utilizar la política
más efectiva los costes de mantenimiento, y deben hacerlo bien, por ejemplo, asegurarse de que tienen la
competencia derecho. Además, deben aplicar el ciclo de mejora sin fin, es decir, Plan-Do-Check-Act, que
requiere la identificación de las áreas problemáticas mediante la evaluación de los ahorros y beneficios
generados por el mantenimiento y seguimiento del impacto económico del mantenimiento aplicado
v
política. Por lo tanto, sabrían donde las inversiones deben ser asignados para eliminar las razones
básicas por las pérdidas y aumentar el ahorro.
vi
Agradecimientos
Esta tesis es el resultado del trabajo que se ha llevado a cabo durante varios años de investigación y de la
enseñanza en la Universidad de Växjö. De diversas maneras, se ha involucrado a varias personas y
organizaciones que estoy ansioso por reconocer.
Me gustaría dar las gracias a mi supervisor profesor Basim Al-Najjar por su comentarios útiles y
constructivas, sugerencias, debate fructífero y apoyo continuo. Es co-autor de seis de los ocho
trabajos de investigación. También, gracias a mi colega Anders Ingwald que es un co-autor de dos de
los seis papeles.
Me gustaría dar las gracias a la Universidad de Jordania para la financiación de parte de mis gastos de
manutención durante el período que se registró en la Universidad de Lund. Además, estoy agradecido con
el Centro de Competitividad Industrial (CIC) para financiar mis estudios en la Universidad de Växjö desde
2001. Me gustaría dar las gracias a la Junta Nacional Sueca para el Desarrollo Tecnológico Industrial,
NUTEK y las empresas suecas StoraEnso Hylte AB, Volvo componentes camiones AB en Köping,
SKFCondition de vigilancia, ABB Alstom Power AB en Växjö y el terminal del anuncio sueca en la ciudad de
Alvesta y CIC para apoyar los proyectos de investigación.
Por otra parte, me gustaría expresar mi reconocimiento y agradecimiento a aquellos que han
contribuido a mi trabajo de investigación en una forma u otra, el Prof. David Sherwin, el Dr.
Dhananjay Kumar, los revisores anónimos de mis artículos publicados, los investigadores de los
que he aprendido mucho leyendo y analizando su trabajo de investigación publicado. Asimismo,
agradezco al personal del departamento de mantenimiento en Stora Enso Hylte AB y Magnus
Magnusson en la terminal del anuncio sueca en Alvesta. Además, reconozco las aportaciones de
los responsables de mantenimiento de las empresas suecas que participaron contestando el
cuestionario. Expreso mi reconocimiento de mis colegas y amigos en la escuela de ingeniería
industrial de la Universidad de Växjö que participó en una forma u otra. Me gusta agradecer al Dr.
Staffan Klintborg para la corrección de la tesis,
Al final, agradezco a todos los que han contribuido a mi trabajo al menos por una palabra o un pensamiento
durante este largo período.
vii
Lista de documentos adjuntos
Esta tesis para el grado de Doctor en Filosofía (Terotecnología) se basa en una colección de ocho
trabajos de investigación. Los documentos se adjuntan en el Apéndice C y se hace referencia en el
texto de la tesis por su respectivo número,
por ejemplo, papel que, teniendo en cuenta que se clasifican con respecto a su adecuación a las preguntas de
investigación y no de acuerdo a su orden cronológico:
papel II Al-Najjar, Alsyouf y Ingwald (2004), un modelo práctico para la selección y mejora
de la más rentable Política de Mantenimiento: Parte I. Para ser presentados en la
Conferencia Internacional de Sociedades de mantenimiento (ICOMS), 25-28 de
mayo de 2004, Sydney , Australia.
papel Alsyouf (2001), Concepto cuadro de mando integral adaptado para medir el
VIII rendimiento de mantenimiento: Un estudio de caso, Actas de COMADEM 2001,
Universidad de Manchester, Reino Unido, 227-234 (Elsevier).
viii
Explicación de algunos términos
Definiciones adoptadas por los investigadores a menudo no son uniformes, por lo que algunos términos clave y
controvertidos se explican para establecer posiciones tomadas en la investigación de doctorado.
Fabricación ágil
Una respuesta global a los desafíos de negocios de sacar provecho de rabiosamente cambiantes, que
fragmentan continuamente, los mercados globales de alta calidad, alto rendimiento, y los bienes de los
clientes configurados y servicios. Goldman (1995)
comunalidad
La cantidad total de la varianza un original acciones variables con todas las demás variables incluidas en el
análisis. Hair et al. (1998)
Eficacia
La realización de lo 'correcto' a tiempo, y dentro de los requisitos de calidad especificados. Sink y
Tuttle (1989)
Eficiencia
Es una medida del grado de economía se utilizan los recursos de la empresa en la prestación de un
determinado nivel de requisitos. Sink y Tuttle (1989)
valor propio
La cantidad de varianza explicada por un factor. Se calcula como la suma de las cargas de la columna al
cuadrado para un factor. Hair et al. (1998)
ix
Análisis factorial
Es un método estadístico que se puede utilizar para analizar las interrelaciones entre un gran número de
variables y explicar estas variables en función de sus dimensiones subyacentes comunes (factores). Se
considera como una base objetiva para la creación de escalas sumadas. Hair et al. (1998)
Medida KMO
Una prueba estadística, llamada Kaiser-Meyer-Olkin medida, de Muestreo Adecuación utiliza con el análisis de
factor. Hair et al. (1998)
Producción Lean
El término 'Lean' proviene del uso de menos de todo en comparación con 'producción en masa'. Se centra en la
mejora de la productividad mediante la reducción de costes. Goldman (1995)
Mantenimiento
Se define como la combinación de todas las acciones administrativas y técnicas asociadas destinadas a
retener un elemento en, o restaurarlo a un estado en el que puede realizar su función requerida. BS 3811: 1993
concepto de mantenimiento
El conjunto de las diversas intervenciones de mantenimiento (correctivo, preventivo, conditionbased, etc.),
y la estructura general en la que estas intervenciones son llevados juntos. Pintelon et al. (1999)
función de mantenimiento
Se define por la Sociedad de Ingeniería de Mantenimiento de Australia (MESA) como: “Las decisiones de ingeniería y
acciones asociadas necesarias y suficientes para la optimización de la capacidad especificada”. Cuando en esta
definición “capacidad” es la capacidad de realizar una función específica dentro de un rango de niveles de
rendimiento que pueden relacionarse con la capacidad, velocidad, calidad y capacidad de respuesta. Tsang et al. (1999)
Manejo de mantenimiento
Todas las actividades de la gestión que determinan la estrategia de mantenimiento, objetivos y
responsabilidades y aplicarlos recurriendo a medios tales como la planificación de mantenimiento, control
de mantenimiento y supervisión, mejora de los métodos de la organización, incluyendo los aspectos
económicos. BS 3811: 1993
X
operaciones
Se trata del proceso de transformación que consiste en tomar las entradas y convertirlos en la salida junto
con las diversas funciones de apoyo estrechamente asociados con esta tarea. Hill (2000)
Escala ordinal
Es el siguiente nivel más alto de precisión de medición de escalas no métricas después de la escala nominal.
Las variables pueden ser ordenados o clasificados con escalas ordinales en relación a la cantidad del atributo
poseído. Cada subescala se puede comparar con otro en términos de una “mayor que” o “menor que” las
relaciones. Hair et al. (1998)
Actuación
El nivel al que se alcanza una meta. Dwight (1999)
Mantenimiento preventivo
Cualquier tarea diseñada para prevenir fallos o mitigar sus efectos. Sherwin (2000)
Productividad
Es la relación entre lo que sale de un sistema de organización (suponiendo que la salida cumpla con los
atributos establecidos para ellos) dividido por lo que entra en un sistema de organización (es decir, mano
de obra, capital, materiales, etc.) durante un período de tiempo determinado . Sumanth (1998)
Rentabilidad
Es el mejor indicador global del rendimiento de la empresa; que mide el resultado de todas las
decisiones de gestión sobre ventas y precios de compra, los niveles de inversión y producción, y la
innovación, así como el reflejo de la eficiencia subyacente con el que las entradas se convierten en
productos. Rantanen (1995)
Escala de proporción
Es una de las dos escalas métricas, es decir, escalas de intervalos y escalas de razón, proporciona el más alto nivel
de precisión de la medición, lo que permite casi todas las operaciones matemáticas a realizar. Sin embargo, a
diferencia de las escalas de intervalos que tienen un punto cero arbitrario tal como escalas de temperatura Fahrenheit
y Celsius, escalas de razón
tener un punto de cero absoluto que indica una cantidad de cero. Hair et al. (1998)
xi
Tasa de respuesta
El porcentaje de encuestados en la muestra inicial de la que se obtienen respuestas completas. Es el
principal índice de la calidad de los datos en una encuesta, ya que define el alcance de un posible sesgo
de no respuesta. Judd et al. (1991)
componente significativo
Un componente se considera significativa sólo si su probabilidad de causar un fallo de costosos o peligrosos
no es despreciable. Al-Najjar (1997)
Terotecnología
Lo que solía ser definido como una combinación de gestión, financieros, de ingeniería, construcción y otras
prácticas que se aplican a los activos físicos en la búsqueda de los costos del ciclo de vida económico. Pero
más tarde en el siguiente se añadió “Su práctica se ocupa de la especificación y diseño para la fiabilidad y
facilidad de mantenimiento de las instalaciones, maquinaria, equipos, edificios y estructuras, con su instalación
y sustitución, y con la retroalimentación de la información sobre el diseño, rendimiento y costes ”. “Los costos
del ciclo de vida” podrían ahora, con ventaja, ser sustituidos por “ganancias Vida de ciclo” en lo anterior. Sherwin
(2000)
xii
Total de la eficacia del proceso global (TOPE)
Es una extendida OEE / OPE. Se calcula como el producto de la OEE / OPE con un nuevo índice llamado
el índice operativa planificada. El índice operativa prevista se calcula como {el tiempo de producción
teórica, por ejemplo, 1 año menos (el vacaciones planificadas y mayor tiempo de descuento planificada)}
dividido por el tiempo de producción teórica. Alsyouf (2001)
xiii
abreviaturas
xiv
Contenido
xv
4.3.1 MCDM Fuzzy para la selección mantenimiento ................................. 42
4.3.2 Un modelo práctico para la selección de mantenimiento ........................... 44
4.4 condición efectiva toma de decisiones de mantenimiento basado ................ 47
4.4.1 modelo mecanicista para predecir el nivel de vibración .............. 48
4.4.2 La mejora de la eficacia de los sistemas de toma de decisiones ....... 49
4.5 Mantenimiento de la contribución a los objetivos estratégicos de negocio .............. 51
4.5.1 Valoración de coste de mantenimiento, el ahorro y el beneficio. .............. 51
4.5.2 Mantenimiento impacto sobre la productividad y la rentabilidad ............. 54
4.5.3 Un enfoque estratégico para medir el rendimiento de mantenimiento .... 58
5. Resultados, Conclusiones e implicaciones ........................................... .............. 60
5.1 Resultados de la investigación y conclusiones ............................................. ......... 60
5.1.1 Prácticas de mantenimiento ............................................. .................. 60
5.1.2 Mantenimiento selección ............................................. .................. 61
5.1.3 Selección de la política de mantenimiento más rentable .............. 62
5.1.4 Eficacia de la toma de decisiones CBM ................................... 62
5.1.5 Evaluación de la contribución de mantenimiento ................................. 63
5.2 Tesis contribución ................................................ ............................ 64
5.3 Implicaciones para la teoría ............................................... ........................sesenta y cinco
xvi
1. Introducción
1.1 Antecedentes
La creciente competencia en el mercado crea una necesidad urgente de buscar nuevas formas en
que las empresas de fabricación pueden diferenciarse y obtener mejor posición competitiva. Al
examinar el debate sobre los mercados y los recursos se podría darse cuenta de la existencia de
dos puntos de vista opuestos, es decir, el punto de vista de adentro hacia afuera y la perspectiva
de afuera hacia adentro, De Wet y Mayer (1998). Los defensores de la 'perspectiva de adentro
hacia afuera' establecer claramente los dos supuestos fundamentales sobre los que se apoya la
teoría de recursos: las empresas tienen diferentes recursos y estos recursos no pueden ser
transferidos fácilmente a o copien, otras empresas, Barney (1991). Se argumenta que estos
recursos pueden ser la base de una ventaja competitiva si cumplen cuatro criterios: ser valiosos,
raros y difíciles de imitar y sustituir. Recientemente mucha investigación en la escuela de recursos,
es decir, el punto de vista de adentro hacia afuera, del pensamiento estratégico ha pasado de
centrarse en los activos tangibles como fuente de ventajas a los activos intangibles, que incluyen
los conocimientos, competencias básicas, el aprendizaje, y 'activos invisibles' imagen de marca o la
cultura corporativa, Pehrsson tales como ( 2000). Por otro lado, los partidarios de la 'perspectiva de
afuera hacia adentro' argumentan que dos cuestiones centrales subyacen a la elección de la
estrategia competitiva. En primer lugar, los estrategas deben seleccionar un dominio competitivo
con características atractivas y luego deben posicionar a la empresa en lo que respecta a las cinco
fuerzas competitivas encontradas. Estas cinco fuerzas son la entrada de nuevos competidores, la
amenaza de sustitutos, el poder de negociación de los compradores, el poder de negociación de
los proveedores, y la rivalidad entre los competidores existentes, Porter (1985).
El éxito en cualquier contexto competitivo depende de ofrecer un valor superior al cliente (es decir, valor
ventaja) o el funcionamiento con menores costos relativos (es decir ventaja de costos) o, idealmente, ambos,
véase, por ejemplo, Porter (1985), Christopher (1998) y De Wet y Mayer (1998). La supervivencia de cualquier
negocio depende de su capacidad para competir de manera efectiva, Madu (2000). Las ventajas competitivas
se producen cuando una empresa utiliza sus recursos y capacidades para desarrollar las competencias de
organización que, a su vez, crean valor para los clientes, sagú (2003). Como respuesta a los retos planteados
por un entorno empresarial, por ejemplo, aumento de la competencia global, muchas empresas de fabricación
están buscando maneras de obtener ventajas competitivas con respecto al costo, el servicio, la calidad y el
tiempo
1
entrega. Por otra parte, el foco se ha movido de la producción con bajo, que se centra en la reducción de los
costos totales hacia 'Fabricación ágil' que se centra en el aumento de los ingresos totales, Goldman et al. (1995).
En consecuencia, la estructura de la empresa de fabricación ha pasado de ser una industria intensiva en mano
de obra a una capital intensivo en tecnología, es decir, intensiva, la industria. El patrón de producción ha pasado
de la producción en masa a la producción de muchas variantes para satisfacer las necesidades diversificadas, es
decir, de empleo-shop, y, finalmente, a un modelo separado para cada cliente o la personalización en masa.
Muchos cambios en el ambiente interno de las empresas están llevando a cabo: el aumento del uso
de la mecanización y automatización de las operaciones, tales como sistemas de fabricación flexible
(FMS), robots, almacenamiento automático,
vehículos guiados automáticos (AGV); el aumento de las tendencias de la utilización de Just-In-Time
(JIT), y la filosofía TQM, Yamashina (1995), Luxhoj et al. (1997) y Suito (1998). Estos cambios enteras
atan de capital mucho invertido, por ejemplo, las empresas dentro de las industrias de proceso y
químicos, tales como fábricas de papel y refinerías, un uso extremadamente caros y totalmente
automatizados líneas de producción, Swanson (2003). Por otra parte, existe una creciente presión para
proteger el entorno ecológico del peligro de los residuos industriales nocivos y la contaminación. Esto
significa que la planta de fabricación se debe utilizar con eficacia, eficiencia, y proporciona productos de
alta calidad a un precio competitivo además de mostrar preocupación por el medio ambiente y la
seguridad.
En el movimiento hacia la fabricación de clase mundial muchas empresas se están dando cuenta de
una necesidad crítica para el uso de un mantenimiento adecuado, es decir, eficiente y eficaz, de las
instalaciones y sistemas de producción, Luxhoj et al. (1997) y Stephen (2000). plantas industriales,
maquinaria y equipos son cada vez más avanzados tecnológicamente y al mismo tiempo más complejo y
difícil de controlar. sistemas de gestión de JIT, fabricación magra y ágil y el uso de sistemas
automatizados e integrados han hecho los sistemas de producción cada vez más vulnerables a los
riesgos y susceptibles a diversos efectos consecuentes debido a averías, Luce (1999), Vineyard et al.
(2000), y Holmberg (2001). Por ejemplo, la implementación de JIT requiere un mantenimiento efectivo y
eficiente que puede asegurar un flujo suave de producción e, idealmente, un coste de calidad 100 por
ciento de eficacia, Charlene (1989) y Al-Najjar (1996). El mantenimiento es una función de negocio que
sirve y apoya el proceso primario en una organización. El proceso de mantenimiento se suma al valor del
cliente en términos de beneficio, calidad, tiempo y servicio, Zhu et al. (2002). Por lo tanto, la función de
mantenimiento se hizo más esencial
Por lo tanto, la importancia de la función de mantenimiento ha sido mayor que antes, debido a su papel
en el mantenimiento y la mejora de la disponibilidad, eficiencia en el rendimiento, productos de calidad,
entregas a tiempo, el medio ambiente, la seguridad
2
et al. (1997), McKone y Weiss (1998) y Bevilacqua y Braglia (2000). Por otra parte, una conciencia cada
vez mayor de mantenimiento y su influencia tanto para las empresas industriales y de la sociedad en su
conjunto puede ser reconocida. Muchos investigadores y profesionales han destacado las pérdidas
totales debido a la omisión de mantenimiento o ineficacia, Ahlmann (1984 y 1998), Jardine et al. (1996),
Alnajjar (1997), Davies (1998), Ljungberg (1998), Luce (1999), Vineyard et al. (2000) y Holmberg (2001).
Sin embargo, el mantenimiento sigue siendo considerado como un centro de coste y poca investigación
se ha hecho para poner de relieve el impacto de la función de mantenimiento en el rendimiento global de
la planta, es decir, la productividad y la rentabilidad, Ahlmann (1984 y 1998), Al-Najjar (2000a), Al -Najjar
et al. (2001), Carter (2001) y Kutucuoglu et al. (2001).
Podemos ver que la tarea de mantenimiento es cada vez más compleja debido a los cambios
en la producción y el medio ambiente de las empresas. Estos cambios pueden ser descritos por
factores tales como el nivel de automatización y la intensidad de capital asociado a las líneas de
producción automatizada, la globalización, la reestructuración y las estrategias de reducción de
personal, estructuras de organización, el desarrollo de la competencia del personal y la dificultad
de evaluar el impacto del mantenimiento de la competitividad de las empresas. Se ha dado
cuenta de que un sistema de fabricación típico consiste no sólo de los componentes mecánicos,
sino también de otros elementos tales como,, elementos electrónicos hidráulicos
electromecánicos, software y los seres humanos.
problemas de decisión de mantenimiento podrían clasificarse con respecto a la escala de tiempo en cuestión. Se inicia
temprano en la fase de diseño de los sistemas; el tipo de equipo, el nivel de redundancia, y la accesibilidad que afecta
fuertemente la capacidad de mantenimiento, Dekker y Scarf (1998). Además, una decisión muy crítico debe hacerse con
respecto a qué evento (por ejemplo, fallo, el paso del tiempo, etc) desencadena qué tipo de mantenimiento, es decir, inspección,
reparación o reemplazo. Por lo general, el objetivo es reducir el mantenimiento fallos de instalación, maquinaria y equipo /
componente, mejorando así la productividad general de la planta. Este objetivo puede alcanzarse usando varios enfoques:
mantenimiento correctivo; el cambio de un componente a la vez pre-especificada usando modelos estadísticos basados en los
datos fracaso histórico recogido; mantenimiento basado en la condición a través de la vigilancia de la condición del componente
utilizando una (o más) de la monitorización de estado técnicas (CM). En todos los casos, el tomador de decisiones trata de
seleccionar entre todos los posibles de mantenimiento se acerca un enfoque para cada pieza de equipo o componente. Sin
embargo, las prácticas actuales en las decisiones de plantas y mantenimiento de los equipos y servicios de sustitución se basan
frecuentemente en opiniones informadas tales como seguir las recomendaciones del fabricante de equipos originales (OEM), o
las respuestas subjetivas a situaciones comunes, tales como la reacción a un fallo de un componente crítico mediante la
introducción de una el tomador de decisiones trata de seleccionar entre todos los posibles de mantenimiento se acerca un
enfoque para cada pieza de equipo o componente. Sin embargo, las prácticas actuales en las decisiones de plantas y
mantenimiento de los equipos y servicios de sustitución se basan frecuentemente en opiniones informadas tales como seguir las
recomendaciones del fabricante de equipos originales (OEM), o las respuestas subjetivas a situaciones comunes, tales como la
reacción a un fallo de un componente crítico mediante la introducción de una el tomador de decisiones trata de seleccionar entre
todos los posibles de mantenimiento se acerca un enfoque para cada pieza de equipo o componente. Sin embargo, las prácticas actuales en las decisiones de plantas
3
programa de toda la compañía de reemplazo o sustitución preventiva basado en la condición de tales
componentes.
Sin embargo, mientras tales procedimientos para establecer un programa de mantenimiento puede mejorar
la fiabilidad de la planta, es de ninguna manera garantizado para proporcionar la solución más rentable, Jardine
et al. (1996). La identificación y
implementación del enfoque de mantenimiento adecuado permitirá a los administradores para evitar los
costos de sustitución prematuras, mantener la capacidad de producción estables y controlar el deterioro del
sistema y de sus componentes, Viña et al. (2000). Esto significa que la industria podría mejorar su
rendimiento si se implementa el enfoque de mantenimiento adecuado para la eliminación de las causas de
las perturbaciones de producción, Swanson (2001 y 2003).
Esta investigación trata teórica y empíricamente el grado en que se implementan las prácticas de
mantenimiento adecuadas y los vínculos entre las prácticas de mantenimiento aplicadas y el rendimiento
general de la empresa. Con base en los resultados de la investigación presentada en la fase de licenciatura,
se demostró en un estudio de caso que el costo de mantenimiento eficaz mejoraría la calidad, la eficiencia y
eficacia de las operaciones de una empresa. Por lo tanto, esto podría mejorar su competitividad, ventajas
de productividad es decir, ventajas de valor y rentabilidad a largo plazo. En consecuencia, los accionistas,
los clientes, y la sociedad se verían afectados positivamente, como se ilustra esquemáticamente en la
Figura 1.1.
Por otra parte, se ha demostrado que el uso de uno de los enfoques de mantenimiento,
mantenimiento basado en la vibración es decir (VBM) para planificar actividades de mantenimiento podría resultar
en grandes ahorros, especialmente cuando el costo tiempo de inactividad es alta, por ejemplo, en la industria de
procesos. Para generalizar estos resultados, hay una necesidad de investigar las prácticas de mantenimiento, que
se utilizan en diversos tipos de industrias y cómo estas prácticas influyen en la competitividad industrial, Adebiyi et
al. (2003) y Tse (2002). prácticas de mantenimiento en este estudio incluyen actividades como el mantenimiento
previsto, monitoreo de la condición (CM), mantenimiento autónomo, el análisis técnico, la educación y la formación
del personal, los sistemas de planificación y control de trabajo, los sistemas expertos, la multitarea y el trabajo en
equipo. La extensión del uso de tales actividades influye en los resultados de rendimiento de negocio.
Normalmente, diversas acciones de mantenimiento se utilizan para reducir las fallas de la planta industrial,
maquinaria y equipos y sus consecuencias tanto como sea posible. Estas acciones pueden tomar varias formas
tales como el mantenimiento de fallos basada (FBM), el mantenimiento preventivo (PM), es decir, la sustitución de
componentes en un tiempo pre-especificada usando modelos estadísticos sobre la base de datos de fallas
históricos recopilados, o mantenimiento conditionbased (CBM). En todos los casos, la toma de decisiones, sin
embargo, tiene que seleccionar entre todas las políticas de mantenimiento aplicables al más rentables para cada
componente, módulo o equipos.
4
accionistas
Sociedad Cliente
valor rentabilidad
Ventajas de Ventajas de
productividad
Ventajas competitivas
Mantenimiento rentable
Figura 1.1. modelo conceptual que muestra cómo el mantenimiento podría afectar a la competitividad de las empresas
5
1.3 Objetivo y preguntas de investigación
El propósito de esta investigación es estudiar el impacto de las prácticas de mantenimiento en el resultado del
desempeño de las empresas. El objetivo se logra a través de la investigación de las siguientes preguntas de
investigación:
RQ5: ¿Cómo evaluar el impacto de las prácticas de mantenimiento de los objetivos estratégicos
del negocio?
RQ1
papel I
papel II
PI2
papel III
PI3
Documento IV
PI4
papel V
papel VI
papel VIII
Figura 1.2. Vínculo entre las preguntas de investigación y los trabajos de investigación
6
La dependencia y la relación entre las preguntas de investigación se pueden describir de la siguiente manera. La
primera pregunta de investigación examina empíricamente las prácticas de mantenimiento en la industria sueca. A
partir de este se obtiene una mejor comprensión de la importancia
cuestiones tales como el estado de mantenimiento, organización de mantenimiento, sistemas de
gestión de mantenimiento y las políticas implementadas mantenimiento. La segunda pregunta de investigación
investiga empíricamente (dentro de la industria sueca) qué métodos se utilizan para la selección de la política de
mantenimiento aplicada. A partir de este identificamos los métodos de selección de mantenimiento utilizados y su
capacidad de satisfacer las necesidades de la empresa. La tercera pregunta de investigación ayuda a la toma de
decisiones en la selección de la política de mantenimiento más rentable basado en los resultados teóricos obtenidos
en el Libro I y II de papel y los resultados empíricos obtenidos en el Libro III.
Una vez que hemos seleccionado la derecha, es decir, el más rentable, la política de mantenimiento,
entonces deberíamos aumentar su eficacia. Por lo tanto, la cuarta pregunta de investigación ayuda al gerente de
mantenimiento en la mejora de la eficacia de la toma de decisiones en la aplicación de mantenimiento basado en
condición (CBM). Por último, la quinta pregunta de investigación es sobre cómo evaluar teórica y empíricamente
el impacto financiero de las prácticas de mantenimiento en los objetivos estratégicos del negocio, es decir, la
competitividad de las empresas.
1.4 Relevancia
A continuación se discute la relevancia del problema de investigación y las preguntas de investigación, mostrando
por qué es importante hacer una investigación en esta área. También se muestra cómo el problema de
investigación se aborda en este estudio es muy importante tanto por razones teóricas y prácticas como podría ser
justificado a través de:
7
sólo un gasto impredecible e inevitable, Al-Najjar (2000a), Ralph (2000), Sherwin (2000) y
Al-Najjar et al. (2001).
La necesidad de obtener resultados estadísticamente rigurosas y generalizadas utilizando métodos sistemáticos y
cuantitativos, es decir, una encuesta a través de una muestra grande, para mejorar los resultados de la investigación que
1.5 Delimitaciones
En esta tesis, las tres primeras preguntas de investigación están restringidas a la industria sueca.
La población estudiada se compone de un conjunto de empresas de fabricación sueca que tienen
al menos 100 empleados 1. Sin embargo, en la cuarta pregunta de investigación, nos ocupamos de
mantenimiento basado en condición (CBM) en el mantenimiento general y basada en la vibración
(VBM), en particular. Esto se debe a la monitorización de vibraciones se considera generalmente
como una de las herramientas clave de la mayoría de los programas de monitoreo de condiciones
para la rotación y máquinas de movimiento alternativo, Collacott (1977), Bloch y Geitner (1994) y
Al-Najjar (1997). En la quinta pregunta de investigación, la empresa estudio de caso se selecciona
de una industria que se caracteriza por tener el capital invertido-alta con un alto costo de tiempo
de inactividad, es decir, la industria del papel, donde suele ser justificado el uso de la política de
VBM. Por último, aunque estamos estudiando la función de mantenimiento en general, hay
algunas áreas de investigación en el campo de mantenimiento que no se tratarán, como el
mantenimiento y el diseño, métodos de reparación, programación de mantenimiento,
Capítulo Dos describe los principales aspectos metodológicos: el concepto metodología, métodos de
investigación, diseño de investigación y diseño de investigación de tesis.
Capítulo Tres revisa y resume la literatura relevante para establecer un marco de referencia
para las preguntas de investigación, que vale la pena investigar
---------
1 Más detalles acerca de las características de la población van a aparecer en el capítulo de metodología.
8
porque son controvertidos o no se ha respondido adecuadamente por los investigadores anteriores. El
propósito de este capítulo es guiar al lector en la literatura existente que se considera que es
importante para el problema de investigación.
Capítulo Cuatro presenta los resultados empíricos de la investigación que se llevó a cabo para resolver el problema
de investigación y sus preguntas de investigación establecidos. Muestra los datos recogidos y el tratamiento y análisis
de los mismos en función de su relevancia a las preguntas de investigación y trabajos de investigación.
Capítulo Cinco continuación se analizan los resultados de la investigación, es decir, resultados y conclusiones, para
cada pregunta de investigación y el problema de investigación. Por otra parte, se analiza implicaciones para la teoría y la
práctica, sugerencias para futuras investigaciones y finalmente mi propia crítica del estudio.
9
2. Metodología
Es muy importante tener una posición clara y evidente con respecto a la metodología de un estudio de
investigación. La metodología seleccionada afecta a la validez, la fiabilidad y resultados de la investigación.
Por lo tanto nos ilustran cómo se puede seleccionar y diseñar la metodología de investigación.
Teoría de la Metodología
ciencia
Paradigma paradigma
presunciones enfoques Área de estudio
operativa
Ultimate metodológicos
metodología de investigación
Realidad
ideográfico
ética Analítico
caso encuesta
Ciencia
Conceptual
10
presunciones Ultimate
Las presunciones en última instancia definen vista del investigador del mundo social y la forma
en que puede ser investigada. Muestra cómo el investigador se ve en la realidad, los ideales, la
ciencia, la ética, etc. Estos supuestos, según lo sugerido por Burrel y Morgan (1979), podrían ser
agrupados utilizando la dimensión subjetiva-objetiva como enfoque “objetivista” o enfoque
“subjetivista”. El investigador puede ver la realidad como objetivo “ahí fuera” independiente del
investigador, algo que se puede medir de manera objetiva mediante el uso de un cuestionario o
un instrumento. Por otro lado, uno puede ver la realidad como única construida por los individuos
involucrados en la situación de investigación, por lo tanto existen múltiples realidades en
cualquier situación dada. Sin embargo, Arbnor y Bjerke (1997) añade que muchos científicos
analíticos se refieren a la intersubjetividad en lugar de la objetividad.
Al considerar los supuestos sobre terrenos del conocimiento, el investigador debe determinar su
posición sobre la cuestión de si el conocimiento es algo que puede ser adquirido por un lado, o algo que
tiene que ser experimentado personalmente, por otro lado. En cuanto a la relación entre el investigador y
lo investigado, el enfoque del “objetivista” implica que el investigador debe permanecer distante e
independiente de lo que está siendo investigado. Así, en las encuestas y experimentos, los investigadores
intentan controlar el sesgo, la selección de una muestra sistemática y ser objetivo en la evaluación de una
situación. Mientras que en la postura “subjetivista”, los investigadores interactúan con los que estudian,
viviendo con la observación de los informantes o durante un período prolongado de tiempo, o mediante la
colaboración real.
En esta tesis, el autor considera a sí mismo como un investigador más cerca del enfoque de
“objetivista”, debido a su formación científica en ingeniería. Esto significa que el autor ve la realidad
como objetivamente accesible, independiente y objetivamente medible. El conocimiento puede ser
adquirido, y el investigador debe permanecer distante e independiente de lo que se investiga. Sin
embargo, dado que lo que se considera como objetivo por un cierto entorno social podría considerarse
subjetivo por el punto de vista de otra comunidad, por lo tanto, hago hincapié en que esta tesis se centra en una
comunidad de investigadores dentro de la ciencia de mantenimiento y personas dentro de una fábrica o de un entorno
de negocios.
enfoques metodológicos
Arbnor y Bjerke (1997) ilustran que el conocimiento puede ser desarrollado usando uno de los
siguientes tres enfoques metodológicos: enfoque analítico, enfoque de los sistemas, o enfoque de los
actores. El enfoque analítico asume que la realidad es objetiva y tiene un carácter sumativo: “el todo
es la suma de sus partes”. El enfoque de sistemas asume que la realidad objetiva es accesible. La
realidad está dispuesto de tal manera que el conjunto se diferencia de la suma de sus partes. La
relación entre las partes en sí y entre las partes y el medio ambiente son muy importantes. El
conocimiento depende de los sistemas. Las partes se explican (entendida) en términos de todo el
sistema. El enfoque de los actores asume que el todo existe sólo como estructuras de significado, que
se construyen socialmente.
11
El conocimiento depende de los individuos. El conjunto se entiende a través de ámbitos finitos de los actores de
significado. Se asume que la realidad es una construcción social. Los estudios cualitativos se utilizan con este
enfoque.
Mantenimiento como una función de soporte es parte del sistema de fabricación de la compañía, lo
que significa que afecta y se ve afectada por las otras partes del sistema, por ejemplo la estrategia de
producción. Por lo tanto, el autor cree que el enfoque metodológico más cercano es el enfoque de
sistemas, ya que mediante este enfoque se puede estudiar los efectos mutuos de cualquier parte del
sistema con respecto a las otras partes y el medio ambiente, y que todavía mantiene los supuestos
“objetivistas”.
paradigma operativo
El paradigma operativo describe la relación entre el enfoque metodológico y el área bajo estudio. Se
determina en función de los procedimientos metodológicos utilizados para capturar los datos,
analizar y sacar conclusiones. Jonsson (1999) declaró que las técnicas de investigación y
resolución son ya sea de carácter empírico o conceptual (teórico). Burrel y Morgan (1979), Larsson
(1993), y Bengtsson et al. (1997), entre otros, los términos utilizados nomotéticos (leyes generales y
procedimientos para la ciencia exacta) e ideográfico (la comprensión de los casos particulares) para
representar las metodologías de investigación cuantitativa y cualitativa, respectivamente. Las
características de los enfoques nomotético e ideográfico, además de la metodología de la encuesta
caso se ilustran en la Figura 2.2.
ideográfico nomothetic
Number of aspects
(Questions)
Caso encuesta
Numero de casos
El enfoque Idiographic se basa en un enfoque de estudio de caso orientado al proceso que hace
hincapié cualitativos (estudio interpretativo y explicativo) multi-aspectos
12
y pocos estudios en profundidad, cubriendo a menudo un largo periodo de tiempo con el objetivo de explicar y
comprender. Su objetivo es proporcionar descripciones ricas y hacer generalizaciones teóricas. Esto está en
contraste con el enfoque Nomothetic, que se ocupa de los análisis cuantitativos de algunos aspectos a través
de grandes muestras (casos) con el fin de probar las hipótesis y hacer generalizaciones estadísticas utilizando
métodos sistemáticos y cuantitativos para describir y explicar la causalidad.
En esta tesis tanto la Idiographic (cualitativa) y Nomothetic enfoques (cuantitativos) fueron utilizados
con diferentes preguntas de investigación y en diferentes períodos de estudio de investigación. Por
ejemplo, papel VI, VII papel y papel VIII se basan en un enfoque ideográfico, mientras que el papel III
se basa en un enfoque nomotético. Sin embargo, el resto de los papeles podría considerarse teórica.
Estudios de caso
Yin (2002) definió estudio de caso como una investigación empírica que investiga un fenómeno
contemporáneo dentro de su contexto real, especialmente cuando los límites entre el fenómeno y el contexto
no son claramente evidentes. Sekaran (2000) aclaró que implican estudios de casos en profundidad, análisis
contextual de las situaciones similares en otras organizaciones. Puede basarse en cualquier combinación de
pruebas cuantitativas y cualitativas. No siempre es necesario incluir, observaciones detalladas directos como
fuente de pruebas. Los estudios de casos son considerados como una forma menos deseable de la
investigación de los experimentos o encuestas debido a la falta de rigor, la poca base científica para la
generalización, y debido a que tardan demasiado tiempo, Gomm et al. (2000). Sin embargo, Yin (2002) aclara
que la falta de rigor también existe en las otras estrategias de investigación, pero en la investigación de
estudio de caso se ha encontrado con mayor frecuencia y superar con menos frecuencia. Criticó la
13
vista jerárquica mediante el cual se disponen los diversos métodos de investigación. Puso en duda la concepción común de que
los estudios de casos sólo son adecuados para la fase exploratoria de una investigación que las encuestas son apropiadas para
la fase descriptiva, y que los experimentos son la única manera de hacer preguntas explicativas o causales. Hizo hincapié en que
lo que distingue a los métodos no es esta jerarquía, pero las siguientes tres condiciones: el tipo de pregunta de investigación
planteada, el grado de control tiene un investigador, y el grado de enfoque en los acontecimientos contemporáneos. “Cómo” y
“por qué” son propensos a favorecer el uso de estudios de casos si no hay necesidad de tener el control de la conducta caso, y
cuando se centra en los acontecimientos contemporáneos. En cuanto al problema de la generalización, que mostró que los
estudios de casos, como los experimentos, podría generalizarse a propuestas teóricas y no a las poblaciones o universalmente.
Ilustró que un defecto fatal en la realización de estudios de casos es concebir la generalización estadística como el método de
generalizar los resultados del estudio de caso. Esto se debe a que los casos no son “unidades de muestreo” y no deben ser
elegidos por este motivo. Por lo tanto, el modo de generalización utilizando estudios de caso es “generalización analítica”, en el
que se utiliza una teoría desarrollada previamente como una plantilla con la que comparar los resultados empíricos del estudio de
caso. Esto se debe a que los casos no son “unidades de muestreo” y no deben ser elegidos por este motivo. Por lo tanto, el
modo de generalización utilizando estudios de caso es “generalización analítica”, en el que se utiliza una teoría desarrollada
previamente como una plantilla con la que comparar los resultados empíricos del estudio de caso. Esto se debe a que los casos
no son “unidades de muestreo” y no deben ser elegidos por este motivo. Por lo tanto, el modo de generalización utilizando
estudios de caso es “generalización analítica”, en el que se utiliza una teoría desarrollada previamente como una plantilla con la
experimentos
Los experimentos se llevan a cabo por los investigadores en prácticamente todos los campos de la investigación, por lo
general para descubrir algo acerca de un proceso o sistema en particular. Montgomery (2001) experimento se define
como una prueba o una serie de pruebas en las que se realizan cambios intencionados a las variables de entrada de un
proceso o un sistema de modo que se puede observar e identificar las razones de los cambios que se pueden observar en
la respuesta de salida. Gomm et al. (2000) contrasta la investigación experimental con estudios de casos. Demostraron
que la investigación experimental por lo general implica la investigación de un pequeño número de casos en comparación
con la encuesta de trabajo, y lo que la distingue de estudio de caso no es tanto la cantidad de datos recopilados como el
hecho de que se trata de un control directo de las variables. En los experimentos, los investigadores crean los casos
estudiados (s), mientras que los investigadores del estudio de casos construir casos de situaciones que ocurren
naturalmente sociales. Yin (2002) hizo hincapié en que “cómo” y “por qué” son propensos a favorecer el uso de
experimentos si hay una necesidad de tener un control de la prueba de comportamiento, y cuando se centra en los
acontecimientos contemporáneos.
Encuesta
Graziano y Raulin (2000) mostraron que la investigación de la encuesta utiliza varios procedimientos básicos
de investigación para obtener información de las personas en sus ambientes naturales. El instrumento
básico que se utiliza es la encuesta, que es un conjunto de una o más preguntas que pregunta a la gente
acerca de varias cuestiones. La encuesta se pudo realizar por la elección entre dos tipos de instrumentos de
estudio: (1) cuestionarios, en la que los participantes lean las preguntas y luego escribir sus respuestas, y (2)
encuestas de entrevista, en la que los participantes escuchan las preguntas y hablar de sus respuestas ,
Mitchell y Jolley (2001).
14
A diferencia del experimento de investigación del investigador de la encuesta no manipula variables, pero impone
algunas restricciones en los participantes mediante el uso de los instrumentos de la encuesta. Podría ser utilizado para
poner a prueba la relación entre las variables. Yin (2002) puso de relieve que “qué” y “quién” y “dónde” (o sus derivados
“cuántos” y “cuánto”) son propensos a favorecer los métodos de encuesta. Un estudio de encuesta es un método
apropiado para utilizar, por ejemplo, cuando las preocupaciones estudio de búsqueda de características distintas en una
población (ibid). Puede ser una forma relativamente barata de obtener información acerca de las actitudes, creencias y
comportamientos de las personas mediante la recopilación de una gran cantidad de información sobre una muestra a
gran escala en un corto período de tiempo, Mitchell y Jolley (2001). Una comparación entre los tres métodos de
investigación, estudios de casos es decir, experimentos y la encuesta se ilustra en la Tabla 2.1.
Tabla 2.1 Comparación entre los tres métodos de investigación (desarrollado a partir de Gomm
et al., 2000, Yin (2002), entre otros)
Forma de pregunta de ¿Cómo por qué? ¿Cómo por qué? Quién, qué, dónde, cuántos, ¿cuánto?
investigación
el caso estudiado en sí, sin tampoco, la inferencia partir de una muestra a una población finita,
ningún interés en la teórica, el desarrollo y aunque esto es a veces visto como una
de la intervención.
(preguntas)
Tipo de casos y nivel Estudio de casos que ocurren Estudio de casos creados Estudio de una muestra de casos de origen
de control de forma natural, donde la de tal manera como para natural; seleccionado de tal manera que se
está controlando las variables. importantes. la población más grande, donde la principal
Tipo de datos La cuantificación de los datos no La cuantificación de los La cuantificación de los datos es una
como superior
15
2.3 Diseño de la investigación
diseño de la investigación es la secuencia lógica que conecta los datos empíricos a preguntas de investigación
iniciales de un estudio y, en última instancia, a sus conclusiones, Yin (2002). Describió diseño de la investigación
como un plan de acción para ir desde aquí a ahí,
dónde aquí puede definirse como el conjunto inicial de preguntas a ser respondidas, y
ahí es un conjunto de conclusiones sobre (a) las respuestas a estas preguntas. La definición de la pregunta de
investigación es la condición más importante entre otras condiciones, tales como la necesidad de control sobre la
conducta de eventos, y el grado de enfoque en los acontecimientos contemporáneos. Otra forma de pensar en el
diseño de investigación es como un “modelo” de la investigación, se trata de al menos cuatro aspectos: qué
preguntas para estudiar, ¿cuáles son relevantes los datos, qué datos para recoger, y cómo analizar los resultados,
según lo declarado por Yin ( 2002). El objetivo principal del diseño es ayudar a evitar la situación en la que la
evidencia no se ocupa de las preguntas iniciales de investigación. Yin (2002) identificó cinco componentes que se
consideran importantes para un diseño de investigación: las preguntas de un estudio; su propuesta, en su caso; su
unidad (s) de análisis; la vinculación lógica de los datos a las proposiciones; y los criterios de interpretación de los
resultados. Aunque la sustancia de la preguntas de investigación variará, la forma de pregunta - en términos de
“quién”, “qué”, “dónde”, “cómo” y “por qué” proporciona una pista importante en relación con el método de
investigación más relevantes para ser utilizado. En cuanto a proposición de estudio, si los hubiere, que ayuda a dirigir
la atención hacia algo que debe ser examinada dentro del alcance del estudio. El tercer componente, es decir, el unidad
de Análisis, está relacionada con la definición de lo que la investigación está a punto. Podría ser sobre un individuo,
un evento, una entidad tal como una máquina, un departamento, una función, una empresa, una industria, etc.
(ibid).
La validez de constructo: que tiene que ver con el establecimiento de medidas operativas correctas de los
conceptos que se está estudiando, por ejemplo, los nombres dados a las medidas son correctas, es
decir, para estar seguro de que el instrumento mide lo que se supone que debe medir. Esto requiere que
el instrumento entonces se debe administrar de una manera apropiada, normalizada de acuerdo con los
procedimientos prescritos.
Validez interna: para estudios explicativos o causales, que tiene como objetivo garantizar que un determinado
evento observables (variables de entrada) era responsable de la influencia o un cambio en el comportamiento (de
salida). Arbnor y Bjerke (1997) añadió que podría
dieciséis
estar preocupado con la relación lógica (relevancia) entre un estudio y la teoría existente en
la zona.
Validez externa: la posibilidad de generalizar los resultados más allá de la superficie real en estudio.
los fiabilidad prueba, por el contrario, tiene como objetivo minimizar los errores y sesgos en el
estudio. Esto demuestra que las operaciones del estudio, tales como procedimientos de
recolección de datos se pueden repetir con los mismos resultados y conclusiones, Yin (2002). El
enfoque de sistemas considera que el problema de la validez de manera algo diferente. Las
conexiones entre la teoría, las definiciones, y la fiabilidad no son tan fuertes como lo son en el caso
de análisis. Un enfoque de sistemas comunes para garantizar que las mediciones son correctas es
reflejar el sistema real desde tantos ángulos como sea posible. Los investigadores tratan de
estudiar el sistema, siempre y tan a menudo como sea posible, hablar con tantas personas como
sea posible, y para estudiar el material pertinente que sea posible. En el enfoque de los actores
hablan de la calidad y la credibilidad del análisis cualitativo.
rigurosas técnicas y métodos para la recopilación de datos de alta calidad, que se analizan con
cuidado, con atención a las cuestiones de validez, fiabilidad, y la triangulación.
Yin (2002) añade que, además de la fiabilidad, la credibilidad, conformabilidad y de datos pruebas de
confiabilidad que se utilizan para juzgar la calidad de diseño de investigación, pruebas tales como la validez de
constructo, la validez interna, validez externa y pruebas de fiabilidad también son relevantes para los estudios de
caso .
17
diseño de la investigación 2.4 Tesis
Esta tesis consiste en cinco preguntas de investigación que se relacionan con ocho trabajos. La metodología utilizada en
estudiar
Participant
The empirical study was performed by conducting a cross sectional survey to obtain information
from the maintenance managers of the Swedish firms (production plants) that have at least 100
employees. Large plants will probably invest significantly in technology and are likely to require
an internal maintenance group to care for equipment. We wanted to make comparisons across a
wide range of groups of Swedish industries, aiming to make rigorous and statistical generalisation
from the studied sample to a larger population, i.e. Swedish industry. We got the addresses of
1440 Swedish firms from Statistics Sweden “Statistiska Centralbyrån” (SCB). The population was
selected from the Swedish Standard Industrial Classification (SE-SIC) 2002.
18
At first, it was decided that we survey every member of the selected population, i.e. 1440
firms. However Mitchell and Jolley (2001) warned that even if one starts with an unbiased
sample, by the end of the study the sample may become biased because people often fail or
refuse to respond to a questionnaire. However, they showed that a typical mail survey response
rate might reach only 10 percent. A mail survey with a return of 30 percent or so is often
considered satisfactory, Emory and Cooper (1991). However, Jonsson (2000) mentioned that
response rates from similar studies in the USA, Australia, New Zealand and Singapore vary from
about 10 to 40 percent, with a median response rate around 20 percent.
Out of the 1440 questionnaires that we sent by surface mail, 38 questionnaires were removed
from the population for various reasons. However, the total number of respondents was 185. This
means that the response rate is 13.2%. Furthermore, it was found that there is bias to certain
types of industries. Therefore, in order to improve the possibility to generalise the results we
decided to restrict the studied population to the industries that have a high response rate, they
were also characterised by having expensive down time due to highinvested capital, which also
mean that the maintenance could have big role. This means that we can generalise to only the
studied population and not to the Swedish industry in general. The size of the restricted population
becomes 539 and the total number of respondents is 118, hence, the overall response rate
increased to about 22%, as shown in Table 2.3.
Table 2.3 Distribution of the response rate with respect to type of industry.
Petrochemical 5 4 80.0%
19
promote rapport the questionnaire started with general questions that are considered more
interesting and also easier to answer.
To guarantee that the questionnaire fulfils the construct validity requirements,
i.e. being adequate and measuring what it is supposed to measure, a special table was created to
test if the questionnaire covered all the aspects of the research objectives considered necessary,
such as technical, financial, organisational, environmental and educational aspects. As can be seen
in Paper III, every aspect as well as every objective was covered by at least one question.
Furthermore, one of the most important criteria of questionnaire design is that the questions should
be short, relevant and easy to understand. At the same time it is very important to make sure that
the respondents have interpreted the questions as intended. Therefore, the questionnaire was
designed with structured questions where the respondent could choose an answer from a set of
listed possibilities. This is recommended for mail surveys, and when necessary an open-ended
option is used to guarantee that the respondent has interpreted the question as intended. This will
improve the validity and reliability of the instrument. To guarantee a high degree of content and
construct validity, e.g. a concept being studied such as the names given to the measures, the
questionnaire was based on the related theory and literature and pre-tested by academics and
practitioners. At the same time all the procedures for data collection were written down in detail to
ensure the repeatability of the process to enable good reliability.
Procedure
The designed questionnaire passed through the procedures of drafting the questionnaire,
pre-testing, questionnaire finalising and production, first mailing, first reminder using mail, and
second reminder using telephone. Based on the study objective and the main research problem,
questions were drafted to cover all relevant aspects. Both the question content and the wording
were selected carefully. Then the questionnaire (written in English) underwent a series of tests.
The first test was performed by the designers as part of the research project activities, and then it
was presented in a seminar as part of the yearly meeting with the centre of industrial
competitiveness (CIC). After that other colleagues and PhD students at the department tested a
draft of the questionnaire translated into Swedish. All their comments were considered and
incorporated in a revised version. The quality of translation was controlled and tested in a way that
guarantees interactive and back translation during all the stages.
Then the Swedish version of the questionnaire was tested on the field by sending the
questionnaire with a covering letter to one of the companies participating in financing the research
project. The maintenance manager of the Swedish terminal post in the town of Alvesta was asked
to write down his reflections on how clear the questions were and how easy they were to answer.
After that, the questionnaire was revised and retested again with the same manager. When we
were satisfied with the questionnaire it was reproduced and mailed.
A mail survey was used to collect the empirical data. In a mail survey the respondent can take
more time to collect facts and talk with others. Another
20
advantage of a written questionnaire is the avoidance of potential interviewer bias. However, the
major weakness of the mail survey is the non-response, Emory and Cooper (1991).
Paper V is related to the fourth research question. It is a conceptual study. Its objective is to
investigate how to improve the effectiveness of plant monitoring and enhance the reliability of
decision-making systems. The study was based on performing a critical analysis of the related
literature.
21
paradigms, these papers are classified as “Idiographic”, i.e. qualitative
methodology, using case study research. However, quantitative data and analysis were
implemented according to the scientific approach which is common among natural scientists and
engineers. A case study was conducted in the natural environment of the studied organisation,
where researchers did not interfere with the normal flow of work, i.e. the researchers remained
distant and independent of what is being researched. The unit of the analysis of the study was one
of the Swedish paper companies, where technical and economic data were collected from one of
the company’s four machines, i.e. PM2. In-depth and practically relevant analyses of the complex
organisational and manufacturing processes took place for a long period of time, i.e. about 16
months.
We used previously collected data, i.e. secondary data covering the period from 1997 to 2000,
about the machine and its environment. We designed our own tables for collecting data, because
the required data were not possible to find in one database in the required format. To avoid any
confusion during data collection, and ensure that the instrument, i.e. data collection table,
measures what it is supposed to measure, all the terms used in the tables of data collection were
discussed, explained, and agreed on by the company personnel involved in the case study. In
addition, to guarantee that the measurements were correct, we tried to study the system as long
and as often as possible, to talk to as many people as possible, and to study as much relevant
material as we could. Furthermore, the research team was characterised by being well-trained,
experienced in methodological skills, competence, and rigorous supervision, which ensured the
credibility of the researchers. Finally, we can ensure the validity of the measurement, i.e. the
extent to which the results are true and represent reality by showing how the results compare to
the following criteria suggested by, among others, Arbnor and Bjerke (1997), Mitchell and Jolley
(2001), Patton (2002) and Yin (2002):
Face validity
The degree of acceptance of the results, i.e. their credibility, was examined in several ways. At
first by discussing the findings with the company of the case study, and then by presenting and
discussing the results with the project’s steering committee, which consists of representatives of
the organisations that supported the research project financially, i.e. SKF, Volvo, Alstrom Power,
and StoraEnso. Next, the project results were printed as a report and distributed to the project
supporting organisations and universities. After that, the findings were presented and discussed
as scientific papers at
international scientific
conferences both in Sweden and the UK. Also, the result in the form of a scientific paper, i.e.
Paper VI, was published in a scientific journal of a high reputation, i.e. ranked with an “A” rating 2. Furthermore,
two other papers, i.e. Paper VII and Paper VIII have also been sent for publication in an “A” rated
journal.
–––––––––
2
The European Journal of Operational Research, for the ranking see
http://hal.boku.ac.at/fao/journal_ranking
22
Internal validity
We can say that the study also has a high internal validity, because the findings of the study are
relevant to and a logical output of the theory developed at the beginning of the project. This study
was able to prove quantitatively for the first time that the maintenance function should no longer be
considered as a cost centre, but that it is a profit generating function.
External validity
Yin (2002) showed that normally the findings of a case study could be generalised to theoretical
propositions. He argues that the findings of case studies could not be generalised to populations
or universally because of the lack of rigorousness, statistical and quantitative evidence. However,
Lukka-Kari and Kasanen-Eero (1995) believe that statistical and case studies are not as far from
one another with regard to generalising as is commonly considered. They argue that
generalisation from statistical studies is only one of the possible modes of generalisation. They
claimed that “in case studies an attempt is usually made to counterbalance the impossibility of
applying statistical inference by, for example, the large theoretical or practical relevance of the
research subject, the thoroughness of analysis and interpretation, and the triangulation of
research methods”.
Therefore, properly conducted case studies of a high quality can produce generalisable results,
where inferences are not drawn to some larger population based on sample evidence, but
generalisation are rather made back to theory. However, Yin (2002) believes that generalisation is
not automatic, because a theory must be tested through replications of the findings in a second or
even a third neighbourhood. Lukka-Kari and Kasanen-Eero (1995) stated that the significance of
replication is emphasised by the idea that the generalisability of research results could be
substituted by their transferability. Therefore they argue that generalisation is possible from a
properly conducted case study. One way of realising this is by building an argument that the
substantial results of a case study also hold true for other cases.
Based on the previous discussion, we argue that the findings of this study can be generalised
both as theoretical propositions and to other similar populations or universally. This is because the
case study was characterised by being an indepth study for about sixteen months. We believe that
the findings can be transferred, to other paper machines within the same company and among
other paper mills. Furthermore, the theory can be applied to similar types of industries such as
chemical, process industries, and fully automated manufacturing industries, which have high
downtime costs.
23
3. Literature Survey
This chapter aims to build a theoretical foundation upon which the research is based by reviewing
the literature relevant to identifying research issues, which are worth researching. Based on this
chapter the research problem and research questions that were presented in Sections 1.2 and 1.3
are discussed with respect to other researchers’ work and publication. Figure 3.1 shows the
study’s theoretical foundation, illustrating the relationship to the theory of knowledge related to the
research problem. However, more a specific and comprehensive related theory is presented in the
respective research papers.
Maintenance (3.2)
RQ1 RQ4
RQ2 RQ3
Processes (3.3)
RQ5
The chapter starts by discussing the overall management strategy in Section (3.1) showing the
lack of evidence of linkage between maintenance practices and overall corporate strategy. Then
Section (3.2) discusses maintenance practices, reviews maintenance approaches, considers
maintenance selection methods, illustrates maintenance efficiency and discusses the need to
improve plant monitoring decision-making accuracy. Next, Section (3.3) discusses the impact
24
of maintenance practices on competitive advantages through its impact on business processes. It
discusses maintenance costs, savings and profits. And finally it deals with business performance
measurement.
Each business or firm needs to develop a clearly established overall management strategy to
satisfy its customers’ needs at minimum costs and keeping up the best image. Then each
function, e.g. the maintenance function, will have its own strategy within the overall management
strategy context. This is despite the fact that many authors and practitioners have acknowledged
that maintenance is a major contributor to the performance and profitability of manufacturing
companies, see among others Maggard and Rhyne (1992), Foelkel (1998), Ralph (2000),
Swanson (2001) and Mitchell et al. (2002). There is evidence of the lack of linkage between
maintenance practices and overall corporate strategy, see Ahlmann (1998), Tsang (1998),
Coetzee (1999), Tsang et al. (1999), Kutucuoglu et al. (2001) and Mitchell et al. (2002). The
maintenance manager must be able to convince top management that maintenance strategies or
policies and overall management strategy are interdependent, and that maintenance could
produce economic benefits. He must therefore learn to communicate in the language of money,
Ahlmann (1984 and 1998).
3.2 Maintenance
Different authors have defined maintenance variously, see among others Moubray (1991),
Pintelon and Gelders (1992), (BS 3811:1993), Al-Najjar (1997), Mike et al. (1997). The definitions
differ in their scope, i.e. the target of maintenance varies from concentrating on an item only to
including the whole process. Tsang et al. (1999) stated that the scope of maintenance
management should cover every stage in the life cycle of technical systems (plant, machinery,
equipment and facilities): specification, acquisition, planning, operation, performance evaluation,
improvement, replacement and disposal.
25
engineering and technical analysis, personnel education and training, systems for planning and
controlling work, expert systems, multitasking and team work are considered very essential to
achieving world-class performance, Liptrot and Palarchio (2000), Mitchell (2002), and Mitchell et
al. (2002). Although good maintenance practice could be “common sense” there is a need to know
if it is adopted as “common practice”, and to what extent good maintenance practices are
deployed. Furthermore, one must know how to assess the impact of maintenance practices on
performance outcomes, Mitchell et al. (2002).
Proactive + Predictive
Diagnostic/Prognostic
Preventive
Reactive
Kelly (1997) showed that maintenance strategy involves the identification, researching and
execution of many repairs, replacement and inspection decisions. It is concerned with formulating
the best life plan for each unit of the plant, and formulating the optimal maintenance schedule for
the plant, in coordination with production and other functions concerned. Maintenance strategy
describes what events (e.g. failure, passing of time, condition) trigger what type of maintenance
(inspection, repair or replacement). Maintenance strategy consists of a mix of policies/techniques,
which varies from facility to facility, Dekker (1996), Al-Najjar (1997) and Zeng (1997). It depends
on several factors such as the goals of maintenance, the nature of the facility or the equipment to
be maintained, work flow patterns (process focus, product focus), and the work environment, see
among others Gallimore and Penlesky (1988), Pintelon and Gelders (1992) and Al-Najjar (1997).
Usually, maintenance actions are used to control failures of industrial plant, machinery and
equipment. These actions can take several forms and make use of various approaches:
corrective (breakdown) maintenance, preventive
maintenance (PM), i.e. replacing components at a pre-specified time using
26
statistical models based on collected historical failure data, or condition-based maintenance using
data from monitoring the condition of the
component/equipment through utilising one (or more) of the condition monitoring (CM)
techniques. In all cases, the decision maker needs to select from all the applicable maintenance
policies the most cost effective one for each component, module or equipment that suits the
operating context. The
identification and implementation of the appropriate maintenance policy will enable managers to
avoid premature replacement costs, maintain stable production capabilities and prevent the
deterioration of the system and its component parts, Vineyard et al. (2000). Knowing what is the
right maintenance program for an asset is no easy task. It has nothing to do with the number of
years the company has been doing maintenance. For the most part, companies are doing either
too much maintenance too early, or too little too late, all of which has cost consequences to the
organisation, Liptrot and Palarchio (2000).
Making a cost-effective maintenance decision is not an easy task, especially when the
production system consists of several different components with different maintenance
characteristics and the maintenance program must combine technical requirements with the
firm’s managerial and business strategies.
In the literature, it is possible to find methods that can help in selecting the most cost effective
maintenance policy or action such as the models used for maintenance optimisation, e.g. age,
block and GTTT-plots, see for example Sherwin (2000), Campbell and Jardine (2001) and
Al-Najjar (2003). Furthermore, there are sets of rules, such as the systematic approach of
maintenance planning applied by reliability centred maintenance (RCM) for selecting the suitable
maintenance action, O’Connor (2002). Additionally, there is the method of multiple criteria
decision-making (MCDM), which is implemented in some cases. All these methods have different
strengths and weaknesses, for example block and age models only deal with PM and require
historical failure data, GTTT-plots demand condition-based replacement or failure data, the
RCM-procedure does not consider the organisational aspects and MCDM-method does not
consider technical analysis before data gathering. Thus, there is a need to develop a method for
maintenance selection that is characterised by incorporating the strengths of the above-mentioned
methods and avoiding their weaknesses.
27
3.2.4 Plant monitoring and decision making accuracy
Mechanical and electrical systems in power stations, paper mill machines, hydraulic systems, etc.
consist of many sub-systems, components and modules such as rolling element bearings, pumps,
motors and gearboxes. Replacing the mechanical parts at the right time, which could be achieved
by using conditionbased maintenance (CBM) or age replacement at cost-optimised interval,
reduces the contribution of each part to the system rate of occurrence of failures (ROCOF) and
therefore reduces the ROCOF itself, Sherwin (2000).
If we can assess the condition of the machine’s significant components accurately, we can
prolong the mean effective life length of the component, AlNajjar (2000b). Thus, the machine will
not fail so often, and we will not perform maintenance far too frequently. This would improve the
overall equipment effectiveness (OEE) and consequently result in higher company savings and
enhanced competitiveness.
However, in many cases, in spite of using cost effective maintenance approaches which may
be based on sophisticated and advanced techniques, e.g. vibration based maintenance (VBM),
machines still experience failure and unplanned but before failure replacement (UPBFR), which
result in high economic losses, Al-Najjar (1997). UPBFR replacements are performed at
unplanned but before failure stoppages to prevent the occurrence of failures. This situation arises
because of a sudden increment in the measured variable(s), e.g. the vibration level, due to the
inability to detect the cause(s) of the defect at an early stage. Davies (1998) reported that at
present there is insufficient knowledge to understand and predict the behaviour of rotating
machinery, which may lead to unpredictable systems failure and an expensive shutdown.
However, Sheppard and Scicon (1993) believe that the vital data required for accurate condition
monitoring are now available. However, extracting the important information from the mass of
data is the problem. Therefore, further research and development are needed in the area of
evaluation and analysis techniques of condition monitoring and diagnostics, to provide more
effective, accurate and precise knowledge of plant or system conditions, see among others
Davies (1998), Al-Najjar (1997), and Williams et al. (1994).
28
equipment/component useful life, equipment redundancy, extra spare parts inventory, buffer
inventory, damage to equipment due to breakdown, extra energy consumption, etc. On the other
hand, the capability of the machine to produce quality products, e.g. products that satisfy
customer requirements, is highly affected by maintenance effectiveness, Henning (1989),
Taguchi et al. (1989), Oakland (1995), Al-Najjar (1997), Ahlmann (1998), and Edwards et al.
(1998). Maintenance affects the technical performance and cost effectiveness of the production
department, according to Ollila and Malmipuro (1999). The technical performance of the
production function can be assessed by so-called overall equipment effectiveness (OEE) in TPM,
see Nakajima (1988), or a modified version of OEE, i.e. the Overall Process Effectiveness (OPE),
see Sherwin (2000). Efficient maintenance contributes by adding value through better resource
utilisation (higher output), enhanced product quality, and reduced rework and scrap (lower input
production costs). In addition, it avoids the need for additional investment in capital and people
due to expanding the capacity of existing resources, Git (1992 and 1994), Ben-Daya and Duffuaa
(1995), AlNajjar (1997), Ahlmann (1998), Dunn (1998), Ralph (2000), and Swanson (2001).
Maintenance can have an impact on customers, society, and shareholders. Dearden et al.
(1999) showed that firms try to capture new customers, satisfy them and retain existing customers
by giving them assurance of supply on time, which in turn depends on adequate production
capacity with a minimum of disturbances and with high quality products. The impact of
maintenance on society can be traced through its effects on safety, on the environment and on
ecology, see Keller and Huwaishel (1993) and Rao (1993). Finally, the impact of maintenance on
shareholders can be traced by analysing the effect of maintenance on the generated profit, which
is usually measured by indexes such as Return on Investment (ROI) percentage, Ahlmann (1984
and 1998). From this we can see that there is a need to know how to assess the impact of
maintenance practices on performance outcomes, Mitchell et al. (2002).
Recently more emphasis has been put on maintenance as a profit generating function. There
is talk now of life cycle profit rather than life cycle cost, see among others Ahlmann (1994 and
1998), Jonsson (1999), Wilson (1999), and
29
Sherwin (2000). Many researchers have been talking about the savings, gains or profits that could
be made when implementing more effective maintenance approaches, see among others Ahlmann
(1984 and 1998), Maggard and Rhyne (1992), Foelkel (1998), Coetzee (1999), Walsh (1999), Miller
(2000), Ralph (2000), Carter (2001), Kutucuoglu et al. (2001), and Swanson (2001). Nothing,
however, has been mentioned about how to calculate/estimate the relevant life cycle cost factors,
or where to find the required information parameters from the available accountancy system. This
is because the impact of the maintenance function can be found in many areas in the company
such as production, quality, logistics, etc., Al-Najjar (2000a) and Al-Najjar et al. (2001). Dunn
(1998) illustrated that when a breakdown happens, it is often easy to show that a lack of
maintenance was responsible. But when breakdowns do not happen, it is not easy to demonstrate
that maintenance had prevented them. It is easy to say that maintenance costs so much per year
but not what is the gain of that maintenance, and how it can be measured.
The importance of maintenance to a business strategy can be paradoxical, see Dunn (1998)
and McGrath (1999). On the one hand, the more maintenance contributes positively to the overall
strategic goals of an organisation the less noticeable as a value adding activity it becomes to top
management. Instead it might be noticed as just more costs. On the other hand, poor maintenance
programs can obstruct the addition of value, retard the advantage of a capital resource, and
destroy a business strategy. Appropriate performance measurement systems are crucial to ensure
the successful implementation and execution of strategies, since measurement provides the link
between strategy and action, see Neely et al. (1995) and Schalkwyh (1998). Therefore, integrating
maintenance
30
into the overall company strategy is essential especially in capital-intensive industries. Actually,
there is a need for a holistic performance measurement system that can assess the contribution
of the maintenance function to the business strategic objectives, see among others Ahlmann
(1998), Tsang (1998), Tsang et al. (1999), and Muthu et al. (2000).
31
4. Cost Effective
Maintenance for Competitive
Advantages
This chapter presents the results of the research that are achieved through solving the research
questions. It introduces the collected empirical data and their analyses according to the sequence
of the research questions discussed in Chapter Two. The relevant literatures, problem formulation
and data collecting procedures are presented in the respective research papers.
Notice! The connection between the research questions and research papers was illustrated in
Section 1.3. Chapter Five will discuss the findings of Chapter Four and draw general conclusions
and implications.
In the following we present the results obtained for the first research question connected to Paper
III in Appendix C.
Data were collected from 118 Swedish manufacturing firms in a survey conducted in the way
discussed in the methodology chapter. It was found that about 28% of the firms have no
maintenance strategy or policy at all, about 48% have a written maintenance strategy or policy and
24% have an oral one. Also, it was found that about 39% of the firms have a maintenance
department that is organi s ationally independent of the production department, while about 56 % are
organised as part of the production, whereas about 5% have other organisational relations with
production such as an independent company within the mother company that sells service to the
production plants. On the other hand, it was found that about 41% have a centralised organisation,
15% have a decentralised organisation, 41% have a combination of centralisation and
decentralisation, and about 3% have other types of organisation.
32
It was found that about 50% of the time is spent on planned tasks, about 37% spent on unplanned
tasks and 13% allocated for planning. However, the causes of planned maintenance actions were
distributed on average so that 34% were the recommendation of the original equipment
manufacturer (OEM), 33% the use of condition monitoring (CM) techniques, 9% the use of statistical
modelling of failure data, 7% the use of Key performance measures and 17% other factors such as
those based on the company’s own experience, see Figure 4.1.
40
30
Percent
20
34 33
10 17
9 7
0
Modelling
Use of CM
Statistical
Others (Own
Experience)
Key Performance
Recommendations
Measure
OEM
The respondents were asked if they use a computerised maintenance management system at
their plants. About 22% answered that they use a manual maintenance management system, about
36% use a computerised maintenance management system (CMMS), about 35% use a
combination of a manual system and CMMS and about 7% answered that this question is not
applicable in their plant, see Figure 4.2.
40
35
30
25
Percent
20
36 35
15
10 22
5
7
0
Manual Ccomputerised Combination of Others
Manual and
Computerised
33
The respondents were asked if they use a CMMS, and if they do that to rate, using a 5-point
Likert scale, the level of integration of the CMMS with other company information systems. The
Likert scale measures the extent to which a person agrees or disagrees with the question. The
scale used was “1=Not integrated” and “5= completely integrated”. It was found that about 29%
so not use a CMMS, an answer that confirms results illustrated in Figure 4.2 where the percent of
the respondents who use CMMS was 71%. However, among the users of CMMS about 20% are
using a CMMS that is not integrated with other company information systems, about 14% of these
systems are somewhat integrated, 17% are moderately integrated, 10% are highly integrated and
10% are totally integrated computerised maintenance management systems, see Figure 4.3.
25
20
Percent
15
10 20
17
14
5 10 10
level of Integration
Regarding the attitude to maintenance, it was found that about 70% consider maintenance as
a cost centre, 26% consider it as a profit centre and 4% think that it is both a cost and a profit
centre. On the other hand, the respondents were asked to estimate the maintenance budget in the
year 2002 as a percentage of the production cost. Only 49 valid answers were received, in which
the average estimate was about 11% with a range from one to fifty percent. Also, when estimating
the budget as a percentage of the turnover, it was found to be on average about 4% with a range
from 0.3 to sixty percent. However, the distribution of the maintenance budget among the different
tasks and resources is illustrated in Figure 4.4
34
40
30
Percent
20 37
32
10 19
8 4
0
s
rts
ri e
es
in
pa
la
g
rc
iti
in
Sa
iv
ou
e
in
t
ar
ac
tra
s
Sp
ut
er
O
d
an
th
O
n
io
at
uc
Ed
Maintenance Task
Here we tried to empirically find a construct that can describe the maintenance practices in
Swedish industry using exploratory factor analysis. Factor analysis is a statistical approach that
can be used to analyse interrelationships among a large number of variables and to explain these
variables in terms of their common underlying dimensions (factors), Hair et al. (1998). Therefore,
we analysed the 26 variables, i.e. maintenance activities, mentioned in question M1, see Appendix
(A), using principal component factor analysis with varimax rotation.
At first we tried to ensure that the data are suitable for factor analysis. Therefore we tested all
the variables using both the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and
Bartlett’s test of sphericity. Thus we got a KMO value of 0.789 and a Chi-Square value of 1225
with significance 0,000, see Table b1 in Appendix B. This indicates that the data could be used
with factor analysis. The communality values, also confirm that the data are suitable for factor
analysis, since they range between 0.450 to 0.835, i.e. there are no values close to zero or one.
In addition, the measure of sampling adequacy (MSA) is in the range of 0.480 to 0.905. This
means that it is suitable for use in factor analysis since all the MSA values are greater than
one-half, and furthermore, the majority are greater than 0.70. Thus, we can derive new factors
from the 26 variables, Hair et al. (1998).
35
We determine the number of factors using an eigenvalue over one as an extraction criterion.
Consequently, we got seven factors that account for about 64% of the variation, see Table b1 in
Appendix B. Analysing the factor loading,
i.e. the correlation of each variable and the factor, can help deriving the new construct. The
loading indicates the degree of correspondence between the variable and the factor. It is the
means of interpreting the role each variable plays in defining each factor. It ranges between
{minus one: plus one}; the higher loading absolute value makes the variable more representative
of the factor. A loading was considered significant if it has an absolute value higher than 0.30, Hair
et al. (1998). Knowing that the communalities are defined as the amount of variance accounted for
by the factor solution for each variable, the researcher may as a guideline specify that at least
one-half of the variance of each variable must be taken into account (ibid). Thus, we can say that
more or less all the variables have a communality value greater than or equal to one-half, e.g.
only two variables had a value less than 0.5. Since we have got a factor solution in which all the
variables have at least one significant loading on a factor, we identified the following seven
factors:
36
4) The fourth factor ( diagnostic “ expert system” approach) includes:
Variable Loading
In the following we present the results obtained from the second research question that is
connected to Paper III in Appendix C.
As can be seen in Paper III, by means of factor analysis, we could identify three main groups from
the respondents’ answers of how much emphasis is placed on a set of aspects when deciding on or
selecting a maintenance policy. The first group is called the business oriented group because there
they try to achieve competitive advantages through concentrating on aspects such as machine
availability, reducing lost production costs, machine reliability, smooth production, product quality,
on-time delivery, replacement costs, cost effectiveness and investment costs. The second group is
labelled the greens because they focus mainly on aspects related to health, safety and the
environment. The third group was labelled the followers because they only work according to the
original equipment manufacturers’ recommendation and company policy.
It was found that most of the firms, i.e. 81%, do not use a specific method when selecting their
maintenance policy, but use the knowledge and experience accumulated within the company.
Besides, about 31% of the respondents use a method based on modelling the time to failure and
optimisation when selecting a
37
maintenance policy. While about 10% use FMECA and decision trees in their selection and only 2%
use MCDM. On the other hand, about 6% have mentioned using other maintenance selection
methods such as monthly lists, documentation and experience, major overhauls twice a year,
maintenance cost, manufacturer recommendations, risk analysis and their own databases, see
Figure 4.5.
100
75
Percent
50
81
25
31 10
2 6
0
MCDM
Experience and
Modelling the
Other Methods
time to failure
DecisionTrees
optimisation
Company's
FMECA or
and
knowledge
It should be noticed that about 30 percent of the respondents use combination of at least two
methods at the same time, for example, experience and FMECA or experience and modelling and
optimisation. This explains why the total percent illustrated in F igure 4.5 exceeds 100%.
Dependability on failure-data F3
User friendliness F4
38
The respondents were asked to rate the importance of these features using a 5point Likert scale.
The Likert scale measures the extent to which a person agrees or disagrees with the question. The
scale used was 1=Not Important and 5=Very Important. As can be seen in Figure 4.6, all the features
got a mode value greater than three, which confirms that these features should be available and
employed in an ideal maintenance selection method. The mode measure was used since we are
dealing with an ordinal scale, which means that the answers have an inherent order or sequence, and
therefore the mode or median could be used. However the mode is probably the most suitable
alternative for easy interpretation.
4
Importance (Mode)
3
5
2
4
4
3
F1 F2 F3 F4 F5 F6 F7 F8 F9 F10
Figure 4.6 The rank order of importance of maintenance selection method features as assessed by the
respondents
The respondents were asked to evaluate the maintenance selection method(s) that they use the
most at their plants with respect to the ideal features. They were asked to rate the capability of the
method using the 5-point Likert scale, where 1=Bad and 5=Good. An example of the assessment of
the method depending on
accumulated knowledge and experience is illustrated in Figure 4.7. As we can see, utilizing
personnel knowledge and experience, i.e. F4, obtained a mode value of five, which means that this
method is rated as completely good in 10% of the features. Also, we can see that this method is
deemed rather good in 30% of the features, i.e. its ability to be used across all maintenance
approaches (F1), its user friendliness (F3) and its ability to consider organisational aspects (F7). On
the other hand, it is considered completely bad in 10% of the features, i.e. its ability to measure cost
effectiveness (F8). The assessment results of the other methods are presented in Paper III.
39
Accumulated Knowledge and Experience
4
How Good in fullfilling the
(Mode Value)
3
feature
5
2 4 4 4
3 3 3 3 3
1
F1 F2 F3 F4 F5 F6 F7 F8 F9 F10
Figure 4.7 Assessment of how well accumulated knowledge and experience fulfil maintenance selection
method features
The maintenance selection methods used by Swedish industry were evaluated using MCDM
methodology. They were assessed with respect to their capability to fulfil the features of an ideal
maintenance selection method. A rank ordering was performed based on the mode value of the
respondents’ assessment of both the importance of the features of maintenance selection methods
and the capability of each maintenance selection method to fulfil these features. It was assumed
that the value of the mode corresponds to a metric measure with a ratio scale ranging between zero
and five. By means of the MCDM methodology suggested in Paper I a score representing the
capability of each selection method to satisfy the ideal features was obtained using the simple
additive weighting (SAW) method for decision-making. Then, a normalised rank order, with a range
between zero and one, was obtained by dividing the SAW score of each selection method by the
summation of the total SAW scores.
It was found that FMECA, which is related to the RCM procedures, obtain ed
the first rank according to the evaluation methodology. The Modelling and Optimisation method
came next, third came the knowledge and experience
method and finally the multiple criteria decision making (MCDM) maintenance selection method with
a small marginal difference, see Figure 4.8. Note that the
40
sample size used to evaluate the MCDM maintenance selection method is very small.
0,270
0,260
Normalised rank
0,250
0,267
0,240
0,250
0,244
0,230
0,239
0,220
Knowledge and Modeling and FMECA MCDM
experience optimisation
Figure 4.8 Rank order of maintenance selection method using MCDM methodology
41
4.3 A technique for selecting the most cost
effective maintenance policy
In the following we present the results obtained from the third research question connected to Paper
I, Paper II and Paper III in Appendix C. In this section we introduce a practical model that enables
the maintenance manager to select and improve the most cost effective maintenance policy. This
section describes the results achieved by answering the third research question:
A1 R~ 11 . R~ 1j . R~ 1n
. . . . . .
Ai R~ i1 . R~ ij . R~ in
. . . . . .
Am R~ m1 . R~ mj . R~ mn
42
Technical analysis of manufacturing processes, machines Past data
and components.
(4) Rank ordering the approaches using the simple additive weighting (SAW)
Assessment of the importance of considering the failure causes (criteria) using a fuzzy
linguistic variable.
Assessment of the capability of the maintenance approaches, using a Fuzzy Inference System
(FIS) that results in an assessment of the fuzzy linguistic variable
For more details, see Paper I, where two examples illustrate how the suggested evaluation
methodology could identify the most informative approach.
43
4.3.2 A practical model for maintenance selection
In Paper II, a review of the relevant literature identified the main methods that are suggested or
being used for selecting maintenance policies. These methods are: Maintenance Optimisation ( both
graphical and non-graphical); Reliability Centred Maintenance (RCM) and Multiple Criteria
Decision-Making (MCDM).
In the following we present a model that was suggested as a result of analysing the matrix diagram
illustrated in Table 1 in Paper II. It was possible to identify the strength and weakness of the available
maintenance selection methods, which enabled us to highlight the gaps in these methods.
Consequently, we developed a practical model for selecting and improving the most cost effective
maintenance policy.
Model development
The model proposed in Paper II and shown in Figure 4.10, is based on the continuous
improvement concept (i.e. Plan, Do, Check, Act) known as the PDCA-cycle. It is recommended
that an improvement group that includes representatives from all the relevant working areas and
is authorised by the top management perform this work. The planning phase consists of four main
stages:
44
Act Plan
Ini tiation of the selection process.
(Periodically or at need)
7. Maintenance
policy adoption
No Yes
2. Is past data available?
Check
Do
Figure 4.10 Conceptual model for selecting and improving the most cost effective maintenance policies
45
Stage 1
The initiation of the selection process could take place periodically or whenever there is a need.
The first stage depends on gathering relevant data and performing technical analyses to assess
the current state of the component, equipment or plant. In this stage it is necessary to identify the
production characteristics, the most critical production machine or line, damage causes, failure
modes and consequences, the maintenance concept used, and the current estimate of the overall
equipment effectiveness (OEE), which includes availability, performance efficiency and quality
rate. In addition, all the technically applicable maintenance policies should be identified.
Stage 2
In the second stage, when considering significant components or equipment (e.g. rolling element
bearings in a machine or sub-system), the GTTT-plot method can be used, if historical failure data or
on condition replacement data or both are available for all the applicable maintenance policies Thus
the decision-maker can get a list of the applicable maintenance policies ranked according to the
longterm cost per unit time. When considering a new plant, machine, component and maintenance
policy or when there are no historical (failure or condition) data, the selection can be made using the
fuzzy MCDM method. That could end up with a list of the maintenance policies ranked according to
their ability to provide and utilise more information.
Stage 3
The selection of the most cost effective maintenance policy should be based on a method that
enables the user to build a holistic view of the production process through considering all relevant
aspects. Therefore, the ranking of the applicable maintenance policies is based on the following:
Then the benefit/cost ratio of each maintenance policy should be calculated. After that, the
applicable maintenance policies could be ranked with respect to the value of the benefit/cost ratio.
Consequently, the decision maker can select which maintenance policy is the most cost-effective
based on the highest benefit cost ratio and the available budget.
Stage 4
The previous procedures illustrated in stage 3 should be repeated for all machines or equipment in
order to select the most cost-effective maintenance policy for the whole machine, production line or
plant. When we are ready with the plan phase in the PDCA-cycle, we continue with the remaining
phases, i.e. DoCheck-Act, which are represented by the following three stages:
46
Stage 5 (Do)
Based on the results obtained by the planning phase, the working group is given the task of
implementing the most cost effective maintenance policy(ies). It is of great importance to make
everyone involved fully aware of the requirements of the implementation process such as the
technical and organisational aspects.
Stage 6 (Check)
When appropriate steps have been taken, the results of the implementation process should be
investigated to control its cost effectiveness. This could be achieved by measuring and evaluating
the results with respect to the expected performance.
Stage 7 (Act)
If the results obtained in stage 6 were not satisfactory, we have to go through the cycle once more.
However, if the results were successful, the implemented maintenance policy(ies) should be adopted
for longer application. It is very important to repeat the whole process periodically, i.e. every three to
five years or when there is a need, due to the fact that market and technical changes occur.
Complex manufacturing systems are becoming more sensitive to disturbances, and cannot
tolerate expensive and unpredictable behaviour. Therefore the availability and reliability of
manufacturing systems are vital, and the importance of using effective decision-making systems is
increasing. In order to ensure the optimum performance of automated manufacturing systems,
various conditionmonitoring techniques are used.
When dealing with vibration-based maintenance (VBM), the condition of significant parts (e.g.
rolling element bearings) cannot be assessed effectively,
i.e. with high certainty, without considering both probabilistic and deterministic aspects of the
deterioration process. Modelling the time for maintenance action and predicting the value of the
vibration level when damage of a significant component is detected are examples of the
probabilistic part, which is discussed in Paper IV. However, issues related to machine function,
failure analysis and diagnostics are examples of the deterministic part that is discussed in Paper V.
47
4.4.1 Mechanistic model for predicting the vibration level
When a potential failure (damage under development) of a significant component is detected,
predicting the value of the CM parameter, e.g. vibration level, during the interval until the next
measurement or planned stoppage, accurately, would enhance the effectiveness of
decision-making process. A model for predicting the CM parameter, e.g. vibration level, during the
next period and until the next measuring moment was developed by Al-Najjar (2001). In Paper IV
we reformulated the model as illustrated by equation 4.1. Let Y be the dependent variable
representing the predicted value of the vibration level. It is the function of three independent
variables (X, Z & T) and three parameters (a, b and c). For i=1, 2…n, and i the number of
measuring opportunities after damage initiation.
exp(
i)
i 1 i
Z T b a X Y1 ciii
E i
(4.1)
Where
Yi : 1The predicted value of the vibration level at the next planned measur-
ing time.
T i : The
1
elapsed time since the damage is initiated and its development is
detected.
To test and verify the model two tests were performed. The first test was based on conducting
an experiment at the department’s CM laboratory and the second test was based on real data
collected from a Swedish paper mill. In both cases the model was able to predict the value of the
vibration level accurately.
Figure 4.11 shows the results achieved when testing the model using laboratorial experiment
data. As we can see, the model can predict the value of the vibration level with an absolute error
that ranges from –1.14 to 0.77 mm/sec with an error mean value of 0.15 and a standard deviation
of 0.51. The model is capable of adapting to the changes that are taking place within the
machine’s operating condition. It can be seen that the model learns from the experience
encountered in the recent performance, i.e. the predicted value follows the trend and latest
development in the recent vibration level actual value. For example, in measuring opportunity
number eight the actual value was smaller than the
48
predicted value (which could happen in real life applications), and therefore, for the next point
(measuring opportunity number nine) the model was adapted and predicted a value that goes with
the actual changes experienced recently, i.e. with a decreased rate. Then, again since the actual
reading at measuring opportunity number nine was higher than the predicted value, the model
adapted again and predicted a value with an increasing rate at point 10, and so forth. Furthermore,
by eyeballing the trend for the actual and predicted points we can see that both sets of points can be
fitted by more or less the same curve. We can see that the error, i.e. the dispersion between the
predicted value of the vibration level and its actual value, is decreasing with time.
Y a c tu a l Y p r e d
ic te d E r r o r
CM Parameter Value 1X RPM mm/sec
11
10 X P = 4 .0 m m /s e c
89
X o= 3 .1 4 m m /s e c
67
45
23
01
-1
0 1 2 3 4 5 6 7 8 9101112131415161718
M e a s u r in g o p p o r tu n i ty in H o u r s , D a y s , W e e k s o r M o n th s
Fig. 4.11. The actual CM value, the predicted CM value and the error obtained when predicting the CM
parameter value
To verify the model in a real life application, real vibration data collected about a spherical
roller bearing installed at the lead roller of a paper mill machine were used. It was found that the
model could predict the value of the CM parameter with an absolute error that ranges from –0.47
to 0.04 mm/sec with an error mean value of –0.18 and a standard deviation of 0.26. For more
details see Paper IV
To improve the accuracy of decision making, more attention, in the literature, is paid to integrating
and interpreting richer information from the fusion of multiple sensor data, reducing noise and
obtaining the most reliable extracted features. But less attention was devoted to the integrated CM
system, which enables the user to evaluate a multi-variant system based on the data collected from
different
49
sources such as CM transducers, maintenance, quality, production, accountancy, machine tools and
process monitoring. On the other hand, although expert systems help analysts dealing with
information and decision-making by accepting data from multiple sources, and then using a model
and rule-based programming to diagnose a problem, we believe that there exist additional elements,
which should be considered to increase the expert system’s effectiveness.
The manufacturing system that consists of the machines, tools, materials, product, quality
control system, manufacturing methods and management, can be monitored and controlled by
implementing the most cost effective maintenance policy. In Paper V, a new approach to an expert
system concept is suggested. It benefits from the advantages of a common database, artificial
neural network and fuzzy logic, and applies an expert system method in a more effective and
reliable way. Using a common database ensures that the required data for particular purposes are
available and continuously updated. The user interface is added to the expert system to provide a
user-machine-expert system link, which in addition to the common database enables the expert
system to be more flexible and continuously improved based on the new experience gained by the
user or others. At the same time, artificial neural network, fuzzy logic and fuzzy neural network
algorithms can be utilised in the inference engine part to enhance the certainty of data and remove
any ambiguity or noise from the signal. Also, using fuzzy neural networks enables the controller to
take proper action at the suitable time, due to their ability to reduce the time needed for learning.
Finally, the beneficial integrated information available in the common database, which provides
more accurate and comprehensive perception of the condition of the manufacturing system,
production, product quality and costs results in more accurate diagnosis/prognosis and decisions.
For more details see Paper V.
50
4.5 Maintenance contribution to business strategic
objectives
In the following we present and discuss the results of solving the fifth research question.
How to assess the contribution of maintenance function to the business strategic objectives?
We demonstrate some of the results of a case study that was conducted at StoraEnso Hylte
AB (a paper company in Hyltebruk in southern Sweden). The complete results can be found in
Paper VI, Paper VII and Paper VIII in Appendix (C).
51
Identification of the model's cost factors
E) Maintenance profits
tools, e.g.
F) Maintenance measures
Pareto diagram
Maintenance
Recoverable Changes
expenses
investment
52
Case study results
The case study was conducted at a Swedish paper mill company. The data collected were
delimited to only stoppages of mechanical components that were monitored by vibration signals.
The study was conducted at PM2, one of the company’s four machines. It was selected due to its
valuable database, especially during the period studied (1997-2000). A special data sheet was
designed for collecting manually the relevant technical and economic information parameters from
the company databases. We think that one of the reasons behind the unavailability of certain data
is that these data had not been needed before for analysis. The other is that these data were
either hidden or confused with other data.
Model validation
The conceptual model shown in Figure 4.1 2 was validated using the data collected from the case
company. As the economic data were confidential, the data used in the analysis were transformed
by several suitable factors, which still allowed accurate analysis. The total maintenance investment
in PM2 both in general and training, on average, was about 0.455 MSEK (about 45.5 thousand
USD) per year. The total economic losses (potential savings) consist of the summation of profit
losses and the costs of unutilised resources, e.g. the fixed cost of an idle machine, during the time
in which the machine is not producing due to failures, UPBFR, planned stoppages, and short
stoppages. The economic losses due to bad quality products caused by maintenance deficiency
and tied up capital due to extra spare parts inventory are considered as part of the potential
savings, too. On average the total potential saving was about 30 MSEK (about three million USD).
The maintenance department has been implementing VBM for several years. Therefore, their
long experience and competence in VBM enabled them to achieve high technical efficiency and
precision. For example, the average number of failures was only one failure per year with an
average time of about
1.6 hours. Furthermore, the average number of UPBFR was about 3.25 per year, with an average
time of about 4.07 hours per stop. Moreover, they managed to integrate the VBM with the
production schedule. According to the production schedule, the paper-mill machine is stopped
every other week on average for about eight hours on the basis of technical production reasons.
Therefore, the maintenance department
planned and performed based on VBM
recommendations, throughout the time window initiated by the production department, on
average about 12 replacements of rolling element bearings per year. Consequently, the minimum
saving achieved due to performing these maintenance tasks was estimated to be on average
about four MSEK (about 0.4 million USD), see Figure 4.13.
53
35
30
25
Million SEK
20
15 30,07
10
0,45 4,04
0
Figure 4.13. Estimated yearly average total investment, potential savings, and maintenance savings for one
paper mill machine
We assume that Q1 is the quantity of quality product produced when using a certain
maintenance policy, which resulted in total manufacturing cost TC1, see position 1 in Figure 4.14.
If the company improves the implemented maintenance policy, or uses a better maintenance
policy that requires a new investment of (I), this could result in increasing the quantity produced to
Q2. Assuming that no other actions were performed,
then, the new total
manufacturing cost (following the same curve) has a value of TC2. Consequently the impact on the
company profitability can be estimated using the equations 4.2 to 4.4 as:
54
1 2
Currency/unit
TC1
TC2 Quantity
Q1 Q2
Figure 4.14 Detailed relation between quantity of quality product produced and total cost per unit
If the net profit achieved is greater than the cost of improvement, i.e. I, required for achieving
the increase in output, then the investment is cost effective. The relation between the costs of
investment needed to improve maintenance effectiveness, e.g. reduced breakdowns, and the
expected savings is not easy to predict. As far as the absolute level of maintenance improvement
is concerned, there is a finite limit to the impact that maintenance improvements can have upon
the generated savings. In other words, after some point diminishing returns will start. Beyond this
point we are in a region where additional investment on improving maintenance does not give
payback.
55
It was found that 5.8% of the planned working time, the machine was stopped for several reasons
such as failures and UPBFR, planned stoppages and short stoppages. The total stoppage time was
distributed as short stoppages, 48%, unplanned stoppages, 34%, and regular planned stoppages,
18%, see Figure 4.15.
60
50
40
Percent
30
48
20
34
10 18
0
Short Stoppages Unplaned Planned Stoppages
Stoppages
The short stoppage constitutes the largest portion of the stoppage time; on average it amounted
to about 1624 short stoppages. Regular planned stoppages (about eight hours each) were planned,
in general, every other week to perform certain tasks ordered by the production department. This
planned stoppage time creates a great opportunity that could be utilised to perform pre-planned
maintenance tasks if the right maintenance policy is implemented. The causes of the unplanned
stoppages and the percentage of each with respect to the total unplanned stoppages are illustrated
in Figure 4.16.
50
40
Percent
30
45
20
10
10 10 11 11 13
0
s
ng
er
ca
al
t-u
ge
ni
th
tri
ni
ar
an
a
O
ec
le
ha
St
ch
C
El
ec
ng
M
ni
Li
56
In this study, it was found that the actual annual quantity produced during the case study period,
i.e. the years 1997-2000, was Q1= 176 963 tons. The average selling price was about 3000 SEK per
ton. The average total production cost, i.e. TC1, at Q1 was about 1949 SEK per ton, and the fixed
cost per ton was about 724 SEK. The average quantity of production lost because of unavailability
due to all types of unplanned stoppages such as mechanical, electrical and hydraulic, was 3775
tons. The average quantity of bad quality production lost due to causes related to maintenance
problems was about 432 tons. This figure was estimated according to the company personnel’s
experience at about 7.5% of the total bad quality production. This means that if, ideally, an effective
maintenance policy is used which can get rid of all the unplanned stoppages and all the bad quality
production related to maintenance problems, the new output quantity, i.e. Q2, could amount to
181170 tons. Thus, the new fixed cost per ton would be about
707. Thus total production costs, i.e. TC2, would be about 1932 SEK per ton. Based on the
above-mentioned data, the productivity index calculated at point Q1 is about 1.539. Also, the value
of productivity index could be improved to about 1.553 at Q2. This means that the productivity
index of one paper machine,
i.e. PM2, could be improved by an increment of about 0.014 if a better maintenance policy is
used. Thus, the ideal net impact on the company profit without the cost of investment can be
calculated using equations 4.2 to 4.4 as:
This means that in this case, ideally, at least 7.8 million SEK (approximately US$ 0.975
million) per year could be gained as a result of the productivity improvement of one paper
machine, if a better maintenance policy had been used. This value will increase according to how
the maintenance actions are linked to the causes of other elements of the overall equipment
effectiveness, i.e. short stoppages and planned stoppages. Practically, we cannot avoid all
unplanned stoppages. To assess the cost effectiveness of a new investment in maintenance, the
savings increment due to the output achieved by improving maintenance could be compared to
the investment needed.
This represents the economic effect of maintenance due to its impact on the profit margin value
only. In addition, there are other factors such as the lost expenses due to not utilising the fixed cost
elements such as idle labour or idle machine. On the other hand, there is a possibility in the long
run to decrease the total manufacturing costs (when using a more efficient maintenance policy) due
to the maintenance effect of elements such as less tied capital in raw materials, work in progress
(WIP), and finished goods. Furthermore, the price could be improved by providing value
advantages to the end customer such as on time and consistent delivery.
57
4.5.3 A strategic approach to measure maintenance performance
When the BSC model was tested, i.e. the case study discussed in the previous section, it was
found that the calculated Overall Equipment Effectiveness for one paper machine is about 91%.
However, when considering the major planned stoppages and yearly vacations, a new measure
called Total Overall Equipment Effectiveness (TOEE) was suggested and calculated to be about
85%. Figure
4.17 illustrates the causes of unproductive time that resulted in the TOEE value. They are
distributed among 44% planned inoperative time that consists of major planned stoppages and
yearly vacations; 19% bad quality representing all unaccepted production; 19% unavailability
consisting of the time lost due to unplanned stoppages and 18% performance inefficiency
representing short stoppages.
50
40
Percent
30
44
20
10 19 19 18
0
inoperative
Bad quality
Performance
Planned
Unavailability
products
inefficiency
time
The economic consequence, i.e. profit losses, of the measured TOEE has the potential to increase
the Return On Investment (ROI) by an absolute value of about 1.47, i.e. about a 9% increase in the
ROI value. The potential increment in ROI is equivalent to about 67 million SEK.
58
Finally, the cause-effect relationship for the suggested BSC model when used to measure
maintenance performance at one paper machine is illustrated in Figure
4.18.
Lost Profit due to Unutilised Resources expenses due to Potential to increase Potential to improve
TOPE=66.93 MSEK TOPE=33.81 MSEK product prices Stake-Value
Planning Performance Quality On Time Zero injury due Extra Capital Tiedup in
Index Availability Efficiency Rate to main. Inven. (w.r 97)
Delivery
93.7% 97.1% accidents. 0.41 MSEK
97% 97%
Man. Pl. Stoppages Failures/UPBFR Short Bad quality Zero Potential avail.
116 hrs Maj. Stop. 160 hrs 83 hrs Stoppages 5492 Tons accidents information about
435 hrs Vacations Pl.Stop. 231 hrs 7.5% Maint. spare parts needs
Good relation with Good co-operation & Implementing New Invest in maint. High competence 50%
machines contacts with Research maintenance Methods 0.46 millions SEK. of Invest. in
manufacturer Centres & Universities & Tech. maintenance
Figure 4.18 Impact of maintenance on company performance as measured by the balance scorecard
59
5. Results, Conclusions and
Implications
This chapter shows the main research results, the conclusions and how this research makes a
distinct contribution to the body of knowledge. As was illustrated in Chapter One, the main
research problem is how to select and improve the most cost-effective maintenance policy and
how to assess its financial impact. The overall objective of the thesis is to study the impact of
maintenance practices on companies’ performance outcome. The objective was investigated
based on five research questions related to eight research papers.
The main conclusion is that although preventive and predictive maintenance approaches are
emphasised in Swedish industry, there is much room for improvements to be made. For example,
about one third of the planned tasks is initiated based on the OEM recommendations only and this
may not be the best policy to use because the context (e.g. the working environment, speed, load,
etc) may be not considered. Furthermore, another third of the planned tasks is based on the use of
various CM techniques, in which visual inspection is the most emphasised but may not be the most
cost effective one. On the other hand, the majority, i.e. about 70% of the respondents, still consider
maintenance as a cost centre.
60
5.1.2 Maintenance selection
Based on the data collected from a survey conducted in Swedish industry related to the second
research question (i.e. How are maintenance policies selected in Swedish industry?) and the
empirical results and analyses illustrated in Paper III, the following findings were arrived at:
Using an MCDM method to rank order the implemented maintenance selection methods. Hence,
it was found that FMECA is the most satisfactory method, followed by modelling the time to
failure and optimisation method, after which comes the companies’ own knowledge and
experience method and finally the MCDM maintenance selection method.
From the above results we conclude that none of the implemented maintenance selection
methods is adequate and capable to satisfy all the firm’s needs, i.e. the ideal maintenance
selection method features. This means that there is a need in industry for a maintenance
selection method that integrates the strengths of the available maintenance selection methods
and avoids their weaknesses.
61
5.1.3 Selecting the most cost effective maintenance policy
The main results achieved from the third research question (i.e. How to select the most cost effective
maintenance policy?), which is related to Paper I, Paper II and Paper III, can be summarised as follows:
In Paper I a fuzzy MCDM methodology for selecting the most informative maintenance
approach was developed and illustrated using two examples based on typical data.
In Paper II a model for selecting and improving the most cost effective maintenance policy
was developed. It is based on the continuous
improvement method, i.e. Plan, Do, Check and Act. It is characterised among other things by
being general, since it can be used for selecting from all types of applicable maintenance
policies and also for selecting a maintenance policy for both new and existing plants or
machines regardless of the availability of historical failure data.
In Paper III it was found that there is a need in industry for a maintenance selection method that
integrates the strengths of the available maintenance selection methods and avoids their
weaknesses.
This means that the model suggested in Paper II for selecting and improving the most cost
effective maintenance policy will, with a high probability, satisfy the requirements of ideal
maintenance selection method features, since it is based on the integration of the four methods
used by industry. This is because it integrates the strength of the maintenance selection methods
used in industry and avoids their weaknesses. Therefore, it could be considered very strong in
fulfilling all the ideal maintenance selection feature requirements as discussed in Paper II.
We can conclude that using the practical model developed and suggested in Paper II, it would
be possible to select the most cost effective maintenance policy(ies) and improve it (them)
continually. This will enhance the overall equipment effectiveness (OEE) of the machine and
process, which could be achieved, for instance, through the maintenance impact on stoppages
and on poor quality caused by maintenance ineffectiveness. Improving the OEE cost effectively
will improve the productivity and profitability of the company, i.e. its competitive advantage.
In Paper IV a mechanistic model for predicting the vibration level in the near future, e.g. at the
next planned measuring time, was verified at a laboratory experiment in addition to using data
collected in a case study.
62
In Paper V, a new approach to an expert system was suggested. It is based on the use of a
common database, an artificial neural network (ANN) and fuzzy logic in the inference engine and
user interface.
The following conclusions can be drawn from the results achieved in Paper IV and Paper V:
2) In complex systems, signals from only one condition monitoring parameter are
insufficient and cannot provide accurate prediction of failure and accurate
diagnostics. Thus, the integration and
management of data obtained from maintenance, operations, production, quality
control, surroundings and accountancy are important to achieve an efficient and
effective plant-monitoring, diagnosis/prognosis and decision-making system.
3) The main conclusion is that better data coverage and quality will reduce the
uncertainty and improve the effectiveness of CBM decision-making. This would
reduce the number and duration of planned and unplanned stoppages, which has a
direct impact on the company’s productivity, and hence, on its competitiveness.
In Paper VI a model to assess the economic impact of maintenance was developed and
validated in a case study. Using the model one can identify, reclassify and assess
maintenance-related costs, so that the user can reveal maintenance benefits, highlight
profits, analyse the situation, identify the problem area, and help the decision-maker to
perform the continuous improvement process, i.e. KAIZEN. When the model was validated in
a case study, it was found that implementing VBM in a paper machine achieved yearly
maintenance savings and profit, on average, of at least about 4.0 and
In Paper VII, a conceptual model that illustrates the impact of maintenance on a firm’s
profitability was developed and tested in a case study. It was found that, ideally, at least 7.8
million SEK per year could be gained as a result of the productivity improvement of one paper
machine, if an effective maintenance policy is used which can get rid of all the unplanned
stoppages
63
and the bad quality products related to maintenance problems. This figure could be increased in
accordance with how maintenance actions affect other factors such as short and planned
stoppages.
In Paper VIII, a modified balanced scorecard suitable for measuring the performance of
production support functions such as maintenance, quality and logistics was developed. The
model widens the perspectives of BSC to cover the extended enterprise such as the suppliers
and machine designers. The model was validated in a case study.
The main conclusion can be stated like this: effective maintenance affects the production systems’
ability to provide quality products and timely services to customers through its direct impact on
factors such as number and duration of stoppages, quality, environmental and employee safety.
The better the data coverage and quality, the greater the ability to detect deviations in
maintenance performance and the higher the possibility of identifying problem areas at an early
stage. Thus, assessing maintenance costs, savings and investments needed to eliminate the
causes of deviations, enables the decision-maker to make the right decision maximising the
contribution to the performance and profitability of manufacturing companies.
Studying which maintenance practices are deployed in Swedish industries, and showing,
empirically, that there is a need for improving the implemented maintenance practices. In
particular it was shown that the most used maintenance selection method is not satisfactory
to its users. In addition it was found that none of the applied maintenance selection methods
fulfils the requirements of the ideal maintenance selection method features.
Developing an evaluation methodology based on fuzzy MCDM to select the most informative
maintenance approach.
Developing a practical model for selecting and improving the most cost effective maintenance
policy. The model is based on a literature review and analyses of the available maintenance
selection methods and is supported by the results of a survey conducted about Swedish
industry.
Verifying a mechanistic model for predicting the vibration level at the next planned measuring
time.
64
Suggesting a new approach to an expert system for improving the effectiveness of plant
monitoring and its diagnostic/prognostic decisionmaking systems.
Suggesting and testing a model that illustrates the impact of maintenance on the profitability of
firms.
Modifying the balanced scorecard (BSC) to suit the measuring of the performance of a
production support function, i.e. the maintenance function, showing its effect on the strategic
objectives of the manufacturing company.
Proving that maintenance could improve the quality, efficiency and effectiveness of
production systems and consequently enhance the competitiveness and profitability of
manufacturing companies. Thus, it was proved that maintenance is no longer a cost centre;
but a profit generating function.
Finally, we can say that just like every other function in production systems, maintenance has
a role in gaining and maintaining competitive advantages. However, the weight of this role varies
according to the type of production system. Nevertheless, this role is not easily seen due to the
lack of linkage between maintenance practices and performance, e.g. the profitability, of
manufacturing companies. In summary, the contribution of the thesis can be developing tools that
improve the efficiency and effectiveness of maintenance practices, e.g. selecting and improving
the most cost effective maintenance policy, and assessing the financial impact of maintenance
practices on business strategic objectives.
The models developed and suggested to assess the impact of maintenance practices on
business strategic objectives (discussed in Paper VI, Paper VII
65
and Paper VIII) can be used to assess the impact of other support functions employed in
production systems such as quality, logistics.
Proving that maintenance is not a cost centre could affect the view of the maintenance staff
whose function is usually given a low status in manufacturing companies. This may have
positive psychological effects.
66
The conceptual model for an expert system suggested in Paper V could be developed further
through a prototype test in lab and validated in a case study.
The low response rate achieved in the survey is considered the main serious problem affecting
the ability to generalise the results to a larger population. However, since this survey is part of
a current project, efforts are being made now to call the surveyed firms and encourage them
to respond to the questionnaire.
Because the research work was performed during a long period and involved two main research
projects and other conceptual research work performed at different times, it was not possible to
present the research papers and research questions in a chronological order. Therefore, the
research papers were numbered according to their relevance to the research questions.
67
References
Ben-Daya, M. and Duffuaa, S. (1995). Maintenance and Quality: The Missing Link.
Journal of Quality In Maintenance Engineering 1:1, 20-26.
Bengtsson, L., Elg, U. and Lind, J. (1997), Bridging The Transatlantic Publication Gap:
How North American Reviewers Evaluate European Idiographic Research, Scandinavian Journal
Management 13:4, 473-492.
68
Bevilacqua M., Braglia M. (2000). The Analytical Hierarchy Process Applied To
Maintenance Strategy Selection. Reliability Engineering and System Safety 70, 71-83. Blanchard,
Benjamin S. (1986), Logistics Engineering and Management, third edition,
Prentice-Hall Inc., New Jersey.
Blanchard, Benjamin S. (1997). An enhanced approach for implementing total productive
maintenance in the manufacturing environment. Journal of Quality in Maintenance Engineering 3:2,69-
80.
Bloch, H.P. and Geitner, F.K. (1994). Machinery failure, analysis and troubleshooting.
Gulf publishing company, London.
British standard glossary of terms used in terotechnology , BS 3811:1993. Burrel,Gibson and Morgan, Gareth
(1979). Sociological Paradigms and Organisational
Analysis: Elements of the Sociology of Corporate Life. Heinemann, London. Campbell JD, Jardine AKS
(2001). Maintenance excellence: Optimising equipment life
cycle decisions. New York: Marcel Dekker.
Carter, Russell A. (2001). Shovel Maintenance Gains from Improved Designs, Tools and
Techniques, August 106:8, S7. Elsevier Engineering Information. Charlene, B. (1989). The JIT
Challenges For Maintenance. The Magazine Of
Manufacturing Performance, September-issue, 34-35.
Christopher, Martin (1998). Logistics and Supply Chain Management: Strategies for
Reducing Cost and Improving Service. Second-Edition, Financial Times Prentice Hall. Coetzee, Jasper
(1999). A holistic approach to the maintenance “problem”. Journal of
quality in maintenance engineering 5:3, 276-280.
Collacott, R.A. (1977). Mechanical fault diagnosis and condition monitoring. Chapman
and Hall, London.
Cooke, Roger and Paulsen, Jette (1997). Concepts for measuring maintenance
performance and methods for analysing competing failure modes. Reliability engineering and system
safety 55, 135-141.
Cooke, Fang Lee. (2000). Implementing TPM in plant maintenance: some organisational
barriers. International Journal of Quality & Reliability Management 17:9, 1003-1016. Davies, A. (1998).
Handbook of Condition Monitoring, Techniques and Methodology,
Chapman & Hall.
De Wet Bob and Mayer Ron (1998), “Strategy Process, Content, Context: an international
perspective”.
Deardeen, J. Lilien, G. and Yoon, E. (1999). Marketing and Production Capacity Strategy
For Non-Differentiated Products: Winning And Losing At The Capacity Cycle Game. International Journal
Of Research In Marketing 16:1, 57-74.
Dekker, Rommert (1996). Applications Of Maintenance Optimisation Models: A Review
And Analysis, Reliability Engineering And System Safety 51,229-240. Dekker Rommert and Scarf Philip
(1998). On the impact of optimisation models in
maintenance decision-making: the state of the art, Reliability Engineering & System Safety 60(2), pp.
111-119
Dwight, R. (1995). Concepts For Measuring Maintenance Performance, In Martin , H. H.
(Ed), New Development In Maintenance: An International View, IFRIM, Eindhoven, 109-125.
69
Emory William and Cooper Donald (1991). Business research methods, fourth edition,
IRWIN.
Foelkel, Celso (1998). A business-oriented approach to maintenance, Tappi Journal,
September 1998.
Gallimore, K. and Penlesky, R. (1988). A Framework For Developing Maintenance
Strategies, Production And Inventory Management Journal, First Quarter, 16-22. Gits, C. W. (1992). Design
Of Maintenance Concepts. International Journal Of Production
Economics 24, 217-226.
Gits, C. W. (1994). Structuring Maintenance Control Systems. International Journal of
Operations and Production Management 14:7, 5-17.
Goldman, L., Nagel, R. and Preiss, K. (1995). Agile Competitors and Virtual
Organisations – Strategies for Enriching the Customers, Van Nostrand Reinhold, New York, USA.
Gomm, R., Hammersley, M., Foster, P. (2000). Case Study Methods. SAGE Publications,
London.
Graziano, A. and Raulin, M. (2000). Research methods a process of inquiry, fourth
edition, Allyn and Bacon, USA.
Hair Joseph, Anderson Rolph, Tatham Ronald Black William (1998), Multivariate data analysis, fifth edition,
USA, Prentice-Hall International, Inc.
Henning, Kirsten (1989). Quality Improvement through Preventive Maintenance. ASQC
Quality Congress Transactions-Toronto, 408-415.
Hill, Terry (2000). Operations Management: Strategic Context and Managerial Analysis.
MACMILLAN Business, UK.
Holmberg (2001). Competitive Reliability 1996-2000, Technology Programme Report
5/2001, Final Report Edited By Kenneth Holmberg, National Technology Agency, Helsinki 2001.
70
Liptrot D. And Palarchio G. (2000), Utilising Advanced Maintenance Practices And
Information Technology To Achieve Maximum Equipment Reliability, International Journal Of Quality&
Reliability Management 17:8, 919-928.
Ljungberg, Orjan (1998). Measurement of Overall Equipment Effectiveness as A Basis
for TPM Activities, International Journal of Operations & Production Management 18:5, 495-507.
Mitchell Mark and Jolley Janina (2001). Research Design Explained, fourth edition,
Harcourt College Publishers.
Mcgrath, Robert (1999). Maintenance The Missing Link In Supply Chain Strategy,
Industrial Management, Norcross; Jul/Aug 1999.
Mike, Organ, Tony, Whitehead, and Mike, Evans (1997). Availability-Based Maintenance
Within An Asset Management Program. Journal Of Quality In Maintenance Engineering 3:4, 221-232.
Miller, D. (2000). Profit from preventive maintenance. Bulk solids handling 20:1, 57-61.
Elsevier Engineering Information.
Mitchell Ed., Robson Andrew Prabhu Vas B. (2002), The Impact Of Maintenance
Practices On Operational And Business Performance, Managerial Auditing Journal 17:5, 234-240.
Mckone, Kathleen and Wiess, Elliott (1998). TPM: Planned and Autonomous
Maintenance: Bridging The Gap Between Practice and Research. Production and Operations
Management 7:4, 335-351.
Montgomery Douglas C. (2001). Design and Analysis of Experiments, fifth edition, John
Wiley and sons, inc.
Moubray, John (1991). Reliability Centred Maintenance, Butterworth Heinemann,
Oxford, UK.
Muthu, S., Devadasan, S.R., Ahmed, Saleem, Suresh, P., and Baladhandayutham R.
(2000). Benchmarking for strategic maintenance quality improvement. Benchmarking 7:4, 292-303.
71
Nakajima, S. (1988). Introduction to Total Productive Maintenance, TPM. Productivity
Press, Cambridge.
Neely, A., Gregory, M. and Platts, K. (1995). “Performance Measurement System Design.
International Journal Of Operations And Production Management 15:4, 80-116. Oakland, John (1995).
Total Quality Management. Butterworth Heinemann, UK. O’Connor, Patrick D.T., (2002), Practical Reliability
Engineering, 4 ed., Wiley
Chichester, UK.
Ollila, A. and Malmipuro, M. (1999). Maintenance Has A Role. The TQM Magazine
11:1, 17-21.
Patton, Michael Quinn (2002). Qualitative Research and Evaluation Methods, Third
Edition, Sage Publications, London.
Pehrsson Anders (2000), Strategy Competence: A Key Profitability Driver, Strategic
Change 9, 89-102.
Pintelon, L. M. and Gelders, L. F. (1992). Maintenance Management Decision Making.
European Journal Of Operational Research 58, 301-317.
Pintelon, L. and Van, Puyvelde F. (1997). Maintenance performance reporting systems:
some experiences. Journal of quality in maintenance engineering 3:1, 4-15. Pintelon, L. Nagarur, N. and
Puyvelde, F. (1999). Case Study: RCM-Yes, No, Or
Maybe?”, Journal Of Quality In Maintenance Engineering 5:3, 182-191. Porter Michael (1985).
Competitive Advantage: Creating and Sustaining Superior
Performance. Free Press, New York.
Ralph, W Peter (2000). Maximising Value from Maintenance Operations. Material
Handling Management, Cleveland, June.
Rao, B.K.N. (1993). Profitable Condition Monitoring and Diagnostic Management.
Profitable Condition Monitoring, Kulwer Academic, London, pp.37-44. Rantanen, H. (1995). The Effects of
Productivity on Profitability: A Case Study at Firm
Level Using an Activity-Based Costing Approach. Doctoral Thesis, Lappeenranta University of
Technology, Finland.
Riis, Jens, Luxhoj, James and Thorsteneinsson, Uffe, (1997). A Situational Maintenance
Model. International Journal of Quality and Reliability Management, 14:4, 349-366. Sago Brad, D.B.A.
(2003) Building organisational competencies for competitive
advantages, Business Credit 105 (2), 16-17.
Schalkwyk, Johan C. (1998). Total quality management and the performance
measurement barrier. The TQM Magazine 10:2, 124-131.
Sekaran, U. (2000). Research Methods for Business. John Wiley and Sons, Third Edition,
USA.
Shepard, Colin, and Scicon, EDS (1993). Neural Networks - Theory and Practice,
Profitable Condition Monitoring, Kulwer Academic, London, 205-216. Sherwin, D. J. (2000). A review of
overall models for maintenance management. Journal
of quality in maintenance engineering 6:3, 138-164.
Shonder, John and Hughes, Patrick (1997). Estimated maintenance cost savings from a
geothermal heat pump energy savings performance contract at Fort Polk Louisiana, ASHREA
Transactions 103:2, 757-766. Elsevier Engineering Information. Sink, Scott D. and Tuttle, Thomas C.
(1989). Planning and Measurement in Your
Organisation of the Future, Norcross, Industrial Engineering and Management Press. Sumanth, David
(1998). Total productivity management- a systematic and quantitative
approach to compete in quality, price, and time. CRC Press LLC, USA. Suito, Kiyoshi (1998). Total
Productivity Management, Work Study 47:4, 117-127. Swanson, Laura (2001). Linking maintenance strategies
to performance. International
Journal Of Production Economics 70, 237-244.
72
Swanson, Laura (2003). An information-processing model of maintenance management,
International Journal Of Production Economics 83(1), 45-64.
Stephen T. Desirey (2000). Positioning maintenance as a competitive advantage, Plant
Engineering, 54(5), 66-69.
Taguchi, G., Elsayed, E., and Hsiang, T. (1989). Quality Engineering in Production
Systems, McGraw-Hill, New York.
Tsang, A. (1998). A Strategic Approach To Managing Maintenance Performance. Journal
of Quality in Maintenance Engineering 4:2, 87-94.
Tsang, Albert H.C., Jardine, Andrew K.S. and Kolodny, Harvey (1999). Measuring
maintenance performance: a holistic approach. International Journal of Operations and Production
Management 19:7, 691-715.
Tse peter w., (2002) maintenance practices in honk kong and the use of the intelligent
scheduler, journal of quality in maintenance engineering 8:4, 369-380. Vineyard M., Amoako-Gyampah K.,
Meredith J., (2000). An evaluation of maintenance
policies for flexible manufacturing systems: a case study. International journal of operations and
production management 20(4), 409-426.
Walsh, Kimberly (1999). Predictive maintenance profits from sensor diagnostics. InTech,
instrument Society of America 46:6 June, 36-42.
Williams, J., Davies A. and Drake, P. (1994). Condition Based Maintenance and Machine
Diagnostics, Chapman & Hall.
Wilson, Alan (1999). Asset Maintenance Management: A Guide to Developing Strategy
& Improving Performance, First Edition, Conference Communication, Oxford, UK. Yamashina, Hajime
(1995). Japanese Manufacturing Strategy And The Role Of Total
Productive Maintenance. Journal Of Quality In Maintenance Engineering 01:01, 27-
38.
Yin, Robert K. (2002). Case study research: design and methods, third edition, Thousand
Oaks, Sage Publication.
Zhu Guojum, Gelders Ludo And Pintelon Liliane (2002). Object/objective-oriented
maintenance management, journal of quality in maintenance engineering 8(4), 306-
318.
Zeng, Shuo Wei (1997). Discussion on maintenance strategy, policy and corresponding
maintenance systems in manufacturing. Reliability engineering and system safety 55, 151-162.
73
Appendixes
75
Cross functional groups (for instance improvement groups)
76
Appendix (B): Factor analysis results for
maintenance activities
77
Scree Plot
2
Eigenvalue
23 21 19 17 15 13 11 9 7 5 3 1 8
0
25
Component Number
Figure b1 Scree plot used for determining the number of factors extracted from the
maintenance activities
78
Table b2 Rotated Component Matrix a, Communalities, Eigenvalues and Percentage of Variance
Before and After Rotation
Component
1 2 3 4 5 6 7 Communalities
Restoring equipment to - 0.01 0.00 - 0.12 0.00 0.00 0.64 0.15 0.45
operation (acute)
Installing new 0.12 - 0.11 0.00 - 0.01 0.16 0.73 0.12 0.61
equipment
Keeping the level low in - 0.01 0.01 0.00 0.30 - 0.01 0.01 0.74 0.65
Having inventory - 0.18 0.26 0.01 0.15 0.26 0.49 - 0.25 0.49
between machines
(WIP)
Decreasing the repair time 0.01 0.11 0.16 0.00 0.34 0.23 0.68 0.68
Investing in improving the 0.01 0. 33 0.61 - 0.01 0.00 0.00 0.44 0.68
maintenance staff
Use of (CMMS) - 0.01 0.12 0.67 0.01 0.00 - 0.13 0.26 0.55
Analysing equipment failure 0.38 0.29 0.41 - 0.20 0.32 0.11 0.01 0.55
Using failure data 0.26 0.32 0.28 0.00 0.62 0.00 0.18 0.66
Off-line Monitoring 0.19 0.00 0.58 0.41 0.00 0.00 0.00 0.54
On-line Monitoring 0.15 0.18 0.52 0.23 0.28 0.00 - 0.23 0.51
Following (OEM) 0.11 0.00 0.00 0.16 0.80 0.24 0.00 0.73
recommendations
Performing the 0.39 0.00 0.51 0.21 0.23 - 0.01 - 0.12 0.53
maintenance tasks
based on CM
Performing the 0.10 0.37 0.51 0.16 0.16 0.36 - 0.17 0.61
on statistical modelling of
failure data
Helping the purchasing 0.74 - 0.01 0.25 0.01 0.01 0.16 - 0.10 0.66
department in OEM
selection
Performing periodic 0.42 0.13 0.29 - 0.01 0.36 0.01 0.21 0.46
planned replacement
Automatic diagnosis 0.01 0.25 0.25 0.75 0.10 0.01 0.12 0.72
(expert system)
Remote diagnosis 0.15 0.13 0.11 0.82 0.00 0.00 0.14 0.75
79
Use of company wide 0.56 0.13 0.30 0.31 0.11 0.00 0.19 0.56
Cross functional groups (for 0.64 0.33 0.12 0.01 0.01 - 0.26 0.01 0.61
instance improvement
groups)
Helping improve the 0.79 0.33 - 0.01 0.00 0.01 0.00 0.01 0.75
production process
Helping design the 0.62 0.46 - 0.01 0.25 0.17 0.11 - 0.14 0.73
production process
Recording the period and 0.16 0.84 0.22 0.01 0.18 - 0.43 0.16 0.84
frequency of failures
Recording the period and 0.22 0.82 0.23 0.11 0.00 0.00 0.12 0.80
Recording the poor 0.25 0.73 0.01 0.26 0.00 0.12 0.00 0.68
quality rate.
Annual overhaul 0.35 0.26 0.22 0.20 - 0.37 0.48 0.12 0.66
Eigenvalues
rotation
% of variance
rotation
Notes. Extraction method: Principal Component Analysis. Rotation Method: Varimax with Kaiser
Normalisation
a. Rotation converged in 16 iterations.
80
Appendix (C): Research papers
Pages
Paper I [85-100]
Paper II [1-16]
Paper IV [1-15]
Paper V [267-276]
Paper VI [1-15]
81