Está en la página 1de 12

Microe/ectron. Re/iab., Vol. 35, Nos 9-10, pp.

1245-1256, 1995
Pergamon Copyright 1995 Elsevier Science Ltd
Printed in Great Britain
0026-2714/95 $9.50+.00

THE OPTIMUM ADJUSTMENT OF MOTOR PROTECTION RELAYS


IN AN INDUSTRIAL COMPLEX

P. Massee 1 and H. Rijanto 1

Abstract
In this paper a distribution network is considered supplying electricity to a
number of factories belonging to one industrial complex. The costs of
interruption may vary more than one order of magnitude from one factorY
to another. The problem studied in this paper is how to adjust the time
lags of the undervoltage protection relays in each factory in order to
obtain a minimum in the total costs of interruptions of all factories
together. By properly adjusting the time lags of the relays it may be
provoked that factories with low interruption costs are disconnected from
the network rather fast after the occurrence of a short circuit through
which action the chance that a factory with high interruption cost is
switched off is decreased.
1. Introduction
In this paper the distribution network for an industrial complex is
considered which provides electrical power to a large number of factories.
Characteristic for this situation is that the load at each factory rail
consists of a large number of asynchronous motors. Each factory rail is
provided with an undervoltage relay with the aim to protect the motors
connected to it. The problem adressed in the paper is how to adjust the
different undervoltage relays such that the sum of the interruption costs
of the different factories over a chosen period of time is minimum. The
costs per interruption for the various factories are determined by the
temporary market position for the different products and may vary by t w o
orders of magnitude from one factory to another. Thus it is reasonable to
switch off a factory with small interruption costs rather fast after the
occurrence of a short circuit in order to try to prevent a factory with large

Eindhoven University of Technology, Group Electrical Energy


Systems, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
1245
1246 E Massee and H. Rijanto

interruption costs from being switched off. The effect on the distribution
network when a certain factory is switched off, depends on the
behaviour of the motors comprising the considered plant. In the paper
each factory is represented by one motor, the behaviour of which during
the disturbance is described by a simple model. This model takes into
account the loss of kinetic energy of the motor when the voltage is less
than nominal and the fact that the motor requires 6 times the nominal
current in order to reaccelerate after the short circuit has been taken out
of the network.
A computer program has been written to calculate the interruption costs
after choosing the various adjustments of the undervoltage relays which
are used as inputs. The occurrence of short circuits in the network is
described by negative exponential distributions and the duration of the
short circuits depends on the chosen protection strategy for the various
connections and rails. The optimum adjustment of the undervoltage relays
has to be reached through an interactive approach in which previous
results are coupled back. In this way the optimum adjustment of the
motor protection relays can be approached very closely.

2. The distribution network


The electricity distribution network considered in the paper is shown in
figure 1. The configuration in this figure is a simplified version of the
actual situation but contains the essential features of an electricity
network in practice (Fransen,1989). The typical, characteristic that the
loads mainly consist of a number of large motors has also been indicated
in figure 1. It can also be seen that the principle of single reserve has
been used in the supply to the factories. When one of the connections to
a factory has to be switched off due to the presence of a short circuit the
above implies that the electricity supply can be resumed very fast through
the parallel connection. However, this does not imply that the voltage
returns very fast to its nominal value due to the behaviour of the
Motor protection relay adjustment 1241

asynchronous motors as has been sketched in figure 2 (Schreurs,l989).

As long as the short circuit is present in the network the motors

decelerate and loose kinetic energy. After the short has been eliminated

from the network by the protection the motors have to regain the lost

kinetic energy and during this process they draw six times the nominal

current. In this way the fact that the voltage at the factory rail builds up

rather slowly after a sh,ort circuit, as shown in figure 2, can be explained.

More details of the model which is used to describe the motor behaviour

are given below.

factory 1 2 3 4 5 6 7
power(MVA) 2 1 2 1 0.1 5 5

Figure 1: The industrial electricity network

3. Short circuits and the protection

Short circuits can occur anywhere in the distribution network and this has

to be taken into account in the computer program. However, the chance

on the occurrence of a short circuit depends on the involved component.

Since we concentrate on electrical phenomena statistical calculations are

kept simple. For instance the total chance on a short circuit in a

connection is obtained by adding the chances on a short


1248 E Massee and H. Rijanto

VlVn

i ............... J

I I I
o
o 0.5 t (s) 1.0 1.5

Figure 2: The reaction of a network with a motor to a voltage


dip and the used approximation (dotted line).

circuit in the different components comprising that connection. Similarly


the total chance on interruption of a certain factory is obtained by adding
the chances on interruption due to the different short circuits. The
chances on short circuits in the different components have been taken
from the industrial practice (Fransen,1989) and are given in table 1.
Another point to decide is how long a short circuit will be present in the
network which depends on the chosen protection strategy. Here a
simplification is introduced by assuming that the primary protection will
never fail. Since the transformer connections are protected by differential
relays a short circuit in such a connection is normally disconnected after
0.2 sec. Short circuits at rails occur less frequent than short circuits in
connections so that special rail protections are not installed in the
considered industrial application. Thus short circuits at rails have to be
disconnected by overcurrent-time backup protections in the connections.
In order to maintain selectivity the short circuits at the 10 kV rails are
Motor protection relay adjustment 1249

disconnected after 0.5 see and at the 30 kV rails after 0.8 sec. Finally
there is one other type of protection which has to be discussed and that
is the protection of the motors inside a certain factory. For this protection
an undervoltage relay is used which starts its time counter when the
voltage at the factory rails sinks below for instance 85% of the nominal
voltage and disconnects the motor after a preset time. The problem
adressed in this paper is how to adjust the time lags of the motor
protection relays in order to obtain a minimum in the total of the costs of
interruptions of all factories together. In order to avoid unacceptably large
damage to the motors it is assumed that the time delay may not be larger
than one second. It should be mentioned that short circuits at the factory
rails are not taken into account since their contribution to the costs of
interruptions of the factories can not be influenced by the adjustment of
the undervoltage relays.

Table 1 : Chance on the occurrence of a failure (short circuit) in the


most important components comprising the connections in
an electricity network (Fransen, 1989).

Component Chance (per year)

Cable 0.0051
Transformer 0.0114
Circuit breaker 0.0011
Rail 0.01
Relay 0.0071

4. The costs of interruption


The costs of interruption of the various factories have also been taken
from the industrial practice (Fransen,1989) and appear to vary from Dfl
1250 P. Massee and H. Rijanto

7 , 0 0 0 to Dfl 6 7 5 , 0 0 0 per interruption. In the paper only the fixed costs


(the costs that have to be made when the factory is started up again) are
considered and not
the time dependent costs of interruption since the duration of the
interruptions is not determined. Note that the costs depend on the type of
factory and especially on the value of its product. Thus the costs depend
on the market situation and when this changes the costs may change
considerably.

5. The model of the asynchronous motor


As has been discussed above and as shown in figure 2 the voltage at the
factory rail comes back rather slowly after a short circuit has been
disconnected from the network. This is caused by the behaviour of the
motors. Also indicated in figure 2 is the approximation to the actual
voltage curve, which is used in this paper. From this approximation three
different states of a motor can be distinguished namely nominal
operation, "short circuit" operation and "starting" operation. In contrast
to a realistic motor model in which the motor is represented by 5
impedances which are partly connected in series, partly parallel
(Lerch,1991) a very simple model is used with only one impedance
representing the motor. The value of this impedance, however, depends
on the state of operation of the motor. In the model the motor behaviour
is described by the impedance of the motor ZM,.the power factor of the
motor cos q), the voltage at the factory rail V and the power
consumption of the load PL- As load a mechanical tool is considered such
as a pump, a compressor, a stirrer etc. It is assumed that the power
consumption of the load PL is constant. In nominal operation the power
produced by the motor equals the power consumption of the load so that

p, = V2cs(I)- (1)
IZMol
Motor protection relay adjustment 125 I

This equation can be used to calculate the motor impedance in nominal


operation ZM, from the power of the load taking into account that in this
situation cos ~ , = 0.8.
As can be seen from the approximation curve in figure 2 it is assumed in
the model that the parameters describing the motor behaviour change
instantaneously. During the short circuit the voltage at the factory rail will
be very low. It is assumed that the motor will draw the motor starting
current in this situation which equals 6 times the nominal current. This
implies that the "short .circuit" motor impedance ZM,c then equals the
starting impedance ZM,t so that ZM,~ = ZM,t = 1/6 ZM,. In this situation
the power factor will also be very low and is assumed to have the value
cos ~,c = 0.3. Because of the lower values of V, ZM and cos q~ the
power produced by the motor PM = V~ 2 cos q~,~/IZM,] is less than PL SO
that the motor will loose kinetic energy AE. The kinetic energy lost during
the "short circuit", operati.o.n can then be calculated from the formula

AE= V~cosq~= _pL ~ At= [2)


I

where At~c equals the duration of the short circuit.


After the short circuit has been taken out of the network the motor has
to reaccelerate and is therefore still characterized by the starting
impedance (ZM,t = 1/6 ZM,) and by the low value of the power factor (cos
q~,t = 0.3). However, not only the lost kinetic energy has to be replenis-
hed but also the magnetic field inside the motor which has decreased
strongly while the short circuit was present, has to be built up again. It
has been estimated that the energy needed to build up the magnetic field
again equals the lost kinetic energy AE. Thus the time that the motor
needs to reaccelerate during the "startin.q" operation At,~ can be
calculated from the formula

MR 35:9/i0-0
1252 E MasseeandH.Rijanto
}-1
At,, = -2Ae ,
v,=,cos *., - p'
(3)

After the time At, t the motor has completely reaccelerated and the motor
impedance and the power factor will then change instantaneously to their
nominal values. Note that the lost kinetic energy AE is negative (see
formula (2)) so that equation (3) leads to a positive value for At, t when
the power produced by the motor V,t 2 cos q~,t /IZM,tl is larger than the
power consumption of the load PL. When the power produced by the
motor is smaller than PL the motor cannot reaccelerate anymore and the
value from equation (3) is negative. This is interpreted as At,, is infinite
which means in practice that the motor is connected to the network until
the undervoltage protection disconnects it. The advantage of the simple
motor model described above is that it is not necessary to solve
differential equations in order to determine the motor behaviour. This
implies that the transient behaviour of the motor is neglected which is a
reasonable assumption since it will hardly influence the chance on
interruption of the factories in which we are mainly interested. Thus it is
only necessary to calculate the voltages at any moment that changes
take place in the network i.e. when a short circuit occurs or is removed
or when the operating state of one of the motors is changed. Note that
each factory is represented by one motor only which is consistent with
the simple motor model described above. Since the motors are
represented by only one impedance the voltages in the network are easily
calculated by means of the bus impedance matrix approach which has
been described before already (Massee,1990).

6. Results
In the computer program each possible short circuit is considered and at
Motor protection relay adjustment 1253

any moment that changes take place in the network the voltages are
calculated and it is decided if a factory is disconnected by its motor
protection relay. Thus after each possible short circuit has been worked
out, it is known which factories have been switched off. Using the
probability of occurrence of the various short circuits the chance that a
certain factory is switched off due to the considered short circuit is thus
known and adding these chances the total chance on disconnection of
the considered factory is obtained. Thus the result of the calculations can
be given as the expected number of interruptions for the various factories
over a period of for instance 10 years (see figures 3 and 4).

f/
//

/!
f/
//
//
//
/J
I
//
//
/J

J/
i/ /J
2 fJ //
i1 //
N
//
//
//
// //

N
//
lJ
/J
/J
//
lJ
//

N
/J
,,,/

i ,,,j

fJ
~4
Z/
0
1 2 3 4 6 e 7
factory number
O.S 1 1 1 0.6 0,3 0.3

Figure 3: Number of interruptions in the base case and with


"optimally" adjusted time lags of motor protection
relays (hatched)

The estimated cost of interruption per year is easily calculated by


multiplying the total chance that a certain factory is disconnected due to
the occurrence of short circuits by its cost of interruption. For ease of
comparison, however, the figures 3 and 4 do not show the result of one
1254 P. Massee and H. Rijanto

calculation but the difference between the results of two calculations. In


the base case the undervoltage protection relays of all factories are set at
the standard time delay of 1.0 sec. The next step is to adjust the time
delay for factories with small interruption costs to a smaller value and in
this way to try to prevent factories with high interruption costs from
being disconnected. In figure 3 the difference in the expected number of
interruptions between the base case and the case with "optimally"
adjusted motor protection relays is indicated.

//
7"7 //

/ / /J
/ /
/ /
I / / i

/ /

II
/ i
n

ii
ii / /
.-A
ii i/
ii / /
/i I i
ii i i
ii i i
ii /i

",4 ~A

i/
ii ~ .r A

II f/ /I e'iA
li ,1~
ii
Ii /.4
ii /I II elA
1A ~
fl 1A
II /I ~A
ii /I /Jl
ii /I /A iA ~
~1 1J

~ 1A '/'A
/A r'/A
l/ rl~
~ 1A
".6 Ii
F/A
1.1

1 2 3 4 6 6 7
factory number
0.3 I 0.3 1 0.4 1 0.3

Figure 4: Number of interruptions in the base case and with


"optimally" adjusted time lags of motor protection
relays (hatched) when the market situation has
changed.

The economic benefit, however, cannot be judged from figure 3 alone. It


can be mentioned that
the adjustment of the motor protection relays which resulted in figure 3
led to a decrease in yearly interruption costs for all factories together
from Dfl. 185,000 for the base case to Dfl. 9 6 , 0 0 0 for the "optimum"
Motor protection relay adjustment 1255

case. Thus it is clear that a considerable reduction of interruption costs


can be obtained by adjusting the time delays of the various motor

protection relays.
It is clear, however, from figure 3 that the expected number of
interruptions varies much from one factory to the other both in the base
case as in the "optimum" case. The way in which the number of
interruptions is distributed over the factories depends on the layout of the
electricity network as well as on the different powers of the various
factories (see figure 1). This implies that the expected number of
interruptions cannot be adjusted at wish by adjusting the time lags of the
motor protection relays. This is shown in figure 4 for which it has been
assumed that the market situation has changed which led to an exchange
of the interruption costs of the factories 3 and 6. Figure 4 again shows
the difference in the expected number of interruptions between the base
case and the situation with "optimally" adjusted motor protection relays.
A comparison of figure 4 with figure 3 shows that now less reduction in
number of interruptions could be obtained for the factories with high
interruption costs. This is reflected by the fact that the yearly interruption
costs for all factories together is now only reduced from Dfl. 185,000 for
the base case to Dfl. 154,000 for the "optimum" case.

7. Discussion
When an electricity network for a large industrial complex is designed it
may turn out that the expected number of interruptions due to short
circuits is unevenly distributed over the different factories. Depending on
the market situation the difference in interruption costs of the various
factories may vary to a much larger extent. By properly adjusting the
undervoltage relays protecting the motors in the factories it can be
stimulated that factories with small interruption costs are disconnected
rather fast. In this way it may be hoped that the expected number of
interruptions will be small for factories with large interruption costs so
1256 E Massee and H. Rijanto

that a net economic benifit may result. This has been verified in this
paper by means of a computer program showing that significant
reduction in interruption costs for all factories together may be obtained
in this way.

References
Fransen, H., "Single reserve in the electricity supply of a chemical
industry", Elektrotechniek, Vol. 67, 1989, p.719-722 (in dutch).

Lerch, E. and Simons, J., "Identification methods to solve complex


network problems", Elektrotechnische Zeitschrift, Vol. 112, 1991,
p.712-716 (in german).

Massee, P. and Bollen, M.H.J., "Reliability analysis of industrial electricity


supply", Proceedings Third International Conference on Probabilistic
Methods Applied to Electric Power Systems, London, July 1991,
p.220-223.

Schreurs, C., "The behaviour of motors and adjustable drives during short
duration voltage dips", Elektrotechniek, Vol. 67, 1989, p.706-714 (in
dutch).

También podría gustarte