You are on page 1of 5

Journal of International Council on Electrical Engineering Vol. 1, No. 1, pp.

44~48, 2011


44
Protectability Evaluation for Overcurrent Relay based on
a Probabilistic Method


Wen-Hao Zhang*, Seung-Jae Lee** and Myeon-Song Choi



Abstract - This paper proposes a probability measure for evaluation of relay performance from
two aspects - correct trip and no-unwanted trip. They are developed based on the relationship
between relay settings and relay measurements which follows a Gaussian probability model. The
proposed method is applied to protection capability evaluation of overcurrent relays.

Keywords: Overcurrent relay, Protectability evaluation, Probabilistic model



1. Introduction

Overcurrent relay is a well known device used in power
system protection in the sub-transmission and distribution
side. Its purpose is to isolate electrical equipment in the
presence of a fault. If the main overcurrent relay fails in
operating in the presence of a fault, the backup relay should
be operating at a specified delayed time [1, 2]. In other
words, there should be some coordination constraints
between the main and the backup relay [3-5]. The
overcurrent relays are also used as a backup unit protection
device for other protection relays such as the distance
protection relays [6]. Presently the Inverse Definite
Minimum Time (IDMT) overcurrent relay is used for the
purpose of overcurrent protection. Using the inverse time
characteristics, the relay delays the trip according to the
ratio of the fault current to the set point current according to
the Time Dail Setting (TDS). The ratio of the fault current
and the set point current is called the Plug Setting
Multiplier (PSM).
A detailed mathematical description of this relay can be
viewed in the inverse time relay standards defined by IEEE
[7]. The use of IDMT overcurrent relay has some
disadvantages. Most of the problems relate to the
coordination between the main and the backup relay. One
of the problems faced by the overcurrent relay is fault
severity variation. This would mean that for a single setting,
the relay would require more time to trip for a less severe
fault such as a single phase to ground fault and a shorter
time if the fault was more severe such as a double or a triple
phase to ground fault [8]. It would be more desirable for the
relay to trip with nearly the same time delay regardless of
the fault severity. Another problem associated with the
conventional overcurrent relay includes the inability to
detect the presence of a High Impedance Fault (HIF) [9].
During a HIF, the overcurrent relay would detect a smaller
fault current. This current would at times be similar to the
load current. The relay would then look at the fault as if it
was a normal operation.
Overcurrent relays are connected and set in such a way
that the minimum amount of load is lost when the relays are
activated in an event of a fault. However when the relays
have activated, the other relays which are in the system
have not been configured to the post fault conditions and
parameters. With the pre fault setting on the other relays,
there is a possibility that the coordination of the relays will
be lost in an event of another fault [10, 11]. The potential of
a mal-tripping exist in these kinds of situations.
The distance of the fault from the main relay also plays
an important part in determining the time delay of an
inverse time overcurrent relay. The longer the distance of
the fault from the relay, the smaller the fault current
detected by the current transformer will be. The effect of
this situation will be a larger time delay for the inverse time
overcurrent relays. Previous studies have shown that there
has been a lot of work being done in coordination of the
inverse time
Overcurrent relay to ensure that the relays would trip as
fast as possible without violating the coordination
constraints. A linear programming method could be used in
order to optimize the settings of the overcurrent relays such
that the tripping time is minimized and the coordination
constraints followed [12-16].
In this paper, the measurement errors are modeled in the
second section and concepts of correct trip probability and
no-unwanted trip probability are proposed to evaluate the
protectability of overcurrent relays.


Corresponding Author: Dept. of Electrical Engineering, Tongji
University, China (xiwen.ren@gmail.com)
* Dept. of Electrical Engineering, Tongji University, China
(xiwen.ren@gmail.com)
** Dept. of Electrical Engineering, Myongji University, Korea
(sjlee@mju.ac.kr, mschoi@mju.ac.kr)
Received: May 1, 2010; Accepted: October 16, 2010
Wen-Hao Zhang, Seung-Jae Lee and Myeon-Song Choi



45
2. Probabilistic Modeling

Due to errors that are a consequence of a relays own
imperfection, CT and VT errors and inaccuracy of line
constants, the operation of protective relays would not
follow its apparent setting strictly. In other words, the relay
may not be able to trip correctly.
A probabilistic model for relay errors was presented in [9,
10] for the distance relay protection, which can be extended
to all types of protective relays. The combination of
measurement errors modeled by uniform probability
distributions yield an approximate Gaussian function.
Therefore the effective protection reach that determines a
real operation of a protective relay can be assumed to
follow the following probability distribution function:

2
2
( ) 1
( ) exp
2 2
E S
E
X X
f X


=


(1)

Here X
E
represents an effective range that determines the
actual operation and X
S
represents relay setting. Note that it
is a Gaussian distribution function with a mean of X
S
and
standard deviation of which can be determined by
MATLAB simulation of generic equation of protective
relay operation.
Relays usually have inputs from several current
transformers (CTs), and the zone of protection is bounded
by these CTs. In order to explain about the relationship
between relay setting and its effective protection zone, the
types of relays could be classified into two types [11]:
Protective relays with open protection zone have a
degree of uncertainty in determining the location of
protection boundary. Non-pilot relays usually employ the
open zone protection.
Protective relays with closed protection zone have
absolutely selective characteristic for the internal and
external fault, also known as differential.
Here, take the Relay in Fig. 1 with open protection zone
as example since its much easier to be understood.



Fig. 1. Distance relay setting and its effective reach.

Seen from Fig. 1, if there are no relay errors, the trip
characteristic of relay R1 will seriously follow its setting L
S
,
however, due to the errors, the setting L
S
would have an
effective reach L
E
which determines the actual trip
characteristics of the relay. This effective reach L
E
can be
considered to have a probabilistic behavior resulted by
relay errors. To be more specific, let's suppose an apparent
setting L
S
that is determined by a relay engineer and its
effective reach L
E
would locate around L
S
as a Gaussian
distribution so the relay has the probability of under-
reaching the fault within its protection zone, and it may also
overreach the fault outside the protection zone.
For this Gaussian distribution model, the most important
factor is its deviation. Considering 5% CT error, 5% VT
error, 3% impedance error, and 5% calculation error, the
generic equation for a single line to ground fault is given by
(2) that represents fault distance seen by the relay from its
measurements:

[ ]
0 0 1 1 1
1 )
( )
m
E
m m m m m m
V
L
I I Z Z Z Z
=
+

(2)

Here, V
m
, I
m
, I
0m
are measurements, Z
0m
, Z
1m
are the line
impedances assumed and is the calculation error.
In MATLAB, uniform values of variable r can be
generated from interval [a, b] as

r=a+(b-a)*rand(k,1) (3)

Here, k is the random data number.
Take the 5% CT error as an example. I
m
can be a uniform
distribution in the interval [1-5%, 1+5%] with 1e6 data
number. All parameters generated using (3) can be
substituted into (2) and the density of L
E
would follow the
Gaussian distribution shown in Fig. 2. The shadow area
represents a probability of 99% with [0.871, 1.129] which
equals [1-2.58, 1+2.58], from which the deviation is
calculated as 5%, which can be used for the protectability
evaluation.
In similar process, the deviations of other relays can be
obtained.



Fig. 2. Probability of 99% with the shadow area.
Protectability Evaluation for Overcurrent Relay based on a Probabilistic Method



46
Seen from the above probabilistic analysis, the setting of
a protective relay has an effective zone of protection
following the Gaussian distribution.
The operation of power systems highly depends on the
performance of relays. However, protective relaying
performance is not easy to be quantified. The relay
performance is high affected by the measurement errors
which come from the CT, VT, inaccuracy of line impedance,
A/D conversion and so on. All these errors can be modeled
by uniform probability distributions, and the combination
of a number of distributions would yields an approximate
Gaussian function.
Due to these errors, the relay setting would not trip
exactly as the setting value and it would have an effective
reach which determines the actual trip characteristic. Its trip
characteristic would follow this Gaussian distribution with
the setting as mean value, the effective reach as variable,
and a deviation which could be decided by the combination
effect of errors. The concepts of under-reach and
overreach need to be specified before evaluating the relay
performance. Under-reach means that the relays effective
reach fails to reach the setting value which might result in a
failure to trip. On the contrary, overreach means that the
relays effective reach exceeds the setting value which
might lead to a mis-tripping. Note the additional restriction
on overreach to avoid loss of coordination for some relays.


3. Protectability Definition

The operation of power systems highly depends on the
performance of relays. However, protective relaying
performance is not easy to be quantified. The relay
performance is high affected by the measurement errors
which come from the CT, VT, inaccuracy of line impedance,
A/D conversion and so on. All these errors can be modeled
by uniform probability distributions, and the combination
of a number of distributions would yields an approximate
Gaussian function.
Due to these errors, the relay setting would not trip
exactly as the setting value and it would have an effective
reach which determines the actual trip characteristic. Its trip
characteristic would follow this Gaussian distribution with
the setting as mean value, the effective reach as variable,
and a deviation which could be decided by the combination
effect of errors. The concepts of under-reach and
overreach need to be specified before evaluating the relay
performance. Under-reach means that the relays effective
reach fails to reach the setting value which might result in a
failure to trip. On the contrary, overreach means that the
relays effective reach exceeds the setting value which
might lead to a mis-tripping. Note the additional restriction
on overreach to avoid loss of coordination for some relays
Protectability is defined as an index to denote the
protection degree of a relay for its supposed protection
target under variable relay setting and variable system
conditions.
In order to evaluate the performance of a protective relay,
protectability is evaluated in two aspects from the
viewpoint of a relay operation in this study:
1) Correct trip probability(CTP) for a fault in its
supposed protection line , which is related to the
sensitivity
2) No unwanted trip probability(NUTP) for a fault
outside its supposed protection line, which related to
the selectivity---protection coordination
Note that sensitivity and selectivity are the most critical
requirements of protection. Protectability index derived
based on the probabilistic model introduced above.


4. Protectability Evaluation

Using the difference between the currents flowing
through the relay installed place at normal state and fault
state to judge the exact operating sate is the basic principle
of the overcurrent relay. There must be a substantial
reduction of short-circuit current as the fault is moved
away from the relay toward the far end of the line. The
inverse time delay overcurrent relay is proposed based on
this current change along the line. However, the closer the
fault is to the source, the greater the fault current
magnitude, yet the longer the tripping time. The
instantaneous overcurrent relays makes this system of
protection viable as an addition. The instantaneous relay
can be set to see almost up to, but not including, the next
bus to lower the fault-clearing times. A typical application
of overcurrent relays is shown in Fig. 3.



Fig. 3. Application of overcurrent relays.

Basically, as the increase of setting, the CTP would
become smaller. If the current setting decreases, there
would be increasing NUTP for the fault in the next line or
Wen-Hao Zhang, Seung-Jae Lee and Myeon-Song Choi



47
potential coordination problems.
The protectability of different types of overcurrent relays
can be evaluated as follows by considering their specific
characteristics.


Fig. 4. Sample system.

A. Instantaneous overcurrent relay
The instantaneous OCR must be set to provide high-
speed protection for an appreciable portion of the circuit
and at the same time, to prevent the relay from operating
for faults beyond the end of the line.
For a fault on its own line l
1
, with a larger current setting,
it would have smaller probability to trip the fault, which
means that the correct trip probability (CTP) decreases.
Also, the fault current in the next line l
2
would be smaller
comparing to the fault close to the source so the increase of
current setting for R
1
would have a smaller mis-tripping
probability, which means theres higher no-unwanted trip
probability (NUPT) if the setting increases.
As for the optimal setting, its correct trip probability
(CTP) of instantaneous overcurrent relay must keep 1 for
its protection zone with no probability to mal-operate for
fault in the next line. The instantaneous OCR must be set
not to overreach the bus at the remote end of the line, so its
no-unwanted trip probability (NUTP) must also be 1 for
the fault in the next line.

B. Time-Delay Overcurrent Relay
The definite, inverse time-delay overcurrent relays are
set to coordinate with the instantaneous overcurrent relay to
protect the rest of the line.
1) Definite Time Delay Overcurrent Relay
Definite time delay overcurrent relay is adopted to
protect the rest of the line. The correct trip probability
(CTP) decreases as the setting increases comparing to
the fault current. And the no-unwanted trip probability
(NUTP) would also decrease if the setting increases
because it would have smaller mis-tripping probability
for the fault in next line.
The optimal setting for definite time delay
overcurrent relay must have a correct trip probability
(CTP) as 1 for the fault in its own line. And its no-
unwanted trip probability (NUTP) considering the
coordination with instantaneous overcurrent relay in
the next line must be as large as possible under the
constraint of correct trip probability (CTP) as 1.
2) Inverse Time-delay Overcurrent Relay
The correct trip probability (CTP) and no-unwanted
trip probability of the inverse time-delay overcurrent
relay are very difficult to define since as the relay setting
changes it could trip with different time delay because of
the measurement errors. Normally, the inverse time delay
overcurrent relay would has smaller correct trip
probability (CTP) and also smaller no-unwanted trip
probability (NUTP) as the larger time delay brought by
current setting increase.


5. Conclusion

A probabilistic model is proposed to describe the
effective protection zone of protective relay setting under
the influence of measurement errors and system parameter
errors. A concept of protectability is defined to indicate
the protection degree of the protective relay. The
protectability is evaluated from two aspects: correct trip
probability (CTP) and no unwanted trip probability (NUTP)
by which to elaborate the performance for different types of
overcurrent relays. Protectability evaluation here is
proposed from a qualitative view based on this Gaussian
model and the quantitative analysis would be done in the
further study.


References

[1] Stanley H. Horowitz and Arun G. Phadke, Power
System Relaying, Research Studies Press Ltd, John
Wiley And Sons Inc. 1993.
[2] Warrington, ARC, The Protective Relays. Theory
and Practice, John Wiley & Sons, New York, 1969.
[3] Westinghouse Electric Corp., Applied Protective
Relaying, Westinghouse Electric Corp., Newark, NJ.,
1979.
[4] General Electric Co. Measurements, Protective
Relays Application Guide, GEC, United Kingdom,
1975.
[5] M. Gopal Control Systems Principles and Design,
Mc. Graw Hill, second edition 2003.
[6] Gerhard Ziegler Numerical Distance Protection
Principles and Applications, Publicis MCD, Munich
and Erlangen, 1999.
[7] IEEE Standard Inverse-time Characteristic Equations
for overcurrent relays, IEEE Std C37.112-1996
[8 ] Arturo Conde and Ernesto Vazquez Operation logic
proposed for time overcurrent relay IEEE Trans.
Power Delivery, Vol. 22 No 4 October 2007.
[9] A.H. Etermadi and Sanaye-Pasand High impedance
fault detection using multi resolution signal
decomposition and adaptive neural fuzzy interface
IET Gener. Transm. Distrib. 2008.2.(1). pp. 110-118.
Protectability Evaluation for Overcurrent Relay based on a Probabilistic Method



48
[10] Alberto J. Urdaneta, Harold Restrepo, Saul Marquez
and Jorge Sanchez Coordination of directional
overcurrent relay timing using linear programming
IEEE Trans. Power Delivery, Vol. 11, No. 1, January
1996.
[11] Alberto J. Urdaneta, Ramon Nadira and Luis G. Perez
Jimenez Optimal coordination of directional
overcurrent relays in interconnected power systems
IEEE Trans. Power Delivery, Volume 3, No 3, July
1988.
[12] J.P. Whiting and D. Lidgate, Computer prediction of
IDMT relay settings and performance for
interconnected power systems, IEE Gen., Transmiss.,
Distrib., Vol. 130, No. 3, pp. 139-147, 1983.
[13] Hossien Askarian Abyaneh, Majid Al-Dabbagh,
Hossien Kazemi Karegar Seyed Hesameddin Hossien
Sadeghi, A new optimal approach for coordination
of overcurrent relays in interconnected power system,
IEEE Trans. Power Delivery, Vol. 18, No. 2, April
2003.
[14] Bijoy Chattopadhyay, M.S. Sachdev, T.S. Sidhu, An
on-line relay coordination algorithm for adaptive
protection using linear programming technique,
IEEE Trans. Power Delivery, Vol. 11, No. 1, January
1996.
[15] Perez L. G. and Urdaneta. A. J.Optimal computation
of distance relays second zone timing in a mixed
protection scheme with directional overcurrent relays,
IEEE. PWRD, Vol. 16, No. 3, July 2001.
[16] A.H. Eternadi and Sanaye-pasand High resolution
fault detection using multi-resolution signal
decompositionand adaptive neural fuzzy inference
system IET Gener. Transm. Distrib., 2008, 2 (1),
pp.110-118.


Wen-Hao Zhang was born in Shanxi,
China, in 1982. He received his B.E.
degree in Electrical Engineering from
the Harbin Institute of Technology,
Harbin, China in 2003. He received his
M.S. degree in Electrical Engineering
from Xian Jiao Tong University,
Xian, China in 2006. He received his
Ph.D. at Myongji University, Yongin, Korea in 2010.
Currently he is working in Dept. Electrical Engineering at
Tongji University, China. His research interests are power
system protection and control and power system automation.
Seung-Jae Lee was born in Seoul,
Korea in 1955. He received his B.E.
and M.S. degrees in Electrical
Engineering from Seoul National
University, Korea in 1979 and 1981,
respectively. He received his Ph.D. in
Electrical Engineering from the
University of Washington, Seattle, USA
in 1988. Currently, he is a Professor at Myongji University
and a Director at the Next-Generation Power Technology
Center (NPTC). His major research fields include
protective relaying, distribution automation, and Multi-
Agent applications to power systems.


Myeon-Song Choi was born in
Chungju, Korea in 1967. He received
his B.E., M.S., and Ph.D. degrees in
Electrical Engineering from Seoul
National University, Korea in 1989,
1991 and 1996, respectively. He was a
Visiting Scholar at the University of
Pennsylvania State in 1995. Currently,
he is a Professor at Myongji University. His major research
fields include power system control and protection,
including Multi-Agent applications.

You might also like