1140

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 4, OCTOBER 2003

Load Model Effects on Distance Relay Settings

Kuo-Hsiung Tseng, Wen-Shiow Kao, and Jia-Renn Lin

AbstractThe purpose of this paper is to study load model effects on distance protective relay settings in Taipowers transmission system. Five load models have been used in order to explore
the effects of load characteristics on relay settings: (1) static ZIP
model, (2) dynamic motor model, (3) composite model, (4) PTI
IEEE model, and (5) exponential model. The major points of this
study include (1) setting of impedance values for the three distance protective relay zones and the protection against loss of synchronicity in the outermost periphery, (2) exploring the effects of
load models on impedance angle at the time of distance relay tripping, (3) study whether the load model will cause a malfunction in
the distance relay when the system swings due to a sudden change
in the power system, (4) analyze the coordination of the blocking
time of the out-of-step blocking relay during system power swings,
and (5) propose how to select a better load model for accuracy in
relay settings.
Index TermsBlocking time, distance relay, load model, power
swing.

I. INTRODUCTION

OWER transmission lines are vital to power systems

because they constitute power systems backbone and
supply the highest proportion of electricity throughout the
power-supply area. Due to their exposure to outdoor environment, power transmission lines fault rate is usually much
higher than other facilities, and therefore, their protection is of
particular importance [1].
Studies on the use of distance relays in power swing blocking
have been widely discussed. As pointed out in [2], during a
power swing, there will be a time delay between relay action
and blocking action. In [3], the rate of change in power is used
as a blocking signal. Reference [4] proposes a new blocking
theory for detecting rapid power swings when a phase or
grounding fault occurs which can be cleared immediately
when the fault occurs in the relay tripping zone. Reference [5]
simulates the relay performance during short fault or temporary
swings. None of the above mentioned publications analyze the
effect of load models on protective relays. It is hoped that this
study will clearly point out the effect of load characteristics on
the relay, and that it will be of substantial help in the protection
of the power transmission system.
This study consists of three parts: (1) Establish impedance
values used for distance relays three protection zones which
will also be the basis for time domain simulation used in this
paper. (2) Determining which load model causes the distance

Manuscript received February 26, 2002; revised April 26, 2002. This work
was supported by the Taiwan Power Company of the Ministry of Economic
Affairs.
The authors are with the National Taipei University of Technology, Department of Electrical Engineering, Taipei 106, Taiwan, R.O.C.
Digital Object Identifier 10.1109/TPWRD.2003.817507

Fig. 1. Dynamic induction motor equivalent circuit.

relay to malfunction when the system is subject to a power

swing, and (3) Explore the setting values for blocking time when
distance relays are integrated with out-of-step blocking relays.
II. LOAD MODELS
This paper adopts the following five load models appropriate
for the Taipower system for comparative analysis.
1) Static ZIP Model: As shown in (1) and (2), the load is
decomposed into its active and reactive parts each represented
by three components, namely, the constant impedance part (Z),
constant current part (I), and the constant power (P) part
(1)
(2)
where
,

active and reactive power load,

respectively;
active and reactive power load at
specified bus;
constant impedance load part;
constant current load part;
constant power load part;

,
,
,
,
,

2) Dynamic Motor Model: In this model, active and reactive parts of the load are expressed as a function of the past
and present system voltages and frequencies [5] and represented
by an induction motor equivalent circuit as shown in Fig. 1 [6]
where
static impedance;
rotor impedance;
excited reactance;
static reactance;
rotor reactance;
rotor slip.
3) Composite Load Model: This is a combination of the
static and dynamic load models above [6] whose equivalent
circuit is shown in Fig. 2.

0885-8977/03$17.00 2003 IEEE

TSENG et al.: LOAD MODEL EFFECTS ON DISTANCE RELAY SETTINGS

Fig. 2.

1141

Equivalent circuit of composite model.

Fig. 4.

Distance relay protective zones after amendments.

out-of-step protective zone, the second circle represents zone

3, and the innermost circle represents zone 1. The flowchart for
determining the impedance values settings of each protective
zone is shown in Fig. 5 while part of the calculated results are
listed in Table I.
IV. TIME DOMAIN SIMULATION
Fig. 3. Traditional distance relay protective zones.

4) PTI IEEE Model [7]: According to this model, active and

reactive part of the load shall be represented by (3) and (4) which
are both voltage and frequency dependent
(3)
(4)
.
where frequency deviation
5) Exponential Model: As represented by (5) and (6), using
an arbitrary exponential of system voltages, this load model has
the characteristics of using fewer parameters and is usually used
to represent mixed load
(5)
(6)
where
voltage dependent parameter of the active power;
voltage dependent parameter of the reactive power;
actual voltage;
nominal voltage.
III. DISTANCE RELAY SETTINGS BETWEEN PROTECTIVE ZONES
In order to accurately set the relays on both ends of every
transmission line, the transmission line parameters must be
accurately measured. Protective zones of distance relays are
usually divided into three as shown in Fig. 3 [8]: zone 1 protects
85% of the circuit, zone 2 protects 150%, and zone 3 protects
225%. Applying this rule to Taiwan Powers 345-kV transmission network and incorporating the out-of-step blocking
relay requirement, the amended distance relay protective zones
are shown in Fig. 4 where the outermost circle represents the

Fig. 6 shows part of Taiwan Powers 345-kV network where

a three-phase ground fault had occurred causing blackout of
northern Taiwan in 1999 and which is used in this paper for
time domain simulation. The three-phase ground fault will be
duplicated by computer simulation in order to find out the load
models effects on impedance trajectory, power swing as observed by distance relay, and impedance relays blocking performance.
A. Description of the Event
At 11:31 P.M. on July 29, 1999, Tower #326 of Taiwan
Powers 345-kV transmission network located in Tsuochen
Village, Tainan County, fell down and caused two 345-kV
transmission lines: Lungchi-Chungliao Mountain Line and
Lungchi-Chiamin Sea Line, attaching to that tower to trip and
also Hsingta Power Plant to disconnect. The incident eventually
triggered a series of relay operations isolating Taiwans power
network into two islands [9].
Before the incident, the system was operated under normal
conditions with both system frequency (60 Hz) and voltages
within normal range. At the time of the occurrence, total generation of the system was 20 090 MW. Power generation in the
area north to Lungchi was 12 100 MW where the corresponding
load was 15 500 MW. The shortage of 3,400 MW was transferred from south to north by the two double-circuits of 345-kV
transmission line in Lungchi (1,880 MW from Chiamin 1 and 2,
and 1,540 MW from Chiamin sea line and Chungliao mountain
line).
B. Load Model Effects on Impedance Trajectory
Fig. 7 shows the impedance trajectory seen by relay #114 of
the Lungchi (N)3rd Nuclear line as simulated by the computer for different load models. As can be seen, a series of chain
reactions caused relay #114 to trip in all cases.
Fig. 8 shows the impedance trajectorys entry angle at each
protection zone. It can be noted that, regardless of the load

1142

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 4, OCTOBER 2003

TAIPOWER

Fig. 6.

TABLE I
VALUES FOR DISTANCE RELAYS IN
EXTRA-HIGH VOLTAGE POWER TRANSMISSION LINES
(IMPEDANCE MODEL)

Diagram of a portion of the 345-kV(E/S) transmission line in Taiwan.

namic motor, composite, static ZIP, exponential, and PTI IEEE

model.
C. Effect of Load Models on Swing Curve

Fig. 5.

Program flowchart for settings between distance relay protective zones.

model used, the impedance angle increases gradually as they

move from outer to inner circles. Difference in entry angle
among various load models are the greatest in zone 1 where
the angle simulated based on the dynamic model is the largest
and that of the composite model is the smallest. In zone 3 and
out-of-step zones, there is only noticeable angle difference
between the composite model versus the rest.
Table II lists the impedance angle at the time of tripping
for different load models as compared to that simulated using
Taiwan Powers experienced model (constant current model). It
can be seen that due to instantaneous action, the tripping angle of
all load models are quite close to that of Taiwan Powers experienced model with a difference range of 0.017 51 0.106 25 .
Rank of the differences, from large to small, is as following: Dy-

Following the system failure and line tripping, significant

changes can be observed in system voltages and currents which
caused the distance relay at the opposite end to detect a swing in
impedance value. To analyze this power swing, this paper uses
the Chungliao (N) (marked in black) relay #79 for its simulation
and the study results are shown in Fig. 9. From Fig. 9, it can be
seen that: (1) Outside the protective zone, significant swing can
be observed for the dynamic load model and composite load
model cases. (2) After entering the out-of-step protection zone,
the swing range, from large to small, is ranked as follows: Dynamic motor, composite, exponential, static ZIP, and PTI IEEE
model. The former three models have much more significant
swings than the other two with impedance trajectories even entering protection zone 3 of the distance relay.
D. Effect of Load Models on Blocking Time
Referring to Fig. 4 in the previous section, the out-of-step
relay blocking system used in this paper is as following.

TSENG et al.: LOAD MODEL EFFECTS ON DISTANCE RELAY SETTINGS

(c)

1143

(f)

Fig. 7. Impedance diagrams observed from the relay #114 further away from the point of failure (the end at the 3rd Nuclear Power Station). (a) Static ZIP model.
(b) Dynamic motor model. (c) Composite model. (d) PTI IEEE model. (e) Exponential model. (f) Constant current.

1144

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 4, OCTOBER 2003

(4.998 cycles) are longer than 3.6 cycles as such the distance
relay will not trip. For the PTI IEEE model, a rapid swing is
experienced with the blocking time being only one cycle; however, the impedance trajectory never gets to the inner circle as
such the line is not tripped either.
V. DISCUSSION OF SIMULATION RESULTS

Fig. 8. Angles at which different types of load models enter into individual
protective zones (from the end at the 3rd Nuclear Power Station).
TABLE II
DIFFERENCES

IN

ANGLE

AT THE POINT OF
OF RELAY TRIP

FAILURE

AT THE

TIME

If the impedance trajectory moves so quickly such that the

time it takes to traverse the region from Zos to 21-2
is less than the default out-of-step action time [i.e., 60 ms
(3.6 cycles)], the distance relay will trip the line and no
blocking action will be initiated.
If the impedance trajectory moves slowly such that the
time it takes to traverse the region from Zos to 21-2
is greater than 60 ms (3.6 cycles), the blocking relay will
then block the distance relay so that no tripping action
is taken even if the impedance trajectory enters the 21-2
tripping zone.
The time it takes for the impedance trajectory to traverse
is referred to in this paper as
region from Zos to 21-2
blocking time.
Table III lists the blocking time for each load model based
on simulation results in Fig. 7 for relay #114 of the Lungchi
(N)3rd Nuclear Line and Fig. 9 for relay #79 on the Chungliao
(N)Chiamin 1 Line.
It can be noted from Table III that: (1) For the relay #114 case
in Fig. 7, the static ZIP model (2.478 cycles), dynamic motor
model (3.48 cycles), PTI IEEE model (2.562 cycles), exponential model (2.844 cycles), and constant current model (Taipower
experienced model, 2.52 cycles) have blocking time less than
3.6 cycles while that of composite model (6.87 cycles) is greater
than 3.6 cycles. The Lungchi (N)3rd Nuclear line will therefore be tripped for all load models except for the composite
model case whose blocking relay will block the distance relay
action. (2) For the Relay #79 case in Fig. 9, the blocking time
for static ZIP model (5.25 cycles), dynamic motor model (4.70
cycles), composite model (6.99 cycles), and exponential model

Based on the above simulation, it can be observed that

1) When the distance relay is installed very close to the point
of failure, the relay will trip regardless of load model
used. The main reason is that under this condition, the
impedance trajectory will enter the trip zone instantly and
cause the relay to trip.
2) When the distance relay is installed further away from the
point of failure, the trip time will be different for each load
model used. This may be due to the large differences in
impedance angle in zone 1 for each load model.
3) The most significant swings in impedance trajectory
are observed when using the dynamic motor model and
composite model followed by exponential, static ZIP,
and PTI IEEE models. This should be due to that when
dynamic motor and composite models are used, they
cause more significant changes in power flow, and thus,
the impedance observed.
4) In Fig. 7, all load models have blocking time less than 3.6
cycles except for the composite model case. In Fig. 7(c),
blocking action was initiated for the composite model and
caused the simulated result to differ from the actual result.
This should be due to the parameters of composite model
which is a combination of dynamic and static models.
VI. CONCLUSIONS
This paper has proposed the use of load models to explore
their effects on distance protective relay settings in the Taiwan
Powers extra-high voltage transmission lines system. The following conclusions can be reached through comparisons between simulation results and the actual situation.
1) It should be noted that load models can significantly influence the blocking time simulation and the accurate setting
of distance relays.
2) By comparing the simulated impedance trajectory of various load models versus that based on Taiwan Powers
experienced model (i.e., constant current model), it was
found that the PTI IEEE load model can best represent the
load characteristics when a system swing occurs or when
the fault was cleared. However, when using the dynamic
motor and composite model, the distance relay tends to
malfunction and the swing is more serious.
3) Blocking action simulation based on: (1) PTI IEEE
model, (2) static ZIP model, (3) exponential model, and
(4) dynamic model shows good agreement with actual
situation. However, when using the composite model,
significant difference is noticed between the simulated
result and actual situation.
4) Based on study in this paper using Taiwan Powers experience, the authors suggest that when distance relays

TSENG et al.: LOAD MODEL EFFECTS ON DISTANCE RELAY SETTINGS

1145

(a)

(b)

(c)

(d)

(e)

Fig. 9. Impedance diagram of the relay #79 in North of Chungliao outside the point of failure. (a) Static ZIP model. (b) Dynamic motor model. (c)
Composite model. (d) PTI IEEE model. (e) Exponential model.

1146

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 4, OCTOBER 2003

TABLE III
BLOCKING TIME FOR EACH LOAD MODEL

are the main system protection, the preferred sequence

for load model selection is: PTI IEEE, static ZIP, exponential, dynamic motor, and composite model. Further
studies and field tests are needed to better estimate the
accurate load model parameters and distance relay characteristics.
ACKNOWLEDGMENT
The authors also want to thank vice director Ho-Chang Lee,
Taiwan Power Company, Taiwan, R.O.C., for providing valuable suggestions on the improvement of the paper.

[9] Ministry of Economic Affairs, Survey report from Taiwan Power Company regarding the July 29, 1999 power-outage, in Research Rep. From
the Ministry of Economic Affairs, Executive Yuan, Taipei, Taiwan, Sept.
1999.

Kuo-Hsiung Tseng received the M.S. degree in electrical engineering from

Monmouth University, West Long Branch, NJ, in 1988.
Currently, he is an Associate Professor with the Department of Electrical Engineering at National Taipei University of Technology, Taipei, Taiwan, R.O.C.
His current research interests include power system protection, load modeling,
and distribution automation.

REFERENCES
[1] IEEE Guide for Protective Relay Applications to Transmission Lines,
IEEE Std. C37. 113-1999, Feb. 29, 2000.
[2] A. R. van C. Warrington, Protective Relays: Their Theory and Practice. New York: Wiley, 1977.
[3] J. Machowski, New power swing blocking method, in Proc. 6th
Int. Conf. Power Syst. Protection, Mar. 1997, Conf. Publ. no. 434, pp.
218221.
[4] A. Mechraoui et al., A new blocking principle with phase and earth
fault detection during fast power swings for distance protection, IEEE
Trans. Power Delivery, vol. 10, pp. 12421248, July 1995.
[5] A. E. Efthymiadis et al., Simulation of protective relay performance
under short-circuit and transient swing conditions, IEEE Trans. Ind.
Applicat., vol. 26, pp. 11081115, Nov./Dec. 1990.
[6] W. S. Kao et al., Dynamic load modeling in Taipower system stability
studies, IEEE Trans. Power Syst., vol. 10, pp. 907914, May 1995.
[7] J. C. Wang et al., Development of a frequency-dependent composite
load model using the measure approach, IEEE Trans. Power Syst., vol.
9, pp. 15461556, Aug. 1994.
[8] K. W. Leung, Computer-aided setting calculation for distance zone 2
and zone 3 protection, in Proc. Int. Conf. Advances in Power Syst.
Contr., Operation Manage., Nov. 1991, pp. 152157.

Wen-Shiow Kao was born in Taiwan Hua-lien, R.O.C., on August 24, 1953.
He received the M.S. degree in electrical engineering from National Taiwan
University, Taipei, Taiwan, R.O.C.
Currently, he is a Professor at National Taipei University of Technology,
Taipei, Taiwan, R.O.C. His research interests include analysis of power system
dynamics, load modeling, and stability.

Jia-Renn Lin received the B.S. and M.S. degrees in electrical engineering from
National Taipei University of Technology, Taipei, Taiwan, R.O.C., in 1999 and
2001, respectively.
He was an Assistant Researcher with National Science Council R.O.C. cooperating project from 2000 to 2001. His research interests include power system
control, power system protection, and control system analysis.