2005 IEEE/PES Transmission and Distribution

Conference & Exhibition: Asia and Pacific

Dalian, China

Dynamic Reduction of Large Power System in

PSS/E
J. P. Yang, G. H. Cheng, Z. Xu, Member, IEEE

suggests that those buses which are not easily affected by the
Abstract-- This paper presents the procedure to establish the disturbances in some areas can be eliminated. Since it is
dynamic equivalent models of large-scale power systems in the difficult to find these undisturbed buses and additional
PSS/E program. It consists of identification of coherent modification has to be made in the original dynamic
generators, aggregation of coherent generators and network simulation programs in order to make use of the state matrix
reduction. Coherent generators are identified by means of time of the equivalent equations, this method is not widely applied.
domain simulation. Aggregation of coherent generators based on
the classical model is performed in this paper. Network is
In the 1970s, a novel coherency based method was presented,
reduced through topological reduction. The procedure is with the corresponding dynamic reduction programs,
illustrated on the East China Power Grid (ECPG) and DYNRED [4] and DYNEQ [5], developed under the
simulations are performed to verify its accuracy, with fault American power research institute (EPRI). It identifies the
applied on the equivalent system and the original system coherent generators on the basis of the swing curves obtained
respectively. Results of the simulation show that the developed by time domain simulations. This method has found some
equivalent system has high accuracy in representing the dynamic applications at that time. The NETOMAC program of
characteristics of the original system. The developed equivalent Siemens developed the evaluation equivalent method [6], in
model can be used for electromagnetic transient analysis, on-line which the complete structure and parameters of the external
dynamic security assessment and system control design. system are not required. Slow coherency technique [7] based
on singular perturbation theory was developed in the early
Index Terms-- large-scale power system, dynamic equivalent,
coherency, identification, aggregation, reduction 1980s, which defines the coherent generators by means of
modal analysis and coherency analysis. Both the approaches,
I. INTRODUCTION however, have difficulties in the application in large-scale and
nonlinear time domain simulations. Since PSS/E can
THE power system in China has increased in both size and accomplish nonlinear time domain simulations for the large-
complexity due to the rapid growth of widespread scale power system efficiently, this paper attempts to construct
interconnections. For such a large-scale system it is neither the equivalent model for the power system under PSS/E with
practical nor necessary to perform the electromagnetic the procedure directly implemented in it.
transient analysis, on-line security assessment and system This paper mainly describes the procedure to establish the
control design with the detailed system model. Therefore, the dynamic equivalent model of power systems in the program
investigation of dynamic behavior of large interconnected PSS/E. The overall procedure can be divided into three stages:
power system requires the use of dynamic equivalents for
identification of coherent generators, aggregation of coherent
particular parts of the system. Actually, when analyzing a generators and network reduction. The classical generator
large-scale power system, we are usually interested in the
behavior of a certain part of the system which is called the
models are assumed in this paper, without consideration of
study system, while the rest of the system is called the external turbine and governing system, excitation system, and PSS.
system. Dynamic equivalent is the process of reducing the
complexity of the external system model while retaining its II. PROCEDURE
effect on the study system [1]. The large-scale power system The process for dynamic reduction of a power system can
models can be reduced significantly with this method while be divided into three steps:
maintaining acceptable accuracy with respect to a specific 1. Identification of coherent generator groups.
phenomenon. 2. Aggregation of generators in the group.
The research in dynamic reduction of power system can 3. Reduction of the network.
date back to the late 1960s when several methods such as The following subsections provide a brief description of the
modal equivalent, coherency-based equivalent and evaluation techniques used in each of the three steps.
equivalent have been developed to determine a dynamic A. Identification of coherent generators
equivalent of the power system. The modal equivalent method
In this stage, coherent generators are identified [8]-[11].
The coherency identification technique is based on the fact
Project Supported by National Science Foundation of China (50277034) that some generators tend to swing together following a
J. P. Yang, G. H. Cheng and Z. Xu are with the Department of Electrical
Engineering, Zhejiang University, Hangzhou, 310027 P. R. China (e-mail: disturbance in the multimachine power systems. Two
hvdc@cee.zju.edu.cn).
0-7803-9114-4/05/$20. 00 ©2005 IEEE.

Authorized licensed use limited to: STATE UNIV NY BINGHAMTON. Downloaded on May 12,2021 at 14:31:24 UTC from IEEE Xplore. Restrictions apply.
generator buses are to be defined as coherent if the angular
difference between them keeps constant within a certain
tolerance over a certain time interval when the power system
is perturbed. The mathematical expression can be defined as
follows:
Vi (t)- Vi (t) ey[5i (t)-,5j(t)]
V1(t) V1(t)
e
~~1
a1 :1

I
2

(1)
Vj (O) Fig.Aggregatino oeetgnrtrbss
1.
Fig.1.Aggregtionofoheentge
= const
Assuming the voltage magnitude of the coherent buses to be
constant, (1) can be further simplified as
15i (t) 5j (t) = J54 (t) = d54 (°) = const (2)
-

where 8i (0) and 8, (0) = 8i (0) - 5j (0) are the initial values The mechanical and electrical power of the equivalent
of the variables calculated for the reduced model. generator are the sum of the mechanical and electrical power
This paper determines the coherent generators by means of of all the generators in the same coherent group, separately.
time domain simulations, making use of the great capability of They can be represented as follows:
PSS/E in performing dynamic simulation of large-scale power n
systems. The swing curves of the generators in the external Pm E/Pmi (6)
system are observed by applying a fault in the study system i=l
n
and those generators with the most identical swing curves are
classified as coherent. Pe E Pei
(7)
i=l
B. Dynamic Aggregation ofgenerators When the generators are represented by classical models,
After coherent groups of generators are identified, the aggregation of the mechanical parameters as well as the
generators in a group can be aggregated to an equivalent transient reactance are to be performed.
generator, while the generator buses are lumped together to The swing equation of the rotor is
form an equivalent bus. The aggregation of coherent generator do) .
buses [11-[12] is based on the Zhukov's method in which M.1 ' =P .-P .-Dw,i i=12 n (8)
each terminal bus is connected through an ideal transformer dt ml el i
with complex turns ratio to the equivalent bus. The voltage of where co, Pmi Pei, M, and Di are respectively the angular
the equivalent bus is defined as the average voltage of the speed, mechanical power, electrical power, inertia and
coherent generator buses, which can be mathematically damping constants of generator i; n denotes the number of
expressed as the coherent generators. All the parameters in (8) are based on
n
the same MVA base.
E Vk As the angular frequencies of the coherent generators are
k=1 (3) most identical and thus assumed to be equal to c, the swing
t

n equation of the equivalent generator can then be described by

n
n do n n n

0 k=1
Ok
(4)
(Mi ) dt
i=1
=
I
i=1
Pmi-E Pei- Di )o
n (9)
The turns ratio of the ideal transformer is given by
i=1,2 n
-k (5) Therefore the inertia and damping constant of the equivalent
generator can be defined as the sum of the inertia and
where Vk and Vt are the voltages at buses k and t damping constants of the coherent generators, respectively:
n
respectively. This process can be illustrated in Fig. 1 where
the buses k, k-I and k-2 represent the generator buses in a
M* =YM i=l
(10)
coherent group, while the bus t indicates the equivalent bus n
formed by these generator buses. D* = D (1 1)
i=l

As all the coherent buses are connected to the equivalent

bus through the ideal transformers, the transient reactance of
2

Authorized licensed use limited to: STATE UNIV NY BINGHAMTON. Downloaded on May 12,2021 at 14:31:24 UTC from IEEE Xplore. Restrictions apply.
 3

the equivalent generator can be obtained by paralleling the in order to determine the coherent generators. The results
transient reactance of all the coherent generators: show that the swing curves of the generators in the same area
1 are most identical. Therefore, four coherent groups of
Xd (12) generators can be defined which include the generators in
Xd 1
Y,/ Shanghai, Zhejiang, Anhui and Fujian area respectively.
i=l Xdi After the coherent generators in each group are lumped into an
C. Reduction of network [11]-[14] equivalent generator, network is reduced by deleting the load
buses whose voltages are below 500kV in the external system.
Network reduction is the process of deleting some load The resulting equivalent network is reduced to 709 buses,
buses in order to reduce the size and complexity of the
network. The topological reduction method [11] is adopted in 1481 branches, 163 generators and 459 loads. Table I gives a
this paper, which eliminates the selected load buses by comparison between the original and equivalent system.
TABLE I
performing Gauss elimination operation on the admittance COMPARISON BETWEEN THE ORIGINAL AND EQUIVALENT SYSTEM
matrix of the external system. The node equation of the
original network can be described by:
[La] [~~~K~~.][~4] (13)
LIR j YRR YRE HVR j(3
where the subscripts R and E refer to the variables at the
nodes to be retained and eliminated, respectively.
The equivalent is obtained by rearranging the second row To assess the accuracy of the equivalent model, a three-
of (13) as phase fault is applied on the 500kV bus at Yangzhou plant and
cleared after 0. Is. Simulation is performed on the original and
VE YEE(E -YERVR) (14) equivalent system respectively for 2.Os and the results are
Thereby the current equation of the retained buses can be shown in Fig. 2 to Fig.4.
obtained by substituting (14) into the first row of (13)
YRVR+ KIIE (15)
1.0-
where
e 0.8
YR YRR YRE YEE ER (16) Original system
K = YREYEE (17) 006 Equivalent system
The admittance matrix YR corresponds to a reduced m 0.4-
equivalent network that consists of the retained buses and the
equivalent branches linking them. The matrix KI is often 0.2-

referred to as the distribution matrix which passes the node

currents from the eliminated buses to the retained buses. The 0-. 0.5 1.0 1.5 2.0
Time (s)
first term of (15) specifies a set of equivalent branches and Fig.2. Voltage responses of the faulted bus in the original and equivalent
static shunt elements connecting the retained nodes, while the system.
second term specifies a set of equivalent currents which must
be impressed on the retained nodes to reproduce the effect of
load currents at the deleted nodes. These equivalent currents
may be translated into equivalent constant real and reactive 15-
Original system
power loads and added to the original loads at the retained Equivalent system
buses. 10- u

\~~~~~-
Pa)
III. APPLICATION 5-
The East China Power Grid (ECPG) is used as the sample
system to evaluate the equivalent model of the external system. 0-
The ECPG consists of 2055 buses, 2896 branches, 469
generators and 874 loads. Since the focus of the ECPG study
is the transient stability performance of Jiangsu area, Jiangsu 0.0 0.5 1.0 1.5
Time (s)
2.0
area is identified as the study system whereas the rest of the
system including Shanghai, Zhejiang, Anhui, Fujian areas is Fig.3. Active power response of the generator at 345# bus in the original and
equivalent system.
referred to as the external system. There are 19 branches
between these two subsystems. Time domain simulation is
performed on the ECPG with fault applied in the study system,
3

Authorized licensed use limited to: STATE UNIV NY BINGHAMTON. Downloaded on May 12,2021 at 14:31:24 UTC from IEEE Xplore. Restrictions apply.
 4

[6] Kang Yi IX Zhou Xianlin IX Xie Guoen, "Study of Power System

Dynamic Equivalents in NETOMAC," Power Technology, Vol. 22, No. 5,
pp. 21-24, May 1998.
[7] JOE.H.CHOW, R.GALARZA, RACCARI, WWPRICE, "Inertial and
slow coherency aggregation algorithms for Power System Dynamic
model reduction," IEEE trans. On Power Systems, Vol. 10, No. 2,
pp.680-685, May 1995.
[8] R. Podmore, "A Comprehensive Program for Computing Coherency-
based Dynamic Equivalents," 1979 Power Industry Computer
Applications Conference, pp. 298-306.
[9] R. Podmore, "Identification of Coherent Generators for Dynamic
Equivalents," IEEE Trans. on Power Apparatus and Systems, Vol. PAS-
97, pp. 1344-1354, July/August, 1978.
[10] M.L.Ourari, L.A.Dessaint, V.Q.Do, "Coherency Approach for Dynamic
Equivalents of Large Power Systems," International Conference on
Power systems Transients - IPST 2003 in New Orleans, USA.
1.5 2.0
[11] Jan Machowski, Janusz W.Bialek, James RlBumby, "Power System
Time (s)
Dynamics and Stability," John Wiley & Sons Ltd, 1997.
Fig.4. Rotor angle response of the generator at 345# bus in the original and [12] A.J.Germond, R.Podmore, "Dynamic Aggregation of Generating Unit
equivalent system. Models," IEEE Trans. on Power Apparatus and Systems, Vol. PAS-97,
No.4, pp. 1060-1069, July/Aug 1978.
It can be seen from the figures that responses of the [13] PTI, PSS/E Application Guide. Vol. I [M]. PSS/E Brochure, 2002.
derived equivalent model agree very well with the original [14] Thomas L.Baldwin, Lamine Mili, Arun G.Phadke, "Dynamic Ward
Equivalents for Transient Stability Analysis," IEEE Trans. on Power
system, thus proving the accuracy of the reduced order model Systems, Vol. 9, No. 1, pp. 59-67, February 1994.
for transient stability analysis. Therefore, the proposed
procedure is efficient in developing dynamic equivalents for VI. BIOGRAPHIES
large-scale power systems which can provide significant Jingping Yang was born in Shanxi, China, in September
reduction in system model size while maintaining acceptable 1974. She received the BS and MS degrees from Taiyuan
accuracy. University of Technology, China, in 1996 and 2002
respectively, all in Electrical Engineering. She is now a Ph.D.
IV. CONCLUSION student in the E.E. Department of Zhejiang University. Her
current research interests include power system dynamic
This paper presents the procedure to construct the dynamic equivalents.
equivalent models of large-scale power systems in the PSS/E
program. The procedure mainly consists of identification of Gaihong Cheng was born in Shanxi, China, in April 1977.
coherent generators, aggregation of coherent generators and She received the BS and MS degrees from Taiyuan
network reduction. The procedure is illustrated on the East University of Technology, China, in 1999 and 2002
China Power Grid (ECPG) and simulations are performed to 1- .17 respectively, all in Electrical Engineering. She is now a
Ph.D. student in the E.E. Department of Zhejiang University.
assess the accuracy of the equivalent model, with fault applied Her research interests are power system restoration and
on the equivalent system and the original system respectively. HVDC.
Results of the simulation show that the developed equivalent
Zheng Xu was born in Zhejiang, China, in September 1962.
system has high accuracy in representing the dynamic He received the BS, MS and Ph.D. degrees from Zhejiang
characteristics of the original system. The developed University, China, in 1983, 1986 and 1993 respectively, all
equivalent model can be used for electromagnetic transient in Electrical Engineering. He has been with the Electrical
analysis, on-line dynamic security assessment and system Engineering Department of Zhejiang University since 1986.
Since 1998 he is a professor of Zhejiang University. His
control design. research area includes HVDC, FACTS, power harmonics
and power quality.
V REFERENCES
[1] Ni Yixin, Chen Shousun, Zhang Baolin, "Theory and Analysis of
Dynamic Power System," Tsinghua University Publisher, BeiJingZMay
2002.
[2] Sebastiao E.M.de.Oliveira, J.F.de Queiroz., "Modal Dynamic Equivalent
for Eletric power systems Prat I: Theory," IEEE Trans. on power system,
Vol. 3, No. 4, pp. 1723-1730, November 1988.
[3] Sebastiao E.M.de.Oliveira, Alexandre G.Nassaud, "Modal Dynamic
Equivalent for Eletric power systems Prat II: Stability Simulation Tests,"
IEEE Trans. on power system, Vol. 3, No. 4, pp. 1731-1737, November
1988.
[4] Lei Wang, Meir Klein, Solomon Yirga, Prabha Kundur, "Dynamic
Reduction of Large Power Systems for Stability Studies," IEEE trans. On
Power Systems, Vol. 12, No. 2, pp. 889-895, May 1997.
[5] WW.PRICE, JOE.H.CHOW, A.WHAQGAVE, M.HIRSCH, "Large-
scale System Testing of Power System Dynamic Equivalencing
program," IEEE Trans. on Power Systems, Vol. 13, No. 3, pp. 768-774,
August 1998.

Authorized licensed use limited to: STATE UNIV NY BINGHAMTON. Downloaded on May 12,2021 at 14:31:24 UTC from IEEE Xplore. Restrictions apply.