The Journal of CPRI,

Vol. 13, No. 2, June 2017 pp. 263-268

Network reduction of power system for transient stability studies

Ashwin N* Sreedevi J** and Pradip Kumar Dixit***

Abstract: Modern power systems are complicated and heavily interconnected systems. Hence critical
analysis of the network for power system planning is very important. Challenges in analyzing large
interconnected networks have given rise to the need to find small, equivalent networks in order to
improve computational efficiency. There are many methods for determining an equivalent network
which accurately represents the original network. In this paper Thevenin’s method is used to calculate
the equivalent of a system. The PSS/E software is used to demonstrate the method for separation
of a particular area network from all-India network by placing dynamic equivalents at the boundary
buses.

Keywords: boundary buses, dynamic equivalent, internal and external systems, transient stability.

1.0 INTRODUCTION complete information among one another. (ii)

Practical limitations on computational resources.
Uncertainties in load growth, increase in (iii) As the electrical distance from the point of
penetration of renewable energies, increased interest increases, the requirement of detailed
use of power electronic converters and growing modelling of remote location decreases (iv)
competitiveness of electricity markets have led With the help of optimally placed measuring
to the operation of the power system under instruments, such as Phasor Measurement Units
heavily loaded and critical conditions. The (PMU), it may be more efficient to calculate
overall characteristics of the power systems are equivalents at boundary buses.
constantly under change. Hence it is imperative
for the power systems planners, to not only It must be noted that achieving faster simulation
analyze the power system accurately but also in and optimization times is not the only goal for
a quick and efficient manner. Hence, modelling applying the model reduction [2]. Sometimes,
of the power system plays an important role in its it is most important to get the model with
efficient analysis. the lowest number of variables. For example,
world-wide power systems increase in size and
Large interconnected power systems are difficult complexity year-by-year due to the rapid growth
to handle. By using small, equivalent systems of widespread interconnections and much higher
which are representative of the original network penetration of distributed resources. Today,
the computational requirements can be reduced. the interconnected power systems cover large
Some of the reasons for using a reduced model geographical areas and comprise thousands of
are [1]: (i) A typical interconnected power devices, so the dimension of the models may
system consists of number of areas owned by easily reach the order of several thousands
various utilities which are reluctant to share of state variables and more. For such large-scale
*Senior Research Fellow Power Systems Division Central Power research Institute Bengaluru, India [email protected]
**Joint Director Power Systems Division Central Power Research Institute Bengaluru, India [email protected]
***Professor Dept. of Electrical and Electronics Engineering M S Ramaiah Institute of Technology Bengaluru, India [email protected]
264 The Journal of CPRI, Vol. 13, No. 2, June 2017

power systems and their full detailed models, is connected to the internal system via tie lines.
it is neither practical nor necessary to perform Boundary buses are those buses connecting the
dynamic studies such as stability studies and external system and the tie lines, as shown in Fig.
dynamic security assessment. 1. In power system analysis tools, the internal
system is modelled in detail and the external
Network reductions techniques are generally system is usually represented by equivalent
classified as static and dynamic based on the models. According to the authors in [9], there
model representation and its intended use [3]. are three modes in which an equivalent can
Static reductions model the equivalent as a be calculated based on the availability of data.
snapshot of the system and is used for static
analysis such as power flow calculations, system i). Mode 1: The solved load-flow model of the
planning and operational studies. Dynamic entire interconnected system is available. This
reduction techniques are intended for mode is applicable in power system planning.
analysis of dynamic effects such as large scale This mode is used only for simple reduction of
power system transient analysis for large and system size.
small disturbances and power system security
assessment. The term network reduction in this ii). Mode 2: The solved load flow model of the
paper refers to 50Hz equivalent of the system internal system is available. Data for corresponding
that can be used for static and dynamic studies. external system not precise. This mode is used
for producing the best possible external model
The idea of partitioning the interconnected using whatever information available. Mostly
system into internal and external systems was used in on-line applications like state estimation.
first given by Ward, who originally proposed the
technique of network reduction in [4]. In Ward’s iii). Mode 3: An unsolved model of the internal
method, the external system is removed by system is available. A separately obtained external
Gaussian elimination. This method was developed equivalent is to be attached to it, before load-flow
along the years and modified as extended Ward solution can be run. This mode is used in power
equivalent method by A. Monticelli et.al [5]. system expansion planning.
The Ward method and its variations are the most
popularly used methods for finding equivalents. This paper deals with construction of equivalent
in accordance to Mode 1, where the solved load
Another type of method is the REI (Radial flow model of the entire interconnected system is
Equivalent independent) network reduction available. The main aim of such equivalencing is
technique, first discussed in [6]. Further to achieve reduced network for transient stability
development of the method was done in [7] and studies.
[8]. The main idea is to identify a group of similar
nodes and to replace each group by one virtual
node. The power injection at this virtual node is
equal to the aggregated injections of the group of
nodes that will be replaced, and the aggregated
nodes become passive.

The generalities of equivalencing is explained in

detail in [9]. The interconnected system can be
Fig. 1 Equivalencing in interconnected
divided into two parts: an “internal” subsystem systems.
and an “external” subsystem as shown in Fig.1. In
practice, the area under consideration is usually
The authors in [9] have proposed the following
modelled as the internal system. The rest of the
basic properties for an equivalent system
network is modelled as the external system and
The Journal of CPRI, Vol. 13, No. 2, June 2017 265

1) The equivalent system should be a reliable could alternatively be calculated as VTH divided
accurate model of the external system’s effect by the short circuit current between A and B
on the internal system. when they are connected together. Thevenin’s
equivalent is similar to REI reduction, where
2) The equivalent model must not cause any
group nodes are replaced by a voltage source in
load flow solution convergence or precision
series with an impedance.
problems.
3) The equivalent model must be as
computationally economical as possible.
4) The equivalent model should be such
that the study system can be solved by
standard load flow and other programs.
This maybe mandatory feature in some
equivalencing methods where inter-state
data exchange is not available. Fig. 2 Thevenin’s equivalent circuit

5) Certain external buses may be retained in the

equivalent system for simulating generation Various software packages like E-TRAN, PSS/E,
outage, network enhancement, to adjust Dig SILENT Power Factory and Power World
power flow interchanges and other such simulator can be used for obtaining equivalent
studies. network. In this paper, Short circuit (Thevenin’s)
equivalents are calculated at the boundary buses
The equivalents attached to the internal system with the help of PSS/E (Power System Simulation
must not alter the present internal system states. for Engineering) software. The proposed method
In other words, the power flows from the is tested on the All-India network (external
equivalents into the boundary buses must exactly system) for which the solved load flow data and
match the tie line flows. Boundary matching sub-transient reactance data of the synchronous
is inherent in some equivalencing methods, machines was available. The Kerala network
where the equivalent can be derived from an (internal system) is separated from the rest of the
external system state (Mode 1). Otherwise, the network and the methodology adopted is given in
equivalent model being used must undergo subsequent section.
some adjustments, to contribute to the required
power flows at the boundary buses (Mode 2). 2.0 DETERMINATION OF EQUIVALENTS

In [1], the authors have compared the various When the external system is replaced by a
methods for determining the equivalents and have Thevenin’s equivalent, the equivalent must
concluded that the REI method is best suited for supply the same fault current as the external
static reduction of power systems. system would have supplied at the boundary bus.
This is illustrated in Fig. 3
According to Thevenin’s theorem [10], any black
box that contains impedances, linear voltage Once the short circuit current at the point of
and current sources, can be replaced by an equivalencing in known, then the Thevenin’s
equivalent circuit consisting of an equivalent equivalent impedance can be calculated from
voltage source connected in series with an Thevenin’s theorem. The short circuit MVA of
equivalent impedance as shown in Figure 2. the equivalent source can be calculated from (1)

In the Thevenin’s equivalent circuit, if the

terminals A and B are shorted, the current flowing (1)
from A to B is VTH/ZTH. This means that ZTH
266 The Journal of CPRI, Vol. 13, No. 2, June 2017

Where, FC is Fault Current in kilo amperes, 3.0 SIMULATION RESULTS FROM PSS/E
VL-L is the line-to-line RMS voltage in kilo Volts
at the boundary bus and MVASC is the short circuit All-India network system, which consists of
MVA and
6731 buses (50No. of 765 kV buses, 540No.
(2) of 400 kV buses, 1556 No. of 220 kV buses,
130 No. of 230 kV buses, 3078 No. of 132 kV
buses, 1015No. of 110 kV buses, 33No. of 100
ZTH is the Thevenin’s impedance of the
kV buses, 282 No. of 66 kV buses, 44No. of 33
equivalent in ohms.
kV buses and 3No. of 11 kV buses). 2027 No. of
synchronous machines, 13246 No. of AC lines
The power injections (or power absorption, and 17No. of 2 terminal DC lines. This network is
depending on power exchange at the boundary used to extract the Kerala state system, consisting
bus) are the same for the equivalent source as that of 38 buses (with 12 tie lines, 3No. of 230 kV
of the external system. buses, 30No. of 220 kV buses and 5No. of 400
kV buses) which is considered as the internal
The PSS/E software can be used to quickly and system for Network reduction concepts in PSS/E.
efficiently calculate the short circuit currents at The reduced network consists of seven boundary
a particular bus. The software calculates short buses where the Thevenin’s equivalent has been
circuit currents in accordance to IEC-60909 calculated. In order to find the equivalent at the
standard taking into account the sub-transient boundary buses, three phase-to-ground fault is
reactances of the synchronous machines. applied and the short circuit currents flowing into
Hence, it is very accurate method to determine the boundary buses are determined. The short
the equivalent of the external system at the circuit currents supplied by the external system
boundary buses using PSS/E and the equivalent source are compared at each
boundary bus and tabulated in Table I.

TABLE 1
COMPARISON OF SHORT CIRCUIT
CURRENTS BETWEEN EXTERNAL SYSTEM
& EQUIVALENT SOURCE

Short circuit
Short circuit
Boundary current
current
Bus supplied by
supplied by external
number equivalent source
system (A)
(A)

Bus 1 17887.39 17887.39

Bus 2 20927.13 20927.13
Bus 3 6045.44 6045.44
Bus 4 8089.73 8089.73
Bus 5 5301.40 5301.40
Bus 6 6312.00 6312.00
Bus 7 12470.51 12470.51

The comparison of bus voltages at the boundary

Fig. 3 Thevenin’s equivalent at boundary buses between the original and the reduced
buses supplying the same short
circuit current as the external
network is tabulated in Table II.
system
The Journal of CPRI, Vol. 13, No. 2, June 2017 267

TABLE 2 method for calculating the frequency dependent

AC equivalents using the concept of resonance
COMPARISON OF BUS VOLTAGE BETWEEN matching in parallel RLC systems.
ORIGINAL AND REDUCED NETWORK
Complex bus voltage Complex bus
Boundary voltage 4.0 CONCLUSIONS
in original network in reduced network
Bus number
(p.u.) (p.u.)
0 0 Network reductions are widely used in
Bus 1 1.00068 ∠-68.35 1.00069 ∠-68.35
0 0
power system applications. Constraints in
Bus 2 0.96036 ∠-75.56 0.96032 ∠-75.56 computation, data storage and communication
0 0
Bus 3 0.96986 ∠-74.07 0.96986 ∠-74.08 capacity necessitates the concept of network
Bus 4 0.96784 ∠-81.02
0
0.96783 ∠-81.02
0
reduction. All network reduction methods
Bus 5 1.00880 ∠-71.40
0
1.00880 ∠-71.41
0 utilize some form of aggregation of an external
0 0 network. The basic theory is to consider the
Bus 6 0.96941 ∠-81.53 0.96941 ∠-81.54
0 0
effect of external network as representative as
Bus 7 0.97420 ∠-77.85 0.97421 ∠-77.85 possible when reducing the network to carry
The comparison of the complex power flow out the studies in the desired frequency range.
through the boundary buses between the original This paper aims to compute the dynamic
and reduced network is tabulated in Table III. equivalents in a simple and accurate way for
transient stability studies by using the concept
TABLE 3 of Thevenin’s equivalents. The PSS/E software
COMPARISON OF COMPLEX POWER AT is used to calculate short circuit currents
BOUNDARY BUSES BETWEEN ORIGINAL & at the boundary buses, which leads to quick
REDUCED SYSTEM determination of Thevenin’s equivalent. At
Complex Power Complex Power boundary buses bus voltages, power flows and
Boundary Bus Injection Injection short circuit currents are compared between
Number In Original In Reduced the original networks and reduced network and
Network (Mva) Network (Mva) found that they are exactly matching. Hence
Bus 1 471.4 + j 22.2 471.4 + j 22.2 this method of equivalencing is accurate for
Bus 2 84.9 – j 73.3 84.9 – j 73.3 dynamic analysis.
Bus 3 22.0 – j 46.2 22.0 – j 46.0
Bus 4 423.2 + j134.0 423.3 + j 134 ACKNOWLEDGMENTS
Bus 5 446.8 + j 113.6 446.8 + j113.5
Bus 6 135.2 + j 21.0 135.2 + j 21.0
The authors would like to acknowledge the
Bus 7 375.6 + j 69.6 375.6 + j 69.6
support of K. S. Meera, HOD, Power Systems
As seen from Table II and Table III, the Thevenin’s Division and wish to thank the authorities of
equivalent method gives accurate results with CPRI for permitting to publish this paper.
respect to both bus voltage and power flowing
at the boundary buses. It is assumed that no RERENCES
boundary bus is part of a HVDC terminal, i.e.
there are no DC boundary buses. The Thevenin’s
method is a static method and used for [1] S. M. Ashraf, B. Rathore and S. Chakrabarti,
computation at fundamental frequencies. While "Performance analysis of static network
carrying out Electro Magnetic Transient studies reduction methods commonly used in
like Switching transient studies and Insulation power systems," 2014 Eighteenth National
co- ordination studies the frequency dependent Power Systems Conference (NPSC),
network equivalent has to be calculated in Guwahati, 2014, pp. 1-6.
order to include the contribution of the other [2] S. Djukic and A. Saric, “Dynamic model
frequencies. In [11], the authors have proposed a reduction: An overview of available
268 The Journal of CPRI, Vol. 13, No. 2, June 2017

techniques with application to power IEEE PICA Conference Proceedings,

systems,” Serbian Journal of Electrical Minneapolis, pp. 384390, July 1983.
Engineering, Vol. 9, No. 2, pp. 131–169,
[8] E. C. Housos, G. Irisarri, R. M. Porter
2012.
and A. M. Sasson, “Steady State Network
[3] A. Papaemmanouil and G. Andersson, Equivalents for Power System Planning
“On the Reduction of Large Power System Applications,” IEEE Trans. Power App.
Models for Power Market Simulations,” Syst., Vol. PAS-99, no. 6, Nov./Dec. 1980.
[Online]. Available: http://www.eeh.
[9] S. Deckmann, A. Pizzolante, A. Monticelli,
ee.ethz.ch/uploads/tx_ethpublications/
B. Stott and O. Alsac, "Studies on Power
PSCC2011_Antoni os .pdf.
System Load Flow Equivalencing," in IEEE
[4] J. B. Ward, "Equivalent circuits for power- Transactions on Power Apparatus and
flow studies," in Electrical Engineering, Systems, Vol. PAS-99, No. 6, pp. 2301-
Vol. 68, No. 9, pp. 794-794, Sept. 1949. 2310, Nov. 1980.
[5] A. Monticelli, S. Deckmann, A. Garcia and [10] W. Richter, "Applications of Thevenin's
B. Stott, "Real-Time External Equivalents theorem," in Electrical Engineering, Vol.
for Static Security Analysis," in IEEE 64, No. 3, pp. 103-105, March 1945.
Transactions on Power Apparatus and
[11] J. Sreedevi, K. G. V. S. B. Raju and K.
Systems, vol. PAS-98, No. 2, pp. 498-508,
S. Meera, “Computation of Equivalent
March 1979.
Network at HVDC Terminals,” National
[6] P. Dimo, Nodal Analysis of Power Systems, Conference On ZEN and TAO of Electrical
Abacus Press, Kent, England, 1975. And Electronics Engineering, Jan. 2014.
[7] W. F. Tinney and W. L. Powell, “The REI
Approach to Power Network Equivalents,”