Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol.

15(2), (2015), 1937-1944

TRANSIENT STABILITY SIMULATION OF

MULTI-MACHINE POWER SYSTEMS USING SIMULINK
Serdar EKINCI1 and Aysen DEMIROREN2

1,2
Electrical Engineering Department, Istanbul Technical University, Istanbul, Turkey
ekinciser@yahoo.com, demiroren@itu.edu.tr

Abstract: This paper describes a generalized dynamic model of multi-machine power systems for transient stability
analysis and its computer simulation using MATLAB/SIMULINK. The generalized model of the power systems can be
used for teaching the power system transient phenomena, as well as for research works particularly to improve
generator controllers with advanced technologies. Constructional details of various sub-models for the whole power
systems are given and their implementation in SIMULINK environment is outlined. The developed simulation model is
tested on 3-machine 9-bus power system and 10-machine 39-bus New England power system under different large
disturbances. For the studied cases, the critical clearing times (CCT) are calculated and the simulation results are
presented and discussed. Nonlinear time-domain simulation results obtained from several case studies validate the
effectiveness of proposed model for transient stability analysis. The proposed dynamic model has been employed to
support and develop power engineering education at both the undergraduate and graduate levels. Likewise, for
academic and educational use, all component sub-models are transparent and can simply be modified or extended.
Keywords: Transient stability, multi-machine power systems, MATLAB/SIMULINK, power system modeling, power
system simulation..

1. Introduction upon which the system is at the critically stable level. If the
actual fault clearing time in a system is below the CCT
The term "transient stability," in the power system level, the system will be stable [4].
stability studies, mostly denotes the capacity of Simulation programs for power system stability analysis
synchronous machines for remaining in synchronism can be divided into two classes of tools: commercial
for the short period of time subsequent to large software packages and education/research-aimed software
disturbances, including a fault on transmission packages. Various commercial software packages, such as
facilities, loss of a large load and sudden loss of Power System Simulator for Engineering (PSS/E), Power
generation. The system response to such disturbances System Simulator (Simpow), DigSilent, EuroStag,
contains generator rotor angles' large excursions power NEPLAN and PowerWorld, are on sale. These programs
flows, bus voltages and other system variables. It is allow users to access comprehensive component/system,
worth mentioning that, while steady-state stability is a models and computationally efficient algorithms for the
function only of operating conditions, transient stability analysis. Nevertheless, these programs are not appropriate
is a function of both the operating conditions and the for educational and research fields because they generally
disturbances [1]. This entangles the transient stability do not provide modification or incorporation of novel
analysis noticeably. System nonlinearities, in large component models and algorithms. In education and
disturbances, have a significant role. research fields, flexibility and potential for simple
To control transient stability or instability prototyping are much more crucial aspects than its
subsequent to a large disturbance, or a number of computational efficiency.
disturbances, time-domain simulation analysis is This paper discusses the use of SIMULINK software of
commonly used to solve the nonlinear equations set MATLAB in the dynamic modelling of multi-machine
explaining the dynamic behavior of system. In that power systems for transient stability simulation.
case, conclusion pertaining to stability or instability SIMULINK is a software package developed by
can be obtained from an inspection of the solution [2- MathWorks Inc., which is one of the most widely used
3]. Fault clearing time (FCT) and critical clearing time software in academia and industry for modeling, analysis
(CCT) are significant parameters so as to preserve and simulating dynamical systems. It can be used for
power systems transient stability. The FCT is the time modeling linear and nonlinear systems, either in continuous
upon which the fault is cleared following the fault time frame or sampled time frame of even a hybrid of the
occurrence, while the CCT is the fault clearing time two. It provides a very easy drag-drop type graphical user

Received on: 13.01.2015

Accepted on: 26.06.2015
1938
Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol. 15(2), (2015), 1937-1944

interface to build the models in block diagram form. It other words, it allows the display of a signal at any point
has many built-in block library components that you readily available; all one has to do is to add a Scope block
can use to model complex systems. If these built-in or, alternatively, an output port. What's more, giving a
models are not enough for you, SIMULINK allows you feedback signal is as easy as drawing a line. A parameter
to have user defined blocks as well. Over the last within any block can be regulated from MATLAB
decade, the topic of power system simulation in command line or through an m-file program. This is a
MATLAB/SIMULINK has been provided in [5-12]. predominantly helpful for multi-machine power system
The use of the MATLAB/SIMULINK for the transient stability study because the power system
enhancement of power system component can allow configurations change before, during and after a fault.
users to take full advantage of dealing with control Loading conditions and control measures can be employed
blocks and power system elements, corroborating new correspondingly.
component through comparison of the simulation
results for numerous events, and understanding of the 2.1. Differential Equations
basic concepts of power system modeling and
simulation. The differential equations of the machine and exciter for
In this paper, we have developed SIMULINK- the m machine, n bus system are expressed as follows [13]:
based generalized dynamic model and an efficient
approach so as to examine the transient stability d i
performance of practical power systems, with s i 1 i 1,2, , m (1)
dt
SIMULINK as a tool. The aim of the generalized
dynamic model is to provide an environment within
which students can quickly get started and provides d i Pmi Pei Di ( i 1)
i 1,2, , m (2)
enough modeling flexibility to allow modification or dt Mi Mi Mi
addition of new generator, load, and control system
models. Whereas ease of use is mostly important for dEqi Eqi ( xdi xdi )idi E fdi
coursework, the modeling flexibility is useful mostly i 1,2, , m (3)
for research applications. All component sub-models dt Tdoi Tdoi Tdoi
are transparent and can easily be modified or extended.
Emphasis has been given to keeping the component dE fdi E fdi K Ai
sub-models transparent and simple. All component Vrefi Vi i 1,2, , m (4)
dt TAi TAi
sub-models are designed for use by undergraduate and
graduate students in the learning of power system
transient stability and for rapid testing of research 2.2. Stator Algebraic Equations
ideas. The proposed model for transient stability
simulation has been applied to different examples of The stator algebraic equations describe the electrical
multi-machine power systems such as IEEE 9-bus and variables pertaining to the stator windings. The stator
IEEE 39-bus power systems. The nonlinear simulation algebraic equations are expressed as
results have been carried out to assess the effectiveness
of the developed transient stability simulation model Vi sin( i i ) xqi iqi 0 i 1,2, , m (5)
under various large disturbances. We hope that this
attempt will add some more practical information in Eqi Vi cos( ) xdi idi 0 i 1,2, , m (6)
i i
this important and unexhausted domain.
2.3. Network Equations
2. Multi-machine Power System Model
Constructed Using SIMULINK The network equations can be expressed in power-
balance or current-balance form. In this study, the current-
Regarding transient electromechanical phenomena balance form is employed and the loads are assumed to be
analysis of a power system, power flow algebraic of the constant impedance type. In power system with m
equations for the transmission network and for the generators, the nodal equation can be formulized as:
stator windings of the synchronous machines, together
with the differential equations for the rotor of the
I1
synchronous machines are commonly used. For that
reason, the power system mathematical model can be V1
formulated by a set of differential and algebraic Im
equations (DAEs) [13]. In this study, the flux-decay Y (7)
0
model with static exciter is employed to discuss the
synchronous machines. Vn
The whole system is given with regard to 0
SIMULINK blocks in three main single sub-models.
One of the most significant characteristics of a model j /2
in SIMULINK is its remarkable interactive ability. In Ii idi jiqi e i
i 1,2, , m (8)
 1939
Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol. 15(2), (2015), 1937-1944

where I1 , , I m are the complex injected generator 3. SIMULINK Models

currents at the generator buses. Assume that the
The complete multi-machine power system embodied
modified Ybus represented as Y be divided as
with regard to SIMULINK blocks in a single integral model
by using Eqs. (1)-(10) for transient stability study. The
m n-m
complete system consists of three main sub-models:
m Y1 Y2
Y = (9) differential equations sub-model, stator algebraic equations
n-m Y3 Y4 sub-model, and network equations sub-model. The
SIMULINK-based general sub-models for computing of
Inasmuch as there are no injections at buses differential equations and stator algebraic equations are
m 1, , n , we can leave them out in order to obtain modeled and are given for all generators in Figure 1 and
Figure 2 respectively. We can see in the complete sub-
model of Figure 1 that Scope 1 and Scope 2 display the
I1 V1 individual generator angles and the angular velocities of the
YR (10) all machines, respectively. In Figure 2, Scope 5 and Scope
Im Vm 6 monitor electrical power outputs and machine terminal
voltages responses respectively, for all generators. It is
worth mentioning that in Figure 1 and Figure 2; , ,
where YR = Y1 - Y2 Y4-1 Y3 is the desired reduced M , D , Eq , E fd , K A , TA , V , Vref , Pm and Pe are vector
matrix. The reduced matrices for every network signals having as many rows as the number of generators.
condition (before, during and after fault) are computed
on account of the power system under study.

Pmech
1 Integrator6 To Workspace deltas
Pm
1
2*pi*f 180/pi delta_angle
s
1
[Pe] 1./M Power Angles
s
Pe Speed Deviations
Integrator [del]
Sum10
[dw] Scope 1
Scope 2

D
Scope 3
Dampings

Efd_min_ max
Vref
1
Ka 1./Ta
s Integrator 5 Eqd
[V] Integrator 1 1
1./Tdod [Eqd]
s
Add10

Scope 4
xd-xdd Add 11

[Id]

Product 3

Figure 1. Differential equations sub-model for transient stability simulation

Figure 3 illustrates the complete block diagram of Likewise, prior to starting transient stability simulations,
network equations for all generators in SIMULINK it is mandatory to specify the initial conditions of a number
environment. The SUBSYSTEM in Figure 3 is of quantities for all machines, consequently, a power flow
expressed to calculate the value of electrical current calculation is performed by using the same associated
outputs for different generators and internal view of the MATLAB program, in which the pre-fault and the fault
SUBSYSTEM is given in Figure 4. In Figure 4 it clearing time are specified. The main SIMULINK-based
includes a block to specify network admittance sub-models are simply modified for multi-machine power
matrices required for numerous conditions of the power systems with different number of generators and also
transmission network: before, during and after a different network configurations. In addition, the
specified fault. These admittance matrices are generalized dynamic model of multi-machine power
calculated by using an associated MATLAB program systems in SIMULINK environment also facilitates the
prior to starting transient stability simulations of multi- choice of simulation parameters, including start and stop
machine power systems, and fault location can be times, types of solver, step sizes, tolerance and output
randomly specified. options.
1940
Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol. 15(2), (2015), 1937-1944

Product 1 Scope 5

[ Id]
[Iq]

xq [Pe]
Product 2
P_e
Product 12 [ Iq]

Sum 3
Complex to
Product 13
Magnitude-Angle Scope 6
[ Id] |u|

xdd u
j [V ]

[Eqd] Sum 5 pi/2 [ theta]

Sum 4
[del]

Figure 2. Stator algebraic equations sub-model for transient stability simulation

[ V] V

I
Re [ Id]
[theta] theta

Im [Iq]
Trigonometric SUBSYSTEM
Function Product 6 Complex to
Real- Imag
[del] sin

cos j

Trigonometric
Function 1

Figure 3. Network equations sub-model for transient stability simulation

Yaf Ydf
Clock
Yaf Ydf
V Matrix Matrix
1 |u| Multiply Multiply |u|
2 u u
theta Magnitude -Angle Switch Product of Magnitude-Angle
Product of
to Complex Elements to Complex
Elements

I Clock 1
1

I0
Switch initial value of I

Figure 4. Internal view of the SUBSYSTEM.

 1941
Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol. 15(2), (2015), 1937-1944

4. Results of Transient Stability numerical integration of the differential equations and step
Simulations and Discussions of integration is chosen as 0.005 s. All calculations were
done on a personal computer with 2.10 GHz Intel Core
In this section, we illustrate the DAE model Processor and 2.00 GB of RAM running
discussed in the previous section and to validate MATLAB/SIMULINK 7.11.0.
SIMULINK-based generalized model, the simulations
are carried out on a 3-machine 9-bus and 10-machine 4.1. Illustrative System example 1: Nine-bus three-
39-bus New England power systems. To assess the machine power system
effectiveness of the proposed dynamic model,
nonlinear time-domain simulation studies are In this example, we have considered the popular
implemented on different severe fault conditions. In Western System Coordinated Council (WSCC) 3-machine,
transient stability simulation of multi-machine power 9-bus power system shown in Figure 5. This is also the
systems, Runge-Kutta 4 technique is used for system appearing in [13-14] and widely used in literature.

Load C

G2 G3
7 8 9
2 3

5 6

Load A Load B

G1

Figure 5. WSCC 3-generator 9-bus test system

We can see in the complete sub-model of Figure 1 Figures 6-7 show the difference angles ( 21 2 1
that Scope 1 display the values of the 3-machine angles and 31 ) and the relative rotor angular speed
T 3 1
in vector and Scope 2 display the deviations ( 21 1 and 1 ), respectively,
1 2 3
2 31 3
values of the 3-machine angular speeds in vector for the system with FCT = 0.15 s. It is obvious from
T
1 2 3 . It is worth mentioning that the Figures 6-7 that the power system is stable in Case (a).
computation time for 3-machine 9-bus power system Figures 8-9 show the system response for FCT = 0.20 s.
3.28 s for 12 s of simulated real time. For time-domain From the results, we see that the system is unstable in Case
simulations, different three-phase faults have been (b) because all machines will lose synchronism.
applied to demonstrate the effectiveness of the
= 2-
proposed dynamic model as follows: 100
21

31
= 3-
1

The performance of the proposed dynamic model 80

under transient conditions is substantiated by applying
(degree)

a three-phase fault at t 1 s, on bus 9 at the end of line 60

8-9 is considered. The fault is then cleared by opening

31

40
the line 8-9. The CCT for this scenario was found to be
and

tcr 0.192 s by using a trial-and-error approach. The

21

20

system responses are given for different values of fault 0

clearing time (FCT).
Case (a): FCT < CCT, the fault occurs at 1 s and is -20
0 2 4 6 8 10 12
Time (sec)
cleared at 1.15 s (FCT = 0.15 s).
Case (b): FCT > CCT, the fault occurs at 1 s and is Figure 6. System response of 21 and
2 1 31 3 1
cleared at 1.20 s (FCT = 0.20 s).
for FCT = 0.15 s
1942
Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol. 15(2), (2015), 1937-1944

0.03 0.7
21
= 2
- 1
= -
21 2 1
= - = -
31 3 1 0.6 31 3 1
0.02

0.5
(pu)

(pu)
0.01
0.4
31

31
0 0.3
and

and
0.2
21

21
-0.01

0.1
-0.02
0

-0.03 -0.1
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time (sec) Time (sec)

Figure 7. System response of 21 2 1 and Figure 9. System response of 21 2 1 and 31 3 1

31 3 1 for FCT = 0.15 s for FCT = 0.20 s

1000
21
= 2- 1
4.2. Illustrative system example 2: Thirty nine-bus
800
31
= 3- 1 ten-machine power system
(degree)

600 In this part of the study, the 10-machine 39-bus New

England power system shown in Figure 10 is considered
31

400
further demonstrate the versatility of the suggested dynamic
and

200
model. This is also the system appearing in [15-16] and
21

widely used in the literature.

0

-200
0 2 4 6 8 10 12
Time (sec)

Figure 8. System response of 21 2 1 and

31 3 1 for FCT = 0.20 s

G8
G9
37
G10 38

30 28
25 26
29
27

2 24
18 G6
3 17
16 35
1
15
G1
21

4 14 22
39
5 19
12
6 23
7 13
11
20
8
31 10
33

G2 32 34 G4 36
9

G3 G5 G7

Figure 10. 10-machine, 39-bus New England test system

 1943
Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol. 15(2), (2015), 1937-1944

We can see in the complete sub-model of Figure 1 81

= 8
- 1

that Scope 1 display the values of the 10-machine 0.01 91

= 9
- 1

angles in vector
T
0.005

(pu)
1 2 3 4 5 6 7 8 9 10

and Scope 2 display the values of the 10-machine

91
angular speeds in vector 0

and
T

81
1 2 3 4 5 6 7 8 9 10 .
-0.005
The simulation was performed for 12 s and the
computation time for this large power system 13.56 s. -0.01
0 2 4 6 8 10 12
To assess the effectiveness of the proposed dynamic Time (sec)
model, nonlinear time-domain simulation studies are
implemented on different severe fault conditions. The Figure 12. System response of 81 8 1 and
following cases are taken into consideration: 91 9 1 for FCT = 0.10 s
A three-phase fault is applied on bus 29 (near
generator 9) at the end of line 29-26 at t 1 s. The 1000
= 8-
81 1
fault cleared without line tripping and the original 900
91
= 9- 1

system is restored upon the clearance of the fault. The 800

CCT for this scenario was found to be tcr 0.125 s by

(degree)
700

600
using a trial-and-error approach. The system responses
91
500
are given for Case (a) and Case (b). and
400
Case (a): FCT < CCT, the fault was initiated at 1 s
81

300
and cleared at 1.10 s (FCT = 0.10 s).
200
Case (b): FCT > CCT, the fault was initiated at 1 s
100
and cleared at 1.13 s (FCT = 0.13 s).
0
In this scenario, G8 and G9 are nearest generators 0 0.5 1 1.5 2
Time (sec)
2.5 3 3.5 4

to the fault location and therefore system responses are

given for only these machines. For Case (a) and Case Figure 13. System response of 81 8 1 and
(b), the power system angle responses of G8 and G9 91 9 1 for FCT = 0.13 s
with respect to G1 , ( 81 8 1 and 91 9 1 ),
0.14
are shown in Figure 11 and Figure 13, and the 81
= 8
- 1

0.12 91
= 9
- 1
variations of the relative speed deviation of G8 and G9
0.1
with respect to G1 , ( 81 8 1 and 91 9 1 ),
(pu)

0.08
are shown in Figure 12 and Figure 14. In Figures 11-
91

12, the FCT is set at 0.10 s while in Figures 13-14 the 0.06
and

FCT is set at 0.13 s. Figure 11 shows that the relative 0.04

81

rotor angles of the Generator 8 and Generator 9 0.02

oscillate and the system is said to be stable whereas
0
Figure 13 shows that the relative rotor angle of the
Generator 9 go out of step after a fault is cleared and -0.02
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec)
the systems becomes unstable.
130 Figure 14. System response of 81 8 1 and
81
= 8- 1
120 91
= 9- 1 91 9 1 for FCT = 0.13 s
110
(degree)

100 The effectiveness of the suggested dynamic model

90 in multi-machine power systems is verified through the
91

80
nonlinear simulation results. In addition, it can be
and

70
deduced from Figures 6-9 and Figures 11-14 that the
81

FCT setting is a significant factor to determine the

60
power systems transient stability. If the FCT is set at a
50
longer time than the CCT of the faulted line, the system
40
0 2 4 6 8 10 12 will be unstable; otherwise the system will be stable.
Time (sec)

Figure 11. System response of 81 8 1 and 5. Conclusion

91 9 1 for FCT = 0.10 s
SIMULINK is a powerful software package for the
study of dynamic and nonlinear systems. Using
SIMULINK, the simulation model can be built up
1944
Serdar EKINCI and Aysen DEMIROREN/ IU-JEEE Vol. 15(2), (2015), 1937-1944

systematically starting from simple sub-models. A self- [8] Demiroren A., Zeynelgil H.L., “Modelling and
sufficient generalized dynamic model for transient Simulation of synchronous machine transient analysis
stability simulation of multi-machine power systems using SIMULINK”, Int J Elec Eng Educ, Vol. 39, No.
has been given with full details, which can work as a 4, pp. 337-346, 2002.
[9] Cho Y.S., Park J., Jang G., “A novel tool for transient
basic structure for advanced and detailed studies. The
stability analysis of large-scale power systems: Its
models built in this way are easy to be understood by application to the KEPCO system”, Simul Model Pract
students and engineers. The proposed dynamic model Th, Vol. 15, No. 7, pp. 786-800, 2007.
has an open structure and all sub-models can be [10] Panda S., “Differential evolutionary algorithm for
modified or extended using various SIMULINK TCSC-based controller design”, Simul Model Pract Th,
constructs. The efficiency of the proposed Vol. 17, No. 10, pp. 1618-1634, 2009.
MATLAB/SIMULINK-based transient simulation [11] Gelen A., Yalcinoz T., “An educational software
model has been tested and demonstrated through the package for Thyristor Switched Reactive Power
transient stability simulations under various Compensators using Matlab/Simulink”, Simul Model
Pract Th, Vol. 18, No. 3, pp. 366-377, 2010.
disturbance conditions and different fault clearing [12] Nguimfack-Ndongmo J.D.D., Kenné G., Kuate-Fochie
times, by using the three and ten machine study R., Cheukem A., Fotsin H.B., Lamnabhi-Lagarrigue F.,
systems. Several cases have been also carried out to “A simplified nonlinear controller for transient stability
determine the effect of the clearing time of a fault on enhancement of multi-machine power systems using
the transient stability of multi-machine power systems. SSSC device”, Int J Electr Power Energy Syst, Vol. 54,
The proposed dynamic model shows to be a pp. 650-657, 2014.
powerful didactic and research tool, able to provide [13] Sauer P.W., Pai M.A., Power system Dynamics and
important information about various aspects of Stability. Prentice Hall, 1998.
transient stability phenomena. This educational model [14] Anderson P.M., Fouad A.A., Power System Control and
Stability. Ames, IA: Iowa State Univ. Press, 1977.
is also used for illustration purposes during lectures, as
[15] Pai M.A., Energy function analysis for power system
well as by students preparing personal assignments and stability. USA: Kluwer Academic publishers, 1989.
design projects. The authors believe that SIMULINK [16] Padiyar K.R., Power System Dynamics: Stability and
will soon become an indispensable tool for the teaching Control. Hyderabad, India: B.S. Publications, 2002.
and research of power systems.
Serdar Ekinci received B.Sc. degree
6. References from Istanbul Technical University
(ITU) Control Engineering
Department in 2007. He earned his
[1] Kundur P., Power system stability and control, New York,
M.Sc. degree in Electrical
USA: Mc Graw-Hill, 1994.
Engineering from the same university
[2] Bettiol A. L., Souza A., Todesco J. L., Tesch JR Jr.,
“Estimation of critical clearing times using neural in 2010. He is currently a Ph.D.
student in Istanbul Technical
networks”, In: IEEE Power Tech Conference
University. His areas of interest are
Proceedings; 23-26 June 2003; Bologna, Italy: IEEE. pp.
electrical power system, stability,
1-6.
control technology and the applications of heuristic
[3] Karami A., Esmaili S.Z., “Transient stability assessment
optimization to power system control.
of power systems described with detailed models using
neural networks”, Int J Electr Power Energy Syst, Vol.
45, No. 1, pp. 279-29, 2013. Aysen Demiroren received her B.Sc.,
[4] Ahmed M.A. Haidar, M.W. Mustafa, Faisal A.F. Ibrahim, M.Sc. and Ph.D. degrees from
Ibrahim A. Ahmed., “Transient stability evaluation of Istanbul Technical University in
electrical power system using generalized regression Electrical Engineering in 1982, 1985
neural networks”, Applied Soft Computing, Vol. 11, No. and 1994, respectively. Currently, she
4, pp. 3558-3570, 2011. is a full professor in Electrical
[5] Ong C., Dynamic Simulation of Electric Machinery Engineering Department of Istanbul
Using MATLAB/SIMULINK. Upper Saddle River, NJ: Technical University. Her research
Prentice Hall, 1998. interests are power system dynamics,
[6] Allen E., White N.L., Yoon Y., Chapman J., Ilic M., stability and control, power generation
“Interactive object-oriented simulation of interconnected and alternative power sources, and applications of AI
power systems using SIMULINK”, IEEE Trans. Educ, techniques in power systems.
Vol. 44, No. 1, pp. 87-94, 2001.
[7] Patel R., Bhatti T.S., Kothari D.P., “MATLAB/Simulink-
based transient stability analysis of a multi-machine
power system”, Int J Elec Eng Educ, Vol. 39, No. 4, pp.
320-336, 2002.