IEEE AFRICON 2004 199

Assessment of Commercially Available

K. K. Kaberere, K. A. Folly, and A. I. Petroianu

Software Tools for Transient Stability:

Experience Gained in an Academic
Environment

paper are PSS/E (PTI), DIgSILENT (PowerFactory), and

Abstract There are many power system simulation software PSCAD (Manitoba HVDC Research Centre).
tools for transient stability analysis that are commercially As a basis for the software assessment, the following
available today. The choice of the most adequate tool for a features are investigated.
specific type of stability analysis is a difficult task, and is (i) Power system component models available in each
influenced by economical reasons. It is imperative that a
comparative study of various tools be carried out to help users tool. These include the synchronous generator,
make informed decisions. This paper evaluates DIgSILENT, generator saturation, transmission line representation,
PSS/E and PSCAD, which are commercial software tools used and external network model.
for digital simulation of power systems. As criteria for software (ii) Software flexibility which is based on how much the
assessment, the following features are investigated: (i) modelling user is able to interact with the program during
capabilities synchronous generator, generator saturation, simulation, and method of data input.
transmission line representation, and external network, (ii)
The paper is organized as follows. In section II, we discuss
flexibility allowed to the user.
A well known benchmark power system model from literature
the following modelling aspects: - synchronous generator,
is used to validate the simulation results of the commercial tools. generator magnetic saturation, transmission line
representation, and external network. In section III, we
Index Terms Commercial software tools, excitation system, discuss the flexibility of the three software tools. In section
power system stabiliser, transient stability IV, we describe the case study, single machine infinite bus
(SMIB) system [1] used to validate the simulation results and
I. INTRODUCTION discuss the simulation results. In section V, we present the
Power systems are increasingly operating under stressed conclusions drawn from the study.
conditions with very narrow stability margins. This poses a
threat to the successful operation of the network as the system II.MODELLING OF COMPONENTS
could become unstable even for minor disturbances. An
unstable condition may cause the protection system to isolate A. Synchronous generator
the affected areas of the network. Such action by the In power system dynamic studies, the synchronous
protection system may cause instability in other parts of the generator is commonly represented using the dq-axis. The
network (e.g., due to overloading) which may eventually sixth order model has been found to adequately represent the
culminate in total system blackout. synchronous generator in stability studies and thus, it is the
The advancement of computer technology has permitted most extensively used model for round rotor generators [2],
the development of powerful power system analysis tools. One [3]. The equivalent circuit used to represent this model is
of the important functions of these tools is transient stability shown in Fig. 1. It has six electrical circuits namely stator d-
analysis. Digital simulation of power system response to and q-axis circuits (Ll), the field circuit (Rfd, Lfd), one d-axis
changes in operating conditions plays a very important role in damper winding (R1d, L1d) and two q-axis damper windings
system analysis, planning, and operation. (R1q, L1q, R2q, L2q). Lad and Laq represent the mutual inductance
The power system analysis software tools are expensive between the stator and the rotor circuits. The fifth order
and moreover, the mastering of a new tool is a time generator model, which is similar to the sixth order model but
consuming exercise. Hence a lot of users cannot afford to with only one damper winding on the q-axis, is used to
invest (time and money) in a number of tools and carry out a represent salient pole machines. These two models are
comparison to determine which tool is best suited for a available in all the three software tools under study.
particular application. The motivation of this study is the
need to bridge this gap.
The industrial-grade software tools investigated in this
Manuscript received March 2, 2004.

K. Kaberere is with the Department of Electrical Engineering, University of

Cape Town, Private Bag, Rondebosch, 7701, South Africa. (e-mail: keren@
powerelec.ee.uct.ac.za).
K. A. Folly is with the Department of Electrical Engineering, University of Fig. 1: Sixth order generator model
Cape Town (e-mail: [email protected]).
A. Petroianu is with the Department of Electrical Engineering, University of
Cape Town (e-mail: [email protected]). PSS/E has several synchronous generator models of
 IEEE AFRICON 2004 200

varying degrees of complexity [4]. Fig. 2: Typical generator open-circuit saturation curve
Both DIgSILENT and PSCAD have fifth and sixth order
C. Transmission line representation
models, [5], [6].
In DIgSILENT, the transmission line is represented by the
B. Generator Magnetic saturation equivalent model. In both PSS/E and PSCAD, the user may
In both PSS/E and DIgSILENT, the user gives two choose between equivalent model and distributed parameter
saturation parameters S1.0 and S1.2 corresponding to 1.0 and (travelling wave) model depending on the length of the line.
1.2 pu terminal voltage (flux linkage) respectively [4], [5].
These parameters are determined from the open circuit
saturation curve of the generator [1], [4], [5], [7]. They are D. External network (Infinite bus) model
computed using (1) (3) as given below. It should be An infinite bus in power systems is defined as a constant
mentioned that these equations do not apply if the generator frequency and constant voltage (both magnitude and angle)
is fully saturated. source [1], [7]. The infinite bus is modelled differently in all
B sat at T 1 three software tools.
I Asat e In PSS/E, the classical generator model is used to represent
(1) the infinite bus. For the classical generator to be a constant
I A1.0 I B1.0 voltage and constant frequency source, the inertia constant of
S1.0 the machine is set to zero as required in [4]. In this model,
I B1.0 the d-axis transient reactance is equal to the synchronous
(2) reactance (Xd = Xd). In transient calculations, this reactance
I A1.2 I B1.2 represents the short circuit reactance at the infinite bus.
S1.2 In DIgSILENT, the infinite bus is modelled as an external
I B1.2 grid.
(3) In PSCAD, the infinite bus is modelled as a constant
where voltage source with an internal reactance. In transient
at - Flux linkage at a point on the non-linear curve calculations, the source reactance is equal to the short circuit
I - Flux linkage drop due to saturation reactance of the external network.
T1 - Linear characteristic threshold flux linkage In all the three software tools, it is recommended that the
Asat, Bsat - Constants short circuit reactance at the infinite bus be low to ensure
IA1.0, IB1.0, IA1.2, IB1.2 - Field currents electrical stiffness.
A typical saturation curve is shown in Fig. 2.
In PSCAD the user gives data for up to ten points on the III. SOFTWARE FLEXIBILITY
non-linear open circuit saturation curve. The first point must All the three programs have graphical user interface. In the
be (0, 0) and the second point must lie on the linear part of graphics environment, the user can draw the system one-line
the curve (i.e., line OP on Fig. 2). The other points on the diagram (or three phase in PSCAD) and populate the system
non-linear part of the curve are determined using (1). with data using the pop-up windows. The one distinct
advantage of the graphical user interface is that the user is
able to quickly modify the network topology and input data.
In PSS/E, the user can introduce data in text format. The
program reads load flow data separately from the dynamic
data. These are given as two distinct raw data files.
In PSCAD, the user has online access to input variables,
i.e., the user can change parameters during a simulation.
Switches, push buttons, sliders, and ammeters are some of the
control and meter interfaces available in the program.
In both PSS/E and DIgSILENT, the user is able to run a
load flow before the dynamic simulation. PSCAD does not
perform load flow calculations and hence the user should
have pre-calculated initial conditions for all elements in the
power system. The user starts a simulation without a
disturbance and brings the system to steady state operation
after which the disturbance can be applied.

IV. CASE STUDY SINGLE MACHINE INFINITE BUS (SMIB)

A. System model
We used the SMIB system [1] in Fig. 3 to run the
simulations. System data is the same for the three software
tools and are given in appendix A.
 IEEE AFRICON 2004 201

In all the three software tools, we used the sixth order PSCAD. For DIgSILENT, we used two values of short circuit
generator model for G1. We computed the machine saturation reactance XSC = 0.0073 pu and XSC = 0.073 pu
parameters S1.0 and S1.2, and the points on the saturation curve We observed that the results obtained using PSS/E and
using (1) (3). We used a linear flux linkage threshold T1 = PSCAD in all cases were similar and compared well with the
0.8 pu and air-gap line slope of 1.66. results in [1].
We used a short circuit reactance at the infinite bus of However, the DIgSILENT results, with infinite bus Xsc =
0.0073 pu on 2220 MVA, 24kV base referred to the low 0.0073 pu, for both Case I and Case II deviated significantly
voltage side. from results of the other two software tools. The DIgSILENT
results exhibited better damping than with the other tools.
B. Excitation system When an AVR was used to control the excitation, instability
To investigate the transient stability, we applied a three- was observed after the third swing in DIgSILENT whereas
phase fault on transmission line L2 at point F shown on Fig. this behaviour was observed after the second swing in
3. The pre-fault system condition in pu on 2220 MVA, 24 kV PSCAD and PSS/E. All three software tools agreed well when
base is: both the AVR and PSS were modelled. This may be attributed
P = 0.9; Q = 0.436 (overexcited) to the presence of the PSS, which damps the oscillations.
VT = 1.028.34 VB = 0.900810 To match the DIgSILENT results with those of the other
The fault is cleared by simultaneously opening the circuit two software tools, we increased XSC to 0.073 pu. Figures 11-
breakers on both ends of the line after 0.07s, which is the 16 show the comparison of DIgSILENT results with Xsc =
critical fault clearing time [1]. We simulated this system for 0.0073 pu and Xsc = 0.073 pu. It can be observed that the
three different excitation control scenarios. results obtained with Xsc = 0.073 pu closely matched those
Case I: Constant excitation obtained using PSCAD and PSS/E for all the three cases of
Case II: Excitation control with automatic voltage excitation control as observed in figures 5-10. This suggested
regulator (AVR) only that the modelling of the external grid in DIgSILENT does
Case III: Excitation control with AVR and power system not only depend on the short circuit reactance as in the other
stabiliser (PSS). two software tools, but also depends on other parameters
The block diagram of the generator excitation system is which we were not able to identify in this study.
shown in Fig. 4. This is an IEEE type ST1A excitation model The DIgSILENT results for Case I, with XSC = 0.073 pu
[1] and is the model used in PSS/E and PSCAD. We used exhibited better damping than with XSC = 0.0073 pu. When
VCO type 5 model in DIgSILENT. the excitation was controlled using an AVR only, DIgSILENT
results for the two values of XSC did not differ much during
the first swing. This can be associated with the action of an
AVR, which improves the recovery of the terminal voltage
after a disturbance.

Fig. 3: SMIB network representation

Fig. 4: Thyristor excitation system with AVR and PSS

C. Time domain simulation results
The aim of this study was to assess the three software tools.
The simulation results in [1] were used as a basis for the Fig. 5: Generator terminal voltage, constant excitation
validation of our results. It should be mentioned that some of
the data required to reproduce the case was not available e.g.,
modelling and data of the infinite bus was not given, and
hence matching the absolute values of the results was
difficult.
The simulation results for the machine load angle and
terminal voltage using the three tools are shown in figures 5-
10. All the figures show results for PSS/E, DIgSILENT, and
 IEEE AFRICON 2004 202

Fig. 6: Generator load angle, constant excitation Fig. 9: Generator terminal voltage, AVR and PSS

Fig. 7: Generator terminal voltage, AVR only Fig. 10: Generator load angle, AVR and PSS

Fig. 11: DIgSILENT: Generator terminal voltage for infinite bus

Fig. 8: Generator load angle, AVR only XSC = 0.0073 and XSC = 0.073 constant excitation
 IEEE AFRICON 2004 203

Fig. 12: DIgSILENT: Generator load angle for infinite bus XSC = Fig. 15: DIgSILENT: Generator terminal voltage for infinite bus
0.0073 and XSC = 0.073 constant excitation XSC = 0.0073 and XSC = 0.073 AVR and PSS

Fig. 13: DIgSILENT: Generator terminal voltage for infinite bus Fig. 16: DIgSILENT: Generator load angle for infinite bus XSC =
XSC = 0.0073 and XSC = 0.073 AVR only 0.0073 and XSC = 0.073 AVR and PSS

V. CONCLUSION
The results obtained using PSS/E and PSCAD with infinite
bus XSC = 0.0073 pu agree reasonably well. They also
compare well with the results in [1]. DIgSILENT results only
match the other two tools when XSC = 0.073 pu. There is need
for further investigation into the modelling of the external
network particularly in DIgSILENT where the bus short
circuit reactance has significant effect on the simulation
results.
In future, the authors will extend the scope and the depth
of the study by comparing results with other software tools
and investigating other aspects of stability like small signal
stability.

APPENDIX A
Fig. 14: DIgSILENT: Generator load angle for infinite bus XSC = A1. SMIB system data
0.0073 and XSC = 0.073 AVR only
The network reactances are in per unit on 2220 MVA,
24kV base referred to the low voltage side of the step-up
transformer. We have neglected all transmission lines and
 IEEE AFRICON 2004 204

transformer resistances, and charging capacitance.

Transformer reactance = j0.15 pu
L1 reactance = j0.5 pu
L2 reactance = j0.93 pu

Generator rating: 2220 MVA, 24 kV

Xd = 1.81 Xq = 1.76 Xl = 0.15
Xd = 0.30 Xq =0.65 Ra = 0.003
Xd = 0.23 Xq = 0.25 H = 3.5
Tdo = 8.0 s Tqo = 1.0 s KD = 0
Tdo = 0.03 s Tqo = 0.07 s
d- and q-axes have similar saturation characteristics with
Asat = 0.031 Bsat = 6.93 T1 = 0.8 T2 =

A2. Points on the saturation curve

I (pu) (pu) I (pu) (pu) I (pu) (pu)
0 0 0.8134 1.1 3.313 1.5
0.482 0.8 1.024 1.2 5.782 1.6
0.58 0.9 1.386 1.3
0.6778 1.0 2.048 1.4

A3. Excitation control

Excitation system parameters
KA = 200 TR = 0.015 s EF max = 7.0
KSTAB = 9.5 TW = 1.41 s EF min = -6.4
Vs max = 0.2 Vs min = -0.2
T1 = 0.154 s T2 = 0.033 s

REFERENCES
[1] P. Kundur, Power System Stability and Control, McGraw-Hill, 1994.
[2] IEEE Guide for Synchronous Generator Modelling Practices in Stability
Analysis, IEEE Std 1110-1991.
[3] K. R. Padiyar, Power System Dynamics: Stability and Control, John Wiley
& Sons, 1996.
[4] PTI, PSS/ETM 29 Program Application Guide, October 2002.
[5] DIgSILENT Manuals, DIgSILENT PowerFactoryVersion 12
[6] EMTDC Transient Analysis for PSCAD Power System Simulation, Users
Guide
[7] P. M. Anderson, A. A. Fouad, Power System Control and Stability, IEEE
Press, 2nd edition, 2003
[8] Paul Forsyth et al., Comparison of Transient Stability Analysis and Large-
Scale Real Time Digital Simulation, Accepted for presentation at PPT
2001 IEEE Porto Power Tech Conference, Porto, Portugal, 10 th 13th
September, Paper 0-7803-7139-9/01.
[9] J. Persson, J. G. Slootweg, L. Rouco, L. S der, and W. L. Kling, A
Comparison of Eigenvalues Obtained with Two Dynamic Simulation
Software Packages, Accepted for presentation at 2003 IEEE Bologna
Power Tech Conference, 23rd 26th June, Bologna, Italy, Paper 0-7803-
7967-5/03.
[10] Emad S. Ibrahim. (2002, December), A Comparative Study of PC Based
Software Packages for Power Engineering Education and Research,
[Online] International Journal of Electrical Power and Energy Systems,
Volume 24 (Issue 10), pp. 799 805. Available:
http://www.sciencedirect.com/science/journal/01420615.
[11] J. G. Slootweg, J. Persson, A. M. van Voorden, G. C. Paap ,W. L. Kling, A
Study of the Eigenvalue Analysis Capabilities of Power System Dynamics
Simulation Software, 14th PSCC, Sevilla, 24th 28th June, 2002.
[12] Juan A. Martinez, Jacinto Martin-Arnedo, Expanding Capabilities of
EMTP Like Tools: From Analysis to Design, IEEE Transactions on
Power Delivery, vol. 18, No. 4, October 2003, pp.1569-1571.