Andrzej BOBOŃ, Stefan PASZEK, Piotr PRUSKI, Tomasz KRASZEWSKI, Maria BOJARSKA

Silesian University of Technology

Computer-aided determining of parameters of generating unit

models based on measurement tests
Abstract. The paper presents a computer program for determining parameters of generating unit element models (synchronous generators and their
excitation systems, turbines with governors) based on transient waveforms recorded in the generating unit after introducing disturbances of the
steady working point.

Streszczenie. W artykule przedstawiono program komputerowy przeznaczony do wyznaczania parametrów modeli elementów zespołów
wytwórczych (generatorów synchronicznych i ich układów wzbudzenia, turbin z układami regulacji prędkości) na podstawie przebiegów
nieustalonych zarejestrowanych w zespole wytwórczym po wprowadzeniu zakłóceń ustalonego punktu pracy. (Komputerowo wspomagane
wyznaczanie parametrów modeli zespołów wytwórczych na podstawie przebiegów pomiarowych)

Keywords: Matlab/Simulink program, generating unit models, parameter estimation.

Słowa kluczowe: Program w środowisku Matlab/Simulink, modele zespołów wytwórczych, estymacja parametrów.

Introduction
Simulation investigations of the power system are an
important tool used for estimation of static and dynamic
properties as well as prediction of the results of disturban-
ces and system failures. Reliability of these investigations
depends mainly on the accuracy of used mathematical
models of particular system elements and their parameters.
The paper presents a computer program (PARZW) for
determining parameters of generating unit element models
(synchronous generators and their excitation systems,
turbines with governors), based on transient waveforms
recorded in the generating unit after introducing
disturbances of the steady working point.

U
Uref Excitation Power
system system
UPSS G
Fig. 2 Structural model of the generating unit in Matlab – Simulink
environment
T ref

 Synchronous generator models:

Turbine o Approximated models expressed by the standard
governor reactances and time constants of the subtransient, transient
and steady state (Xd, Xd’, Xd”, Tdo’, Tdo”, and so on [1, 2, 3,
4]). In these models the stator transformation voltages were
Fig. 1. Elements of the generating unit
neglected, while saturation of magnetic cores was taken
Fig. 1 shows the elements of a typical generating unit. into account. There were included one equivalent damping
One of them is a synchronous generator (G) connected with circuit in d axis and two equivalent damping circuits in q
a power system and driven by a turbine (T). The generator axis (turbogenerator) or one equivalent damping circuit in q
excitation system is controlled by a signal dependent on the axis (hydrogenerator). Four kinds of the generator models
generator voltage (U), the reference voltage of the voltage expressed by the standard parameters: GENROU and
regulator (Uref) and the output signal of the power system GENROE (turbogenerators) as well as GENSAE and
stabilizer PSS (UPSS). The turbine is equipped with a GENSAL (hydrogenerators) differing in the way of
governor controlled by a signal dependent on the generator approximation of the magnetic core magnetization
angular () and reference (ref) speed. characteristic [3] are implemented in the program. Fig. 3
A general model of the generating unit was developed in presents the GENROU turbogenerator model implemented
the framework of the research realized. Due to blocks of in Simulink.
“Configurable Subsystems” used in this model, it is o Models expressed by resistances and inductances of
convenient to construct a concrete model of the generating the generator electric circuits (R, L, Lmd, and so on [1, 2]).
unit (Fig. 2) selecting the models of its particular elements. These models take into account the transformation and
It is possible to neglect the influence of the excitation rotation voltages in the stator and different number of
system, turbine and PSS in the generating unit model. damping circuits in the rotor. Fig. 4 shows the equivalent
circuits of the generator model including two damping
Mathematical models of generating unit elements circuits in d axis and three damping circuits in q axis, the
The following mathematical models of generating unit so-called model of type (3,3). In these circuits index d and q
elements were implemented in the program: denotes the stator circuit parameters in d and q axis, index

PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 87 NR 5/2011 17

D and Q denotes the rotor damping circuit quantities, while
U ref
index f is used for the excitation system quantities.
o Models of the excitation systems with voltage regulators U R max
recommended by the IEEE Committee [4, 5]: IEEET1, U 1 sTi  1 U rov
K1
EXAC1, EXAC2, EXDC2, EXPIC1, EXST1, EXST2, sT1  1 sTi
IEEET2, IEEET3 as well as models of static and U R min
electromachine (Fig. 5) excitation systems working in the
Polish Power System [6]. K 2  sT 4 1
 Models of the IEEEG1 steam (Fig. 6), HYGOV water and sT 3  1 sT 2  1
GAST gas turbines and their governors [4, 6, 7].

K 4  U R max
U rov sT5  1 1 I fe
K3
sT5 sT6  1

K 4 U R min

c) I fe 1 E fd
K5
sT 7  1

Fex _ s

I fd
KC

Fig. 5. Structural diagram of the electromachine excitation system

installed in Power Plant Rybnik; a) voltage regulator, b) exciter and
additional regulator of the excitation system, c) system forming the
excitation voltage

P0
Uo
 sT9  1 1 1 Pd
K6
sT8  1 T10 s
Uc

Fig. 3. Model of the synchronous generator GENROU: a) in d axis,

Pm
b) in q axis, c) including saturation of magnetic cores

K7 K8 K9

Pd 1 1 1
sT11  1 sT12  1 sT13  1

Fig. 6 Model of the IEEEG1 steam turbine with a governor; a)

governor, b) turbine

According to the operating conditions of the generating

unit, the equation of mechanical motion or equations of the
equivalent power system are taken into account in the
generating unit model.

Measurement tests
There were selected such measurement tests that they
should be safe and easy to realise for a working or being
launched generating unit. The exemplary test signals are
Fig. 4. Equivalent circuits in d and q axis of the synchronous small disturbances such as a step change or pseudorandom
generator of type (3,3) expressed by resistances and inductances signal (PRBS) of the voltage regulator reference voltage
of the generator electric circuits introduced in the synchronous generator excitation system or
load rejection of the generator operating under the
determined conditions.

18 PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 87 NR 5/2011

 Appropriate selection of the test and measured signals  measurement data acquisition module containing proce-
makes it possible to separate the tested element of the dures for preprocessing and filtering the signals,
generating unit by reducing or eliminating the influence of the  library of mathematical models of the generating unit
other unit elements, and thus to simplify estimation of its elements implemented in Simulink,
parameters. For instance, during the no-load or load rejection
 parameter estimation module of the generating unit
test of the generator, the influence of the external power
models,
system (PS) is eliminated.
The measured waveforms, being the basis for parameter  module for sensitivity analysis of dynamic waveforms to
estimation, have to be processed numerically, which is aimed parameter changes of particular models,
at preliminary processing and filtration of disturbances [8].  simulation module of generating unit disturbance states,
 database containing the models of generating units and
Methods for parameter estimation their parameters.
Parameter estimation of the generating unit particular Fig. 8 shows the functional diagram of program PARZW.
models is performed with use of the least square method.
The mean square error was defined as the difference at A generating unit configuration
particular time instants between the waveforms obtained and the choice of the object model
from measurements and those from computations when (generator, excitation system, turbine)
using the mathematical models expressed by the
parameters searched [9, 10]:
The choice of input signals

 
n 2 (measured transient waveforms)
(1)  w  P    Wi( m )  Wi( s )  P 
i 1
Numerical data preprocessing
where: index m denotes the measured waveform of quantity (filtering, detrending, smoothing)
W, index s denotes the simulated waveform of quantity W
calculated for the parameter vector P.
In case of taking into account several waveforms, the Parameter estimation Sensitivity analysis
mean square error is a sum of the errors defined for of transient waveforms
particular waveforms when taking into consideration The choice of parameters to changes in
appropriately chosen weight functions wk: to be estimated, estimating parameter values
initial and bound values
(2)   P    wk  wk  P 
k Minimization of an Simulation of transient
objective function states
The following optimization algorithms are used for
minimization of the mean square error [6, 11]:
Presentation of resulting
 Gradient algorithm implemented in the package parameters. Comparison
Optimization Toolbox of Matlab program. Gradient Database containing
of measured and generating unit models
algorithms belong to local optimization algorithms, that simulated waveforms
is why the initial set of parameters should be selected and their parameters
carefully in order to obtain correct results.
 Genetic algorithm which finds the global minimum of the Fig. 8. Functional diagram of program PARZW
objective function. Examples of parameter estimation of the generating
 Hybrid algorithm consisting of genetic and gradient unit
algorithms. The hybrid algorithm ensures the highest Parameter estimation of the synchronous generator model
accuracy of approximation. of type (3,3)
Estimation of the model of type (3,3) parameters
Fig. 7 shows the general diagram of the algorithm for
(resistances and inductances of the equivalent circuits
parameter estimation of the generating unit elements.
shown in Fig. 4) was performed for the 589 MVA
synchronous generator. The dynamic waveforms in the
machine caused by introducing a small disturbance in the
form of -5% change in the field voltage were used for
 P estimation. Before introducing the disturbance, the
generator worked in the steady state of the rated load. In
the program window presented in Fig. 9 there are shown
the stator current waveforms in d axis calculated for the
initial and final, estimated parameter values as well as the
P true waveform. At the window bottom part there are given
the initial values of the parameters and their upper and
lower constraints taken into account by the gradient
optimization algorithm. At the bottom row, distinguished with
Fig. 7. Algorithm for parameter estimation of the generating unit colour, there is given the list of the model final, estimated
parameters. The program enables the choice of estimated
Program structure parameters when assuming that the others are known and
The program was developed in Matlab/Simulink the comparison of the waveforms of selected quantities.
environment. Its structure includes:
 menu controlled graphical user interface,

PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 87 NR 5/2011 19

Fig. 9. Results of parameter estimation of the type (3,3) generator Fig. 11. Generator exciting current (measured, approximated) for a
for the step change in field voltage by -5% step change of the voltage regulator reference voltage by +5%, the
approximating waveforms and results of parameter estimation in d
Parameter estimation in d axis of the GENROU generator axis of the GENROU generator model
model

For parameter estimation of the TWW-200-2A generator

GENROU model there were used the exemplary (recorded
during launching test of the generating block in Power Plant
Rybnik [6, 8]), dynamic waveforms of the field voltage (input
signal – Fig. 10), exciting current (Fig. 11) and the
generator armature voltage (Fig. 12) for a step change of
the voltage regulator reference voltage of the no-load
generator by +5%. During measurements the generator was
operating with the so-called automatic regulation in the
excitation system.

Fig. 12. Generator terminal voltage (measured, approximated) for a

step change of the voltage regulator reference voltage by +5%, the
approximating waveforms and results of parameter estimation in d
axis of the GENROU generator model

Fig. 10. Program window with the measured waveform of the

generator field voltage (input signal for estimation) for a step
change of the voltage regulator reference voltage by +5%

Parameter estimation of the excitation system

For calculation of the electromachine excitation system
parameters of Fig. 5 there were used exemplary dynamic
waveforms of selected quantities measured on the unit in
Power Plant Rybnik for a step change of the voltage
regulator reference voltage of the no-load generator by
+5%. In order to improve the calculation efficiency, the
parameter estimation procedure was divided into three
substages related to the excitation system submodels of
Fig. 5 for which separate calculations were carried out [6]. Fig.13. Field voltage (measured, approximated) for a step change
of the voltage regulator reference voltage by +5%, the
Fig. 13 shows the waveforms of the field voltage as well as
approximating waveforms and results of parameter estimation of
the estimation results of the system forming the excitation the system forming the excitation voltage
voltage.
Parameter estimation of the turbine
For parameter estimation of the IEEEG1 steam turbine
model there were used the exemplary dynamic waveforms
for the generator load rejection in q axis (P0 = 0.2 and

20 PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 87 NR 5/2011

Q0 = -0.0723 in relative units were assumed to be the initial REFERENCES
values of the generator load). Parameter estimation of the [1] K u n d u r P . : Power System Stability and Control, M c G r a w -
turbine with its governor was performed by dividing the Hill, Inc., 1994.
computational process into two substages [6]. In the model [2] A n d e r s o n P . M . , F o u a d A . A .: Power system control and
stability, John Wiley & Sons, Inc., 2003.
there were distinguished a governor model and a turbine
[3] d e M e l l o F . P . , H a n n e t t L . H .: Determination of
model, for which calculations were carried out separately. Fig. Synchronous Machine electrical characteristics by test. IEEE
14 shows the exemplary dynamic waveforms of the power Transaction on Power Apparatus and Systems, Vol.102, No.12,
transmitted by the steam flow at the turbine inlet and the Dec 1983, pp. 3810-3815.
parameter estimation results of the steam turbine model. [4] PSS/E-29 Program Application Guide, Siemens Power
Technologies Inc., 2002.
[5] IEEE Guide for Identification, Testing, and Evaluation of the
Dynamic Performance of Excitation Control Systems, IEEE Std
421.2-1990.
[6] M a j k a Ł., P a s z e k S.: Wykorzystanie algorytmu
hybrydowego do estymacji parametrów modelu
matematycznego zespołu wytwórczego, Przegląd
Elektrotechniczny, 86 (2010), nr 8, 70-76.
[7] IEEE Committee Report: Dynamic Models for Steam and Hydro
Turbines in Power System Studies, IEEE Transactions on
Power Apparatus and Systems, Vol. PAS-92, November.
(1973), No. 6, 1904-1915.
[8] M a j k a Ł., P a s z e k S.: Turbogenerator and electromachine
excitation system parameter identification on the ground of
measuring tests made in Power Plant Rybnik. 30-th
International Conference on Fundamentals of Electrotechnics
and Circuit Theory, SPETO’2007, Ustroń May 2007, pp. 179-
180.
[9] F e l t e s J . W . , O r e r o S . , F a r d a n e s h B . , U z u n o v i c
E . , Z e l i n g h e r S . , A b i - S a m r a N . , Deriving Model
Fig. 14. Power transmitted by the steam flow at the turbine inlet Parameters from Field Test Measurements, IEEE Computer
(measured, approximated) for load rejection in q axis, the Applications in Power, (Oct 2002), 30-36.
approximating waveforms and parameter estimation results of the [10] P a s z e k S . , B o b o n A . , K u d l a J . , B i a l e k J . , A b i -
steam turbine model S a m r a N ., Parameter estimation of the mathematical model
of a generator, excitation system and turbine, Przegląd
Summary Elektrotechniczny, 81 (2005), nr 11, 7-12.
[11] N o c o ń A . , P a s z e k S .: Polioptymalizacja regulatorów
The computer program PARZW developed in napięcia zespołów prądotwórczych z generatorami
Matlab/Simulink environment (being improved continuously) synchronicznymi, Wydawnictwo Politechniki Śląskiej, Gliwice,
is a convenient tool for calculating parameters of different (2008).
models of generating unit elements (synchronous
generators, excitation systems, turbines with governors)
working in the power system. The basis for estimation is the This work was partly financed by the Polish Ministry of Science
analysis of dynamic waveforms obtained from appropriate and Higher Education from means of budget on science in
measurement tests carried out on generating units in power years 2009-2012 as the research project N N511 352137.
plants. The program is equipped with modules of: measured
signal filtration, generating unit operation simulation and
sensitivity analysis of waveforms to parameter changes as Authors: dr inż. Andrzej Boboń, dr hab. inż. Stefan Paszek prof.
well as enables entering the results in the database of the Pol. Śl., mgr inż. Piotr Pruski, mgr inż. Tomasz Kraszewski, Instytut
dynamic parameters of the Polish and European PS. Elektrotechniki i Informatyki, ul. Akademicka 10, 44-100 Gliwice, dr
The program PARZW should be used widely in the inż. Maria Bojarska docent Pol. Śl., Instytut Metrologii, Elektroniki i
future for analysis of PS operation, since the knowledge of Automatyki, E-mail: [Link]@[Link], [Link]
the actual parameters of generating units can enable @[Link], [Link]@[Link], [Link]@[Link],
carrying out reliable calculations of static and dynamic [Link]@[Link].
states, and in consequence, avoiding occurrence of system
failures.

PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 87 NR 5/2011 21