Testing & Modeling of Generator Controls
G.R. Bérubé, L.M. Hajagos
Kestrel Power Engineering
Toronto, Ontario, Canada
INTRODUCTION
NERC planning standards approved in September 1997
have reiterated the need for accurate and timely model data
for the transmission system and connected equipment. The
data requirement includes steady-state and dynamic models,
and covers control system small-signal performance and
limits.
GENERATION EQUIPMENT MODELLING DATA
REQUIREMENTS
The subject of this paper is the generating unit and its
controls. The applicable NERC requirements from Sections
IIA and IIB of the Planning Standards are summarized
below.
Steady-State Data
• Generator minimum and maximum ratings (net real and
reactive power), regulated bus and voltage set-point
• Station service and auxiliary loads
Dynamic Data
• Unit-specific dynamic data for generators, excitation
systems, voltage regulators, turbine-governor systems and
power system stabilizers
Testing Requirements
• Generator reactive limits shall be periodically reviewed
and field tested to ensure that reported var limits are
attainable
• Dynamic modelling data shall be field verified through
testing
• Dynamic field testing shall include voltage regulators,
speed/load governor controls, excitation systems, power
system stabilizers and other devices if applicable
• Test shall ensure coordination of generator controls
(minimum and maximum excitation limiters) with
generator short-term capability and protective relays
• Tests are required at least every 5 years
• Unit-specific data is required for all generators greater
than 50 MVA or installed after 1990
1
Benefits
Some of the benefits of modelling tests are listed below
• To accurately determine the reliability of the system
through operating security limit studies
• To plan equipment outages and required enhancements
through accurate estimates of active and reactive
margins
• To identify and correct control instability problems
• To identify and correct equipment problems
TESTING EXPERIENCE
Kestrel Power Engineering has routinely performed all of
the recommended tests on units ranging in size from 2 to
1200 MVA. We perform acceptance and tuning tests and
modelling of exciters, power system stabilizers and limiters.
Generators are tested to derive model parameters and to
establish their reactive capability. We have also performed
tests and coordinated tuning of multiple governors in
islanded conditions.
Much of our testing and modelling effort takes place during
the commissioning phase of new or retrofit equipment.
Control performance and validation is routinely specified as
part of any equipment modification. Regular re-verification
has not been a priority in the past, although routine tests
have been incorporated into regular scheduled outages.
A number of problems in recent years have forced us to re-
visit generator control performance and settings. In cases
where we have tested controls to analyze and avoid system
disturbances we have found numerous changes to the model
parameters: equipment parameters have been altered from
their reported settings, latent uncorrected defects have been
discovered, and system changes have occurred without
corresponding alteration of tunable controls.
The attached paper describes some of the issues and our
experiences in performing testing and modelling on the
North American bulk electricity system. An attachment
provides some examples of actual results obtained using the
described test procedures. These tests were performed on
behalf of utility clients wishing to meet the recent WSCC
modelling requirements.
 Testing & Modeling of Generator Controls
G.R. Bérubé, L.M. Hajagos
Kestrel Power Engineering
Toronto, Ontario, Canada
INTRODUCTION
The testing and modeling of generators, and their associated
controls, is a complex topic. This is in part because of the
variety of different generators and controls found on a
practical power system, and the manner in which they
interact during normal and abnormal power system
operating conditions. There is a consensus within the
industry that the development of accurate models for
generator control systems is a key step in establishing
operating-security limits and in simulating and
understanding the operation of the system during
disturbances.
The most common objectives of utilities embarking on
systematic test programs, to establish models of their
facilities, include:
• Simulate system or local disturbances, to compare with
measured data or reconstruction of actual events and
derive remedial actions
• Simulate system disturbances in the planning or operating
time frame to establish secure operating limits and assess
alternative plans
• Coordinate existing or new protection and control systems
• Study the addition of new facilities
Regardless of the reason for testing, the first step should be
a thorough needs analysis. The needs analysis identifies the
utility’s requirements and translates them into specific
criteria: location of critical stations, list of critical units, list
of critical components (e.g. generator, excitation system,
governor…), type of models and data to be collected, and
finally, types of tests to be performed.
This paper describes some of the issues in performing a
needs analysis, and our experience in performing testing and
modelling in North America and worldwide.
GENERATING UNIT CONTROLS
Figure 1 provides an overview of the turbine-generator
control loops for a typical fossil-fired unit. These controls
allow the power system to meet the continually changing
customer load requirements by adjusting the active and
reactive outputs of the generators. Briefly, the major
control loops are:
2
Power System Generation Dispatch includes automatic
generation control systems (AGC) for the maintenance of
overall system frequency and tie-line interchanges and
manual operator controls such as economic dispatch.
Prime Mover Energy Supply System for fossil-fired units
(shown) includes fuel supply and boiler control loops, while
for hydroelectric machines, penstock and conduit systems
will participate in the dynamic behaviour.
Turbine Controls include the governor (speed control
loop) as well as supplementary controls which are active
during full or partial load rejections.
Excitation System includes the amplification stage(s),
automatic voltage regulator (AVR) function, excitation
limiters and supplementary controls such as power system
stabilizers (PSS), and outer control loops (e.g. power factor
or reactive power regulators).
Included within these control loops are the turbine and
synchronous generator. Clearly, the modelling of these
components can have a significant impact on the accuracy
of the representation of each of the major control loops.
While not explicitly shown in Figure 1, plant auxiliaries and
large system loads may require detailed modeling,
especially for conditions involving large variations in
system frequency and voltage /2/.
The need to represent each of these components and control
loops, and the type of model required, is established by
determining the type of simulations which are to be
performed, and the types of operating conditions to be
represented.
MATCHING MODELS TO REQUIREMENTS
A critical stage, which is often overlooked, is a Needs
Analysis. The needs analysis identifies the utility’s
requirements and translates them into the information
required to proceed with testing and modelling. The various
stages are described below.
 Figure 1: Turbine-Generator Control Loops
Define the Need
The first step is to define why the tests and models are
required. Some questions that must be answered are:
• Is the sole purpose of performing the tests to develop
simulation models?
• Are the tests intended to reproduce disturbance conditions
in order to identify deficiencies and improper
coordination of limiters and controls?
• Will the models be used to produce recommendations for
changes to equipment settings?
Once these questions have been answered, the modeling
requirements may be identified.
Select Location(s)
This is often the simplest stage, and involves considering
the following factors:
• Was a station or specific unit a critical participant in a
system event?
• Did the station, or specific equipment, exhibit unexpected
behaviour during the system event?
• Is existing dynamic data available at a station?
• Is the available data supported by previous tests and
measurements or well-documented manufacturer’s data
sheets?
3
A prioritized list of stations and units can usually be
compiled from a quick review of a past system event or a
forecast operating condition.
Define the Equipment Modelling Requirements
In order to determine which pieces of equipment must be
modelled in detail, the type of simulations that will be
performed must be defined. This involves establishing the
time frame and the system conditions that are to be
simulated (e.g. interconnected and/or islanded operation).
The latter is often specified as a range of voltages and
frequencies for which the models must be valid.
Table 1 provides example guidelines for equipment to be
modelled for different purposes. These are only guidelines;
location-specific details should be combined with the study
objectives to arrive at a final decision. The entries can be
interpreted as follows: “Y” indicates dynamic model
normally included, “N” normally not included, “?” may be
included depending on the specific conditions and
simulation goals.
The following are definitions of the study types used in
Table 1 (typical bounds on study time scales for each
phenomenon are indicated in parentheses):
SSR subsynchronous resonant interaction between
synchronous machines and transmission system elements
and controls. (< 1s)
Transient large disturbance, first-swing rotor-angle
stability(< 10 s)
Small-Signal linearized analysis of oscillatory modes
associated with rotor electromechanical oscillations and
control system modes(<20 s or eigenvalue)
Islanded Operation operation of isolated portions of the
system; large frequency and voltage excursions (10 s to a
few minutes)
Long-Term Dynamics slow dynamics associated with
operation following widespread disturbances (minutes) /3/
Voltage analysis of voltage stability for large and small
disturbances (all time frames are possible, depending on
type of phenomenon)
significant nonlinearities, such as saturation and limits.
Linearized representations may be derived from the large-
signal models, often within the simulation software used in
the studies without intervention by the user.
Normally, the models are developed to match industry-
accepted standards /5,6,7/. In some cases more complex
models are developed; these are usually used for unusual
equipment or when a direct correspondence is required
between the model parameters and the settings of the
physical system. Figure 2 provides a portion of an overview
block diagram of an excitation system in which external
components are identified for each module. One of the
blocks, the damping feedback circuit shows the relationship
between the external settings and transfer function
constants. The transfer functions and defining equations for
each of the blocks is developed in a similar manner to
include gains, time constants and limits in terms of the
physical components and available settings (e.g.
potentiometer settings, resistor values, jumper positions).

Table 1: Equipment Modelling Requirements The advantage of this type of

STUDY TYPE model is that it allows the results
EQUIPMENT SSR Transient Small- Islanded Long-Term Voltage of studies to be translated into
Signal Operation Dynamics recommended changes which can
Generator Y Y Y Y Y Y be implemented in the field to
Shaft System Y N ? N N N obtain performance
Excitation Systems improvements, and the variables
AVR ? Y Y Y Y Y are represented in engineering
Stabilizer ? Y Y ? ? Y
units (e.g. V, A,…) which can be
Limiters N ? N ? Y Y
compared directly with measured
PF/VAR N N N ? Y Y
results /8/.
Turbine Controls
Speed Governor N N ? Y Y N
Overspeed N N N Y Y N
The disadvantage of this structure
Controls is that the models are complex,
AGC N N N N ? N manufacturer-specific and are not
Prime Mover N N N ? Y N in a standard format that can be
Loads N N ? Y Y Y used in commercial simulation
software. Therefore, the detailed
models are often used as a starting point for developing
There is obviously overlap in these selected study types, as standard non-linear or linear models. Figure 3 is an
certain simulations will address multiple categories example of one such model, IEEE Standard AC1A /5/,
simultaneously. One row has been added for loads; representing an alternator-rectifier excitation system with
although they are not part of the generator control systems, non-controlled rectifiers.
inclusion of accurate load models is often critical to the
success of detailed simulations /4/. For many studies this kind of representation, the exciter
power stage and AVR control, will be adequate, while for
Select Model Structures other studies the excitation limiters or supplementary
controls must be added to reproduce the unit’s behaviour
The types of models and tests required are a function of the /9/. While industry standards also provide guidance for
intended application. For small-signal studies, linearized modeling these features, there tends to be a higher level of
representations are adequate. For transient or other large customization, often requiring user-defined representations.
disturbance studies, the models must explicitly include all

4
 Figure 2: Overview of Block Diagram of Exciter Including External Settings

Vs VAMAX
limiters VRMAX
+
- 1 + s TC KA VR 1 VE Efd
Vc GATE +
X
+ 1 + s TB 1 + s TA s TE
- -
FEX
0
Vref VAMIN VRMIN
FEX=f[IN]
VX = VE SE[VE]

+
+
KE KCIFD
IN =
VE

s KF +
+
KD Ifd
1 + s TF VFE

Figure 3: Excitation System Model Type AC1A

TESTING METHODOLOGY validation. The following is a brief description of the steps

which can be included in this process /10/. Not all of these
Once the locations and level of modeling detail have been steps are required for each location and each piece of
selected, the model parameters must be determined. Field equipment.
testing is normally part of the model development and

5
Step 1 - Review available data: Collect data from the
station and from the manufacturer of the equipment. This
includes operating manuals, diagrams, capability and rating
data, previous test results, and, where available, block
diagrams. This stage often provides a significant amount of
information reducing the amount of site testing required.
Step 2 - Select model structures: Based on the available
data, develop a block diagram of the equipment to be
modeled. This often involves selecting the standard model
that is the best fit to the equipment.
Step 3 -Develop test plan: The test plan should include
measurements to identify all of the key parameters, and
provide additional data for validating the final product - the
completed model. The plan should clearly identify any
special equipment requirements, the station/unit operating
state during each phase, and the safety/reliability issues.
Step 4 - Perform site tests: Included in the actual testing
should be a review of nameplate data and any equipment
modifications and a comparison to the manufacturer’s data
gathered earlier. In our experience, it is not uncommon to
find differences between field wiring, installed component
values, equipment settings and documentation. The
functional and dynamic tests also frequently identify latent
deficiencies, such as defective components or incorrect
settings. These may not be apparent during typical day-to-
day operation.
Step 5 - Validate data: Compare measured and modeled
responses for modules and/or complete system. In order to
expedite this stage we have developed data acquisition
systems which format the data for use with our simulation
software. Whenever possible parameters are calculated on-
site and compared with test results. This provides an
opportunity to identify inconsistencies or missing data,
which can then be rectified prior to removing the test
equipment.
Step 6 - Produce reduced-order model: Eliminate
feedback loops, time constants and limits not relevant to the
target model. Convert all parameters to per-unit values for
use within simulation software.
Step 7 - Train station staff: This step is not always
included as part of the process, but it is actually critical to
ensuring that the relationship between the derived model
and equipment settings is maintained. Ideally, station staff
should be provided with enough training to periodically test
the unit and confirm that its dynamic performance has not
changed significantly. Regular courses are offered, and test
facilities have been added to most excitation systems and
stabilizers to permit verification of the closed-loop voltage
regulator and PSS response /11/.
6
In order to perform tests on a wide variety of equipment
designs, specialized test instrumentation is required. Over
the years custom transducers have been developed for
measuring all of the quantities associated with generators,
excitation systems and governors (e.g. active/reactive
powers, field voltages and currents, shaft speed, frequency,
valve positions…). These transducers and associated
instrumentation differ from multi-purpose hardware in that
they have been specifically designed for the bandwidth,
isolation and accuracy requirements of these dynamic tests.
In order to support different client requirements, we utilize
most of the common commercially-available power system
and control simulation packages, along with several custom-
designed tools. This allows us to develop and test models
on different target platforms.
EXAMPLE RESULTS AND CASE STUDIES
Generator Testing
In many cases, utilities rely on manufacturer-supplied data
in impedance and time-constant form. Depending on the
manufacturer, this data may be measured through enhanced
short-circuit and stator decrement tests, or derived from
numerical analysis of design data /12/. Our experience has
been that there can be significant differences between the
parameters provided by manufacturers and those obtained
through direct measurement. The tests we have performed
range from simple static and open-circuit measurements to
sophisticated standstill and on-line frequency response tests
/13/.
Figure 4 provides a comparison of the quadrature-axis
operational inductance of a round-rotor generator obtained
through standstill frequency response tests and data
provided by the manufacturer. Detailed impedance and time
constant values may be derived from SSFR data.
Figure 4: Comparison of SSFR and Manufacturer Data
for Thermal Generator
Although the models derived through detailed tests can be valid for a variety of operating conditions. Regardless of
significantly different, one of the conclusions of /13/ is that, the technique applied, the final step should include
for modern generators, there is not always a significant validating the model, by comparing simulated time-domain
improvement in the accuracy of simulations when compared results against measured field data.
with those conducted using manufacturer’s data. In fact,
our experience has been that the accuracy of the excitation
system models is often far more significant in determining
the outcome of most simulations. Once accurate exciter and
governor models are available, then improving the generator
representation can provide incremental improvement, where
special controls are being simulated or tuned.

Excitation System Modelling

It has been our experience that generator excitation system

settings and performance can have a dramatic effect on the
overall behaviour of individual units and the power system
as a whole. Some form of excitation system model is
required in virtually every type of power system simulation.
Our approach has been to maximize transmission capability
through the use of high-initial-response (HIR) excitation
systems and supplementary controls such as power system
stabilizers (PSS) and transient excitation controls (TSEC).
Consequently, we test all excitation systems and derive Figure 5: Comparison of Measured and Simulated
detailed large-signal models for inclusion in the system Response for Alternator-Rectifier Excitation System
dynamic database.
Figure 5 provides an example comparison between
Excitation system tests are normally performed in one, or simulated and measured data for a generator equipped with
more, of the following operating conditions: an alternator-rectifier excitation system.

• excitation system energized via test supplies and isolated
from the generator field winding. These tests are often
scheduled during routine maintenance outages, and
provide an opportunity to obtain detailed measurements of
the transfer functions of the various sub-assemblies (e.g.
AVR, UEL, PSS, …)
• generator on open circuit at rated speed. The excitation
system is operated in its normal configuration and
generator excitation and field voltage are adjusted to
various levels. This provides an opportunity to measure
closed-loop response and test the various power stages
under normal and forcing conditions.
• generator synchronized to the grid, operating at a variety
of active and reactive power loads. These tests provide an
opportunity to test excitation limiters, and supplementary
controls (e.g. PSS, VAR regulator) under realistic
operating conditions.
A wide variety of testing techniques have been developed
for determining the model parameters. In general, they all
involve performing measurements of the time or frequency
response of the system, and comparing the result against
simulated responses calculated using different model
parameters. Various system identification techniques are
available for extracting the model parameters directly from
overall test results /14/. Some caution must be used when
interpreting results from overall tests to ensure that they are
Thermal Governor Modelling
Detailed thermal governor models are normally only
developed when simulation of partial/full load rejections are
required. Reference /2/ describes simulations performed to
understand the nature of a failed load rejection test on a
nuclear unit. The failure was a result of large induction
motors stalling, causing protective relays to operate. Potential
remedial measures were simulated and a final solution, using
a temporary voltage boost on the AVR, adopted and
implemented to prevent further failures.
Simulations were conducted to answer the following
questions:
• Under the conditions of the test, do the pump motors stall?
• Are the initial generator conditions important?
• Is the sustained overspeed important?
• Is the distribution of the motors loads significant?
• Does the performance of the AVR play a significant role?
The simulations also allowed different remedial measures to
be studied, thereby eliminating costly re-testing on the unit. In
order to simulate the generator speed and motor stalling,
detailed generator, governor and load models were required.

7
The generator was represented with the usual power system
stability model having two damper windings on the
quadrature axis and one on the direct axis, using accurate
reactances and time-constants parameters.
The excitation system consists of a main rotating exciter and
twin digital AVRs, each with a power system stabilizer (PSS).
It was represented by a continuous equivalent IEEE standard
model /5/.
A governor model which represented the control and
intercept valve action and the overspeed controls was used. In
particular, the shaft acceleration detector controls were
modelled as shown in /2/. This control acts to close the
intercept valves for 5 seconds following detection of shaft
acceleration greater than 4.5%/s /15/. The governor load set-
point runback at 10%/s following detection of full load
rejection was also explicitly modelled. Governor parameters
were adjusted to produce a speed signal that closely matched
the measured results. This non-standard model was
implemented within the user-defined capabilities of the time-
domain simulation program.
The station auxiliary service load consists of large induction
motors driving pumps, the generator excitation system, lights,
and other smaller loads. The most important motors are those
driving the Primary Heat Transport (PHT) pumps.
Each type of motor was modelled as a detailed induction
motor having a deep bar rotor, and with saturation of the
stator and rotor leakage reactances represented /1/. The model
parameters were obtained by converting from the standard
specifications supplied by the motor manufacturer using
commercial software.
Figure 6. Simulation of Failed Load Rejection Test
The simulation of the unsuccessful load rejection test is
shown in Fig. 6. The shaft speed of one of the Primary Heat
Transport pumps from bus A/B (feeding 5 motors) and one
8
from bus C/D (feeding 7 motors) are shown at the top of the
graph with generator speed. The simulation shows that the
PHT pump motors on the more heavily-loaded transformer
secondary are unable to track the generator speed. The high
motor current lowers the motors' terminal voltage sufficiently
to cause their speed to start to reduce after 1 s, i.e. the motors
stall. The duration of the high inrush current eventually causes
the operation of the bus protective relays.
Simulation results, and past experience, demonstrated that if
the generator terminal voltage could rapidly be increased to
over 1.05 per unit for several seconds following a full load
rejection, the motor stalling problem could be overcome. The
necessary logic to activate a terminal voltage boost signal was
obtained from the very flexible logic already present in the
digital excitation system. This solution was subsequently
implemented without exhaustive testing because of the
fidelity of the computer models.
Hydro Governor Modeling
Detailed hydro governor models are normally only included
in simulations intended to reproduce long-term dynamics
and system operation during which large frequency
deviations occur, such as islanded operation. The structure
of the local bulk electricity system includes numerous
pockets of hydroelectric generation, which are connected to
the remainder of the system via limited transmission
connections. As a result, we have performed measurements
and simulations of islanded operation on several occasions,
and have developed standard procedures for tuning for
optimal damping /16/.
The approach taken for hydro governor testing is similar to
that used for excitation system testing, although additional
specialized test equipment is required to introduce test
inputs and measure the intermediate and output quantities.
With the advent of electro-hydraulic governors, which
employ analog and digital regulators, the need for
specialized hardware has been reduced, as all critical
quantities are converted to electrical form.
Figure 7 shows a comparison of the measured and simulated
response of a hydro-governor during islanded operation.
Clearly, the model developed for the PI governor, provides
an excellent tool for studying the performance of the unit
during isolated operation.

Figure 7. Governor Response - Islanded Operation
CONCLUSIONS & RECOMMENDATIONS
Our experience has been that the most critical models for
system stability limit studies are those of the exciter and
associated controls, such as power system stabilizers. Up-
to-date manufacturer’s data, confirmed with spot
measurements usually provides adequate representation of
generators. Governor models are required for islanded
operation studies or for simulation of plant dynamics, such
as load rejections. Induction motor and nonlinear load
models may be required for specific event reconstruction.
Representation of over- and under-excitation limiters,
voltage and volts/Hertz limiters and power factor/var
controllers may be required for detailed simulations; in
addition, site tests provide an opportunity to confirm
settings for adherence to contractual agreements for
available reactive support
Prior to embarking on a test program, the utility should
complete a “needs analysis” to define exactly what the
results will be used for or describe the problem which has
initiated the requirement for testing and modelling. Once
this has been done, determine what pieces of equipment
should be tested at what locations. Select the types of tests
based on the types of models required. Produce a test plan
for the location based on combining the standard tests and
location-specific details. As much as possible, validate the
models on-site, and train staff to collect appropriate data for
future performance review. Finally, the testing methodology
should conform to the utility’s normal operating and
maintenance practices.
REFERENCES
1. Extended Transient Midterm Stability Program: Version 3.0,
EPRI TR-102004, Ontario Hydro, Project 1208-9, April 1993.
2. G. Rogers, R.E. Beaulieu, L.M. Hajagos Performance of
Station Service Induction Motors Following Full Load Rejection of
a Nuclear Generating Unit, IEEE Transactions on Power Systems,
Vol 10, No 3, August 1995.
9
3. Long-Term Power System Dynamics, EPRI Report EL-6627,
Final Report of Project 2473-22, Ontario Hydro, December, 1989.
4. L. Hajagos, B. Danai, Laboratory Measurement of Modern
Loads Subjected to Large Voltage Changes for Use in Voltage
Stability Studies, CEA project 113T1040, final report, May 1996.
5. IEEE Recommended Practice for Excitation System Models
for Power System Stability Studies, IEEE Standard 421.5-1992.
6. Hydraulic Turbine and Turbine Control Models for System
Dynamic Studies, IEEE Working Group Report, IEEE Trans., Vol.
PWRS-7, No. 1, pp. 167-179, Feb 1992.
7. Dynamic Models for Steam and Hydro Turbines in Power
System Studies, IEEE Trans., Vol. PAS-92, Nov/Dec 1973.
8. K. Shah, G.R. Bérubé, R.E. Beaulieu, Testing and Modelling of
the Union Electric Generator Excitation Systems, presented at the
1995 Missouri Valley Electrical Association meeting in Kansas City,
MO, April, 1995.
9. G.R. Bérubé, L.M. Hajagos, R.E. Beaulieu, A Utility
Perspective on Underexcitation Limiters, IEEE Transactions on
Energy Conversion, Vol 10, No 3, September 1995.
10. IEEE Guide for Identification, Testing and Evaluation of the
Dynamic Performance of Excitation Control Systems, IEEE
Standard 421.2-1990.
11. G.R. Bérubé, L.M. Hajagos, Utility Experience with Digital
Excitation Systems, IEEE PES Winter Meeting, Feb 1997, NY.
12. IEEE Guide: Test Procedures for Synchronous Machines,
IEEE Standard 115-1995.
13. J. Service, L.M. Hajagos, Practical Aspects of On-Load
Generator Testing, EPRI TR-102351, Project 2328-02, Final
Report, May 1993.
14. T. Guo et al, Identification of Model Parameters of Excitation
System and Power System Stabilizer of Mingtan #6 via Finalization
Field Tests, paper 94-SM-566-0-PWRS presented at IEEE/PES
1994 Meeting.
15. P. Kundur, D.C. Lee, J.P. Bayne and P.L. Dandeno, Impact
of Turbine Generator Overspeed Controls on Unit Performance
Under System Disturbance Conditions, IEEE Trans. on PAS, June
1985, Vol PAS_104, No. 6.
16. P.L. Dandeno, P. Kundur, J.P. Bayne, Hydraulic Unit Dynamic
Performance under Normal and Islanding Conditions - Analysis
and Validation, IEEE Trans., Vol. PAS-97, Nov/Dec 1978.
17. Rowen, W.I Simplified Mathematical Representations of
Heavy-Duty Gas Turbines, Journal of Engineering for Power,
October 1983, Vol 105, p865.
G. Roger Bérubé graduated from McGill University in Montréal,
Canada, with a B.Eng. and M.Eng. in electrical engineering in 1981
and 1982 respectively. Since 1986, he has worked mainly on the
modeling, testing and development of excitation and governor
controls for synchronous generators. He is a registered Professional
Engineer in the Province of Ontario and a member of the IEEE PES.
E-mail: roger@kestrelpower.com ph:(416)767-7704
Les M. Hajagos received his B.A.Sc. in 1985 and his M.A.Sc. in
1987 from the University of Toronto. Since 1988 he has worked
mainly in the analysis, design, testing and modelling of generator,
turbine and power system control equipment and power system
loads. He is a registered Professional Engineer in the Province of
Ontario and an active member of the IEEE PES. E-mail:
les@kestrelpower.com ph:(905) 272-2191
APPENDIX A: WSCC TESTING EXPERIENCE
40

Out Overexcited
The following examples are drawn from our recent testing
experience in the WSCC system and elsewhere.

Generator Testing 30

Generator parameters may be confirmed with some straight-

forward static and dynamic tests. The open-circuit 20
saturation curve should be measured and compared with the ROTOR LIMIT 1140 A
manufacturer’s measurements as shown in Figure 8.

Reactive Power (MVAr)

10
16
Active Power(MW)
14 0
0 10 20 30 40 50
12
-10
Terminal (kV)

10

8
In Underexcited -20
6
Air gap field current UEL LIMITER
@ 13.8 kV = 560 Adc
4 -30
measured points
fitted curve
2 LOE RELAY
-40
0
0 200 400 600 800 1000
Figure 9. Capability Curve
Field Current (Adc) 1.00
simulated
Terminal V

Figure 8. Open Circuit Saturation Curve 0.95

measured
(pu)

0.90
Static measurements of field current, terminal voltage,
active and reactive power, and rotor angle allow the 0.85
calculation of d- and q-axis and leakage reactances. During 0.80
these tests, the capability curve may be verified, and limits
such as field current, bus voltage, under-excitation limiter or 140
loss of field relay identified (Figure 9). 135
(Vdc)
Field

Two load rejection tests have been used to verify transient 130
parameters. A zero power factor test with excitation on 125
manual, shown in Figure 10, allows d-axis parameters to be
120
fitted to terminal voltage and field current response.
465
A partial load rejection test, as in Figure 11, allows
450
confirmation of generator inertia, and is also useful in
(Adc)

435
Field

analyzing off-line governor performance.

420
405
390
0 5 10 15

Time (seconds)
Figure 10. 0 pf Load Rejection Test

10
 filters, gain, and phase lead-lag stages are measured with the
60
Power (%) stabilizer off-line. The stabilizer tuning requirement may be
determined using an on-line frequency response test, shown
40
in Figure 12.
20
The desired stabilizer gain and phase lead were
0 implemented and on-line step responses with the stabilizer
on and off, shown in Figure 13, were used to demonstrate
140 the results.
130
Speed (%)

1.00

Active Power
120

(pu)
0.95
110
100 0.90
90
initial slope with inertia 2.55 MW-s/MVA 1.04
60
Gate Pos'n (%)

Terminal V
(pu)
cushioning rate = -0.8 %/s 1.03
40
closing rate -6%/s PSS ON
PSS OFF
20 1.02

0 0.0010
0 10 20 30 40
delta speed
(pu) 0.0005
0
Time (seconds)
-0.0005
Figure11. Partial Load Rejection Test
-0.0010

0.005
150
PSS Output

stabilizer phase compensation 0

Tw = 10 s, Tlead = 0.18 s, Tlag = 0.05 s
(pu)

measured phase lead requirement

-0.005

-0.010
0 1 2 3 4 5
100
Phase (degrees)

Time (seconds)

Figure 13. Power System Stabilizer Test

Electro-Hydraulic Governor Testing

50 Electronic governors may be tested in the same way as

analog electronic exciters, with a combination of time- and
frequency-response tests. Static measurements are used to
establish the permanent droop and turbine operating curve
(gate or fuel flow versus power output). Step tests, shown in
Figure 14, may be used to confirm PID gain parameters.
0
0.1 0.2 0.5 1 2 5 Gas Turbine Governor Modeling
Frequency (Hz)
The advent of gas turbines and co-generation has created a
Figure 12. Stabilizer Phase Lead Requirement new class of facility, which is becoming widespread as the
choice for new construction. A simplified simulation model
Power System Stabilizer Tuning and Testing of a single-shaft gas turbine is presented in /17/.

Power system stabilizer testing and tuning is performed The controls are typically analog-electronic or digital
using a combination of off-line and on-line time- and electronic. In both cases, step-response tests and steady-
frequency-domain tests. Transfer functions of the input state measurements have been successfully used to

11
characterize both the turbine and governor (Figure 16). Gas Turbine Governor
Instrumenting and performing dynamic tests on digital Off-Line Reference Step Response
versions presents more of a challenge. However, even if the 1.5
manufacturer has not provided built-in test facilities, 1.0

speed
dynamic response tests can usually be performed with some

(%)
0.5 measured
guidance from the manufacturer. The testing environment simulated
0
(typically the plant control room) is decidedly more
-0.5
attractive.
1.0
2
0

Pos'n (pu)
Prop

simulated
(V)

Valve
-2 0.5
-4 measured
-6
8 0
6
(V)
Int

4 600
2

(deg F)
0

Temp
575
2
1
Deriv
(V)

0 550
-1
-2 0.2

Demand (pu)
3
Power

Valve
(MW)

2 0
1
0 -0.2
0.5 0 10 20 30 40
0.4
Gate

Time (seconds)
(pu)

0.3
0.2 Figure 16. Gas Turbine Governor Response
0.1
0
0.8
0.6 CONCLUSIONS
Lvdt
(pu)

0.4
0.2
0 The preceding paper and case studies are intended to
0 5 10 15 illustrate that models, test techniques, and control guidelines
are available to meet the NERC testing and modelling
Time (seconds) recommendations. The equipment models shown have been
Figure 14. PID PID Governor On-Line Response drawn from IEEE standards and are widely available in
power system simulation programs.
Standards for control tuning for such equipment have not
yet evolved, so reporting a model, rather than tuning is the G. Roger Bérubé
primary test objective. It should be noted that the model Senior Engineer
shown includes several control modes: droop control (the Kestrel Power Engineering
familiar mode for partial load conditions), acceleration Ph: (416) 767-7704
control (for startup and maneuvering), and temperature E-mail: roger@kestrelpower.com
control. For economic reasons, these units are invariably
loaded to near full output, and are controlled to respect Les M. Hajagos
temperature, pressure and emission constraints. In this Senior Engineer
mode, the speed/load droop control is not active. As these Kestrel Power Engineering
facilities may represent a significant portion of the total Ph: (905) 272-2191
generation in an area, control guidelines will need to be E-mail: les@kestrelpower.com
developed to correctly model their governor performance.

12