Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Guidelines for Generator Stability

Model Validation Testing


IEEE Task Force on Generator Model Validation Testing of the Power System Stability
Subcommittee*

Abstract— This paper constitutes the final report of the IEEE


Task Force on Generator Model Validation Testing, which II. SIMULATION MODELS
was created in 2002 and served under the Power System
Stability Subcommittee. The objective of the TF was to Power system planning and operational studies require the
provide technical guidance to utilities undertaking testing and simulation of the response of synchronous machines and
modeling of generation facilities. This paper outlines accepted their respective control systems. For these studies it is
industry methods for establishing dynamic models for essential that the control systems of the synchronous
generators and their control systems. machines be modeled in sufficient detail (see [1]). The
desired models must be suitable for representing the actual
Index Terms—generator modeling, model validation, field equipment performance for large, severe disturbances as
testing of power plants
well as for small perturbations. To obtain accurate
simulations, not only must the models be of an adequate
I. INTRODUCTION level of detail, but the values of the parameters in the
Few people disagree that good quality simulation models models must also correspond to actual field values.
of power system equipment is beneficial to all power The equipment to be tested and modeled includes the
system participants. There are, however, differing opinions generator and its control systems; excitation systems,
on the required level of detail, best methods to obtain data, power system stabilizers and turbine governors. Protective
and frequency of verification. The goal of this paper is not relay coordination with equipment capabilities and control
to prescribe one answer to any of these issues. Rather, it is system limiters is equally important and is discussed in
to identify a body of available techniques and describe the detail in appropriate standards [2-11, 62].
benefits and drawbacks of each, to help industry
participants determine the best approach for their types of To some extent, the model structures used in system
equipment, capabilities and objectives. studies are intended to facilitate the use of field test data as
a means of obtaining model parameters. The models are,
The purpose of the model validation is to ensure the proper however, reduced order models and they do not necessarily
performance of the control systems and validate the represent all of the control loops on any particular system.
computer models used for stability analysis. In some cases, the model may represent a substantial
Specific goals include the measurement and verification of reduction, resulting in large differences between the
the following: structure of the model and the physical system. Model
structures are shown in the references [2, 7, 12, 13 and 14].
• generator reactive capability
Useful models should be valid for frequency deviations of
• generator dynamic model data +/-5% from rated frequency and oscillation frequencies up
• excitation system, stabilizer and limiter models to about 3 Hz. Voltage deviations are typically between
• turbine/governor models 95% and 105% for steady-state generator operation, but
• coordination between protective devices, control system the models must be capable of representing large-signal
limiters and equipment capability disturbance performance. The models discussed here
would not normally be adequate for use in studies of sub
In addition to obtaining model data, the tests performed to synchronous resonance or other shaft torsional interaction
gather this information may uncover latent defects that problems. Slow-acting control and limiter functions (such
could lead to inappropriate unit response during system as tap-changing transformers and generator field current
disturbances, thereby improving the reliability of the unit limiters) that may come into play in long term dynamic
and the power system. performance studies are required for some studies and are
included.

*Task Force Members: Les Hajagos (chair), Joel


III. REVIEW OF EXISTING DATA AND MODELS
Barton, Roger Berube, Murray Coultes, Jim Feltes, Gary
Lanier, Shawn Patterson, Les Pereira, Pouyan Pourbeik, Prior to undertaking a testing program, a thorough review
Alex Schneider, Robert Thornton Jones should be made of existing simulation models,

1
1-4244-1298-6/07/$25.00 ©2007 IEEE.
manufacturer’s data sheets, and control and protective needed. In addition, many of the tests are designed to be
equipment tunable settings. Priority should be given to performed by the manufacturers and do not lend
testing parameters which are tunable, or which may have themselves to application in the field to installed
been changed by refurbishment, maintenance or equipment. With the increasing usage of sophisticated
replacement. Often, new excitation or governing controls computer-based design tools, detailed commissioning tests
are installed without corresponding changes to simulation of this type are often only performed as type tests on the
models. Conversely, some model parameters, such as unit initial design.
inertia, may be fixed by the system design, and once TABLE 1. MODEL PARAMETERS
measured or calculated, may not change except by
equipment modification. Examples of model parameters to Conditions Generator Exciter Power Governor
be validated are shown in Table 1. System
Stabilizer
Some common mistakes that may be found in existing
database models are as follows: A. Set by Reactances, Limits, Input Water
design (not type(s) Starting
• Use of typical or bid data rather than as-built data. changed Time,
• Incorrect scaling of data based on changes to unless rebuilt)
Time Exciter time No Load
ratings, most commonly up-rating of unit apparent Constants, constant, Gate,
power
• Use of saturated instead of unsaturated generator Inertia, Voltage Full Load
sensing time Gate,
parameters
constant,
• Inertia not including the contribution of the
turbine Saturation Saturation Turbine
• Exclusion of features such as reactive current (rotating Power
exciter) Fractions,
compensation in excitation systems
Maximum
Power,
IV. DESCRIPTION OF TESTING AND VALIDATION METHODS
This section will focus on the testing methods generally Dead band,
employed to determine model parameters for generators, Turbine
excitation systems, and governors. damping,
Existing methodologies for generator testing can be B. Tunable Voltage Gains, Droop,
broadly divided into two categories: Regulator
Gains,
• Time-response tests, involving step changes to
controller set points, partial load rejections and Time Time Time
other disturbance techniques with the machine Constants, Constants, Constants,
operating at pre-defined conditions [16, 17, 18, Reactive Limits Gains,
19, 20, 29, 30]; and Compensation
• Frequency response tests conducted with the unit
at standstill [21, 22] or in operation. Limit Levels Rate Limits

These methodologies are well established and can produce A. Synchronous Generator Tests
accurate models. The major drawbacks are associated with
Historically, manufacturers performed sudden short circuit
the time the machine is unavailable for normal commercial
tests to determine dynamic synchronous generator
operation (economics) and the technical problems that may
parameters. These test procedures are specified in IEEE
arise during the tests. In some cases the technical
Standard 115-1995 [4]. These tests provide only the d-axis
problems are associated with the discovery of latent
transient and sub-transient constants. In addition, they do
defects. In this case, although problems may arise during
not include measurement of field current during the short
testing, this is considered preferable to the consequences
circuit tests and consequently the field circuit is not
that could occur during an actual system disturbance.
specifically identified. The limitations of these procedures
Note that the tests and methodologies defined in equipment for providing data suitable for stability studies have been
standards [3 through 11, 60] are not specifically designed recognized for some time.
to determine model parameters, but generally to specify a
Several different testing and analytical methods have been
desired level of performance and usually a methodology to
proposed and used to obtain better models. These include:
measure that performance. While these tests can give very
useful information for deriving parameters, it is not their
• Enhanced sudden short-circuit tests
main focus to determine the complete set of parameters

2
• Partial load rejection tests control, the main generator breaker is opened which results
in decay in terminal voltage. As the generator dynamic
• Frequency response tests
response is only a function of the generator d-axis time
o Standstill frequency response constants and reactances, this test can be used to determine
o Open-circuit frequency response those generator parameters. Figure 1 shows a typical
response for this test.
o On-line frequency response
26
• Analysis of design data, e.g., finite element 24
analysis.

Terminal Voltage (kV)


22

These improved methods of obtaining machine 20

characteristics are briefly discussed below. 18

16

1) Enhanced Short Circuit Tests 14

12
References [33] and [34] describe improved methods of 0 5 10 15 20 25 30

utilizing results from sudden short-circuit tests to T im e ( S e c o n d s )

determine more accurate d-axis parameters. The most


important feature of these methods is the utilization of Fig. 1. Typical response for d-axis test showing decay in
rotor current measurements during the short circuit tests to terminal voltage
identify the field circuit characteristics more accurately.
Data is generally recorded using digital recording
Among the disadvantages of the short-circuit approach are: equipment. This allows direct comparison of
the inability to provide q-axis parameters, the necessity for measurements with simulations. Simulations are
subjecting the machine to a severe shock imposed by the performed and the generator model parameters are
sudden short circuit and the complexity of setting up this adjusted iteratively to produce a match. The advantage of
test. this approach is that its setup is relatively simple and it
does not require a high degree of training to execute.
2) Partial Load Rejection Tests
Some of the drawbacks with this approach include the need
Partial load rejection tests are among the most commonly for a manual exciter control that maintains constant field
employed means of validating generator model parameters voltage, and the need to overcome plant controls that may
for in-service generating units. They have been performed prevent completely unloading the unit or operating at a low
on hundreds of units and regional testing programs, such as off-line voltage for an extended period.
the ongoing program in the WECC [15], have been
If power angle can be measured, the q–axis parameters can
primarily accomplished with these techniques. This
be measured in a manner similar to that used in the test
method is one of the simplest approaches to developing an
employed for the d-axis. By matching the power angle
adequate equivalent circuit model, or for verifying
relative to terminal voltage with the power factor angle, the
manufacturer’s data.
steady state conditions for the q-axis test are represented
The load rejection dynamic tests are selected to provide a by the phasor diagram shown in Figure 2, which reveals
simple and safe disturbance to excite the response of the that the armature current is aligned with the quadrature
unit and, as much as possible, limit the set of parameters axis. This requires a generator output in the neighborhood
affecting the response to simplify the derivation process. of 10 percent of rated output and a reactive power
The usual approach is a series of load rejection tests with absorption of a few percent of rated. As in the d-axis test,
the unit initially carrying partial load. This approach is an the generator is brought to the desired loading, excitation
extension of the work documented in References [17, 18, placed on manual and the main generator breaker is
35, and 36]. References [19, 20, 23, 31 and 37] give more opened, resulting in a response in terminal voltage. Since
details on the application of such tests and the procedures the unit is producing real power, the partial load rejection
involved. will cause a rise in generator speed and hence care must be
taken to select the initial loading to limit over speed and
The initial conditions for the load rejections are selected to
also the resulting voltage change.
isolate specific parameters. Two tests are needed to
determine, respectively, the parameters of the generator’s d
and q axes.
To determine d-axis parameters, the generator is
dispatched on-line at approximately rated voltage, no
active power output, and absorbing reactive power. This
results in a steady state condition where the generator
armature current is completely aligned with the direct axis.
With the excitation system in manual (field voltage)

3
This technique has been extensively used on round rotor
generators [48, 49, 50, and 51]. In addition, it is now part
q axis of IEEE standard No.115-1995 [4]. SSFR testing can be
performed either in the factory or during a maintenance
outage on a unit. Other utilities have begun using these
techniques on hydro generators. Reference [52] gives
details of the SSFR test procedures and method of
~ analyzing data. Details of models developed from SSFR
jX q I
tests on three large units are given in reference [50].
~
I
The advantage of this approach is that it can be performed
δ during outages, thus eliminating lost commercial operating
time. The disadvantage is that there is that a high level of
~ expertise is required to perform the test properly and
Ra I
~ analyze the results.
V
b) Open Circuit Frequency Response (OCFR)
Open circuit frequency response testing allows
confirmation of some of the SSFR data in the middle of the
frequency range for the d-axis only [48]. For this test, the
d axis unit is operated on open circuit at reduced voltage. The
field is excited at various frequencies and the field to stator
frequency response measured. The differences between
Fig. 2. Phasor diagram for quadrature axis test this response and the equivalent one from the standstill
tests gives some indication of rotational effects.
In practice, load rejection tests may be repeated to
characterize the dynamic response for the excitation The test is normally done at more than one voltage to
system and turbine/governor if generator field voltage and examine saturation effects. By conducting the test with
governor signals are monitored. Although this approach various signal amplitudes, slot wedge conduction effects
simplifies testing since it does not require a detailed can be assessed.
knowledge of each Manufacturer’s control design, it has c) On Line Frequency Response (OLFR)
some drawbacks. In particular on newer systems, on-line
and off-line response may not be identical as the systems In many respects, on line frequency response testing is the
may switch state depending on the status of the generator "proof of the pudding" as far as small signal verification
breaker. Also these tests are not likely to fully exercise all machine models is concerned. Here the machine is being
parts of the model and provide sufficient data for complete tested under the same conditions as the model is expected
model verification. They may however be useful as part of to perform, although over a restricted operating range.
a re-verification process once detailed models have been For this test, the machine is operated near rated (or at
developed. reduced) load preferably over substantial impedance to the
3) Frequency Response Tests system. The excitation is modulated either by sinusoidal
or random noise. Components are resolved on the two axes
a) Standstill Frequency Response (SSFR) and data similar to that of the SSFR tests is used to derive a
In the SSFR technique, all tests are conducted with the unit model.
at rest and disconnected from the bus. The rotor must be The frequency range of usable data in this test is more
aligned to two particular positions with respect to the stator limited than that of the SSFR tests, but the SSFR data is
during the tests; because of this, it is normally only expected to be good at the frequency extremes. The OLFR
performed on round-rotor (thermal) generators. The stator testing allows the middle of the frequency range to be
is excited by a low level source over the range of filled in with data for both axes that include rotational
frequencies from 1 mHz to 1 kHz. The frequency effects.
responses from the stator winding with the field open and
shorted are analyzed to derive the parameters for a higher The disadvantage of this test is that it requires testing on an
order d- and q-axis model. operating unit connected to the system, possibly under
special system or unit conditions. In addition, it does not
Because the tests are conducted at very low flux levels, the provide large signal response information. Care must be
results must be corrected to bring them from the "toe" of exercised to avoid exciting critical unit or system
the saturation curve to normal unsaturated levels. This is frequencies.
done by minor adjustment of the mutual reactance in each
axis. Details of OLFR measurement techniques and model
identification procedure are given in reference [53].

4
Reference [54] gives results of OLFR tests on two large physical operating limits of the unit. These tests are
thermal generators and improved models developed from required regularly since the reactive capability of each
the tests. generator is critical to power system reliability and can be
affected by many different factors within the plant.
In reference [55], models for three large generators based
on SSFR and OLFR tests are validated by comparing the
A. Voltage and Reactive Capability
results of simulations with measured responses involving
line switching. For one of the generators; the models There is often confusion about the actual reactive
derived from short circuit tests and decrement tests are capability of generators and the relationship between
also validated. The results show that models derived from voltage and reactive power limits during staged testing and
frequency response measurements are generally superior. during actual power system disturbances.

d) Calculation of Parameters from Design Data Utility-scale generators supplied in North America are
typically designed to operate continuously without de-
References [56, 57] describe improved generator models rating for operating voltages between 95% and 105% of
developed from design information. Reference [58] the nameplate voltage. Sustained operation outside of
presents more recent work using finite element analysis. these limits is not recommended and staged tests should
Generator manufacturers use these methods almost not be conducted beyond these limits. Station and unit
exclusively today, rather than tests, when supplying model service loads fed from the generator should be capable of
data for their products. continuous operation at voltage levels corresponding to
Work done on two and three dimensional finite element this range of generator terminal voltage. If a unit is not
models in the steady-state have shown good agreement capable of operating within the 95% to 105% range
with measured results. In addition, two dimensional finite because a station service voltage level restriction is
element models have been developed which provide reached first, this needs to be corrected in order to be
frequency response data that compares well with measured compliant.
results. The degree to which conduction occurs across slot Excluding voltage restrictions, the reactive output
wedges appears to have a significant effect on the model at capability of a generator will be affected by some or all of
higher frequencies. The significance of rotational effects the following factors:
and the effect of disturbance amplitude on the model
require additional investigation. • field winding thermal limits
• stator winding thermal limits
This approach requires detailed involvement by the
• stator core-end overheating limits
Manufacturer and may not be possible for older units
• excitation output capability
unless original design material is still readily available.
• excitation limiter settings
• relay operating characteristics (e.g. loss-of-
V. REACTIVE CAPABILITY TESTING excitation relay settings)
Generator steady state measurements normally consist of Each generator’s reactive capability is depicted in a
tests to either determine or validate the synchronous capability curve, such as the one shown in Figure 3. This
reactances and saturation data associated with the curve plots physical limitations, such as stator and rotor
generator and tests to confirm the reactive capability of the heating limits, in the (MW, MVAr) plane. By maintaining
unit including auxiliaries. The modeling tests involve: operation within the limits depicted on the capability curve,
• Off-line measurements to measure the open- the unit’s continuous ratings will be respected.
circuit saturation curve of the generator Prior to performing reactive capability tests it is important
• On-line measurement of generator terminal to understand the difference between the conditions that
quantities at different active and reactive power exist during staged tests and those that exist during actual
levels system events when the generator may be called upon to
The off-line measurements are used to generate/validate operate at the extremes of its reactive capability. During
the saturation coefficients for the generator. The measured normal operation under Automatic Voltage Regulator
ac terminal conditions of the generator are used to (AVR) control, the Operator sets the AVR reference to
calculate generator field current and rotor angle and this is achieve a specific reactive power or terminal voltage level.
then compared against the measured values to validate the If the power system voltage drops, the AVR control will
synchronous reactances. These tests are normally only respond by boosting excitation resulting in reactive power
required when a generator is first registered or after output from the unit to the system (i.e. lagging power
significant changes have been made. factor operation). The converse is true if the system
voltage rises. If the generator voltage is maintained near
Many ISOs and Reliability Organizations also require that rated by AVR action, then the limits to be respected will be
Utilities confirm compliance with minimum levels of those shown on the capability curve. Under test conditions,
lagging and leading reactive capability or identify the the situation is different. Unless the Utility can
5
intentionally lower nearby system voltage levels, over- B. Prerequisites and Preparation
excited operation is normally achieved by raising the unit’s Station personnel are responsible for the following
voltage set point. Reactive power will then flow from the preparation:
unit to the system. The limits shown on the capability
curve must still be respected, however, it is most possible • Obtain a copy of the latest reactive capability
that the generator or auxiliary bus voltages will reach their curve for the unit under test and become familiar
maximum allowed values before the stator or rotor with each of the limitations.
capabilities are reached for either over-excited or under- • Ensure that all voltages and currents monitored
excited operation and possibly for both. during the test on the station Human Machine
Interface (HMI) accurately represent the
A OVEREXCITATION R EGION quantities measured from the primary PTs and
CTs. Spot checks should be conducted using a
calibrated meter of each station service bus
B
voltage and the current levels on any critical load.
50 • Ensure that recent calibration records are
available for protective relays (O/C, O/V, U/V) on
any critical loads and unit service buses.
REACTIVE PO WER IN MVAR


SUPPLIED

Calibrate any relays that do not have recent


MVARS

SAF E OPERATION calibration records.


Most of the measurements involve use of existing station
POWER IN M W
transducers and therefore do not introduce any risks other
10 50 85 MW 100 than those normally associated with working in a
ABSORBED

generating station environment. During the course of these


MVARS

tests the unit will be run at the extremes of its normal


continuous capability and all participating staff should be
C
vigilant for any sign of potential problems such as
overheating. It is recognized that Operators may not be
D
called upon to operate units at the limits of their reactive
50
UNDER EXC ITATION R EGION
capability on a routine basis and may require the support of
their Engineering Staff to properly prepare for these tests.

C. Measurements and Operation


Fig. 3. Capability curve and steady-state measurement The on-line measurements are performed with the unit
points (100 MVA, 0.85 PF Unit) synchronized to the electrical network and operating at a
specified active power load. In some cases the number of
For typical transformer reactance levels and constant measurements is minimal. An active power level close to
system voltage conditions it will normally not be possible rated power is specified and the generator reactive power is
to measure both the lagging and leading reactive power adjusted until both extremes of operation are reached. In
limits if these correspond to typical levels of 0.9 pf lagging other cases, the ISO or Reliability Organization may
and 0.95 pf leading. In some cases the transmission require that measurements be performed at several
system Operator will be able to adjust other reactive different active power levels.
resources sufficiently to allow the full range of testing but
this is not typical. It may also be possible to dispatch other At each load level, the generator field excitation is varied
units within the same Plant to obtain the desired range. to change the reactive power output. The measurement
This is most effective when multiple units are connected points would be selected based on the generator capability
together at their low-voltage terminals, sharing a common curve and any other limits such as excitation system limits,
Generator Step-Up transformer. generator ac terminal voltage limits (typically 95% to
105% of rated) and auxiliary supply voltage. Figure 3
Regardless of which limits are reached the Utility should depicts typical measurement points superimposed on the
be prepared to supplement these measurements with unit capability curve.
calculations based on the steady-state representation of the
generator. These can be used to establish the actual As noted earlier, each jurisdiction may have specific
reactive capability of the generator over the full operating requirements however a typical reactive capability test
voltage range. would include up to five different reactive load levels to
allow for extrapolation of results to limiting output
conditions.

6
• unity power factor time manufacturers' data are normally available, and a
• over-excited (reactive power = 50% of maximum manufacturer’s representative is normally on site. The
over-excited requirement) manufacturer may provide schematic or block diagrams for
• over-excited (reactive power = up to100% of the equipment, which is a good starting point. Target
maximum over-excited requirement) simulation models should be chosen to match the
• under-excited (reactive power = 50% of simulation software to be used. Standard models [2] should
maximum under-excited requirement) be used where possible.
• under-excited (reactive power = up to 100% of For model validation purposes, normally two distinct types
maximum under-excited requirement) of tests are required: open-circuit or on-line dynamic
Local guidelines will dictate quantities that should be response tests that allow confirmation of the closed-loop
monitored. Not all measurements will be required or response of the Automatic Voltage Regulator (AVR),
available on all units. Station metering is acceptable as Power System Stabilizer (PSS) and selected limiters, and
long as it has been maintained in reasonable calibration. off-line tests used to measure detailed transfer functions or
The source of each measurement should be noted along measure limiter operating characteristics.
with the test data. The following is a typical list of points The following is a brief overview of the types of tests
to be monitored on a conventional round rotor generator performed, based on the unit operating condition. Each of
connected to a steam turbine: these tests is described in more detail in the following sub-
• ac terminal voltage sections, along with specific examples of equipment
• active and reactive power (generator gross) connections and results. Excitation system testing
techniques are discussed in detail in [3].
• generator field voltage and current (excitation
system)
• Off-line tests are performed on individual
• generator stator temperature (average and
modules within the excitation system while it is
maximum measurement)
energized from test supplies and isolated from the
• generator core temperature (average and field winding.
maximum measurement)
• station ambient and unit cooling temperatures • Open Circuit tests are performed with the
• unit service bus voltages if these are fed from a generator on open circuit running at rated speed.
Unit Service Transformer and are therefore
subject to change during the course of the test. • On-Line tests are performed with the generator
Hydroelectric units normally require fewer measurements synchronized to the grid and operating at a variety
since unit service voltage issues stator core-end heating are of active and reactive power load levels.
rarely a factor.
A. Off-Line Tests
The unit is normally operated under AVR excitation During the off-line tests, the exciter is energized using a
control throughout the test. An Operator should monitor test supply, and test signal generator outputs are
the unit to maintain relatively constant reactive output in substituted for the normal input signals (e.g. three-phase ac
the event that system conditions change during the course voltage input to the voltage regulator, dc millivolt source in
of the test. MVAr control loops can also be enabled for place of field current shunt feedback). Individual modules
this period, as long as they are voltage-supervised, to or sub-modules are isolated and tested separately to
prevent voltage from moving outside of the required limits. validate the derived transfer functions. Step response and
If any tests reveal problems with over or under-excited frequency response techniques are normally used to
operation, the excitation should be adjusted to restore unity measure the small signal performance. Non-linearities in
power factor operation, or operation at rated terminal the electronic controls, such as limits, can be measured at
voltage. The choice of operating level will depend on the this stage by inputting large signal changes at selected
nature of the problem (i.e. voltage or reactive current). inputs.

The station service load fed from the unit under test should Testing of limiter and protective relay operating settings,
match normal operating conditions for the maximum associated with the excitation system, can also be
continuous load level. performed at this stage using test sources as inputs. This is
important since it is not practical to operate the unit on-line
at the high excitation levels necessary to reach the typical
VI. EXCITATION SYSTEM TESTS limiter settings. Testing is complicated with excitation
The testing of excitation systems to validate their limiters since it involves verifying that once engaged, the
performance specifications and to construct models can be limiter is capable of controlling the excitation level in a
a time consuming task. The best time to do this testing is in stable fashion. For summing limiters this involves the
conjunction with the commissioning of a new unit. At this complex interaction with other control loops such as the
AVR and PSS. For this reason, excitation limiter testing is
7
normally performed in two stages. First the operating limit suddenly and provides a good measure of the overall
characteristic (e.g. field current or reactive output points at response of the excitation system.
which the limiters take over control) is measured during
The type of exciter will dictate the method used to inject
off-line tests through secondary signal injection. Once the
the AVR reference step change. On analog-electronic
limit characteristic is known, the limit is temporarily
systems, the signal is normally applied as a low-voltage
adjusted to lower limit settings that permit the limiter to be
(e.g. 0-12 Vdc) signal applied to the input of an
engaged during open-circuit operation or on-line operation
operational-amplifier summing-junction. On modern digital
at low active and reactive power load levels. Under these
systems, the exciter is normally equipped with a built-in
controlled conditions, the control-loop dynamics can be
facility for introducing controlled step changes to the
measured with minimal risk to the unit or power system.
voltage reference signal. On a system equipped with a
A new class of test sets that combine static and dynamic magnetic amplifier, the signal may be applied as a current
closed-loop simulating capability may allow both types of injection into a spare winding. On older systems such as
tests to be performed during outages reducing the need for those equipped with discontinuous regulators, the signal
some or all of the on-line tests. These test sets simulate may be introduced by altering the three-phase PT feedback
the dynamic response of the generator and power system. signals at the AVR input. This approach can also be used
Prior measurements or a detailed representation of the on more modern systems if detailed design information or
generator is necessary for this approach to be used with interfacing software is unavailable. This technique has the
confidence to verify dynamic regulator and limiter added advantage of explicitly including the terminal
response. voltage transducer in the forward path of the disturbance.
Another step response test can also be accomplished by
B. Open–Circuit Tests
transferring from constant field (manual) control mode to
Open circuit tests are performed at rated speed and rated the AVR mode with a small unbalance between the set
voltage with the generator remaining on open circuit. The points. This method has the advantage of not requiring
open circuit tests normally consist of the following stages: equipment to inject a change in AVR reference.

• Steady-state measurements of the exciter and Figure 4 presents a response typical of the changes in
generator quantities. This is often performed terminal voltage that are obtained in the excitation system
concurrent with the main exciter and/or generator tests.
saturation characteristic measurements.
1.00
simulated
Terminal V

• Dynamic tests (time-response or frequency- measured


0.99
(pu)

response) of the closed-loop AVR. The measured


data can be compared with simulations performed 0.98
using the block diagram model.
200
• Dynamic tests of the excitation limiters. The over-
150
excitation limiters, which normally operate to
(Vdc)
Field

100
limit terminal voltage, generator field current
and/or exciter output current, can be tested on 50
open-circuit, by lowering the limit set points to 0
the open-circuit operating levels. The unit's 520
operation is then forced into the limit set point, by
(Adc)
Field

injection of a test signal into the AVR or firing 510


circuit or by simply increasing the AVR setpoint
until the limit is reached.
500
0 2 4 6 8 10
One of the most common of the open circuit tests is the
AVR step response test. This test is specifically identified Time (seconds)
in many North American regulatory compliance
procedures for confirmation of excitation system response. Fig. 4. Typical step response of static excitation system
As a minimum, generator terminal voltage, generator field
An alternative to AVR step response tests are load
voltage, or pilot exciter field voltage for brushless systems,
rejections with the unit absorbing reactive power [20].
should be measured during the application of a step change
Upon opening of the generator circuit breaker, the
signal that will produce a 1% to 2% change in the terminal
excitation system will respond to the step change in
voltage reference. This test is performed by injecting a
reactive current in a manner similar to that seen for a step
small, short duration change in the AVR reference level.
test. The AVR response in the opposite direction can be
This causes the generator terminal voltage level to change
tested with the unit producing reactive power. The amount
of reactive power produced or absorbed prior to opening
8
the circuit breaker will determine the magnitude of the required compensation characteristics, measure the effect
AVR response. This method also has the advantage of not of the settings on the closed-loop response and finally
requiring any equipment to inject a change in AVR perform special measurements which could alert us to
reference, but requires the coordination involved in a load possible side-effects of the AVR and PSS settings during
rejection. On some digital systems care must be exercised both normal and disturbance operating conditions.
in interpreting results since AVR or limiter dynamic References [24 to 26] describe various aspects of testing of
settings or setpoints may be switched when the ac breaker these controls.
is opened.
The selection of phase compensation is critical to the
proper functioning of the stabilizer. As a result it is
C. On–Line Tests
important to have accurate measurements of the phase
On-line tests are performed with the generator relationship between the generator terminal quantities and
synchronized to the grid and operating at different active the stabilizer signals. Any transducers or amplifiers used
and reactive power levels. Among the tests performed for this process must be scrutinized to ensure that they do
with the unit on-line, are the following: not introduce unnecessary filtering and associated phase
lag in the frequency range of interest, 0.1 Hz to 10 Hz.
• Steady-state measurements of exciter and Analog electronic hardware is being superseded by digital
generator quantities. These measurements measurement systems, which incorporate the transducer
supplement the open-circuit measurements, and functions with the data storage and analysis software.
can be performed over a wide range of generator
field current levels and are normally performed at In many cases, some or all of the required signals are
the same time as reactive capability testing directly available from the stabilizer itself, as these
quantities are used in the stabilizing signal generation.
• Dynamic tests (time-response and frequency- Prior to investing in specialized transducers it is
response) of the closed-loop AVR. Many of these worthwhile to review the manufacturer’s manuals and
tests would only be required during schematics showing the available test points.
commissioning or on units equipped with power Two types of tests are normally performed: time-domain
system stabilizers. and frequency-domain. For the time-domain tests, the
quantities should be recorded with a bandwidth of at least
• Dynamic tests of the excitation limiters and other
50 to 100 Hz to ensure that higher-frequency components
features. All of the limiters and other features that
in some of the signals (e.g. turbine shaft torsional
are not operational on open circuit (e.g. UEL,
components in the speed signal) are captured. For the
RCC) are tested at this stage.
frequency-domain tests a transfer function analyzer is very
Dynamic measurement of the field current limiter useful. It should be capable of generating the required
operation is typically performed with reduced limit settings disturbance signal (e.g. swept-sine, pseudo-random binary
as described in the earlier sections on off-line testing. noise sequences, etc.) and of performing input/output
Settings are reduced to avoid excessive bus voltage and transfer functions of any pair of signals. At one time, these
stator current during the performance of the test. Limit pieces of equipment were only available to small numbers
levels are reduced, and a terminal voltage step is applied of specialists; however with the proliferation of computer
which forces the field current into the new limit. and DSP-based test equipment, they should be widely
available to most utilities.
Care must be taken when tuning the UEL dynamic
performance under all circumstances, but particularly when Stabilizer testing must validate several different model
it is used in conjunction with a power system stabilizer. parameters: gains, time constants, limits and calibration of
When testing and calibrating UELs, several factors must be measurement transducers.
considered, all of which suggest performing dynamic tests
If detailed models exist from commissioning tests or
with reduced settings (e.g. with reactive power settings
studies, simple step response tests may suffice to confirm
closer to zero). Low bus voltage will result from operation
continued validity of an existing model. For equipment
at extreme under excited levels and should be avoided. In
where no model exists, bench tests to confirm transducer
some cases, where there is limited margin between the
calibrations and output limits may be required, as the
UEL and loss-of-excitation relay or generator core-end
required operating ranges of inputs and outputs may be
overheating characteristic, there is the possibility of
impossible to test with the equipment in service on line.
entering one of these regions if the limiter does not
Transfer function tests will likely also be necessary for
function as expected.
various stages of the stabilizer characteristic, requiring
access to internal stabilizer signals and appropriate
D. Power System Stabilizers
isolation transducers, with suitable bandwidth as discussed
The techniques used to test the AVR and power system above. As with excitation systems, the techniques and
stabilizer (PSS) mirror the techniques that are used in their details are discussed in the reference [3, 24, 25 and 26].
simulation and tuning. That is, we seek to measure the
9
VII. GOVERNOR TESTS 60.04

A. Permanent Droop: 60.02

Frequency
(Hz)
The extent to which each generating unit responds to 60.00
system frequency variations is determined by its permanent
59.98
droop setting, within its turbine limits. In most
jurisdictions, the governor droop and speed sensing dead 59.96
band are required to be confirmed by test.
95
Permanent droop is often set through a calibrated
adjustment with an indication of the actual value. While 94

Gate
(%)
this calibration is often reasonably accurate, permanent
droop should be measured directly as a confirmation, 93
especially on older mechanical units where changes may
have been made to governor components. 92

This measurement may be performed without the benefit 0.04


of any special equipment. The simplest approach is
0.02

Deadband
described below.

(Hz)
0
With the unit operating off-line, apply excitation such that
the unit is operating close to rated terminal voltage. Adjust -0.02
the speed reference to different settings above speed no- -0.04
load (e.g. +/-4 % in 1% increments) and tabulate the unit’s 0 500 1000 1500 2000 2500 3000 3500
frequency versus reference setting. In cases where the Time (seconds)
speed reference does not have marked settings this will
sometimes involve adding temporary markings to dials or Fig. 5. Governor Deadband
gear wheels.
Clearly in this case the unit's deadband is well within the
Synchronize the unit and adjust the speed reference to the required limits. Care must be taken in interpreting this
values above speed-no-load that were tabulated in the kind of curve to ensure that there is sufficient ambient
previous step. Tabulate the final gate or fuel valve position frequency variation to make a reasonable assessment.
and active power versus speed reference setting.
Although in theory it is possible to introduce very small
Plot the measured off line speed and on line feedback reference changes in modern electronic or digital
signal used for permanent droop for corresponding speed governors this is rarely an effective method of assessing
reference settings. The slope of this curve at any point is the overall deadband.
the permanent droop. The average slope over the entire
operating range is usually quoted as the droop value. C. Hydraulic Governor Dynamic Tests
This test is equally applicable to hydraulic and thermal For hydraulic turbines, it is necessary to provide governor
turbine governors. compensation for stable off-line and islanded speed
governing. Testing this compensation should be performed
B. Deadband: on a routine basis as well as following any major
maintenance outage.
The other quantity that is most frequently requested for all
governors is deadband. Deadband is defined as the largest The following is a simplified test description suitable for
frequency change for which the governor does not respond. use with most mechanical-hydraulic governors for
This can obviously be difficult to assess, especially if hydraulic turbines. The test equipment requirements may
system frequency doe not change sufficiently during on- be quite modest: gate position transducer, hand tools, stop
line monitoring. One method of evaluating governor watch, multimeter capable of measuring frequency.
deadband is to simulate the response of the governor to
Prior to testing, tabulate as-found settings and adjust, if
system frequency changes using the known governor droop
necessary, to match any recommended settings.
and time constants. The "ideal" simulated response is then
subtracted from the actual measured response and Connect slide wire or other transducer to produce a voltage
converted to equivalent frequency as shown in in the signal proportional to servomotor position.
"deadband" plot of Figure 5.
Operate the unit on line with the speed reference at a
calibrated setting (e.g. 2% above speed-no-load). On units
with small versus large system settings, operate with the
small system (damped) settings.

10
Tabulate or record the initial gate position and rapidly are one method used for this purpose. It should be noted
change the speed reference to a new (calibrated) position. that many governors switch modes or set-points when the
Allow the gates to settle to a new steady-state position and unit synchronizing breaker opens. As a result, on many
tabulate the final gate opening. On mechanical systems, units neither on-line dynamics nor droop can be validated
adjust the needle valve opening as required to obtain the from a load rejection test. In this case, some means of
recommended reset time. disturbing the speed reference or one of the feedback
signals must be employed with the unit operating on-line as
To obtain a record of the transient response of the
described above.
governor, follow the steps outlined above and use a
slidewire mounted on the wicket gate servomotor to obtain
D. Thermal Unit Testing and Modeling
a voltage proportional to gate position as shown in Figure
6. Whereas there is a long history of tuning and testing
hydraulic generator governors, there has been less focus on
0.45 testing of steam and gas turbines for tuning and model
validation. For operating security limit studies, the focus
has moved to overall plant frequency response.
0.40
The use of detailed thermal governor models has not
simulated
measured produced load and frequency response simulations which
match measured system events. Reference [14] presents
0.35
the development of a new turbine-governor modeling
Gate Position (pu)

approach that correctly represents thermal units that have


demonstrated unresponsive characteristics such as “base
0.30
loaded” units, or as units with load-controllers. This
approach focuses on the use of data collected from station
0.25
recorders during actual events rather than staged tests.

E. Combined-cycle power plant tests and models


0.20 The reader should refer to [12] for a more detailed account
of combined cycle power plants.
From the perspective of the electrical generator(s), its
0.15
0 5 10 15 20 excitation system, associated supplementary controls,
Time (seconds)
limiters, etc. a combined cycle power plant is no different
than any other similarly size thermal unit. The key
Fig. 6. Mechanical Governor Damped Response difference is in the turbine controls since the plant consists
of the combination of both gas (Brayton) and steam
New governing systems are almost exclusively of the (Rankine) thermodynamic cycles.
electro-hydraulic design. The flyball head speed sensing
system is replaced by speed probes or frequency input Modern combined-cycle power plants use digital turbine
circuits and frequency-to-voltage converter circuitry. The controls. In these systems, governor droop is typically
gate position feedback signal is obtained from a rotational implemented as a constant relationship between turbine
or linear displacement transducer. In some cases electrical speed and generator electrical power (in units of
power is used in place of gate position for on-line megawatts per Hertz). An intentional deadband is often
regulation purposes. The speed reference, measured speed programmed into the controls, to prevent constant motion
and droop feedbacks are combined electronically, either of the fuel valve. For multi-shaft units this droop setting is
with analog circuitry or using a digital implementation. implemented in the governor controls of the individual gas
turbines. The effective droop of the entire plant is a
In this case, speed reference step response tests may be combination of the gas turbine response and the
initiated and recorded in the same way as they are subsequent dependence of steam turbine power on gas
performed for analog electronic or digital voltage turbine output. The steam turbine output follows the gas
regulators, and the same precautions, recording, and turbine, with a large time constant associated with the heat-
analysis techniques apply. recovery steam-generator (the steam turbine is typically
Appropriate simulation models may be found in references operated with valves wide open – under sliding pressure
[13 and 39]. Testing techniques for governor dynamic control). The droop of the gas turbine(s) may be measured
tuning and modeling may be found in references [40-47]. by plotting the steady-state variation in power versus the
governor speed reference and thus determining the slope of
Once individual parameters have been identified, to the this line [12]. For single-shaft units the governor droop
extent possible, an overall test can be performed to verify can be, but is not necessarily, implemented as a
the correct performance of the model. Load rejection tests

11
relationship between the total electrical power of the unit repeat model validation testing need only be performed if
to shaft speed. measured responses to system disturbances disagree with
model predictions.
Off-line speed and on-line power response to step-changes
in the governor reference input may be used to validate the Validation of tunable settings (Table 1, B) should be
time constants and rate limits associated with the gas performed periodically (e.g. following a major unit outage)
turbine fuel valve and turbine [12]. In addition, as determined by experience with the particular equipment
particularly for multi-shaft units, the time constant in use and every time a change or upgrade is made. The
associated with the heat-recovery steam-generator may be history of the equipment, such as component failures and
estimated by effecting a change in the gas turbine power the need for adjustments should be used to determine when
and recording the corresponding response in the power re-testing of a model may be warranted.
output of the steam turbine. During this test the steam
Validation of equipment characteristics set by design
valves should be kept wide open (in sliding pressure mode)
(Table 1, A) are typically only performed during
and the gas turbine kept in its new steady-state condition
commissioning or at the start of a test program. Simple re-
for many minutes while the steam turbine is allowed to
verification tests may be performed occasionally or when
fully respond. The gas turbine(s) in many modern
evidence of changes to unit behavior occurs.
combined-cycle power plants will have an outer-loop
megawatt controller that will act to maintain a pre-selected Rotating Equipment –Most generator and rotating exciter
megawatt output level on the gas turbines as specified by model data is fixed by design. Generator inertia will not
the operator. This control loop will effectively override the change unless generator or turbine modifications are made,
initial response of the turbine droop-governor in the event and thus repeated testing is not a necessity. Generator
of a system disturbance bringing the unit’s output back to rewinding is not expected to introduce changes to the
the pre-selected level [12, 14]. Ideally, this loop should be generator impedances; however periodic reactive capability
disabled on units participating in primary frequency tests and confirmation of the open circuit saturation
regulation. characteristic will reveal unexpected reductions in unit
capability, for instance, shorted rotor turns, or control
The maximum power output of a gas turbine, whether in
system automation enforcement of operational limits.
combined-cycle power plants or operated as simple-cycle
Reactive capability verification is mandatory in many
units, is dependant on both the operating speed of the unit
jurisdictions.
and ambient air conditions [12, 27 and 28]. In some
applications, inlet-air may be kept at near constant Closed-Loop Controls –Whenever the controls are
conditions by air-conditioning units to minimize variations modified or upgraded a new model may be needed and
in the turbine capability due to ambient conditions – this is model validation testing will need to be performed.
done at the expense of overall plant efficiency. Where Equipment technology plays a large role in determining the
such inlet-air conditioning is not performed and the gas necessity and frequency of testing.
turbine capability varies significantly with ambient
Older mechanical or magnetic amplifier equipment is more
conditions, manufacturer data should be sought to estimate
likely to be refurbished, has fewer calibrated settings and is
the turbine megawatt capability under various ambient and
more susceptible to drift than more modern equipment, and
frequency conditions for system studies.
hence should be tested following major outages.
Both analog electronic and digital electronic equipment
VIII. TIMING OF VALIDATION TESTING
rely on input and output electronics which require
Timely validation of computer models is necessary to calibration and may exhibit drift. In these cases, equipment
ensure their continued validity. Equipment wear, upgrading history may provide the best guide for frequency of testing.
and refurbishment, component drift, adjustments to
settings and configuration management all contribute to Digital electronic equipment may have drift-free settings,
possible changes in the dynamic response of the unit. but may be subject to change control issues (modification
Guidelines for frequency of model validation testing are of settings), which should be periodically checked. This
summarized below. equipment typically has both data logging and testing
facilities built-in, which remove two of the barriers to re-
Model validation testing should be part of equipment verification tests.
commissioning. If model validation was not performed
during commissioning then it should be done as soon as
practical. IX. AMBIENT MONITORING
The concept of an ambient test methodology involves
The measured responses to actual power system
passive monitoring of generator or system events rather
disturbances should be compared with simulations
than active testing. In this case, the recorded responses of
performed using the models, and differences used to
normal or abnormal system or plant events may provide
prioritize testing. If no modifications have been made to
sufficient data for model validation.
the equipment since the last model validation test, then

12
In conventional dynamic performance testing, specialized the limiter setting for this test to avoid possible instability
equipment is temporarily connected to the equipment being or tripping generator auxiliaries on low voltage.
tested and the unit is then subjected to some form of Differences between dynamic responses at different
artificial step change in one of its input signals in order to operating levels can be simulated.
observe the response.
Other functions, such as over excitation limiters, may be
For ambient testing, the design concept of the equipment tested by simulating the input signal with the machine shut
would be more task-specific, hopefully reducing its cost to down.
the point that it could be left connected to the unit for
Step responses with the generator operating at rated speed,
longer periods or even permanently. This test equipment
not synchronized are a good check on the dynamic
would record unit operation under normal operating
performance of the voltage regulator/exciter. A change in
conditions and buffer the data for post-event retrieval, or
the generator voltage of approximately 1-2% should be
possibly trigger the capture of data when system conditions
sufficient. An exception is some older voltage regulators
are such that dynamic performance can be adequately
that had different gains for different-sized errors; in such
measured and evaluated.
cases, a larger step may be necessary to check the alternate
It is hoped that the development of an ambient test gain. In either case, the size of the step should be increased
methodology would fulfill some of the objectives of gradually from zero to avoid unsafe conditions.
equipment performance testing without incurring the costs
Step response tests are also required to check the dynamic
and risks associated with scheduled testing.
performance on load. As with the open-circuit tests, the
Ambient testing methods are presently evolving and not yet size of the step should be increased gradually from zero.
fully established methods, although their use has already
If frequency response tests are done with the generator
proved valuable in some cases [14, 60 and 61], and are
synchronized, care must be taken to avoid shaft torsional
being encouraged as acceptable means of model validation
frequencies. The generator manufacturer can usually
[32].
provide the necessary information.
One area where ambient measurements have provided
Care must be taken when performing load rejection tests to
useful results is in governor droop and dead band
ensure that an over speed condition is not reached. This
assessment. High-resolution measurement of system
can be done by simple calculations prior to the test, based
frequency and measurement of unit output (gate position,
on the expected unit inertia and governor response.
fuel valve position and/or active power) are required, but at
low sampling rate (see Figure 5). System frequency
disturbance data may then be analyzed to confirm that the XI. RECOMMENDED BUILT-IN TEST FACILITIES IN NEW
unit meets system requirements for frequency response. EQUIPMENT
With most generator excitation control systems now being
X. PREPARATION, TEST SET-UP AND PRECAUTIONS TO BE implemented using digital technology, opportunities exist
TAKEN DURING TESTING to design equipment with useful features to aid the
validation testing process.
Testing should be undertaken with knowledge of available
industry standards where possible, such as those listed in When testing such systems the commissioning engineer is
the references [2-11, 40]. normally able to see the intended settings displayed on
either a built-in display or on a laptop computer. However,
If offline exciter or PSS tests are planned, preparation is
simply examining settings in this way can only be
required to safely energize the various exciter test supplies.
considered as a quick and easy check to ensure that the
The ease with which voltage regulator functions can be
system has correctly retained programmed settings.
enabled in this state varies from machine to machine, but
the tester should be prepared to deal with numerous Built in test features should normally include the ability to
interlocks that will have to be bypassed to permit shutdown assign analog input and output signals to the inputs and
testing. outputs of the various internal stages of the control system.
This allows separate verification of the operation of each
Some tests must be done with the generator running. In
stage and facilitates confirmation of the time constants and
these cases, personnel who are familiar with this type of
gains.
testing should be present. Dangerous voltages are present
in the generator field circuit where transducers need to be It can be very useful to have an internal signal generator
connected. Care should be taken to avoid introducing noise incorporated into the equipment. The internally generated
into high gain circuits when test leads are connected or signal may then be used as an alternative to an analog test
disconnected. signal generated using separate equipment. It should be
possible to easily adjust the frequency, amplitude and wave
The dynamic performance of some equipment, such as
shape of the internally generated signal.
under excitation limiters, will probably have to be tested
with the generator on load. It is often advisable to reduce
13
Another ideal use of digital technology to aid validation More work is required by both generators and market
testing is the addition of data recording facilities that operators to determine a suitable threshold for unit size,
enable sampled waveforms to be viewed on laptop below which performance testing is not required.
computers. When this is provided it should be possible to
XIII. CONCLUSIONS
save the sampled data for later reference. This facility
should be provided with flexibility to allow signals to be This paper outlines the general approach and guidelines for
recorded from various stages in the control system, with field testing of generating equipment for the purpose of
various sampling periods and with various archiving and deriving and verifying parameters for computer simulation
triggering options. models of the power plant equipment. This document is a
brief outline of the methodologies used. The reader should
While it is recommended that the internal recording and
refer to the many references outline through the discussion
signal generation features should be incorporated into
for further detail on the testing procedures and techniques.
equipment, on occasions it may be necessary to use
separate signal injection and data recording equipment to
provide independent verification of the operation of XIV. ACKNOWLEDGEMENTS
excitation control equipment. The working group would like to thank the many
participants who contributed to this effort both at
XII. DEVELOPMENT OF CRITERIA FOR DETERMINING A committee meetings and via email correspondence.
MINIMUM THRESHOLD UNIT SIZE FOP PERFORMING TESTING
Typically, units connected to a wholesale market grid tend XV. REFERENCES:
to be large. Smaller units are often connected into local [1] Byerly R.T. and Kimbark, E.W. Ed., Stability of Large Electric Power
distribution systems, which typically do not have the same Systems. New York: IEEE Press, 1974.
requirements for equipment performance testing. [2] IEEE 421.5-2005 Recommended Practice for Excitation System
Models for Power System Stability Studies - models
However, for every rule, there is an exception and some
small units or stations can be found connected directly to [3] IEEE 421.2-1990 Guide for Identification, Testing and Evaluation of
wholesale market grids. In some cases, the cost of periodic the Dynamic Performance of Excitation Control Systems - testing
performance testing of these smaller stations could [4] IEEE Std 115-1995 IEEE Guide: Test Procedures for Synchronous
adversely affect the economic viability of continued Machines
operation within the market. At the same time, the smaller [5] IEEE Std 67-1990 IEEE Guide for Operation and Maintenance of
the units, the less the impact they have on overall system Turbine Generators
performance. [6] IEEE Std 492-1999 IEEE Guide for Operation and Maintenance of
Hydro Generators.
Therefore as unit size decreases a) the benefits to the
market from performance testing decreases while b) the [7] IEEE Std 1110-1991 IEEE Guide for Synchronous Generator
Modeling Practices in Stability Analyses - models; testing
cost of these tests per MWhr delivered increases. Clearly,
at some threshold of unit size, performance testing is no [8] ANSI Std. C50.10-1977 General Requirements for Synchronous
longer viable. Machines
[9] ANSI Std. C50.12-1982 Requirements for Salient-Pole Synchronous
The cost of conducting testing can be quantified relatively Generators for Hydraulic Turbine Operations
easily. This includes the actual cost of staff engaged to
[10] ANSI Std. C50.13-1989 Cylindrical-Rotor Synchronous Generators
conduct the testing plus the lost opportunity cost of not
having bid the unit optimally into the market. It should be [11] ANSI Std. C50.14-1977 Requirements for Combustion Gas Turbine
kept in mind, however, that these costs might be reduced in Driven Cylindrical Rotor Synchronous Generator.
the re-verification phase of testing. For instance, suitably [12] CIGRE Technical Brochure 238, Modeling of Gas Turbines and Steam
trained/knowledgeable local site staff might be substituted Turbines in Combined-Cycle Power Plants, December 2003.
for testing specialists. Also, lost opportunity costs might [13] "Dynamic Models for Steam and Hydro Turbines in Power System
be reduced as testing schedules are shortened. Studies", IEEE Committee Report, IEEE Trans, Vol PAS-92, Nov-Dec
1973, pp. 1904-1915.
More difficult is the assessment of the impact on overall
[14] L. Pereira, J. Undrill, D. Kosterev, D. Davies, and S. Patterson, “A
system reliability, especially in trying to express this in New Thermal Governor Modeling Approach in the WECC”, IEEE
terms of a financial benefit. Generally, system impact can Trans. PWRS, May 2003, pp 819-829.
be taken as meaning the impact on the complete system
[15] Generator Test Guidelines, WSCC Control Work Group and Modeling
within the market operator’s jurisdiction and beyond. & Validation Work Group, March 1997.
However, in some cases, what might be considered a non-
[16] L. N. Hannett and J. W. Feltes, "Testing and Model Validation for
critical unit on such a system wide basis might be critical Combined-Cycle Power Plants," in Proc. 2001 IEEE Power
on a local basis. Engineering Society Winter Meeting Conf., pp. 664-670, vol. II.
[17] F. P. de Mello and J. R. Ribeiro, “Derivation of Synchronous Machine
Parameters from Tests”, IEEE Trans. on Power Apparatus and
Systems, vol. 96, no. 4, pp. 1211-1218, July/August 1977.

14
[18] Determination of Synchronous Machine Stability Study Constants [37] P. Kundur and P.L. Dandeno, "Implementation of Advanced Generator
Volume 3. EPRI EL-1424, Palo Alto, CA, June 1980. Models into Power System Stability Programs", IEEE Trans. on Power
Apparatus and Systems, Vol PAS-102, No 7, July 1983, pp 2047-54.
[19] F. P. de Mello and L. N. Hannett, “Determination of Synchronous
Machine Electrical Characteristics by Tests,” IEEE Trans. on Power [38] R.H. Park, “Two-Reaction Theory of Synchronous Machine -
Apparatus and Systems, vol. 102, no. 12, pp. 3810-3815, December Generalized Method of Analysis - Part I”, AIEE Trans. Vol. 48, 1929,
1983. pp. 716-727.
[20] L. N. Hannett and J. W. Feltes, “Derivation of Generator, Excitation [39] IEEE PES Working Group on Prime Mover & Energy Supply Models
System and Turbine Governor Parameters from Tests,” presented at the for System Dynamic Performance Studies, "Hydraulic Turbine and
CIGRÉ Colloquium on Power System Dynamic Performance, Turbine Control Models for System Dynamic Studies”, IEEE
Florianópolis, Brazil, 1993. Transactions on Power Systems, Vol 7, No 1, Feb. 1992, pp.167-79.
[21] P. L. Dandeno et all., “Experience with Standstill Frequency Response [40] "International Code for Testing of Speed Governing Systems for
(SSFR) Testing and Analysis of Salient Pole Synchronous Machines”, Hydraulic Turbines", IEC Publication 308, 1970.
IEEE Trans. on Energy Conversion, vol. 14, no. 4, pp. 1209-1217,
December 1999. [41] J.M. Undrill and J.L. Woodward, “Nonlinear Hydro Governing Model
and Improved Calculation for Determining Temporary Droop", IEEE
[22] P. A. E. Rusche, G. J. Brock, L. N. Hannett and J. R. Willis, “Test and Trans., Vol. PAS-85, pp. 750-756, July 1966.
Simulation of Network Dynamic Response using SSFR and RTDR
Derived Synchronous Machine Models,” IEEE Trans. on Energy [42] J.L. Woodward, 'Hydraulic-Turbine Transfer Function for Use in
Conversion, vol. 5, no. 1, pp. 145-155, March 1990. Governing Studies", Proc IEE, Vol 115, No. 3, March 1968.

[23] G. R. Bérubé and L. M. Hajagos, “Testing and Modeling of Generator [43] P.L. Dandeno, P. Kundur, J.P. Bayne, "Hydraulic Unit Dynamic
Controls,” presented at SERC Generator Testing Workshop, Atlanta, Performance Under Normal and Islanding Conditions - Analysis and
GA, December 2000. (http://www.kestrelpower.com/Docs) Validation", IEEE Trans., Vol PAS-97, Nov-Dec 1978, pp 2134-2143

[24] Application of Power System Stabilizers for Enhancement of Overall [44] L.N. Hannett, B. Fardanesh, “Field Tests to Validate Hydro Turbine-
System Stability, P. Kundur, M. Klein, et al, IEEE Trans. on Power Governor Model Structure and Parameters”, IEEE Trans. on Power
Systems, Vol. 4, May 1989, pp 614-626. Systems, Vol. 9, No. 4, November 1994.

[25] Applying Power System Stabilizers, Parts I, II and III, E.V. Larsen, [45] S. Hagihara et al, “Stability of a Hydraulic Turbine Generating Unit
D.A. Swan, IEEE Trans., Vol. PAS-100, June 1981, pp 3017-3046. Controlled by PID Governor”, IEEE Trans., Vol. PAS 98, No. 6, pp
2294-2298, Nov/Dec 1979.
[26] IEEE Tutorial Course: Power System Stabilization Via Excitation
Control, Sponsored by the IEEE Power Engineering Education [46] F.R. Sckleif, G.E. Martin, R.R. Angell, “Damping of System
Committee, Center, Piscataway, NJ. Oscillations with a Hydro Generating Unit”, IEE Trans., PAS-86,
1967, pp 438-442.
[27] K. Kunitomi, A. Kurita, H. Okamoto, Y. Tada, S. Ihara, P. Pourbeik,
W. W. Price, A. B. Leirbukt and J. J. Sanchez-Gasca, “Modeling [47] J.C. Agee, G.K. Girgis, “Validation of Mechanical Governor
Frequency Dependency of Gas Turbine Output”, Proceedings of the Performance and Models Using an Improved System for Driving
IEEE PES Winter Meeting, Jan 2001. Ballhead Motors”, IEEE Trans. on Energy Conv., Vol. 10, No. 1,
March 1995.
[28] P. Pourbeik, “The Dependence of Gas Turbine Power Output on
System Frequency and Ambient Conditions”, paper 38-101, [48] EPRI Report EL-1424, "Determination of Synchronous Machine
Proceeding of CIGRE Session 2002, August 2002, Paris, France. Stability Constants", Vol. 2, prepared by Ontario Hydro, Dec. 1980.

[29] P. Pourbeik, C. E. J. Bowler and V. L. Crocker, “Model Validation [49] M.E. Coultes and W. Watson "Synchronous Machine Models by
Testing for the Purpose of Determining Generation Equipment Standstill Frequency Response Tests", IEEE PAS-100, April 1981, pp.
Dynamic Performance and Torsional Mechanical Response”, 1480-1489.
Proceedings of IEEE PES General Meeting, June 2004, Denver, [50] P. L. Dandeno and A. T. Poray, "Development of Detailed
Colorado. Turbogenerator Equivalent circuits from Standstill Frequency
[30] L. M. Hajagos and G. R. Berube, “Utility Experience with Gas Response Measurements", IEEE Trans. Vol. PAS-100, April 1981,
Turbine Testing and Modeling”, Proceedings of the IEEE PES Winter pp. 1646-1653.
Power Meeting, January 2001. [51] P. A. E. Rusche, G. J. Brock, L. N. Hannett and J. R. Willis, “Test and
[31] "Testing Methods - An Overview," presented at SERC Generator Simulation of Network Dynamic Response using SSFR and RTDR
Testing Workshop, Atlanta, GA, December 2000 (http://www.pti- Derived Synchronous Machine Models,” IEEE Trans. on Energy
us.com/pti/company/technical_papers.cfm). Conversion, vol. 5, no. 1, pp. 145-155, March 1990.

[32] Generator Test Policy documents at Western Electricity Coordinating [52] M.E. Coultes, "Standstill Frequency Response Tests", presented at the
Council web site, www.wecc.biz. IEEE Symposium on Synchronous Machine Modeling for Power
System Studies (IEEE Power Engineering Society 1983 Winter
[33] G. Shackshaft, "New Approach to the Determination of Synchronous Meeting), IEEE Publication 83TH0101-6-PWR, pp. 26-30.
Machine Parameters from Tests". Proc. IEE (London), 1974,
121(11), pp. 1385-1392. [53] M.E. Coultes, P. Kundur and G.J. Rogers. "On-Line Frequency
Response Tests and Identification of Generator Models", presented at
[34] Y. Takeda and B. Adkins, "Determination of Synchronous Machine the IEEE Symposium on Synchronous Machine Modeling for Power
Parameters Allowing for Unequal Mutual Inductances", Proc. System Studies (IEEE Power Engineering Society 1983 Winter
IEE(London), Vol. 121. No. 12, Dec 1974, pp. 1501-1504. Meeting), IEEE Publication 83TH0101-6-PWR, pp. 38-42.
[35] G. Shackshaft and A.T. Poray, "Implementation of New Approach to [54] P.L. Dandeno, P. Kundur, A.T. Poray and H.M. Zein El-Din,
Determination of Synchronous Machine Parameters from Tests", Proc. "Adaptation and Validation of Turbogenerator Model Parameters
IEE (London), 1977, 124(12) pp. 1170-1178. Through On-Line Frequency Response Measurements". IEEE Trans.
Vol. PAS-100, April 1981, pp. 1656-1661.
[36] F.P. deMello and L.N. Hannett, "Validation of Synchronous Machine
Models and Determination of Model Parameters from Tests", IEEE
Trans. Vol. Pas-100, February, 1981, pp. 662-672.

15
[55] P.L. Dandeno, P. Kundur, A.T. Poray and M.E. Coultes, "Validation of
Turbogenerator Stability Models by Comparison With Power System
Tests", IEEE Trans. Vol. PAS-100, April 1981, pp. 1637-1645.
[56] R.P. Schulz, W.D. Jones and D.N. Ewart, "Dynamic Models of
Turbine Generators Derived from Solid rotor Equivalent Circuits",
IEEE Trans. Vol. PAS-92, May/June 1973, pp. 926-933.
[57] I.M. Canay, "Extended Synchronous Machine Model for the
calculation of Transient Processes and Stability", Electrical
Machines & Electromechanics, 1:137-150, 1977.
[58] J.W. Dougherty and S.M. Minnich, "Finite Element Modeling of Large
Turbinegenerators: Calculations versus Load Test Data", IEEE Trans.
Vol. PAS-100, August 1981, pp. 3921-3929.
[59] L. N. Hannett and J. W. Feltes, "Testing and Model Validation for
Combined-Cycle Power Plants," in Proc. 2001 IEEE Power
Engineering Society Winter Meeting Conf., pp. 664-670, vol. II.
[60] D. Kosterev, "Hydro Turbine-Governor Model Validation in the
Pacific Northwest," IEEE Trans Power Systems, Vol. 19, No. 2, pp.
1144-1149, May 2004.
[61] B. Agrawal, D. Kosterev, “Model Validation Studies for a Disturbance
Event on June 14, 2004 in the Western Interconnection”, submitted for
IEEE PES Transaction publication, December 2006.
[62] IEEE Guide for AC Generator Protection, IEEE Standard C37.102-
1995, December 1995.

16

You might also like