1

Abstract— This paper constitutes the final report of the IEEE Task Force on Generator Model Validation Testing, which was created in 2002 and served under the Power System Stability Subcommittee. The objective of the TF was to provide technical guidance to utilities undertaking testing and modeling of generation facilities. This paper outlines accepted industry methods for establishing dynamic models for generators and their control systems.

Index Terms—generator modeling, model validation, field testing of power plants

I. INTRODUCTION

Few people disagree that good quality simulation models of power system equipment is beneficial to all power system participants. There are, however, differing opinions on the required level of detail, best methods to obtain data, and frequency of verification. The goal of this paper is not to prescribe one answer to any of these issues. Rather, it is to identify a body of available techniques and describe the benefits and drawbacks of each, to help industry participants determine the best approach for their types of equipment, capabilities and objectives.

The purpose of the model validation is to ensure the proper performance of the control systems and validate the computer models used for stability analysis.

Specific goals include the measurement and verification of the following:

• generator reactive capability

• generator dynamic model data

• excitation system, stabilizer and limiter models

• turbine/governor models

• coordination between protective devices, control system limiters and equipment capability

In addition to obtaining model data, the tests performed to gather this information may uncover latent defects that could lead to inappropriate unit response during system disturbances, thereby improving the reliability of the unit and the power system.

*Task Force Members: Les Hajagos (chair), Joel Barton, Roger Berube, Murray Coultes, Jim Feltes, Gary Lanier, Shawn Patterson, Les Pereira, Pouyan Pourbeik, Alex Schneider, Robert Thornton Jones

II. SIMULATION MODELS

Power system planning and operational studies require the simulation of the response of synchronous machines and their respective control systems. For these studies it is essential that the control systems of the synchronous machines be modeled in sufficient detail (see [1]). The desired models must be suitable for representing the actual equipment performance for large, severe disturbances as well as for small perturbations. To obtain accurate simulations, not only must the models be of an adequate level of detail, but the values of the parameters in the models must also correspond to actual field values.

The equipment to be tested and modeled includes the generator and its control systems; excitation systems, power system stabilizers and turbine governors. Protective relay coordination with equipment capabilities and control system limiters is equally important and is discussed in detail in appropriate standards [2-11, 62].

To some extent, the model structures used in system studies are intended to facilitate the use of field test data as a means of obtaining model parameters. The models are, however, reduced order models and they do not necessarily represent all of the control loops on any particular system. In some cases, the model may represent a substantial reduction, resulting in large differences between the structure of the model and the physical system. Model structures are shown in the references [2, 7, 12, 13 and 14].

Useful models should be valid for frequency deviations of +/-5% from rated frequency and oscillation frequencies up to about 3 Hz. Voltage deviations are typically between 95% and 105% for steady-state generator operation, but the models must be capable of representing large-signal disturbance performance. The models discussed here would not normally be adequate for use in studies of sub synchronous resonance or other shaft torsional interaction problems. Slow-acting control and limiter functions (such as tap-changing transformers and generator field current limiters) that may come into play in long term dynamic performance studies are required for some studies and are included.

III. REVIEW OF EXISTING DATA AND MODELS

Prior to undertaking a testing program, a thorough review should be made of existing simulation models,

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

Guidelines for Generator Stability Model Validation Testing

1-4244-1298-6/07/$25.00 ©2007 IEEE.

2

manufacturer’s data sheets, and control and protective equipment tunable settings. Priority should be given to testing parameters which are tunable, or which may have been changed by refurbishment, maintenance or replacement. Often, new excitation or governing controls are installed without corresponding changes to simulation models. Conversely, some model parameters, such as unit inertia, may be fixed by the system design, and once measured or calculated, may not change except by equipment modification. Examples of model parameters to be validated are shown in Table 1.

Some common mistakes that may be found in existing database models are as follows:

• Use of typical or bid data rather than as-built data. • Incorrect scaling of data based on changes to

ratings, most commonly up-rating of unit apparent power

• Use of saturated instead of unsaturated generator parameters

• Inertia not including the contribution of the turbine

• Exclusion of features such as reactive current compensation in excitation systems

IV. DESCRIPTION OF TESTING AND VALIDATION METHODS

This section will focus on the testing methods generally employed to determine model parameters for generators, excitation systems, and governors.

Existing methodologies for generator testing can be broadly divided into two categories:

• Time-response tests, involving step changes to controller set points, partial load rejections and other disturbance techniques with the machine operating at pre-defined conditions [16, 17, 18, 19, 20, 29, 30]; and

• Frequency response tests conducted with the unit at standstill [21, 22] or in operation.

These methodologies are well established and can produce accurate models. The major drawbacks are associated with the time the machine is unavailable for normal commercial operation (economics) and the technical problems that may arise during the tests. In some cases the technical problems are associated with the discovery of latent defects. In this case, although problems may arise during testing, this is considered preferable to the consequences that could occur during an actual system disturbance.

Note that the tests and methodologies defined in equipment standards [3 through 11, 60] are not specifically designed to determine model parameters, but generally to specify a desired level of performance and usually a methodology to measure that performance. While these tests can give very useful information for deriving parameters, it is not their main focus to determine the complete set of parameters

needed. In addition, many of the tests are designed to be performed by the manufacturers and do not lend themselves to application in the field to installed equipment. With the increasing usage of sophisticated computer-based design tools, detailed commissioning tests of this type are often only performed as type tests on the initial design.

TABLE 1. MODEL PARAMETERS

Conditions Generator Exciter Power System

Stabilizer

Governor

Reactances, Limits, Water Starting Time,

Time Constants,

Exciter time constant,

No Load Gate,

Inertia, Voltage sensing time

constant,

Full Load Gate,

Saturation Saturation (rotating exciter)

Turbine Power

Fractions,

Maximum Power,

Dead band,

A. Set by design (not changed unless rebuilt)

Input type(s)

Turbine damping,

Voltage Regulator

Gains,

Gains, Droop,

Time Constants,

Time Constants,

Time Constants,

Reactive Compensation

Limits Gains,

B. Tunable

Limit Levels Rate Limits

A. Synchronous Generator Tests

Historically, manufacturers performed sudden short circuit tests to determine dynamic synchronous generator parameters. These test procedures are specified in IEEE Standard 115-1995 [4]. These tests provide only the d-axis transient and sub-transient constants. In addition, they do not include measurement of field current during the short circuit tests and consequently the field circuit is not specifically identified. The limitations of these procedures for providing data suitable for stability studies have been recognized for some time.

Several different testing and analytical methods have been proposed and used to obtain better models. These include:

• Enhanced sudden short-circuit tests

3

• Partial load rejection tests

• Frequency response tests

o Standstill frequency response

o Open-circuit frequency response

o On-line frequency response

• Analysis of design data, e.g., finite element analysis.

These improved methods of obtaining machine characteristics are briefly discussed below.

1) Enhanced Short Circuit Tests

References [33] and [34] describe improved methods of utilizing results from sudden short-circuit tests to determine more accurate d-axis parameters. The most important feature of these methods is the utilization of rotor current measurements during the short circuit tests to identify the field circuit characteristics more accurately.

Among the disadvantages of the short-circuit approach are: the inability to provide q-axis parameters, the necessity for subjecting the machine to a severe shock imposed by the sudden short circuit and the complexity of setting up this test.

2) Partial Load Rejection Tests

Partial load rejection tests are among the most commonly employed means of validating generator model parameters for in-service generating units. They have been performed on hundreds of units and regional testing programs, such as the ongoing program in the WECC [15], have been primarily accomplished with these techniques. This method is one of the simplest approaches to developing an adequate equivalent circuit model, or for verifying manufacturer’s data.

The load rejection dynamic tests are selected to provide a simple and safe disturbance to excite the response of the unit and, as much as possible, limit the set of parameters affecting the response to simplify the derivation process. The usual approach is a series of load rejection tests with the unit initially carrying partial load. This approach is an extension of the work documented in References [17, 18, 35, and 36]. References [19, 20, 23, 31 and 37] give more details on the application of such tests and the procedures involved.

The initial conditions for the load rejections are selected to 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)

control, the main generator breaker is opened which results in decay in terminal voltage. As the generator dynamic response is only a function of the generator d-axis time constants and reactances, this test can be used to determine those generator parameters. Figure 1 shows a typical response for this test.

1 2

1 4

1 6

1 8

2 0

2 2

2 4

2 6

0 5 1 0 1 5 20 25 3 0

T im e (S e c o n d s )

Ter

min

al V

olta

ge

(kV

)

Fig. 1. Typical response for d-axis test showing decay in terminal voltage

Data is generally recorded using digital recording equipment. This allows direct comparison of measurements with simulations. Simulations are performed and the generator model parameters are adjusted iteratively to produce a match. The advantage of this approach is that its setup is relatively simple and it does not require a high degree of training to execute. Some of the drawbacks with this approach include the need for a manual exciter control that maintains constant field voltage, and the need to overcome plant controls that may prevent completely unloading the unit or operating at a low off-line voltage for an extended period.

If power angle can be measured, the q–axis parameters can be measured in a manner similar to that used in the test employed for the d-axis. By matching the power angle relative to terminal voltage with the power factor angle, the steady state conditions for the q-axis test are represented by the phasor diagram shown in Figure 2, which reveals that the armature current is aligned with the quadrature axis. This requires a generator output in the neighborhood of 10 percent of rated output and a reactive power absorption of a few percent of rated. As in the d-axis test, the generator is brought to the desired loading, excitation placed on manual and the main generator breaker is opened, resulting in a response in terminal voltage. Since the unit is producing real power, the partial load rejection will cause a rise in generator speed and hence care must be taken to select the initial loading to limit over speed and also the resulting voltage change.

4

V~

I~

IRa

~

IjX q

~

q axis

d axis

δ

Fig. 2. Phasor diagram for quadrature axis test

In practice, load rejection tests may be repeated to characterize the dynamic response for the excitation system and turbine/governor if generator field voltage and governor signals are monitored. Although this approach simplifies testing since it does not require a detailed knowledge of each Manufacturer’s control design, it has some drawbacks. In particular on newer systems, on-line and off-line response may not be identical as the systems may switch state depending on the status of the generator breaker. Also these tests are not likely to fully exercise all parts of the model and provide sufficient data for complete model verification. They may however be useful as part of a re-verification process once detailed models have been developed.

3) Frequency Response Tests

a) Standstill Frequency Response (SSFR)

In the SSFR technique, all tests are conducted with the unit at rest and disconnected from the bus. The rotor must be aligned to two particular positions with respect to the stator during the tests; because of this, it is normally only performed on round-rotor (thermal) generators. The stator is excited by a low level source over the range of frequencies from 1 mHz to 1 kHz. The frequency responses from the stator winding with the field open and shorted are analyzed to derive the parameters for a higher order d- and q-axis model.

Because the tests are conducted at very low flux levels, the results must be corrected to bring them from the "toe" of the saturation curve to normal unsaturated levels. This is done by minor adjustment of the mutual reactance in each axis.

This technique has been extensively used on round rotor generators [48, 49, 50, and 51]. In addition, it is now part 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 tests on three large units are given in reference [50].

The advantage of this approach is that it can be performed 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 analyze the results.

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 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 this response and the equivalent one from the standstill tests gives some indication of rotational effects.

The test is normally done at more than one voltage to examine saturation effects. By conducting the test with various signal amplitudes, slot wedge conduction effects can be assessed.

c) On Line Frequency Response (OLFR)

In many respects, on line frequency response testing is the "proof of the pudding" as far as small signal verification machine models is concerned. Here the machine is being tested under the same conditions as the model is expected to perform, although over a restricted operating range.

For this test, the machine is operated near rated (or at reduced) load preferably over substantial impedance to the system. The excitation is modulated either by sinusoidal or random noise. Components are resolved on the two axes and data similar to that of the SSFR tests is used to derive a model.

The frequency range of usable data in this test is more limited than that of the SSFR tests, but the SSFR data is expected to be good at the frequency extremes. The OLFR testing allows the middle of the frequency range to be filled in with data for both axes that include rotational effects.

The disadvantage of this test is that it requires testing on an operating unit connected to the system, possibly under special system or unit conditions. In addition, it does not provide large signal response information. Care must be exercised to avoid exciting critical unit or system frequencies.

Details of OLFR measurement techniques and model identification procedure are given in reference [53].

5

Reference [54] gives results of OLFR tests on two large thermal generators and improved models developed from the tests.

In reference [55], models for three large generators based on SSFR and OLFR tests are validated by comparing the results of simulations with measured responses involving line switching. For one of the generators; the models derived from short circuit tests and decrement tests are also validated. The results show that models derived from frequency response measurements are generally superior.

d) Calculation of Parameters from Design Data

References [56, 57] describe improved generator models developed from design information. Reference [58] presents more recent work using finite element analysis. Generator manufacturers use these methods almost exclusively today, rather than tests, when supplying model data for their products.

Work done on two and three dimensional finite element models in the steady-state have shown good agreement with measured results. In addition, two dimensional finite element models have been developed which provide frequency response data that compares well with measured results. The degree to which conduction occurs across slot wedges appears to have a significant effect on the model at higher frequencies. The significance of rotational effects and the effect of disturbance amplitude on the model require additional investigation.

This approach requires detailed involvement by the Manufacturer and may not be possible for older units unless original design material is still readily available.

V. REACTIVE CAPABILITY TESTING

Generator steady state measurements normally consist of tests to either determine or validate the synchronous reactances and saturation data associated with the generator and tests to confirm the reactive capability of the unit including auxiliaries. The modeling tests involve:

• Off-line measurements to measure the open-circuit saturation curve of the generator

• On-line measurement of generator terminal quantities at different active and reactive power levels

The off-line measurements are used to generate/validate the saturation coefficients for the generator. The measured ac terminal conditions of the generator are used to calculate generator field current and rotor angle and this is then compared against the measured values to validate the synchronous reactances. These tests are normally only required when a generator is first registered or after significant changes have been made.

Many ISOs and Reliability Organizations also require that Utilities confirm compliance with minimum levels of lagging and leading reactive capability or identify the

physical operating limits of the unit. These tests are required regularly since the reactive capability of each generator is critical to power system reliability and can be affected by many different factors within the plant.

A. Voltage and Reactive Capability

There is often confusion about the actual reactive capability of generators and the relationship between voltage and reactive power limits during staged testing and during actual power system disturbances.

Utility-scale generators supplied in North America are typically designed to operate continuously without de-rating for operating voltages between 95% and 105% of the nameplate voltage. Sustained operation outside of these limits is not recommended and staged tests should not be conducted beyond these limits. Station and unit service loads fed from the generator should be capable of continuous operation at voltage levels corresponding to this range of generator terminal voltage. If a unit is not capable of operating within the 95% to 105% range because a station service voltage level restriction is reached first, this needs to be corrected in order to be compliant.

Excluding voltage restrictions, the reactive output capability of a generator will be affected by some or all of the following factors:

• field winding thermal limits • stator winding thermal limits • stator core-end overheating limits • excitation output capability • excitation limiter settings • relay operating characteristics (e.g. loss-of-

excitation relay settings)

Each generator’s reactive capability is depicted in a capability curve, such as the one shown in Figure 3. This curve plots physical limitations, such as stator and rotor heating limits, in the (MW, MVAr) plane. By maintaining operation within the limits depicted on the capability curve, the unit’s continuous ratings will be respected.

Prior to performing reactive capability tests it is important to understand the difference between the conditions that exist during staged tests and those that exist during actual system events when the generator may be called upon to operate at the extremes of its reactive capability. During normal operation under Automatic Voltage Regulator (AVR) control, the Operator sets the AVR reference to achieve a specific reactive power or terminal voltage level. If the power system voltage drops, the AVR control will respond by boosting excitation resulting in reactive power output from the unit to the system (i.e. lagging power factor operation). The converse is true if the system voltage rises. If the generator voltage is maintained near rated by AVR action, then the limits to be respected will be those shown on the capability curve. Under test conditions, the situation is different. Unless the Utility can

6

intentionally lower nearby system voltage levels, over-excited operation is normally achieved by raising the unit’s voltage set point. Reactive power will then flow from the unit to the system. The limits shown on the capability curve must still be respected, however, it is most possible that the generator or auxiliary bus voltages will reach their maximum allowed values before the stator or rotor capabilities are reached for either over-excited or under-excited operation and possibly for both.

A

50

50

MV

AR

SA

BS

OR

BE

DM

VA

RS

SU

PPL

IED

RE

AC

TIV

E P

OW

ER

IN

MV

AR

10 50 85 MW 100

C

B

D

UNDER EXC ITATION R EGION

SAF E OPERATION

POWER IN MW

OVEREXCITATION R EGION

Fig. 3. Capability curve and steady-state measurement points (100 MVA, 0.85 PF Unit)

For typical transformer reactance levels and constant system voltage conditions it will normally not be possible to measure both the lagging and leading reactive power limits if these correspond to typical levels of 0.9 pf lagging and 0.95 pf leading. In some cases the transmission system Operator will be able to adjust other reactive resources sufficiently to allow the full range of testing but this is not typical. It may also be possible to dispatch other units within the same Plant to obtain the desired range. This is most effective when multiple units are connected together at their low-voltage terminals, sharing a common Generator Step-Up transformer.

Regardless of which limits are reached the Utility should be prepared to supplement these measurements with calculations based on the steady-state representation of the generator. These can be used to establish the actual reactive capability of the generator over the full operating voltage range.

B. Prerequisites and Preparation

Station personnel are responsible for the following preparation:

• Obtain a copy of the latest reactive capability curve for the unit under test and become familiar with each of the limitations.

• Ensure that all voltages and currents monitored during the test on the station Human Machine Interface (HMI) accurately represent the quantities measured from the primary PTs and CTs. Spot checks should be conducted using a calibrated meter of each station service bus voltage and the current levels on any critical load.

• Ensure that recent calibration records are available for protective relays (O/C, O/V, U/V) on any critical loads and unit service buses.

• Calibrate any relays that do not have recent calibration records.

Most of the measurements involve use of existing station transducers and therefore do not introduce any risks other than those normally associated with working in a generating station environment. During the course of these tests the unit will be run at the extremes of its normal continuous capability and all participating staff should be vigilant for any sign of potential problems such as overheating. It is recognized that Operators may not be called upon to operate units at the limits of their reactive capability on a routine basis and may require the support of their Engineering Staff to properly prepare for these tests.

C. Measurements and Operation

The on-line measurements are performed with the unit synchronized to the electrical network and operating at a specified active power load. In some cases the number of measurements is minimal. An active power level close to rated power is specified and the generator reactive power is adjusted until both extremes of operation are reached. In other cases, the ISO or Reliability Organization may require that measurements be performed at several different active power levels.

At each load level, the generator field excitation is varied to change the reactive power output. The measurement points would be selected based on the generator capability curve and any other limits such as excitation system limits, generator ac terminal voltage limits (typically 95% to 105% of rated) and auxiliary supply voltage. Figure 3 depicts typical measurement points superimposed on the unit capability curve.

As noted earlier, each jurisdiction may have specific requirements however a typical reactive capability test would include up to five different reactive load levels to allow for extrapolation of results to limiting output conditions.

7

• unity power factor • over-excited (reactive power = 50% of maximum

over-excited requirement) • over-excited (reactive power = up to100% of

maximum over-excited requirement) • under-excited (reactive power = 50% of

maximum under-excited requirement) • under-excited (reactive power = up to 100% of

maximum under-excited requirement)

Local guidelines will dictate quantities that should be monitored. Not all measurements will be required or available on all units. Station metering is acceptable as long as it has been maintained in reasonable calibration. The source of each measurement should be noted along with the test data. The following is a typical list of points to be monitored on a conventional round rotor generator connected to a steam turbine:

• ac terminal voltage • active and reactive power (generator gross) • generator field voltage and current (excitation

system) • generator stator temperature (average and

maximum measurement) • generator core temperature (average and

maximum measurement) • station ambient and unit cooling temperatures • unit service bus voltages if these are fed from a

Unit Service Transformer and are therefore subject to change during the course of the test.

Hydroelectric units normally require fewer measurements since unit service voltage issues stator core-end heating are rarely a factor.

The unit is normally operated under AVR excitation control throughout the test. An Operator should monitor the unit to maintain relatively constant reactive output in the event that system conditions change during the course of the test. MVAr control loops can also be enabled for this period, as long as they are voltage-supervised, to prevent voltage from moving outside of the required limits.

If any tests reveal problems with over or under-excited operation, the excitation should be adjusted to restore unity power factor operation, or operation at rated terminal voltage. The choice of operating level will depend on the nature of the problem (i.e. voltage or reactive current).

The station service load fed from the unit under test should match normal operating conditions for the maximum continuous load level.

VI. EXCITATION SYSTEM TESTS

The testing of excitation systems to validate their performance specifications and to construct models can be a time consuming task. The best time to do this testing is in conjunction with the commissioning of a new unit. At this

time manufacturers' data are normally available, and a manufacturer’s representative is normally on site. The manufacturer may provide schematic or block diagrams for the equipment, which is a good starting point. Target simulation models should be chosen to match the simulation software to be used. Standard models [2] should be used where possible.

For model validation purposes, normally two distinct types of tests are required: open-circuit or on-line dynamic response tests that allow confirmation of the closed-loop response of the Automatic Voltage Regulator (AVR), Power System Stabilizer (PSS) and selected limiters, and off-line tests used to measure detailed transfer functions or measure limiter operating characteristics.

The following is a brief overview of the types of tests performed, based on the unit operating condition. Each of these tests is described in more detail in the following sub-sections, along with specific examples of equipment connections and results. Excitation system testing techniques are discussed in detail in [3].

• Off-line tests are performed on individual modules within the excitation system while it is energized from test supplies and isolated from the field winding.

• Open Circuit tests are performed with the generator on open circuit running at rated speed.

• On-Line tests are performed with the generator synchronized to the grid and operating at a variety of active and reactive power load levels.

A. Off-Line Tests

During the off-line tests, the exciter is energized using a test supply, and test signal generator outputs are substituted for the normal input signals (e.g. three-phase ac voltage input to the voltage regulator, dc millivolt source in place of field current shunt feedback). Individual modules or sub-modules are isolated and tested separately to validate the derived transfer functions. Step response and frequency response techniques are normally used to measure the small signal performance. Non-linearities in the electronic controls, such as limits, can be measured at this stage by inputting large signal changes at selected inputs.

Testing of limiter and protective relay operating settings, associated with the excitation system, can also be 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 limiter settings. Testing is complicated with excitation limiters since it involves verifying that once engaged, the limiter is capable of controlling the excitation level in a stable fashion. For summing limiters this involves the complex interaction with other control loops such as the AVR and PSS. For this reason, excitation limiter testing is

8

normally performed in two stages. First the operating limit characteristic (e.g. field current or reactive output points at which the limiters take over control) is measured during off-line tests through secondary signal injection. Once the limit characteristic is known, the limit is temporarily adjusted to lower limit settings that permit the limiter to be engaged during open-circuit operation or on-line operation at low active and reactive power load levels. Under these controlled conditions, the control-loop dynamics can be measured with minimal risk to the unit or power system.

A new class of test sets that combine static and dynamic closed-loop simulating capability may allow both types of tests to be performed during outages reducing the need for some or all of the on-line tests. These test sets simulate the dynamic response of the generator and power system. Prior measurements or a detailed representation of the generator is necessary for this approach to be used with confidence to verify dynamic regulator and limiter response.

B. Open–Circuit Tests

Open circuit tests are performed at rated speed and rated voltage with the generator remaining on open circuit. The open circuit tests normally consist of the following stages:

• Steady-state measurements of the exciter and generator quantities. This is often performed concurrent with the main exciter and/or generator saturation characteristic measurements.

• Dynamic tests (time-response or frequency-response) of the closed-loop AVR. The measured data can be compared with simulations performed using the block diagram model.

• Dynamic tests of the excitation limiters. The over-excitation limiters, which normally operate to limit terminal voltage, generator field current and/or exciter output current, can be tested on open-circuit, by lowering the limit set points to the open-circuit operating levels. The unit's operation is then forced into the limit set point, by injection of a test signal into the AVR or firing circuit or by simply increasing the AVR setpoint until the limit is reached.

One of the most common of the open circuit tests is the AVR step response test. This test is specifically identified in many North American regulatory compliance procedures for confirmation of excitation system response. As a minimum, generator terminal voltage, generator field voltage, or pilot exciter field voltage for brushless systems, should be measured during the application of a step change signal that will produce a 1% to 2% change in the terminal voltage reference. This test is performed by injecting a small, short duration change in the AVR reference level. This causes the generator terminal voltage level to change

suddenly and provides a good measure of the overall response of the excitation system.

The type of exciter will dictate the method used to inject the AVR reference step change. On analog-electronic systems, the signal is normally applied as a low-voltage (e.g. 0-12 Vdc) signal applied to the input of an operational-amplifier summing-junction. On modern digital systems, the exciter is normally equipped with a built-in facility for introducing controlled step changes to the voltage reference signal. On a system equipped with a magnetic amplifier, the signal may be applied as a current injection into a spare winding. On older systems such as those equipped with discontinuous regulators, the signal may be introduced by altering the three-phase PT feedback signals at the AVR input. This approach can also be used on more modern systems if detailed design information or interfacing software is unavailable. This technique has the added advantage of explicitly including the terminal voltage transducer in the forward path of the disturbance.

Another step response test can also be accomplished by transferring from constant field (manual) control mode to the AVR mode with a small unbalance between the set points. This method has the advantage of not requiring equipment to inject a change in AVR reference.

Figure 4 presents a response typical of the changes in terminal voltage that are obtained in the excitation system tests.

0

50

100

150

200

Fie

ld(V

dc)

500

510

520

0 2 4 6 8 10

Time (seconds)

Fie

ld(A

dc)

0.98

0.99

1.00simulatedmeasured

Ter

min

al V

(pu)

Fig. 4. Typical step response of static excitation system

An alternative to AVR step response tests are load rejections with the unit absorbing reactive power [20]. Upon opening of the generator circuit breaker, the excitation system will respond to the step change in reactive current in a manner similar to that seen for a step test. The AVR response in the opposite direction can be tested with the unit producing reactive power. The amount of reactive power produced or absorbed prior to opening

9

the circuit breaker will determine the magnitude of the AVR response. This method also has the advantage of not requiring any equipment to inject a change in AVR reference, but requires the coordination involved in a load rejection. On some digital systems care must be exercised in interpreting results since AVR or limiter dynamic settings or setpoints may be switched when the ac breaker is opened.

C. On–Line Tests

On-line tests are performed with the generator synchronized to the grid and operating at different active and reactive power levels. Among the tests performed with the unit on-line, are the following:

• Steady-state measurements of exciter and generator quantities. These measurements supplement the open-circuit measurements, and can be performed over a wide range of generator field current levels and are normally performed at the same time as reactive capability testing

• Dynamic tests (time-response and frequency-response) of the closed-loop AVR. Many of these tests would only be required during commissioning or on units equipped with power system stabilizers.

• Dynamic tests of the excitation limiters and other features. All of the limiters and other features that are not operational on open circuit (e.g. UEL, RCC) are tested at this stage.

Dynamic measurement of the field current limiter operation is typically performed with reduced limit settings as described in the earlier sections on off-line testing. Settings are reduced to avoid excessive bus voltage and stator current during the performance of the test. Limit levels are reduced, and a terminal voltage step is applied which forces the field current into the new limit.

Care must be taken when tuning the UEL dynamic performance under all circumstances, but particularly when it is used in conjunction with a power system stabilizer. When testing and calibrating UELs, several factors must be considered, all of which suggest performing dynamic tests with reduced settings (e.g. with reactive power settings closer to zero). Low bus voltage will result from operation at extreme under excited levels and should be avoided. In some cases, where there is limited margin between the UEL and loss-of-excitation relay or generator core-end overheating characteristic, there is the possibility of entering one of these regions if the limiter does not function as expected.

D. Power System Stabilizers

The techniques used to test the AVR and power system stabilizer (PSS) mirror the techniques that are used in their simulation and tuning. That is, we seek to measure the

required compensation characteristics, measure the effect of the settings on the closed-loop response and finally perform special measurements which could alert us to possible side-effects of the AVR and PSS settings during both normal and disturbance operating conditions. References [24 to 26] describe various aspects of testing of these controls.

The selection of phase compensation is critical to the proper functioning of the stabilizer. As a result it is important to have accurate measurements of the phase relationship between the generator terminal quantities and the stabilizer signals. Any transducers or amplifiers used for this process must be scrutinized to ensure that they do not introduce unnecessary filtering and associated phase lag in the frequency range of interest, 0.1 Hz to 10 Hz. Analog electronic hardware is being superseded by digital measurement systems, which incorporate the transducer functions with the data storage and analysis software.

In many cases, some or all of the required signals are directly available from the stabilizer itself, as these quantities are used in the stabilizing signal generation. Prior to investing in specialized transducers it is worthwhile to review the manufacturer’s manuals and schematics showing the available test points.

Two types of tests are normally performed: time-domain and frequency-domain. For the time-domain tests, the quantities should be recorded with a bandwidth of at least 50 to 100 Hz to ensure that higher-frequency components in some of the signals (e.g. turbine shaft torsional components in the speed signal) are captured. For the frequency-domain tests a transfer function analyzer is very useful. It should be capable of generating the required disturbance signal (e.g. swept-sine, pseudo-random binary noise sequences, etc.) and of performing input/output transfer functions of any pair of signals. At one time, these pieces of equipment were only available to small numbers of specialists; however with the proliferation of computer and DSP-based test equipment, they should be widely available to most utilities.

Stabilizer testing must validate several different model parameters: gains, time constants, limits and calibration of measurement transducers.

If detailed models exist from commissioning tests or studies, simple step response tests may suffice to confirm continued validity of an existing model. For equipment where no model exists, bench tests to confirm transducer calibrations and output limits may be required, as the required operating ranges of inputs and outputs may be impossible to test with the equipment in service on line. Transfer function tests will likely also be necessary for various stages of the stabilizer characteristic, requiring access to internal stabilizer signals and appropriate isolation transducers, with suitable bandwidth as discussed above. As with excitation systems, the techniques and details are discussed in the reference [3, 24, 25 and 26].

10

VII. GOVERNOR TESTS

A. Permanent Droop:

The extent to which each generating unit responds to system frequency variations is determined by its permanent droop setting, within its turbine limits. In most jurisdictions, the governor droop and speed sensing dead band are required to be confirmed by test.

Permanent droop is often set through a calibrated adjustment with an indication of the actual value. While this calibration is often reasonably accurate, permanent droop should be measured directly as a confirmation, especially on older mechanical units where changes may have been made to governor components.

This measurement may be performed without the benefit of any special equipment. The simplest approach is described below.

With the unit operating off-line, apply excitation such that the unit is operating close to rated terminal voltage. Adjust the speed reference to different settings above speed no-load (e.g. +/-4 % in 1% increments) and tabulate the unit’s frequency versus reference setting. In cases where the speed reference does not have marked settings this will sometimes involve adding temporary markings to dials or gear wheels.

Synchronize the unit and adjust the speed reference to the values above speed-no-load that were tabulated in the previous step. Tabulate the final gate or fuel valve position and active power versus speed reference setting.

Plot the measured off line speed and on line feedback signal used for permanent droop for corresponding speed reference settings. The slope of this curve at any point is the permanent droop. The average slope over the entire operating range is usually quoted as the droop value.

This test is equally applicable to hydraulic and thermal turbine governors.

B. Deadband:

The other quantity that is most frequently requested for all governors is deadband. Deadband is defined as the largest frequency change for which the governor does not respond. This can obviously be difficult to assess, especially if system frequency doe not change sufficiently during on-line monitoring. One method of evaluating governor deadband is to simulate the response of the governor to system frequency changes using the known governor droop and time constants. The "ideal" simulated response is then subtracted from the actual measured response and converted to equivalent frequency as shown in in the "deadband" plot of Figure 5.

92

93

94

95

Gat

e(%

)

-0.04

-0.02

0

0.02

0.04

0 500 1000 1500 2000 2500 3000 3500

Time (seconds)

Dea

dban

d(H

z)

59.96

59.98

60.00

60.02

60.04

Fre

quen

cy(H

z)

Fig. 5. Governor Deadband

Clearly in this case the unit's deadband is well within the required limits. Care must be taken in interpreting this kind of curve to ensure that there is sufficient ambient frequency variation to make a reasonable assessment.

Although in theory it is possible to introduce very small reference changes in modern electronic or digital governors this is rarely an effective method of assessing the overall deadband.

C. Hydraulic Governor Dynamic Tests

For hydraulic turbines, it is necessary to provide governor compensation for stable off-line and islanded speed governing. Testing this compensation should be performed on a routine basis as well as following any major maintenance outage.

The following is a simplified test description suitable for use with most mechanical-hydraulic governors for hydraulic turbines. The test equipment requirements may be quite modest: gate position transducer, hand tools, stop watch, multimeter capable of measuring frequency.

Prior to testing, tabulate as-found settings and adjust, if necessary, to match any recommended settings.

Connect slide wire or other transducer to produce a voltage signal proportional to servomotor position.

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.

11

Tabulate or record the initial gate position and rapidly change the speed reference to a new (calibrated) position. Allow the gates to settle to a new steady-state position and tabulate the final gate opening. On mechanical systems, adjust the needle valve opening as required to obtain the recommended reset time.

To obtain a record of the transient response of the governor, follow the steps outlined above and use a slidewire mounted on the wicket gate servomotor to obtain a voltage proportional to gate position as shown in Figure 6.

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0 5 10 15 20

simulatedmeasured

Time (seconds)

Gat

e P

osi

tion

(pu)

Fig. 6. Mechanical Governor Damped Response

New governing systems are almost exclusively of the electro-hydraulic design. The flyball head speed sensing system is replaced by speed probes or frequency input circuits and frequency-to-voltage converter circuitry. The gate position feedback signal is obtained from a rotational or linear displacement transducer. In some cases electrical power is used in place of gate position for on-line regulation purposes. The speed reference, measured speed and droop feedbacks are combined electronically, either with analog circuitry or using a digital implementation.

In this case, speed reference step response tests may be initiated and recorded in the same way as they are performed for analog electronic or digital voltage regulators, and the same precautions, recording, and analysis techniques apply.

Appropriate simulation models may be found in references [13 and 39]. Testing techniques for governor dynamic tuning and modeling may be found in references [40-47].

Once individual parameters have been identified, to the extent possible, an overall test can be performed to verify the correct performance of the model. Load rejection tests

are one method used for this purpose. It should be noted that many governors switch modes or set-points when the unit synchronizing breaker opens. As a result, on many units neither on-line dynamics nor droop can be validated from a load rejection test. In this case, some means of disturbing the speed reference or one of the feedback signals must be employed with the unit operating on-line as described above.

D. Thermal Unit Testing and Modeling

Whereas there is a long history of tuning and testing hydraulic generator governors, there has been less focus on 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.

The use of detailed thermal governor models has not produced load and frequency response simulations which match measured system events. Reference [14] presents the development of a new turbine-governor modeling approach that correctly represents thermal units that have demonstrated unresponsive characteristics such as “base loaded” units, or as units with load-controllers. This approach focuses on the use of data collected from station recorders during actual events rather than staged tests.

E. Combined-cycle power plant tests and models

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 excitation system, associated supplementary controls, limiters, etc. a combined cycle power plant is no different than any other similarly size thermal unit. The key difference is in the turbine controls since the plant consists of the combination of both gas (Brayton) and steam (Rankine) thermodynamic cycles.

Modern combined-cycle power plants use digital turbine controls. In these systems, governor droop is typically implemented as a constant relationship between turbine speed and generator electrical power (in units of megawatts per Hertz). An intentional deadband is often programmed into the controls, to prevent constant motion of the fuel valve. For multi-shaft units this droop setting is implemented in the governor controls of the individual gas turbines. The effective droop of the entire plant is a combination of the gas turbine response and the subsequent dependence of steam turbine power on gas turbine output. The steam turbine output follows the gas turbine, with a large time constant associated with the heat-recovery steam-generator (the steam turbine is typically operated with valves wide open – under sliding pressure control). The droop of the gas turbine(s) may be measured by plotting the steady-state variation in power versus the governor speed reference and thus determining the slope of this line [12]. For single-shaft units the governor droop can be, but is not necessarily, implemented as a

12

relationship between the total electrical power of the unit to shaft speed.

Off-line speed and on-line power response to step-changes in the governor reference input may be used to validate the time constants and rate limits associated with the gas turbine fuel valve and turbine [12]. In addition, particularly for multi-shaft units, the time constant associated with the heat-recovery steam-generator may be estimated by effecting a change in the gas turbine power and recording the corresponding response in the power output of the steam turbine. During this test the steam valves should be kept wide open (in sliding pressure mode) and the gas turbine kept in its new steady-state condition for many minutes while the steam turbine is allowed to fully respond. The gas turbine(s) in many modern combined-cycle power plants will have an outer-loop megawatt controller that will act to maintain a pre-selected megawatt output level on the gas turbines as specified by the operator. This control loop will effectively override the initial response of the turbine droop-governor in the event of a system disturbance bringing the unit’s output back to the pre-selected level [12, 14]. Ideally, this loop should be disabled on units participating in primary frequency regulation.

The maximum power output of a gas turbine, whether in combined-cycle power plants or operated as simple-cycle units, is dependant on both the operating speed of the unit and ambient air conditions [12, 27 and 28]. In some applications, inlet-air may be kept at near constant conditions by air-conditioning units to minimize variations in the turbine capability due to ambient conditions – this is done at the expense of overall plant efficiency. Where such inlet-air conditioning is not performed and the gas turbine capability varies significantly with ambient conditions, manufacturer data should be sought to estimate the turbine megawatt capability under various ambient and frequency conditions for system studies.

VIII. TIMING OF VALIDATION TESTING

Timely validation of computer models is necessary to ensure their continued validity. Equipment wear, upgrading and refurbishment, component drift, adjustments to settings and configuration management all contribute to possible changes in the dynamic response of the unit. Guidelines for frequency of model validation testing are summarized below.

Model validation testing should be part of equipment commissioning. If model validation was not performed during commissioning then it should be done as soon as practical.

The measured responses to actual power system disturbances should be compared with simulations performed using the models, and differences used to prioritize testing. If no modifications have been made to the equipment since the last model validation test, then

repeat model validation testing need only be performed if measured responses to system disturbances disagree with model predictions.

Validation of tunable settings (Table 1, B) should be performed periodically (e.g. following a major unit outage) as determined by experience with the particular equipment in use and every time a change or upgrade is made. The history of the equipment, such as component failures and the need for adjustments should be used to determine when re-testing of a model may be warranted.

Validation of equipment characteristics set by design (Table 1, A) are typically only performed during commissioning or at the start of a test program. Simple re-verification tests may be performed occasionally or when evidence of changes to unit behavior occurs.

Rotating Equipment –Most generator and rotating exciter model data is fixed by design. Generator inertia will not change unless generator or turbine modifications are made, and thus repeated testing is not a necessity. Generator rewinding is not expected to introduce changes to the generator impedances; however periodic reactive capability tests and confirmation of the open circuit saturation characteristic will reveal unexpected reductions in unit capability, for instance, shorted rotor turns, or control system automation enforcement of operational limits. Reactive capability verification is mandatory in many jurisdictions.

Closed-Loop Controls –Whenever the controls are modified or upgraded a new model may be needed and model validation testing will need to be performed. Equipment technology plays a large role in determining the necessity and frequency of testing.

Older mechanical or magnetic amplifier equipment is more likely to be refurbished, has fewer calibrated settings and is more susceptible to drift than more modern equipment, and hence should be tested following major outages.

Both analog electronic and digital electronic equipment rely on input and output electronics which require calibration and may exhibit drift. In these cases, equipment history may provide the best guide for frequency of testing.

Digital electronic equipment may have drift-free settings, but may be subject to change control issues (modification of settings), which should be periodically checked. This equipment typically has both data logging and testing facilities built-in, which remove two of the barriers to re-verification tests.

IX. AMBIENT MONITORING

The concept of an ambient test methodology involves passive monitoring of generator or system events rather than active testing. In this case, the recorded responses of normal or abnormal system or plant events may provide sufficient data for model validation.

13

In conventional dynamic performance testing, specialized equipment is temporarily connected to the equipment being tested and the unit is then subjected to some form of artificial step change in one of its input signals in order to observe the response.

For ambient testing, the design concept of the equipment would be more task-specific, hopefully reducing its cost to the point that it could be left connected to the unit for longer periods or even permanently. This test equipment would record unit operation under normal operating conditions and buffer the data for post-event retrieval, or possibly trigger the capture of data when system conditions are such that dynamic performance can be adequately measured and evaluated.

It is hoped that the development of an ambient test methodology would fulfill some of the objectives of equipment performance testing without incurring the costs and risks associated with scheduled testing.

Ambient testing methods are presently evolving and not yet fully established methods, although their use has already proved valuable in some cases [14, 60 and 61], and are being encouraged as acceptable means of model validation [32].

One area where ambient measurements have provided useful results is in governor droop and dead band assessment. High-resolution measurement of system frequency and measurement of unit output (gate position, 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 unit meets system requirements for frequency response.

X. PREPARATION, TEST SET-UP AND PRECAUTIONS TO BE

TAKEN DURING TESTING

Testing should be undertaken with knowledge of available industry standards where possible, such as those listed in the references [2-11, 40].

If offline exciter or PSS tests are planned, preparation is required to safely energize the various exciter test supplies. The ease with which voltage regulator functions can be enabled in this state varies from machine to machine, but the tester should be prepared to deal with numerous interlocks that will have to be bypassed to permit shutdown testing.

Some tests must be done with the generator running. In these cases, personnel who are familiar with this type of testing should be present. Dangerous voltages are present in the generator field circuit where transducers need to be connected. Care should be taken to avoid introducing noise into high gain circuits when test leads are connected or disconnected.

The dynamic performance of some equipment, such as under excitation limiters, will probably have to be tested with the generator on load. It is often advisable to reduce

the limiter setting for this test to avoid possible instability or tripping generator auxiliaries on low voltage. Differences between dynamic responses at different operating levels can be simulated.

Other functions, such as over excitation limiters, may be tested by simulating the input signal with the machine shut down.

Step responses with the generator operating at rated speed, not synchronized are a good check on the dynamic performance of the voltage regulator/exciter. A change in the generator voltage of approximately 1-2% should be sufficient. An exception is some older voltage regulators that had different gains for different-sized errors; in such cases, a larger step may be necessary to check the alternate gain. In either case, the size of the step should be increased gradually from zero to avoid unsafe conditions.

Step response tests are also required to check the dynamic performance on load. As with the open-circuit tests, the size of the step should be increased gradually from zero.

If frequency response tests are done with the generator synchronized, care must be taken to avoid shaft torsional frequencies. The generator manufacturer can usually provide the necessary information.

Care must be taken when performing load rejection tests to ensure that an over speed condition is not reached. This can be done by simple calculations prior to the test, based on the expected unit inertia and governor response.

XI. RECOMMENDED BUILT-IN TEST FACILITIES IN NEW

EQUIPMENT

With most generator excitation control systems now being implemented using digital technology, opportunities exist to design equipment with useful features to aid the validation testing process.

When testing such systems the commissioning engineer is normally able to see the intended settings displayed on either a built-in display or on a laptop computer. However, simply examining settings in this way can only be considered as a quick and easy check to ensure that the system has correctly retained programmed settings.

Built in test features should normally include the ability to assign analog input and output signals to the inputs and outputs of the various internal stages of the control system. This allows separate verification of the operation of each stage and facilitates confirmation of the time constants and gains.

It can be very useful to have an internal signal generator incorporated into the equipment. The internally generated signal may then be used as an alternative to an analog test signal generated using separate equipment. It should be possible to easily adjust the frequency, amplitude and wave shape of the internally generated signal.

14

Another ideal use of digital technology to aid validation testing is the addition of data recording facilities that enable sampled waveforms to be viewed on laptop computers. When this is provided it should be possible to save the sampled data for later reference. This facility should be provided with flexibility to allow signals to be recorded from various stages in the control system, with various sampling periods and with various archiving and triggering options.

While it is recommended that the internal recording and signal generation features should be incorporated into equipment, on occasions it may be necessary to use separate signal injection and data recording equipment to provide independent verification of the operation of excitation control equipment.

XII. DEVELOPMENT OF CRITERIA FOR DETERMINING A

MINIMUM THRESHOLD UNIT SIZE FOP PERFORMING TESTING

Typically, units connected to a wholesale market grid tend to be large. Smaller units are often connected into local distribution systems, which typically do not have the same requirements for equipment performance testing.

However, for every rule, there is an exception and some small units or stations can be found connected directly to wholesale market grids. In some cases, the cost of periodic performance testing of these smaller stations could adversely affect the economic viability of continued operation within the market. At the same time, the smaller the units, the less the impact they have on overall system performance.

Therefore as unit size decreases a) the benefits to the market from performance testing decreases while b) the cost of these tests per MWhr delivered increases. Clearly, at some threshold of unit size, performance testing is no longer viable.

The cost of conducting testing can be quantified relatively easily. This includes the actual cost of staff engaged to conduct the testing plus the lost opportunity cost of not having bid the unit optimally into the market. It should be kept in mind, however, that these costs might be reduced in the re-verification phase of testing. For instance, suitably trained/knowledgeable local site staff might be substituted for testing specialists. Also, lost opportunity costs might be reduced as testing schedules are shortened.

More difficult is the assessment of the impact on overall system reliability, especially in trying to express this in terms of a financial benefit. Generally, system impact can be taken as meaning the impact on the complete system within the market operator’s jurisdiction and beyond. However, in some cases, what might be considered a non-critical unit on such a system wide basis might be critical on a local basis.

More work is required by both generators and market operators to determine a suitable threshold for unit size, below which performance testing is not required.

XIII. CONCLUSIONS

This paper outlines the general approach and guidelines for field testing of generating equipment for the purpose of deriving and verifying parameters for computer simulation models of the power plant equipment. This document is a brief outline of the methodologies used. The reader should refer to the many references outline through the discussion for further detail on the testing procedures and techniques.

XIV. ACKNOWLEDGEMENTS

The working group would like to thank the many participants who contributed to this effort both at committee meetings and via email correspondence.

XV. REFERENCES: [1] Byerly R.T. and Kimbark, E.W. Ed., Stability of Large Electric Power

Systems. New York: IEEE Press, 1974.

[2] IEEE 421.5-2005 Recommended Practice for Excitation System Models for Power System Stability Studies - models

[3] IEEE 421.2-1990 Guide for Identification, Testing and Evaluation of the Dynamic Performance of Excitation Control Systems - testing

[4] IEEE Std 115-1995 IEEE Guide: Test Procedures for Synchronous Machines

[5] IEEE Std 67-1990 IEEE Guide for Operation and Maintenance of Turbine Generators

[6] IEEE Std 492-1999 IEEE Guide for Operation and Maintenance of Hydro Generators.

[7] IEEE Std 1110-1991 IEEE Guide for Synchronous Generator Modeling Practices in Stability Analyses - models; testing

[8] ANSI Std. C50.10-1977 General Requirements for Synchronous Machines

[9] ANSI Std. C50.12-1982 Requirements for Salient-Pole Synchronous Generators for Hydraulic Turbine Operations

[10] ANSI Std. C50.13-1989 Cylindrical-Rotor Synchronous Generators

[11] ANSI Std. C50.14-1977 Requirements for Combustion Gas Turbine Driven Cylindrical Rotor Synchronous Generator.

[12] CIGRE Technical Brochure 238, Modeling of Gas Turbines and Steam Turbines in Combined-Cycle Power Plants, December 2003.

[13] "Dynamic Models for Steam and Hydro Turbines in Power System Studies", IEEE Committee Report, IEEE Trans, Vol PAS-92, Nov-Dec 1973, pp. 1904-1915.

[14] L. Pereira, J. Undrill, D. Kosterev, D. Davies, and S. Patterson, “A New Thermal Governor Modeling Approach in the WECC”, IEEE Trans. PWRS, May 2003, pp 819-829.

[15] Generator Test Guidelines, WSCC Control Work Group and Modeling & Validation Work Group, March 1997.

[16] 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.

[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.

15

[18] Determination of Synchronous Machine Stability Study Constants Volume 3. EPRI EL-1424, Palo Alto, CA, June 1980.

[19] F. P. de Mello and L. N. Hannett, “Determination of Synchronous Machine Electrical Characteristics by Tests,” IEEE Trans. on Power Apparatus and Systems, vol. 102, no. 12, pp. 3810-3815, December 1983.

[20] L. N. Hannett and J. W. Feltes, “Derivation of Generator, Excitation System and Turbine Governor Parameters from Tests,” presented at the CIGRÉ Colloquium on Power System Dynamic Performance, Florianópolis, Brazil, 1993.

[21] P. L. Dandeno et all., “Experience with Standstill Frequency Response (SSFR) Testing and Analysis of Salient Pole Synchronous Machines”, IEEE Trans. on Energy Conversion, vol. 14, no. 4, pp. 1209-1217, December 1999.

[22] P. A. E. Rusche, G. J. Brock, L. N. Hannett and J. R. Willis, “Test and Simulation of Network Dynamic Response using SSFR and RTDR Derived Synchronous Machine Models,” IEEE Trans. on Energy Conversion, vol. 5, no. 1, pp. 145-155, March 1990.

[23] G. R. Bérubé and L. M. Hajagos, “Testing and Modeling of Generator Controls,” presented at SERC Generator Testing Workshop, Atlanta, GA, December 2000. (http://www.kestrelpower.com/Docs)

[24] Application of Power System Stabilizers for Enhancement of Overall System Stability, P. Kundur, M. Klein, et al, IEEE Trans. on Power Systems, Vol. 4, May 1989, pp 614-626.

[25] Applying Power System Stabilizers, Parts I, II and III, E.V. Larsen, D.A. Swan, IEEE Trans., Vol. PAS-100, June 1981, pp 3017-3046.

[26] IEEE Tutorial Course: Power System Stabilization Via Excitation Control, Sponsored by the IEEE Power Engineering Education Committee, Center, Piscataway, NJ.

[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 Frequency Dependency of Gas Turbine Output”, Proceedings of the IEEE PES Winter Meeting, Jan 2001.

[28] P. Pourbeik, “The Dependence of Gas Turbine Power Output on System Frequency and Ambient Conditions”, paper 38-101, Proceeding of CIGRE Session 2002, August 2002, Paris, France.

[29] P. Pourbeik, C. E. J. Bowler and V. L. Crocker, “Model Validation Testing for the Purpose of Determining Generation Equipment Dynamic Performance and Torsional Mechanical Response”, Proceedings of IEEE PES General Meeting, June 2004, Denver, Colorado.

[30] L. M. Hajagos and G. R. Berube, “Utility Experience with Gas Turbine Testing and Modeling”, Proceedings of the IEEE PES Winter Power Meeting, January 2001.

[31] "Testing Methods - An Overview," presented at SERC Generator Testing Workshop, Atlanta, GA, December 2000 (http://www.pti-us.com/pti/company/technical_papers.cfm).

[32] Generator Test Policy documents at Western Electricity Coordinating Council web site, www.wecc.biz.

[33] G. Shackshaft, "New Approach to the Determination of Synchronous Machine Parameters from Tests". Proc. IEE (London), 1974, 121(11), pp. 1385-1392.

[34] Y. Takeda and B. Adkins, "Determination of Synchronous Machine Parameters Allowing for Unequal Mutual Inductances", Proc. IEE(London), Vol. 121. No. 12, Dec 1974, pp. 1501-1504.

[35] G. Shackshaft and A.T. Poray, "Implementation of New Approach to Determination of Synchronous Machine Parameters from Tests", Proc. IEE (London), 1977, 124(12) pp. 1170-1178.

[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.

[37] P. Kundur and P.L. Dandeno, "Implementation of Advanced Generator Models into Power System Stability Programs", IEEE Trans. on Power Apparatus and Systems, Vol PAS-102, No 7, July 1983, pp 2047-54.

[38] R.H. Park, “Two-Reaction Theory of Synchronous Machine - Generalized Method of Analysis - Part I”, AIEE Trans. Vol. 48, 1929, pp. 716-727.

[39] IEEE PES Working Group on Prime Mover & Energy Supply Models for System Dynamic Performance Studies, "Hydraulic Turbine and Turbine Control Models for System Dynamic Studies”, IEEE Transactions on Power Systems, Vol 7, No 1, Feb. 1992, pp.167-79.

[40] "International Code for Testing of Speed Governing Systems for Hydraulic Turbines", IEC Publication 308, 1970.

[41] J.M. Undrill and J.L. Woodward, “Nonlinear Hydro Governing Model and Improved Calculation for Determining Temporary Droop", IEEE Trans., Vol. PAS-85, pp. 750-756, July 1966.

[42] J.L. Woodward, 'Hydraulic-Turbine Transfer Function for Use in Governing Studies", Proc IEE, Vol 115, No. 3, March 1968.

[43] 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, pp 2134-2143

[44] L.N. Hannett, B. Fardanesh, “Field Tests to Validate Hydro Turbine-Governor Model Structure and Parameters”, IEEE Trans. on Power Systems, Vol. 9, No. 4, November 1994.

[45] S. Hagihara et al, “Stability of a Hydraulic Turbine Generating Unit Controlled by PID Governor”, IEEE Trans., Vol. PAS 98, No. 6, pp 2294-2298, Nov/Dec 1979.

[46] F.R. Sckleif, G.E. Martin, R.R. Angell, “Damping of System Oscillations with a Hydro Generating Unit”, IEE Trans., PAS-86, 1967, pp 438-442.

[47] J.C. Agee, G.K. Girgis, “Validation of Mechanical Governor Performance and Models Using an Improved System for Driving Ballhead Motors”, IEEE Trans. on Energy Conv., Vol. 10, No. 1, March 1995.

[48] EPRI Report EL-1424, "Determination of Synchronous Machine Stability Constants", Vol. 2, prepared by Ontario Hydro, Dec. 1980.

[49] M.E. Coultes and W. Watson "Synchronous Machine Models by Standstill Frequency Response Tests", IEEE PAS-100, April 1981, pp. 1480-1489.

[50] P. L. Dandeno and A. T. Poray, "Development of Detailed Turbogenerator Equivalent circuits from Standstill Frequency Response Measurements", IEEE Trans. Vol. PAS-100, April 1981, pp. 1646-1653.

[51] P. A. E. Rusche, G. J. Brock, L. N. Hannett and J. R. Willis, “Test and Simulation of Network Dynamic Response using SSFR and RTDR Derived Synchronous Machine Models,” IEEE Trans. on Energy Conversion, vol. 5, no. 1, pp. 145-155, March 1990.

[52] M.E. Coultes, "Standstill Frequency Response Tests", presented at the IEEE Symposium on Synchronous Machine Modeling for Power System Studies (IEEE Power Engineering Society 1983 Winter Meeting), IEEE Publication 83TH0101-6-PWR, pp. 26-30.

[53] M.E. Coultes, P. Kundur and G.J. Rogers. "On-Line Frequency Response Tests and Identification of Generator Models", presented at the IEEE Symposium on Synchronous Machine Modeling for Power System Studies (IEEE Power Engineering Society 1983 Winter Meeting), IEEE Publication 83TH0101-6-PWR, pp. 38-42.

[54] P.L. Dandeno, P. Kundur, A.T. Poray and H.M. Zein El-Din, "Adaptation and Validation of Turbogenerator Model Parameters Through On-Line Frequency Response Measurements". IEEE Trans. Vol. PAS-100, April 1981, pp. 1656-1661.

16

[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.