Generator Testing and

Model Validation

Lei Wang

(lei.wang@powertechlabs.com)

Powertech Labs Inc.

November 2012

� Introduction

� WECC generator testing program

� Generator testing and model validation approaches

� Examples

� Current trend

2

Topics

� Generator is the single most important equipment in

a power system

� The adequacy and accuracy of its model have

significant impact on the results of the system

analysis performed using computer simulations

� Three approaches to obtain generator models:

� From “default” or “typical” model or data – usually

erroneous

� From design – may not be fully consistent with the actual

equipment in use

� From field testing – able to provide good models suitable

for system studies3

Introduction

� Unfortunately, not impressive in many cases

� Typical models are almost certainly inaccurate; design models

are often questionable

� There may even be serious deficiencies for models obtained

from testing by manufacturer

� Tests done may not stick to common standards so results may be hard

to apply

� Models provided may be incomplete

� Models provided may be incompatible with the designated simulation

tools so modifications or simplifications may be necessary

� Tests may be done long time ago so recent equipment retrofitting or

replacement may not be reflected

� Results: simulated system behaviors may be far from

acceptable!4

How good are our models?

� Following the August 10, 1996 WSCC (WECC) blackout, a

simulation was performed using the WSCC standard planning

models

5

A classical case to show importance of models

Measured

response

Simulated

response

� Rush Island Incident in Mid-US (June 12, 1992)

� Fault at Tyson Bus 3 tripped one circuit from

Tyson to Rush Island

� Protection malfunction tripped another

circuit from Tyson to Rush Island

� As a result, two units at Rush Island were

connected to the main grid through only

one 345 kV circuit

� A 1.0 Hz sustained oscillation occurred

6

Another case

RushIsland

#1 #2

Lutesville

Tyson

Labadie

Bus 3

Generator speed at Rush Island

� Initial analysis for this incident using the standard planning

model could not replicate the oscillations

� Therefore, operational studies done prior to the incident could not

identify this problem

� It was later found that the excitation system at the Rush Island

plant had been replaced, but the planning model was not

updated

� Therefore the model in the planning case was totally incorrect

� A field test was done and a new exciter model was developed

� The oscillations could be observed in simulations with the new exciter

model

7

Analysis of this incident

� WECC mandated generator testing policy applies to all

generators

� Single generator > 10 MVA, or

� Generating facilities > 20 MVA, or

� Connected to transmission system at 60 kV or higher

� About 95% of generators (2000 units in total approximately)

have complied with the full

baseline testing requirements

� Major utilities have already

started model re-validation

testing

� The results submitted must be

compatible to WECC approved

models8

WECC generator testing program

� Full baseline testing (full testing and once in life time)

� Existing generators that have never been tested

� New generators (within 180 days)

� Generators with major equipment modifications

� Generators identified having different responses than model

� Model performance re-validation (every 5 years)

� Validate excitation system response

� Disturbance monitoring or

� Voltage step tests or

� Frequency scan

� Validate governor response

� From observed unit response to grid frequency change

� Verify generator reactive capabilities

9

WECC generator testing program (cont’d)

� Mod-025-1 Verification of Generator Gross and Net Real

Power Capability

� Mod-026-1 Verification of Models and Data for Generator

Excitation System Functions

� Mod-027-1 Verification of Generator Unit Frequency

Response

� Field test, among other methods, is acceptable to verify the

above models and data

10

Related NERC standards

� This is a common method to obtain and to validate generator

models and parameters

� A number of standard and customized field tests can be

performed

� Models and parameters are then derived from the testing

results

� This often requires optimization of model parameters through

simulations to fit testing results

� The resulting models can be made compatible with the required

simulation software

11

Overview of generator model testing and validation

� Problems with plant equipment are identified

� Examples: 1) Reactive protections/controls improperly set or not

working; 2) PSS out-of-service

� Measurements can be used to derive good models

� Predicting unit/system performance and identifying operational

problems

� Is there oscillations between units or between areas under certain

contingencies?

� Is there a risk of instability and/or unit trip?

� System Design

� Design of load-shedding scheme – inertias/governors are critical

� Assessment of stability limits & identification of need for new

facilities

� Post-mortem studies

12

Benefits from generator model testing

� Controls can be tuned for optimal plant and system

performance

� To get maximum output from plants, controls and protection must be

working and tuned properly (local tuning)

� To allow maximum transfers between regions requires special

emphasis on excitation and PSS settings (global tuning)

� To provide maximum robustness to disturbances requires all controls

to be properly set

13

Benefits from generator model testing (cont’d)

� Model data quality for generators, excitation systems, turbine

governors, stabilizers, and even excitation limiters has been

improved significantly

� Better tuned excitation and speed control systems

� Plant operators know more about their machine capabilities

� System study engineers have more confidence in the model

data they use for various studies.

� Better correspondence between model simulations and

actual events

14

WECC generator testing achievement

15

The proof

Measurement vs simulation comparison

Frequency at Malin 500 kV California-Oregon Interface (COI) Power

Double Palo Verde trippingMeasurementSimulation

� Open-circuit saturation test

� D-axis test

� Exciter step response test

� Partial load rejection test

� Reactive capability test

16

Generator tests – main

� Generator is running at rated speed and

disconnected from the grid

� Generator excitation is increased to raise the

terminal voltage gradually from the lowest possible

value to the highest possible value

� Saturation factors S(1.0) and S(1.2) are obtained

from the results

17

Open-circuit saturation test

18

Open-circuit saturation test – example

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850

Generator Field Current in Amperes

Gen

erat

or t

erm

inal

vol

tage

in p

er u

nit

Measured

Simulated

Air-gap line

� Generator is running at rated speed without MW

load but with a small negative MVAR load (under

excited)

� Generator excitation is on manual

� Generator is tripped from the system to reject the

MVAR load, and its terminal voltage is recorded

� D-axis parameters (xd, x’d, T’d0, etc.) can be obtained

from this test

19

D-axis test

20

D-axis test – example

0.70

0.75

0.80

0.85

0.90

0.95

1.00

0 5 10 15 20 25 30 35 40 45 50 55 60

Time in seconds

Gen

erat

or t

erm

inal

vol

tage

in p

er u

nit

Measured Simulated

Generator is tripped

� Generator is operated at full speed and disconnected

from the grid

� A step change (usually 5% to 10%) is applied to the

AVR reference set point

� Generator field and terminal voltages are recorded

� Exciter step response test is used to establish

excitation system model parameters

21

Exciter step response test

Where AVR step applied

22

Exciter step response test – example

0.99

1.00

1.01

1.02

1.03

1.04

1.05

1.06

1.07

0 2 4 6 8 10 12 14 16 18 20

Time in seconds

Gen

erat

or t

erm

inal

vol

tage

in p

er u

nit

-0.2

0.1

0.4

0.7

1

1.3

1.6

1.9

2.2

Gen

erat

or f

ield

vol

tage

in p

er u

nit

Measured Vt Simulated Vt

Measured Vf Simulated Vf

Negative step is added

Positive step is added

� Generator is running at rated speed with a small MW

load and a small negative MVAR load (under excited)

� Generator is tripped from the system to reject the

load, and its speed is recorded

� The generator inertia constant (H) and some of the

governor parameters can be obtained from this test

23

Partial load rejection test

24

Partial load rejection test – example

59.5

60.0

60.5

61.0

61.5

62.0

62.5

63.0

63.5

0 10 20 30 40 50 60 70 80Time in seconds

Gen

erat

or s

peed

in H

z

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

Wic

ket g

ate

posi

tion

in p

er u

nit

Measured speed Simulated speedMeasured gate Simulated gate

Generator is tripped

� To determine unit reactive power capabilities in steady-state

operating conditions

� Test Procedure

� The excitation is varied from minimum MVAR (absorbing) to maximum

MVAR (generating) at rated MW

� Hold for 15 minutes to verify that the minimum and maximum MVAR

can be maintained without equipment overheating, alarming, and

other prohibited impact

25

Reactive capability test

26

Reactive capability test – example

MVAR MW Armature

Volts Exciter

Field Volts Exciter

Field Amps Limiting Factor

Max 36.4 52.0 14,705 50.8 11.4 Generator voltage

Min -30.6 51.9 12,937 29.4 6.2 Load angle limiter

-50

-40

-30

-20

-10

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70 80 90

Generator active power output (MW)

Gen

erat

or r

eact

ive

pow

er o

utpu

t (M

VA

R)

Reactive Power Capability Test Points

Load Angle Limiter Setting

Generator Capability Curve (0C)

Generator Capability Curve (50C)

� Field short-circuit test

� Q-axis test

� V-curve measurement

� Reactive load rejection test

� Under-excitation limiter (UEL) test

� Over-excitation limiter (OEL) test

� V/Hz limiter test

� Governor droop test

� PSS frequency response and step response test

� PSS performance test

� Governor step response test

� Water starting time constant test

� Permanent and temporary droop test

� Depending on the situation, only a selected set of these may be performed for a

unit

27

Generator tests – other

� Gather and review plant/unit documents

� Develop a test plan

� Conduct pre-test simulations if necessary

� Perform on-site testing, i.e., execute the test plan

� Develop models based on field test results

� Prepare field test and model validation study report

28

WECC generator testing and model validation procedure

� This is from Powertech’s experience

� Testing team consists of

� 1 testing engineer/technician (Powertech)

� 1 or more plant P&C technicians (Client)

� Other plant personnel (operation, safety, etc.) (Client)

� All WECC compliancy testing can be done in 2 days

� The unit to be tested must be fully available

� The results delivered include

� An engineering report

� A set of models in specified standard format (DSATools,

PSS/E, etc.) ready for use in stability analysis

29

WECC generator testing and model validation practice

� Testing Equipment. Example:

� Tabula data acquisition system (hardware & software)

developed by Powertech

� Simulation Tools. Example:

� DSATools software package developed by Powertech

� Powerflow and Short-circuit Analysis Tool (PSAT)

� Transient Security Assessment Tool (TSAT)

� Small Signal Analysis Tool (SSAT)

� Control Design Toolbox (CDT)

� MATLAB/Simulink

30

Tools required

� There are two types of parameters in generator models that

can be adjusted

� Correctable

� These are the parameters that are entered inconsistently or

incorrectly according to the physical laws

� Such parameters may cause unreasonable or wrong simulation results

� Tunable

� These are parameters that can be set within a range to optimize the

equipment or system performance

� Improperly tuned parameters result in under-performing equipment

or system

31

Generator model validation

� Synchronous machine

� Over-netting

� Reactive capability

� Source impedance

� Inconsistent synchronous, transient, subtransient, and leakage

reactances/time constants

� Saturation factors

� Exciter/AVR

� Inconsistent limits with initial condition

� Inconsistent line-drop compensation

� PSS

� Wrong PSS gain sign for electric power input PSS

� Governor

� Inconsistent turbine rating with generator rating32

Correctable parameters – common problems

� Synchronous machine

� Nothing is tunable

� Exciter/AVR

� Time constants in the transient gain reduction

� AVR gain

� Parameters in stabilizing feedback

� PSS

� Time constants in phase compensator

� PSS gain

� Most of the other parameters (reasonably typical values can be used)

� Governor

� Droop

� Governor controls

33

Tunable parameters

� Messages produced from simulation software

� TSAT limit violation check

� TSAT Dstate check

� PSS/E DOCU check

� Data checking tools built in simulation software

� No-fault simulation check

� Exciter/governor step response check

� Special tools

� NERC/ERAG dynamic database

� Control Design Toolbox (CDT) for PSS tuning

� Other

� Field testing

34

Tools for generator model validation

Generator model/data problem

� Incorrect transient reactance

� Occasionally bad stator resistance

� See the examples from a NERC case

� Inconsistent MW/MVA output with

respect to MVA rating

� See the examples from a NERC case

� Questionable damping coefficient D

� A D of 10 was identified from a recent

production case

35

Model validation example – 1

Incorrect transient reactance

Inconsistent generator outputs

Exciter model/data problem

� Incorrect gain and/or transient gain reduction

� For the exciter model shown

� KA = 1.509; TC = 0.45; TB = 0.1 (transient gain > 1 → normally less than 1)

� This leads to instability of the generator for a fault far from it

� The unit is stable with a set of more reasonable parameters:

KA = 50.0; TC = 1.0; TB = 1.0

36

Model validation example – 2

Generator relative angle (deg) : Reference Generator = 338 [SUND#2GN18.0] ' 2'

Time (sec)

0.000 6.000 12.000 18.000 24.000 30.000-50

0

50

100

150

Generator relative angle (deg) : Reference Generator = 338 [SUND#2GN18.0] ' 2'

Time (sec)

0.000 6.000 12.000 18.000 24.000 30.000-50

0

50

100

150

Generator relative angle (deg) : Reference Generator = 338 [SUND#2GN18.0] ' 2'

Time (sec)

0.000 6.000 12.000 18.000 24.000 30.000-50

0

50

100

150

Generator relative angle (deg) : Reference Generator = 338 [SUND#2GN18.0] ' 2'

Time (sec)

0.000 6.000 12.000 18.000 24.000 30.000-50

0

50

100

150

-50

150

0

50

100

Generator relative rotor angle (deg)

0 306 12 18 24

Time (seconds)

Original data

Modified data

PSS model/data problem

� A growing local oscillation shows from a

nofault simulation

� It turns out that the unit has a PSS

� Electric power is used as input to the PSS

� However, the gain KS is set at a positive value

� This reduces the damping torque of the local

mode

� Setting KS to a negative value resolves the

problem

37

Model validation example – 3

Generator relative angle (deg) : Reference Generator = 490 [GENES 3920.5] ' 3'

Time (sec)

0.000 4.000 8.000 12.000 16.000 20.000-40.000

-39.000

-38.000

-37.000

-36.000

-35.000

2

11

1

sT

sT

++

4

3

1

1

sT

sT

++

W

W

sT

sT

+1 SK

maxSV

minSV

VsPe

Governor model/data problem

� This is a digital governor

� A set of typical governor parameters results

in poor performance for the local mode

under a critical contingency

� Tuning of the governor control (mainly PI

controller) significantly improves the

damping

38

Model validation example – 4

Generator speed (Hz)

Time (sec)

0.000 2.000 4.000 6.000 8.000 10.00059.300

59.540

59.780

60.020

60.260

60.500Generator speed (Hz)

Time (sec)

0.000 2.000 4.000 6.000 8.000 10.00059.300

59.540

59.780

60.020

60.260

60.500Before tuning

After tuning

39

Model validation example – 5

� PSS validation by simulation and field testing

(part of the WECC generator testing and

model validation requirements)

� The unit is a 444 MVA steam unit with PSS

� The existing PSS parameters are implemented

by a third company (method unknown) and

approved by the unit manufacturer

� During the model validation, Powertech found

that the PSS parameters are not the optimal

� A tuning using CDT gave better parameters

� The new parameters were approved and implemented by the

unit manufacturer after the simulation and testing verification

40

Example – 5 (cont’d)

280

285

290

295

300

305

310

315

320

325

0 1 2 3 4 5

Time (seconds)

Act

ive

po

wer

ou

tpu

t (M

W)

Without PSSWith existng PSSWith tuned PSS

280

285

290

295

300

305

310

315

320

325

0 1 2 3 4 5

Time (seconds)

Act

ive

po

wer

ou

tpu

t (M

W)

Without PSSWith exisiting PSSWith tuned PSS

Field testing results

Simulation results

� Use of PMU data in model verification

� PMU is widely installed in US thanks to the Obama stimulus funding

� Generator model verification is one of the promising applications that

are offered by the PMU technology

� WECC/BPA approach is described here

� This requires special “playback” feature in simulation package

� TSAT supports this feature

41

Technical trend in model validation/verification

� The verification is done after a major event (disturbance)

� So there are significant variations in system responses

� A base case is built to represent the operation condition prior

to the disturbance

� Powerflow and the matching dynamics

� An on-line TSA system is ideal in providing these

� It may be beneficial to reduce the full case to include only the

plant to be verified, to speed up the analysis efficiency

� Special simulations are performed

42

Procedure

~Recorded voltage and frequency are injected in the simulations

Generator power (MW and MVAR) is simulated and compared with measurements

� Chief Jo braking resister insertion

� Used to help maintain stability

� Inserted into the system for about

one-half second when certain

abnormal system conditions are

detected

� Consists of five kilometers of

one-half inch stainless steel wire,

each wire strung in a vertical

configuration on a modified

transmission tower.

43

Example – event

44

Example – good model

480.00

605.00

480.0000 a2 1 WECC 500.0 0 0.0 1 1 605.0000 480.0000 pbr 40287 COULEE 500.0 0 0.0 1 1 605.0000

Time( sec )10.0 25.0

TASMO MODEL; OUTPUT GENERATED 2002-07-16 11:52:05SWINGBUS 1520 FOR FC-2001-1:2003-07-14:17:4F--1--1-0-0

ation\grand coulee\unit19

Tue Nov 18 09:56:11 2008

Page 1

coulee19-2008-08-19-1310.chf

Grand Coulee Unit 19 – active power

-100.00

-45.00

-100.0000 a3 1 WECC 500.0 0 0.0 1 1 -45.0000 -100.0000 qbr 40287 COULEE 500.0 0 0.0 1 1 -45.0000

Time( sec )5.0 30.0

TASMO MODEL; OUTPUT GENERATED 2002-07-16 11:52:05SWINGBUS 1520 FOR FC-2001-1:2003-07-14:17:4F--1--1-0-0

ation\grand coulee\unit19

Tue Nov 18 09:56:11 2008

Page 2

coulee19-2008-08-19-1310.chf

Grand Coulee Unit 19 – reactive power

MeasurementSimulation

45

Example – questionable model

Grand Coulee Unit 13 – active power

65.00

100.00

65.0000 a2 1 B1 13.8 0 0.0 1 1 100.0000 65.0000 pg 2 B2 13.8 0 0.0 01 1 100.0000

Time( sec )1.0 7.0

ef joseph\chief jo 230ph4

Tue May 19 00:03:25 2009

Page 1

cjj.chf

MeasurementSimulation