Proceeding of the IEEE 28th

Canadian Conference on Electrical and Computer Engineering

Halifax, Canada, May 3-6, 2015

Commissioning Generator AVR, PSS and Model Validation

Wenyan Gu, Member, IEEE
response time and better steady state error; but it may also
result in higher overshoot and longer settling time.

Abstract - This paper outlines procedures and experience the

author had on commissioning and field tuning ABB Unitrol P
AVR and PSS of a 160 MW steam turbine. Field measurements
are provided in the paper for demonstration of transient
responses. In addition, UEL verification test was conducted and
test method is described in this paper.
Transient responses from computer simulations are also
provided in this paper to demonstrate the techniques used by
todays power engineers. Comparisons between the field
measurements and simulations show good matched.
The tuned AVR and PSS parameters are provided for
reference, and PSS frequency characteristics are presented.

Figure 2. PSS2A block-diagram

PSS2A is a dual input power system stabilizer [1]. It features with
washout components in each of the forward paths of the input
signals and a ramp tracking filter, as well as two lead-lag
components. Therefore, it provides significant flexibility on phase
angle compensation in applications. Gain Ks1 plays the role of
damping effectiveness.

Index Terms - Generator, generator control, AVR, PSS,

simulation, stability, oscillation, damping, overshoot, Bode-plot.

I. INTRODUCTION
An existing 160 MW steam turbine generator required AVR
(automatic voltage regulator) and PSS (power system
stabilizer) upgrades, and ABB Unitrol P control system was
chosen. Unitrol P is a static excitation system with digital
automatic voltage regulator and IEEE PSS2A power system
stabilizer. The AVR and PSS functions are represented by
the block-diagrams below. Figure 1 shows the blockdiagram of the Unitrol P AVR, and Figure 2 shows the
Unitrol PSS2A stabilizer.

Our experience shows that PSS2A is a very reliable and robust

controller, and most importantly, it is relatively easy to tune.
This report outlines the procedures in field commissioning, reprints
measurements taken from the commissioning and compares field
measurements with correspondent PSS/E simulations.
II.

The guidelines for AVR field commissioning include (i) welldamped transient response due to a step change in voltage
reference; (ii) a moderate overshoot in transient response; (iii) a
short response time for fast regulation; and (iv) a balance between
steady state accuracy and transient. It shall avoid over-damped step
responses due to loss of regulation speed; it shall also avoid
aggressive responses because it increases overshoot and increases
settling time. Our experience is to limit the overshot to 4% of the
steady state voltage.
The AVR tuning included two fold tests: off-line (or open-loop)
step response and on-line (or closed-loop) step response. In
addition, UEL (under excitation limiter) and/or OEL (over
excitation limiter) shall be set during AVR field commissioning.

Figure 1. Unitrol P AVR block diagram

This AVR is a proportional (P) type controller, so steady
state error in regulation is expected. It features with two
lead-lag elements for flexibility in arranging zeros and poles.
Adjusting time constants Tc1/Tb1 and Tc2/Tb2 will let the
engineer to gain proper phase angles of the regulator. Kr is
the gain of the AVR. A larger gain Kr will result in fast

Off-line step response

Prior to the field commissioning, PSSE dynamic studies were
conducted for pre-settings of AVR. On site, AVR parameters such
as time constants and gain were set to the values identified from
simulation studies.
AVR tuning always starts with off-line step response test. To
conduct such test, the generator should operate at rated speed
(3,600 rpm for this case) with the unit breaker open and PSS

Manuscript received Nov 12, 2014.

Wenyan Gu is with ATCO Electric, Edmonton, Alberta, Canada.
[email protected].

978-1-4799-5829-0/15/$31.00 2015 IEEE

FIELD AVR COMMISSIOING AND MEASUREMENTS

669

field current, and field voltage are provided in Figure 4. The test
was conducted with stabilizer blocked as well.

blocked. A 3~5% step Vref signal was then applied to the AVR,
and the unit terminal voltage, field voltage, field current and other
signals were recorded. The recorded traces were to be used for
evaluation of the AVR settings on the base of its response time,
oscillation pattern, overshoot, and steady state accuracy. Final
setting of the AVR was made after all the quantities were within
the acceptable level.

A -/+5% step was applied to the AVR when the unit was operated
at 160 MW/0 MVAR (PSS off), i.e. at full load and 1.0 power
factor.
As expected, the terminal voltage step responses (in kV) showed
slower response time (about 2 seconds) comparing to that of the
off-line step response. The field current showed a damped
response, but oscillations were observed comparing to the off-line
step response. The oscillations were due to the induced active
power swings.

Adjusting Kr would result in fast response and larger overshoot or

oscillations, and it was continuously adjusted until satisfactory
response reached.
Figure 3 shows a group of traces recorded from the final AVR
settings. A -/+5% step signal was applied to the AVR. Vt is the unit
terminal phase voltage (in volt), Ifield is the field current (in
ampere), and Vfield is the field voltage (in volt).
The terminal voltage responded to the step signal in less than 300
ms with minimum overshoot, and there was no obvious oscillation.
The filed current was recorded at 488.6 A. When the terminal
voltage was stepped down by 5%, the field current reduced to 455.6
A. When +5% upward step voltage was applied, the field current
came back to 488.6 A. The field current was well damped in open
loop step test.
The field voltage was at 89.2 V before the step changes, and it
stayed at 83.2 V when the 5% step was applied during the time of
t=8~22 sec. It returned to 89.2 V after the step signal was removed
at t=22 sec.
The field voltage pulsed to 141.4 V in responding the 5% step
change, which was about 1.6 pu.
It is clear that all the responses were well damped, and therefore,
stable.

Figure 4. On-line -/+5% step response

The field voltage had a 100 V drop in responding to the 5% step,
and it reached the ceiling voltage (400 V) when +5% step was
applied.
Under Excitation Limiter (UEL) test
This application did not include OEL, so only UEL was tested and
set. It was done by continuously manually lowering the terminal
voltage until the limiter picked up. The test was conducted when
the generator was operated at 150 MW/-20 MVAR generation
level. The terminal phase voltage started at 9.03 kV. When it
reached 8.7 kV, where the var reached 47 MVAR, the limiter
brought the terminal voltage back to 8.83 kV. The UEL stabilized
the reactive power at 37 MVAR.

Figure 3. Off-line -/+5% step responses

On-line step response
The on-line step response was conducted to evaluate AVR on-line
stability. For comparison purpose, the traces of terminal voltage,

670

Most importantly, the applied disturbance level is too small to

trigger area oscillation mode (~0.4 Hz). Therefore, the PSS was
mainly tunned to damp local oscillation mode (~1.6 Hz).
As described in [4], this PSS tuning started at lower generation
level and increased to higher generation level for preliminary
settings for risk mitigation. As the last step of finalizing the PSS
parameters, the generator was applied with a -/+5% step to the
AVR when generator was operating at 160 WM/-2 MVAR. A
record of the responses was taken during the step changes.
The tuning was evaluated in the time domain. In order to evaluate
the damping effect, a similar step disturbance was applied to the
unit with PSS out of service and with PSS in service.
Figure 6 shows the generator responses with PSS out of service,
and Figure 7 shows the generator responses with PSS in service.
From Figure 6 we estimated that the local mode power oscillation
was at approximately 1.6 Hz. It was the local mode.

Figure 5. UEL response test

III.

FIELD PSS COMMISSIOING AND MEASUREMENTS

In order to tune a PSS a moderate strength disturbance should be

applied, and we use 5% step reference signal on AVR while the
generator is operating at certain power level to serve such purpose.
It is recommended to start at a low power generation level and
eventually advance it to higher level when stability is at no risk. A
high generator loading condition is necessary in order to generate
enough power swing for assisting PSS tuning.
WECC (Western Electricity Coordinating Council) has issued a
white paper in 2002 [2] which provide good guidelines for PSS
tuning. It recommends that while the PSS is on-line, slowly
advancing the gain until a small rapid oscillation is just sustained or
growing. The final PSS gain shall be set to 1/2~1/3 of that value.
We consider such method too risky in forcing the generator into
instability.

Figure 6. Step responses PSS out of service

In stead, we set our tuning criteria to (i) effective damp on secondthird oscillation; (ii) because the PSS output is fed to AVR, so the
gain of the PSS shall not saturate the field voltage for the applied
disturbance.
PSS tuning
Similar to AVR tuning, we conducted studies for pre-settings of the
PSS prior to field commissioning. The time constants and gain of
the PSS were set to the values obtained from pre-setting simulation
studies initially. For safety concern, one might initially set the gain
Ks1 to a value lower than the one from simulations, and eventually
increase it to an optimal value.

Figure 7. Step responses PSS in service

PSS is used for damping transient oscillations experienced in a

power grid. For an interconnected power system there are two
modes of oscillations normally triggered by a disturbance.
Depending on the location and severity of a disturbance, the interarea mode [3] may not be always obvious and measurable. As for
this application, the generator is moderately small comparing its
capacity to the grid capacity, and it is located far from the inter-ties.

The PSS output limiter was set at +/-10%. Higher limit will
introduce stronger PSS regulations which might negatively impact
terminal voltage performance. This value was check and verified
by PSSE simulations. WECC recommended a minimum limiter
value of +/-5% limit [5], and the final setting met WECCs
requirement.
671

IV.

SIMULATION STUDIES

Prior to the field tuning, PSSE simulations were conducted for

AVR and PSS parameter settings to assist the commissioning. After
the field commissioning, simulation studies were also conducted to
verify and to validate PSSE model. It also serves the purpose of
providing validated model to WECC and local Independent System
Operator (ISO).
A. Simulations
Simulations were conducted to verify the PSS/E dynamic models.
The correspondent traces were recorded for comparison with the
field data.
The AVR and PSS models were chosen to represent the physical
installations as closely as possible, and the parameters were set at
the values from commissioning.
Off-line step response

Figure 9. On-line step response: -/+5% step on AVR.

Traces shown in Figure 8 are the terminal voltage and field voltage
of the measured and simulated. The plots show a 5% step change
on AVR reference. Although the simulated response of the terminal
voltage was a little slower than the measured correspondence, it is
considered acceptable from an engineering point of view. The next
simulation of on-line step response will further confirm the
accuracy of the model used in the simulation.

The well matched terminal voltage and the field voltage between
the field measurement and the simulation further confirmed that the
AVR model used in the simulation is accurate.
PSS damping
Simulation for the PSS effect is presented in Figure 10, where the
field measured MW and the simulated MW responses subject to a /+5% AVR step were plotted. The unit was operated at 160 MW/-2
MVAR production. The mismatch of the MW swings between the
field measurement and the simulation is about 2 MW. But the
damping time and the oscillation pattern are very close. Therefore,
the PSS parameters used in the simulation are considered accurate.

Figure 8. Off-line step response: 5% AVR reference step applied

as in Figure 1.

Figure 10. MW -Field measurement and simulation result

comparison for -/+5% AVR step disturbance.

On-line step response

For those who are interest in frequency domain analysis, Bode

Plots of the tunned PSS are provided in Appendix for reference.

A -/+5% step on AVR was applied when unit was running at 160
MW/-2 MVAR with PSS in service. Figure 9 shows the terminal
voltage response and field voltage response for both the field
measurement and PSS/E simulation.

V.

FINAL AVR AND PSS SETTINGS

Frequency domain analysis [6] is a powerful engineering tool in

such applications. The final tunned parameters for AVR and PSS
are given below:

672

AVR parameters:
TC1= 0.01
Ts=0.003

TB1=0.007

in the range of 10~20 db, and it attenuates sharply for higher

frequencies. For a 45 degree for stability margin, the tunned PSS
has approximately 3 dB of amplitude margin.

TC2=1.6 TB2=9.85 Kr=400

Input #2 reduced the bandwidth of PSS from above 100 Hz to 30

Hz. This is likely to improve noise resistance to higher frequency
signals.

PSS parameters:
TW1=2.0 TW2=2.0 T6=0.0 TW3=2.0 TW4=2.0 T7=2.0
Ks2=0.2 KS3=1 T8=0.0
T9=0.1 KS1=5.0
T1=0.2
T2=0.04 T3=0.36 T4=0.12 USTMAX=0.1 USTMIN=-0.1

The known WECC inter area mode is approximately at 0.4 Hz.

Although the inter area model was not observed from the applied
5% step voltage reference, the Bode Plot shown in Figure 12 a 16
dB magnitude at 0.4 Hz with a 40 degree leading angle. So the
tunned PSS shall provide damping to this inter area mode.

The PSS employed two input signals: input signal #1 is the bus
frequency deviation; input signal #2 is generator electrical power.

VI. CONCLUSSIONS

ACKNOWLEDGMENT

The work reported in this paper was a result of AVR and PSS
upgrades on an existing 160 MW generator by using ABB Unitrol
P exciter and stabilizer. The procedure of commissioning was
provided in this paper and the recorded field data show that the
parameter settings of the AVR and the PSS were satisfactory in
terms of the response time, damping, and stability.

This work was conducted and financed by ATCO Electric

and ATCO Power, and their support is sincerely appreciated.

REFERENCES

Simulations on PSS/E program were also conducted, which shown

good match with the field data. The AVR model used for the
simulation was ESST1A, and the PSS model used for simulations
was PSS2A.

[1] G.R. Berube, L.M.Hajagos, Roger Beaulieu , Practical Utility

Experience with Application of Power System Stabilizers,
Engineering Society Summer Meeting, 1999, p 104-109 vol 1.
[2] WECC White Paper on Power System Stabilizer Tuning, May
2, 2002.
[3] P. Kundur, Power System Stability and Control, 1993
[4] Wenyan Gu, Paul Smulders, Ken Mushens, Power System
Stabilizer Tuning Simulations and Commissioning, IEEE
PES 2000 Summer Meeting, Seattle, WA, USA, July 15-20,
2000, Paper 2000SM-371.
[5] WECC Criteria VAR-502-WECC-CRT-0, Power system
stabilizer design and performance criteria, April 23, 2004.
[6] Benjamin C. Kuo, Automatic Control Systems, third
Edition, 1975.

APPENDIX PSS FREQUENCY CHARACTERISTICS

The final tuned stabilizer is charactered by the Bode Plots shown in
Figure 12, where Trace 1 represents input #1 path, Trace 2
represents input #2, and Trace 3 is the combined function with both
inputs.
PSS Bode Plot

40
25

Amplitude (dB)

10
5
20
35
50
65
80
0.01

0.1

1
Frequency (Hz)

10

100

10

100

trace 1
trace 2
trace 3
180
135

Phase (Degree)

90
45
0
45
90
135
180
0.01

0.1

trace 1
trace 2
trace 3

1
Frequency (Hz)

Figure 12. Tuned PSS Bode Plots

Comparing trace 1 and trace 3 we see that they are very closely
matched each other. This means that input #1 dominates the
functionality of the tunned PSS up to the frequency range of 30 Hz.
In the frequency range of 0.1~9.0 Hz, the PSS gain (amplitude) is
673