Citation
A ramp loading response governor controller

Material Information

Title:
A ramp loading response governor controller
Creator:
Vu, Hoa Dinh
Publication Date:
Language:
English
Physical Description:
vi, 47 leaves : illustrations ; 29 cm

Subjects

Subjects / Keywords:
Hydroelectric generators -- Automatic control ( lcsh )
Governors (Machinery) ( lcsh )
Governors (Machinery) ( fast )
Hydroelectric generators -- Automatic control ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 44-47).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Electrical Engineering.
General Note:
Department of Electrical Engineering
Statement of Responsibility:
by Hoa Dinh Vu.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
34593779 ( OCLC )
ocm34593779
Classification:
LD1190.E54 1995m .V8 ( lcc )

Downloads

This item has the following downloads:


Full Text
A RAMP LOADING RESPONSE GOVERNOR CONTROLLER
by
Hoa Dinh Vu
B.S., The University of Colorado at Denver, 1983
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Electrical Engineering
1995
iAL


This thesis for the Master of Science
degree by
Hoa Dinh Vu
has been approved
by
Miloje S. Radenkovic
William R. Roemish


Vu, Hoa Dinh (M.S., Electrical Engineering)
A Ramp Loading Response Governor Controller
Thesis directed by Professor Pankaj K. Sen
ABSTRACT
A new controller with a ramp loading response
feature has been developed and tested for hydrogenerator
speed governors that controls Francis-type turbines.
This controller produces stable and fast ramp loading
responses. The optimum ramp rate is determined by the
unit's penstock characteristic or the unit's loading
rate. The ramp loading characteristic is an improvement
from a time constant loading characteristic, which most
of the existing governors have.
This controller also provides zero-speed-error
control when the unit is operating disconnected from a
power system (off-line) and speed-droop-characteristic
control (low gain) when the unit is synchronized to a
power system (on-line). The controller also prevents
excessive valve, servomotor, and gate movement to
minimize wear and tear in mechanical parts.
This abstract accurately represents the content of the
candidate's thesis. JE recommend its publication.
Signed


CONTENTS
Figures ................................................ v
Acknowledgements ...................................... vi
Chapter
1. Introduction ........................................ 1
2. The New Governor Control Descriptions ............. 9
2.1 The New Controller ................................ 9
2.2 Droop Feedback Control ........................... 10
2.3 Ramp Loading Control Feature ..................... 13
2.4 Valve and Servomotor Control System .............. 17
3. Testing the New Governor Controller ............. 25
3.1 Unit Off-Line Test ............................... 26
3.2 Unit On-Line Test ............................. 26
3.3 System Disturbance Test ................... 27
4. Conclusions ....................................... 34
4.1 Discussion of Results ............................ 34
4.2 Future Work ...................................... 35
4.3 Alternative Design ............................... 35
4.4 Trend of Governor Control Systems ............... 36
Appendix .............................................. 37
Glossary .............................................. 42
Bibliography .......................................... 44


FIGURES
Figure
1-1 Speed Governor Control System ..................... 4
1-2 Analog Electronic PID Governor [6,15] ........... 5
1-3 Analog Electronic Double Derivative Governor [7] .. 6
1-4 Mechanical Rate Feedback Governor [15] ........... 7
1- 5 Comparison of Time-Constant & Ramp
Loading Responses ................................. 8
2- 1 New Proposed Controller (Governor) ........... 19
2-2 New Governor with Speed Droop Feedback ........... 20
2-3 The Reference Control .......... 21
2-4 New Governor with Ramp Loading Feature ............22
2-5 Valve and Servomotor Control System ...............23
2-6 A Typical Relationship between Gate Position
and Power Output ..................................24
3-1 Unit Off-Line Step Response with the New Governor 28
3-2 A Typical Step Response of a Mechanical Governor .. 29
3-3 A Ramp Loading Response with the New Governor ..... 30
3-4 Field Test of a Ramp Loading Response T...........31
3-5 A Typical Load Response of a Mechanical Governor .. 32
3-6 A System Disturbance Response Test .................33
A-1 The Model of a Machine Swing Equation [13] .......3 9
A-2 The Model of a Hydro Turbine [13,14,15] 40
A-3 The Model of a Valve and Servomotor System .........41
v


ACKNOWLEDGEMENTS
This thesis is submitted to the University of
Colorado at Denver as the final requirement for the
degree of Master of Science in Electrical Engineering.
This study would not have been possible without the
cooperation and contributions of many people.
I especially wish to thank Dr. Pankaj K. Sen,
Professor of Electrical Engineering at CU-Denver for
serving as my academic and thesis advisor and providing
me with his suggestions, criticism, and encouragements.
I wish also to thank the Hydroelectric Research and
Technical Services Group, Bureau of Reclamation. The
research for this thesis was done in Reclamation
facilities and with Reclamation funding. I am
particularly grateful to Mr. J. C. Agee, my team leader,
and Mr. Bert Milano, Head of the Group, for making "it
possible for me to undertake this study.
I would further like to take this opportunity to
thank the faculty of the University of Colorado at Denver
for their efforts in providing an excellent education.
vx


1. Introduction
Over the years, a great deal of effort has been
expended in speed control for hydrogenerators with
Francis-type turbines [1-12]. A complete governor
control system is shown in figure 1-1. Wicket gates
control the amount of water flow into a Francis turbine.
The wicket gates are positioned by a valve and servomotor
control system (blocks 2, 6, 97, and 98 in fig. 1-1).
The valve and servomotor control system is driven by a
speed controller (governor, super block 7 in fig. 1-1).
Most existing analog electronic governors are
Proportional-Integral-Derivative (PID) and Double
Derivative types as shown in figures 1-2 and 1-3. Also,
most existing mechanical governors are rate feedback
controller types as shown in figure 1-4. All of these
controllers have time-constant loading response
characteristic [1-10].
These types of control system can be tuned so that
the governor system has fast, and stable (well damped)
operation. However, this kind of tuning results in
excessive valve, servomotor, and wicket gate movement
(activity) because of excessive derivative control
1


action. This excessive valve, servomotor, and wicket
gate movement results in more system down time for
maintenance because of wear and tear on mechanical parts.
Two options are available to reduce this excessive
activity. The first option is to reduce the damping
factor for synchronized operation (on-line). This option
results in a less stable system. This problem will be
critical when the unit is connected to an isolated power
system. For a strong-tie power system, the instability
of a unit will be stabilized by the power system.
The second option is to reduce the control system
bandwidth. This option results in a slow response
(sluggish) system. The sluggish response will be
magnified by a ramp input instead of a step input in
reference signal and the speed reference adjustment has a
ramp characteristic.
A hydrogenerator governor controls its generator
speed during unsynchronized operation (off-line).
However, the process of controlling speed will have the
effect of controlling the unit's power output when the
unit is on-line. During on-line operation, adjusting a
unit's speed reference will not change the power system
speed because the inertia of the interconnected system is
2


much larger than the unit's inertia. However, as the
wicket gates move, the unit's load angle is changed.
Therefore, the unit's power output is changed.
During on-line operation, the power system speed
(frequency) is almost constant. Therefore, most of the
existing controllers act like a fixed gain and fixed
time-constant controller. This condition results in slow
time-constant loading performance. The proposed ramp
loading control feature provides a ramp loading
characteristic instead of a time-constant loading
characteristic as shown in figure 1-5. The ramp loading
response reaches the final value faster than the time-
constant loading response. The new controller with the
ramp loading feature improves the loading performance
without sacrificing the unit's dynamic stability and the
minimization of valve, servomotor, and gate movement.
3


Valves and'
Servomotors
Controller Gain LX-'
98
Gate limit cntl
Actuator Driver
CD*
Valves and servomotors control
Gate limit
Gate position
Hydro_Turbine
XJ
(V+1) (^+1)
(r^+1) (V+1) (3>+i)-
2~;
Gate error

G> 2*
Generator
Swing equation
cz>
Unit breaker
Torque ang
Electrical torque
Speed______
governor_control
I Controller Output
-U
< SUPER 0
BLOCK
Continuous
Reference
Unit breaker
<32
<32
Figure 1-1
Speed Governor Control System


Governor Electronics
Valve & Servomotor Controller
/ \ / \
Type Feedback
Figure 1-2
Analog Electronic PID Governor [6,13]


Governor Electronics
Valve & Servomotor Controller

Unit MW
Signal
Figure 1-3
Analog Electronic Double Derivative Governor [7]


Figure 1-4
Mechanical Rate Feedback Governor [13]


Constant Rate Vs. Time Constant
Figure 1-5
Comparison of Time-Constant & Ramp Loading Responses


2. The New Governor Control Descriptions
2.1 The New Controller
A new controller (governor) for speed control of a
Francis-type turbine hydrogenerator is proposed as shown
in figure 2-1. This controller includes one integrator,
block 97, in the forward path and two zeroes/two poles
(lead/lag), block 7, in the feedback path of the speed
signal. The input,signal to the lead/lag block is the
speed deviation signal. Therefore, the speed feedback
signal is subtracted by one per unit to compute the speed
deviation, block 9.
The integrator is necessary for the zero-speed-error
control. One-percent change in reference is one-percent
change in speed. The integrator gain, Kit is a factor of
the unit's response time (bandwidth) and dynamic
stability. A relatively low gain, K;, ensures that the
unit is dynamically stable under all loading conditions
and off-line operation. However, a low gain, K;, results
in a sluggish response control system.
The lead/lag controller is used to improve the
unit's dynamic stability by its phase-lead compensation.
Also, with the phase-lead compensation, the integrator
9


gain, Kif does not have to be too low; therefore, the
response time can be improved.
The lead/Lag controller is put in the feedback path
so that the ramp loading control feature is possible with
the integrator. The ramp loading control feature is used
to improve the unit's loading performance. For more
details of ramp loading control feature, please refer to
the ramp loading control feature section.
With the ramp loading control feature, the new
controller can be tuned for relatively slow response.
This slow response ensures the unit's dynamic stability
and minimum valve, servomotor, and gate activities.
Minimization of valve, servomotor, and gate activities
reduces wear and tear in mechanical parts. Therefore,
the unit down time for maintenance is also reduced.
2.2 Droop Feedback Control
The power system speed (frequency) is almost
constant under normal operating condition and can not be
controlled by a single unit's governor. This results in
a very little dynamic effect of the lead/lag controller
on the control system performance. Therefore, the new
controller has the same effect as a single pure
10


integrator.
With a single integrator controller, a small change
in the power system frequency causes the controller to
integrate all the way up or all the way down
(overreaction to power system conditions). This
overreaction happens because the controller tries to keep
the speed-error at zero, which it cannot. The
overreaction activity results in either a fully loaded or
a motoring unit condition. This kind of operation is not
preferred. Therefore, a droop characteristic is needed
to avoid the overreaction.
A droop characteristic is accomplished by a constant
gain (droop setting) feedback, block 6, from the
controller output to the summing junction, block 8, as
shown in figure 2-2. This feedback changes the
controller from an integrator to a fixed-gain and fixed-
time- constant controller.
This feedback goes through a switch, block 18, and
the switch is controlled by the unit breaker signal.
When the unit breaker is open (off-line), the feedback is
disconnected. The controller will have a zero-speed-
error control characteristic (integrator). When the unit
breaker is closed (on-line), the feedback is connected.
11


The controller will have a fixed-gain and fixed-time-
constant characteristic.
The controller's fixed gain is determined by the
inverse proportion of the droop setting. This setting
determines the controller's response to a power system
frequency change. For an 8-percent (0.08) droop setting,
the controller's fixed-gain is 12.5 (1/0.08). With 1
percent change in power system speed or speed reference,
the controller output changes by 12.5 percent. This will
change the wicket gates by 12.5 percent (not all the way
up or down).
Â¥
The fixed-time-constant is determined by the inverse
proportion of the product of droop setting and integrator
gain, K;, (1 / (K; droop) With a small integrator
gain, the time-constant is large. Therefore, the loading
response is slow. For an 8-percent (0.08) droop setting
and an integrator gain of 0.1, the controller's time
constant is 125 seconds. With this kind of setting, the
controller takes more than 400 seconds (3 to 4 times the
time-constant value) for a load change performance. This
setting results in a very slow loading response.
Before the droop feedback is connected, the
controller output has some value. This value multiplied
12


by the droop setting is called the speed-no-load (SNL)
offset. For minimum disturbance when connecting the
droop feedback signal (unit breaker closed), the droop
feedback signal has to be subtracted by a constant (block
99 in fig. 2-2) equal to the speed-no-load offset. The
SNL offset value shown is determined from field test
data.
2.3 Ramp Loading Control Feature
With the droop feedback, the controller becomes a
fixed-gain and fixed-time-constant controller. The
control system has a time-constant response
characteristic which is very slow,, especially with low
integrator gain.
However, the unit's loading response time can be
improved with a ramp loading control feature. This
feature only works when a reference signal is adjusted.
Under steady state conditions, this feature does not
function and the governor reverts to a fixed-gain and
fixed-time-constant controller.
The ramp loading controller has two features. The
first feature is the reference control as shown in figure
2-3. It has a reference adjustment detection and a
13


variable gain. The reference signal is a constant signal
and it either ramps up or down from an initial to a final
value. If the reference signal ramps up, then it is
multiplied by a large positive gain. If the reference
signal ramps down, then it is multiplied by a large
negative gain. Under steady state conditions, the
reference signal is multiplied by a gain of 1.
The reference-control input is the reference signal.
A reference adjustment is detected by a derivative (block
2). The output of the derivative block is connected to
an absolute value block (block 12). With the absolute
value block, a reference adjustment up or down can be
detected. The absolute value block output is connected
to a logic block (block 13). This block compares the
reference adjustment signal to an epsilon (small number)
to be sure that the reference is adjusted. This logic
signal controls a switch (block 14). This switch output
will be a gain of 1 or a large gain (+100 or -100). This
output gain will be multiplied to the reference signal
(block 34).
The derivative output is also connected to two other
logic blocks (block 32 and block 5). Block 5 determines
whether the reference is adjusted down, and its output
14


controls a switch (block 31). This switch output will be
-100 if the reference is adjusted down. Otherwise, its
output is zero. Block 32 determines whether the
reference is adjusted up, and its output controls a
switch (block 23). This switch output will be +100 if
the reference is adjusted up. Otherwise, its output will
be the output of block 31, which can be -100 or zero.
The block 23 output will be a large gain value (+100
or -100). This large gain has to be reversed in sign if
the reference signal is negative (unit motoring). The
reference signal is checked for negative condition by the
logic block 99. Block 99 controls a switch (block 25).
This switch output will be a large gain value (+100 or -
100) or the reverse of the large gain (-100 or +100) .
The second feature of the ramp loading controller
is the constant-rate limiter, block 70, as shown in
figure 2-4. The constant-rate limiter and the reference-
control features are used only when the unit is on-line
(unit breaker closed) because they are used to improve
the loading performance. They are bypassed by switches,
blocks 76 and 98, during off-line operation.
During a reference adjustment, the reference signal
is multiplied by a large gain. Therefore, the speed
15


error signal will be large and will be limited by the
constant-rate limiter. The effect of the limiting
process is that there is no droop feedback and the
controller becomes a pure integrator with a different
gain. The controller output will either ramp up or down
depending on whether the reference signal ramps up or
down.
The rate of the controller output ramping is
controlled by the integrator gain and the limiter value.
With an integrator gain of 0.1 and a limiter value of 1,
the controller output will change by 0.1 (10 percent) in
1 second. If the controller output needs to change by
62.5 percent from speed-no-load to full load and the
limiter value is 6.25, then the controller output changes
by 0.625 (62.5 percent) in 1 second (integrator gain of
0.1). If the limiter value is 0.0625, then the
controller output changes by 62.5 percent in 100 seconds.
The optimum ramp rate is limited by the gate rate of a
unit. This gate rate is set mechanically based on the
unit's penstock characteristic or the unit1s loading-
rate, whichever is lower. If the gate rate is set too
fast, a vacuum space may be created in the penstock
during a gate opening and the penstock may collapse. If
16


the gate rate exceeds the unit's loading rate, the unit
may heat up too fast. Therefore, the ramp rate should be
set lower than the gate rate.
If the ramp rate of the controller from speed-no-
load to full load is 100 seconds, then the reference
ramp rate from speed-no-load to full load (0 percent to 5
percent) should also be set at 100 seconds.
2.4 Valve and Servomotor Control System
The valve and servomotor control system is shown in
figure 2-5. The valve and servomotor system is shown as
the super block 97. The valve and servomotor system
controller is shown as blocks 6 and 98. This control
system controls the wicket gate position (its output) to
its input signal. The input signal to this control
system is the controller (governor) output. If this
control system has a large bandwidth (fast), then the
wicket gate position will respond correspondingly to the
governor output. A typical governor control system
bandwidth is about 0.1 to 0.3 Hertz. Therefore, the
valve and servomotor control system bandwidth should be
about 0.9 to 1.0 Hertz.
If the wicket gates respond at the ramp rate, then
17


the unit's power output will respond at almost the same
ramp rate. A typical relationship between a unit's power
output and its wicket gate position is shown in figure 2-
6.
The controller for this system has a gain, K, and
two sets of lead/lag. Two additional lags are included
for filtering. A typical model of a valve and servomotor
system is illustrated in the appendix.
18


H
VD
Lead Lag Controller Speed Deviation
Kj = Integrator gain
T, = First lead time constant
T2 = Second lead time constant
T3 = First lag time constant
T4 = Second lag time constant
Figure 2-1
New Proposed Controller (Governor)


to
o
Lead Lag Controller Speed Deviation
Figure 2-2
New Governor with Speed Droop Feedback


Derivative
to
H
Figure 2-3
The Reference Control


to
to
Lead Lag Controller Speed Deviation
Figure 2-4
New Governor with Ramp Loading Feature


Valves and
Servomotors
K = Actuator gain
T,j = First lead time constant
T,2 = Second lead time constant
T13 = First lag time constant
T14 = Second lag time constant
Tf = Filter time constant
Figure 2-5
Valve and Servomotor Control System


Gate
Figure 2-6
A Typical Relationship between Gate Position
and Power Output


3. Testing the New Governor Controller
The new governor with a ramp loading control feature
was tested by a real-time digital computer simulation and
field tested at Mt. Elbert powerplant. The real-time
simulation system includes a Sun Sparc 10 station, an
AC100 real-time simulator, and a Matrixx-System-build
software package. The field test was performed at Mt.
Elbert powerplant using the same hardware and software
used during simulation tests.
For simulation tests of the new controller, a model
of a generator was needed. The models of a generator
swing equation, hydro turbine, and valve and servomotor
system used for simulation tests are illustrated in the
appendix.
The real-time simulator simulated all blocks shown
in figure 1-1. The Sun station was set up as a unit
control board to control and monitor the simulated unit.
However, for the field test situation, the models of the
valve and servomotor, hydro-turbine, and generator swing
equation were eliminated. The models of the new governor
and the valve and servomotor controller were connected to
the actual unit's signals.
25


3.1 Unit Off-Line Test
A two-percent step in reference signal response is
shown in figure 3-1. The 10- to 90-percent rise time of
the speed signal is about 18 seconds. No overshoot
occurs. This response is still relatively slow. If the
controller is tuned for a faster response, the control
system will be less dynamically stable and will exhibit
higher overshoot. In addition, the valve, servomotor,
and gate system will be more active. A typical step
response of a mechanical governor is shown in figure 3-2.
The control system takes more than two minutes to settle
to the final value. Also, the response exhibits high
overshoot and some oscillations.
3.2 Unit On-Line Test
A ramp loading response of 38 percent is shown in
figure 3-3. The controller output, gate position, and
power output respond at a ramp rate. A ramp loading
response of the actual unit (field test) is shown in
figure 3-4. The controller output, gate position, and
power output (load) respond at a ramp rate. A typical
loading response of a mechanical governor is shown in
figure 3-5. The controller output, gate position, and
26


power output (load) exhibit a time-constant response
characteristic.
3.3 System Disturbance Test
A system disturbance is simulated by lowering the
power system frequency. The new governor control system
response data are shown in figure 3-6. The governor
raises the gate position to help the power system. The
response data indicate that the new governor controller
responds correctly to power system disturbances.
However, the response is a slow time-constant response
because the ramp loading control feature only works
whenever a reference adjustment exists. Without a
reference adjustment, the new governor controller reverts
to a fixed-gain and fixed time-constant controller.
27


Controller
Output
Gate
position
Speed
Figure 3-1
Unit Off-Line Step Response with the New Governor
28


Figure 3-2
A Typical Step Response of a Mechanical Governor
29


30


Figure 3-4
Field Test of a Ramp Loading Response
31


Figure 3-5
A Typical Load Response of a Mechanical Governor
32


100
Controller
Output
0
100
Gate
position
0
125
Speed
75%
100%
Load
0%
FFFFF!*.. F;:!
[L. :::: j: r TL: Sill
:::: :::: Ini
. - iiLiLij v s:i.i
Fi ! f'
*-j~r iililii ill: IF F: *rr: p ii :u£
:::: 1:::: ill: [iFiF: :::: :::i iii
iiLp;:: lin j--::
[F: : i:::: : : 1:::: FIJ: :FiFP
LLiv FFFFF


: . ' :
:f F
. :::: FiF'N:-.! i iii . F L
;;:i iiiii;: "i.. ; :p
::: :::: :::: : ! * < -: .j:1:-!-* :.* : :.:L
' 1 i> ...... :. {I i lJ_ : Hii-Hi
.. T 1!: 1 :::: :F LP
i- - ::::i:: ll ::::
| FLiFF"! L :F: : llLf
t jLi-r i rr ..... ... ,....
Cleveland. Ohio Pr ted ip U.S.A.

.::: i:.:: JIM : :::\ ::::: { ::: FijiF HL
! '
* ;F Li: i-.ii'iiH T- L
System Distur bance i i:
F: Fi :::: r:..
rF.:: .
.'L
Mi ... : L:Vi :i- . i!:F ' 1 ::
' ! _
:::: - . : ' '


.... .. : .
::: .. : ::
: . :1 rl : - - - " f;
.... -- L-: r 'vr - r-77T
:::
:':j F i-. : F::;i 1
F: ::: . '
Li..,::: i n :>nds
;;;; nL
iLbi ..

Figure 3-6
A System Disturbance Response Test
33


4. Conclusions
4.1 Discussion of Results
The new governor control system has the following
performance:
a) Well damped, no overshoot, and relatively slow
response for off-line conditions. Zero-speed-error
control provides 1 percent change in speed output with 1
percent change in reference.
b) Well damped, no overshoot, but very slow time-
constant response for on-line conditions. However, with
the ramp loading control feature, the loading response is
improved.
c) Minimized wear and tear on mechanical parts by
slow response tuning. This improvement will minimize a
unit's down time for maintenance.
The new governor has almost the same structure as
the analog electronic double derivative governor (fig. 1-
2). Therefore, the new governor performance for startup
and load rejection will be similar to that of the double
derivative governor.
34


4.2 Future Work
Future work is needed to improve the off-line
response because the response is still relatively slow.
A feedforward controller may be the solution. Also,
future work is needed to extend this ramp loading control
feature to power (load) reference control instead of
speed reference control.
4.3 Alternative Design
Alternative designs are available that could improve
the load-change response. For a mechanical governor,
there is a scheme that bypasses the transient droop
(reduces the value of Tr in fig. 1-3) Reducing the .time-
constant, Tr, results in a faster loading response time.
However, the unit is less dynamically stable and the
response is still a time-constant response type. In
fact, the response shown in figure 3-5 occurs with the
transient droop bypassed.
Analog electronic governors use two controllers with
different parameters for each unit. One controller is
for off-line operation and one is for on-line operation.
The one for on-line operation produces faster loading
response, but the unit's dynamic stability depends on
35


connection to a large system. This stability problem
will become critical if the unit becomes isolated on a
small portion of the system. In addition, the loading
response is still a time-constant type.
Another design, called the feedforward controller,
will improve the load-change performance. However, the
response of this controller depends on the input signal.
Various input sizes and rates may cause the feedforward
controller time constant to reset and the load response
to revert back to the slow time-constant response type.
4.4 Trend of Governor Control Systems
Today's trend for speed governors is toward digital
controllers. Mechanical and analog electronic governors
are not standard products anymore. Therefore, the cost
of these governors is higher than digital governors. In
addition, digital governors can do advanced control
algorithms and can do more than just speed governing,
such as start and stop control, programmable logic
control (PLC) types, temperature monitoring, etc.
36


APPENDIX
This appendix includes the models of a machine swing
equation, a hydro turbine, and a valve and servomotor
system. The model of a machine swing equation is shown
in figure A-l with some typical parameter values
determined from field test data. For a detailed
description of a machine swing equation, please refer to
reference 15, chapter 3, page 128.
However, the torque angle output has a reset feature
so that the angle.will not integrate to saturation. An
angle of pi (3.1416 radians/second) is the same as an
angle of negative pi (-3.1416 radians/second) and vice
versa. Therefore, whenever the angle integrates up past
3.1416 radians/second, it is reset to -3.1416
radians/second and when the angle integrates down past -
3.1416 radians/second, it is reset to 3.1416
radians/second. This feature is accomplished by blocks
10, 12, 95, 4, 5, 13, and 90.
The model of a hydro turbine is shown in figure A-2.
For a detailed description of the model, please refer to
references 13, 14, and 15-chapter 9. The addition of
high frequency dynamic blocks 83, 90, 94, 95, 97, and 99
37


is intended to produce a better match with field test
data of a turbine characteristic.
The model of a valve and servomotor system is shown
in figure A-3 with typical parameter values determined
from field test data. This figure also illustrates the
model of the gate limit control. The gate limit is
compared to the gate position (block 96). The gate error
is then compared to the actuator driver signal (block
92). The lower value will have the control of the pilot
valve and servomotor system (block 14). The gate-rate
limiter, block 15, is set based on the unit's penstock
characteristic or the unit's loading rate. This limiter
limits the rate of closing and opening of the wicket
gates. Block 97 represents the model of the main
servomotor with an integrator.
38


OJ
V£)
Bus
Figure A-l
The Model of a Machine Swing Equation [13]


o
Linear_
L22 2.0000s2+ 240.00s + 2000.0 2 m 0.62500s2+ 62.5s + 125.00 Hi 0.60000s2+ 3.0000s + 20.000 98
:on tinu
s2 + 20.000s + 2000.0 V0= 0.01333 s2 + 3.1250s + 125.00 V0* 0.01333 s2 + 0.40000s + 20.000 YQ= 0.01 333
:S!LCI>
ilil 941 991
9.0000 3 1
s2 + 0.60000s + 9.0000 s2+ 2,7000s + 3.0000 s + 1.00000 yo= n.oirn
static head
pressure
EE
+ 1
92 85
PP^Gate position Flow over gate rVa pressure
o
.V
Water Column
1 Hi
T.>
YQ= 0.01392
Friction fp
^ Kt _ Flow square
A/

I low
o
14
Pm
Figure A-2
The Model of a Hydro Turbine [13,14,15]


gain higher
Integrator
Figure A-3
The Model of a Valve and Servomotor System


GLOSSARY
Off-line: Unit is operating disconnected from a power
system.
On-line: Unit is operating and is synchronized to a power
system.
Isolated condition: Unit is on-line, and changing unit's
speed reference changes system speed
and the unit's power output.
System stability: How power system quantities such as
voltage, megawatt, megavar, speed, and
power settle to their steady state
values.
Valve activity: Valve movement under steady state
conditions.
Speed-no-load value: Governor controller output value
with the unit operating off-line at
rated speed.
Wicket gate: Gate mechanisms that control the water flow
from a penstock into the turbine of a
hydrogenerator.
Gate Position: Position of the wicket gates, ranging from
zero (closed) to 100 percent (wide open).
42


Gate limit: A mechanical system that limits the wicket
gate position. This system takes over the
control of the wicket gate from the actuator
driver signal when the gate limit is active.
43


BIBLIOGRAPHY
[1] L. M. Hovey, "Optimum adjustment of hydro governor on
Manitoba hydro system," AIEE Trans. Vol 81, Part III, pp.
581-587, Dec. 1962.
[2] M. H. Chaudhry, "Governing stability of a hydro-
electric powerplant," Water Power, pp. 131-136, April,
1970.
[3] D. H. Thorne and E.F. Hill, "Extensions of stability
boundaries of a hydraulic turbine generating unit," IEEE
Trans. Power Apparatus and Systems, Vol. PAS-94, No. 4,
pp. 1401-1408, 1975.
[4] S. Hagihara, H. Yokota, K. Goda, and K. Isobe,
"Stability of a hydraulic turbine generating unit
controlled by PID governor," IEEE Trans. Power Apparatus
and Systems, Vol. PAS-98, No. 6, pp. 2294-2298, 1979.
[5] D. T. Phi, E. J. Bourque, D. H. Thorne, and E. F.
Hill, "Analysis and application of the stability limits
of a hydro-generating unit," IEEE Trans. Power Apparatus
44


and Systems, Vol PAS-100, No. 7, pp. 3203-3211, 1981.
[6] N. S. Dhaliwal and H. E. Wichert, "Analysis of PID
governors in multimachines system," IEEE Trans. Power
Apparatus and System, Vol PAS-97, No. 2, pp. 456-463,
1978 .
[7] L. D. Murphy, L. Wozniak, and T. A. Whittemore, "A
digital governor for hydrogenerators," IEEE Trans, on
Energy Conversion, Vol. 3, No. 4, pp. 780-784, 1988.
[8] D. Findlay, H. Davie,. T. R. Ford, A. G. Marshall, and
D. J. Winning, "Microprocessor-based adaptive water-
turbine governor," IEE Proc. Vol. 127, pt. C, No. 6, pp.
360-369, 1980.
[9] G. Orelind, L. Wozniak, J. Medanic, and T.
Whittemore, "Optimal PID gain schedule for
hydrogenerators decision and application," IEEE Trans,
on Energy Conversion, Vol. 4, No. 3, pp. 300-307, 1989.
[10] D. H. Thorne, E. F. Hill, "Field testing and
simulation of hydraulic turbine governor performance,"
45


IEEE Trans. Power Apparatus and Systems, Vol. PAS-93, pp.
1183-1188, 1974.
[11] L. Wozniak, and T. L. Filbert, "Speed loop
cancellation governor for hydrogenerators, Part I:
development," IEEE Trans, of Energy Conversion, Vol. 3,
pp. 85-90, 1988.
[12] IEEE Committee, "Dynamic models for steam and hydro
turbines in power system studies," IEEE Trans. Power
Apparatus and Systems, Vol. PAS-92, pp. 1904-1915, 1973.
[13] Prabha Kundur, Power System Stability and Control,
Electric Power Research Institute, Power System
Engineering Series, McGraw-Hill, Inc. 1221 Ave. of the
Americas, New York, NY 10020.
[14] E. De Jaeger, N. Janssens, B. Malfliet, F. Van De
Meulebroeke, "Hydro turbine model for system dynamic
studies," IEEE paper 94 WM 186-7 PWRS, Presented at the
IEEE/PES 1994 Winter Meeting.
[15] Louis N. Hannett, James W. Feltes, B. Fardanesh,
46


"Field test to validate hydro turbine-governor model
structure and parameters," IEEE paper 94 WM 190-9 PWRS,
Presented at IEEE/PES 1994 Winter Meeting.
47


Full Text

PAGE 1

A RAMP LOADING RESPONSE GOVERNOR CONTROLLER by Hoa Dinh Vu B.S., The University of Colorado at Denver, 1983 A thesis submitted to the Faculty of the Graduate School of the University of Colorado at Denver in partial fulfillment of the requirements for the degree of Master of Science Electrical Engineering 1995

PAGE 2

This thesis for the Master of Science degree by Hoa Dinh Vu has been approved by William R. Roemish / Date

PAGE 3

Vu, Hoa Dinh (M.S., Electrical Engineering) A Ramp Loading Response Governor Controller Thesis directed by Professor Pankaj K. Sen ABSTRACT A new controller with a ramp loading response feature has been developed and tested for hydrogenerator speed governors that controls Francis-type turbines. This controller produces stable and fast ramp loading responses. The optimum ramp rate is determined by the unit's penstock characteristic or the unit's loading rate. The ramp loading characteristic is an improvement from a time constant loading characteristic, which most of the existing governors have. This controller also provides zero-speed-error control when the unit is operating disconnected from a power system (off-line) and speed-droop-characteristic control (low gain) when the unit is synchronized to a power system (on-line) The controller also prevents excessive valve, servomotor, and gate movement to m1nimize wear and tear in mechanical parts. This abstract accurately represents the content of the candidate's thesis. J recommend its publication. Pa aj K. Sen

PAGE 4

Figures Acknowledgements Chapter 1. Introduction CONTENTS v vi 1 2. The New Governor Control Descriptions .... ...... 9 2 .1 The New Controller . . . . . . . . . . . . . . 9 2.2 2.3 Droop Feedback Control Ramp Loading Control Feature 2.4 Valve and Servomotor Control System 3 Testing the New Governor Controller ............ 3 .1 Unit Off-Line Test ........................... 3.2 Unit On-Line Test ............................ 3.3 System Disturbance Test 4. Conclusions 4.1 Discussion of Results 4.2 4.3 4.4 Future Work Alternative Design Trend of Governor Control Systems Appendix Glossary Bibliography 1 0 13 17 25 26 26 27 34 34 35 35 36 37 42 44

PAGE 5

FIGURES Figure 1-1 Speed Governor Control System 4 1-2 Analog Electronic PID Governor [6,15] 5 1-3 Analog Electronic Double Derivative Governor [7] 6 1-4 Mechanical Rate Feedback Governor [15] 7 1-5 Comparison of Time-Constant & Ramp Loading Responses . . . . . . . . . . . . . . . . 8 2-1 New Proposed Controller (Governor) ................. 19 2-2 New Governor with Speed Droop Feedback ............. 20 2-3 The Reference Control ............................. 21 2-4 New Governor with Ramp Loading Feature ............. 22 2-5 Valve and Servomotor Control System .............. 23 2-6 A Typical Relationship between Gate Position and Power Output ................................ 24 3-1 Unit Off-Line Step Response with the New Governor 28 3-2 A Typical Step Response of a Mechanical Governor .. 29 3-3 A Ramp Loading Response with the New Governor ..... 30 3-4 Field Test of a Ramp Loading Response ............ 31 3-5 A Typical Load Response of a Mechanical Governor . 32 3-6 A System Disturbance Response Test ............... 3 3 A-1 The Model of a Machine Swing Equation [13] ........ 39 A-2 The Model of a Hydro Turbine [13,14,15] ........... 40 A-3 The Model of a Valve and Servomotor System ........ 4 1 v

PAGE 6

ACKNOWLEDGEMENTS This thesis is submitted to the University of Colorado at Denver as the final requirement for the degree of Master of Science in Electrical Engineering. This study would not have been possible without the cooperation and contributions of many people. I especially wish to thank Dr. Pankaj K Sen, Professor of Electrical Engineering at for serving as my academic and thesis advisor and providing me with his suggestions, criticism, and encouragements. I wish also to thank the Hydroelectric Research and Technical Services Group, Bureau of Reclamation. The research for this thesis was done in Reclamation facilities and with Reclamation funding. I am particularly grateful to Mr. J. C. Agee, my team leader, and Mr. Bert Milano, Head of the Group, for making it possible for me to undertake this study. I would further like to take this opportunity to thank the faculty of the University of Colorado at Denver for their efforts in providing an excellent education. Vl

PAGE 7

1. Introduction Over the years, a great deal of effort has been expended in speed control for hydrogenerators with Francis-type turbines [1-12] A governor control system is shown in figure 1-1. Wicket gates control the amount of water flow into a Francis turbine. The wicket gates are positioned by a valve and servomotor control system (blocks 2, 6, 97, and 98 in fig. 1-1). The valve and servomotor control system is driven by a speed controller (governor, super block 7 in fig. 1-1). Most existing analog electronic governors are Proportional-Integral-Derivative (PID) and Double Derivative types as shown in figures 1-2 and Also, most existing mechanical governors are rate feedback controller types as shown in figure 1-4. All of these controllers have time-constant loading response characteristic [1-10] These types of control system can be tuned so that the governor system has fast, and stable (well damped) operation. However, this kind of tuning results in excessive valve, servomotor, and wicket gate movement (activity) because of excessive derivative control 1

PAGE 8

action. This excessive valve, servomotor, and wicket gate movement results in more system down time for maintenance because of wear and tear on mechanical parts. Two options are available to reduce this excessive activity. The first option is to reduce the damping factor for synchronized operation (on-line) This option results in a less stable system. This problem will be critical when the unit is connected to an isolated power system. For a strong-tie power system, the instability of a unit will be stabilized by the power system. The second option is to reduce the control system bandwidth. This option results in a slow response (sluggish) system. The sluggish response will be magnified by a ramp input instead of a step input in reference signal and the speed reference adjustment has a ramp characteristic. A hydrogenerator governor controls its generator speed during unsynchronized operation (off-line) However, the process of controlling speed will have the effect of controlling the unit's power output when the unit is on-line. During on-line operation, adjusting a unit's speed reference will not change the power speed because the inertia of the interconnected system is 2

PAGE 9

much larger than the unit's inertia. However, as the wicket gates move, the unit's load angle is changed. Therefore, the unit's power output is changed. During on-line operation, the power system speed (frequency) is almost constant. Therefore, most of the existing controllers act like a fixed gain and fixed time-constant controller. This condition results in slow time-constant loading performance. The proposed ramp loading control feature provides a ramp loading characteristic instead of a time-constant loading characteristic as shown in figure 1-5. The ramp loading response reaches the final value faster than the timeconstant loading response. The new controller with the ramp loading feature improves the loading performance without sacrificing the unit's dynamic stability and the minimization of valve, servomotor, and gate movement. 3

PAGE 10

.p. Actuator Driver .Pressure Flow and servomotors control Continuous (T11s + 1} + 1} (T1_rS0+1} (Tw+1} (Tf+1)2 r Gate error
PAGE 11

lJl Governor Electronics Valve & Servomotor Controller s Speed Droop {/Feedback I I ______ J +VEL oPEN Q 1 +Tps SeleCt ________ \ ___________ Unit MW Signal Feedback Speed Regulation Type Feedback Figure 1-2 Analog Electronic PID Governor [6,13]

PAGE 12

0"\ Governor Electron los Valve & Servomotor Controller Controller Output I Droo .J + Speed Feedback Regulation Type V Feedback I I UnltMW Signal Figure 1-3 +VEL oPEN -VEL cLOSE Analog Electronic Double Derivative Governor [7]

PAGE 13

s -E ...J s D RP -1 Q 1 --1 + Tp s s 1 + T9s -Rt TRs 1 + TRs Figure 1-4 Mechanical Rate Feedback Governor (13] Gat Posi1 e ion

PAGE 14

00 21 18 15 c "' 0 (.) ., E 12 I= ui > "' a: 9 c .!!! II) c: 0 (.) 6 3 0 0 : : l i : . 0 ;--------r r : : : : i ... ............ ...... j ............ ;.. ... . ... .... ... ......... ... ... ; .... ........... . : : : : : : 1 f 1 Time'-const'ant i j l : i i y t ............. [ j 'tt t t !. Ramp Loading Response! i i fi : : : 1 : l :: : : : : : : ... ........ ..:.:. . ... . .. . .. .. : ............. .... :. .. .... ... ......... ............... . : ..... ... ... .... .l ... . ..... . ...... : ... ..... .......... .!. ............. .... : .............. . . _/:.1 [_ ] __ j_ j__[ L J _LJ I J J I J I J I . 1 .......... l ................. t . ........... l ................ I ............... J . ........ . ....... ...... ... ...... I i : : : : : i I I I : ; i ; : 3 6 9 12 15 second Figure 1-5 18 21 24 27 30 Comparison of Time-Constant & Ramp Loading Responses

PAGE 15

2. The New Governor Control Descriptions 2.1 The New Controller A new controller (governor) for speed control of a Francis-type turbine hydrogenerator is proposed as shown in figure 2-1. This controller includes one integrator, block 97, in the forward path and two zeroes/two poles (lead/lag), block 7, in the feedback path of the speed signal. The input.signal to the lead/lag block is the speed deviation signal. Therefore, the speed feedback signal is subtracted by one per unit to compute the speed deviation, block 9. The integrator is necessary for the zero-speed-error control. One-percent change in reference is one-percent change in speed. The integrator gain, Kit is a factor of the unit's response time (bandwidth) and dynamic stability. A relatively low gain, Ki, ensures that the unit is dynamically stable under all loading conditions and off-line operation. However, a low gain, Kit results in a sluggish response control system. The lead/lag controller is used to improve the unit!s dynamic stability by its phase-lead compensation. Also, with the phase-lead compensation, the integrator 9

PAGE 16

gain, Kit does not have to be too low; therefore, the response time can be improved. The lead/Lag controller is put in the feedback path so that the ramp loading control feature is possible with the integrator. The ramp loading control feature is used to improve the unit's loading performance. For more details of ramp loading control feature, please refer to the ramp loading control feature section. With the ramp loading control feature, the new controller can be tuned for relatively slow response. This slow response ensures the unit's dynamic stability and minimum valve, servomotor, and gate activities. Minimization of valve, servomotor, and gate activities reduces wear and tear in mechanical parts. Therefore, the unit down time for maintenance is also reduced. 2.2 Droop Feedback Control The power system speed (frequency) is almost constant under normal operating condition and can not be controlled by a single unit's governor. This results in a very little dynamic effect of the lead/lag controller on the control system performance. Therefore, the new controller has the same effect as a single pure 10

PAGE 17

integrator. With a single integrator controller, a small change in the power system frequency causes the controller to integrate all the way up or all the way down (overreaction to power system conditions) This overreaction happens because the controller tries to keep the speed-error at zero, which it cannot. The overreaction activity results in either a fully loaded or a motoring unit condition. This kind of operation is not preferred. Therefore, a droop characteristic is needed to avoid the overreaction. A droop characteristic is accomplished by a constant gain (droop setting) feedback, block 6, from the controller output to the summing junction, block 8, as shown in figure 2 -2. This feedback changes the controller from an integrator to a fixed-gain and fixedtime-constant controller. This feedback goes through a switch, block 18, and the switch is controlled by the unit breaker signal. When the unit breaker is open (off-line), the feedback is disconnected. The controller will have a zero-speederror control characteristic (integrator) When the unit breaker is closed (on-line), the feedback is connected. 11

PAGE 18

The controller will have a fixed-gain and fixed-time-constant characteristic. The controller's fixed gain is determined by the inverse proportion of the droop setting. This setting determines the controller's response to a power system frequency change. For an 8-percent (0.08) droop setting, the controller's fixed-gain is 12.5 (1/0.08). With 1 percent change in power system speed or speed reference, the controller output changes by 12.5 percent. This will change the wicket gates by 12.5 percent (not all the way up or down). The fixed-time-constant is determined by the inverse proportion of the product of droop setting and integrator gain, (1 I droop) With a small integrator gain, the time-constant is large. Therefore, the loading response is slow. For an 8-percent (0.08) droop setting and an integrator gain of 0.1, the controller's time constant is 125 seconds. With this kind of setting, the controller takes more than 400 seconds (3 to 4 times the time-constant value) for a load change performance. This setting results in a very slow loading response. Before the droop feedback is connected, the controller output has some value. This value multiplied 12

PAGE 19

by the droop setting is called the speed-no-load (SNL) offset. For minimum disturbance when connecting the droop feedback signal (unit breaker closed) the droop feedback signal has to be subtracted by a constant (block 99 in fig. 2-2) equal to the speed-no-load offset. The SNL offset value shown is determined from field test data. 2.3 Ramp Loading Control Feature With the droop feedback, the controller becomes a fixed-gain and fixed-time-constant controller. The control system has a time-constant response characteristic which is very slow,_ especially with low integrator gain. However, the units loading response time can be improved with a ramp loading control feature. This feature only works when a reference signal is adjusted. Under steady state conditions, this feature does not function and the governor reverts to a fixed-gain and fixed-time-constant controller. The ramp loading controller has two features. The first feature is the reference control as shown in figure 2-3. It has a reference adjustment detection and a 13

PAGE 20

variable gain. The reference signal is a constant signal and it either ramps up or down from an initial to a final value. If the reference signal ramps up, then it is multiplied by a large positive gain. If the reference signal ramps down, then it is multiplied by a large negative gain. Under steady state conditions, the reference signal is multiplied by a gain of 1. The reference-control input is the reference signal. A reference adjustment is detected by a derivative (block 2) The output of the derivative block is connected to an absolute value block (block 12). With the absolute value block, a reference adjustment up or down can be detected. The-absolute value block output is connected to a logic block (block). This block compares the adjustment signal to an epsilon (small number) to be sure that the reference is adjusted. This logic signal controls a switch (block 14) This switch output will be a gain of 1 or a large gain (+100 or -100). This output gain will be multiplied to the reference signal (block 34) The derivative output is also connected to two other logic blocks (block 32 and block 5) Block 5 determines whether the reference is adjusted down, and its output 14

PAGE 21

controls a switch (block 31) This switch output will be -100 if the reference is adjusted down. Otherwise, its output is zero. Block 32 determines whether the reference is adjusted up, and its output controls a switch (block 23). This switch output will be +100 if the reference is adjusted up. Otherwise, its output will be the output of block 31, which can be -100 or zero. The block 23 output will be a large gain value (+100 or -100) This large gain has to be reversed in sign if the reference signal is negative (unit motoring) The reference signal is checked for negative condition by the logic block 99. Block 99 controls a switch (block 25). This switch output will be a large gain value (+100 or 100) or the reverse of the large gain (-100 or +100). The second feature of the ramp loading controller is the constant-rate limiter, block 70, as shown in figure 2-4. The constant-rate limiter and the referencecontrol features are used only when the unit is on-line (unit breaker closed) because they are used to improve the loading performance. They are bypassed by switches, blocks 76 and 98, during off-line operation. During a reference adjustment, the reference signal is multiplied by a large gain. Therefore, the speed 15

PAGE 22

error signal will be large and will be limited by the constant-rate limiter. The effect of the limiting process is that there is no droop feedback and the controller becomes a pure integrator with a different gain. The controller output will either ramp up or down depending on whether the reference signal ramps up or down. The rate of the controller output ramping is controlled by the integrator gain and the limiter value. With an integrator gain of 0.1 and a limiter value of 1, the controller output will change by 0.1 (10 percent) in 1 second. If the controller output needs to change by 62.5 percent from speed-no-load to full load and the limiter value is 6.25, then the controller output changes by 0.625 (62.5 percent) in 1 second (integrator gain of 0.1). If the limiter value is 0.0625, then the controller output changes by 62.5 percent in 100 seconds. The optimum ramp rate is limited by the gate rate of a unit. This gate rate is set mechanically based on the unit's penstock characteristic or the unit's loadingrate, whichever is lower. If the gate rate is set too fast, a vacuum space may be created in the penstock during a gate opening and the penstock may collapse. If 16

PAGE 23

the rate exceeds the unit's loading rate, the unit may heat up too fast. Therefore, the ramp rate should be set lower than the gate rate. If the ramp rate of the controller from speed-noload to full load is 100 seconds, then the reference ramp rate from speed-no-load to full load (0 percent to 5 percent) should also be set at 100 seconds. 2.4 Valve and Servomotor Control System The valve and servomotor control system is shown in figure 2-5. The valve and servomotor system is shown as the super block 97. The valve and servomotor system controller is shown as blocks 6 and 98. This control system controls the wicket gate position (its output) to its input signal. The input signal to this control system is the controller (governor) output. If this control system has a large bandwidth (fast), then the wicket gate position will respond correspondingly to the governor output. A typical governor control system bandwidth is about 0.1 to 0.3 Hertz. Therefore, the valve and servomotor control system bandwidth should be about 0.9 to 1.0 Hertz. If the wicket gates respond at the ramp rate, then 17

PAGE 24

the unit's power output will respond at almost the same ramp rate. A typical relationship between a unit's power output and its wicket gate position is shown in figure 2-6 The controller for this system has a gain, K, and two sets of lead/lag. Two additional lags are included for filtering. A typical model of a valve and servomotor system is illustrated in the appendix. 18

PAGE 25

1-' I.D Integrator I I B-Lead Lag Cont roller Speed Deviation (T1s + 1) (T zS + 1) (T sS + 1) (T4s + 1) Controller Output Ki s 1 Speed error @,.+Reference YO= 0 Ki = Integrator gain Tt = First lead time constant T2 = Second lead time constant T3 = First lag time constant T4 = Second lag time constant Figure 2-1 New Proposed Controller (Governor)

PAGE 26

N 0 Controller Output I Droop 6 Lead Lag Controller Speed Deviation 1J I A (T1s + 1) (T zS + 1) Integrator (T sS + 1) (T4s + 1) B-Ki s 1 Speed error $+ Reference + YO= 0 Switch Unit breaker ul u;:J y ..l2.JlQl Figure 2-2 New Governor with Speed Droop Feedback

PAGE 27

Derivative Reference adjusted s Reference (TS+l) Switch y u > Reference adjusted Is -------I I I r-Ju3-' I 0.001 N 1-' I I Switch I y = u < i 1 Ll1lG a1n 1 lu3-_ Reference lowered ul +J 0.00! Ref erence with aaain Y = U l 'U2 Reference Figure 2-3 The Reference Control ..

PAGE 28

1\.) 1\.) Controller Output I Kl I s YO= 0 0.001 Unit breaker Constant Rate Control llpeede I Limiter Switch Unit breaker __,...,.., u1 u3x un-:-001 Figure 2-4 8-+ Lead Lag Controller Speed Deviation (T1s + 1) (T + 1) (T:rS'+1) (T4s+1) Switch u1 Unit breaker .:..2.91 New Governor with Ramp Loading Feature

PAGE 29

N w Gate limit cntl Controller Gain 98 Actuator Driver and servomotors control SUPER BLOCK Continuous Gate limit Gate position (Tlls + 1) + 1) 2-Gate error Controller output + 1 ) (T uS + 1) (T f + 1 )2 K = Actuator gain Tu = First lead time constant Tll = Second lead time constant T13 = First lag time constant T14 = Second lag time constant Tr = Filter time constant Figure 2-5 Valve and Servomotor Control System

PAGE 30

N ..,. i 0 0.. 1 2 i i i i .8 6 .4 .2 0 .1 ! i j : : : : -r-J-t--r-tr m m : : I 1b4, , . .............. + .. "+"'"'"'''''t"'1'''"''''"''"'''"'i . ...... + ................... + .. + .. : I i i 1 i i .2 .3 .4 .5 .6 .7 8 .9 Gate Figure 2-6 A Typical Relationship between Gate Position and Power Output

PAGE 31

3. Testing the New Governor Controller The new governor with a ramp loading control feature was tested by a real-time digital computer simulation and field tested at Mt. Elbert powerplant. The real-time simulation system includes a Sun Spare 10 station, an AC100 real-time simulator, and a Matrixx-System-build software package. The field test was performed at Mt. Elbert powerplant using the same hardware and software used during simulation tests. For simulation tests of the new controller, a model of a generator was needed. The models of a generator swing equation, hydro turbine, and valve and servomotor system used for simulation tests are illustrated in the appendix. The real-time simulator simulated all blocks shown in figure 1-1. The Sun station was set up as a unit control board to control and monitor the simulated uni t However, for the field test situation, the models of the valve and servomotor, hydro-turbine, and generator swing equation were eliminated. The models of the new governor and the valve and servomotor controller were connecte d to the actual unit's signals. 25

PAGE 32

3.1 Unit Off-Line Test A two-percent step in reference signal response is shown in figure 3-1. The 10to 90-percent rise time of the speed signal is about 18 seconds. No overshoot occurs. This response is still relatively slow. If the controller is tuned for a faster response, the control system will be less dynamically stable and will exhibit higher overshoot. In addition, the valve, servomotor, and gate system will be more active. A typical step response of a mechanical governor is shown in figure 3-2. The control system takes more than two minutes to settle to the final value. Also, the response exhibits high overshoot and some oscillations. 3.2 Unit On-Line Test A ramp loading response of 38 percent is shown in figure 3-3. The controller output, gate position, and power output respond at a ramp rate. A ramp loading response of the actual unit (field test) is shown in figure 3-4. The controller output, gate position, and power output (load) respond at a ramp rate. A typical loading response of a mechanical governor is shown in figure 3-5. The controller output, gate position, and 26

PAGE 33

power output (load) exhibit a time-constant response characteristic. 3.3 System Disturbance Test A system disturbance is simulated by lowering the power system frequency. The new governor control system response data are shown in figure 3-6. The governor raises the gate position to help the power system. The response data indicate that the new governor controller responds correctly to power system disturbances. However, the response is a slow time-constant response because the ramp loading control feature only works whenever a reference exists. Without a reference adjustment, the new governor controller reverts to a fixed-gain and fixed time-constant controller. 27

PAGE 34

Gate position Speed _ ___ .....__ +--:;.." i Figure 3-1 Unit Off-Line Step Response with the New Governor 28

PAGE 35

W! I " ''" ;:r < r:: : 1'0:1 :::. [!212 ill 1::: : I"' Speed C : 1::' ad::; .. ;::: :\ ::': '" '" ::::Iii:; ::: .llii Gate @ :; ;:;; '!" :::: : .::: :;;: :l!l:lt-:::: kF! 'j;,:: 2% .:: : ::::: : ::. :::=:=: ;:: !'!' :: ::i::!J:::k ... " :iiT'i ::::1;:: =:: 10 ;:: I '" '"'T:: ::: :::r:= x "" 1:: 1::: !' t=: 1 : :;: :;:;,::: 1::= [i ::: [i :::: : :: t:::: : li:i bJ if:: !Y ['l ::: .. = .:..:..:!!..::. !::::t:::: . ::::, ::!:1::: 1::: H l"i E' t:: 511:2 :. :::r::: 1 i:ii:t:::i''" 1""1!''! '"" '"' Figure 3-2 A Typical Step Response of a Mechanical Governor 29

PAGE 36

100% :i:::t::IiJ::::r:: : : : : ::: :ii}O:iii:::: :::::::: :::r:JL i:=: :::r:n:s:' / :L :::: ::: : =u> :::; : ; ]:::r:c:nr:: : J :: : :': : : : : ;:,i ::;; ;;;: < '::: ::::c:c:.::::k:;;:: : :: =::: ::=' :=:r:: :::;12.:::: :::: :::: 'i'' :: : -s ::;: =::i '!:Ta=:::ctu::=: = r:: :d=.Sl2 -:r :5 :::: :m < nu :t ilf: ::': 0 -s ::::lA :r: r:: ::: !lif:im 0% -. --+-"__..__ -----!--+ .. -,----+---+-; ... -100% -I Position r=P:f:-0 ::J::: ii' ,, .. ;::; :::; ;:::r::s;::::: == = : p : ; 0 m; ;ill 8; :;mm-::=:-;-:.::m: ,;:: 0% = ::::= : =m mmm :_:: ACCUCHARTo!t Goul Speed 100% Figure 3-3 A Ramp Loading Response with the New Governor 30

PAGE 37

Figure 3-4 Field Test of a Ramp Loading Response 31

PAGE 38

::;==! ''"' tllll Controller EE 10% Gate Figure 3-5 A Typical Load Response of a Mechanical Governor 32

PAGE 39

Gate. position 125% O% Figure 3-6 A System Disturbance Response Test 33

PAGE 40

4. Conclusions 4.1 Discussion of Results The new governor control system has the following performance: a) Well damped, no overshoot, and relatively slow response for off-line conditions. Zero-speed-error control provides 1 percent change in speed output with 1 percent change in reference. b) Well damped, no overshoot, but very slow timeconstant response for on-line conditions. However, with the ramp loading control feature, the loading response is improved. c) Minimized wear and tear on mechanical parts by slow response tuning. This improvement will minimize a unit's down time for maintenance. The new governor has almost the same structure as the analog electronic double derivative governor (fig. 1-2). Therefore, the new governor performance for startup and load rejection will be similar to that of the double derivative governor. 34

PAGE 41

4.2 Future Work Future work is needed to improve the off-line response because the response is still relatively slow. A feedforward controller may be the solution. Also, future work is needed to extend this ramp loading control feature to power (load) reference control instead of speed reference control. 4.3 Alternative Design Alternative designs are available that could improve the load-change response. For a mechanical governor, there is a scheme that bypasses the transient droop (reduces the value of Tr in fig. 1-3) Reducing the _timeconstant, Tr, results in a faster loading response time. However, the unit is less dynamically stable and the response is still a time-constant response type. In fact, the response shown in figure 3-5 occurs with the transient droop bypassed. Analog electronic governors use two controllers with different parameters for each unit. One controller is for off-line operation and one is for on-line operation. The one for on-line operation produces faster loading response, but the units dynamic stability depends on 35

PAGE 42

connection to a large system. This stability problem will become critical if the unit becomes isolated on a small portion of the system. In addition, the loading response is still a time-constant type. Another design, called the feedforward controller, will improve the load-change performance. However, the response of this controller depends on the input signal. Various input sizes and rates may cause the feedforward controller time constant to reset and the load response to revert back to the slow time-constant response type. 4.4 Trend of Governor Control Systems Today's trend for speed governors is toward digital controllers. Mechanical and analog electronic governors are not standard products anymore. Therefore, the cost of these governors is higher than digital governors. In addition, digital governors can do advanced control algorithms and can do more than just speed governing, such as start and stop control, programmable logic control (PLC) types, temperature monitoring, etc. 36

PAGE 43

APPENDIX This appendix includes the models of a machine swing equation, a hydro turbine, and a valve and servomotor system. The model of a machine swing equation is shown in figure A-1 with some typical parameter values determined from field test data. For a detailed description of a machine swing equation, please refer to reference 15, chapter 3, page 128. However, the torque angle output has a reset feature so that the angle.will not integrate to saturation. An angle of pi (3.1416 radians/second) is the same as an angle of negative pi (-3.1416 radians/second) and vice versa. Therefore, whenever the angle integrates up past 3.1416 radians/second, it is reset to -3.1416 radians/second and when the angle integrates down past -3.1416 radians/second, it is reset to 3.1416 radians/second. This feature is accomplished by blocks 10, 12, 95, 4, 5, 13, and 90. The model of a hydro turbine is shown in figure A 2 For a detailed description of the model, please refer to references 13, 14, and 15-chapter 9. The addition of high frequency dynamic blocks 83, 90, 94, 95, 97, and 99 37

PAGE 44

is intended to produce a better match with field test data of a turbine characteristic. The model of a valve and servomotor system is shown in figure A-3 with typical parameter values determined from field test data. This figure also illustrates the model of the gate limit control. The gate limit is compared to the gate position (block 96) The gate error is then compared to the actuator driver signal (block 92). The lower value will have the control of the pilot valve and servomotor system (block 14). The gate-rate limiter, block 15, is set based on the unit's penstock characteristic or the unit's loading rate. This limiter limits the rate of closing and opening of the wicket gates. Block 97 represents the model of the main servomotor with an integrator. 38

PAGE 45

w \0 1 :-p Accelerati on toraue Damping power factor 23 Figure A-1 The Model of a Machine Swing Equation [13]

PAGE 46

,p,. 0 2.0000s2+ 240.00s + 2000 0 0.6250052+ 62,5s + 125.00 I T T
PAGE 47

,p.. I-' gain high e r gate rate 99 Integrator Gate lim! t cnt l .1. Actuator driver Actuator driver P ilot v alves and servomotors Gate Rate Limit Gat e lim! t C L C P Main Servomot ..!. 3 26 0 7 (9 + 4 30001 (S+43.0001 (S+25 0001 (S +63.0 0 0 1 2 025 ... IMain valve Figure A-3 The Model of a Valve and Servomotor System

PAGE 48

GLOSSARY Off-line: Unit is operating disconnected from a power system. On-line: Unit is operating and is synchronized to a power system. Isolated condition: Unit is on-line, and changing unit's speed reference changes system speed and the unit's power output. System stability: How power system quantities such as voltage, megawatt, megavar, speed, and power settle to their steady state values. Valve activity: Valve movement under steady state conditions. Speed-no-load value: Governor controller output value with the unit ,operating off-line at rated speed. Wicket gate: Gate mechanisms that control the water flow from a penstock into the turbine of a hydrogenerator. Gate Position: Position of the wicket gates, ranging from zero (closed) to 100 percent (wide open) 42

PAGE 49

Gate limit: A mechanical system that limits the wicket gate position. This system takes over the control of the wicket gate from the actuator driver signal when the gate limit is active. 43

PAGE 50

BIBLIOGRAPHY [1] L. M. Hovey, 110ptimum adjustment of hydro governor on Manitoba hydro system,11 AIEE Trans. Vol 81, Part III, pp. 581-587, Dec. 1962. [2] M. H. Chaudhry, "Governing stability of a hydroelectric powerplant," Water Power, pp. 131-136, April, 1970. [3] D. H. Thorne and E.F. Hill, "Extensions of stability boundaries of a hydraulic turbine generating unit," IEEE Trans. Power Apparatus and Systems, Vol. PAS-94, No. 4, pp. 1401-1408, 1975. [4] S. Hagihara, H. Yokota, K. Goda, and K. Isobe, "Stability of a hydraulic turbine generating unit controlled by PID governor," IEEE Trans. Power Apparatus and Systems, Vol. PAS-98, No. 6, pp. 2294-2298, 1979. (5] D. T. Phi, E. J. Bourque, D. H. Thorne, and E. F. Hill, "Analysis and application of the stability limits of a hydro-generating unit," IEEE Trans. Power Apparatus 44

PAGE 51

and Systems, Vol PAS-100, No. 7, pp. 3203-3211, 1981. [6] N. s. Dhaliwal and H. E. Wichert, "Analysis of PID governors in multimachines system," IEEE Trans. Power Apparatus and System, Vol PAS-97, No. 2, pp. 456-463, 1978. [7] L. D. Murphy, L. Wozniak, and T. A. Whittemore, "A digital governor for hydrogenerators," IEEE Trans. on Energy Conversion, Vol. 3, No. 4, pp. 780-784, 1988. [8] D. Findlay, H. Davie,. T. R. Ford, A. G. Marshall, and D. J. Winning, "Microprocessor-based adaptive waterturbine governor," IEE Proc. Vol. 127, pt. C, No. 6, pp. 360-369, 1980. [9] G. Orelind, L. Wozniak, J. Medanic, and T Whittemore, 110ptimal PID gain schedule for hydrogenerators-decision and application," IEEE Trans. on Energy Conversion, Vol. 4, No. 3, pp. 300-307, 1989. [10] D. H. Thorne, E. F. Hill, "Field testing a:hd simulation of hydraulic turbine governor performance,11 45

PAGE 52

IEEE Trans. Power Apparatus and Systems, Vol. PAS-93, pp. 1183-1188, 1974. [11] L. Wozniak, and T L. Filbert, "Speed loop cancellation governor for hydrogenerators, Part I : development," IEEE Trans. of Energy Conversion, Vol. 3, pp. 85-90, 1988. [12] IEEE Committee, "Dynamic models for steam and hydro turbines in power system studies," IEEE Trans. Power Apparatus and Systems, Vol. PAS-92, pp. 1904-1915, 1973. [13] Prabha Kundur, Power System Stability and Control, Electric Power Research Institute, Power System Engineering Series, McGraw-Hill, Inc. 1221 Ave. of the Americas, New York, NY 10020. [14] E. De Jaeger, N. Janssens, B. Malfliet, F. Van De Meulebroeke, "Hydro turbine model for system dynamic studies," IEEE paper 94 WM 186-7 PWRS, Presented at the IEEE/PES 1994 Winter Meeting. [15] Louis N. Hannett, James W. Feltes, B. Fardanesh, 46

PAGE 53

"Field test to validate hydro turbine-governor model structure and parameters," IEEE paper 94 WM 190-9 PWRS, Presented at IEEE/PES 1994 Winter Meeting. 47