Professional Documents
Culture Documents
00900506
00900506
4, DECEMBER 2000
Abstract—The action of excitation system protective limiters standards. For illustration purposes, we will focus on the lim-
and var regulation are often not taken into account in planning iter designs used in the GE EX2000 excitation system. The input
studies, and actual units may be quite different than simulations. points for the OEL and UEL limiters are shown in Fig. 1, for the
This may lead to incorrect predictions of actual stability limita-
tions, or even voltage collapse phenomena. The design of the exci- IEEE ST4B model representing a static potential or compound
tation system protective limiters such as Over Excitation Limiters source excitation system [3]. The current IEEE standards [2],
(OEL), Under Excitation Limiters (UEL) and Volts/Hertz (V/Hz) [3] do not include limiter models but there is a recent focus to
is explained, and models presented. For system events where these include limiter models so that their performance can be included
controls and limiters may become active, modeling in transient in stability studies [4], [5].
stability studies is important to correctly predict volt/var perfor-
mance. Specific models applicable to the GE EX2000 digital-based
control systems are shown for reference, although the general con- II. OVER EXCITATION LIMITER (OEL)
cepts discussed here are applicable to any other excitation equip-
ment. An important aspect of the new digital-based excitation con- The purpose of the Over Excitation Limiter (OEL) is to pre-
trols is that they allow for more intelligent limiter functions and vent excessive current versus time (heating) in the field circuit
protective features not convenient with older equipment. Test pro- of the generator [3], [5]–[7]. The monitored variable is typically
cedures and results are shown to illustrate the limiter performance. field current or field voltage. There are a number of ways the
Index Terms—Excitation controls, generator controls, protective field current can exceed full load levels. The two most common
limiters, system stability, voltage stability. would be during an overload when the system voltage is re-
duced, and transiently when there is a close in fault on the ma-
I. INTRODUCTION chine. The system voltage reduction (collapse) scenario typi-
cally results in long time sustained forcing. The transmission
Fig. 1. Potential source model (Type ST4B) showing input for OEL and UEL limiters.
Fig. 3. Block diagram showing the protection inverse time calculation used in
the OEL logic.
Fig. 4. Plot of PRIT function compared to ANSI/IEEE Standards C50.13 field circuit short-time capability.
in Fig. 4. There are two curves given in the figure, one marked Increases in the vars absorbed by the machine result in low flux
as trip level and the second marked as limit level. The accumu- levels in the machine which tend to decrease both the transient
lator on field current error in the PRIT module has actually three and dynamic stability margins. Also, the flux distribution leads
comparator circuits on the output, the highest level for the trip to concentration in the end winding region of the stator and re-
output, a second level (typically 85%) for a transfer output (if sults in eddy currents and heating in the core end region [4],
there are two regulators and there is a transfer to a back-up regu- [8]–[13]. As in the OEL the effects are thermal in nature and
lator), and the lowest level for activation of the limiter. The lim- we can tolerate transient penetration of the limits. The steady
iter level is typically set to somewhere between 50 and 75% of state limits are normally set slightly within (or sometimes per-
the trip level. In Fig. 4, a level of 70% was assumed in the plot- haps equal to) the generator capability curve. The transient or
ting of the limit curve. A comparison of the requirements from dynamic stability may require increasing the settings above the
the IEEE/ANSI Standards C50.13 [5] is also plotted in Fig. 4. capability curves in some cases. The UEL limit is re-calibrated
The design point for the PRIT curve is to have coincidence of the based on square of terminal voltage in the control since the ma-
120 second trip time at 112% IFFL per the IEEE/ANSI curve. chine capability is different as the voltage changes.
For higher values of field currents the PRIT trip curve (and limit The UEL control functional block diagram is shown in Fig. 5.
curve) are conservatively within the requirements set forth in the The form of the UEL regulator is a PI control, similar to the AVR
ANSI standards. regulator and the OEL field current regulator. The setpoint for
The OEL limiter, and the UEL limiter which will be described the UEL regulator is a bandwidth limited signal from the user
next, have the capability of modifying their limits based on ei- entered UEL curve, a function of watts level and re-calibrated by
ther hydrogen pressure (if the generator is hydrogen cooled) or the square of terminal voltage. There is a time constant, , in
inlet air temperature measurement. Also, it should be mentioned the setpoint calculation to minimize interaction of the UEL with
that the OEL is monitoring both the heating and cooling of the local mode oscillations through the watts signal. The setpoint is
generator field. After an OEL event, the integrator in the OEL re- then compared with the VAR output of the generator. Above the
mains active and it takes some time for it to decay to zero value. UEL limit line, the regulator will be driven and held at the lower
If a subsequent OEL event occurs before the machine cools to limit of zero output. After the limit has been reached, there will
normal temperature, the time to accumulate to the limit value be an output of the UEL regulator in a direction to increase the
will be shorter to account for the fact the machine is hotter than voltage and raise the VAR output above the limit line. The re-
normal, hence the limiter controls are smarter. The operation of sponse of the UEL regulator depends on the AVR settings. For
modern OEL algorithms is a significant improvement over some nominal gain AVR (transient gain of 20) the proportional gain
previous designs where the excitation control would simply re- is 0.8, and is reduced by one half if the AVR gain is increased
vert to manual control. A well designed OEL control can help to 50–100 pu, to keep reasonable gain margins in the UEL con-
prevent voltage instability/collapse during critical system con- trol loop. The UEL control is applied to the input to the AVR as
tingencies [7] by maximizing the transient capability of the exci- a summed signal rather than the previous designs which were
tation system while protecting the generator field winding. Re- based on a takeover type control which replaced the AVR input
verting to manual control is not a part of modern OEL strategy. through a high value gate. As the UEL acts through the AVR,
it allows the PSS to continue functioning when the UEL is ac-
III. UNDER EXCITATION LIMITER (UEL) tive [14]. This has a positive impact on stability margins in the
The function of the UEL is to protect the generator against op- underexcited region when at limit. The settings for the UEL reg-
erating too far in the underexcited (leading power factor) region. ulator are determined based on the UEL control loop stability,
MURDOCH et al.: EXCITATION SYSTEM PROTECTIVE LIMITERS AND THEIR EFFECT ON VOLT/VAR CONTROL 443
Fig. 6. LOE trip and transfer curves with mapping of capability/UEL curves
into R–X plane.
V. TEST METHODOLOGY
Fig. 7. Generator capability and UEL curve with mapping of LOE trip/transfer The process of testing the excitation limiters begins with un-
curves into P –Q plane.
derstanding the basic design of the limiter circuits, or algorithms
in the case of today’s newer digital based controls. As the lim-
iters are normally configured to protect the generator and ex-
citation system from operating outside the limits of capability
it may be difficult to arrange tests at these design limits. Such
tests may put the machine and the system in a state of unac-
ceptably high risk. The testing therefore may consist of verifi-
cation of the functionality of the basic components of the lim-
iter using tests done at less extreme operating points. This kind
of functional testing, combined with verification of actual set-
tings is often the best method for insuring the protective limiters
will work correctly when they are needed. It should be recog-
nized that the process of testing may require some settings to be
changed, and careful documentation and resetting of the param-
eters is required to insure correct normal operation. The UEL
Fig. 8. AVR setpoint block (AVRSP) showing V/Hz limiter. and OEL are inherently open loop in the GE EX2000 imple-
mentation during normal operation. A very safe and effective
method of verifying the small signal dynamics of the loops can
to cause arcing flashovers and local core melting. While these
be performed using frequency response techniques. The open
levels of V/Hz are not likely, there is a need to insure the levels
loop frequency response of these loops can be estimated using
do not stay significantly above 105% for an extended time.
commercially available dynamic signal analyzer to measure the
The design of the V/Hz limiter utilizes a limit function in
linear portion of the loop and then analytically completing the
the EX2000 excitation system “Setpoint” software block dia-
loop by including those parts of the loop that are non linear (be-
gram (AVRSP) shown in Fig. 8. First, consider the left side
cause the integrator is being held in limit by a negative error
of this figure which shows all the software blocks used to de-
signal) when the loop is open (which is the normal operating
velop the voltage reference signal. For purposes of definition,
mode). Settings do not require changes nor is the limit actually
the various inputs are considered as parts of the reference set-
placed in service during testing. Operation outside the normal
point. The traditional definition of the voltage setpoint is what
capability of the generator is not required.
is called the voltage ramp. This is equivalent to the setpoint pot
90 ( from the analog systems). The operator inputs of
raise and lower, and a var/pf controller, if used, are the inputs VI. TESTING THE OVER EXCITATION LIMITER (OEL)
to the ramp (SETPOINT ADJUST block in Fig. 8). The other The purpose of the OEL is to prevent excessive current versus
input blocks contain the functions for PSS—the power system time (heating) in the field circuit of the generator [1], [6], [17],
stabilizer, UEL—the under excited limiter, and RCC which is [18]. The monitored variable is typically field current or field
the reactive current compensation. The RCC function is used to voltage. As operation of the OEL normally requires higher than
provide regulation either part way through the generator step up full load levels of field current or voltage there is a problem
transformer, or inside the machine (i.e. droop control for mul- with the testing of this function at normal limit values. Higher
tiple direct bus connected machines, such as cross-compound excitation levels might lead to machine overheating if the limits
MURDOCH et al.: EXCITATION SYSTEM PROTECTIVE LIMITERS AND THEIR EFFECT ON VOLT/VAR CONTROL 445
Fig. 9. Step test of the FCR during commissioning of a 300 MVA steam-turbine generator.
do not function correctly. Also, since most OEL’s are by na- in the right side of this figure. The response of the FCR to the
ture takeover type controls where they replace the main control step changes shows a stable well damped response verifying the
signal while in operation, there is a consequent risk of the ma- stability of the FCR control.
chine not being able to respond if a disturbance occurs during Similar results can be obtained using dynamic signal
the time the machine is being tested because the normal volt/var analyzers to estimate the open loop response of the FCR.
regulation is disabled. Advantages of this technique are that the measurements can be
In the OEL model, described earlier, the functional form of done with the unit operating normally with no readjustment of
the OEL is a field current regulator (FCR). The FCR can be ac- the limiter characteristics. The dynamic signal analyzer is used
tive if an OEL event is detected either off-line or on-line. Nor- to measure the linear portion of the loop and then analytically
mally we are concerned about the on-line setting, but the off-line including the settings of the FCR to complete the open loop
OEL is in service during startup and is transiently active during response. Fig. 10 shows the estimated open loop frequency
the build up of terminal voltage in AVR because of the large response of the FCR for a generator operating on-line. Loop
amortisseur effect on the field current. In this sense the off-line crossover is found at the frequency where the loop gain is unity,
OEL is really protecting against high V/Hz, and acts as backup at 2.84 Hz, and phase margin is found by estimating the phase at
to the normal V/Hz protection, sensing field current rather than the unity gain frequency. Because Fig. 10 is an approximation
terminal voltage. Because the off-line OEL goes into service for of the open loop gain rather than the usual “negative” of the
every start in AVR, the function is verified every time the unit open loop gain, the critical frequency occurs at unity gain and
is started in AVR, sometimes daily for peaking plants. zero phase shift. Positive phase shift at the unity gain frequency
The on-line OEL circuit is also a FCR but with different set- is interpreted as positive phase margin. The phase margin in
tings to correspond to the levels necessary to protect the ma- Fig. 10 of about 51 degrees indicates a very stable control loop.
chine from overheating. On site testing of the OEL can be done An alternative way to test the OEL control is to actually force
by step tests to verify the functional response of the FCR or fre- the generator for an extended time to show the response. As this
quency response tests to verify the open loop response of the test has high risk, it is normally not recommended. The plot of
FCR. For step tests, it is necessary to change the setpoint in the test results from a 250 kVA diesel driven generator in a con-
logic to make the FCR (OEL) active. While the FCR is active, trolled environment test lab is shown in Fig. 11. These plots
steps in reference to the FCR can be programmed in the soft- show terminal voltage in per unit, generator var output in per
ware to demonstrate the stability of the current regulator loop. A unit, field voltage in per unit on air-gap base, and field cur-
sample recording from a 300 MVA fossil unit taken during com- rent in per unit on full load base. For reference, the full load
missioning testing is shown in Fig. 9. This recording shows gen- field current is 3.73 times the air-gap base for this generator.
erator terminal voltage, FCR setpoint, and field voltage during The OEL event was simulated by reducing the terminal voltage
an OEL test. At the left side of the figure is the point at which feedback to 0.8 pu while the generator was operating at full
the FCR becomes active. This point is obvious by inspection of rated kW output and connected to the local utility. The sudden
the field voltage trace. Since the regulation of field current is reduction in feedback signal causes the field voltage to reach
done through a large field time constant, the changes in field ceiling level. Note, as described in the OEL model, full forcing
voltage have to be aggressive in order to have a high bandwidth is available for 1 second, then operation at 125% IFFL for a time
control of current. The required forcing of field voltage during determined by the inverse time curve in the OEL, followed by
the FCR step is also obvious when we consider the step change operation at 100% IFFL. After about 38 seconds the terminal
446 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 15, NO. 4, DECEMBER 2000
event, the integrator in the OEL is still active and it takes some
time for it to decay to zero value. If a subsequent OEL event oc-
curs before the machine had cooled to normal temperature the
time to accumulate to the limit value will be shorter (since it still
has a positive value), to account for the fact the machine has ac-
cumulated excess thermal energy. This is one example of how
today’s limiter controls are now made smarter.
Fig. 11. Test of OEL on an 250 kva diesel turbine-generator in an Exciter Lab.
the UEL setpoint. The plot of unit Mvar output shows a stable
response with some overshoot as it is slightly underdamped. A
simulation of the same event is plotted in Fig. 13, indicating a
close comparison of the responses from the field test. Valida-
tion of the simulation model allows further study of the UEL
response in situations where testing may be difficult.
There are many situations where a plant engineer may be re-
luctant to force the UEL into service, especially at high load
levels. Since the UEL is open loop during normal operation with
the nontakeover implementation, frequency response testing can
be used to estimate the UEL open loop response without the
requirement for placing the UEL in service. A dynamic signal
analyzer can be used to measure the loop characteristics of the
linear part of the loop, then the remaining part of the loop which
Fig. 12. Field test results for UEL response with a 05% step in terminal is nonlinear (in limit) because the loop is open can be included
voltage reference. analytically. Fig. 14 is a plot of the UEL open loop transfer func-
tion. The crossover occurs at unity gain at a frequency of about
damping. A nontakeover UEL should be tested with the PSS in 1.2 Hz, and the phase margin of 86 degrees indicates a very
operation. stable control loop. It has been the experience of the authors
A recording from a test of the UEL on a 155 MVA gas tur- that a non takeover type UEL which has filtering of the power
bine-generator is shown in Fig. 12. The UEL curve was first signal to decouple from local mode oscillation does not require
readjusted to a limit of 5.0 Mvar, just below the operating tuning on site.
point near unity power factor. At this point a 3% step in ter- Additional factors that should be considered in testing the
minal voltage setpoint is applied. Without the UEL, the var re- UEL are a review of the settings to make sure they take into
sponse would be , where is on unit base. For account the classical steady state stability limit (if the unit will
a relatively strong system with pu, the would be switch to manual control), and coordination of the UEL and loss
0.15 pu times 155 MVA or about 23 Mvar. quite a bit more than of excitation (LOE) protection. If an LOE protection scheme
448 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 15, NO. 4, DECEMBER 2000
not so much on functionality as it is verification of the desired [8] G. R. Berube, L. M. Hajagos, and R. E. Beaulieu, “A utility perspective
settings. Older equipment, which has probably not been tested on under-excitation limiters,” Trans. EC, vol. 10, no. 3, pp. 532–537,
Sept. 1995.
since installation, probably requires functional operation testing [9] A. S. Rubenstein and M. Temoshok, “Underexcitation reactive ampere
and settings review on a periodic basis. Increased reliability and limit for modern amplidyne voltage regulator,” AIEE Trans. PAS-15, vol.
capability with newer smart limiters may in fact justify the up- 73, pp. 1433–1438, Dec. 1954.
[10] R. A. Phillips and A. Rubenstein, “Operation of large synchronous
grade of older excitation equipment in today’s power systems. generators in the dynamic stability region with a modem amplidyne
voltage regulator—Part 1: Recommendation for Setting the reactive
APPENDIX ampere limit,” AIEE Trans., Aug. 1956.
[11] J. R. Ribeiro, “Minimum excitation limiter effects on generator response
LIMITER MODEL DATA to system disturbances,” IEEE Trans. EC, vol. 6, no. 1, pp. 29–38, Mar.
1991.
The parameters here are representative settings and do not re- [12] S. E. M. de Olivera and M. G dos Santos, “Impact of under-excitation
flect any particular unit, but provide guidelines for adjustments. limit control on power system dynamic performance,” Trans. PWRS, vol.
10, no. 4, pp. 1863–1869, Nov. 1995.
UEL Model Constants [13] C. G. Adams and J. B. McClure, “Underexcited operation of turbogen-
erators,” AIEE Transactions, vol. 67, pp. 521–528, 1948.
[14] J. M. Soares et al., Operational Aspects of the Actuation of Power
Integral Gain pu System Stabilizers: Coordination with Protection and Other Con-
trollers. Recife, Brazil: V. Sepope, May 19–24, 1996.
Proportional Gain pu [15] J. Berdy, M. Crenshaw, and M. Temeshok, “Protection of large steam
(Nominal Gain AVR) turbine generators during abnormal operating conditions,”, CIGRE
pu (High Gain AVR) Paper 11-05, 1972.
[16] S. S. Choi and R. Larkin, “Steady state stability of a remote gas-turbine
Filter Time Constant seconds generating station: Field test and validation of computer simulation re-
Positive Output limit pu sults,” IEEE Trans. PWRS, vol. 8, no. 4, pp. 1518–1524, Nov. 1993.
[17] N. E. Nilsson and J. Mercurio, “Synchronous generator capability curve
testing and evaluation,” IEEE Trans. PWRD, vol. 9, no. 1, pp. 414–424,
UEL Curve (pu MVA) Jan. 1994.
[18] P. O. Bobo, J. W. Skooglund, and C. L. Wagner, “Performance of
excitation systems under abnormal conditions,” IEEE Trans. PAS, vol.
Segment 1 pu MW pu MVAR PAS-87, no. 2, pp. 547–553, Feb. 1968.
Segment 2 pu MW pu MVAR
Segment 3 pu MW pu MVAR
Segment 4 pu MW pu MVAR Alexander Murdoch received his B.S.E.E. from Worcester Polytechnic Insti-
Segment 5 pu MW pu MVAR tute in 1970, and his M.S.E.E. and Ph.D. from Purdue University in 1972 and
1975, respectively. Since 1975 Dr. Murdoch has worked for General Electric,
OEL Model Constants first in Power Systems Energy Consulting (PSEC). In 1996 he joined GE Drive
Systems, and subsequently rejoined PSEC in 1998. His areas of interest include
rotating machine modeling, excitation system design and testing, and advanced
control theory. He is a Member of the Excitation System Subcommittee in the
Integral Gain pu IEEE.
Proportional Gain pu
Positive Output Limit FCRPLN pu
Negative Output limit FCRNLN pu
Robert W. Delmerico received the B.E.E. degree from SUNY Buffalo in 1974,
and the M.E. degree from RPI in 1975. From 1974 to 1985 Mr. Delmerico
AVRSP Model Constants worked for Westinghouse Electric as a Systems Engineer in drives systems for
large ac and dc machines. He was a Senior Project Engineer for United Tech-
nologies Corp. before joining GE in 1985. Until 1990 he worked in GE’s An-
Setpoint Upper Limit pu alytical Engineering Operation on industrial drives and controls, and joined
V/Hz Gain pu GE Power Systems Energy Consulting where he presently performs advances
system development and studies for drive systems, solid state converters, and
synchronous machine controls.
REFERENCES
[1] IEEE Recommended Practice for Excitation System Model for Power
System Stability Studies, IEEE Standard 421.5-1992.
[2] “Computer models for representation of digital-based excitation sys- Sundar Venkataraman obtained his B.S.E.E. degree from University of
tems,” Digital Excitation System Task Force of the Equipment Working Madras, India, in 1991 and his M.S.E.E. from Iowa State University in 1994.
Group, Trans. EC, vol. 11, no. 3, pp. 607–615, Sept. 1996. He joined GE—Power Systems Energy Consulting in 1994 and has been
[3] “Recommended models for overexcitation limiting devices,” IEEE Task involved in computer simulation studies of excitation systems and power
Force on Excitation Limiters, Trans. EC, vol. 10, no. 4, pp. 706–713, system stabilizers. He is currently Program Manager for generator/excitation
Dec. 1995. system testing. His recent work has involved extensive field testing programs
[4] “Underexcitation limiter models for power system stability studies,” on excitation systems/PSS.
IEEE Task Force on Excitation Limiters, Trans. EC, vol. 10, no. 3, pp.
524–531, Sept. 1995.
[5] Requirement for Cylindrical Rotor Synchronous Generators,
IEEE/ANSI Standard C50.13-1989. Rodney A. Lawson received the B.E.E. and M.E.E. degrees from Univ. of Vir-
[6] G. K. Girgis and H. D. Vu, “Verification of limiter performance in ginia in 1966 and 1973, respectively. He joined the GE as Product Engineer
modem excitation control systems,” Trans. EC, vol. 10, no. 3, pp. in 1966 and is presently Senior Application Engineer in the Drive System De-
538–542, Sept. 1995. partment. Mr. Lawson has extensive experience in the design, application and
[7] C. W. Taylor, Optimizing Voltage Stability: Finding Effective Voltage testing of excitation systems. He is a Registered Professional Engineer in the
Instability Countermeasures. Recife, Brazil: V. Sepope, May 19–24, State of Virginia, a Member of the IEEE Excitation System Subcommittee, and
1996. has been issued 5 patents.
450 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 15, NO. 4, DECEMBER 2000
James E. Curran received his B.Sc. degree in electrical and electronic engi- William R. Pearson received the B.Sc. and M.Sc. degrees from Queen’s Uni-
neering from the University of Strathclyde, Scotland in 1978. He joined GE as versity, Kingston, Ontario, Canada, in 1981 and 1983 respectively. He joined
a Field Engineer in 1980. In 1992 he transferred to GE Drive Systems in Salem, GE Canada in 1984 as a Development Engineer where he worked on the design
VA where he is presently a Software Application Engineer in the excitation sys- of DC motor drives and excitation systems with digital controllers. Mr. Pearson
tems group. He has accumulated extensive experience in the installation and is a Registered Professional Engineer in Ontario. He joined GE Drive Systems
testing of excitation systems. in 1990 in Salem, VA and is presently a Senior Design Engineer. He continues
to work on digital controllers for both excitation systems and drive products.