You are on page 1of 6

724 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 18, NO.

2, MAY 2003

Modeling Combined-Cycle Power Plant for


Simulation of Frequency Excursions
Kazuyoshi Kunitomi, Atsushi Kurita, Yasuyuki Tada, Associate Member, IEEE, Satoru Ihara, Fellow, IEEE,
William W. Price, Fellow, IEEE, Leon M. Richardson, and Gordon Smith

Abstract—The maximum continuous power output of a com- For low-frequency events such as the Malaysia blackout [1],
bined-cycle power plant depends on the system frequency and the the primary issue is how fast the GT can increase its output and
ambient temperature. The temperature control of a gas turbine how the maximum output will be affected by the reduced fre-
limits the exhaust temperature by reducing the fuel flow as the air-
flow decreases with the shaft speed. In this paper, a simple model quency. A previous paper [2] describes a simple, mathematical
of a combined-cycle power plant is derived from a detailed design model of a GT of a CCPP. The model includes the IGV control,
model for an example plant. The model is suitable for long-term the temperature control, and the frequency dependency of the
simulation of power system dynamic performance involving ab- GT output. It is based on physical principles and the detailed
normal frequency conditions. design model of a GT.
Index Terms—Combined-cycle power plants, dynamic mod- Formation of an island with excess generation resulted in par-
eling, gas turbines. tial load rejection and high system frequency. Various controls
act upon partial load rejection to prevent excessive overspeed.
I. INTRODUCTION These controls include a governor, power load unbalance (PLU)
control, GT acceleration control, and ST speed control. The gov-

A TYPICAL combined-cycle power plant (CCPP) consists


of a gas turbine (GT), a steam turbine (ST), a heat-re-
covery steam generator (HRSG), and a generator mounted in
ernor detects the deviation of the shaft speed from a desired
value and attempts to reduce the deviation by adjusting the GT
fuel command under normal and abnormal conditions. The ST
tandem on a single shaft. The GT of a typical CCPP is equipped
speed control also detects the speed deviation and attempts to
with inlet guide vanes (IGVs) that adjust the airflow to maintain
regulate the speed by adjusting the steam valves only under ab-
a high exhaust temperature.
normal conditions. The acceleration control detects the shaft ac-
A mathematical model of a combined-cycle power plant is
celeration and reduces the acceleration by adjusting the GT fuel
needed, including relevant control and protective functions,
command and the steam valves. If the speed recovers, the plant
for long-term simulation of various power system distur-
returns to normal operation as long as the PLU control has not
bances [1]. Long-term simulations of large disturbances are
been activated.
often performed to investigate specific sequences of events
observed in the actual system. Long-term simulations of The PLU control detects excessive unbalance between the
large disturbances are extremely difficult because of possible turbine mechanical power output and the generator electrical
operation of various control limiters and protections. These output and puts the plant in full-speed, no-load condition by ad-
can include generator underexcitation limiters, overexcitation justing the load reference and the steam valves. It prevents the
limiters, and volts-per-hertz limiters as well as underfrequency turbine from tripping. The plant waits for an operator’s com-
protection, overfrequency protection, loss-of-field protection, mand for reloading. Not all of the CCPPs are equipped with
volts-per-hertz protection, and other complex protections in PLU controls.
both plant auxiliaries and the switchyard. Large disturbances, At the design stage of a plant, full load rejection is extensively
especially those involving abnormal frequency operations, studied but partial load rejection is not. It is because full load
cannot be staged and, therefore, it is almost impossible to verify rejection stresses the plant more severely and full load rejection
a complete power plant model through field tests. can be tested while partial load rejection cannot. As long as all
GE power generation, as a supplier of power plants, has de- the valves function properly, the turbine does not trip on full
veloped detailed models of gas turbines, steam turbines, heat-re- load rejection.
covery steam generators, and associated controls for use in plant In this paper, simple, mathematical models of HRSG, ST, and
design and analysis. Critical aspects of these design models have pertinent controls are described and the model of a single-shaft
been verified with field tests or detailed engineering evaluation. CCPP, typical of a GE STAG 109FA is completed. The simple
models have been derived from the detailed design models. Suf-
ficient details have been included in the process and control
Manuscript received February 27, 2002; revised June 24, 2002.
K. Kunitomi, A. Kurita, and Y. Tada are with the Electric Power Research and modeling to represent the response to system disturbances in-
Development Center, Tokyo Electric Power Company, Inc., Tokyo, 230-8510, volving frequency excursions of up to 5%. The model is not
Japan. intended for simulation of startup or shutdown of the CCPP. It is
S. Ihara, W. W. Price, L. M. Richardson, and G. Smith are with GE Power
Systems, Schenectady, NY 12345 USA. generally applicable for initial CCPP loading above 50%. This
Digital Object Identifier 10.1109/TPWRS.2002.807112 minimum is well below the normal operating point of a CCPP.
0885-8950/03$17.00 © 2003 IEEE
KUNITOMI et al.: MODELING COMBINED-CYCLE POWER PLANT FOR SIMULATION OF FREQUENCY EXCURSIONS 725

Fig. 2. Gas turbine configuration.

Fig. 3. Configuration of HRSG HP section.

Fig. 1. Critical components of example plant (GE STAG 109FA). controls. The airflow speed factor characterizes the frequency
dependency of the GT output. The physically based structure of
II. MODEL STRUCTURE AND ASSUMPTIONS the model reveals the possibility of determining the parameters
of the air flow speed factor from the loading tests made at
Fig. 1 shows the critical components of an example CCPP. Its different ambient temperatures. The GT model is based on the
HRSG has three drums for high-pressure (HP), intermediate- following assumptions:
pressure (IP), and low-pressure (LP) steam. Eleven heat ex-
• solely determines ;
changers are shown excluding the economizers that preheat the
• since is negligibly small compared to ;
water admitted to the drums. The heat exchangers consist of
• the shaft power required to drive the compressor is
bundles of tubes through which water or steam flow. The ex-
constant;
haust from the gas turbine goes through and around each of the
• the only significant dynamic effects are the time lags be-
heat exchanger tube bundles and transfers heat to the water or
tween changes in fuel command and changes in and
steam. The main control valve (MCV), intercept control valve
between changes in and changes in .
(ICV), intermediate-pressure admission valve (IPAV), and low-
pressure admission valve (LPAV) adjust the steam admission to A. HRSG and Steam Turbine
the HP, IP, and LP turbines.
The desired performance range of the CCPP model is above The performance of the detailed design model of an ex-
50% load level and between 95% and 105% of nominal fre- ample plant led to the following observations and simplifying
quency. Selection of pertinent variables, model structure and assumptions.
parameters are based on careful inspection of the structure and • The feed water energy gains in the economizers and in the
performance of the detailed design model of an example plant evaporators are additive and there is no need to model the
within the desired performance range. economizers separately. The feed water adjustment is fast,
Gas Turbine: The GT consists of an axial compressor, and there is no need to model the feed water system or
a combustion chamber, and a turbine (Fig. 2). The air is drum level controls.
compressed through the compressor and then mixed with fuel • The temperature and flow variations in the condenser have
in the combustion chamber. The hot gas resulting from the insignificant effects and there is no need to model the at-
combustion process is expanded through the turbine to drive temperators (ATT) or the bypass flows. The water flows to
a generator and the compressor. Pertinent input variables of the attemperators in the HP and reheat lines have insignifi-
the gas turbine are fuel mass flow ( ) and air mass flow cant effects and they can be assumed constant without sub-
( ), which depends on IGV angle ( ), shaft speed ( ), stantial loss of accuracy.
ambient temperature ( ), and atmospheric pressure ( ). • The enthalpy of the steam at the each turbine inlet is con-
Pertinent output variables are net mechanical power output stant since the steam temperature is held constant.
( ), exhaust gas mass flow ( ), exhaust gas temperature Figs. 3–5 show pertinent components of the HP, IP, and LP
( ), and compressor pressure ratio (CPR). A previous paper sections, respectively. The pertinent variables are steam pres-
[2] describes a simple model of a GT including pertinent sures and flows, and valve positions of MCV, IPAV, ICV, LPAV,
726 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 18, NO. 2, MAY 2003

Fig. 4. Configuration of HRSG IP section.

Fig. 7. Model of HRSG IP section and reheater.

Fig. 5. Configuration of HRSG LP section.

Fig. 8. Model of HRSG LP section.

model are the ambient air temperature ( ) and pressure ( ),


electrical power output of the generator ( ), and load reference
( ). The pertinent controls are
• GT controls—as identified previously [2], including tem-
perature control, IGV control, governor, and acceleration
control. The acceleration control limits the acceleration
rate of the turbine by limiting the fuel flow. Its primary
function is to limit turbine acceleration during plant
Fig. 6. Model of HRSG HP section. start-up. It may become active during a partial load rejec-
tion. The acceleration control signal ( ) is transferred
and bypass valves. Figs. 6–8 show models of the HP, IP, and LP to the ST valve controls.
sections, respectively. • ST controls—including inlet pressure controls, inlet pres-
sure limiters, and bypass valve controls for MCV, IPAV,
B. Controls and Interconnections and LPAV, which regulate, limit, and relieve the steam
Controls are site specific, and the structure and parameters of pressure at the valve inlet points. Also included is the
the control models need to be adjusted for each site. Therefore, ST speed control, which operates upon an excessive over-
the contribution of this paper is limited to the identification of speed of the turbine.
controls of an example plant that are pertinent to the desired • PLU control—which affects both GT and ST. The PLU
performance range of the model. Figs. 9–11 define the model control issues two signals, the PLU signal and load rejec-
structure of the CCPP example. The input variables to the plant tion occurrence (LRO) signal, if the difference between
KUNITOMI et al.: MODELING COMBINED-CYCLE POWER PLANT FOR SIMULATION OF FREQUENCY EXCURSIONS 727

Fig. 9. Plant model structure.

Fig. 10. Model structure for GT and controls.

Fig. 12. Load reference change (solid curve: simple model, +: detailed design
model).

Load Reference Change: A load reference change is per-


formed as one of the commissioning tests for demonstration of
plant behavior and correct control performance during normal
operation without any system disturbances. Fig. 12(a) shows a
series of load reference change starting from 1.1 p.u. The gov-
Fig. 11. Model structure for HRSG, ST, and controls.
ernor adjusts the output to the load reference except toward the
beginning and end of the test when the load reference exceeds
the mechanical power to the generator and the electrical
the 1.0-p.u. value and the temperature control limits the output
power exceeds a specified value (e.g., 0.4 p.u.) and if the
to the base load value [Fig. 12(b)]. The IGV control holds the
electrical power decreases faster than a specified rate (e.g.,
GT exhaust temperature essentially constant [Fig. 12(c)]. The
0.35 p.u./s). With a PLU signal, the ICV and LPAV close
exhaust temperature and flow [Fig. 12(d)] determine the ex-
completely. With a LRO signal, the MCV and IPAV close
haust power [Fig. 12(e)]. As shown in Fig. 12(f) and (g), MCV
completely, and is set for a no-load, full-speed value
and IPAV are kept fully open under sliding pressure control,
of 100.3%. Once the unbalance is removed, the PLU signal
while LPAV regulates the LPA pressure under inlet pressure
is automatically reset but not the LRO signal.
control. The steam enthalpy and flow [Fig. 12(h)] determine the
ST output shown in Fig. 12(b). The performance of the simple
III. MODEL PERFORMANCE
model (shown with solid curves) matches well with the detailed
The model performance has been evaluated for scenarios of design model performance (shown with points).
load reference change, low-frequency excursion, and partial The GT exhaust power correlates linearly to the GT mechan-
load rejection. ical power during the load reference change [Fig. 13(a)]. The
728 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 18, NO. 2, MAY 2003

Fig. 13. Mechanical power and GT exhaust power (solid curve: simple model, Fig. 15. Load rejection by 30%.
+ : detailed design model).

Fig. 14. Low-frequency excursion. Fig. 16. Load rejection by 70%.

-axis intercept is the no-load exhaust power that corresponds rejection is less than 40%, and the load reference remains the
to the GT mechanical power driving the compressor. The ST same. The speed decreases to the final value of 101.5%, which
mechanical power follows the GT exhaust power linearly with is above the rated speed due to the governor droop of 5%.
a dynamic delay during the load reference change [Fig. 13(b)]. Fig. 16 shows 70% load rejection. The speed increases rapidly
These relationships suggest a simpler approach to modeling the upon load rejection, and the governor reduces the GT fuel flow;
steam system as previously suggested [3]. consequently, the GT exhaust temperature decreases and the
Low-Frequency Excursion: Fig. 14 shows a low-frequency temperature control closes IGV. At the same time, the PLU con-
excursion simulated with the simple model. Starting with the trol puts the ST at no-load full-speed condition since the load
initial plant output of 90% of the base load of 366 MW, the rejection is greater than 40%. The MCV, ICV, IPAV, and LPAV
system frequency is decreased by 5% over 25 s, similarly to the are closed (a second later, the ICV and LPAV are opened so that
Malaysia incident [1]. This is one of the cases reported previ- the trapped steam is released) and the load reference is set to
ously for the GT only [2]. The governor responds to the low the no-load value of 100.3%. The speed decreases rapidly to a
frequency first and increases the GT output to more than base value substantially below the rated speed. The governor now in-
load; consequently, the exhaust temperature increases. After a creases the GT fuel flow in response to the speed deviation, and
few seconds, the temperature control fully opens the IGV, re- the temperature control opens the IGV to keep the desired ex-
duces the GT output, and effectively lowers the exhaust tem- haust temperature. The final speed of 98.8% is below the rated
perature to a specified value. The final output is close to the speed due to the governor droop and the load reference set to
original level of 90% of the base load instead of 100% due to the no-load value.
the reduced airflow from the compressor running more slowly. The ST speed control does not operate during these simulated
The steam turbine output is essentially invariant during the sim- partial load rejection scenarios; however, this control would act
ulated low-frequency excursion. to protect the unit from excessive overspeed if the PLU is not
Partial Load Rejection: Full load rejection is extensively equipped or fails to operate properly upon partial load rejection.
studied in the design stage of a plant since it is more critical Various controls substantially and rapidly change the GT air-
for the design. Load rejection is tested during commissioning flow and fuel flow during system frequency excursions. The
by tripping a loaded generator. It is extremely difficult to test simulated plant performance is very sensitive to the control pa-
partial load rejection while keeping the generator connected to rameters and the system dynamic characteristics. The system
the system as in an islanding situation with excess generation. frequency deviation may exceed 5%, and the overfrequency or
Fig. 15 shows 30% load rejection. The speed increases rapidly underfrequency protection might trip the turbines. One cannot
upon load rejection, and the governor reduces the GT fuel flow; assume that all pertinent controls will operate to the same ef-
consequently, the GT exhaust temperature decreases and the fect during severe system frequency excursions as they do in
temperature control closes IGV to recover the exhaust temper- the normal frequency band. The plant performance during ab-
ature. The GT acceleration control closes the MCV, IPAV and normal frequency condition cannot be verified by tests. There-
LPAV temporarily. PLU control does not operate since the load fore, simulation results should be evaluated with care.
KUNITOMI et al.: MODELING COMBINED-CYCLE POWER PLANT FOR SIMULATION OF FREQUENCY EXCURSIONS 729

IV. CONCLUDING REMARKS Atsushi Kurita received the B.S. degree from Ibaraki University, Ibaraki, Japan,
in 1979 and the M.S. degree from Shizuoka University, Shizuoka, Japan, in
The GT response is rapid and dominant, and the governor 1981.
and GT temperature control play critical roles during abnormal Currently, he is the Group Manager of the Power System Technology Group,
Electric Power Engineering R&D Center, Tokyo Electric Power Company
frequency operations. The dependency of the GT maximum (TEPCO), Tokyo, Japan, where he has been since 1981.
continuous output on the frequency and ambient temperature is
a significant factor. The PLU control, GT acceleration control,
and ST speed control also influence the plant performance
Yasuyuki Tada (A’96) received the Ph.D. degree in engineering from the Uni-
during partial load rejection. versity of Tokyo, Toko, Japan, in 1998.
These controls and their settings are site specific, and Currently, he is a Researcher in the Power System Technology Group, Electric
therefore, some effort is required to identify control models Power Engineering R&D Center, Tokyo Electric Power Company (TEPCO),
where he has been since 1981.
and verify them with field tests. It may be desirable to consult
the manufacturer on modeling the frequency and ambient
temperature characteristics of a specific GT.
Satoru Ihara (F’01) received the B.S.E.E. and M.S.E.E. degrees from Kyoto
The ST power output follows the GT exhaust power with a University, Kyoto, Japan, and the Ph.D. degree in applied mathematics from
long-time constant on the order of 100 s. The simple model pre- Harvard University, Cambridge, MA.
sented in this paper can be further simplified by replacing the Currently, he is a Senior Research Fellow at GE Power Systems Energy Con-
sulting, Schenectady, NY. He is also a former chairman of the IEEE working
steam part of the model with a simple time-lag model—at least group on load representation for dynamic performance.
for the simulated scenarios. Dr. Ihara is a member of IEE of Japan.

ACKNOWLEDGMENT
William W. Price (F’95) received the B.S.E.E. degree from Clarkson Univer-
J. Undrill of GE Power Systems Energy Consulting provided sity, Potsdam, NY, in 1968, and the M.Eng. degree from Rensselaer Polytechnic
constructive discussions. His contribution is acknowledged. Institute, Troy, NY, in 1971.
Currently, he is a Consulting Engineer at GE Power Systems Energy Con-
sulting, Schenectady, NY. He is a former chairman of the IEEE power system
REFERENCES dynamic modeling subcommittee.
[1] CIGRE TF38.02.14, “Analysis and modeling needs of power systems
under major frequency disturbances,” Rep., CIGRE Brochure no. 148,
1999.
[2] K. Kunitomi et al., “Modeling frequency dependency of gas turbine Leon M. Richardson received the B.S.E.E. degree from Rochester Institute
output,” in IEEE Power Eng. Soc. Winter Meeting, Columbus, OH, IEEE of Technology, Rochester, NY, in 1966, and the M.S.E.E. degree from Union
Publication no. 01CH37194. College, Schenectady, NY, in 1969.
[3] J. Undrill and A. Garmendia, “Modeling of combined cycle plants in Currently, he is a Consulting Engineer at GE Power Systems Energy Con-
grid simulation studies,” in IEEE Power Eng. Soc. Winter Meeting, sulting, Schenectady, NY.
Columbus, OH, 2001, IEEE Publication no. 01CH37194. Mr. Richardson is a former member of the IEEE working group on prime
mover and energy supply models for system dynamic studies.

Kazuyoshi Kunitomi received the B.S. and M.S. degrees from Tokyo Institute Gordon Smith received the B.S. degree from the University of Louisville,
of Technology, Tokyo, Japan, in 1986 and 1988, respectively. Louisville, KY, in 1977.
Currently, he is a Researcher in the Power System Technology Group, Currently, he is a Senior System Engineer at GE Power Systems, Schenectady,
Electric Power Engineering R&D Center, Tokyo Electric Power Company, Inc. NY, engaged in the GE 7H GT/combined cycle system launch project. He joined
(TEPCO), Tokyo, Japan. He joined TEPCO in 1988. GE as an international field engineer in 1981.

You might also like