Sie sind auf Seite 1von 6

A Thermal Power Plant Model for Dynamic

Simulation of Load Frequency Control


Toshio Inoue, Member, IEEE, and Hiroyuki Amano

Abstract-- A thermal power plant model for dynamic


simulation of the Load Frequency Control (LFC) of electric
power systems is presented. In the model, MW response of the
thermal power plant is represented using two components. One is
the slow component responding to the MW demand change from
the LFC, and the other is the fast component due to the primary
frequency (governor) control. For the former, demand rate limit,
boiler response delay, and boiler steam sliding pressure control
are considered. For the latter, turbine governor response, turbine
load reference control and steam pressure change due to turbine
control valve movement are considered.
Index Terms-- Power generation control, Thermal power
generation, Dynamic response, Modeling, Simulation

I. INTRODUCTION

oad frequency control (LFC) takes important part in the


reliable operation of electric power systems. The main
contribution to the generation control required for the LFC
comes from the thermal power plants in Japan.
In the dynamic simulation of the LFC, the turbine
governor model added with the demand rate limit and the time
delay to the MW demand from the LFC is conventionally used.
However, recent experience of the authors on comparisons of
the simulated response with the measured one clearly shows
that the use of conventional model is not enough for
simulating the MW response of the recent thermal power
plants.
It has been concluded that the effects of sliding pressure
control of boiler steam, the steam pressure change due to the
turbine control valve movement and the MW control (turbine
load reference control) by the coordinated boiler-turbine
control should be considered. This consideration is similar to
that given in the recent paper [1].
In this paper, a thermal power plant model suitable for
dynamic simulation of the LFC under the power system
normal conditions is presented. In the model the plant effects
mentioned above are included. The number of parameter
values to be tuned through the comparison of simulations with
Toshio Inoue is with the System Engineering Research Laboratory,
Central Research Institute of Electric Power Industry (CRIEPI), 2-11-1
Iwado-kita,
Komae-shi,
Tokyo
201-8511
Japan
(e-mail:
inotoshi@criepi.denken.or.jp).
Hiroyuki Amano is with the System Engineering Research Laboratory,
Central Research Institute of Electric Power Industry (CRIEPI), 2-11-1
Iwado-kita,
Komae-shi,
Tokyo
201-8511
Japan
(e-mail:
hamano@criepi.denken.or.jp).

142440178X/06/$20.002006IEEE

measurements is small. Most of the parameter values are


available from the plant operational data.
The main purposes of the LFC simulation under the
normal condition are evaluations of the daily performance of
the present LFC and improvements of the performance with
more considering the plant effects. In addition, a need of
studying effective use of the coal-fired thermal power plant in
the LFC is arising in some of the electric power companies.
On the late night-time in the period with light electricity
demand such as spring or fall, most of the electricity is
supplied from nuclear power plants and coal-fired thermal
power plants. There the necessity of utilizing the coal-fired
thermal power plants in the LFC is recognized and some
plants are designed for the use in the LFC.
The normal conditions mean here that there is no severe
accident such as loss of large-capacity generating plants, and
hence the maximum frequency deviation from the standard
frequency stays always within the permissible range. In Japan,
the range is 60 0.2Hz in the western interconnected system,
50 0.2Hz in the eastern interconnected system and 50 0.3Hz
in Hokkaido island system, the north part of Japan.
II. TRANSIENTS OF INTEREST
The MW response of the thermal power plant to the
frequency regulation can be divided into two components; one
is the slow component responding to the MW demand change
from the LFC, and the other is the fast component around the
slow component due to the action of the primary frequency
(governor) control to the generator speed error.
A. Slow Component (Central Component) of MW Response
The slow component is the central component of the MW
response. In representing the slow component, not only the
time delay in transmission of the MW demand signal from the
control center to the plant but also the demand rate limit and
the boiler response delay are significant. The MW demand
signal is transmitted in the form of either the MW value or the
rising/falling pulse, depending on the electric power
companies.
The value of the demand rate limit is usually corresponding
to the generator output power zone. The range of generator
output power for the LFC operation is usually divided into
three zones in Japan. For an example of a large-capacity coalfired unit with sliding pressure control, the high power zone is
0.75 ~ 0.95 pu, the middle power zone is 0.55 ~ 0.75 pu and
the low power zone is 0.4 ~ 0.55 pu. The rate limit is ranging
from 0.03 to 0.05 pu/min, and the value of rate limit is

1442

PSCE2006

smaller at the lower power zone.


B. Fast Component of MW Response around Slow Component
The fast component around the slow component mentioned
above is produced through the turbine governor action to the
speed deviation (and hence frequency) of the turbinegenerator. In representing the fast component, not only the
governor action but also the MW control by the action of the
coordinated boiler-turbine control is important. The MW
control is conducted through the turbine control valve (CV)
position control to fit the generator output power to the MW
demand. In addition, the effects of the sliding pressure control
of boiler steam, the steam pressure change due to the turbine
control valve movement are significant.

stability simulation studies. For convenience, typical values of


EHC turbine-governor model are shown at the list (1).
As for the LFC part (the lower part of the model),
parameter values of the list (2), except for the time delay of
MW response to demand signal (TD), are available from the
plant operational data. TD can be identified from MW
measurements responding to the MW Demand. According to
the authors experiments corresponding to recent plants in
Japan, TD is about 10 ~ 80 sec, generally dependent on the
boiler type and the fuel type. The value of the coal-fired unit is
longer than the oil/LNG fired one, and the unit with unit
sliding pressure control shows longer value than one with the
fixed pressure control.

B. Proposed Model (Fig. 2)


1) Fast Component of MW Response
The turbine-governor part (the upper part of the model) is
A. Conventional Model (Fig. 1)
basically almost same as the conventional model, but the
This is the turbine governor model added with the demand turbine mechanical power deviation (PM) due to primary
rate limit and the time delay to the MW demand from the LFC. frequency (governor) control around the slow component
As for the turbine-governor part (the upper part of the (PM0) is represented. PM is the fast component of the MW
model), parameter values are obtained from those used in the response.
III. MODEL DESCRIPTION

sCVmax
Speed
Error (pu)

CVmax

Droop

T1

s
sCVmin

CV

T2.s+1

T3.s+1

1-K

CVmin
K

LFCmax sLFCmax
LFC
MW Demand (pu)
LFCmin

sLFCmin

Parameter
Droop
T1
sCVmax
sCVmin
CVmax
CVmin
T2
T3
K

Parameter
LFCmax
LFCmin
sLFCmax
sLFCmin
TD

TD

(1) Turbine-governor Parameters List


Description
Speed governor droop (pu)
CV (Control Valve) servomotor lag time constant (sec)
Maximum rate of CV opening (pu/sec)
Minimum rate of CV closing (pu/sec)
Maximum limit of CV position (pu)
Minimum limit of CV position (pu)
HP inlet piping lag time constant (sec)
Reheater, crossover and LP inlet piping lag time constant (sec)
Fraction of power developed by HP

Typical value of EHC [2]


0.05
0.1
0.1
-0.2
1.0
0.0
0.3
7.5
0.3

(2) LFC Parameters List


Description
Note for setting parameter value
Maximum limit of MW demand in each zone (pu)
Available from plant operational data
Minimum limit of MW demand in each zone (pu)
Ditto
Maximum rate limit of MW demand rising (pu/min)
Ditto
Minimum rate limit of MW demand falling (pu/min)
Ditto
Time delay of MW response to demand signal (sec)
Can be identified from MW response measurements

Variable
CV
PM

(3) Variables List


Description
CV position (pu)
Turbine Output Power (pu)

Fig. 1. Conventional Thermal Power Plant Model for LFC Simulation

1443

PM

PM is calculated from the deviation of turbine inlet steam


flow rate (W) using the control valve position deviation
(CV) and the main steam pressure deviation (MSP). W is
calculated by
W = CV MSP 0 + CV 0 MSP

Speed
Error (pu)

sCVmaxCVmax
1

Droop

In the middle part of the proposed model, the effects of the


MW control, the sliding pressure control of boiler steam and
the steam pressure change due to the turbine control valve
movement are represented.

(1)
CV

CV
dCV

W
dW

s
sCVmin CVmin

T1
CV0

T2.s+1

T3.s+1

1-K

PM
dPM

Main Steam Pressure Dynamics


MSP
dMSP

MW Control
Gp

PM
dPM
MSP
dMSP

(T7-T6)s
T6*T7.s 2 +(T6+T7)s+1

MWD
dMWD

T9.s
FXFB

Main Steam Sliding Pressure Control

Frequency
Error (Hz)

1
T8.s+1

FXMSP
MSP0

Boiler Output Power Dynamics

sLFCmax

LFCmax

T5.s+1

LFC
MW Demand

PM0

T4.s+1
LFCmin

MW control

Parameter
Gp
T9
T8
FXFB

Main steam sliding


pressure control
Main steam pressure
dynamics
Boiler dynamics

TD

sLFCmin

FXMSP
T6, T7
T4, T5

(1) Parameters List Additional to Conventional Model of Fig. 1.


Description
Note for setting parameter value
Proportional control gain
Available from plant control system data
Integration control time constant (sec)
Ditto
MW demand correction time constant (sec)
Ditto
Function of MW demand correction (pu)
Ditto
(an example is shown in Fig. 3a.)
Function of main steam pressure set-point (pu)
Ditto
(an example is shown in Fig. 3b.)
Pressure response time constant (sec)
Can be determined through comparing simulation
results with measurements or reference data
Lag and lead time constant of boiler output
Ditto
power response to MW demand (sec)
(2) Variables List

Turbine mechanical power

Turbine-governor

MW control
Main steam pressure
dynamics

Variable
PM
PM
PM0
CV
CV0
CV
W
MWD
MSP0
MSP

Description
Turbine mechanical power (=PM0+PM) (pu)
Turbine mechanical power deviation from PM0 due to primary frequency (governor) control (pu)
Slow (Central) component of turbine mechanical power responding to LFC MW demand (pu)
CV position (=CV0+CV)
CV position or turbine load reference considering sliding pressure control (pu)
CV position deviation from CV0 due to primary (governor) control (pu)
CV steam flow rate deviation due to deviations in CV position and steam pressure (pu)
MW demand correction by frequency bias (pu)
Main steam pressure set-point (pu)
Pressure deviation from MSP0 (pu)
Fig. 2. Proposed Thermal Plant Model for LFC Simulation

1444

PM

(3)

2) Slow (Central) Component of MW Response


In the lower part of the model, the slow (central)
component of the MW response (PM0) is represented. In
addition to the time delay of MW response to demand signal
(TD), the first-order lag and lead time constants (T4, T5) of
boiler output power response to MW demand are included. T4
is generally dependent on the boiler type and fuel type, and the
value of coal-fired unit is longer than the oil/LNG fired one.
T5 is dependent on individual unit.
3) Parameter Values
From the authors experiments on setting the parameter
values corresponding to recent thermal power plants in Japan,
some comments are given for convenience as follows.
a) MW Control
An example of FXFB is shown in Figs. 3(a). As for the PI
control, the proportional gain (G9) is about 0.5 ~ 1.0 and the
integration time constant (T9) is about 20 ~ 60 sec. T8 is about
60 ~ 120 sec.
b) Main Steam Pressure Dynamics
An example of FXMSP is shown in Figs. 3(b). T6 is about
60 ~ 200 sec, generally dependent on the fuel type and the
coal-fired unit has longer value than the oil/LNG fired one. T7
is about 20 ~ 60 sec, generally dependent on the boiler type
and the drum unit has longer value than the once-through one.
IV. ACCURACY OF PROPOSED MODEL
To examine the accuracy of the proposed model, simulation
results of the MW response of large-capacity coal-fired oncethrough unit with sliding pressure control are compared with
the previously developed detailed model [3] or measured data.
In the coal-fired unit, the effects of the MW control and the
main steam pressure seems to be conspicuous since the boiler

Frequency Error
(Hz)

0.24

-0.24

-0.05

(a) FXFB

(2 )

where PM0 is the slow component of the MW response,


MSP0 is the main steam pressure set-point
determined by the pressure program (FXMSP).
b) Main Steam Pressure Dynamics
The dynamics of the main steam pressure is represented by
the first-order lag time constants (T6, T7). T7 is the pressure
response time constant to the control valve movement, and T6
is the response time constant to the boiler pressure control.
The main steam pressure deviation (MSP) is approximated
by
MSP = {1 /(1 + T 6 s ) 1 /(1 + T 7 s )}CV

0.05

Main Steam Pressure


Set-point (pu)

CV 0 = PM 0 / MSP 0

response usually very slow. This is the reason why the coalfired unit is chosen here.

MWD (pu)

a) MW Control
The correction to the turbine control valve position (CV0) is
represented by the PI control with input variables. CV0 is
equal to the turbine load reference. The input variables consist
of the MW demand correction (MWD) from the frequency
bias setting (FXFB) with the first-order lag time constant (T8),
the main steam pressure deviation (MSP) and the turbine
mechanical power deviation (PM). CV0 is calculated
backward from the slow (central) component of the MW
response and the main steam pressure set-point by

1.0

1.0

Generator Output Power (pu)

(b) FXMSP
Fig. 3 Examples of FXFB and FXMSP

A. Accuracy of Fast Component of MW Response


Comparison of the fast component of the MW response due
to the primary frequency (governor) control with the detailed
model has been conducted. The detailed model was already
verified that it showed good agreement with the measured
MW response of a 1000MW-class coal-fired unit in the
primary frequency control [4]. For reference, results of the
conventional model are also included in the comparisons.
1) Frequency Decline of 0.1 Hz
The comparison result is shown in Fig. 4 (a). The proposed
model (indicated by symbol P) shows good agreements with
the detailed model (D) on the changes in the CV position,
main steam pressure and turbine output power. The
conventional model (C) is obviously not proper to represent
this MW response.
2) Frequency Deviations of Normal Condition
The comparison result is shown in Fig. 4 (b). The proposed
model shows almost same responses as the detailed model on
the changes in the CV position, main steam pressure and
turbine output power. The effect of the sliding pressure control
appears in the turbine output power (PM) between the 50%
power and 90% power. The change in PM at 50% power is
clearly smaller than that at 90% power. The conventional
model cannot be used to simulate this MW response.
B. Accuracy of Slow and Fast Components of MW Response
Comparison of simulation result with the measured MW
response of other coal-fired unit in the LFC operation and the
primary frequency control is shown in Fig. 5. In the simulation,
both the MW demand and the frequency are input to the
proposed model and conventional model. The proposed model
shows good agreement with the measured data.

1445

Frequency or Speed (Hz)

PM (pu)

C
Inp ut to Each Model

D
P
Turb ine Outp ut Po wer: PM

Time (sec)

Time (sec)

C
Main Steam Pressure: MSP

MSP (pu)

CV (pu)

D
P

C
P

Contro l Valve Positio n: CV

Time (sec)

Time (sec)

Frequency or Speed (Hz)

(a) Simulation results of plant response under frequency decline of 0.1 Hz

Input to Each Model

Time (sec)

Operation at 50% Power

Operation at 90% Power


C

PM (pu)

PM (pu)

PM
Turbine Output Power: PM

Turbine Output Power: PM

Time (sec)

Time (sec)

Waveform of P is almost same as D.

Waveform of P is almost same as D.

C
CV (pu)

CV (pu)

P
D

CV

Time (sec)

Waveform of P is almost same as D.

MSP (pu)

Waveform of P is almost same as D. Since CV position of conventional


model is varying around 0.5, it is not shown.

MSP (pu)

P
Control Valve Position: CV

Control Valve Position: CV

Time (sec)

MSP

D
P
Main Steam Pressure: MSP

D
P
Main Steam Pressure: MSP

Time (sec)

Time (sec)

Since main steam pressure of conventional model is 1.0, it is not shown.


(b) Simulation results of plant response under frequency deviation in normal condition
P: Proposed Model (Fig. 2), C: Conventional Model (Fig. 1), D: Detailed Plant Model [3]
Fig. 4. Comparison of simulation results of plant response under primary frequency (governor) control

1446

Frequency or Speed Error


(Hz)

MW Demand (pu)

Measurements (Input to Each Model)

VI. ACKNOWLEDGMENT
The valuable contribution to the present study by N.
Kawaguchi, Hokuriku Electric Power Company is greatly
acknowledged.
VII. REFERENCES

Generator Output Power (pu)

Time (sec)

[1]

Measurements
Proposed Model
Conventional Model

[2]
[3]

[4]
Time (sec)

Fig. 5. Comparison of MW response in LFC and primary frequency control


(Large-capacity coal-fired once-through unit with sliding pressure control)

L. N. Bize and J. D. Hurley, Frequency Control Considerations for


Modern Steam and Combustion Turbines, in Proc. 1999 IEEE Power
Engineering Society Winter Meeting, Vol. 1, pp. 543-553. JanuaryFeburuary 1999
P. Kundur, Power System Stability and Control, McGraw-Hill, 1994
T. Inoue, H. Taniguchi, and Y. Ikeguchi, A Model of Fossil Fueled
Plant with Once-through Boiler for Power System Frequency
Simulation, IEEE Trans. Power Systems, vol. 15, No. 4, pp. 1322-1328,
November 2000
H. Amano, T. Inoue, N. Uehara, and H. Kosaka, Comparison of Large
Scale Thermal Plant Model Response with Measured Plant Response in
a Field Test, 2001 National Convention of IEE Japan, pp.2343-2344,
March 2001 (in Japanese).

VIII. BIOGRAPHIES

V. CONCLUSION
In this paper, thermal plant model for the dynamic
simulation of LFC under the normal conditions is presented.
In the model, the effects of sliding pressure control of boiler
steam, the steam pressure change due to the turbine control
valve movement and the MW control (turbine load reference
control) by the coordinated boiler-turbine control is included.
The propose model is presently tested with the extended
comparisons with the measured MW response data. At present,
it is revealed from the comparisons of about 60 thermal power
units that almost all of the difference between the simulation
and measured data stays within 1.5% ~ 2.0% of each unit
rated capacity.

1447

Toshio Inoue received M.E. degree in electrical


engineering from Waseda University, Tokyo, Japan, in
1982. He is now working for the Central Research
Institute of Electric Power Industry. From 1988 to
1989, he was a visiting researcher for the University of
Texas at Arlington. His current research interests are
modeling and analysis of power plant dynamics,
automatic generation control, and analysis of longterm dynamics. He is a member of IEEE and IEE of
Japan.
Hiroyuki Amano received M.E. degree in electrical
engineering from Kyoto University, Kyoto, Japan, in
1998. He is now working for the Central Research
Institute of Electric Power Industry. His current
research interests are nonlinear dynamics in power
systems and modeling of power plant dynamics. He is
a member of IEE of Japan.

Das könnte Ihnen auch gefallen