Combined Cycle Power Plant Optimization Based On Supervisory Predictive Controllers

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Proceedings of the European Control Conference 2009 • Budapest, Hungary, August 23–26, 2009 WeB14.

Combined Cycle Power Plant Optimization Based on Supervisory


Predictive Controllers
Doris Sáez, Senior Member, IEEE, Freddy Milla, Student Member, IEEE and Luis S. Vargas, Senior
Member, IEEE




Abstract—This work considers the optimization of a controllers at regulatory level. In this case, the typical
combined cycle power plant by using supervisory controllers predictive objective function is considered. Suárez et al. [7]
for the turbines and boiler. The design of a hybrid predictive describe an adaptive predictive control scheme for the steam
supervisory controller for a gas turbine is based on a state generator (or boiler) startup of a fossil power plant. A
space discrete model including the switching behavior of a PI
different application is provided by del Real et al. [8], where
control system. The control design is based on an objective
function that represents the economic and regulatory the application of hybrid modeling control techniques to a
performance of a gas turbine, the economic and regulatory two–generator power system connected to the grid. The
performance of a boiler and the regulatory performance of a plant consists of a solar field and a secondary power source
steam turbine by using a dynamic optimal set-point for the formed by an electrolyzer, hydrogen tank and fuel cell stack.
regulatory level. The proposed algorithms are applied to the The system is inherently hybrid as it combines both
gas turbine, boiler and steam turbine of a thermal power plant
continuous and discrete dynamics, it can operate in four
and compared with the standard control strategy with constant
optimal set-points. different modes, depending on the power circuit
Keywors—Predictive control, combined cycle power plant, configuration and the fuel cell stack state.
supervisory control. This work proposes the design of different predictive
supervisory controllers for combined cycle power plants. At
I. INTRODUCTION supervisory level, both linear and hybrid predictive
controllers are applied to the gas turbine. In addition, linear
A S the energy demand increases, especially in big cities,
where the environmental concerns are very important
and resources to produce energy are limited, the efficiency
supervisory predictive controllers are applied for the boiler
and steam turbine. Regarding the gas turbine supervisory
of the operation of power plants becomes of paramount level, the first algorithm uses the classical predictive control
importance. The supervisory controller provides the theory where the switching mode is simplified by the
regulatory level set-points, based on the objective function assumption that only one controller is active at all time. The
optimization. This objective function may represent the second algorithm includes the hybrid behavior by adding the
plant profit, operational costs, process energy consumption switching control problem which is solving by using genetic
and other criteria. algorithms. The control design is based on an objective
Some papers have dealt with supervisory control based on function that represents the economic and regulatory
dynamic models. For example, de Prada [2] proposes a performance of the gas turbine and the boiler and the
predictive control strategy based on the optimization of an regulatory performance of a steam turbine by using a
economic index. This strategy is applied to a chemical dynamic optimal set-point for the regulatory level.
reactor. Katebi [3] describes a decentralized control strategy, Comparisons of the designs and applications to a combined
based on the optimization of a GPC objective function. The cycle unit indicate that the supervisory predictive approach
objective function has only regulatory objectives. The can improve the performance of gas turbine, boiler and
control strategy is applied to a thermal power plant steam turbine control systems.
simulator. In this field there are some industrial applications, The paper is organized in six sections. In section II
for example, a supervisory dynamic matrix controller control loop for combined thermal power plant is described.
(DMC) for a petrochemical process (YHP) [4] that gives Section III explains the supervisory predictive control
good economic results based on linear dynamic models. strategies. In section IV applications to a combined thermal
Tadeo et al [5] propose a constrained predictive supervisory power plant are presented. Finally, section V presents the
controller that can deal with the retuning of the PID work conclusions.

II. TRADITIONAL CONTROL STRATEGY FOR COMBINED


CYCLE POWER PLANT
D. Sáez, F. Milla and L. Vargas are with the Electrical Engineering
Department, Universidad de Chile, Chile (corresponding author: (+562) Combined cycle power plants have high efficiencies and
6784207; fax:(+562) 6720162; e-mail: dsaez@ing.uchile.cl).
require comparatively lower investment costs than other

978-3-9524173-9-3
© Copyright EUCA 2009 4558

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 03:23:49 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the European Control Conference 2009 • Budapest, Hungary, August 23–26, 2009 WeB14.6

technologies. These plants consist of a gas turbine, a boiler and steam mixture takes place in the drum and the risers.
and a steam turbine to generate electricity [1@. Turbines are The steam generated in the risers is separated in the drum,
combined into only one cycle where the energy is from where it flows through the superheater to the high-
transferred from one turbine to the other. The exhaust gases pressure turbine. Then, this steam is recycled to the boiler in
from the gas turbine are used to provide the necessary heat the reheater where its energy content is increased.
for the boiler steam production. Boiler control strategy based on PI controllers is shown in
Figure 2. In this case, the controlled variables and their set-
A. PI Controllers for a Gas Turbine
points are: superheated steam pressure (ps, prs), drum water
Figure 1 shows the gas turbine configuration [1]. The gas level (L, Lr), furnace gas pressure (pG, prG) and superheated
turbine contains a combustion chamber, a compressor and a steam temperature (Ts, Trs). The manipulated variables are:
turbine to convert the fuel energy into mechanical energy. fuel flow (wF), feedwater flow or valve position (x1), air
Air at atmospheric pressure enters the compressor, where it flow to the furnace (wA) and attemporator water flow or
is compressed and mixed with the fuel gas in the combustion valve position (x2).
chamber. The compressed products of combustion are then PI 1
p sr wF
expanded through the turbine and they finally exit to the 1
1

atmosphere as the escape gases. Part of the work generated ps

2
in the turbine is used to drive the compressor, which is Lr
PI
2 x1
usually mounted on the same axis as the turbine. 2
L
In Figure 1, a gas turbine control strategy based on PI PI 3
pGr 3 wA
controllers is shown. This strategy is commonly used in 3
pG
industrial gas turbines for power generation [6]. The
controlled variables and their set-points are: exhaust gas Ts r
PI 4
x2
4
temperature (TTout, TrTout), the power of the gas turbine (Pmech- 4

r Fig. 2. Control strategy based on PI controllers for boiler

(gcNOx, grcNOx) and Z mechanical frequency. The manipulated


G, P mech-G) and the NOx concentration in the exhaust gases
C. PI Controller for a Steam Turbine
variables are the air flow to the compressor (wa), the fuel The steam turbine is a rotating machine that converts the
flow (Fd) and the flow of the steam injected into the potential energy of steam into shaft rotation. First, the
combustion chamber (wis). Notice that the regulatory control superheated steam enters the high-pressure chest via the
level includes the switching controllers, that uses the main admission valves. Then, it goes to the boiler reheater,
minimum signal (f(x)) of three PI controllers (PI2, PI3 and where the steam temperature rises at constant pressure.
PI4) in order to calculate the fuel flow control action. Next, the steam goes through the intermediate-pressure stage
1 -K - H
Cx + Du ux and finally to the low-pressure stage before returning to the
TTout G4 Thermocouple
t
-K -
u
PI
PI1 -K - 1 condenser. The power of the steam turbine is generated
wa
TTout
r G5 G1 through the steam expansion at low pressure.
PI
PI 2
Figure 3 shows the control strategy for the steam turbine.
r
Pmech-G
2 - K- PI
PI
3 f(x) -K - 2
Manipulating the flow of steam to the high-pressure turbine
G3 Saturationn
Saturatio
Pmech-G G2
Fd (Win) controls the steam turbine power (Pmech-S) and set-
Z
-K -
PI
PI4

S
r
r G6 points for the steam turbine power ( Pmech ).
Z
4
3 1
G7
- PI 1
r
gcNOx P rmech-S G1 1 Win
3
4
x' = Ax+Bu PI
PI5 3 Valve
gcNOx y = Cx + Du
Analyser
Dead Zone wis
Fig. 1. Control strategy for the gas turbine 1 -
Pmech-S G2
Fig. 3. Control strategy for steam turbine
B. PI Controllers For a Boiler
In a typical boiler configuration the steam boiler converts Although the generated power of the plant is assigned by
the fuel energy into an increase of water temperature. The the economic dispatch, the generation share of each turbine
water is transformed into steam, which runs through sections can be modified within the space of the technical constraints
of the boiler and finally reaches the condition of superheated of the thermodynamics process. The challenge for the plant
steam. This superheated steam is directed into the steam operators is to deliver the assigned power by the central
turbine. The feedwater runs through the initial heating up economic dispatch by satisfying the constraints and
stage in the economizer and is supplied to the drum. The minimizing the total cost.
drum is connected to the risers, where the furnace heat is In Figure 4, the traditional control configuration for a
used to increase the water temperature and eventually causes combined cycle power plant is presented. In this diagram,
G
r
evaporation. The circulation of water, steam, and the water the set-points for the gas turbine power Pmech and steam

4559

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 03:23:49 UTC from IEEE Xplore. Restrictions apply.
D. Sáez et al.: Combined Cycle Power Plant Optimization Based on Supervisory Predictive Controllers WeB14.6

S
r
turbine power Pmech satisfies: A. Supervisory Controllers for a Gas Turbine
P r
+P r
= PCCPP (1) The supervisory level, shown in Figure 6, is based on a
mech-G mech-S
predictive optimizer that will provide the optimum set-points
Where PCCPP is given for a economic dispatch.
at the regulatory level. The optimization variables proposed
Both turbines and boiler have a regulatory level given by are the set-points r1=Trtout* and r2= Prmech* because they
Proportional Integral (PI) controllers. depend on the manipulated variables u1 = wa and u2 = Fd.
P rmech-S The external set-point Trtout and Prmech for the temperature
Pmech-S
PI Steam and gas turbine power are constant and previously fixed.
Controller Win Turbine
The set-point grcNOx for the controlled variable gcNOx, will
be constant, because it does not affect the optimizer of
supervisory level.
p rs PI Boiler ps In this application, the proposed objective function
Controller
wF contains two terms. The first term is related to the
economical performance, in this case the plant profit (JCp).
The second term is a regulatory criterion (JCr), which takes
Pmech-G
T rTout PI Gas into account the set-point trajectory error and the control
P rmech-G Controller Turbine
Fd ,Wa action effort. The regulatory criterion ensures that the
T Tout
solution is stable within technical constraints. Then, the total
Fig. 4. Control loop for combined cycle power plants objective function to be optimized at the supervisory level is
given in (3).
J (Prmech -G , TrTout)
III. PROPOSED SUPERVISORY PREDICTIVE CONTROLLERS
FOR A COMBINED CYCLE PLANT
In figure 5, the main control loops for a combined cycle
power plant are presented. The power set-point for the
r
complete power plant ( PCCPP ) is determined by the central
dispatcher. The supervisory predictive controllers of the gas
turbine provide the optimal set-points for both exhaust gas
temperature ( Ttout ) and gas turbine power ( Pmech G ).
r * r *

The supervisory linear controller of the boiler and steam


turbine generates the corresponding optimal set-points for
r*
the superheated steam pressure ( ps ) and steam turbine Fig. 6. Proposed supervisory control strategy

mechS mechS J Cp  KGT J Cr


r *
power ( P whit P
r
= PCCPP - P r *
) respectively. Max J

¦C (t  i  1)  ¦ CFd Fd (t  i  1)
mech-G r1 , r2
The proposed supervisory controllers exchange information
based on the equations (1) and Pr mechS K ˜ pr s
N N

Pmech mech G
(2) J Cp P (3)


JCr CrTtout ¨ ¦ Tˆtout (t  j) TTout r OWa ¦'Wa2 (t  i 1) ¸ +
§N ·
i 1 i 1

2 N
Lineal

©j1 ¹
Predictive


CrPmech ¨ ¦ PmechG (t  j)  PmechGr OFd ¦'Fd 2 (t  i 1) ¸
Supervisory

§N ˆ ·
i 1
Control
P rmech-S* 2 N

©j1 ¹
Pmech-S
PI Steam
Controller Turbine i 1
Win
Lineal

Where TˆTtout (t  j ) is the j-step ahead prediction for the


Predictive Optimization variables are reference r1 and r2 [6].
Supervisory
Control
r*
p s PI Boiler ps
Controller exhaust gas temperature, CrT 103 is the weighting factor
of the regulatory level and OW
wF Tout

Hybrid
Predictive a
1015 is the control
P rCCPP
Supervisory
Control weighting. The external set-point trajectory TTout r
1015.8

fixed. Besides Pˆmech  S (t  j ) is the j-step ahead prediction


P r
mech-G
*
for the exhaust gas temperature is constant and previously
Pmech-G
PI Gas
T rTout* Controller Turbine
Fd ,Wa
T Tout
for the gas turbine power, CrPmech 1 is the weighting factor
Fig. 5. Proposed supervisory control strategy for CCPP

4560

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 03:23:49 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the European Control Conference 2009 • Budapest, Hungary, August 23–26, 2009 WeB14.6

of the regulatory level, O F


G X >G 0 " G N 1 @
1017 is the control weighting The optimization vector of genetic algorithm:
10 5 is the
d
(7)
where CF 100 is the fuel price, C Pmech
power price factor and K is a practical weighting factor. The
d
The optimization vector for the quadratic problems is
X ª¬Ttout PmechG Wa Fd 'Wa 'Fd r1 r2 ua ub yab º¼
G G G G G G G G G G G T
external set-point trajectory Pmech G 3.386 u 107 for the gas
(8)

turbine power is constant and previously fixed. N=10 is the the control law is given by the optimal set-points:
G G (9)
prediction and control horizon. r1* (t ) r1 (1) r2* (t ) r2 (1)
Figure 7 shows a detail of a regulatory level block in
B. Supervisory Controllers for a Boiler
Fig.1. As shown in Fig. 6, the gas turbine system is
regulated by proportional integral controllers (PI). The The supervisory control design is based on the
element H (Thermocouple), from figure 7 is the first order optimization of the objective function defined in (10). The
transfer function of ux ; G1, G2 , G3 , G4 an G5 are static predictive optimizer requires the linear models for the boiler
gains, f(x) is a minimum operator. And r1=Ttout* and r2 = process and the PI controllers.
Pmech*. The following optimization problem at a supervisory level
for the boiler is solved by quadratic programming:

1 -K - H ux Máx. JWf J C  K Bo J Cr

¦C
Cx +Du
TTout G4

W f (t  i  1)
Thermocouple
N
u
PIPI1 -K- JC (10)

¦ Pˆ (t  j )  r
-K - 1
Wf

 OW f ¦ 'W f 2 (t  i  1)
r G5 G1 wa i 1
TTout
N 2 N
PI
PI 2
set  po int_ PS
ua J Cr
r s
Pmech yab yabx j 1 i 1
2 -K- PI
PI 3
ub
f(x) 2

4.5251 *10 6 , OW
Fd
with CWf 10 , K Bo 1 , rset  po int_ Ps
G3 Saturation G2
Pmech
Fig. 7. Details of regulatory level.
f
1011

The switching controller behavior can be modeled as C. Supervisory Controllers For a Steam Turbine
hybrid system with integer variables shown in Fig.7. is
The supervisory control problem for a steam turbine is
given by the following expression:
based on the following optimization problem, solved by

f ( x) min ua (t ), ub (t )
quadratic programming and considering linear models of the
yab (t ) (4)
process and models for the corresponding PI controllers.

Min J Ps =J Pmech-S +OST J ǻU

¦ Pˆ
This expression can be reformulated as:
yab (t  k ) ua (t  k )G k  ub (t  k ) ˜ (1  G k )
(t  j )  rset  po int_ Pmech S
(5)
where Gk is auxiliary integer variable.
N
J Pmech  S
2 (11)
mech  S

¦ 'u
j 1

(t  i  1)
N
J 'U Pmech  S
2
In general, the supervisory controller minimizes the i 1
objective function defined in equation (3), considering the
process models and the regulatory level models. Therefore, a with rset  po int_ Pmech  S 1.1*107 and OST 105
mixed integer non linear optimization problem is solved.
The solving of the supervisory hybrid predictive
controller based on genetic algorithms is: IV. SIMULATION RESULTS

A. Evaluation Basis

defined by means of N states of binary variables G k (with


1. At each instant, each individual of the population will be
The proposed supervisory controllers for Gas Turbine are

0,", N  1 )
compared with a control strategy, where the optimum set-
k points are constant and calculated from a static optimization
2. The fitness of each individual is obtained as a result of of the objective function defined in equation (1). Then, the
the solution of the resultant quadratic optimization problem. static set-point is:
 
Max J J Cp  KGT J Cr
The problem of optimization to be solved is: CP CF
2CrPmechGKGT 2kPmech CrPmechGKGT
(12)
mechG
mechG mech G
r *

G X
P P d

(6)

subject to: The process model equation, linear equations


with k Pmech static model constant.
of the regulatory system equations, non linear equations of
the regulatory system equations, and equation (4). On the other hand, for the comparative analysis, a
disturbance is included, given by changes in the temperature

4561

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 03:23:49 UTC from IEEE Xplore. Restrictions apply.
D. Sáez et al.: Combined Cycle Power Plant Optimization Based on Supervisory Predictive Controllers WeB14.6

of the air mass flow into the compressor in order to produce linear hybrid predictive controller of the steam turbine
different operating conditions. The disturbance values (Tcin) present the best power regulation of the CCPP (see Figure
move between 276 and 294 [ºK] for 450 seconds. 12).
In order to quantify the improvement of the proposed
strategy, a standard profit indicator, based on the V. CONCLUSIONS
comparison of the control strategy with constant set-point is
used: This paper presents a new approach for the problem of

§ ·
switching modes for the regulatory control of gas turbine of

¨100  100 ¸%
J Cp with supervisory level a combined cycle power plant. The solution proposed is
¨ ¸
© ¹
Profit (13) based on a hybrid predictive control theory and is applied at
J Cp with constant set-points
supervisory level.
The control design is based on an objective function that
where JCp is given by the economic objective function in represents the economic and regulatory performance of a gas
equation (3). turbine and boiler, and also the regulatory performance of a
A. Supervisory Control Strategies steam turbine by using a dynamic optimal set-point for the
regulatory level.
Figures 8 and 9 present the response systems (Pmech-G and
The total regulation of the system can be improved and
TTout) using the multivariable predictive controller (MPC)
savings up to a 3.64% in fuel for the boiler can be obtained
and a hybrid predictive controller algorithm (HPC-GA) with
with an adequate tuning of this series of predictive
an economic level objective function. It has better regulatory
controllers.
behavior than the control strategy with constant set-points.
Figure 10 shows steam superheated pressure fuel at the TABLE 1. SIMULATION RESULTS FOR GAS TURBINE
Boiler. Figure 11 shows a closed-loop power response with
the proposed supervisory controllers in the Steam Turbine. Supervisory predictive Regulatory level
controllers in Gas Turbine with constant
Figure 12 shows a closed-loop total power response PCCC MPC HPC-GA set-point
with all the proposed supervisory controllers.
J Cp 2798.7 2800.1 2770.5
The mean values of set-point trajectory error and the
control action effort behaviour indexes in the Gas Turbine J Cr 1.451*1015 1.479*1015 0.978*1015 KGT 3

(13), for KGT =3 and KGT =1.5 are shown in Table 1; the
and the mean values of the objective functions (3) and profit Economic 2.088 % 2.1388 % -
profit
J Cp 2828.3 2830.0 2770.5

KGT 1.5
mean values of set-point trajectory error and the control 15 15
J Cr 5.321*10 5.657*10 0.928*1015
action effort behavior indexes in the Boiler (10) are shown
Economic 3.1637 % 3.2274 % -
in Table 2. profit
There are savings of a 3.45% in the boiler when
implementing the supervisor controllers in the CCPP gas
turbine, and the effect when including a linear supervisor TABLE 2. SIMULATION RESULTS FOR BOILER
controller in the boiler produces a benefit of 0.14% over the
Without linear supervisory With linear supervisory
solely hybrid controller of the gas turbine from the predictive controllers in predictive controllers in
combined cycle thermoelectrical central. Boiler Boiler
This can also be observed in Figure 10, where the Controllers JWf profit JWf profit
implementation of predictive controllers in the gas turbine over Gas w/c PI w/c PI
Turbine
produce the effect of diminishing the flow of fuel use by the MPC 129.9748 3.4127 129.7974 3.5446
boiler, nevertheless, the steam flow is enough to (together HPC-GA 129.8386 3.5140 129.6661 3.6422
with the steam turbine), reaching the power set-point of the PI 134.5673 - 134.5674 -
r
CCPP ( PCCPP ).
All the supervisor predictive controllers applied to the gas
turbine with an objective function with economic and
regulatory characteristics, produce an improvement of two
types of magnitude in regulation of the total power of the
CCPP regarding the controller with a constant set-point.
Now, if we consider predictive controllers with objective
functions with only regulatory characteristics, the controller
with explicit enumeration HPC-EE and the controller with
genetic optimization HPC-GA for the gas turbine plus the

4562

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 03:23:49 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the European Control Conference 2009 • Budapest, Hungary, August 23–26, 2009 WeB14.6

6
x 10
1050
10.2
1040
(t) [K]

1030 10.1

Pmech (t) [W]


Tout r

1020 10 MPC
HPC-GA
T

S
1010 9.9 Controller with constant set-point
1000
1700 1800 1900 2000 2100 2200 2300 2400 9.8

9.7
1050
9.6
1040 1000 1500 2000 2500 3000 3500 4000
(t) [K]

1030
Fig. 11. Closed-loop response of steam turbine power with the proposed
Tout

1020
supervisory controllers (Steam Turbine).
T

1010 7
x 10
1000
1700 1800 1900 2000 2100 2200 2300 2400 4.45 MPC

Output Pmech of CCPP [W]


MPC HPC-GA
HPC-GA 4.44 Controller with constant set-point
48
Controller with constant set-point
Wa(t) [Kg/s]

4.43
46
4.42
44
4.41
1700 1800 1900 2000 2100 2200 2300 2400
time [s] 4.4

3380 3390 3400 3410 3420 3430 3440 3450 3460 3470 3480
Fig. 8. Closed-loop response of exhaust gas temperature and air flow with time [s]
the proposed supervisory controllers (Gas Turbine). Fig. 12. Closed-loop response of total power PCCPP with the proposed
7
x 10
supervisory controllers (CCPP).
3.44
PmechG r (t) [W]

3.42
3.4
3.38 VI. REFERENCES
3.36
1700 1800 1900 2000 2100 2200 2300 2400
7 [1] A.Ordys, A. Pike, M. Jonson, R. Katebi, M. Grimble, “Modeling and
x 10
simulation of power generation plants”, Springer-Verlag, London,
3.44
Great Britain, 1994.
PmechG(t) [W]

3.42
3.4
[2] C. de Prada, A. Valentin, “Set-point optimization in multivariable
3.38
constrained predictive control”, Proceedings of the 13th World
3.36
Congress of IFAC International Federation of Automatic Control, San
1700 1800 1900 2000 2100 2200 2300 2400 Francisco, June 30 - July 5, 1996, pp. 351-356
MPC [3] M. Katebi, M. Johnson, “Predictive control design for large-scale
0.64 HPC-GA systems” Automatica, Vol. 33, Nº 3, pp. 421-425,1997
Controller with constant set-point
0.63 [4] C. Ramirez, M. Ruiz, C. Lago, C. Ruiz, O. Santollani, Ch. Johston,
Fd(t) [Kg/s]

0.62 “Controladores por matriz dinámica aplicados a procesos de


0.61
refinación”, Proceeding of the International Society for Measurement
and Control, ISA Show Brasil, 1995.
0.6
1700 1800 1900 2000 2100 2200 2300 2400 [5] F. Tadeo, T. Alvarez, M. Grimble, R. Vilanova, “Implementation of
time [s ] constrained predictive outer-loop controllers: Application to a boiler
Fig. 9. Closed-loop response of gas turbine power and fuel flow with the control system”, Proceedings of the IFAC 15th Triennial World
proposed supervisory controllers. (Gas turbine) Congress, Barcelona, Spain, 2002.
[6] D. Sáez, F. Milla, A. Ordys, “Hybrid predictive supervisory control
6

4.25
x 10 based on genetic algorithms for a gas turbine of combined cycle power
plants”, European Control Conference, July, Kos, Greece, 2007.
4.2
[7] D. Suarez, A. Sanchez, “Adaptive predictive control for the steam
4.15
generator of a fossil power plant”, Systems, Man and Cybernetics”,
PS(t) [Pa]

4.1 2007. ISIC. IEEE International Conference. Page(s):2806 - 2813


4.05 [8] A. del Real, J. Arce, C. Bordons, “Hybrid model predictive control of
4
a two-generator power plant integrating photovoltaic panels and a fuel
cell”, Decision and Control, 46th IEEE Conference, 2007.
3.95
1000 1500 2000 2500 3000 3500 4000
MPC
HPC-GA
Controller with constant set-point ACKNOWLEDGEMENTS
13.5
13.4
This article has been partially financed by Project Anillo
Wf(t) [Kg/s]

13.3
13.2 ACT-32, and by FONDECYT Project 1080009.
13.1
13

1000 1500 2000 2500 3000 3500 4000


time [s]

Fig. 10. Closed-loop response of superheated steam pressure and fuel flow
(Boiler)

4563

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 03:23:49 UTC from IEEE Xplore. Restrictions apply.

You might also like