Professional Documents
Culture Documents
Shabaninia 2012
Shabaninia 2012
ISRN Electronics
Volume 2012, Article ID 134580, 8 pages
doi:10.5402/2012/134580
Research Article
Using LQG/LTR Optimal Control Method to Improve Stability
and Performance of Industrial Gas Turbine System
Copyright © 2012 F. Shabaninia and K. Jafari. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
The gas turbine is a power plant, which produces a great amount of energy for its size and weight. Its compactness, low weigh, and
multiple fuels make it a natural power plant for various industries such as power generation or oil and gas process plants. In any of
these applications, the performance and stability of the gas turbines are the end products that strongly influence the profitability of
the business that employs them. Control and analyses of gas turbines for achieving stability and good performance are important
so that they have to operate for prolong period. Effective control system design usually benefits from an accurate dynamic model of
the plant. Characteristic component parts of the system are considered in this model. Gas turbine system is described by specified
thermodynamic equations that can be used for defining its model. This paper introduces an optimal LQG/LTR control method
for a gas turbine. Analysing the gas turbine dynamic in time and frequency domain by using proposed control compared to PID
controller is followed. Applying this optimal control method can provide good performance and stability for the component parts
of system.
and outputs are related by nonlinear mathematical equa- Nonlinear state equations of gas turbine system in [9, 20]
tions. describe the gas turbine model and are used for control
Next sections are concerned with a general description of objects in this paper. This model is based upon the character-
the gas turbine system, definition of its model, LQG/LTR istics of the component parts of system such as pressure and
method, and introducing state space model and simulation temperature in compressor, plenum, turbine, and rotational
results. Matlab mfile commands are used for this idea. PID speed.
(Proportional Integrated Derivative) control [1] is applied to The compressor pressure (pcomp ) is relative to air mass
this system to compare results. flow rates in compressor (ṁcomp ), plenum (ṁbl ) blow off
valve (ṁbl ) inlet pressure ambient temperature, and com-
pressor temperature according to nonlinear dynamic equa-
2. System Description tion as follows [9, 20]:
Gas turbines are designed for many different purposes. In d pcomp γR
the petroleum industry, they are commonly used to drive = ṁcomp Tcpout − ṁpl ṁbl Tcomp , (1)
dt Vcomp
compressors for transporting gas through pipelines or
generators that produce electrical power. The gas turbine is a where
constant flow cycle with a constant addition of heat energy. It (γ−1)/γηcomp
is commonly referred to as the Brayton cycle [16, 17]. Gas pcomp
Tcpout = Tin ,
turbine consists basically of a compressor, combustion pin
chamber, and turbine. The compressor draws air from the (2)
ambient and increases the pressure of the incoming air. The 2pcomp A2 comp pcomp − ppl
pressurized and hot air then used as combustion air in ṁpl = .
ξRTcomp
combustion chamber, in which a fuel is injected and com-
busted continuously. The hot pressurized gases are then Nonlinear dynamic equation of turbine temperature is
expanded slightly above atmospheric in turbine and cause to
produce rotational speed at output. When compressing the dTtbin
air, power is required and when expanding the hot gas dt
through the turbine, power is generated [11]. The gas turbine
is the best suited prime mover when the needs at hand such as RT Ttbin cpcomp power
= γT ṁthr Tccin − ṁtb Ttbin +
capital cost, time from planning to completion, maintenance ptbin Vcc cpT cpT
costs and fuel cost, are considered [18].
The design of any gas turbine must meet essential criteria −Ttbin (ṁthr − ṁtb ) ,
based on operational considerations such as high efficiency,
high reliability and thus high availability, ease of service, ease (3)
of installation and commission, flexibility to meet various
service and fuel needs, conformance with environmental where
standards, and auxiliary and control systems. The two fac- (γ−1)/γ
Ptbin
tors, which most effect high turbine efficiencies, are pressure Tccin = Tpl . (4)
ppl
ratios and temperature. High-pressure ratios and turbine
inlet temperatures improve efficiencies on the simple cycle Rotational speed described by
gas turbine. It should also be noted that the very high-pres-
sure ratios tend to reduce the operating range of the turbine dN
compressor [18]. dt
Auxiliary systems and control systems must be designed 1
carefully, since they are often responsible for the downtime in = ṁtb cpT (Ttbin − Ttbout ) − ṁcomp cpcomp Tcpout − Tin ,
NI
many units. Control systems provide acceleration time and (5)
temperature time controls for startups as well as control
various antisurge valves. At operating speeds they must reg- where γ = c p /cv is specific heat capacity ratio, R is gas
ulate fuel supply and monitor vibrations, temperatures, and constant (Nm/kg K), I is inertia (Kg m2 ), ξ is friction factor,
pressures throughout the entire range [18]. and η is efficiency.
Other parameters state by specific nonlinear dynamic
equations same as these at [9, 20].
3. Gas Turbine Model The state space model of gas turbine is defined with 5
inputs as
Gas turbine system is described by nonlinear mathematical
equations that used thermodynamic principle rules such as U = ṁcomp ṁbl ṁthr ṁtb power . (6)
conservation of total mass, energy, and ideal gas equation
[2, 12, 17, 19]. The investigation of steady state behavior of And 7 states as
gas turbine in terms of static turbine characteristic is a
traditional area in engineering [12]. X= pcomp Tcomp ppl Tpl ptbin Ttbin N . (7)
ISRN Electronics 3
Figure 1: The Kalman filter. where x(t) ∈ Rn , u(t) ∈ Rm , and y(t) ∈ Rq are the states,
inputs, and outputs vectors, respectively, A ∈ Rn×n , B ∈
Rn×m , Γ ∈ Rn× p , C ∈ Rq×n are the system states, inputs of
To CRT display plant, inputs of disturbance, and outputs matrices, respec-
Fuel tively. The system disturbance w(t) and the measurement
Temperature noise v(t) are p- and q-dimensional uncorrelated Gaussian
white noise processes with zero-mean, and the associated
To CRT display
covariance matrices are defined as
Minimum FSR
value Fuel
Speed system E wwT = W ≥ 0, E vvT = V 0,
select
logic
(10)
Acceleration
rate E wvT = 0,
To turbine
To CRT
display
Start up where E{·} is an expectation function operator, and W and
Shut V are the system disturbance and measurement noise covari-
down
Manual ance matrices, respectively. The problem is then to devise a
Figure 2: Simplified control schematic.
feedback control low which minimizes the cost:
T
T T
J = lim E z Qz + u Ru dt , (11)
T →∞ 0
Parameters of this model are defined as bellows:
where Z = Mx is some linear combination of the states, and
ṁcomp , ṁbl , ṁthr , , ṁtb : Mass flows through compres- R = RT 0, Q = QT ≥ 0 are weighting matrices.
sor, blow off, throttle, and turbine, respectively The solution to the LQG problem is prescribed by the
(kg/s). seperation principle, which states that the optimal result is
achieved by adopting the following procedure. First obtain
pcomp , ppl , ptbin : Pressure at compressor, plenum, an optimal estimate x of the state x, optimal in the sense that
and turbine, respectively (pa). E{(x − x)T (x − x)} is minimized, and then use this estimate as
Tcomp , Tpl , Ttbin : Temperature at compressor, ple- if it was an exact measurement of the state to solve the deter-
num, and turbine, respectively (k). ministic linear quadratic control problem. The point of this
procedure is that it reduces the problem to two sub problems,
the solutions to which are known.
The system dynamic equations can be expressed as The solution to the first subproblem that of estimating
follows the state is given by Kalman filter theory. Figure 1 shows the
·
block diagram of a Kalman filter, which is seen to have the
X = f (X, U). (8) structure of a state observer; it is distinguished from other
observers by the choice of gain matrix k f . Note that the
inputs to the Kalman filter are the plant input and output
vectors, u and y, and that its output is the state estimate
vector x.
4. LQG/LTR Method The second subproblem is to find the control signal
which will minimize the cost:
Linear quadratic Gaussian or LQG problem is a method
based on optimal control theory. The main results of this T
theory are stated in this section. But fuller details can be J= zT Qz + uT Ru dt. (12)
0
referred to [1, 3, 5–7]. This method is one which allows the
designer to shape the principle gains of the return ratio at
either the input or the output of the plant, to achieve required On the assumption that
performance or robustness specifications. Stability is ob-
·
tained automatically [3]. x= Ax + Bu (13)
4 ISRN Electronics
×105
3.5 8
7
3
6
2.5
5
2 4
1.5 3
2
1
1
0.5
0
0 −1
0 5 10 15 0 5 10 15
Time (s) Time (s)
Uncontrolled system LQG controlled system Uncontrolled system LQG controlled system
LQR controlled system LQG/LTR controlled system LQR controlled system LQG/LTR controlled system
(a) (b)
0
−0.01
−0.02
Rotational speed (rev/s)
−0.03
−0.04
−0.05
−0.06
−0.07
−0.08
−0.09
0 5 10 15
Time (s)
Figure 3: Step response of the compressor pressure (a), turbine temperature (b), and rotational speed (c) respect to compressor mass flow
rate.
the solution to this is to let the control signal u be a linear It is shown that the kalman filter and optimal state feedback
function of the state: are asymptotically stable [3].
The equation of the combined Kalman filter-optimal
u = −Kc x, (14)
state feedback scheme is
where Kc is the state feedback matrix gain.
Kc and k f are obtained from solution of algebraic Riccati d x A −BKc x Γw
= + . (17)
equations [3, 5]. dt x K f C A − K f C − BKc x Kf v
By substitute (14) in to (13):
·
If ε define as state estimation error, ε = x − x, then above
x= (A − BKc )x, (15) equations can be rewritten as
as the equation of the close loop system. d x A − BKc BKc x Γw
From Figure 1, The state equation of the Kalman filter is = + , (18)
dt ε 0 A − Kf C ε Γw − K f v
to be
d
x = A − K f C x + Bu + K f y. (16) which shows that the close loop eigenvalues of the LQG
dt compensated plant are just the union of eigenvalues of the
ISRN Electronics 5
×104
12 7
10 6
Compressor pressure (Pa)
−2 −1
0 5 10 15 20 25 30 35 40 45 50 0 20 40 60 80 100 120 140
Time (s) Time (s)
−0.03
−0.04
−0.05
−0.06
−0.07
−0.08
−0.09
−0.1
0 5 10 15 20 25
Time (s)
PID control
LQG/LTR control
(c)
Figure 4: Step response of the compressor pressure (a), turbine temperature (b), and rotational speed (c) respect to compressor mass flow
rate for LQG/LTR and PID control.
optimal state feedback scheme with those of the Kalman model is minimum phase and that it has at least as many
filter. The overall scheme is therefore internally stable under inputs as outputs. Alternatively, the LQG loop transfer func-
the stated assumption [3]. tion can be made to approach state feedback transfer func-
tion by designing k f in the Kalman filter to be large using spe-
cific procedure. Again, it is necessary to assume that the plant
5. Loop Transfer Recovery (LTR) model is minimum phase and that it has at least as many
outputs as inputs [1, 3, 5, 6].
Since both the optimal state feedback regulator and the
Kalman filter have such good properties, it might be expected
that LQG compensator would generally yield good robust- 6. State Space Description
ness and performance. Unfortunately, this is not the case
and the LQG designs can exhibit arbitrarily poor stability The state space detailed model of gas turbine at a given oper-
margins [3]. ation point helps to understand the dynamic behavior of sys-
The LQG loop transfer function can be made to approach tem. A nonlinear model can capture the dynamic behavior of
filter transfer function with its guaranteed stability margins if gas turbine system [12]. By using nonlinear equations de-
Kc , state feedback gain, is designed to be large by specific scribed at [9, 20], a linear time invariant, LTI, model of the
procedure [1, 3, 5, 6]. It is necessary to assume that the plant gas turbine is developed for control purposes. LTI state space
6 ISRN Electronics
×10−8 ×10−5
16 7
14 6
10
4
8
3
6
2
4
2 1
0 0
−2 −1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Time (s) Time (s)
Figure 5: Impulse response of the rotational speed (a) and turbine temperature (b) respect to power.
1
power required by the operation or process is accomplished
using parameters such as fuel flow, compressor inlet pressure,
0.8
compressor discharge pressure, shaft speed, compressor inlet
0.6 temperature, and turbine inlet or exhaust temperature [19].
0.4 The SPPEDTRONIC MARK VI turbine control is the
0.2 current state of the art control for General Electric turbine.
0 It contains a number of control protection and sequencing
systems designed for reliable and safe operation of the gas
−0.2
turbine [22].
−0.4
0 1 2 3 4 5 6 7 8 9 10 Control of the turbine is done mainly by startup, speed,
Time (s) acceleration, synchronization, and temperature control. Fig-
ure 2 illustrates the three control modes and the means of
LQR controller fuel control in relation to the fuel command signal. Sensors
LQG/LTR controller monitor the turbine speed, temperature, and compressor
Figure 6: Impulse response of the turbine temperature respect to discharge pressure to determine the operating conditions of
mass flow rate of compressor. the unit. When it is necessary for the turbine control to alter
the turbine operating conditions because of changes in load
or ambient conditions, it is accomplished by modulating
the flow of fuel to the turbine. For example, if the exhaust
temperature starts to exceed its permitted value for a given
model derived by using Jacobian method [1] at operation operating condition, the temperature control circuit will
condition for MS5002D General Electric gas turbine system. cause a reduction in the fuel supplied to the turbine and
Matlab and Control toolbox used for creating this model. thereby limit the exhaust temperature.
Validation of this model is done by measurement data with Operating conditions of the turbine are sensed and
Matlab mfile commands [21] and so at [9]. utilized as feedback signals to the SPEEDTRONIC control
The MS5002D mechanical drive gas turbine is used to system. There are three major control loops—startup, speed,
drive a centrifugal load compressor to compress treated gas and temperature, which may be in control during turbine
and deliver export gas at 90 bar pressure for export via operation. These loops command fuel stroke reference
pipeline. The gas turbine is that part of the mechanical drive (FSR), the command signal for fuel. The outputs of these
gas turbine, exclusive of control and protection devices, in control loops are connected to a minimum value select algo-
which fuel and air are processed to produce shaft horsepower. rithm as shown in Figure 2. The minimum value select algo-
Gas turbine axial air compressor has 17 stages with 10.75 : 1 rithm selects the lowest FSR called for by the control loops
pressure ratio. The output shaft speed is 4670 rpm with and passes the result to the FSR controller. The action of
ISRN Electronics 7
SV diagram of main system SV diagram of return ratio at input for LQG/LTR controlled system
150 100
100 0
50 −100
0
Singular values (dB)
−200
−250 −600
−300 −700
−350 −800
10−2 100 102 104 106 108 1010 10−2 100 102 104 106 108 1010
Frequency (rad/s) Frequency (rad/s)
(a) (b)
Figure 7: Singular value diagram of main system (a) and return ratio at input for LQG/LTR controlled system.
⎡ ⎤
1.1481 −2055 1.3219 −3967.8 0.0016442 −1.7117 −1.3466e-007
⎢ −1.1648 2057.9 −1.3435 4025.5 −0.0020121 2.0259 1.578e-007 ⎥
⎢ ⎥
⎢
Kc = ⎢ −1.1653 2076.7 −1.3454 4041.4 0.074836 −45.208 −1.4127e-007⎥ ⎥ ∗ 1e−6 , (19)
⎢ ⎥
⎣0.00087393 −19.741 0.0011058 −21.969 −0.12122 60.776 −2.0258e-007⎦
A B C D E F G
where A: 1.463e − 010, B: 7.8539e − 007, C: −8.3254e − Figure 6 shows impulse response of the turbine temperature
010, D: 8.4805e − 006, E: 4.6762e − 008, F: −1.3999e − 00 = respect to mass flow rate of compressor.
and G: 4.3014e − 013. The structure singular value can be used to analyze the
Analysis the gas turbine dynamic in time and frequency ability of a given design on achieves robust performance
domain by using LQG/LTR control method and with com- [1, 3]. Singular value or principal gain plot for main system
pare to PID controller is flowed. Step response, impulse and controlled system by LQG/LTR are shown in Figure 7.
response and principal gain consider for simulation. Figure 3
shows the unit step response of the compressor pressure, tur-
bine temperature and rotational speed respect to compressor
8. Conclusion
mass flow rate for LQR, LQG and LQG/LTR controller design This paper introduces the gas turbine systems. Control and
of proposed gas turbine. So that to compare the results, unit satisfactory performances are important concept for this
step response of these parameters for LQG/LTR and PID system. By using thermodynamic equations, a nonlinear dy-
control is shown in Figure 4. namic model is specified to study control objects. A gas
Simulation results for rotational speed and turbine tem- turbine type is considered and its control system is described.
perature by using impulse function as input to power depict Analyses the system is followed by making LTI state space
in Figure 5 for LQR and LQG/LTR controller design. Also model obtained of linearization nonlinear dynamic model
8 ISRN Electronics
and apply suggested LQG/LTR optimal control method. By [16] H. Cohen, G. F. C. Rogers, and H. I. H. Saravanamuttoo, Gas
LQG/LTR controller, and set parameters of this control by Turbine Theory, Longman, Scientific and Technical, 1996.
specified procedure, simulation results in time and frequency [17] A. M. Y. Razak, Industrial Gas Turbine: Performance and
domain are achieved. To compare results, we use PID control Operability, Woodhead Publishing limited and CRC press
method for described system model. LLC, 2007.
[18] M. P. Boyce, Gas Turbine Engineering Handbook, Gulf Profes-
System analysis shows response of uncontrolled dynam-
sional Publishing, 3rd edition, 2006.
ics and effect of LQR, LQG, and LQG/LTR controller de-
[19] T. Giampaolo, Gas Turbine Handbook: Principles and Practices,
signed and PID control. LQG/LTR controller improved The Fairmont Press, 3rd edition, 2006.
principal gain diagram to attain good performance. With [20] C. H. J. Meuleman, Measurment and Unsteady Flow Modeling
LTR method a satisfactory return ratio is recovered at the of Centrifugal Compressor Surge, University of Technology,
plant input. Eindhoven, Netherlands, 2002.
Settling times of system characteristic component parts [21] Matlab Refrence Guide, The Mathworks, 2010.
by using LQG/LTR controller are shorter, and stability of [22] GE Power System, “SPEEDTRONIC MARK VI Turbine
system makes it better by applying this optimal control Control System,” Phases 9 & 10 Refinery, 2007.
method.
References
[1] S. Skogestad and I. Postlethwaite, Multivariable Feedback
Control Analysis and Design, Wiley, New York, NY, USA, 2000.
[2] A. Maddi, A. Guessoum, and D. Berkani, “Using linear
quadratic gaussian optimal control for lateral motion of
aircraft,” Proceedings of World Academy of Science, Engineering
and Technology, vol. 37, pp. 285–289, 2009.
[3] J. M. Maciejowski, Multivariable Feedback Design, Addison-
Wesley, Boston, Mass, USA, 1989.
[4] H. L. Tsai, “Generalized linear quadratic gaussian and loop
transfer recovery design of F-16 aircraft lateral control system,”
Engineering Letters, vol. 14, p. 1, 2007.
[5] J. P. Hespanha, Linear System Theory, Princeton University
Press, 2009.
[6] G. Stein and M. Athans, “The LQG/LTR procedure for
multivariable feedback control design,” IEEE Transactions on
Automatic Control, vol. 32, no. 2, pp. 105–114, 1987.
[7] M. Athans, “A tutorial on the LQG/LTR method,” in Proced-
dings of the American Control Conference, 1986.
[8] A. J. Chipperfield, B. Bica, and P. J. Fleming, “Fuzzy schedul-
ing control of a gas turbine aero-engine: A multiobjective
approach,” IEEE Transactions on Industrial Electronics, vol. 49,
no. 3, pp. 536–548, 2002.
[9] B. Veroemen, “Model predictive control of gas turbine instal-
lation,” WFW Report, 1997.
[10] N. Chiras, C. Evans, and D. Rees, “Nonlinear gas turbine
modeling using feedforward neural networks,” in Proceedings
of the ASME TURBO Expo 2002:Controls, Diagnostics, and
Instrumentation, Cycle Innovations, Marine, Oil and Gas
Applications, pp. 145–152, nld, June 2002.
[11] O. Bolland, Thermal Power Generation, Compendium, 2008.
[12] P. Ailer, I. Sánta, G. Szederkényi, and K. M. Hangos, “Non-
linear model-building of a low-power gas turbine,” Periodica
Polytechnica Transportation Engineering, vol. 29, no. 1-2, pp.
117–135, 2001.
[13] R. T. Jagaduri and G. Radman, “Modeling and control of
distributed generation systems including PEM fuel cell and gas
turbine,” Electric Power Systems Research, vol. 77, no. 1, pp. 83–
92, 2007.
[14] N. Kakimoto and K. Baba, “Performance of gas turbine-based
plants during frequency drops,” IEEE Transactions on Power
Systems, vol. 18, no. 3, pp. 1110–1115, 2003.
[15] J. Mantzaris and C. Vournas, “Modelling and stability of
a single-shaft combined cycle power plant,” International
Journal of Thermodynamics, vol. 10, no. 2, pp. 71–78, 2007.
International Journal of
Rotating
Machinery
International Journal of
The Scientific
Engineering Distributed
Journal of
Journal of
Journal of
Control Science
and Engineering
Advances in
Civil Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014
Journal of
Journal of Electrical and Computer
Robotics
Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014
VLSI Design
Advances in
OptoElectronics
International Journal of
International Journal of
Modelling &
Simulation
Aerospace
Hindawi Publishing Corporation Volume 2014
Navigation and
Observation
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2010
Hindawi Publishing Corporation
http://www.hindawi.com
http://www.hindawi.com Volume 2014
International Journal of
International Journal of Antennas and Active and Passive Advances in
Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014