R147isbn9512284170 PDF
R147isbn9512284170 PDF
R147isbn9512284170 PDF
Espoo 2006
Report 147
TEKNILLINEN KORKEAKOULU
TEKNISKA HGSKOLAN
HELSINKI UNIVERSITY OF TECHNOLOGY
TECHNISCHE UNIVERSITT HELSINKI
UNIVERSITE DE TECHNOLOGIE DHELSINKI
Report 147
Faisal Mohamed
Abstract: A new concept in power generation is a microgrid. The Microgrid concept assumes a cluster
of loads and microsources operating as a single controllable system that provides both power and heat
to its local area. Not much is known about Microgrid behavior as a whole system. Some models exist
which describe the components of a Microgrid. This thesis aims to model Microgrids at steady state
and study their transient responses to changing inputs. Currently models of a Diesel Engine, a Fuel
Cell, a Microturbine, a Wind turbine, and finally a Photovoltaic cell have been developed. It is intended
that the work completed in this thesis will lay the groundwork for further model development. The long
term goal is to have a highly sophisticated, complete system model of a Microgrid, so as to allow its
simulation to fully understand how microgrids behave. The goal of this thesis is to build a complete
model of Microgrid including the power sources, their power electronics, and a load and mains model
in MATLAB/Simulink.
Keywords: microgrid, diesel engine, fuel cell, microturbine, wind turbine, photovoltaic, genetic
algorithms
Distribution:
Helsinki University of Technology
Control Engineering Laboratory
P.O. Box 5500
FIN-02015 HUT, Finland
Tel. +358-9-451 5201
Fax. +358-9-451 5208
E-mail: [email protected]
http://www.control.hut.fi/
ISBN-13 978-951-22-8417-7
ISBN-10 951-22-8417-0
ISSN 0356-0872
Picaset Oy
Helsinki 2006
Acknowledgements
I joined Helsinki University of Technology, Control Engineering Laboratory as postgraduate
student in July 2003.
First of all I would like to thank Allah for blessing me with ability to complete this work.
This work couldnt be complete without help and support of several people.
First my deep gratitude goes to my advisor, Professor Heikki Koivo who has provided invaluable support, guidance, patience, and encouragement over the past years. I would like to
thank all my Libyan friends here in Helsinki, and in Libya for their care and encouragement. I
wish also to thank all the my friends and colleagues in the Control Engineering Laboratory, for
creating a friendly and stimulating atmosphere.
My deep appreciation are for all my family and relatives at home, for their support and
encouragement. Also I thank my wife and my children for their patience, ultimate support,
great generosity, and lovingness.
This thesis has been supported by grant from Omar Al-Mukhtar University- El-Beida Libya.
Thanks to everyone who has contributed to this work by anyhow directly or indirectly.
Otaniemi, March 20, 2006
FAISAL MOHAMED
Contents
Abbreviations
viii
List of Figures
List of Tables
xi
1 Introduction
1.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.1
Definition of Microgrids . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.2
1.2
Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3
1.3.1
1.3.2
1.3.3
Power quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3.4
Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3.5
Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4
1.5
2.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2
2.2.1
System Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
2.2.2
Parameter estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
2.2.3
12
Controller Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13
2.3.1
PID Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13
2.3.2
Genetic Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13
2.4
15
2.5
Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16
2.3
3 Fuel Cell
3.1
19
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
3.1.1
Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
3.1.2
21
ii
3.2
3.3
3.4
4 Micro Turbines
4.1 Overview of Micro-Turbines . . .
4.2 Construction of Micro-Turbines .
4.3 Major features of Micro- Turbines
4.4 Application of Micro-Turbines . .
4.5 Micro-Turbine Modelling . . . .
4.6 Inverter Model . . . . . . . . . . .
4.7 Simulation Results . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
22
22
23
23
24
27
.
.
.
.
.
.
.
30
30
30
31
31
32
34
38
5 Wind Turbine
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Wind Turbine Generating System . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.1 Squirrel Cage Induction Generator . . . . . . . . . . . . . . . . . . . . . . .
5.2.2 Doubly Fed (Wound Rotor) Induction Generator and Direct Drive Synchronous Generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 Wind Turbine Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3.1 Rotor Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3.2 Generator Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
41
41
42
6 Photovoltaic Cell
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52
52
53
55
58
58
58
59
59
60
60
61
61
iii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
43
43
44
45
48
1. ABBREVIATIONS
AFC
CHP
CI
DG
FC
GA
IGBT
IGBT
MCFC
MG
MPPT
MT
NOCT
PAFC
PCC
PEMFC
PI
PID
PMSG
PRBS
PV
PWM
RLS
SD
SI
SOFC
UPS
2. SYMBOLS
(t)
Nir
e
V
P
0
e
m
s
w
(s)
1 max
1 min
H2
1
2
T
T (t)
Acf
Cp
e(t)
Eac
EGO
E0
Eg
F
gi
G
Hm
i
I(s)
Ir
Ior
Ios
Ish
ISCR
ID
J(s1 )
kc
K
Kan
KB
KH2
KH2 O
K0
K1
K2
K3
KI
Kr
Kv
Lm
Lr
Ls
nH 2
N0
Niin
Ni0
P
P
PH2 O
PH2
Pmin
Pmpp
Pm
Pr u
vi
q
qH2
in
qH
2
out
qH
2
r
qH
2
qH2 O
Q
r
R
Rs
Rsh
s
t
T
T (s)
TD
TI
Tm
Tr
Tt
u(t)
U
Umax
Umin
Uoc
Uopt
Uf
v
V
Vw
V
Vac
Vdc
VLL
Vmpp
Vo
Electron Charge
Molar flows of Hydrogen
Input Hydrogen Molar Flow
Output Hydrogen Molar Flow
Hydrogen Molar Flow Take Part in the Reaction
Molar flows of Water
Reactive Power
Reference Signal
Resistance
Series Resistance
Shunt Resistance
rotor slip
Time
Cell Temperature
Mechanical torque of Diesel Engine
Derivative Time
Integral Time
Mechanical Shaft Torque for the no Loss System
Reference Temperature
Load Torque
System Input
Terminal Voltage of the PV cell
Maximum Fuel Utilization
Minimum Fuel Utilization
Open Circuit Voltage
Optimal Fuel Utilization
Fuel Utilization
Voltage
Cell Volume
Wind Velocity
Output Voltage from the Control Unit
AC Voltage
DC Voltage
Voltage Induced on the Generator Terminal
Voltage at Maximum Power Point
Cell Volume
vii
W
Wan
XL
y(t)
z 1
CO2
3
e
H+
O2
CO2
H2 O
H2
O2
A
Impp
Iph
IL
Jr
Pan
Pr
Rgass
Van
OH H+
Mass Flow
Mass Flow Through the Anode Valve
Reactance
System Output
Backward Shift Operator
Carbonate Ion
Electron
Hydrogen ion
Oxygen Ion
Carbon Dioxide
Water
Hydrogen
Oxygen
Swept Area of Rotor Disc
Current at Maximum Power Point
Light-Generated Current
Load Current
Inertia of the Shaft
Pressure Inside the anode Channel
Cell Pressure
Gas Constant
Volume of the Anode
Hydroxy Ion
viii
List of Figures
1.1
MicroGrid Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1
2.2
2.3
2.4
Unit step response of the studied diesel engine under time delay 1 = 0s . . . . .
16
2.5
Unit step response of the studied diesel engine under time delay 1 = 0.125s . . .
17
2.6
17
2.7
18
18
Operation principle, cathode reactions, and the mobile ion associated with most
common fuel cell types. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
3.2
22
3.3
28
3.4
Responses output voltage, output current, real power output due to the power
demand input. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
3.5
29
3.6
29
4.1
31
4.2
33
4.3
33
4.4
35
4.5
35
4.6
37
4.7
38
4.8
39
4.9
39
2.8
3.1
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
40
40
5.1
42
5.2
42
ix
5.3
5.16
5.17
6.1
6.2
6.3
6.4
6.5
6.6
6.7
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12
5.13
5.14
5.15
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
43
45
47
48
48
49
49
50
50
50
50
51
51
51
51
53
53
55
56
56
57
57
List of Tables
2.1
10
3.1
3.2
20
21
6.1
55
xi
Chapter 1
Introduction
1.1 Introduction
Recent developments in the electric utility industry are encouraging the entry of power generation and energy storage at the distribution level. Together, they are identified as distributed
generation (DG) units. Several new technologies are being developed and marketed for distributed generation, with capacity ranges from a few kW to 100 MW. The DG includes microturbines, fuel cells, photovoltaic systems, wind energy systems, diesel engines, and gas
turbines [1],[2] .
CHAPTER 1. INTRODUCTION
They also enable a more efficient use of waste heat in combined heat and power (CHP) applications, which boosts efficiency and lowers emissions. The CHP systems provide electricity,
hot water, heat for industrial processes, space heating and cooling, refrigeration, and humidity
control to improve indoor air quality and comfort.
Figure 1.1 shows the MG architecture. This consists of a group of feeders. There is a single
point of connection to the main distribution utility called point of common coupling (PCC).
SD means a Separation Device that can disconnect MG immediately when a fault occurs in the
distribution grid. Some feeders, (feeders 1,2) have sensitive loads, which require local generation. The traditional loads are connected to Feeder 3 and do not have any local generation.
Each of the local generation has a LC ( local Controller). This is responsible for local control
that corresponds to a conventional controller (ex. AVR or Governor) but that has a network
communication function to exchange information between other LCs and the upper central
controller to achieve an advanced control, The central controller also plays an important role as
a central load dispatch control center in bulk power systems, which is in charge of distributed
generator operations installed in MG [5].
MG technologies are playing an increasingly important role in the nations energy portfolio. They can be used to meet baseload power, peaking power, backup power, remote power,
power quality, and cooling and heating needs. Customers usually own small-scale, on-site
power generators, but they may be owned and operated by a third party. If the distributed
generator does not provide 100% of the customers energy needs at all times, it can be used
in conjunction with a distributed energy storage device or a connection with the local grid for
backup power. The MG resources support and strengthen the central-station model of electricity generation, transmission, and distribution. The diagram below shows how the grid looks
after the addition of distributed resources. Although the central generating plant continues to
provide most of the power to the grid, the distributed resources meet the peak demands of local
distribution feeder lines or major customers. Computerized control systems, typically operating over telephone lines, make it possible to operate the distributed generators as dispatchable
resources that generate electricity as needed.
1.1. INTRODUCTION
MicroGrid Central
Controller
PCC
SD
Feeder 3
Feeder 2
Feeder 1
Load
LC
LC
Load
Wind
Turbine
Load
Diesel
Engine
Load
LC
LC
Fuel
Cell
Load
Load
Load
PV
array
Heat
load
LC
Load
Micro
Turbine
CHAPTER 1. INTRODUCTION
1.2 Motivation
Currently a lot of research is being undertaken into MGs and some model architectures have
been proposed in the literature such as [3],[4],[6],[7],[8], [5] . Although the components of a
MG are fairly well understood, the system as a whole is not. When several sources are connected to form a MG, the system behaviour is unpredictable. This being the case, modelling
the system and simulating it in order to develop an appropriate control system, is the heart of
micro-grid research. Nowadays several research groups around the world are investigating the
feasibility and benefits that the MGs may provide. Some problems are encountered including
dealing with unbalanced loads and harmonics associated with the system. This work does not
intend to address such problems, rather it is concerned with the modelling of the MG for the
investigation of the transient and steady-state response.
1.3.4 Protection
A number of different aspects of MG protection can be identified [1]:
Protection of the generation equipment from internal faults.
Protection of the faulted distribution network from fault currents supplied by the MGs.
Anti-islanding or loss-of-mains protection.
Impact of MGs on existing distribution system protection.
1.3.5 Stability
For Distributed Generators schemes, the objective of which is to generate power from new renewable energy sources, considerations of generator transient stability tend not to be of great
significance. If a fault occurs somewhere in the distribution network to depress the network
voltage and the Distributed Generator trips, then all that is lost is a short period of generation.
The MGs will tend to overspeed and trip on their internal protection. The control scheme in
the MGs will then wait for the network condition to be restored and restart automatically. In
contrast, if a DG is viewed as providing support for the power system, then its transient stability becomes of considerable importance. Both voltage and/or angle stability may be significant
depending on the circumstances.
CHAPTER 1. INTRODUCTION
Chapter 2
K3 K2
I(s)
(1 + 2 s)
(2.1)
Fuel Flow (s) is then converted into mechanical torque T (s) after a time delay 1 and engine torque constant K1 , which can be represented by the model as shown in Figure 2.2, and
equation ( 2.2).
T (s) = (s)K1 e1 s
(2.2)
Figure 2.1: The Actuator Model and the current driver constant.
10
The values of K3 and K2 can be considered to be constant for a particular engine setup. Gain
K3 is a factor that determines the amount of the mechanical torque obtained per unit of fuel
flow. It depends on the operating point of the prime mover. Time constant 2 is quite dependent on the temperature of the oil flowing into the actuator. Both K2 and 2 are variable, but
the variation is negligible in a small time interval, [15].
As seen from Figure 2.2 and from literature such as [15], [17], [18] and [19], the engine combustion system is commonly represented by the engine torque constant multiplied by a time
delay . In a real system, the dead time is mainly comprised of three components:
1. The time from the actuator signal change until fuel is injected to any cylinder. This is
called "power stroke delay".
2. The time for the fuel to burn in a cylinder and to produce a torque output, which is similar
to the characteristic "combustion delay".
3. The time for a new torque level to produce a sufficient number of cylinders to assignable
to the prime-mover as a whole. This is an effect of the multi cylinder nature of the primemover.
The complete parameter range values for the diesel engine system are given in [15] and
restated in Table 2.1.
Table 2.1: System Parameters of a Typical Diesel Engine .
System parameters
Actuator gain constant K2 (per unit)
Actuator time constant 2 (s)
Engine torque constantK1 (per unit)
Current driver constant K3 (per unit)
Engine dead time 1 (s)
Plant and flywheel acceleration J(s1 )
Friction coefficient (per unit)
Value range
1.0
0.05 0.2
0.8 1.5
1.0
0 1.0
0.1 0.5
0.1
Nominal values
1.0
0.125
1.15
1.0
0.5
0.3
0.1
The speed of the power generation plants driven by diesel prime movers is difficult to control
optimally using a PID controller because the system parameters are changing.
T (s)
K0
=
e1 s
I(s)
s(1 + 2 s)
(2.3)
where the actuator time constant is 2 , engine time delay 1 , and the process gain K0 = K1 K2 K3 .
The parameters are usually unknown or time varying.
First order Pade approximation for the time delay term yields:
e1 s
1 1 /2
1 + 1 /2
(2.4)
11
The following simplified transfer function can now be obtained from ( 2.3):
G0 (s) =
K0
1 1 /2
s(1 + 2 s) 1 + 1 /2
(2.5)
The discrete time model corresponding to equation ( 2.5) can be expressed as:
A(z 1 )y(t) = B(z 1 )u(t 1) + (t)
(2.6)
where u(t) is the system input (current), y(t) is the system output (torque), and (t) is an uncorrelated zero main random sequence. Time t takes usually discrete values t = 0, 1, 2, .......
In ( 2.6) A and B polynomials are
A(z 1 ) = 1 + a1 z 1 + a2 z 2 + ....... + an1 z n
B(z 1 ) = b0 + b1 z 1 + b2 z 2 + ...... + bm z m (b0 6= 0)
)
(2.7)
where z 1 is the backward shift operator. The order of the polynomial A(z 1 ) is n and that
of B(z 1 ) m,.
(t)
(2.8)
P (t 1)(t)
+ T P (t 1)(t)
(2.9)
error covariance
P (t) = M (t)
M (t) diag
1
T
(P (t 1) K(t) (t)P (t 1)) M (t)
1
mi (t)
s
mi (t)
(2.10)
(i = 1, ......, N , N = n + r + 1, 0 < mi 1)
1))2 gi
(y(t) T (t)(t
(i = 1, ......, N )
1 + T (t)P (t 1)(t)
(2.11)
(2.12)
and
T = [a1 ......, an , b00 , ........, b0r ]
(2.13)
T (t) = [y(t 1), ......, y(t n), u(t min ), ......, u(t m min )]
(2.14)
12
b00 b0r , including the dead time of the plant, is more sensitive than that of a1 an so that gi0 s
are set up in the following manner:
g1 gn < gn+1 gn+r+1
(2.15)
Moreover, when the variation of the estimation parameters is too high at the point of initial
control step or when the dynamic characteristics of the system are changed; the control input
assumes an excessively large value. Then, in order to reconstitute the controller, we employ
a renewal condition to apply the estimated parameters. Concretely, when the variation of the
estimated parameters is more than the previous one by 50% or B 0 (1) of equation ( 2.5) has a
very small value, the estimated parameters are stopped being updated.
Since the diesel engine system that is to be estimated is part of the control loop, an additional
problem is introduced. The feedback signal may cause an interruption in the relationship between the input and the output signal causing the RLS estimator to malfunction. The most
effective way to tackle this problem is to add an independent signal into the loop. This extra
dither signal should satisfy the requirement for persistent excitation and should not have large
amplitude components that could drive the system into non-linear operation [21]. We have
used Pseudo Random Binary Signal (PRBS) because it only has two amplitude levels that may
be selected so that the amplitude does not exceed the bands of linearity of the system. The
main advantage of the PRBS is that it has a rich spectrum and is an ideal input for the system
to be estimated.
(2.16)
where r = m + 1 max 1 min , 1 max and1 min are the lower and upper limit values of the
dead time, respectively and 1 is the dead time of the system . When the estimated parameters
b00 b0r of equation ( 2.5) have converged, the obtained data are expected to have the following
character:
b0i = (i 1 min 1)
b0j = (j 1 min + m + 1)
(2.17)
where is a very small value. Therefore, it is possible to estimate the dead time of the plant
using this character without complex calculations.
13
kc Ts
TD 2
e(t) kc ( +
)y(t), t = 0.1, 2, ....
TI
Ts
(2.18)
where u(t) is the control variable, e(t) the control error signal defined as e(t) = yref (t) y(t)
,yref (t) is the reference value, and y(t) is the output of the system. denotes the differential
operator defined by := 1 z 1 . Parameters kc ,TI and TD are the proportional gain, integral
and the derivative time, respectively.
14
15
Mutation
In natural evolution, mutation is a random process that causes one part of a gene to change
which results in an entirely new genetic structure. Mutation in genetic algorithms, like its counterpart in nature, occurs with a low probability - typically in the range from 0.001% to 0.01%.
The role of mutation is thought to have two main functions; firstly, it provides a guarantee that
the probability of searching any given string will never be zero and secondly to act as a safety
net to recover good genetic material lost through the operations of selection and crossover. The
Multi non-uniformly distributed mutation operator, was chosen as the mutation operator as it
is considered to function well with multiple variables. The mutation operator has three options
when the multi non-uniformly distributed mutation function is used. The first is the total number of mutations, normally set with a probability of around 0.1%. The second parameter is the
maximum number of generations and the third parameter is the shape of the distribution. This
last parameter is set to a value of two, three, or four, where the number reflects the variance
of the distribution. The simulation and the genetic algorithm routine were implemented using
The MATLAB/ Simulink environment, with the GA Toolbox [28].
The genetic algorithm was initialized with a population of 20 and iterated for 100 generations. The bounds of the genetic algorithm reduced to a smaller range of numbers. This
ensured quick convergence and a reduction in the number of generations necessary for the algorithm to converge to optimum PID gain values, which in turn will reduce the overall runtime
of the genetic algorithm
The genetic algorithm code simulated the Simulink file, which includes the plant, and the
RLS estimator (S-Function). The genetic algorithm takes in the parameter estimates and then
performs the genetic algorithm search. The genetic algorithm then updates the PID controller
and reads in the parameter estimates.
To ensure that the GA tuned controller produces a stable overall system, additional code
was added that analyzed the stability of the controlled system. The stability of the system was
examined by accessing the magnitude of the real and imaginary parts of the discrete poles.
16
1.4
1.2
Speed [p.u]
0.8
0.6
0.4
0.2
10
20
30
40
50
Time [s]
Figure 2.4: Unit step response of the studied diesel engine under time delay 1 = 0s
17
1.4
1.2
Speed [p.u]
0.8
0.6
0.4
0.2
10
20
30
40
50
Time [s]
Figure 2.5: Unit step response of the studied diesel engine under time delay 1 = 0.125s
Figures 2.4, 2.5, and 2.6 show the unit step responses of the diesel system under the nominal
system parameters.The time delay was chosen to be 1 = 0s , 0.125s , and 0.25s respectively.
There was clear improvement in the unit step response compared to the works [15], [17], and
[19] in the overshoot and settling time at the start up due to fast computing.
1.4
1.2
Speed [p.u]
0.8
0.6
0.4
0.2
10
20
30
40
50
Time [s]
Figure 2.6: Dynamic response of the diesel engine under time delay 1 = 0.25s
The overshoot decreased and the settling time was also about less than 5 s for all the variations of time delay 1 . There was slight improvement in the load handling of the proposed controller. The overshoot and settling times were very satisfactory, and the controller performed
18
1.4
1.2
Speed [p.u]
0.8
0.6
0.4
0.2
10
20
30
40
50
Time [s]
Figure 2.7: Response under the minimum set of parameters,2 = 0.05s, K2 = 0.8p.u, J = 0.1s1 , 1 = 0.125s
1.4
1.2
Speed [p.u]
0.8
0.6
0.4
0.2
10
20
30
40
50
Time [s]
Figure 2.8: Response under the maximum set of parameters 2 = 0.2s, K1 = 1p.u, J = 0.3s1 , 1 = 0.125s
in the same good way for all the variations of the time delay. When the dead time reached the
upper bound,1 = 0.25s, in Figure 2.6, the controller produced very small overshoot and the
settling time was less than 5 s.
Additionally in Figures 2.7, 2.8, there were improvements related to overshoot and settling
time for the start up and when the load for both lower bounds and upper bounds were applied.
Chapter 3
Fuel Cell
3.1 Introduction
Fuel cells generate power through the electrochemical reaction between hydrogen and oxygen. The conversion is highly efficient and leaves only water and heat as by-products, which
is the main motivation for the increasing interest in the technology [29]. Fuel Cells offer lower
emission and higher efficiency than Diesel Engine but are likely to be expensive for many applications. The first fuel cell unit was discovered and developed by Sir William Grove 1842 [30]
with the use of four primitive cells utilizing hydrogen and oxygen. However, it was not practically used until the 1960s when NASA demonstrated a potential fuel cell application. After
such demonstrations, commercial companies became interested in this technology because of
its power quality, high efficiency, modularity, and environmental benefits.
Fuel cells could potentially replace the internal combustion engine and many other energy
generation devices used today. Reduced emissions of greenhouse gases and increased efficiency are two of the major reasons that fuel cells are being seriously researched as a replacement to the internal combustion engine. This work presents an overview of fuel cell systems,
their applications, their benefits and challenges, an explanation of the need for a transient analysis of fuel cell systems, and an overview of the relevant background literature.
A fuel cell (FC) is an electrochemical energy conversion system, where chemical energy is
converted directly into electrical energy and heat. Resulting advantages of this technology are
high efficiency almost at partial load, low emissions, and noiselessness (due to nonexistence of
moving parts), and free adjustable ratio (50 kW to 3 MW) of electric and heat generation. The
basic structure of fuel cells consists of a pair of electrodes and an electrolyte. The fuel which is
usually hydrogen, is supplied to the anode where the fuel then is oxidized, yielding electrons,
which move through the external circuit. At the cathode, the oxidant is reduced, consuming
electrons from the external circuit. Ions move through the electrolyte to balance the flow of
electrons through the external circuit. The anode-cathode reactions and the composition and
direction of flow of the mobile ion vary with the type of fuel cell.
Figure 3.1 explains the operation principle, cathode reactions, and the mobile ion associated
with most common fuel cell types.
The reactions of Alkaline Fuel Cell (AFC), Proton Exchange Membrane Fuel Cell (PEMFC),
Phosphoric Acid Fuel Cell (PAFC), Molten Carbonate Fuel Cell (MCFC) and Solid Oxide Fuel
Cell (SOFC) are summarized in Table 3.1.
19
20
Electrical load
e
Fuel in
Oxident in
+Ions
Porous anode
-Ions
Figure 3.1: Operation principle, cathode reactions, and the mobile ion associated with most common fuel cell types.
Anode reaction
H2 2H+ + 2e
H2 2H+ + 2e
H2 O + CO2
3 H2 O + CO2 + 2e
2
H2 + O H2 O + 2e
Mobile
Ion
H+
OH
CO2
3
O2
Cathode reaction
1/2O2 + 2H+ + 2e H2 O
1/2O2 + H2 O + 2e 2OH
1/2O2 + CO2 + 2e CO2
3
1/2O2 + 2e O2
e : - electron
H2 O:- water
CO2
3 : - carbonate ion
H+ : - hydrogen ion
O2 - oxygen
O2 : - oxygen ion
H2 : - hydrogen
3.1.1 Characteristics
There are four major fuel cell technologies with somewhat different characteristics. The main
apparent difference is the electrolyte, which also has far reaching effects on the design and
operating characteristics of the fuel cell. In Table 3.2. those four technologies are listed with
some key characteristics.
3.1. INTRODUCTION
21
Electrolyte
Operating
temperature (o C)
Electric efficiency based
on natural gas*(%)
PEMFC (PEFC)
Protone Exchange
Membrane
80
PAFC
Phosphoric
Acid
200
MCFC
Molten
Carbonate
650
SOFC
Solid
Oxide
800-1000
30-35
35-40
45-55
45-55
*With hydrogen as fuel the electric efficiency is the same or even higher for low temperature fuel cells, as this is
not a Carnot process. The reason for the higher efficiency with higher temperature for natural gas (or any reformed
fuel) lies primarily in that fuel processing can be thermally integrated with the fuel cell and to a lesser extent to
lower internal electric resistances.
All fuel cells generate a direct current, the voltage depending on cell voltage and the number
of cells in series. Furthermore, the voltage varies with the load and also to some extent with
time as the fuel cell stack ages. To obtain AC current, the equipment should have powerconditioning equipment to handle DC to AC conversion and current, voltage, and frequency
control. Apart from supplying power to the external point of supply, the fuel cell also has to
cover some internal power needs, e.g. pumps, fans, and control system.
As fuel cells are in a development stage, it is difficult to make general statements about
operating characteristics as, for example, operating procedures tend to be on the cautionary
side. Tentatively the following characteristics can be listed:
1. Start-up time depends on type of fuel cell and type of fuel processing system. A low temperature fuel cell (PEMFC) with partial oxidation could probably be started in a couple of
minutes while a high temperature fuel cell takes 3 to 4 hours due to the need of avoiding
thermal stresses during warm up. Generally speaking, high temperature fuel cells are not
suited for start-stop operation.
2. A fuel cell in itself can facilitate nearly instant load changes. However, a fuel cell system
has a limiting factor in the fuel processing system which has a certain time lag (varying
depending on the type) and a truly load following system would need a buffer, e.g batteries or hydrogen storage capacity. A typical turndown ratio of a fuel system is about 1:5
and high efficiencies are kept to at least 50% load.
3. Fuel cells have a potential for high reliability as the number of moving parts is low. It
consists of auxiliary equipment such as fans and pumps. The target for life length of fuel
cells is usually given as 40000 h for the stack and at least twice the number of hours for
the system. This target has been reached for a small number of fuel cells but in general it
still remains to be proven.
22
to the electrical energy. Some other benefits are modularity, simplicity of installation, silent
operation, and suitability for the integration with renewable energy sources.
3.2
The workings of the fuel cells are based on fundamental electrochemical principles [29]. The
reaction of hydrogen gas (H2 ) and oxygen gas (O2 ), to form water, is the form of reaction in the
process as presented in [30]:
2H2 + O2 2H2 O
(3.1)
The fuel cells have an electrolyte between two electrodes. The process occurs naturally and
is caused by the fact that charged particles migrate towards regions of lower electrochemical
energy.
The charged particles of hydrogen and oxygen migrate towards each other and connect together since the final products have lower electrochemical energy [30]. It is essential to separate
electrons from protons and to regulate the movement of the electrons. This can be accomplished by separating the hydrogen and oxygen by an electrolyte, which completely insulates
electrons and allows protons from the hydrogen atoms to move through. An external path is
formed from electrons using an electrical load to generate useful electrical energy [30] as shown
in Figure 3.2.
Electron
Electrical
Load
H2
Heat
Proton
HO
2
Electrolyte
O2
23
Pr V o
)x = Niin Ni0 + Nir
Rgass T
(3.2)
where V o is the cell volume, Niin ,Ni0 are the flow rates (mole/s) of the ith reactant at the
cell input and output (exit), respectively, Nir is the reaction rate (mole/s) of the ith reactant. Pr
is the cell pressure (atm), T is the cell temperature in o K , and Rgass is the gas constant (8.31
J/mole o K).
W
=K M
(3.3)
Pr u
24
where, W
h is the mass
i flow [kg/s], K is the valve constant, mainly depending on the area of
kmol.Kg
the orifice
, and Pr u is the pressure upstream (inside the channel) [atm].
atm.s
For the anode, the concept of fuel utilization Uf can be introduced, as the ratio between the
fuel flow that reacts and the fuel flow injected to the stack. Uf is also a way to express the water
molar fraction at the exhaust. According to this definition, equation ( 3.3) can be written as:
q
Wan
= Kan (1 Uf )MH2 + Uf MH2 O
Pan
(3.4)
where
h Wan is ithe mass flow through the anode valve [kg/s], Kan is the anode valve conkmol.Kg
stant,
, MH2 , MH2 O are the molecular masses of hydrogen and water, respectively
atm.s
[kg/kmol], and Pan is the pressure inside the anode channel [atm].
Assuming that the molar flow of any gas through the valve is proportional to its partial
pressure inside the channel, then according to [33]:
qH2
Kan
= KH2
=p
PH2
MH2
(3.5)
Kan
qH2 O
=p
= KH2 O
PH2 O
MH2 O
(3.6)
where
qH2 O ,qH2 are the molar flows of water and hydrogen, respectively, through the anode valve
[kmol/s]. PH2 O ,PH2 are the partial pressures of water and hydrogen, respectively [atm]. KH2 O ,
kmol
KH2 are the valve molar constants for water and hydrogen, respectively [ s.atm
].
[33] introduces the following expression:
h
i
p
p
W
= Kan . (1 Uf ) MH2 + Uf MH2 O
Pan
(3.7)
Comparison of ( 3.4) and ( 3.7) indicates that for Uf > 70% the error is less than 7%. It
is possible to redefine slightly ( 3.5) and ( 3.6) so that it is even lower. The same study for
the cathode shows that the error in that valve is even lower, because of the similar molecular
masses of oxygen and nitrogen.
(3.8)
where Van is the volume of the anode,nH2 is the number of hydrogen moles in the anode
l.atm
], T is the absolute temperature [K].
channel, R is the universal gas constant [ kmol.K
By isolating the pressure and taking the derivative of the previous equation, we can write:
Rgass T
d
PH2 =
qH2
dt
Van
(3.9)
25
where, qH2 is the time derivative of nH2 , and represents the hydrogen molar flow qH2 [kmol/s].
in ,
There are three relevant contributions to the hydrogen molar flow qH2 : the input flow qH
2
r and the output flow q out [33],[34]. Therefore ( 3.9)
the flow that takes part in the reaction qH
H
2
2
can be expressed as:
Rgass T in
d
out
r
PH2 =
(qH2 qH
qH
)
(3.10)
2
2
dt
Van
r can be calculated according to the basic electroThe molar flow of hydrogen that reacts qH
2
chemical relationship as:
r
qH
=
2
N0 I
= 2Kr I r
2F
(3.11)
where N0 is the number of cells associated in series in the stack, F is the Faradays constant
[C/kmol], I r is the stack current [A], and Kr is a constant defined for modelling purposes
[ kmol
s.A ].
Substituting equation ( 3.11) into ( 3.10):
Rgass T in
d
out
PH2 =
(qH2 qH
2Kr I r )
2
dt
Van
(3.12)
Substituting the output flow of equation ( 3.5) into ( 3.12), taking Laplace transform of both
sides, and solving for the hydrogen partial pressure results in:
sPH2 =
Rgass T in
(qH2 KH2 PH2 2Kr I r )
Van
(3.13)
1/KH2 in
(q 2Kr I r )
1 + sH2 H2
(3.14)
where H2 , expressed in seconds, is the time constant of the system associated with the hydrogen flow. It is a function of temperature and has the form:
H2 =
Van
KH2 Rgass T
(3.15)
For calculating the stack voltage we apply Nernsts equation and Ohms law (to consider
ohmic losses). The stack output voltage V r can be represented by the following expression:
V r = N0
"
#!
PH2 PO0.5
Rgass T
2
ln
rI r
E0 +
2F
PH2 O
(3.16)
where
E0 is the voltage associated with the reaction free energy [V]. R is the same gas constant as
J
previously, but care should be taken with the system unit [ kmol.K
] [33].
r describes the ohmic losses of the stack [].
The above equations provided by [33] from the basic SOFC power section dynamic model
used for performance analysis during normal operation. In [35] the SOFC power generation
was modelled by adding control strategy of the fuel cell system, models of fuel processor, and
the power section:
26
(1) Although CO can be a fuel in SOFC, the CO-shift reaction is chemically favored with
present designs and operations if the fuel gas contains water. The CO-shift reaction is:
CO + H2 O CO2 + H2
(3.17)
Based on this, it is assumed that only H2 and O2 enter into the fuel cells.
(2) Fuel utilization is the ratio between the fuel flow that reacts and the input fuel flow.
Hence, we have
r
Uf = qH2 qin
H2
(3.18)
As in equation ( 3.11) an 80 90% fuel utilization is used and the demand current can be
restricted to the range:
in
in
0.9qH
0.8qH
2
2
I in
(3.19)
2Kr
2Kr
(3) The real output current in the fuel cell system can be measured, so the input fuel flow can
be controlled to control Uf to be 85%, so:
in
qH
=
2
2Kr I r
0.85
(3.20)
(4) From the overall fuel cell reaction ( 3.1), the stoichiometric ratio of hydrogen to oxygen is
2 to 1. Oxygen excess is always taken in to let hydrogen react with oxygen more completely.
Their simulation in fuel cell system shows that rHO should be kept around 1.145 in order to
keep the fuel cell pressure difference below 4 kPa under normal operation. So the input oxygen
flow is controlled to keep rHO at 1.145 by speed control of the air compressor.
(5) The peak power capacity is the ratio of maximum theoretical power delivery to the rated
power in the fuel cell system. It is only determined with the available active fuel cell area.
For the highest possible total efficiency and the dynamic load-following behavior, pk should
be as large as possible. As this value is directly proportional to the effective fuel cell area for
a constant output, cost considerations restrict the upper value to be between 130 and 180% .
In practice, this upper value is also restricted by the safety of system operation. In order to
prevent damage to the electrolyte, the fuel cell pressure difference between the hydrogen and
oxygen passing through the anode and cathode gas compartments should be below 4 kPa under normal operation and 8 kPa under transient conditions. Because different fuel cell systems
have different peak power capacity, simulation shows that pk in the fuel cell system model
should be below 170%, which means the maximum power delivery of the fuel cell system is
below 1.7 times of the rated power.
(6)The chemical reaction is modeled as a first-order transfer function with a 5 s time constant
because of the fuel processor is usually slow as it is associated with the time to change the
chemical reaction parameters after a change in flow reactions.
(7) The electrical response time in the fuel cells is generally fast and mainly associated with
the speed at which the chemical reaction is capable of restoring the charge that has been drained
by the load. This is also modeled as a first-order transfer function but with a 0.8 s time constant.
(8) The fuel cell system can output not only real power but also reactive power. This is done
at the power conditioner (it converts DC power to AC power and includes current, voltage
27
and frequency control). Usually, power factor (P F ) can be in the range of 0.8 to 1. Because the
response time of the power conditioner is less than 10 ms, it is concluded that it is not necessary
to include its detailed model in the slow dynamic fuel cell system. It is assumed that P F can
be adjusted accordingly by the power conditioner.
Based on [33] and the above discussions, the SOFC system dynamic model which is proposed
by [35] is summarized in ( 3.21)-( 3.25). The block diagram of the system is given in Figure 3.3.
dI r
1
r +I
ref ]
dt = Te [I
h
i
in
dqH
1
in + 2Kr I r
2
q
=
dt
Tf h H2
Uopt
i
in
dPH2
1
1
r
=
2K
I
P
q
+
r
H
2
H
dt
H2
KH2
2
h
i
dPH2 O
2Kr r
1
=
P
+
I
H
O
2
dt
H2hO
K
h H2 O
ii
dPO2
1
1
1
in K I r
=
P
+
q
r
O2
dt
O2
KO2 rHO H2
Iref
in Umax
if
qH2 2Kr ,
Umin
in
=
qH2 2Kr ,
if
in
I = Pref /V ,
in Umax
I > qH
2 2Kr
in
min
1
I =
(Pref P )
Vin
(3.22)
(3.23)
V = N0
"
#!
PH2 PO0.5
Rgass T
2
E0 +
ln
rI r
2F
PH2 O
(3.21)
(3.24)
The active (DC) power produced by the fuel cell is then given by the following relation:
Pe = V r I r
(3.25)
where I r is the fuel cell stack current; V r is the DC voltage across the stack of the fuel cells
in stands for the hydrogen input flow; and P , P
is governed by the Nernst equation, qH
H2
O2
2
, PH2 O denote the partial pressures of hydrogen, oxygen, and water, respectively. The time
constants Te , Tf , H2 , H2 O , O2 , designate the electrical response time of the fuel cell, fuel
processor response time, response times of hydrogen, water, and oxygen flows, respectively.
KH2 , KH2 O , and KO2 , denote the valve molar constants for hydrogen, water, and oxygen. The
auxiliary constants U opt, Umax , and Umin stand for the optimal, maximum, and minimum fuel
utilization, respectively. Finally, Kr = N0 /(4F ). The numerical values of the aforementioned
constants can be found in [33] and [35].
28
increase of the demand power is related to the fast electrical response of the fuel cell. After that,
the output power started to increase slowly until it reached the demand power. This is due to
the slow chemical response time of the fuel processor.
1.05
1
Power[p.u]
0.95
0.9
0.85
0.8
0.75
Power Demand
Output Voltage
Output current
Real Power Output
0.7
0.65
10
20
30
40
50
60
Time [s]
Figure 3.4: Responses output voltage, output current, real power output due to the power demand input.
Figure 3.5 illustrates the response of the fuel cell pressure difference between hydrogen and
oxygen. We can notice that it increases to the peak value of 3.5 kPa, which is less than the
maximum safety pressure difference 8 kPa. It can return to the normal operating pressure
difference value around 0 kPa.
29
3.5
3
Difference pressure[kPa]
2.5
2
1.5
1
0.5
0
0.5
20
40
60
80
100
Time [s]
120
140
160
180
In Figure 3.6 the fuel utilization response is presented, due to increase in the power demand,
the fuel utilization increases to the maximum fuel utilization Umax in about 5 s. After staying
at Umax for about 25 s, it decreases to optimal fuel utilization Uopt .
0.9
Fuel Utilization
0.89
0.88
0.87
0.86
0.85
0.84
10
20
30
40
50
60
Time [s]
Figure 3.6: Response of Fuel Utilization.
70
80
Chapter 4
Micro Turbines
4.1 Overview of Micro-Turbines
Micro turbines (MTs) are small high-speed gas turbines powered generators ranging in size
from 25 to 500kW [36]. The operation principle of the MTs follows the same principles of
conventional gas turbine depending on Brayton (constant pressure) cycle [36], [37]. Small gas
turbine engines were initially developed by Alison in the 1960s for ground transportation [37].
The micro-turbine provides input mechanical energy for the MT generator system, which is
converted by the generator to electrical energy. The generator nominal frequency is in the
range of 1.4-4 kHz. This frequency is transformed to the desired power frequency of 50/60
Hz by a converter. The electrical energy, passing through the transformer, is delivered to the
distribution system and the local load. The transformer boosts the converter output voltage up
to the voltage level of the distribution system. The components of the MT generator system are
described in detail in the subsections following [38].
31
32
Because large steam turbines and large synchronous generators have dominated power generation, the steady state and dynamic behaviour for these systems is well understood. The
basic operation and control principles are summarized below [39]:
At steady state, the power of the steam rate into the turbine is equal to the electrical power
removed from the generator. The speed of the generator and turbine is considered to be
synchronous, implying that output electrical sinusoids are in phase with the grid. This
operation requires good speed control of the turbine.
During a load transient, the change in power is taken from the speed of the rotor of the
large turbine and generator. Because these devices are enormous, there is considerable
stored energy in the rotating masses. The speed control of the turbines sees this speed
change and corrects the rate at which the steam is supplied to the turbine, correcting the
speed until the set point is achieved. In this manner, the turbine generator set is capable
of nearly instantaneous load tracking.
The same base of knowledge is not available for micro-turbines and generators. However,
the same basic principles apply and are summarized below.
At steady state, the power of the natural gas combustion and air into the turbine is equal
to the electrical power removed from the generator. The speed of the generator and turbine is not critical, as the output sinusoids from the generator are rectified. The dc link
voltage needs to be supported to ensure that conservation of power requirements are met.
This operation requires good speed control of the turbine.
During a load transient, the change in power is taken from the speed of the rotor of the
micro-turbine. However, because these devices are small, there is very little stored energy
in the rotating masses and the speed of the rotor changes very quickly. The speed control
of the micro-turbines sees this speed change and corrects the rate at which the fuel is
supplied to the micro-turbine, correcting the speed until the set point is achieved. The
speed of micro-turbine needs to be changed quickly to ensure that the generator does not
stall. In this manner, the turbine generator set is capable of load tracking.
33
3-phase AC
Rectifier
Inverter
DC
PMG
Fuel
Control
V dc
V ac
I ac
Control
No load case is considered (Ideal). The voltage induced on the generator terminal VLL can
be expressed as:
VLL = Kv sin(t)
(4.1)
VLL = Kv Im ejt
(4.2)
where Kv is the voltage constant and is the electrical angular frequency. The output DC
voltage is given by:
Vdc =
3
3L
|VLL |
Idc
(4.3)
3
3
Kv L.Idc
(4.4)
34
(4.5)
(4.6)
Eg = Vdc + Kx ..Idc
(4.7)
Eg = Ke =
v
where Ke = 3K
{V /(rad/ sec)}
Then ( 4.5) can be expressed as:
where Kx = 3L
{/(rad/ sec)}
Equation ( 4.5) describes the electromechanical nature of the system. Therefore if the system
has no losses, the input power Pm can be expressed as a function of Idc :
Pm = Vdc Idc
(4.8)
(4.9)
The mechanical shaft torque for the no loss system Tm is expressed as equation ( 4.10) :
Tm =
Pm
2
= Ke Idc Kx Idc
(4.10)
(4.11)
where Jr is the inertia of the shaft, Tm is the mechanical torque, Tt is the load torque.
Additionally, the DC voltage Vdc can also be expressed as:
Vdc
1
=
C
Z
(Idc IL )dt
(4.12)
The relations in equations ( 4.4) and ( 4.12) determine the load current IL and the final output
power. Finally from all of the above equations the block diagram of an MT generator model is
described in Figure 4.4:
35
The DC voltage is denoted by Vdc , Vac corresponds to AC voltage, V represents the output
voltage from the control unit. Coupling to the power system is done through connection reactance XL . The voltage source inverter provides control of both the magnitude and phase of its
36
output voltage, Vac . The vector relationship between the inverter voltage, Vac , and the system
voltage, Eac , along with the connection reactance, XL , determines the flow of real and reactive
power (P ,Q) from the microgrid to the system [5], [39].
As a minimum the inverters needs to control the flow of real power P , and reactive power Q
between the micro sources and the power system. The P and Q are coupled. For small changes,
P is predominantly dependent on the power angle, P , while Q depends on voltage difference.
The power flow equations are:
V Eac
sin P
XL
(4.13)
V
(V Eac cos P )
XL
(4.14)
P =
Q=
The power angle P is:
P = V e
(4.15)
where V is corresponds to inverter terminal voltage phase angel, e denotes grid voltage
phase angle.
In the range of small e , sin(e ) e holds and the relationship between P and e can be
regarded as almost linear [5]. By using these characteristics, inverters are controlled by power
width modulation (PWM) to get the required values of P and Q. However in this thesis inverters are modeled by means of ideal voltage sources, which achieve the same behavior, as
simplified models.
Figure 4.6 shows the details of a droop governor [40]. This governor has two important
characteristics. First, it allows maintaining any desired value of power when the AC grid is
connected. Second, it slowly brings up the frequency near the customary 0 value after the
droop regulation has taken place.
The constant m in Figure 4.6 denotes the frequency droop without the frequency restoration
loop active. It is dependent on the local power setpoint before islanding, and the new power
setpoint to be reached after the grid has failed. k 0 , k 00 are gains dependent on the power setpoint.
The equation that allows the droop to work is:
(t) = 0 m(Pc P (t))
(4.16)
(4.17)
The PI is a block with a proportional and integral gain as indicated below: Kp + Ksi
Based on [5] the microturbine model is summarized in ( 4.18)-( 4.28). The block diagram of
the system is given in Figure 4.4
VLL = Kv sin(t)
Vdc =
3L
3
|VLL |
Idc
(4.18)
(4.19)
37
3
3
Kv L.Idc
(4.20)
3
3
Kv = Vdc + L.Idc
(4.21)
Vdc =
3Kv
(4.22)
Eg = Vdc + Kx ..Idc
(4.23)
Pm = Vdc Idc
(4.24)
2
Pm = Ke Idc Kx Idc
(4.25)
Eg = Ke =
where Ke =
where Kx =
3L
3Kv
{V /(rad/ sec)}
{/(rad/ sec)}
Tm =
Pm
2
= Ke Idc Kx Idc
d
1
= (Tm Tt )
dt
J
Vdc
1
=
C
(4.26)
(4.27)
Z
(Idc IL )dt
(4.28)
38
x 10
7
6
Power[W]
5
4
3
2
1
0
50
100
150
200
Time [s]
250
300
350
The microturbine system response to this power command is displayed in Figures 4.8- 4.11.
Figure 4.8 shows the output power of the system; Figure 4.9 the shaft speed; Figure 4.10 the
DC link voltage; and Figure 4.11 the rotor speed.
x 10
39
Output Power[W]
6
5
4
3
2
1
0
50
100
150
200
Time [s]
250
300
350
18
x 10
10
x 10
16
9
Shaft Speed[RPM]
14
12
60
80
100
10
8
6
4
2
0
50
100
150
200
Time [s]
250
300
350
40
1000
900
800
700
Voltage[V]
600
500
400
300
200
100
0
50
100
150
200
Time [s]
250
300
350
x 10
Rotor Speed[RPM]
50
100
150
200
Time [s]
250
300
350
However, there are some responses which do not follow the reference trajectories, such as
the rotor speed at higher power levels and the dc link voltage steady state value for mid-level
output power. It is believed that a better tuned controller, would result in improved results.
More complex controllers could also be added to improve the response.
Chapter 5
Wind Turbine
5.1 Introduction
Wind energy is expected to be one of the most prominent sources of electrical energy in years to
come. The increasing concerns of environmental issues demand the search for more sustainable
electrical sources. Wind turbines along with solar energy and fuel cells are possible solutions
for the environmental-friendly energy production. In this theses, the wind power as integrated
system will be studied.
Wind turbines are packaged systems that include a rotor, a generator, turbine blades, and
a drive or a coupling device. As wind blows through the blades, the air exerts aerodynamic
forces that cause the blades to turn the rotor. As the rotor turns, its speed is altered to match
the operating speed of the generator. Most systems have a gearbox and a generator in a single
unit behind the turbine blades. As with photovoltaic (PV) systems, the output of the generator
is processed by an inverter that changes the electricity from DC to AC so that the electricity can
be used.
42
Rotor
Generating
System
Grid
Connection
Us
Mechanical
Power
(Translation)
Mechanical
Power
(Rotation)
Grid
Is
Electrical
Power
Rotor
Gear box
Squirrel
cage
induction
generator
Grid
Is
Us
Compensating
capacitors
Figure 5.2: Squirrel cage induction generator is used in wind turbines as generating system.
The slip, and also the rotor speed of a squirrel cage induction generator vary with the generated power. These rotor speed variations are, very small. Therefore, the turbine is normally
considered to operate at constant speed. Because of the squirrel cage induction generator consumes reactive power, capacitors are often added to generate magnetizing currents in the case
of large wind turbines and/or weak grids, and improving the power factor of the system. The
power extracted from the wind needs to be limited, because otherwise the generator could be
overloaded or the pullout torque could be exceeded, leading to rotor speed instability [43], [42].
In such case, this is often done by using the stall effect. This means that the rotor geometry is
designed in such a way that its aerodynamic properties make the rotor efficiency decrease in
high wind speeds, thus limiting the power extracted from the wind and preventing the generator from being damaged and the rotor speed from becoming unstable. Thus, during normal
operation of a stall regulated wind turbine no controllers are reactive.
43
5.2.2 Doubly Fed (Wound Rotor) Induction Generator and Direct Drive Synchronous
Generator
Figure 5.3 shows the other two generating systems. They are used in variable speed turbines.
With these it is possible to increase the energy captured by the aerodynamic rotor by maintaining the optimum power coefficient over a wide range of wind speeds [42]. However it is then
necessary to decouple the speed of the rotor from the frequency of the grid through some form
of power electronic converters. In the doubly fed induction generator, a back-to-back voltage
source converter feeds the three phase rotor winding. In this way, the mechanical and electrical
rotor frequencies are decoupled and the electrical stator and rotor frequency can be matched,
independently of the mechanical rotor speed. In the direct drive synchronous generator, the
generator is completely decoupled from the grid by a power electronics converter. The grid
side of this converter is a voltage source converter, i.e. an IGBT (Insulated Gate Bipolar Transistor) bridge. The generator side can either be a voltage source converter or a diode rectifier.
The generator is excited using either an excitation winding or permanent magnets.
Rotor
Direct drive
synchronous
generator
Us
Is
Convertor
UC
Grid
IC
Doubly fed
(wound rotor)
induction
generator
Rotor
Gear box
Is
Us
Ur
Grid
Convertor
Ir
IC
Figure 5.3: Generating systems used in wind turbines: direct synchronous generator and doubly fed (wound rotor)
induction generator.
44
torque versus speed characteristic than the synchronous generator, thus reducing the mechanical loads and it is also cheaper. This concept is still applied nowadays by some manufacturers.
To limit the power extracted from the wind at high wind speeds, either pitch control or stall
control can be applied. Many papers on modelling of a wind turbine with a directly grid coupled squirrel cage induction generator can be found in the literature, both in combination with
pitch control and with stall control of the mechanical power, e.g. [45].
The problems with design of the pitch control result that the wind turbine with a directly
grid coupled squirrel cage induction generator and pitch control does no longer appear in the
product portfolio of any manufacturer. It has appeared to be rather difficult to limit the output
power to the nominal value by controlling the pitch of the rotor blades. Thus, although models
and analysis of a wind turbine with a directly grid coupled squirrel cage induction generator
still appear in journals and conference proceedings now and then, the value of these is rather
limited [46].
Nowadays, the more modern variable speed wind turbine with a doubly fed induction generator has replaced of the conventional constant speed wind turbine with a directly grid coupled squirrel cage induction generator. The manufacturers have also started to apply a direct
drive synchronous generator grid coupled through a power electronic converter of the full
generator rating. Therefore, modelling efforts have been given to these wind turbine concepts
as well. Because the variable speed wind turbines are complicated systems, most papers addressing their modelling only cover one subsystem, such as the electromechanical conversion
system , the drive train, the control of the generator currents and the DC link voltage or the
rotor speed controller, e.g. [47].
As the power developed is proportional to the cube of the wind speed it is obviously important to locate any electricity generating turbines in areas of high mean annual wind speed, and
the available wind resource is an important factor in determining where wind farms are sited
[1]. Often the areas of high wind speed will be away from habitation and the associated welldeveloped electrical distribution network, leading to a requirement for careful consideration of
the integration of wind turbines to relatively weak electrical distribution networks. The difference in the density of the working fluid(water and air) illustrates clearly why a wind turbine
rotor of a given rating is much larger in size than a hydro-turbine [1].
(5.1)
where:
P = Power (W).
Cp = power coefficient.
Vw = Wind velocity (m/s).
A = swept area of rotor disc(m2 ).
= density of air (1.225 kg/m3 ).
The force extracted on the rotor is proportional to the square of the wind speed and so the
45
wind turbine must be designed to withstand large forces during storms. Most of modern designs are three-bladed horizontal-axis rotors as this gives good value of peak Cp together with
an aesthetically pleasing design [1].
The power coefficient Cp is a measure of how much of energy in the wind is extracted by
the turbine. It varies with rotor design and the relative speed of the rotor and wind (known
as the tip speed ratio) to give a maximum practical value of approximately 0.4 [1]. The power
coefficient Cp is a function of the tip speed ratio and the pitch angle , which will be investigated further. The calculation of the performance coefficient requires the use of blade element
theory [48],[42]. As this requires knowledge of aerodynamics and the computations are rather
complicated, numerical approximations have been developed[44]. Here the following function
will be used:
Cp (, ) = 0.5176(
21
116
0.4 5)e i + 0.0068
i
(5.2)
with
1
0.035
1
=
3
i
+ 0.08 + 1
(5.3)
Figure 5.4 shows the Cp (, ) versus characteristics for various values of . Using the actual
values of the wind and rotor speed, which determine , and the pitch angle, the mechanical
power extracted from the wind can be calculated from equations ( 5.1) to ( 5.3). The maximum
value of Cp (cpmax =0.48) is achieved for = 0 and for = 8.1. This particular value of is
defined as the nominal value (nom ).
= 0
pmax
performance coefficient cp
0.4
= 5
0.3
= 10
0.2
0.1
= 20
0
nom
0.1
5
10
Tip speed ratio Lambda
= 15
15
Figure 5.4: Performance coefficient Cp as a function of the tip speed ratio with pitch angel as a parameter .
46
which means that the currents are outputs instead of inputs and the real power and reactive
power have positive signs when they are fed into the grid. By using the generator convention,
the following set of equations are obtained [43]:
vds = Rs ids s qs + ddtds
d
vqs = Rs iqs + s ds + dtqs
vdr = Rr idr ss qr + ddtdr
d
vqr = Rr iqr + ss dr + dtqr
(5.4)
where
v is the voltage in [V].
i is the current in [A].
R is the resistance in [].
s is the stator electrical frequency in [rad/s].
is the flux linkage in [Vs].
s is the rotor slip.
Subscripts d and q are direct and quadrature axis components respectively; subscripts s and
r indicate the stator and the rotor quantities. All the quantities in equation ( 5.4) are functions
of time. The d-q reference frame is rotating at synchronous speed with the q- axis 90 ahead
of the d-axis. The position of the d-axis coincides with the maximum of the stator flux, which
means that vqs equals the terminal voltage et and vds equals to zero. The flux linkages can be
calculated using the following set of equations in per unit [43].
ds = (Ls + Lm )ids Lm idr
qs = (Ls + Lm )iqs Lm iqr
ds = (Lr + Lm )idr Lm ids
qr = (Lr + Lm )iqr Lm iqs
(5.5)
where Lm is the mutual inductance and Ls and Lr are the stator and rotor leakage inductances, respectively. In equation( 5.5) the generator convention is used again. The rotor slip is
defined as [43]:
s=
s P2 m
s
(5.6)
where P is the number of poles and m is the mechanical frequency of the generator in
[rad/s].
From equations 5.4, 5.5, we can derive the voltage current relationships of the doubly fed
induction generator.
Reference [43] proposes that the rotor and stator transients, represented by the last term in
equation ( 5.4) are to be neglected. Substituting ( 5.5)in to ( 5.4)results in:
vds = Rs ids + s ((Ls + Lm )iqs + Lm idr )
vqs = Rs iqs s ((Ls + Lm )ids + Lm idr )
vdr = Rr idr + ss ((Lr + Lm )iqr + Lm iqs )
vqr = Rr iqr + ss ((Lr + Lm )idr + Lm ids )
(5.7)
47
The active power P and reactive power Q generated by the generator can be written as:
P = vds ids + vqs iqs + vdr idr + vqr iqr
Q = vds ids vqs iqs + vdr idr vqr iqr
(5.8)
From this equation, it can once more be concluded that the reactive power Q is not necessarily
equal to the generated reactive power fed into the grid. Equations ( 5.7) and ( 5.8) describe the
electrical part the generator. However, also the mechanical part must be taken into account in
a dynamic model. The following expression gives the electromechanical torque developed by
the generator:
T e = dr iqr qr idr
(5.9)
The changes in generator speed that result from a difference in electrical and mechanical
torque can be calculated using the generator equation of motion
dm
1
=
(Tm Te )
dt
2Hm
(5.10)
in which Hm is the equivalent inertia constant of the generator rotor [s] and Tm is the mechanical torque [p.u.].
Figure 5.5 shows the speed- power turbine curve which reflects both the aerodynamic power
and the generated power. At low wind speeds, the output power is too low to be exploited.
Normally turbines are started when the wind speed exceeds 3-4 m/s. We can see also that the
wind turbine started at 5 m/s and the output power increases with the cube of the wind speed
until the rated wind speed is reached.
At wind speeds from 12 m/s to 25 m/s the power is limited to the rated power of the wind
turbines by means of stall-regulation or pitch-control. At wind speed over 20-25 m/s wind
turbines are normally stopped to avoid high mechanical loads. The wind speed at which wind
turbines are stopped is called cut-out speed.
1.6
16.2 m/s
1.4
1.2
D
1
0.8
12 m/s
0.6
0.4
0.2
B
5 m/s
0
0.6
A
0.7
0.8
0.9
1
Turbine speed (p.u)
1.1
1.2
1.3
48
Wind Speed(m/s)
13
12
11
10
9
8
10
15
20
25
Time(sc)
30
35
40
45
50
Figure 5.7 shows the active power P. The generated active power starts increasing smoothly
(together with the turbine speed) to reach its rated value of 1.5 MW in approximately 19 s.
1.5
0.5
10
15
20
25
Time [s]
30
35
40
45
50
The response of the reactive power due to a change in the wind speed is shown in Figure
5.8. It can be seen that at nominal power, the wind turbine absorbs 0.11 Mvar (generated Q =
-0.11 Mvar).
Figure 5.9 shows the pitch angle response due to the change in the wind speed. In this
figure, it is clear that, initially, the pitch angle of the turbine blades is zero degrees and the
turbine operating point follows the red curve of the turbine power characteristic up to point
D. Then the pitch angle is increased from 0 deg to 0.078 deg in order to limit the mechanical
power.
In Figure 5.10 the turbine speed increased when the wind speed increased from 0.8 pu to
1.21 pu.
To simulate the wind turbine responses for different values of wind speed, the initial wind
speed is below the nominal wind speed which is assumed to be 14 m/s. After 7 s a wind speed
ramp starts which leads to an increase in the average wind speed in 30 s after a 10 s a wind gust
with an amplitude of -3 m/s and duration of 10 s occurs [50].
49
0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
10
15
20
25
Time [s]
30
35
40
45
50
40
45
50
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
10
15
20
25
Time [s]
30
35
The results are illustrated in Figure 5.11 - 5.13. At 20 s, the nominal power of the wind
turbine is reached because the pitch angle controller is not used which can prevent the rotor
overspeeding.
Next the responses of measured wind sequences is to be simulated. Figures 5.15- 5.17 show
the responses of the active power, the pitch angle, and the rotor speed due to the the input
measured wind speed 5.14 . The wind speed measurements were downloaded from "Database
of Wind Characteristics" which is located at DTU Denmark [51].
It can be seen from the results that the response from the simulated input and measured
input wind speed have almost the same range fluctuations of the output power , the range of
the response of the rotor speed fluctuations are similar, and the behavior of the response of the
pitch angle are different as there were no pitch controller in the design model.
50
1.4
Rotor Speed [P u]
1.3
1.2
1.1
1
0.9
0.8
0.7
10
15
20
25
Time [s]
30
35
40
45
50
1.5
0.5
10
15
20
25
30
35
40
45
50
Time [s]
Figure 5.11: Generated active power due to different values of wind speed.
5
10
15
20
25
30
35
40
45
50
Time [s]
1.2
1.1
0.9
0.8
10
15
20
25
30
35
40
45
50
Time [s]
51
15
14.5
14
13.5
13
12.5
12
10
20
30
40
50
60
Time [s]
1.5
0.5
10
20
30
40
50
60
Time [s]
Figure 5.15: Response of the generated active power due to the measured sequence wind speed input .
1.5
0.5
10
20
30
40
50
60
Time [s]
Figure 5.16: Response of the pitch angle due to measured sequence wind speed input.
1.3
1.2
1.1
0.9
0.8
10
20
30
40
50
60
Time [s]
Figure 5.17: Response of the rotor speed due to measured sequence wind speed input.
Chapter 6
Photovoltaic Cell
6.1 Introduction
The photovoltaic (PV) generation systems are expected to increase significantly worldwide.
PVs are an attractive source of renewable energy for distributed urban power generation due
to their relatively small size and noiseless operation. PV generating technologies have the
advantage that more units can be added to meet load increase demand.
Major advantages of the photovoltaic power are as follows [52]:
Short lead time to design, install, and start up a new plant.
Highly modular, hence, the plant economy is not a strong function of size.
Power output matches very well with peak load demands.
Static structure, no moving parts, hence, no noise.
High power capability per unit of weight.
Longer life with little maintenance because of no moving parts.
Highly mobile and portable because of light weight.
Photovoltaic generation are systems which convert the sunlight directly to electricity. PV
technology is well established and widely used for power supplies to sites remote from the
distribution network [1].
Photovoltaic cells can be divided into four groups: crystalline cells, thin-film cells, dyesensitised solar cells (DYSC or Grtzel-cell) and multilayer cells. The latter can also be considered as several layers of thin-film PV cells. The different types are described in [53].
Figure 6.1 shows the schematic diagram of an inverter for small PV grid connected system.
The inverter typically consists of the following:
Maximum power point tracking ( MPPT) circuit.
Energy storage element, usually a capacitor.
DC:DC converter to increase the voltage.
DC:AC inverter stage.
52
6.2. MODELLING
53
PV
Module
MPPT
Energy
Storage
DC:DC
DC:AC
Output
Filter
Isolation
Figure 6.1: Schematic diagram of small PV inverter for grid connected operation.
6.2 Modelling
An initial understanding of the performance of a solar cell may be obtained by considering it as
a diode in which the light energy, in form of photons with the appropriate energy level, falls on
the cell and generates electron-hole pairs. The electrons and holes are separated by the electric
field established at the junction of the diode and are then driven around an external circuit by
this junction potential. There are losses associated with the series and shunt resistance of the
cell as well as leakage of some of the current back across the p-n junction. This leads to the
equivalent circuit of Figure 6.2 [1], [54].
Rs
I
+
I ph
ID
D
R sh
Rload
__
The PV cell can be modeled as a diode in parallel with a constant current source and a shunt
resistor. These three components are in series with the series resistor.
The output-terminal current I is equal to the light-generated current Iph , less than the diodecurrent ID and the shunt-leakage current Ish .
I = Iph ID Ish
(6.1)
The series resistance Rs represents the internal resistance to the current flow, and depends
on the p-n junction depth, the impurities and the contact resistance. The shunt resistance Rsh
is inversely related to the leakage current to the ground. In an ideal PV cell, Rs = 0 (no series
loss), and Rsh = (no leakage to ground). The PV cell conversion efficiency is sensitive to small
variations in Rs , but is insensitive to variations in Rsh . A small increase in Rs can decrease
the PV output significantly. In the equivalent circuit, the current delivered to the external load
equals the current Iph generated by the illumination, less than the diode current ID and the
54
ground-shunt current Ish . The open circuit voltage Uoc of the cell is obtained when the load
current is zero, i.e., when I = 0, and is given by the following:
Uoc = U + IRs
(6.2)
ID = Id
qUoc
1
Acf KB T
(6.3)
qUoc
Uoc
I = Iph Ios exp
1
AKT
Rsh
(6.4)
where
Iph =
Ios = Ior
G
[ISCR + KI (T 25)]
100
T
Tr
qEGO
exp
BK
1
1
Tr
T
(6.5)
and
I,V = cell output current and voltage.
Ios = cell reverse saturation current.
19
q = electron charge=1.6*10
Coulombs.
A,B= ideality factor of p-n junction.
K=Bolzmann constant.
T =cell temperature [ C].
KI = short circuit current temperature coefficient at ISCR ,KI = 0.0017 A/ C.
G= solar irradiation in W/m2 .
ISCR = short circuit current at 25 C and 1000W/m2 .
Iph = light generated current.
EGO =band gap for silicon.
Tr =reference temperature, Tr =301.18 K.
Ior =cell saturation current at Tr .
Rsh =shunt resistance.
(6.6)
55
Rs =series resistance.
ISCR , the current at maximum power point (Impp ), the voltage at maximum power point
(Vmpp ), and the open circuit voltage of the cell Uoc , are given by the manufacturers. Table 6.1
illustrates the Standard Test Condition (STC) of AM1.5, 1000W/m2 and 25 C, also the date for
80W PHOTOWATT which is used for the simulation study [55]
Table 6.1: Parameters for 80W photowatt panel PWZ750 at STC.
Parameter
Maximum Power Point, (Pmpp )
Minimum Power Point, (Pmin )
Current at MPP,(Impp )
Voltage at MPP,(Vmpp )
Short Circuit Current,(ISCR )
Open Circuit Voltage,(Uoc )
Short circuit current temperature coefficient,scT
Open circuit voltage temperature coefficient,ocT
NOCT (Normal Operating Cell Temperature)
Insolation, G=0.8W/m2 , Ta =20 C,wind speed=1m/s
Value
80W
75.1W
4.6A
17.3V
5A
21.9V
1.57mA/ C
-78.2mV / C
45 C
G = 1000W/m2
5
G = 800W/m2
4
Current [A]
G = 600W/m
3
G = 400W/m2
2
G = 200W/m2
1
10
15
20
25
Voltage [V]
Figures 6.5, 6.6 show the I-U characteristics for different values of temperature and fixed
irradiance of 1000W/m2 respectively.
From the figures, we can conclude that when the irradiation is 1000W/m2 , which corresponds approximately to a cloud-free, sunny day, the upper curve shows that the open-circuit
voltage of the cell is about 22 Volt. As the load (current) of the cell increases, the voltage de-
56
100
90
80
Power [W]
70
Increasing G
60
50
40
30
20
10
0
10
15
20
25
Voltage [V]
60 C
current [A]
40 C
25 C
10 C
2
10
15
20
25
Voltage [V]
creases and at short-circuit (voltage = 0) the current is approximately 5 A. At open circuit and
at short-circuit, no power is produced. At a point called the maximum power point (MPP),
maximum power is gained from the PV-cell. To visualise this, a rectangle can be drawn from a
point on the curve to the x and y-axis. For the point where this rectangle has the largest area,
the maximum power is generated. At a lower irradiation, the short-circuit current decreases
approximately linearly with irradiation. The open circuit voltage does not decrease as much
until a very low irradiation. However, the open circuit voltage is much more affected by the
temperature of the PV-cell. At a higher temperature, the open circuit voltage decreases. The
phenomenon has quite a large impact and it decreases the output power by approximately 15
% at a temperature increase from 25 C to 60 C.
The effect of irradiance and cell temperature on I-V characteristic curve is shown in Figures
57
90
80
70
Power [W]
60
50
40
30
Increasing T
20
10
0
10
15
20
25
Voltage [V]
[1000,25 ]
5
[1000,60 ]
Current [A]
[800,45 ]
[500,25 ]
10
15
20
25
Voltage [V]
6.3 and 6.5. Figure 6.3 shows that the maximum power output varies almost linearly with the
irradiance. Figure 6.5 shows that the maximum output power from the PV decreases as the
temperature increases.
Chapter 7
7.1. CONCLUSIONS
59
lag network., which is characterized by gain, and time constant. Figure 2.1, and equation
2.1 show the actuator and the current driver constant. The output of the actuator is the fuelflow. Fuel flow is then converted to a mechanical torque after a time delay and engine torque
constant, which can be represented by the model of the diesel engine as shown in Figure 2.2,
and equation 2.2.
Based on the results of this work it can be concluded that control of the speed of the diesel
engine using the genetic algorithm based on RLS and PID controller can control the system
efficiently. The proposed controller procedure seems to control the system even if the system
has a time delay variation and load injection. The optimal controller obtained using the genetic
algorithm is far more efficient than that used in [15], [17], and [19] as overshoot, rise time
and settling time are greatly reduced. Load injection does not affect the resulting controller.
The response for the system shows good performance in reducing the overshoot, rise time,
and settling when the load applied. Moreover, it was shown that the proposed method was
effective in controlling the speed of the diesel engine. Genetic algorithm was proven to be an
efficient way of finding optimal controller.
Another useful characteristic of genetic algorithms relevant to our controllers is that they
can be used to obtain a controller with specific specification. Objective functions can be used
to specify certain overshoots, rise time and settling times. For example, multi-objective optimization could also be used. There could be an evaluation function for calculating fitness and
another evaluation function to make sure that the PID values conform to a specific standard.
An example of this would be to reject PID values that cause an overshoot greater than what is
desired.
60
tool suitable to study and to perform accurate analysis of most electrical phenomenon that occurs when a microturbine is connected to the grid. The simulation results obtained for different
levels of power demand show the usefulness of the model and its accuracy. The aim of future
work will be to investigate more questions such as islanding operation.
61
Bibliography
[1] N. Jenkins, R. Allan, P. Crossley, D. Kirschen, and G. Strbac. Embedded Generation. The
Institution of Electrical Engineers, UK, 2000.
[2] T.E Hoff, H.J Wenger, and B.K. Farmer. Distributed generation: An alternative to electric
utility investments in system capacity. Energy Policy, 24(2):137 147, 1996.
[3] B. Lasseter. Microgrids [distributed power generation]. In IEEE Power Engineering Society
Winter Meeting, 2001., volume 1, pages 146149, Columbus, Ohio, Feb 2001.
[4] R. Lasseter. Microgrids. In IEEE Power Engineering Society Winter Meeting, 2002., volume 1,
pages 305308, New York, NY, 2002.
[5] Y. Zoka, H. Sasaki, N. Yorino, K. Kawahara, and C.C Liu. An interaction problem of
distributed generators installed in a microgrid. In Proceedings of IEEE on Electric Utility
Deregulation, Restructuring and Power Technologies Conference, volume 2, pages 795 799,
Hong Kong, April 2004.
[6] Sakis Meliopoulos. Challenges in simulation and design of tgrids. In Proceedings of the
2002 IEEE/PES Winter Meeting, New York, NY, 2002.
[7] G. Venkataramanan and M. Illindala. Microgrids and sensitive loads. In Proc. IEEE Power
Eng. Soc. Winter Meeting, 2002, volume 1, pages 316322, 2002.
[8] J.M. Correa, S. Chakraborty, M.G. Simoes, and F.A. Farret. A single phase high frequency
ac microgrid with an unified power quality conditioner. In Proc. IEEE-Industry Applications
Conference, volume 2, pages 956962, October 2003. Conference Record of the 38th IAS
Annual Meeting, SaltLake City, UT.
[9] P. P. Barker and B. K Johnson. Power system modeling requirements for rotating machine
interfaced distributed resources. In Power Engineering Society Summer Meeting, volume 1,
pages 161166, July 2002.
[10] S. Blazwicz and D. Kleinschmidt.
Distributed generation: System interface.
[Online]. Available: http://www.encorp.com/dwnld/pdf/whitepaper/wp_ADL_2.pdf,
May 1999.
[11] T. Ackermann and V. Knyazkin. Interaction between distributed generation and the distribution network: operation aspects. In Transmission and Distribution Conference and Exhibition 2002: Asia Pacific. IEEE/PES, volume 2, pages 13571362, Oct 2002.
62
BIBLIOGRAPHY
63
[12] A. Amstutz and K. Luigi. EGO sensor based robust output control of EGR in diesel engines. IEEE Transactions on Control System Technology, 3(5):39 48, March 1995.
[13] S. Hggmark, V. Neimane, U. Axlsson, P. Holmberg, G. Karlsson, K. Kauhaniemi,
M. Olsson, and C. Liljergren.
Asppects of different distributed generation technologies, CODGUENT WP.
March 2003.
[Online]. Available:
http://www.energia.fi/attachment.asp?Section=1353&Item=10792.
[14] G. S. Stavrakakis and G. N. Kariniotakis. A general simulation algorithm for the accurate
assessment of isolated diesel-wind turbines systems interaction. i. a general multimachine
power system model. IEEE Transactions on Energy Conversion, 10(3):577 583, Sept 1995.
[15] B. Kuang, Y. Wang, and Y. L. Tan. An H controller design for diesel engine systems. In
Power System Technology, International Conference Proceedings, volume 1, pages 6166, Dec
2000.
[16] S. Roy, O.P. Malik, and G.S. Hope. A least squares based model fitting identification technique for diesel prime movers with unknown dead time. IEEE Transactions on Energy
Conversion, 6(2):251 256, June 1991.
[17] S. Roy, O.P. Malik, and G.S. Hope. An adaptive control scheme for speed control of diesel
driven power-plants. IEEE Transactions on Energy Conversion, 6(4):605 611, Dec. 1991.
[18] Y. Hu, M. Cirstea, M. McCormick, and L. Haydock. Modelling and simulation of a variable
speed stand-alone generator system. In Proceedings of Power Electronics and Variable Speed
Drives, 2000. Eighth International Conference, pages 372 377, Sept. 2000.
[19] S. Roy, O.P. Malik, and G.S. Hope. A k- step predictive scheme for speed control of diesel
driven power plant. IEEE Transactions on Industry Application, 29(2):389 396, March/April
1993.
[20] A. Ozeki, H. Fujikawa, S. Yamada, and H. Tanaka. A design method of self-tuning controller for unknown time-delay system. In Proceedings Industrial Electronics, Control, Instrumentation, and Automation, Power Electronics and Motion Control Conference, volume 3,
pages 1218 1223, Tokyo,Japan, Nov 1992.
[21] P. E. Wellstead and M. B. Zarrop. Self-tuning systems: control and signal processing. John
Wiley & Sons Inc, April 1991.
[22] H. N. Koivo and A. Reijonen. Tuning of PID controller for varying time- delay system.
In Proceedings of the IEEE International Conference on Mechatronics ICM 04, pages 446 451,
Istanbul,Turkey, 2004.
[23] K Zenger. Analysis and Controller Design of Time-Variable Flow Processes. PhD thesis, Laboratory of Control Engineering, HELSINKI UNIVERSITY OF TECHNOLOGY, 2003. [Online].
Available: http://lib.hut.fi/Diss/2003/isbn9512264390/.
[24] M. Dotoli, G. Maione, D. Naso, and E. B. Turchiano. Genetic identification of dynamical
systems with static nonlinearities. In Proceedings of Mountain Workshop Soft Computing
Industrial Applications, pages 65 70, June 2001.
64
BIBLIOGRAPHY
[25] K. Kristinnson and G. A. Dumont. System identification and control using genetic algorithms. IEEE Transactions on System, Man, Cybern, 22:1033 1046, Sept.-Oct 1992.
[26] C. H. Marrison and R. F. Stengel. Robust control system design using random search and
genetic algorithms. IEEE Transactions on Automatic Control, 42:835839, June1997.
[27] F. Cupertino, E. Mininno, D. Naso, B. Turchiano, and L. Salvatore. On- line genetic design of anti- windup unstructured controller for electrical drives with variable load. IEEE
Transactions on Evolutionary Computation, 8(4):347363, August 2004.
[28] A.J. Chipperfield, P.J. Flemming, A. Polheim, and C. M Fonscea. Genetic Algorithm Toolbox.
University of Sheffield. Available: http://www.ie.ncsu.edu/mirage/GAToolBox/gaot/.
[29] W. Morgantown. Fuel Cell Handbook. 6th Edition, EG&G Technical Services, Virginia,
November 2002. [Online]. Available: http://www.rwth-aachen.de/lbz/linksframe.html.
[30] J.A. Smith, M.H. Nehrir, V. Gerez, and S.R Shaw. A broad look at the workings, types, and
applications of fuel cells. In Power Engineering Society Summer Meeting, volume 1, pages
7075, Chicago, USA, July 2002.
[31] A.R Sakhare, A. Davari, and A. Feliachi. Control of stand alone solid oxide fuel cell using
fuzzy logic. In Proceedings of the 35th Southeastern Symposium on System Theory, 2003., pages
473 476, 16-18 March. 2003.
[32] M.W. Ellis, M.R. Von Spakovsky, and D.J Nelson. Fuel cell systems: efficient, flexible
energy conversion for the 21st century. In Proceedings of the IEEE, volume 89, pages 1808
1818, Dec. 2001.
[33] J. Padulles, G.W. Ault, C.A. Smith, and J.R. McDonald. Fuel cell plant dynamic modelling
for power systems simulation. In Proceedings of 3dth universities power engineering conference, volume 34, pages 2125, 1999.
[34] K. Sedghisigarchi and A. Feliachi. Dynamic and transient analysis of power distribution
systems with fuel cells-part i: fuel-cell dynamic model. IEEE Transactions on Energy Conversion, 19(2):423 428, 2004.
[35] Y. Zhu and K. Tomsovic. Development of models for analyzing the load- following performance of microturbine and fuel cells. Electric Power System Research journal, Dec. 2001.
[36] C. Sharma. Modeling of an island grid. IEEE Transaction on Power System, 13(3):971978,
August 1998.
[37] W. G. Scott. Micro- turbine generators for distribution systems. IEEE Industry Applications
Magazine, 4(4):5762, May/June 1998.
[38] H. Nikkhajoei and M.R. Iravani. Modeling and analysis of a micro-turbine generation
system. In IEEE Power Engineering Society Summer Meeting, volume 1, pages 167 169,
Chicago, USA, July 2002.
BIBLIOGRAPHY
65
66
BIBLIOGRAPHY
[53] B. Lindgren. A Power Converter for Photovoltaic Applications. PhD thesis, Department of
Electric Power Engineering, Chalmers University of Technology, Gteborg, Sweden, 2000.
[54] K.H. Hussein, I. Muta, T. Hoshino, and M. Osakada. Maximum photovoltaic power tracking: an algorithm for rapidly changing atmospheric conditions. In Generation, Transmission
and Distribution, IEE Proceedings, volume 142, pages 59 64, Jan 1995.
[55] M.F Ishengoma and E.E Norum. Design and implementation of adigitally controlled
stand-alone photovoltaic power supply. In Nordic Workshop on Power and Industrial Electronics(Norpie 2002), Stockholm, Sweden, 12-14 Aug 2002.
Report 135
Blomqvist, E.
Security in Sensor Networks. February 2003.
Report 136
Zenger, K.
Modelling, Analysis and Controller Design of Time-Variable Flow Processes. March 2003.
Report 137
Hasu, V.
Adaptive Beamforming and Power Control in Wireless Communication Systems. August 2003.
Report 138
Report 139
Report 140
Viitamki, P.
Hybrid Modeling of Paper Machine Grade Changes. May 2004.
Report 141
Pyhnen, S.
Support Vector Machine Based Classification in Condition Monitoring of Induction Motors. June 2004.
Report 142
Elmusrati, M. S.
Radio Resource Scheduling and Smart Antennas in Cellular CDMA Communication Systems. August 2004.
Report 143
Tenno, A.
Modelling and Evaluation of Valve-Regulated Lead-Acid Batteries. September 2004.
Report 144
Hytyniemi, H.
Hebbian Neuron Grids: System Theoretic Approach. September 2004.
Report 145
Hytyniemi, H. (ed.)
Complex Systems: Science at the Edge of Chaos - Collected papers of the Spring 2003 postgraduate seminar.
October 2004.
Report 146
Paanasalo, J.
Modelling and Control of Printing Paper Surface Winding. June 2005.
Report 147
Mohamed, F.
Microgrid Modelling and Simulation. March 2006.
ISBN-13 978-951-22-8417-7
ISBN-10 951-22-8417-0
ISSN 0356-0872
Picaset Oy, Helsinki 2006