Applied Energy 87 (2010) 3051–3064

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Dynamic behavior of a stand-alone hybrid power generation system of wind turbine,

microturbine, solar array and battery storage
M. Kalantar *, S.M. Mousavi G.
Center of Excellence for Power System Automation and Operation, Department of Electrical Engineering, Iran University of Science and Technology, Narmak, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents dynamic behavior and simulation results in a stand-alone hybrid power generation
Received 14 April 2009 system of wind turbine, microturbine, solar array and battery storage. The hybrid system consists of a
Received in revised form 8 January 2010 195 kW wind turbine, an 85 kW solar array; a 230 kW microturbine and a 2.14 kAh lead acid battery
Accepted 23 February 2010
pack optimized based on economic analysis using genetic algorithm (GA). At first, a developed Lyapu-
Available online 8 April 2010
nov model reference adaptive feedback linearization method accompanied by an indirect space vector
control is applied for extraction of maximum energy from a variable speed wind power generation sys-
Keywords:
tem. Then, a fuzzy logic controller is designed for the mentioned purpose and its performance is com-
Wind turbine
Solar array
pared with the proposed adaptive controller. For meeting more load demands, the solar array is
Microturbine integrated with the wind turbine. In addition, the microturbine and the battery storage are combined
Adaptive control with the wind and solar power generation system as a backup to satisfy the load demand under all
Supervisory control conditions.
Indirect space vector control A supervisory controller is designed in order to manage energy between the maximum energy cap-
tured from the wind turbine/solar arrays, and consumed energies of the load, dump load, battery state
of charge (SOC), and generated energy by the microturbine. Dynamic modeling and simulation are
accomplished using MATLAB Simulink™ 7.2.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction 2. Literature review

Renewable energy resources, called sustainable or alternative
energy, are energies generated from natural resources such as
wind, sunlight, tide, hydro, biomass and geothermal which are
naturally replenished. Energy crisis, climate changes such as
atmosphere temperature rise due to the increase of greenhouse
gases emission and the Kyoto Protocol restrictions in generation
of these gases, coupled with high oil prices, limitation and deple-
tion of fossil fuels reserves make renewable energies more
noticeable. Moreover, since many remote rural areas in all over
the world are not electrified, for example about 72 million
households in rural India do not have access to electricity [1],
one of the interesting utilization of the renewable energies is
to electrify many remote villages and rural areas or rugged ter-
rain located so far from power stations and distribution net-
works or utility lines which are uneconomical to install.
Vulnerability to unpredictable climatic changes and dependency
of renewable energy systems on weather conditions remain
among their drawbacks.
* Corresponding author. Tel.: +98 2177240540 50x2665; fax: +98 2177240354.
E-mail address: Kalantar@iust.ac.ir (M. Kalantar).
Among the renewable energy resources, wind power has had
the fastest growth in the world (at the rate of 30% annually) in
many developed and developing countries over the last 20 years
[2]. For example, installed capacity in Germany is more than
20 GW with an annual output of about 40 TWh in 2007 [3]. In or-
der to convert the wind energy to electrical energy, two types of
wind turbines are employed; fixed speed and variable speed wind
turbines installed onshore or offshore [4]. In the fixed speed wind
turbine, electrical generator is connected directly to the power
grid. Therefore, the generator runs at constant frequency and
speed. Active and reactive power control of these turbines is de-
scribed in [5,6]. The variable speed wind turbine (VSWT) is used
for more attraction of energy from the wind. The VSWT, which at-
tracts 10–15% more energy, has lower mechanical stress and less
power fluctuation in comparison with the fixed speed ones. Fur-
thermore, the wind turbine is divided into two large categories:
horizontal and vertical axis wind turbines. Savonius and Sherbious
are the most famous vertical axis wind turbines (VAWT). The aero-
dynamic efficiency of the VAWT is lower than the horizontal type
but complexity and price of these types are lower [7].
One of the most important studies in the VSWT is the applica-
tion of various control schemes for several purposes in the plant.

0306-2619/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2010.02.019
3052 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064
There are some methods to control VSWT. These methods are used
to track the maximum power, to control the voltage and frequency
of the load and reduce power fluctuation [8–10]. Pitch control is a
method to control power extracted from the VSWT [11]. Maximum
power attraction from the VSWT is achieved using linear control;
fuzzy logic control and hill climb searching (HCS) method [12–15].
Due to efficient and economical utilization of renewable ener-
gies, some of renewable energy resources such as wind turbine
and solar array are integrated [16]. Because of dependency on wind
speed and sun irradiance in such systems, their reliability in satis-
fying the load demands decreases under all conditions. Hence,
some studies propose the combination of a diesel generator as a
back up and wind/solar power generation systems [17,18].
Furthermore, every diesel generator has several major disad-
vantages such as low flexibility to respond the demand changes,
high pollution and noise, and low efficiency [19]. The above men-
tioned problems make a diesel generator to be inappropriate for
integrating with clean power generation systems.
Another element that plays an important role in an autonomous
wind–solar hybrid power generation system is battery storage.
Batteries are employed to store superfluous energy derived from
blowing wind and solar irradiance during windy or sunny days
and then to release during cloudy days or at nights. An ignored as-
pect in several previous researches is the lack of a suitable method
to estimate the SOC of the battery for the appropriate management
of energy in the hybrid power generation systems. So far, several
SOC estimation methods for batteries are employed such as fuzzy
logic, neural network, specific gravity measurement, open-circuit
voltage estimation, Kalman/extended Kalman filters and Ampere-
hour counter [20,21]. One of these methods will be utilized for
the SOC estimation of the battery in the hybrid system in the pre-
vious studies.
One of the most important studies in a hybrid system is size
optimization in order to achieve efficient and economical utiliza-
tion in the hybrid system. Various methods and techniques such
as a graphic construction technique, design space approach, genet-
ic programming (GA), and dynamic programming are implemented
for optimal sizing in various hybrid systems such as wind/solar/
battery, PV/fuel cell/battery and wind/solar/diesel/battery consid-
ered in [22–30].
In this paper, integration of microturbine with the wind/solar
hybrid power generation system is proposed to meet the load de-
mands in remote areas rich in wind and solar energies because
microturbine in comparison with diesel generator has more reli-
ability, more fuel flexibility, higher power-to-weight and less
maintenance, noise, pollution and power fluctuation. For example,
a traditional 30 kW diesel generator emits more nitrogen oxide in
1 h than a Capstone unit does in more than 9 days of full-load oper-
ation [31]. The microturbine and the battery storage are used as a
backup for the hybrid system of wind turbine and solar array.
Because of nonlinear behavior of the plant, the methods based
on linear control theory, such as a proportional integral derivative
(PID) controller, are not suitable for the maximum power point
tracking (MPPT). The maximum energy attraction and voltage
and frequency control of the load are performed using fuzzy logic
control method. Generally, a fuzzy logic controller does not guar-
antee an optimal response. Furthermore, it increases complexity
of the control system. The MPPT using the HCS method is appropri-
ate only for the wind turbines with quite small inertia. For the
wind turbines with large inertia, this method is not able to extract
the maximum power. Thus, the input–output feedback lineariza-
tion method is proposed to eliminate the nonlinear behavior of
the plant. Model reference adaptive control method is designed
to estimate the plant parameters such as optimum torque and ro-
tor angular speed in order to capture the maximum energy from
the VSWT when the wind speed change. Furthermore, a fuzzy logic
controller is implemented to extract the maximum power and its
performance is compared with the proposed adaptive controller.
For proper performance of central controller, a suitable SOC
estimation method is needed in order to generate the reference
values for the controllers in the hybrid system. Among the SOC
estimation methods, the Kalman filter, neural network and fuzzy
logic are very complex and also need too many calculations to be
executed. In the specific gravity method, the electrolyte may be-
come laminated resulting in diminishing the sensors lifetime. A dy-
namic mathematical model of a lead acid battery proposed by Durr
et al. [32] is used for this hybrid system. Based on the remaining
battery energy, a new value for the battery open-circuit voltage
is determined using the look-up table. The SOC calculation is rep-
resented in a look-up table relative to the battery open-circuit volt-
age (Voc).
For optimal sizing design of the hybrid system, the GA tech-
nique is implemented because of robust in finding global optimal
solutions. Moreover, a new model of power-fuel consumption is
obtained based on real data given by Capstone Company.
An appropriate supervisory controller is designed for energy
management between maximum energy captured from the wind
turbine/solar array and consumed energies of the load, dump load,
battery energy based on state of charge and the energy generated
by the microturbine. This paper is organized as follows: description
of the proposed system is illustrated in Section 3. Section 4 de-
scribes the control system design. Optimal design and economic
analysis is fulfilled in Section 5. Simulation results and discussion
are presented in Section 6. Section 7 is the conclusions.
3. Proposed system description
The proposed hybrid energy generation system is depicted in
Fig. 1. This system consists of a horizontal axis and variable speed
wind turbine, a solar array, a microturbine, a battery storage,
induction and synchronous generators, indirect space vector con-
trol (ISVC) scheme, sinusoidal pulse width modulation (SPWM)
converters, DC/DC converters, an adaptive feedback linearization
controller, load, dump load and a supervisory controller.
3.1. Wind energy conversion system
Every horizontal axis wind energy conversion system (WECS)
consists of two or three blades, pitch angle control mechanism,
yaw mechanism, gearbox, generator and converters. A squirrel
cage induction generator (SCIG) [33,34], a doubly fed induction
generator [35] and a permanent magnetic synchronous generator
[36] are used in the WECS. The SCIG is chosen to be used in this
system because of its simple and robust mechanical structure. Sta-
tor is connected to the load via converters.
The dynamic modeling of the squirrel cage induction generator
in a stationary dq0 reference frame is demonstrated as following in
[37].
2 3 2 32 3
v qs Rs þ pLs 0 pLm 0 iqs
6 7 6 76 7
6 v ds 7 6 0 Rs þ pLs 0 pLm 76 ids 7
6 7 6 76 7
6 7¼6 76 7 ð1Þ
6 v qr 7 6 pL xr Lm Rr þ pLr xr Lr 76 7
4 5 4 m 54 iqr 5
v dr xr Lm pLm xr Lr Rr þ pLr idr
From stator side, the equations are [37]
kds ¼ Ls ids þ Lm idr
kqs ¼ Ls iqs þ Lm iqr
ð2Þ
Ls ¼ Lls þ Lm
Lr ¼ Llr þ Lm
 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064 3053

VL,ref , fref

VL, fL
Adaptive Controller
Controller

SPWM DC/AC
Wind speed G IG AC/DC DC/DC Inverter
Pbat, ref

Controller

Variable AC Load
Battery Pack DC/DC
Converter

DC/DC
Solar array Converter PDl, ref
& Control
Controller

Ps Pw PL Pbat, est Dump Load Converter

Ac Bus
Supervisory Controller Fuel

PDl, ref PMic,ref Pbat, ref

Combustor

Supervisory Controller
Compressed air

Exhaust gas

Turbine Compressor G S.G.

Fresh air

Fig. 1. Proposed system schematic.

v ds ¼ Rs ids þ dtd kds the DC link voltage to AC output voltage [38]. This inverter is
ð3Þ shown in the Fig. 2. It consists of six insulated gate bipolar-junction
v qs ¼ Rs iqs þ dtd kqs
transistors (IGBT). Each leg has two IGBT. Furthermore, the SPWM
From rotor side, the equations are [37] is used in order to generate the gating pulse for applying to the
gates of the IGBT. The reference signal in the gating circuit for
kdr ¼ Lr idr þ Lm ids
ð4Þ the IGBTs in each leg is displaced by 120° achieved by shifting
kqr ¼ Lr iqr þ Lm iqs the reference signal of the PWM. By comparing a sinusoidal refer-
v dr ¼ Rr idr þ dtd kdr þ xr kqr ence voltage with a triangular waveform, the gating signals are
ð5Þ
v qr ¼ Rr iqr þ dtd kqr  xr kdr
For the air gap flux linkage, the equations are [37]

kdm ¼ Lm ðids þ idr Þ

ð6Þ IGBT1 IGBT3 IGBT5
kqr ¼ Lm ðiqr þ iqs Þ

where Rs, Rr, Lm, Lls, Llr, xr, id, iq, vd, vq, kd and kq are the stator wind-
ing resistor, rotor winding resistor, magnetizing inductance, stator
leakage inductance, rotor leakage inductance, electrical rotor angu- Cdc-link
lar speed, currents, voltages and fluxes of the d–q model, IGBT4 IGBT6 IGBT2
respectively.
Rectifier and inverter used in the WECS are pulse width modu-
lation (PWM) controlled converters. A six-pulse-full-controlled
rectifier is used to convert the AC output of self excited induction
generator (SEIG) into DC voltage. Voltage source inverter (VSI)
using pulse width modulation topology is implemented to convert Fig. 2. Six-pulse inverter topology.
3054 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064

created. In the basic operation of the circuit, if the sinusoidal volt- Power of wind turbine
age is higher than the triangular waveform, the output of the com-
parator will be one; otherwise the output will be zero. Two
modulation indexes, consisting of amplitude modulation index Pmax 4
(ma) and frequency modulation index (mf), are used for the control
of load voltage and frequency [48]. V w1 V w2 V w3 V w4
Pmax 3
D
Vr Pmax 2
ma ¼ ð7Þ
Vc P2 C
Vw 4
fc Pmax 1
mf ¼ ð8Þ P1 B V w3
fr A
Vw1 V w2
where Vr, Vc, fr and fc are reference voltage, carrier voltage, frequen-
cies of reference and carrier signals, respectively. ω1 ω op1 ω op 2 ω op 3 ω op 4 Generator Speed
The output power and torque of turbine (Tt) versus rotational
speed in the WECS are given as follows: Fig. 4. Output power of turbine versus rotor angular speed.
 3
1 R  xopt
Pw ¼ q  A  C p ðk; bÞ ð9Þ
2 kopt
 3 point B by increasing the rotor angular speed to xopt2. If the wind
1 R speed changes from Vw1 to Vw2 while the rotor speed is fixed at
T t ¼ q  A  C p ðk; bÞ  x2opt ð10Þ
2 kopt xopt2, the operating point of system will be C which does not corre-
spond to the maximum power tracking. The rotor speed should be
Furthermore, the output power of a wind turbine versus wind speed increased from xopt2 to xopt3 until to achieve to point D for captur-
is approximated by an equation as follows [28]: ing the maximum power from wind turbine. Therefore, when wind
8 2 2 speed changes, the rotor angular speed of induction generator
> V V cin
< V 2rat V 2cin :P r ; V cin 6 V 6 V rat
> should track these changes in order to extract the maximum power
P w ðVÞ ¼ Pr ; V rat 6 V 6 V cout ð11Þ from the wind turbine.
>
>
: The power coefficient is maximized for an optimum TSR value
0; V 6 V cin and V  V cout
when the blade pitch angle is zero. The optimum rotor speed is gi-
where Pr, Vcin, Vrat and Vcout are the rated power, the cut-in wind ven by the following equations:
speed, the rated wind speed and the cut-off wind speed, respec- kopt
tively. The power coefficient is given according to (12) and is plot- xopt ¼  V wn ð13Þ
R
ted versus tip speed ratio value (k) and blade pitch angle (b) in Fig. 3 R  xopt
[43]. V wn ¼ ð14Þ
kopt
   
151 18:4 where xopt, kopt , R, and Vwn are optimum rotor angular speed (rad/s),
C p ðk; bÞ ¼ 0:53  0:58b  0:002b2:14  13:2 exp 
ki ki optimum tip speed ratio, radius of turbine blade (m) and wind
1 speed (m/s), respectively. For the WECS, power coefficient (Cp) is
ki ¼ 1
k0:02b
 0:003
1þb3
maximized for an optimum TSR value when the blade pitch angle
is zero.
ð12Þ
3.2. Solar array model
The output power of wind turbine versus rotor speed is shown in
Fig. 4 for various wind speeds respectively. If the wind speed is
Each solar array consists of several solar cells converting sun-
Vw1, the rotor angular speed will be x1 and the wind turbine will
light into direct current (DC) power. For a PV module the short-cir-
operate at point A and the power captured using the WECS will
cuit current (Isc) is calculated by [39]
be P1. The output power can be raised to the maximum value at  a
G
Isc ¼ Isc0 ð15Þ
G0
0.5
where Isc0, Isc and a are the short-circuit current of the PV module
under the standard and normal solar irradiance (G0, G); and the
0.4
exponent responsible for all the nonlinear effects, respectively.
The open-circuit voltage (Voc) at any given conditions can be ex-
0.3 pressed by [39]
Cp

 c
0.2 V oc0 T0
V oc ¼   ð16Þ
1 þ b Ln GG0 T
0.1
where Voc and Voc0 are the open-circuit voltage of the PV module
under the normal solar irradiance and the standard solar irradiance;
0 T is the PV module temperature; b and c are a PV module technol-
20 ogy specific related dimensionless coefficient; and the exponent
10 10 15 20 25 30
0 5
TSR considering all the nonlinear temperature–voltage effects, respec-
Beta
tively. Maximum output power of a PV module, Pmax, is written as
Fig. 3. Power coefficient versus tip speed ratio and blade pitch angle. [39]
 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064 3055

Fuel tors. Single- and split-shaft are two types of microturbines. In

the single-shaft microturbine, the turbine and generator are
mounted on the same shaft. Output frequency of the microtur-
Combustor
bine is from about 400 Hz up to several kilo Hertz. It must be
converted to 60 Hz using electronic power converters. In the

Compressed air
Exhaust gas spilt-shaft configuration, utilized in this paper, the turbine shaft
is connected to the generator by a gearbox and the converters
are not needed [40–42]. The GAST model, developed by General
Electric, is one of the most commonly used models to simu-
AC G Compressor Turbine
late a gas turbines [40]. The model and configuration of the
S.G.
Load implemented microturbine are shown in Figs. 5 and 6, res-
pectively.

Fresh air
3.4. Battery storage model

Fig. 5. Configuration of split-shaft microturbine. The battery storage accompanied by the microturbine is used as
a backup for the wind turbine and solar array hybrid system. The
  dynamic battery model used is depicted in Fig. 7. Self discharge
!
V oc
nkTq
V oc
 Ln nkTq þ 0:72  Rs

V oc0
resistance (Rp), internal resistances (Ric, Rid), external resistances
Pmax ¼ V oc
 1    and capacitance (Rco, Rdo, Co), battery capacitance (Cb) are charac-
1 þ nkTq V oc =Isc 1 þ bLn GG0 terized in this figure. The value of each element is a function of
 c  a
T0 G open-circuit voltage (Voc) and state of charge (SOC) [32]. Voc de-
  Isc0 ð17Þ pends on battery discharge current (IB), attracted energy from bat-
T G0
tery (Ecd) and battery temperature (Tb). In this model, remaining
where n is ideality factor (1 < n < 2); k is Boltzmann constant battery energy for determining Voc is calculated using a look-up ta-

(1:38  1023 J= K) and q is the magnitude of electron charge ble. Fig. 8 shows the block diagram for determining the Voc and
19
(1.6  10 C). SOC.
PV modules are connected together in series and in parallel in In this model, the battery’s initial current is used to estimate
order to scale-up the voltage and current. Voltage, current and the maximum available energy. To determine the battery’s
power of a PV array with Ns  Np modules are calculated as below remaining energy and SOC, the energy being drawn from the
IA ¼ N P  IM ð18Þ battery is dynamically calculated and subtracted from the initial
energy.
V A ¼ Ns  V M ð19Þ
PA ¼ NP  Ns  PM ð20Þ
4. Control systems design
where VA, VM, IA, IM, PA and PM are voltages of PV array and module,
currents of PV array and module, and power of PV array and mod- The most important objectives of the control systems used in
ule, respectively. this paper are: (i) capturing the maximum power from the wind
turbine using the ISVC scheme and adaptive feedback lineariza-
3.3. Microturbine model tion method; (ii) controlling the load voltage and frequency
using a SPWM voltage source inverter (SPWM-VSI) control and
A microturbine is a small high-speed gas turbine that gener- (iii) managing energy between generated and consumed energies
ates electrical power within 20–500 kW range. It consists of of the hybrid system components using the supervisory
four components: compressor, combustor, turbine and genera- controller.

ω Dturbine

Ks
- +
+ +
v max
1
ω ref 1 + T1 S L 1 1 -
V + Pm
Pref G
1 + T1 S 1 + T2 S
KF

+ +
vmin
- +
Ki
ΔP s KT -
1
++ +
1 + T3 S

Kk
s

Fig. 6. Used model of microturbine.

3056 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064

Ri Co 4.1.2. Feedback linearization and developed model reference adaptive

controllers design
Rco + The input–output linearization method is used for obtaining a
Rid linear relationship between a new input (l) and the output (y)
Ebat Rdo Vb
Ibat Voc Rp [44]. Fig. 10 shows the driven train of the SEIG. In this figure, Kls,
Bls, Jr, Kr, xr, Jg, Kg and xg are the shaft damping, the shaft stiffness,
the rotor inertia, the rotor external damping, the rotor angular
Fig. 7. Dynamic model of lead acid battery.
seed, the inertia, the external damping and the angular speed of
the generator, respectively. Also, Tae is the aerodynamic torque,
Tls the low speed shaft torque, Ths the high speed shaft torque, Teg
the electromagnetic torque of generator and ng is the gearbox coef-
4.1. Indirect space vector control, fuzzy logic controller, feedback ficient. The dynamic equations for the low speed shaft and the high
linearization, and developed model reference adaptive controller speed shaft are [45]
design
T ls ¼ K ls :ðxr  xls Þ þ Bls :ðhr  hls Þ ð26Þ
4.1.1. Indirect space vector control T hs ¼ ng  T ls ð27Þ
In this section, voltage and current of the stator are controlled
by the stator indirect space vector control based on stator field Relationship between the rotor angular speed (xr), the aerody-
dq0 model. The equations of electromagnetic torque of the induc- namic and the electromagnetic torques can be written as [45]
tion generator are discussed in [38]. dxr
J þ K  xr ¼ T ae  T eg ð28Þ
dt
3P Lm
T eg ¼  ðiqs  kdr  ids  kqr Þ ð21Þ K ¼ K r þ n2g  K g ð29Þ
4 Lr
Assuming l is an additional input and Tae is an aerodynamic estima-
The decoupling control between d and q axis can be obtained by tion of the wind turbine torque; l is given as
imposing the following conditions:
Tb ae  T eg K
d l¼  xr ð30Þ
kdr ¼ kdr ¼ 0 J J
dt ð22Þ
Thus, the generator torque is written
kqr ¼ cte
!
Tb ae
By applying above conditions to induction generator equations [38] T eg ¼ J   l  K xr ð31Þ
J
Lm
kqr ¼  iqs ð23Þ This leads to the following equation:
1 þ Lrrr  S ! !
4 Lr T eg dx Tb ae
ids ¼    ð24Þ J þ K  xr ¼ T ae  J   l  K xr ð32Þ
3P Lm kqr dt J
L r Consequently,
xsl ¼  m  r  ids ð25Þ
Lr kqr
dx T ae  Tb ae Te ae
¼ þl¼ þl ð33Þ
Fig. 9 shows the applied space vector control. Also, current control- dt J J
lers for ids and iqs components were implemented. According to this e ae becomes zero, then the (33) can be written as
If T
figure, the optimum torque (Tae,opt) obtained using adaptive control-
 
ler is used as input of the ISVC. iqs and ids are obtained using the dx
¼l ð34Þ
Tae,opt and the ISVC equations. Furthermore, v ds and v qs are gener- dt
ated by controllers based on these currents. v a , v b and v c are ob-
For the aerodynamic estimation, the optimum rotor angular speed
tained using these voltages by dq0/abc converter based on
is written using (13) as
reference signals for sinusoidal pulse width modulation generator
in order to generate firing pulses for applying to the IGBT gates of kopt
xopt ¼ n v wn ð35Þ
rectifier. R

IB VB t

Ecd = ∫ V .I .dt

Ecd
E max Emax E max,cor Emax,cor Voc Voc Controlled
IB E L = E max − Ecd EL Voltage
I T EL Source
Look- up table

Tb EL
SOC =
Emax,cor E max,cor

Fig. 8. Block diagram for determining Voc and SOC.

 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064 3057

SPWM
IG
AC/DC
θr
d/dt ia
ib iqs
P/2
Lr abc/ dq0
rr ic ids
SPWM

e − jθ e
Rotor flux Generator
ωr angle θe
*
ids
iqs iqs ωs

ωs .σ .Ls .iqs

*
Tem, opt −4 Lr Te ids *
+
. . + vds
3 Lm λqr
_ + dq0
*
iqs * to * *
+ vqs Va Vb V*
abc c
_ +
λqr Lm iqs +
.iqs
L
1 + r .s
rr ωs .( Lm .ids + σ . Ls ) .ids

ωs ids
− Lm rr
. .ids
Lr λqr ωs

Fig. 9. Space vector control scheme implemented in SEIG.

Tls
Ths
Teg

Wind speed
Bls , Kls ng
Tae
Jg , g , Kg

Jr, r, Kr

Low Speed Side High Speed Side

Gearbox

Fig. 10. Driven train of the SEIG.

If e is the tracking error eðtÞ ¼ HðpÞ  ½K  /T ðtÞ  lðtÞ ð39Þ

e ¼ xopt  x
^r ð36Þ where e is the scalar error, H is the strictly positive real transfer
function and / is the vector function of time. The variation of /(t)
The dynamic error at the closed loop system can be expressed with is defined as [44,46]
the following equation
d/ðtÞ
¼ g  e  lðtÞ  sgn ð40Þ
de dt
þ a0 e ¼ 0 ð37Þ
dt g is a positive constant. Assuming l is bounded
Using (34), (36), and (37) are obtained: If t ! 1 then eðtÞ ! 0
dxopt According to the previous equations, the following adaptive relation
l¼ þ a0  e ð38Þ b ae is suggested
for T
dt
ge
a0 is the designed parameter. According to the model reference Tb ae ¼ ð41Þ
adaptive system (MRAS) [46] J
3058 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064

Then the optimum torque (Tae,opt) is obtained

T eg;opt ¼ Tb ae  J  l  K  xr ð42Þ

Configuration of the adaptive feedback linearization controller is

shown in Fig. 11. It is utilized to design the linear controller. It
causes the linear PID controller to encounter with a linear plant in-
stead of a nonlinear one. Thus, the design of the PID controller is ful-
filled using Ziegler–Nichols method based on (38). The performance
of adaptive controller is equal to (41). This controller generates the
estimated torque ( Tb ae ) based on error between the rotor angular
speed and optimum angular speed (xopt  xr). The feedback linear-
Fig. 12. Seven linguistic labels for Pout and xr .
ization makes a linear relationship between input and output of the
plant according to (42).

4.1.3. Fuzzy logic controllers design

Fuzzy logic controller does not require a mathematical model to
control systems. It includes four parts consisted of fuzzification,
rule base, inference engine (reasoning), and defuzzification dis-
cussed in [47,48]. Fuzzification is a process that converts input/
output variables of the system to linguistic labels. Seven linguistic
labels for Pout and xr are NB (Negative Big), NM (Negative Middle),
NS (Negative Small), ZE (Zero), PS (Positive Small), PM (Positive
middle), PB (Positive Big), respectively. Three linguistic labels for
xr are N (Negative), ZE (Zero), and P (Positive), respectively. These
classes of linguistic labels are characterized by membership func- Fig. 13. Three linguistic labels for xr.
tion shown in Figs. 12 and 13, respectively.
Inference engine adjusts and determines the duty and behavior
of the controller based on data and rules related to the input and
output variables of the systems. These rules are stated in order
to extract maximum power discussed in Section 3.1 based on DPout Fuzzy sets & Rule base
and Dxr. For example If ((DPout is NM) AND (Dxr is N)), then (the _
+ Δωr
controller output is PM. ωref *
ωr
Fuzzy Δ ωr
Defuzzification is a process that converts the linguistic labels + Fuzzifier
inference Deuzzifier WECS
Pref Pout
into crisp solution variables. The most important defuzzification ΔPo
_
methods used in several control application are Center of Area
(COA) and Mean of Maxima (MOM). MOM is applied in this paper
as a defuzzifier. Fig. 14 shows the FC controller in the WECS. Fig. 14. Implemented fuzzy controller in the WECS.

4.2. Supervisory controller

To manage the maximum energy captured from the WECS and discussed in Section 3.4, battery reference power, dump load refer-
solar array, a supervisory controller is designed. After adjusting the ence power, microturbine reference power and excess/deficiency
WECS for the maximum energy generation using the adaptive power calculated based on subtraction between generated and
feedback linearization controller, the supervisory controller calcu- consumed energies, respectively.
lates the reference powers for the battery regulator controller, All steps of the supervisory controller operation in the flow
microturbine controller and dump load controller based on sub- chart are explained as below
traction of the load power from sum of the wind turbine and solar
array powers (Pe = Pw + Ps  PL) according to the flow chart shown Step 1: Compare sum of the wind turbine and solar array gen-
in Fig. 15. erated powers with the load power (Pe = Pw + Ps  PL), if
In the flowchart; Pw, Ps, Pb,est, Pb,ref, Pdl,ref, PMT,ref and Pe are the Pe is positive, go to step 2, otherwise go to step 7.
maximum power extracted from the wind turbine, PV array power, Step 2: The battery is in charge mode. Rapid charge is used for
estimated battery power calculated dynamically using the method charging the battery so that it is charged at an hour. In
this step, Pe is positive, this means that generated sum
of wind turbine and solar array powers is more than
ω opt load power; if the estimation of battery power (Pb,est)
Linearized system
is equal to the maximum charge power go to step 3,
+ Feedback
Nonlinear Plant otherwise go to step 5.
e Controller linearization Teg ,op ISCV Teg Step 3: In this step, Pe is positive, if the excess power is less
(Eq. 38) μ system
_ SEIG than the maximum charge power, apply Pb,ref = Pe,
(Eq. 42)
ωr Model otherwise calculate DP2 = Pe  100 and go to step 4.
Step 4: If DP2 = 0 apply Pb,ref = 100 kW, otherwise apply Pb,ref =
T̂ae Adaptation law T̂ae
(Eq. 41) 100 kW and Pdl,ref = DP2.
ωr Step 5: In this step, Pe is positive, if the excess power is more
than the maximum charge power apply Pb,ref = Pb,est
Fig. 11. Configuration of the adaptive feedback linearization controller. and Pdl,ref = Pe  Pb,est, otherwise go to step 6.
 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064 3059

t = 0, Δt = 1s, t k

Receive : Pw , Ps , Pb,est , Pload

Pe = Pw + Ps − PL

y
Pe = 0

N
Pw + Ps = Pb y
Pe > 0

Go to 1 N y
Pb,est = 100kW
y Pb,est < Pe
N
N
y
Pe ≥ 100 Pe ≥ 100

Pe ≥ 100
N
Pb ,ref = Pb ,est Pb ,ref = Pe
y Pb,ref = Pe Pb,ref = Pb,est N N Pe ≥ 100
N PDL ,ref = Pe − Pb ,est PDL ,ref = 0
PDL, ref = 0 P
DL , ref = Pe − Pbat ,est
PMT ,ref = 0 PMTref = 0 y
PMT , ref = 0 P
MT , ref = 0

ΔP1 = Pe − 100 y Pb,est < Pe ΔP2 = Pe − 100

Pb,est < Pe y
Go to 1 Go to 1
N N
y
ΔP1 = 0 Pb, ref = Pe Pb,ref = Pe Pb, ref = 100 y
PDL,re f = 0 ΔP2 = 0
PDL, ref = 0 PDL , ref = 0
N
PMT ,ref = 0 PMT , ref = 0 PMT ,ref = 0
Pb,ref = −100 Pb,ref = −100 N
PMT , ref = 0 PMT , ref = ΔP1 From1
Pb,ref = 100
PDL, ref = 0 PDL, ref = 0 Go to 1
t = t + Δt
Go to 1 PDL, ref = ΔP2
PMT , ref = 0
Go to 1 Go to 1
t > tk
N Go to 1
y

END

Fig. 15. Supervisory controller operation flow chart.

Step 6: If Pb,est < Pe, go to step 5, otherwise apply Pb,ref = Pe.
Step 7: In this step, Pe is negative, this means that the wind/solar
generated powers is less than load power; if the battery
discharge power estimation is less than the power defi-
ciency, go to step 8, otherwise go to step 10.
Step 8: In this step, Pe is negative, if the absolute value of the
power deficiency is less than the absolute value of
the maximum discharge power (100 kW), apply
Pb,ref = Pe, otherwise calculate DP1 = |Pe|  100 and go
to step 9.
Step 9: If DP1 is zero, apply Pb,ref = 100 kW, otherwise apply
Pb,ref = 100 kW and PMT,ref = DP1.
Step 10: In this step, Pe is negative, if the absolute value of the
power deficiency is more than the absolute value of
the maximum discharge power apply Pb,ref = Pb,est and
PMT,ref = |Pe|  |Pb,est|, otherwise go to step 11.
Step 11: If the absolute value of the battery estimated power is
less than the absolute value of Pe, go to step 10, other-
wise apply Pb,ref = Pe.
utilized to satisfy the requirements of given load demands base
on economical analysis. For economical analysis and optimal de-
sign of the hybrid system, total annualized cost of system (ACS)
composed of annualized capital cost (Cacap), annualized replace-
ment cost (Carep), annualized operation and maintenance cost
(Cao&m), and annualized fuel cost (Cafuel) for five main parts, consist-
ing of wind turbine, microturbine (MT), PV array, battery storage
and the other devices included controller, inverter, rectifier and
chopper, are considered [28].
ACS ¼ C acap ðwind þ PV þ MT þ Battery þ othersÞ
þ C arep ðbatteryÞ þ C aom ðwind þ PV þ MT þ Battery
þ othersÞ þ C afuel ðMTÞ ð43Þ
5.1. The annualized capital cost
The annualized capital cost of each component is defined as the
following equation [28]:

C acap ¼ C cap  CRFði; Y proj Þ ð44Þ

5. Optimal sizing and economic analysis
Ccap, Yproj and CRF are the initial capital cost of each component, the
The main purpose of this section is to find optimal sizing of a component lifetime, and capital recovery factor respectively. CRF is
stand-alone wind/microturbine/PV array/battery storage system determined as the following equation [28]:
3060 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064

i  ð1 þ iÞY proj Pw þ PMT þ PPV ¼ PL;max ð58Þ

CRFði; Y proj Þ ¼ ð45Þ
ð1 þ iÞY proj  1 0 6 P w 6 PL;max ð59Þ
0 0 6 P PV 6 PL;max ð60Þ
i is the annual real interest rate related to nominal interested rate, i ,
and the annual inflation rate, f, as follow 0 6 P MT 6 PL;max ð61Þ
0
i f Among several techniques and methods used to achieve the optimal
i¼ ð46Þ configurations of hybrid systems, GA method is implemented be-
1þf
cause of its robust in finding global optimal solutions and particu-
larly in multimodal optimization problems. The GA is categorized
5.2. Annualized replacement cost as global search heuristics that use techniques inspired by evolu-
tionary biology such as inheritance, mutation, selection, and cross-
The annualized replacement cost (ARC) is defined only for bat- over. Two-point crossover, roulette wheel selection methods, a
tery storage replaced periodically during the project lifetime. The method with mutation probability of 0.002 have been allocated to
ARC is calculated as [28] each chromosome. The considered chromosome has been shown
C arep ¼ C rep  SFFði; Y proj Þ ð47Þ in Fig. 16.

where SFF is the sinking fund factor and is stated as [28]

6. Simulation results and discussion
i
SFFði; Y proj Þ ¼ ð48Þ
ð1 þ iÞY proj  1 MATLAB Simulink™ 7.2 is used to evaluate the performance of
the proposed hybrid power generation system and controllers. The
mathematical model and design details of each component in the
5.3. Annualized fuel cost hybrid power generation system are presented in the previous sec-
tions. For the energy conversion system utilized in this paper, pitch
The annualized fuel cost function, Cafuel(MT), is just used for angle is assumed zero and yaw control mechanism is not consid-
microturbine. To obtain the cost function of a microturbine, fuel ered. It consists of an aerodynamic system based on wind speed
cost curve versus output power is modeled as the following model, wind power versus wind speed model and etc., induction
equation: generator (SEIG), a SPWM AC/DC converter, a DC/DC buck-boost
C afuel ¼ a þ b  Pm þ c  P 2m ð49Þ converter, a DC/AC inverter, AC filters, adaptive, PID and ISVC con-
trollers. The specifications of the Vestas™ wind turbine (V-25) and
If the points of fuel cost curve versus output power are clarified, the the SPWM AC/DC converter of the WECS are shown in Tables 1
a, b, c will be determined using least square estimation algorithms and 2, respectively. The DC/DC converter is used in order to fix
as follows: the DC link voltage for the inverter. The specifications of this chop-
X
k per are given in Table 3.
J¼ ða þ b  Pmj þ c  P 2mj  C afuel Þ2 ð50Þ The PV array, employed in the hybrid system, consists of PV
j¼1 modules and buck-boost DC–DC converter. Specifications of the
X
k PV array and the DC/DC converter are given in Tables 4 and 5,
@J=@ a ¼ 2  ða þ b  P mj þ c  P2mj  C afuel Þ ¼ 0 ð51Þ respectively.
j¼1 The microturbine is required when the hybrid system of wind
X
k turbine, solar array and battery storage cannot satisfy the load de-
@J=@b ¼ 2  P mj  ða þ b  Pmj þ c  P 2mj  C afuel Þ ¼ 0 ð52Þ mands. It consists of compressor, combustor, turbine and genera-
j¼1
tor. Mathematical model and the specifications of the
X
k
synchronous generator used for microturbine are presented in
@J=@ c ¼ 2  P 2mj  ða þ b  Pmj þ c  P 2mj  C afuel Þ ¼ 0 ð53Þ
Fig. 6 and Table 6, respectively [40].
j¼1
Accompanied by the microturbine, a lead acid battery and a
Thus a system of algebraic equation versus a, b and c is obtained as boost DC/DC converter are used. A boost converter and a controller
follows: are required in order to control the voltage of battery bus based on
the battery reference power value (Pb,ref). The parameters of this
X
k X
k X
k converter and controller have been shown in Table 7. In the con-
kaþb Pmj þ c  P2mj ¼ C afuel ð54Þ verter, the IGBT is controlled using a proportional and integral
j¼1 j¼1 j¼1
(PI) controller. This controller alters the duty cycles of pulses ap-
X
k X
k X
k
plied to the chopper IGBT gates. The battery capacity and charge
a Pmj þ b  P2mj þ c  P3mj ¼ Pmj  C afuel ð55Þ
efficiency of the charger are 2.14 kAh and 85%, respectively.
j¼1 j¼1 j¼1
The dump load is used to consume the excess power. It consists
X
k X
k X
k
of three phase resistors connected in series with gate turned off
a P2mj þ b  P3mj þ c  P4mj ¼ P2mj  C afuel ð56Þ
j¼1 j¼1 j¼1
(GTO) switches. The power consumed by the resistors of the dump
load is discussed in [49].
Optimal sizing based on economical analysis, discussed in Sec-
5.4. Annualized operation and maintenance cost tion 5, is accomplished using the GA for the maximum load de-
mands (300 kW). The cost and lifetime of each component is
The maintenance cost of the nth year Cao&m(n) is calculated shown in Table 8. In order to obtain the annualized cost of micro-
based on the maintenance cost of the first year Cao&m(1) as [28]

C ao&m ðnÞ ¼ C ao&m ð1Þ  ð1 þ f Þn ð57Þ

Pw Ps PMT Ahb
For minimization of the ACS objective function, the following
constraints must be imposed: Fig. 16. Propounded chromosome for minimization of the ACS function.
 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064 3061

Table 1 Table 6
Vestas 200 kW wind turbine generator, Type V-25-200. Synchronous specifications of microturbine.

Generator type Squirrel cage Rated power 250 kW

Rated power 200 kW Rated line to line voltage 660 V
Nominal power 200 kW Frequency 60 Hz
Nominal voltage (line-to-line) 480 V Inertia constant (H) 0.822 s
Nominal frequency 60 Hz Friction factor 0
Mounted capacitor 75 kVAr
Stator resistance 0.01615 p.u.
Stator inductance 0.05258 p.u.
Table 7
Rotor resistance 0.01515 p.u.
DC/DC converter specifications of the battery storage.
Rotor induction 0.05258 p.u.
Mutual induction(Lm) 2.95 p.u. Chopper type Boost
Inertia constant 1.2 s Semiconductor switch type IGBT
Pairs of poles 2 Converter inductor 75 mH
Nacelle mass 7.1 kg Converter capacitor 2500 lF
Rotor mass 3.2 kg Switching frequency 2 kHz
Hub height 30 m Proportional gain of voltage control system 0.045
Highest point 41.5 m Integral gain of voltage control system 0.065
Cut-in wind speed 3.8 m/s PWM reference voltage 440 V
Cut-off wind speed 25 m/s
Rated wind speed 13.8 m/s
Survival wind speed 56 m/s
solar array, lead acid battery and microturbine are 195 kW,
85 kW, 2.14 kAh and 230 kW, respectively. The total cost of this
hybrid system based on the economical analysis becomes
Table 2 215,580 $.
The SPWM AC/DC converter specifications for the WECS.
There are three models in wind speed modeling. For wind speed
Semiconductor switch type IGBT/DIODE modeling turbulent guest based on Markov chain model presented
Snubber resistance 5 kX by Kantz et al. [50] is used in this paper. The wind speed variations
Snubber capacitor Inf.
Internal resistance 1 mX
are shown in Fig. 17. It changes approximately between 3.6 m/s
Carrier frequency 5 kHz and 17 m/s.
In order to compare operation of the proposed adaptive control-
ler with others, a fuzzy logic controller is designed discussed in
Table 3
Section 4.1.3. The classes of linguistic labels are characterized by
DC/DC converter specifications of the WECS. membership function shown in Figs. 12 and 13, respectively.
MOM is applied in this paper as a defuzzifier method. Control rules
Chopper type Buck-boost
Semiconductor switch type IGBT
of the FC are given in Table 10. Maximum power extracted from
Converter inductor 85 mH the WECS using proposed adaptive controller, designed fuzzy logic
Converter capacitor 1800 lF controller and PID controller is shown in Fig. 18. The performance
Switching frequency 3 kHz of controllers is considered during 50 s because the amount of cal-
Proportional gain of voltage control system 4
culations in the FC controller is very high. This figure shows that
Integral gain of voltage control system 12.5
PWM reference voltage 440 V the operation of Lyapunov adaptive controller is better than the
FC and PID controllers. In some intervals performance of the FC
controller is similar to the adaptive one, but the FC control does
not guarantee the optimal response. Because of nonlinear behavior
Table 4
PV array specifications.
of the plant, the PID controller can not track the maximum power
point.
Maximum power 85 kW The power generated by the WECS is shown in Fig. 19. Fig. 20
Open-circuit voltage (V oc ) 345 V
Short-circuit current (Isc ) 445 A
indicates the power attraction from the solar array. At first, the
Operating voltage at maximum power (V max ) 200 V main load power is 25 kW. At t = 370 min, the load power in-
Operating current at maximum power (Imax ) 425 A creased to 225 kW using a three phase breaker and then at t =
1070 min, the load power decreased up to 125 kW. The load profile
and sum of the powers extracted from the wind turbine and solar
Table 5 array are shown in Figs. 21 and 22, respectively.
DC/DC converter specifications for the solar array. According to sum of the wind turbine/PV array generated pow-
Chopper type Boost
ers and the load power profiles, from t = 0 to t = 370 min, sum of
Semiconductor switch type IGBT the wind turbine/solar array generated powers (Pw + Ps) is more
Converter inductor 75 mH than the load power (25 kW) and the battery will be in charging
Converter capacitor 2500 lF mode. However, in some subintervals of this duration, the excess
Switching frequency 3 kHz
power is more than the maximum charge power of the battery.
Proportional gain of voltage control system 0.46
Integral gain of voltage control system 0.78 In this case, the battery is charged based on the maximum charge
PWM reference voltage 440 V power, and remainder surplus power of generated by this system is
consumed by the dump load in order to prevent overcharging of
the battery. So, the dump load operates as a protector for the
battery.
turbine, the least square estimation algorithm was applied. For cal- From t = 370 to t = 1070 min, sum of the wind turbine/solar ar-
culation of a, b and c, the fuel flow input of some microturbines ray generated powers (Pw + Ps) is less than the load power
produced by Capstone™ company is applied given in Table 9. (225 kW) and the battery will be in discharge mode. However, in
The results of this analysis indicate that the size of wind turbine, some subintervals of this duration, the power deficiency is less
3062 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064

Table 8
The costs and lifetime of each component.

Initial capital cost Replacement cost Fuel cost Maintenance cost in the first year Lifetime (year)
Wind turbine 3500 S$/kW Null Null 90 US$/kW 20
Microturbine 850 S$/kW Null 0.0037 S$/kW 85 US$/kW 20
PV array 6000 S$/kW Null Null 60 US$/kW 20
Battery storage 1500 $/kAh 1500 S$/kAh Null 50 US$/kAh 5
Others 8000 US$ Null Null 80 US$ 20

Table 9

Maximum extracted from wind turbine (kW)

Characteristics of fuel flow input versus output power. 150
Adaptive
Fuel flow input (MBTU/h) Operational power of microturbines Fuzzy
0.433 25 PID
0.842 60
2.28 190 100
6.84 500

18
50
16

14
Wind speed (m/s)

12 0
0 5 10 15 20 25 30 35 40
10 Time (s)

8 Fig. 18. Maximum power extracted from the WECS using proposed adaptive, fuzzy
and PID controllers.
6

4
Maximume power extracted from WECS (kW)

200
0 500 1000 1500
Time (min.)

Fig. 17. Wind speed variations.

150

Table 10
Implemented fuzzy logic rules. 100

DPout Dxr
P ZE N 50
NB NB NB PB
NM NM NB PM
NS NS NM PS
ZE ZE ZE ZE 0
0 500 1000 1500
PS PS PM NS
PM PM PB NM Time (min.)
PB PB PB NB
Fig. 19. Maximum power extracted from the WECS.

than the maximum discharge power. In this case, the battery is dis-
charged based on the maximum discharge power, and unmet re-
minder power of the load demand is satisfied by the 100
microturbine. Furthermore, the supervisory controller adjusts the
Extracted solar power (kW)

reference powers of the microturbine and dump load based on ex- 80

cess and deficiency powers.
The performance of the controllers and hybrid system in the
60
other subintervals is similar to the considered durations.
The battery power, dump load/microturbine powers discussed
in this section are shown in Figs. 23 and 24, respectively. 40
A DC/AC inverter is utilized to adjust the load voltage and fre-
quency to 230 V and 60 Hz to satisfy the required load power using 20
the SPWM method. The switching frequency of the inverter is
2 kHz. In order to eliminate the switching harmonics, a 5 mH
inductive low output passive filter is applied. Specifications of 0
0 500 1000 1500
the SPWM inverter are given in Table 11. The line-to-line output
Time (min.)
voltage amplitude in the SPWM inverter is determined as the fol-
lowing equation [51]: Fig. 20. Solar power extracted from the PV array.
 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064 3063

300 150
PMT

Microturbine/dumpload powers (kW)

PDL
250
Load power (kW)

200 100

150

100 50

50

0 0
0 500 1000 1500 0 500 1000 1500
Time (min.) Time (min.)

Fig. 21. Load power changes. Fig. 24. Microturbind/dump load powers.

250
Table 11
Sum of wind and solar powers (kW)

SPWM DC/AC inverter specifications.

200
Semiconductor switch type IGBT/DIODE
Snubber resistance 6.5 kX
150 Snubber capacitor Inf.
Internal resistance 1 mX
Carrier frequency 2 kHz
100 Modulation index 0.85
Frequency of output voltage 60 Hz

50

0
0 500 1000 1500 61
Time (min.)

Fig. 22. Sum of wind and PV array power.

60.5
Load frequency (Hz)

150
60

100
Battery power (kW)

59.5
50

0
59
0 200 400 600 800 1000 1200 1400
-50 Time (min.)

-100 Fig. 25. Load frequency variations.

-150
0 500 1000 1500
Time (min.) 1.02

Fig. 23. Battery power during charge and discharge time. 1.015
pffiffiffi
Output voltage (p.u.)

3 1.01
V LL ¼ pffiffiffi  ma  V DC-Link ð62Þ
2 2 1.005

VDC-Link is the DC link voltage and ma is the amplitude modulation in- 1

dex. According to the (62), since the amplitude modulation index is 0.995
0.85 and the DC link voltage is adjusted to 440 V by the DC/DC con-
verter, the obtained line to line AC voltage of the inverter for satisfy- 0.99
ing the load demands is approximately 230 V. The profiles of root 0.985
mean square (RMS) value of the load voltage in per unit (p.u.) system
and load frequency (Hz) are evident in Figs. 25 and 26, respectively. 0.98
0 500 1000 1500
According to these figures, the variation of the RMS value of the load
Time (min.)
voltage is less than 0.7% and the variation of the RMS value of the
load frequency is less than 0.8% that are acceptable. Fig. 26. Output voltage of the load.
3064 M. Kalantar, S.M. Mousavi G. / Applied Energy 87 (2010) 3051–3064
7. Conclusions
In this paper, the integration of a wind turbine, a solar array, a
microturbine, and a battery storage which work together in order
to provide sufficient electric energy supply to satisfy the stand-
alone load demands at all times was proposed and considered.
Optimal sizing and economical analysis of the proposed hybrid
system were accomplished using GA for minimizing the annual-
ized cost of system. A model of power-consumption for a micro-
turbine was propounded and obtained using least square
estimation algorithm based on Capstone™ company data and
was suggested for implementing at economic analysis. Due to less
environmental pollution emission, less noise generation, less
power fluctuation, more reliability and more fuel flexibility in
comparison with a diesel generator a microturbine was pro-
pounded as a back up for the hybrid system. A model reference
adaptive controller has been designed and synthesized with feed-
back linearization technique and the ISVC for the maximum
power extraction from the wind energy conversion system. Com-
parison of proposed method with fuzzy logic and PID controllers
indicates that operation of the proposed adaptive controller is the
best. In order to control the load voltage and frequency, a sinusoi-
dal pulse width modulation method was used. For achieving the
proper operation of the supervisory controller to adjust the refer-
ence power values of controllers correctly, the open-circuit volt-
age estimation method for estimating the value of the SOC has
been implemented. The supervisory controller has been designed
for the management of energy between generated and consumed
energies. The proposed system can be used in the isolated rural
and mountainous areas far from the power generation networks.
References
[1] Urban Frauke, Benders Rene MJ, Mol Henri C. Energy for rural India. Appl
Energy 2009.
[2] Ackermann Thomas. Wind power in power system. John Wiley & Sons Press;
2005.
[3] Weigt Hannes. Germany’s wind energy: the potential for fossil capacity
replacement and cost saving. Appl Energy 2009:1857–63.
[4] Snyder Brian, Kaiser Mark J. A comparison of offshore wind power
development in Europe and the US: patterns and drivers of development.
Appl Energy 2009:1845–56.
[5] Tabesh Ahmadreza, Iravani Reza. Small signal dynamic model and analysis of a
fixed-speed wind farms – a frequency response approach. IEEE Trans Power
Deliv 2006;21(2). Apr..
[6] Krichen Lotfi, Francois Bruno, Ouali Abderrazak. A fuzzy logic supervisory for
active and reactive power control of a fixed speed wind energy conversion
system. Electric Power Syst Res 2007.
[7] Kamoji MA, Kedare SB, Prabhu SV. Experimental investigations on single stage
modified Savonius rotor. Appl Energy 2009:1064–73.
[8] De Almeida Rogerio G, Castronuovo Edgardo D, Lopes JA. Optimum generation
control in wind parks when carrying out system operator request. IEEE Trans
Power Syst 2006.
[9] Tapia A, Tapia G, Ostolaza JX. Reactive power control of wind farms for voltage
control applications. Renew Energy 2004.
[10] Thiriger Torbojorn, Linders J. Control by variable rotor speed of a fixed pitch
wind turbine operation in a wide speed range. IEEE Trans Energy Convers
1993;8(3):520–6.
[11] Nagai Baku M, Ameku Kazumasa, Roy Jitendro Nath. Performance of a 3 kW
wind turbine generator with variable pitch control system. Appl Energy
2009:1774–82.
[12] Simoes Marcelo Godoy, Bose Bimal K, Spiegel Roland J. Fuzzy logic based
intelligent control of a variable speed cage machine wind generation system.
IEEE Trans Power Electron 1997;12(1).
[13] Hilloowala Robin M. Rule based fuzzy logic control for a PWM inverter in a
stand-alone wind energy conversion scheme. IEEE Trans Ind Appl 1996:57–65.
[14] Chedid Zriad B, Karaki Sami H, El-Chamali Chadi. Adaptive fuzzy control for
wind-diesel weak power system. IEEE Trans Energy Convers 2000;15(1).
[15] Tanaka T, Toumiya T. Output control by hill-climbing method for a small wind
power generation system. Renew Energy 1997;12(4):389–400.
[16] Tina G, Gagliano S, Raiti S. Hybrid solar/wind power system probabilistic
modeling for long term performance assessment. Solar Energy 2006;80:
578–88.
[17] Kalantar M, Tripathy SC, Balasubramanian R. Dynamic and stability of wind
and diesel turbine generators with superconducting magnetic energy storage
unit on an isolated power system. IEEE Trans Energy Convers 1991:579–85.
[18] Elhadidy MA. Performance evaluation of hybrid (wind/solar/diesel) power
system. Renew Energy 2002:401–13.
[19] http://www.diesel-generator-engine.com/pros-and-cons.html.
[20] Pang S, Farrell J, Du J, Barth M. Battery state-of-charge estimation. IEEE Am
Control Conf 2001;2:1644–9.
[21] Plett GL. Extended Kalman filtering for battery management systems of LiPB-
based HEV battery pack, part 2, modeling and identification. J Power Sources
2004;104:262–76.
[22] Saheb-Koussa D, Haddadi M, Belhamel M. Economic and technical study of a
hybrid system (wind–photovoltaic–diesel) for rural electrification in Algeria.
Appl Energy 2009:1024–30.
[23] Ekren Orhan, Ekren Banu Y, Ozerdem Baris. Break-even analysis and size
optimization of a PV/wind hybrid energy conversion system with battery
storage – a case study. Appl Energy 2009:1043–54.
[24] Yang Hongxing, Zhou Wei, Lu Lin, Fang Zhaohong. Optimal sizing method for
stand-alone hybrid solar–wind system with LPSP technology by using genetic
algorithm. Solar Energy 2008:354–67.
[25] Ekren Banu Y, Ekren Orhan. Simulation based size optimization of a PV/wind
hybrid energy conversion system with battery storage under various load and
auxiliary energy conditions. Appl Energy 2009:1387–94.
[26] Saheb-Koussa D, Haddadi M, Belhamel M. Economic and technical study of a
hybrid system (wind–photovoltaic–diesel) for rural electrification in Algeria.
Appl Energy 2009:1024–30.
[27] Roy Anindita, Kedare Shireesh B, Bandyopadhyay Santanu. Application of
design space methodology for optimum sizing of wind-battery systems. Appl
Energy 2009.
[28] Yang Hongxing, Zhou Wei, Chengzhi Lou. Optimal design and techno-
economic analysis of a hybrid solar–wind power generation system. Appl
Energy 2009:163–9.
[29] Ekren Orhan, Ekren Banu Yetkin. Size optimization of a PV/wind hybrid energy
conversion system with battery storage using response surface methodology.
Appl Energy 2008:1086–101.
[30] Diaf S, Notton G, Belhamel M, Haddadi M, Louche A. Design and techno-
economical optimization for hybrid PV/wind system under various
meteorological conditions. Appl Energy 2008:968–87.
[31] Alliant energy: Capstone™ MicroTurbine, <http://www.alliantenergy.com>.
[32] Durr Matthias, Cruden Andrew, Gair Sinclair, McDonald JR. Dynamic model of
a lead acid battery for use in a domestic fuel cell system. J Power Sources
2006:1400–11.
[33] Singh SP. Self excited induction generator research – a survey. Electric Power
Syst Res 2004:107–14.
[34] Ahshan R, Iqbal MT, Mann George KI. Controller for a small induction-
generator based wind turbine. Appl Energy 2008:218–27.
[35] Lei Yazhou, Mullane Alan, Lightbody Gordon, Yacamini Robert. Modeling of the
wind turbine with a doubly fed induction generator for grid integration
studies. IEEE Trans Energy Convers 2006;21(1):257–64.
[36] Chan TF, Lai LL. An axial-flux permanent-magnet synchronous energy for a
direct-coupled wind turbine system. IEEE Trans Energy Convers 2007;22(1).
[37] Ong Chee Mun. Modeling and dynamic simulation of electric machinery using
Matlab/Simulink. Prentice Hall Press; 1998.
[38] Bose Bimal K. Modern power electronics and AC drives. Prentice Hall Press;
2002.
[39] Zhou W, Yang XH, Fang ZH. A novel model for photovoltaic array performance
prediction. Appl Energy 2007:1187–98.
[40] Li-Shrkht MY, Sisworahardjo NS, Uzunoglu M, Onar O, Alam MS. Dynamic
behavior of PEM fuel cell and microturbine power plant. J Power Sources
2007:315–21.
[41] Nikkhajoei Hassan, Reza Iravani M. A matrix converter based microturbine
distributed generation system. IEEE Trans Power Deliv 2005;20(3).
[42] Juardo F, Saenz JR. Adaptive control of a fuel cell-microturbine hybrid power
plant. IEEE Trans Energy Convers 2003;18(2).
[43] Brahmi Jemaa, Krichen Lotfi, Ouali Abderrazak. A comparative study between
three sensor less control strategies for PMSG in wind energy conversion
system. Appl Energy 2009:1565–73.
[44] Nam Kwangghee, Arapostathis Aristol. A model reference adaptive control
scheme for pure-feedback nonlinear systems. IEEE Trans Autom Control
1988;33(9).
[45] Boukhezzar B, Lupu L, Siguerdidjane H, Hand M. Multivariable control strategy
for variable speed, variable pitch wind turbine. Renew Energy 2007:1273–87.
[46] Slotine E. Wiping, applied nonlinear control. Prentice Hall Press; 1991.
[47] Mousavi G SM, Kalantar M. Voltage and frequency control of a load connected
to a variable speed wind turbine using fuzzy logic controller. MSc Thesis,
August 2004.
[48] Xinyan Zhang, Xiaobo Zhang, Peiyi Zhou, Jing Cheng, Ye Liu. Fuzzy control used
in variable speed wind turbine. In: International conference on automation
and logistics; 2009. p. 1194–8.
[49] Sebastian R, Quesada J. Distributed control system for frequency control in an
isolated wind system. Renew Energy 2006:285–305.
[50] Kantz Holger, Holstein Detlef, Ragwitz Mario, Vitanov Nikoly K. Markov chain
model for turbulent wind speed data. Physica A 2004:315–21.
[51] Mohan Ned, Undeland Tore M, Robbins William P. Power electronics:
converters applications and design. John Wiley& Sons Press; 1995.