Hardware in the Loop Wind Turbine Emulator

Nicolae Muntean, Lucian Tutelea, Diana Petrila, Ovidiu Pelan

Department of Electrical Engineering
University Politehnica
7LPLRDUD5RPDQLD
diana.petrila@et.upt.ro

Abstract- 7KHSDSHUSUHVHQWVDKDUGZDUHLQ WKHORRSHPXODWRU

for a wind turbine system, used to tune and test in the laboratory 




the generator and the associate power electronics and control. The
emulator includes: a Matlab/Simulink model of the wind turbine,
an a.c. drive with direct torque control (the wind turbine 
 
equivalent) coupled with the real generator and the corresponding    

load. Digital simulations compared with experimental results are
presented to validate the wind turbine emulator.





I. INTRODUCTION
Over the last years the wind systems became the fastest
developing renewable energy technology. The control of the
wind turbines becomes more and more important as the power
level of the turbines increases. Power electronic plays an
important role in wind power systems; they have to integrate
the variable-speed wind power generation units to achieve high
efficiency and high performance in power systems [1].
It was observed that the wind turbines behavior is
significantly affected by the control strategy used. Typically,
aerodynamic loop and power electronics are used to control
torque, speed and power [2].
The main control objective of variable speed wind turbines is
power extraction maximization. To operate at its maximum
power point, the tip-speed ratio should be maintained at its
optimum value despite wind variations [3].
To study wind turbine systems, it is necessary to build a real
time emulator, because a wind turbine may not always be
available and neither a wind tunnel test. The emulator is able to
produce the same behavior as real wind turbine can produce, in
a real environment.
The basic requirement for a wind simulator is that its static
and dynamic characteristics must be as close as possible to
those of a real wind turbine. The advantage of the emulator is
that various wind profiles and wind turbines can be
incorporated as desired through software system [4]. Complex
testing of the generator control for grid and stand alone regime
[5, 6] also need a good wind turbine emulator.
A wind turbine emulator consists essentially of (Fig. 1):
A software system which implements the mathematical
model of the wind turbine;
A physical system that provides the similar static and
dynamic characteristics as the real studied system.
Being in a close loop, the two systems communicate with
each other. The real time software system sends reference
signals to the physical system which responds during the real
time simulation [7].
Fig. 1. General schema for hardware in the loop wind turbine
emulator.
II. WIND TURBINE CHARACTERISTICS
The wind turbine output power varies with the wind speed
[8]:
P 0.5 A C p ( ) v3 (1)
where the tip-speed ratio is defined as:
r R v (2)
and:
A : blade swept area [m2];
: specific density of air [kg/m3];
v : wind speed [m/s];
R : radius of the turbine blade [m];
r : rotating speed [rad/s];
Cp : coefficient of power conversion.
Because wind speed varies with environmental conditions,
the system should seek a maximum power operation point. P1,
P2 and P3 in Fig. 2 are the maximum power operating points at
corresponding r1 r2 and r3 rotating speeds, for three
different wind speeds v1, v2 and v3 [9].
For the beginning, a digital simulation of the wind turbine
system, including the wind turbine model and electrical
generator model, was implemented in Matlab/Simulink.
The scheme contains a variable wind speed profile generator,
a wind turbine aerodynamic model and a PMSG model with
the corresponding load (Fig. 3).
Then a hardware emulator for the wind turbine was
proposed. The emulator structure, presented in Fig. 4, contains:
a dSPACE control board, a voltage source inverter with direct


   
 59
 torque control (DTC), a three phase induction machine with a
gearbox (GB) the wind turbine equivalent - and a permanent
magnet synchronous generator (PMSG) with resistive load.
The emulator control was implemented in Matlab/Simulink
and it runs in real time on dSPACE board. The control output
signal (torque reference TIMref) is send to the DTC inverter,
which returns the real (estimated) IM torque (TIM) and the drive
(estimated) rotating speed (IM). The complete block scheme is
shown in Fig. 5.
In steady-state (neglecting the gearbox efficiency):

TIM Twt r IM . (3)

The dynamic equation is:

Twt Tg J (d r dt) (4)

where:
Fig. 2. Wind turbine characteristics. Twt : wind turbine torque [Nm];
Tg : generator torque [Nm];
J : inertia of the wind turbine system [kgm2].

Fig. 3. Wind turbine Matlab/Simulink simulation scheme.

Fig. 4. Wind turbine hardware emulator configuration.

Fig. 5. Wind turbine dSPACE implementation scheme.

60 
   

 The wind turbine torque and power results from the wind Because Jem is different than J, it is necessary to introduce
turbine characteristics: compensation inertia in the emulator control Jc:
Twt 1/ 2 CM v2 S R (5)
Jc J J em . (10)
where:
CM is the wind turbine torque coefficient; expression (6) was The speed derivative was computed using a close loop
obtained from the wind turbine test: observer in order to reject the speed noise (Fig. 6). (Direct
speed derivative would have introduced noise that would be
2.5 (6) passed as a reference torque to the IM. Noise filtering would
CM CM 0 a b
where: have introduced delays in the system and because of the
a, b, CM0 are the constants for the nominal tip-speed ratio 0; reverse reaction would have negatively affected the stability of
the system).
Pwt Twt r . (7)
Twt Testimated J (d r dt ) (11)
The inertia of the wind turbine system is composed by:
TGB T0 kt Tg k r
(12)
J J wt Jg (8)
Twt TIMref TGB J c (d r dt) (13)

where: Twt Tg ( Jc J IM ) (d r dt) (14)

Jwt : inertia of the wind turbine [kgm2];
Jg : inertia of the PMSG [kgm2]; where:
TGB is gearbox friction torque [Nm];
The hardware part of the emulator has itself the inertia Jem, T0 , kt , k , have been experimentally identified, observing
composed by: the control output signal and the generator torque for different
speeds and loads.
J em J IM J GB Jg (9)

where:
JIM : inertia of the IM [kgm2];
JGB : inertia of the gearbox [kgm2];

Fig. 6. Torque observer Matlab/Simulink scheme.






 61
 The PMSG shaft torque, as emulator output, is:
Tgs TIM IM r ( J wt Jc ) d
III. SIMULATION AND EXPERIMENTAL RESULTS
The parameters of the emulator test bench are:
Rated power, Pn = 5.5 [kW];
Rated wind speed, v0 = 11 [m/s];
Maximum speed, n = 126 [rpm];
Pole pairs, pp = 16;
Rated power of the IM, PIM = 7.5 [kW];
Rated IM speed, nIM = 715 [rpm];
Turbine inertia, Jwt = 140 [kgm2];
PMSG inertia, Jg = 1.05 [kgm2];
IM inertia, JIM = 0.156 [kgm2];
Hardware part of the emulator inertia, Jem = 6.68 [kgm2];
Gearbox coefficient, n = 6.03;
Blade swept area, A = 19.6 [m2];
Radius of the turbine blade R = 2.5 [m];
Maximum coefficient of ppower conversion Cp = 0.42;
Nominal tip-speed ratio 0 = rR/ v0;
Constants for the nominal tip speed ratio ( 0 = 3): a =
0.0986, b = 0.0113, CM0 = 0.0222;
r dt . (15) 6SHFLILFGHQVLW\RIDLU 1.225 [kg/m3];
kt = 0.05, k = 3.08, T0 = 17.66 [Nm];
ki = 10 [N/rad], kp = 100 [Ns/rad].
A Tf = 0.02 sec. time constant first order low pass filter was
used to reduce the speed measured noise.
The behavior of the wind power system was studied by
simulation and emulation, considering a step variation of the
wind speed between 9 and 10 [m/s], with a time period of 40
sec., as it is shown in Fig. 7 a) and in Fig. 8 a). The
corresponding shaft speed, wind turbine torque and generator
torque are presented in Fig. 7 b) - d) for simulation study and
in Fig. 8 b) - d) for emulation process. Also the wind turbine
energy is illustrated in Fig. 9; simulation results in Fig. 9 a) and
the experimental (emulation) results in Fig. 9 b). Good
similarities can be observed between simulation and
experimental results, which validates the good accuracy of the
emulator. The system responses were tested considering a
rectangular wind speed shape which is the most difficult
regime that is never meet in real situations.
The experimental setup, containing the dSPACE control
board and DTC inverter is presented in Fig. 10.

Fig. 7. Simulation results: a) Wind speed; b) Rotating speed; Fig. 8. Emulation results: a) Wind speed; b) Rotating speed;
c) Wind turbine torque; d) PMSG torque. c) Wind turbine torque; d) PMSG torque.

62 
   

 

Fig. 9. Wind turbine energy: a) Simulation result; b) Emulation result.



a) b)

c)

Fig. 10. Experimental Setup: a) dSPACE control board; b) DTC Inverter; c) IM, GB and PMSG.






 63
 IV. CONCLUSION
The paper presents a laboratory tool, useful for study and
testing real equipments for wind energy systems, without a real
wind turbine.
The digital simulations and experimental results are in good
agreement, validating the wind turbine emulator.
The advantages of the proposed system are:
flexibility (large scale of wind turbine and generators can
be tested without investments in aerodynamic tunnels and
wind turbines);
it can be easily estimated the energy produced by the
wind system at different wind speed profiles;
possibility to test different control loops, power
converters, storage elements, electrical loads etc.;
universality (the emulator can be used at different
powers; it is necessary to have the proper DTC drive and
the corresponding IM and GB in terms of torque and
power);
rapid prototyping.
Our future work will be focused on PMSG emulation, using
a synchronous d.c. excited machine with the same rated speed
as the IM (without the GB) making the equivalences with the
real generator in terms of inertia and induced voltage.
ACKNOWLEDGMENT
This work was supported by the strategic grant
POSDRU/88/1.5/S/50783, Project ID50783 (2009), co-
financed by the European Social Fund Investing in People,
within the Sectoral Operational Programme Human Resources
Development 2007-2013, the strategic grant POSDRU
107/1.5/S/77265, inside POSDRU Romania 2007-2013 co-
financed by the European Social Fund Investing in People
and The European Economic Area (EEA) project RO 0018
64
,PSURYHPHQW RI WKH 6WUXFWXUHV DQG (IILFLHQF\ RI 6PDOO
Horizontal Axis Wind Generators with Non-Regulated
%ODGHV
REFERENCES
[1] Z. Chen, J. M. Guerrero, F. Blaabjerg, A review of the state of the art of
power electronics for wind turbinesIEEE Trans. Power Electr., vol. 24,
no. 8, August 2009, pp.1859-1875.
[2] J R. Thresher et al., The status and fuWXUH RI ZLQG HQHUJ\WHFKQRORJ\
IEEE Power and Energy Magazine, vol. 5, no. 6, November/December
2007, pp. 34-46.
[3] B. Beltran, M.E.H. Benbouzid, and T. Ahmed-Ali, High-order sliding
mode control of a DFIG-based wind turbine for power maximization and
grid fault tolerance Electric Machines and Drives Conference, 2009,
,(0'&,(((,QWHUQDWLRQDO, pp.183-189.
[4] B. Neammanee, S. Sirisumrannukul and S. Chatratana, Development of
a wind turbine simulator for wind generator testing, International
Energy Journal, Volume 8, 2007, pp.21-28.
[5] M. Fatu, C. Lascu, G.-D. Andreescu, R. Teodorescu, F. Blaabjerg, I.
Boldea, Voltage sags ride-through of motion sensorless controlled
PMSG for wind turbinesIndustry Applications Conference, 2007. 42nd
IAS Annual Meeting. Conference Record of the 2007 IEEE, pp. 171- 178.
[6] M. Fatu, L. Tutelea, R. Teodorescu, F. Blaabjerg, I. Boldea, Motion
sensorless bidirectional PWM converter control with seamless switching
from power grid to stand alone and back, Power Electronics Specialists
Conference, 2007. PESC 2007. IEEE, pp. 1239 - 1244.
[7] G. Caraiman, C. Nichita, V. Mnzu, B. Dakyo, Marine current turbine
emulator design based on hardware iQWKHORRSVLPXODWRU VWUXFWXUH 14th
International Power Electronics and Motion Control Conference, EPE
PEMC 2010, pp. 101-107.
[8] Md. Arifujjaman, M. Tariq Iqbal, J. E. Quaicoe and M Jahangir Khan,
Modelling and control of a small wind turbine, Proc. of 18th Annual
Canadian Conference on Electrical and Computer Engineering
&&(&(, Saskatchewan Canada, May 2005, pp. 778-781.
[9] N. Muntean, O. Cornea and D. Petrila, A new conversion and control
system for a small off grid wind turbine Optim 2010, May 2010,
Brasov Romania, pp. 1167-1173.






 Maximum Power Point Tracking Based on
Sliding Mode Control
K. Boulam and A. Boukhelifa
Instrumentation Laboratory, Faculty of Electronics and Computers
University of Sciences and Technology Houari Boumediene, Algeria
kboulaam@yahoo.fr, aboukhelifa@yahoo.fr

Abstract- In this paper, a non-linear control algorithm based on R: radius of rotor, tutb : the turbine rotor speed.
a sliding mode theory is proposed to maximize the power
extraction of a variable speed wind turbine. This algorithm works The power coefficient Cp defines the aerodynamic efficiency
in the partial load region. It has been used to control the wind of the wind turbine rotor. It is represented by various
turbine speed so that it reaches the desired value which
approximation expressions. In this paper, Cp is expressed by
corresponds to the maximum power point. In order to prove its
efficiency, it has been compared with a classic linear control using (3), [1]:
a PI controller. Simulation tests are established using Matlab-
!"
Simulink. The proposed MPPT algorithm presents good static 
and dynamic performances.
C (, ) = 0.5176 .    0.4 .   5 . e # + 0.0068 . 
  (3)
  %.% *

= $%.%&.'
 -
' $
I. INTRODUCTION 

Wind energy is becoming one of the most important
renewable sources of clean energy. So many studies [1-4][7]
are oriented toward this type of energy production in the aim to
make it more efficient. Variable speed wind turbines are
widely used in this field owing to their ability to maximize
wind power extraction [1-5]. Maximum power point tracking
(MPPT) strategies play an important role in wind power
conversion systems (WECS) because they maximize the power
extracted from the wind, and therefore optimize the conversion
efficiency. Two strategies are used in literature [1-7], with or
without speed control. In this paper we used the strategy with
speed control; it permits to carry the speed wind turbine into
the desired value which corresponds to the maximum power
point.
Using an appropriate control algorithm could improve the
wind power efficiency. For that, linear and non-linear control
are used by authors [1][4][7], we adopted in this paper a non-
linear control which is more suitable for non-linear systems,
such as the wind turbine model. This control is based on sliding
mode theory.
First, we used a PI controller for turbine speed control, and
the results are presented. Then a sliding mode algorithm is
proposed and the results are compared with those of the PI
controller. Finally, the analysis of the results will show the
improvement given by the sliding mode control.
The system consists of an aero turbine, which converts wind
energy into mechanical energy, a gearbox, which serves to
increase the speed and decrease the torque and a generator to
convert mechanical energy into electrical energy.
The mechanical equation of the shaft, including both the
turbine and the generator masses, is given by [5]:
/2
J. /3 = T9  T:  f. ; (4)
J and f are the total moment of inertia and the viscous
friction coefficient appearing at the generator side, Tg is the
gearbox torque, Tem is the generator torque, and  is the
mechanical generator speed.
ABD
@ T9 = E
> 
We have: 3FG = E (5)
? BD
>T =
<  
G is the gear ratio, and Taer is the aerodynamic torque.
The system of equation (1) to (5), permit to us to construct
the bloc diagram of the wind turbine as shown below.

II. WIND TURBINE MODELING   R. 3FG turb 1 

 G
The aerodynamic (mechanical) power developed by a wind Cp
turbine is given by the following expression [2][4][5][7]: Taer 1 Tg 
v 1  +-

 . C . . . R. G H.I$K
P =  . C (, ). . . R .  (1) 2  3FG
Tem

:air density,  : wind speed, :power coefficient, : blade Turbine Gearbox Shaft
pitch angle, : tip-speed ratio, it given by (2).
. Fig. 1. Wind turbine bloc diagram
= 
(2)






 65
 Tg Tg
III. MUXIMUM POWER POINT TRACKING TECHNIQUE
ref
+ Tem ref Tem 1 
The Cp(




 

    
 +- C - -+
J. s + f
  
 
    
 
  
 
 
      !opt) for which Cp is maximum. In
Control Process
 " #$  
%  & 
' !*p max=0.48
 opt=8.1).
According to (2), we can see that if the rotor speed is kept
constant, then any change in the wind speed will change the
tip-speed ratio , leading to degrade the power coefficient Cp,
as well as the generated power output of the wind turbine.
However, if the rotor speed is adjusted according to the wind
speed variation, then the tip-speed ratio can be maintained at an
Fig. 3. Speed Generator Feedback Control
This control structure consists to adjust the torque appearing
on the turbine shaft in order to fix the turbine speed at a
reference which permits to track the maximum wind power.
This reference is deducted from (2).
&  opt, so Cp at its maximum Cpmax which could
yield maximum power output from the system according to (1)
[1][5]. So the MPPT technique consists in varying turbine
speed constantly according to wind speed variations, so that the
tip-speed ratio is maintained in its optimum value, thus the
power generation is optimum.
This operation is obviously valid while the maximum
allowable rotor speed and rated power are not reached.
Arriving to these values; we are no more concerned by the
MPPT, but rather by maintaining them constant, a necessary
proceeding for system protection.
IV. MPPT CONTROL SCHEME
In order to extract the maximum power from the wind, we
adopted the speed turbine control strategy. It permits to carry
the speed wind turbine into the desired value which
corresponds to the maximum power point. The wind turbine
speed control scheme is represented in Fig. 3. We deducted it
from the wind turbine bloc diagram (Fig. 1).
0.5
0.48
0
0.4 4
Power Coefficient
SUVW ..X
NOQ = Y
(6)
In this study, we assume that the generator and its converter
are both ideal. So the electromagnetic torque equals to its
reference all the time.
In a first time, we use a linear control basing on a
proportional integral (PI) controller.
Since the torque Tg is a non-linear function of  
compensation permits the linearization of the process
associated at its control. As a result, the synthesis of the
controller is found largely simplified [6].
The compensation of the term which causes the non-linearity
of the system is just an approach which could a little discard
the real system of its control.
To approach more to reality, we have developed a non-linear
control for the speed generator. It takes into account the non-
linearity of the system.
In addition, the wind speed which is variable and
unpredictable is considered as a perturbation for our system.
Therefore, we need a robust control algorithm which permits
the system to reach its reference without being affected by this
perturbation.
V. SLIDING MODE

Sliding mode control is one of the non-linear techniques. It

0.3
8 is a particular operation mode of variable structure control
systems. Its concept consists in moving the state trajectory of
12
0.2 the system to a predetermined surface called sliding surface
16 and maintaining it around this latter with an appropriate logic
20 commutation (Fig. 4)[9].
0.1
24
28
0 32
0 5 8.1 10 15 20 x1
Tip-Speed Ratio
S(x)>0
Fig. 2. Power Coefficient versus Tip- Speed Ratio for Different Values of the
Pitch Angle
S(x)<0 S(x)=0
x2
Fig. 4. Sliding Mode Principle

66 





 The design of sliding mode controller is done in three steps: Thus the necessary low control to bring back the variable we
1 - Selection of the sliding surface. want control to the selected surface, respecting both the
2 - Establishing the conditions of existence and convergence. existence and attractiveness conditions is given by: u=ueq+ua
3 - Determination of the control law.
Design of sliding mode control algorithm
In general, for a system defined by the state equation (7), for a
vector u of dimension m, we must choose m surfaces. To design a variable structure speed controller [8], we
consider the equation system (9):
xZ = f(x) + g(x). u (7)

Z = H . (T9  T:  f. ) (9)
\ :
With: x [ R and u [ R
We choose the error as being the sliding surface:
Concerning the surface form, [8] propose the following form:
S(2) = e(2) = K   (10)
/ ^
S(x) = /3 +  . e(x) (8) This surface derivative is:
SZ = Z K  Z (11)
With:
e(x): the error between the variable and its reference. To ensure the attractiveness condition SZ. S < 0 , SZ is chosen
' 
 indicating the desired control bandwidth. as follow:
r: relative degree, equal to the number of times to derive the
output to appear the command.
When the control system operates in variable structure
sliding mode, the switching (commutation low) always respects
the condition S(x) = 0. Therefore, the derivative versus time
should also always be zero, ie. SZ(x) = 0.
So that this condition is respected at all times, the magnitude
of command should take a well determined value, designated
by the equivalent control ueq.
So that the evolution trajectory of the system tends to S = 0,
the system must be submitted to the attraction of this surface
Fig. 5. This will be done by the attractive control ua
determined by the condition of attractiveness: S(x). SZ(x) < 0.
This command defines the dynamic behavior of the system
during the trajectory convergence mode to the sliding surface.
It is equal to zero once the sliding mode is achieved. The
simplest solution is to choose ua with the relay shape:
u = k. sign`S(x)a. With: k>0
S(x)=0
Fig. 5. The State Trajectory to the Sliding Surface

SZ = k . sign(S)  k  . S (12)
k1 and k2 are positive constants, sign(S) is the signum
function.
So, the torque command Tem ref can be obtained from (9),
(11) and (12):
T: K = J. `k . sign(S)  k  . S  Z K a + T9  f.  (13)
VI. SIMULATION RESULTS
The wind turbine speed control is tested on a variable wind
power conversion system based on a doubly-fed induction
generator (DFIG) of 4kW. The wind turbine parameters are
given in table I. In order to evaluate the MPPT control strategy
proposed in this paper, the simulation is carried out using the
Matlab/Simuling software. Fig. 6 shows the wind speed used in
this simulation. Simulation results are shown in Fig. 7 to Fig.
13, for the two types of controller chosen.
All the figures show that our objective is attained. The two
controllers have permitted to extract the maximum of power
from the wind. The simulation results show that sliding mode
controller gives a rapid response in comparison with the PI
controller. In addition, Fig. 5 and Fig. 6 show that the
generator speed follows better its reference with a sliding mode
controller. Concerning the power coefficient, it reaches the
optimal value (Cpmax=0.48) in the two cases. Its affected by a
sudden wind speed variation in the PI controller case and not
affected in the sliding mode case. We can clearly see that in the
zoom (Fig. 10). The sliding mode surface (Fig. 11 and Fig. 12)
equals to zero, which shows that the controller parameters are
properly chosen.




 67
 6000

8 4000

Generator Power(W)
Wind speed (m/s)

6 2000

4 0 PI controller

2 -2000

0 -4000
0 5 10 15 20 25 30 SM controller
Time(s) -6000
0 5 10 15 20 25 30
Time(s)
Fig. 6. Wind Speed Profile

Fig. 9. The Generated Power

120
Reel and Reference Generator Speed

0.5
100
0.4
Power Coefficient

80 PI controller
Reel Speed 0.3
60
0.2
40
0.1
20
0
0 0 5 10 15 20 25 30
0 5 10 15 20 25 30
Time(s)
Time(s)

Fig. 7. Generator speed with PI controller Fig. 10. Power Coefficient

Reel and Reference Generator Speed(rd/s)

120 0.481
SM controller
0.48
Power Coefficient Zoom

100

80 0.479
PI controller
60 0.478

40 0.477

20 0.476

0 0.475
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time(s) Time(s)

Fig. 8. Generator Speed with SM Controller Fig. 11. Power Coefficient Zoom

68 
   

 TABLE I
CONCLUSION
WIND T URBINE PARAMETERS
In this paper, we have proposed a MPPT control strategy of a
Rated power 10 kW variable wind turbine based on a sliding mode algorithm. To
show the improvement given by this control algorithm, a
Density of air
= 1.225 kg/m3 comparison with a PI controller applied to the same control
Number of blades 3 structure is effectuated. The simulation results show that the
non-linear control successfully controls the variable speed
Radius of rotor R= 3m
wind turbine, satisfying static and dynamic performances, and
Gear ratio G= 5.4 realizing the power maximization. Due to non-linearity of the
Turbine total inertia J = 0.2 kg.m2 wind turbine model, a robust controller should be more adapted
than a linear one.
Total viscous friction coefficient f = 0.0016 N.m/s
REFERENCES
[1] Oscar Barambones, Patxi Alkorta and Manuel De La Sen+ EUI de
Vitoria, Wind Turbine Output Power Maximization Based on Sliding
120 Mode Control Strategy, IEE2010, pp.364-369.
[2] B. Beltran, M.E.H. Benbouzid and T. Ahmed-Ali, High-Order Sliding
100 Mode Control of a DFIG-Based Wind Turbine for Power Maximization
and Grid Fault Tolerance, 2009IEEE, pp.183-189.
[3] Xuemei Zheng, Lin Li, Dianguo Xu, Jim Platts, Sliding Mode MPPT
80
Control of Variable Speed Wind Power System, 2009 IEEE.
[4] G.Tapia and A.Tapia, Wind Generation Optimization Algorithm for a
Sliding Surface

60 Doubly Fed Induction Generator, IEE Proc.-Gener.Transm. Distrib.,vol.

152, No.2, pp. 253-263, March 2005.
40 [5] S. El Aimani, Modlisation de Diffrentes Technologies dEoliennes
Intgres dans un Rseau de Moyenne Tension, Thse de Doctorat,
Lille 2004.
20 [6] V.Rogez, Modlisation Simplifie de Sources de Production
Dcentralises pour des Etudes de Dynamique des Rseaux. Application
0 lIntgration dune Production Eolienne dans un Rseau de Distribution
Insulaire, Thse de Doctorat, Lille 2004.
[7] Tinglong Pan, Zhicheng Ji, Zhenhua Jiang, Maximum Power Point
-20
0 5 10 15 20 25 30 Tracking of Wind Energy Conversion Systems Based on Sliding Mode
Time(s) Extremum Seeking Control, IEEE Energy 2030, Atlanta, GA USA, 17-
18 November, 2008.
Fig. 12. Sliding Surface [8] Slotine.J.J.E. and W. Li., Applied Non-Linear Control [M], Prentice-
Hall, Englewood Cliffs, NJ. 1991, No. 226-269.
[9] V.I Utkin, Sliding Mode Control Design. Principles and Applications to
Electric Drives, IEEE Trans, on Industrial Electronics, Vol. 40, No. 1
5 pp. 23-36, February 1993.
4

2
Sliding Surface

-1
-2

-3

-4

-5
0 5 10 15 20 25 30
Time(s)

Fig. 13. Sliding Surface Zoom






 69
 Efffects of
o Winnd Turbbines on
Taher Abediinzadeh
Islamic Azad UUniversity,
Shaabestar Branchh, Shabestar,
Iran
o Volltage Profile
P e
Mehdi Ehhsan Hadi Afsharrirad Mohamm mad Nazaralilooo
Sharif University of Isllamic Azad Unniversity, Islamic Azad
A Universitty,
Technoloogy Shabbestar Branch, Shabestar, Shabestar Branch,
B Shabesstar,
Iran Iran

Abstract- For d distribution grrids connected d wind poweer PSCAD/E EMTDC to invvestigate the im
mpact of these tturbines on
generration, voltage quality is an im mportant issuee for the systemm a distributtion grid.
plannning purpose. The operatioon of wind tu urbines in th he
distribution network ks may affect thhe voltage quality offered to th
he II. WIND TUR RBINE
consuumers. One of the most importtant consideratiions is the effecct
on thhe voltage proffile, i.e. the ind
duced slow volltage variationss,
Wind energy
e conversion systems (WECS) are tthe devices
whichh are the subjecct of this paper. which aree used to conveert the wind ennergy to electrical energy.
Th
he wind turbinees can be operrated in two modes:
m constantt- The total wind power enntering the winnd turbine is givven by:
speedd and variable--speed. Each on ne of these cases has differen nt 1
impacct on steady state voltagee of distributiion grids. Foor PT UAV 3C p (1)
2
evaluation of these impacts, we haave simulated SCIGS generatoor
Where  is the denssity of air in kg/m3 and A iis the area
that works
w in constaant-speed operaation and DFIGG generator thaat
is useed in a variablle-speed turbine. We have used a simulation n swept by the blades of wind turbine, in m2 and V iss the speed
progrram of PSCAD D/EMTDC to in nvestigate the impact of thesse of wind inn m/s and CP iss dimensionlesss power coeffiicient.
turbinnes on a distrib
bution grid. Since calculation
c of CP is too comp plicated, usuallly its value
calculatess from simpler formulas. Thee formula that is used for
Keyy words- Wind power,
p Voltagee profile, DFIG CP is[6]:
S (O  c1 )
C p (0.44  0.0167 E ) Sin  0.00184(O  c1 ) E (2)
I. INTROODUCTION c2  0.3E
Wiind energy is said to be one of o the most proominent sourcees Where  is Tip-Speedd Ratio and  is the blade pitch angle.
of eleectrical energyy in years to co ome. The increeasing concernns C1 andd C2 are two coefficients that t equal to 2 and 11,
to environmental
e issues demaand the searrch for morre respectiveely.
sustaiinable electriccal sources. With
W the windd farms ratingg The TS SR dictates thhe operating coondition of a turbine. A
hundrreds megawattts, the concerns start to focuss on the voltage typical poower coefficiennt CP versus TSR is given inn Fig. 1. As
probllems of the pow wer system. the wind sspeed changes, the Tip-Speed Ratio and CP will vary.
Mo ost of the models used to represent
r a wiind turbine arre The CP chharacteristic has single maxiimum at a speecific value
basedd on a non linear relationnship betweenn rotor poweer of TSR. T Therefore if thhe wind turbinee is operating at constant
coeffficient and linear tip speed off the rotor bladee. speed thenn the power cooefficient will be maximum only o at one
Onn the dynamic subject, severral works havee been done onn wind speeed.
the an nalysis of pow
wer quality from m wind turbinees to the poweer
system ms. [4] develooped an extenssive analysis of o the potentiaal
impacts of wind turrbines on the power
p quality, where
w the workk
focussed on the smaall scale integrration. Tande inn [5] explainedd
how to assess the voltage qualitty from wind turbines usingg
wind turbine characcteristics.
WT TG generates electrical pow wer by convertting the kinetic
energgy in the windd into rotary mechanical
m outpput and then too
AC electrical
e energgy. The turbin ne can he operrated in one of o
two modes:
m constannt-speed and variable-speed
v . Presently, thhe
trendd of operating wind
w turbines is
i in the variabble-speed modee.
Reacttive power of constant-sspeed turbinees cannot be
contrrolled but variaable-speed turbbines can be coontrolled in twoo
ways. They can prooduce power with constant poower coefficiennt
or coontrol magnituude of voltage of PCC poinnt. Each one of o Fig. 1. typical CP verssus TSR
thesee cases has different impactt on steady sttate voltage of o
distriibution grids. For evaluationn of these imppacts, we have In a w
wind power sysstem typicallyy the wind turrbine starts
simullated SCIG generator thaat works in constant-speedd operating (cut-in speed)) when the winnd speed exceeeds 3-4m/s,
operaation and DFIG G generator thaat is used in a variable-speedd and is shuut off at speedss exceeding 255 to 30m/s. In between,
b it
turbinne. We haave used a simulation program of o can operaate in the optiimum constannt CP region, the speed-
limited reegion or the poower-limited reegion as shownn in Fig. 2.

70 
   

 Optim
mum power eextractor makees turbine to work in these PWM connverters with an intermediatte capacitor link for DC
regions. bus voltagge regularity. The converterr that is connected to the
rotor called rotor side converter an nd the other named 'grid
A.
A Pitchanglee side convverter'. By conntrolling the grrid and rotor cconverters,
Thhe block diagrram that is ussed to control pitch angle is
i the DFIGG can be adjustted to achieve many capabiliities versus
show
wn in below (Fig .2): conventioonal squirrel caage induction generators.
g

The dq representationn of DFIG macchine are givenn by[2]:

For thee stator side:
Ods  Ls ids  Lm idr (4)
Oqs  Ls iqs  Lm iqr (5)
dOds
vds  Rs ids  Z s Oqs  (6)
dt
dOqs
Fig. 22. Pitch angle conttrol block diagram
m
vqs  Rs iqs  Z s Ods  (7)
dt
Thhis controller iincrease pitch angle if turbiine shaft speedd For thee rotor side:
exceeeds its maximmum speed. Bu ut pitch anglee can only rise
somee degrees (here 20 degrees). O dr Lr idr  Lm ids (8)
Oqr Lr iqr  Lm iqs (9)
B.
B OptimumP PowerExtracto
or
dOdr
Thhis section diictates turbinee operating conditions
c andd vdr Rr idr  sZ s Oqr  (10)
specifies optimum CP C to obtain reeference torque and reference dt
speedd of generatoor. On the othher hand, if turbine poweer dOqr
exceeeds its maximuum value, Optiimum Power Extractor
E shouldd vqr Rr iqr  sZ s Odr  (11)
dt
contrrol it with varyiing CP. The eleectrical output power from thhe induction ggenerator is
given by:
C.
C TurbineSim mulation 3
Beelow equation iis used for turbbine simulationn: Ps (vds ids  vqs iqs ) (12)
2
PT USDT Vw C p
2 3

TT (33) 3
Zr 2Z r Pr d idr  v qr iqr )
(vdr (13)
2
Whhere  = 1.2255 and turbine blade radius is i 23 m. Vw is
i
wind speed and bN is turbine shafft speed. Pg Ps  Pr (14)
Where Ps is the pow wer delivered by the stator, Pr is the
III. DFIG GENERATOR
G power too the rotor, Pg is the totaal power geneerated and
delivered to the grid.
A double-Fed innduction generrator is as a standard woundd And reaactive power:
rotor induction geenerator with its stator winndings directlyy 3
Qs (vqqs ids  vds iqs ) (15)
conneected to the poower grid and rotor connecteed to the poweer 2
grid through
t a frequuency converteer [7] (Fig.3). The electrical torquee generated by b induction m
machine is
given by:
3
Te  P (Ods iqs  Oqss ids ) (16)
2
And thee mechanical equation
e for maachine is:
dZ r
Tm  Te J  DZ r (17)
dt
Where Tm is mechannical torque annd Te is electrromagnetic
torque in Nm, bN is mechanical
m shaaft speed in raad/sec, D is
fiction coefficient in Nmm/rad/s and J iss inertia in kg-m
m2.
In this paper, the pow wer converter that is betweeen grid and
Fiig.3. DFIG overaall system in a wind
w turbine-gennerator rotor is coonsidered ideall.

In modern DFIG G designs, thhe frequency converters arre

usuallly built by twoo, three phase self
s commutateed back-to-backk


   
 71
 IV. VECTO OR CONTROL the quadraature rotor currrent compensaation term, derivved from
Thhe technique off vector controll will be used iin the control of
o (15). Thiss control schem
me is shown in Fig. 5(b).
inducction generatoor to control the generateed voltage foor
differrent loading coonditions and variable
v rotor sppeeds. C. Reactivepower
R rcontrol
Thhe total stator flux is alignned along the d-axis of thhe If we want
w control Q, idrref is obtainned from (26). Its control
frame so that [33]: Oqs 0
rotatiing reference fr scheme iss shown in Fig. 6. The seconddary stage for ccontrolling
reactive power
p is similaar to Fig. 5(b).
So iqs is derived:
Lm
iqs  iqr (188)
Lls  Lm
Substitutinng iqs in (9), (100) will result inn: a) primary
p stage
L
2
L d2
(199)
vdr Rr idr  sZ s Lr  m
iqr  Lr  idr
m

Ls
Ls dt
L2 L V L2 d
vqr Rr iqr  sZ s Lr  m idr  m s  Lr  m iqr (20
0)

Ls LsZ s Ls dt
3 Lm
Te  P Ods iqr (211)
2 Lls  Lm
Neeglecting statorr resistance willl lead to[2]:
Vds 0 (222)
Vqs Z s Ods (233)
Suubstituting for Vds=0, Eqs. (111) and (12) willl be simplifiedd
b) secondary sttage
as folllows:
3 (244) Fig. 4. Speed coontrol scheme baased on vector coontrol
P (vqs iqs )
2
3
Q (vqs ids ) (255)
2

A.
A Electromag gnetictorque//speedcontrol
Opptimum Power Extractor calculates
c desiired torque foor
turbinne. Therefore, iqrref will be derived
d as show wn in Fig.4 (a)).
a) primary stagee
Compparing the reeference variaable to the actual a machinne
current, an error siggnal required foor controlling tthe speed of thhe
machhine is obtainedd. To determine the required rotor voltage, a
standdard PI controlller and the su ummation of the t direct rotoor
current compensatioon term, deriveed from (20), is i implementedd.
This control schemee is shown in Fig.F 4(b).
Suubstituting for ids into Eq. (255) will result inn:
2
Lm Vs Vs
Qs idr  (266)
( Lls  Lm ) Z s ( Lls  Lm )
DFFIG generatorss can be contro
olled in constannt voltage statte
or in constant reactiive power.

B.
B Voltageco ontrol
In order to controol terminal volttage, the ampliitude of idr is
contrrolled. In orderr that control vo
oltage, referencce voltage is b) secondary sttage
comppared with the terminal
t voltagge. Then error control signal
is prooduced to an inntegral controlller, as shown inn Fig. 5(a). F 5. Voltage control scheme based
Fig. b on vector control
c
Thhe secondary sttage of the conntroller is consttructed using
the prrimary stage reeference currennt compared to the direct
compponent of meassured rotor currrent. The requiired rotor
voltagge is obtained from a PI conttroller and the summation
s of

72 
   

  

1
0.995
0.99
0.985
0.98
0.975
0.97
0.965
Fig. 6 Reaactive power contrrol scheme: primarry stage 0.96
0.955
V. VOLTA AGE PROFILE 0.95
Too illustrate the impact of a wind turbine on the voltage 0.945
profille along a line,, a distributionn grid is considdered (Figure 7)
7 4 8 12 16 20
0 24 28 32
[10]. The node 1 iis connected too the high voltage grid. Thhe SCIG DFIG V=
=  DFIIG Q=0
rated voltage is 15 kV. On average, the X/R ratio of thhe
distriibution lines inn the studied grid
g is approxim mately 7.5. Thhe  !

total load is 10 M MVA. A windd turbine is assumed a to be 1.02

conneected to node 99.
Figgure 8 shows tthe wind speedd at hub heightt, used as inpuut 1
for th
he simulation.
Thhree cases forr the turbine type are connsidered in thhe 0.98

followwing scenarioss (Table 1).

0.96

0.94

0.92

0.9
4 8 12 16 20 24 28 32

SCIG
G DFIG V=
=  DFIG Q=0

 

0.99
Fig. 7. Simpliified one-line diaggram of the study case
c feeder.
0.98

0.97

0.96

0.95

0.94

0.93

0.92
4 8 12 16 20 24 28 32

SCIG
G DFIG V=c
c  DFIG Q=0

Fig. 9. Volltages of nodes 1 and

a 9 and 12
`Fig. 8. Asssumed wind speed, as input for the simulation
All the generators aree used in wind turbines is cateegorized in
Tablle 1. Cases considered for simulationns one of thhe above cases. Voltages of two nodes aree shown in
Reactive Fig. 13. As
A can be seenn in the figure, the distributedd generator
Speed Generator
ccontroller type
power has impacct on the voltagge profile alongg the line.
control The reaactive power ooutput of a SC CIG (cases 1) cannot be
Case
fixed speed SCIG no controlledd. The reactivve power outtput is always negative
1
(inductivee behavior). Inn this case, redu uction of voltaage level is
Case
vvariable speed DFIG cos =
2 1
seen in Fiig. 9.
Case voltage The reaactive power forf case 2 is permanently coontrolled to
vvariable speed DFIG
3 control zero. It is seen that in caase 2, voltage levels
l are uppeer than case


   
 73
 1. In case 3 that the DFIG is controlled in constant voltage
state, the voltage in node 9 has been constant in 1 p.u in all
wind speeds
The highest impact on the voltage is for this example seen at
node 9, where the difference between the voltages for the case
1 and 3 is approximately 6.23%.
Also, the lowest impact (2.55%) on the voltage is at node 1,
where distribution grid is connected to high voltage grid.

VI. CONCLUSIONS
Each type of wind turbines has different impact on steady
state voltage of distribution grids. For evaluation of these
impacts, we have simulated SCIG generator that works in
constant-speed operation and DFIG generator that is used in a
variable-speed turbine. DFIG generator can be controlled in
two ways that mentioned before: constant voltage and constant
reactive power. We have used a simulation program of
PSCAD/EMTDC to investigate the impact of these turbines on
a distribution grid.
In inclusion, in weak grids, voltage controlled turbines had
best improvement in the voltage profile of distribution system.
After that, the turbine that was controlled in constant power
coefficient had better impact on the voltage profile than SCIG
generator.

VII. REFERENCES

[1] J.G. Slootweg, S.W.H. de Haan, H. Polinder and W.L. Kling, General
model for representing variable speed wind turbines in power system
dynamics simulations, IEEE Transactions on Power Systems, vol. 18,
no. 1, pp. 144-151 , Feb. 2003.
[2] E. Muljadi and C.P. Butterfield, Pitch-controlled variable-speed wind
turbine generation, IEEE Transactions on Industry Applications , vol.
37, no. 1, pp. 240-246, Jan/Feb 2001.
[3] T. Ackermann and L. Soder, An overview of wind energy-status 2002,
Renew. sustain. Energy Rev., vol. 6, no. 12, pp. 67128, Feb./Apr.
2002.
[4] J. O. Tande, P. Nrgaard, P. Srensen, L. Sndergaard, P. Jrgensen, A.
Vikkels, J. D. Kledal and J. S. Christensen, Power Quality and grid
Connection of Wind Turbines, Summary report, Vol. 1, 2 and 3, Ris
report Ris-R-853, Roskilde, Denmark, October, 1996.
[5] J. O. Tande, Applying Power Quality Characteristics of Wind Turbines
for Assessing Impact on Voltage Quality, In Wind energy, 5, 2002.
[6] Seul-Ki Kim, Eung-Sang Kim, Jae-Young Yoon and Ho-Yong Kim,
PSCAD/EMTDC Based Dynamic Modeling and Analysis of a Variable
Speed Wind Turbine
[7] Shuhui Li, Sitanshu Sinha, "A Simulation Analysis of Double-Fed
Induction Generators for Wind Energy Conversion Using PSpice", IEEE
Transaction on Energy Conversion Systems. VOL. 32, No. 2, November 2006.
[8] I. Erlich, F. Shewarega, Modeling of Wind Turbines Equipped with
Doubly-Fed Induction Machines for Power System Stability Studies ,
IEEE, 2006
[9] L. Holdsworth, X.G. Wu, J.B. Ekanayake and N. Jenkins, Comparison of
fixed speed and doubly-fed induction wind turbines during power system
disturbances, Proceedings of IEE, Vol. 150, No. 3, May 2003
[10] N.G. Boulaxis , S.A. Papathanassiou, M.P. Papadopoulos, Wind
Turbine Effect on The Voltage Profile of Distribution Networks, Renewable
Energy, 2002.

74 
   

 New strategy for optimization of output power of a
DFIG wind turbine
Taher Abedinzadeh Mehdi Ehsan Hadi Afsharirad Mohammad Nazaraliloo
Islamic Azad University, Sharif University of Islamic Azad University, Islamic Azad University,
Shabestar Branch, Shabestar, Technology Shabestar Branch, Shabestar, Shabestar Branch, Shabestar,
Iran Iran Iran

Abstract This paper presents a method to maximize the 1

U AV C p
3
output power of a DFIG wind turbine. In this method, an PT (1)
optimization technique based on mathematical analysis has been 2
used. The modeling of DFIG is performed using MATLAB. Where is the density of air in kg/m3 and A is the area
To evaluate the maximum output power of a DFIG, detailed swept by the blades of wind turbine, in m2 and V is the speed
expressions for output power and losses are derived as functions of wind in m/s and CP is dimensionless power coefficient.
of generator slip, the magnitude and phase angle of rotor
Most of the models used to represent a wind turbine are
excitation voltage in various wind velocities.
The outcomes of this method demonstrate that maximum output based on a non linear relationship between rotor power
power of a DFIG wind turbine do not happen in the maximum coefficient and linear tip speed of the rotor blade. Since
input power (the mechanical power of wind turbine). calculation of CP is too complicated, usually its value
calculates from simpler formulas. The formula that is used for
Keywords Wind turbine, DFIG, Optimization of output CP is [9]:
power, Rotor voltage S (O  c1 ) (2)
C p (0.44  0.0167 E )Sin  0.00184(O  c1 ) E
I. INTRODUCTION c 2  0.3E
Wind energy is said to be one of the most prominent Where is Tip-Speed Ratio (TSR) and is the blade pitch
sources of electrical energy in years to come[1]-[5]. Wind angle. C1 and C2 are two coefficients that equal to 1 and 18,
energy conversion systems (WECS) are the devices which are respectively.
used to convert wind energy to electrical energy and the
variable speed wind turbines with Doubly-Fed Induction
Generator (DFIG) are becoming the most common type of
wind turbines [5]-[10].
In the DFIG, the applied rotor voltage controls the real and
reactive powers and generators speed, when its stator
terminals are connected to a power system and the stator
voltage is held constant by the grid.
Mechanical power of wind turbine varies with its shaft
speed and wind velocity and consists of two parts: the output
power of the machine and the power losses.
In this paper, the optimization technique is used to
maximize the power that is delivered to the electrical grid. The (a)
results are compared with those from the approach in which
only the mechanical power of wind turbine is maximized.
This paper is laid out as follows: Firstly in section II, the
wind turbine model is presented. In section III doubly-fed
induction machine and its equations are touched on. Section IV
explains conventional way to obtain maximum output power.
Section V introduces an optimization technique for output
power of generator and a brief explanation about it. The results
are presented in section VI and the conclusion marks the last
section.

II. WIND TURBINE MODELING

Wind energy converts to mechanical energy by the turbines
in the wind farms. The total wind power entering the wind (b)
turbine is given by: Fig.1. a) power coefficient versus lambda and pitch angle. b) power coefficient
versus lambda when pitch angle is equal to zero


   
 75
 The TSR dictates the operating condition of a turbine. A Rr
I s I s Ts Rs jX ls jX lr s I r I r T r
typical power coefficient CP versus TSR is given in Fig. 1. As
the wind speed changes, the Tip-Speed Ratio and CP will vary.
Is  Ir
The CP characteristic has single maximum at a specific value
of TSR. Therefore if the wind turbine is operating at constant Vr
V s V s 0q E jX m
speed then the power coefficient will be maximum only at one s
wind speed.

III. DIFG AND ITS EQUATIONS
A double-Fed induction generator is as a standard wound
rotor induction generator with its stator windings directly
connected to the power grid and rotor connected to the power
grid through a frequency converter (Fig.2) [10]. In modern
DFIG designs, the frequency converters are usually built by
two, three phase self commutated back-to-back PWM
Fig.3. steady-state equivalent circuit of a DFIG
As mentioned before, the total output power of DFIG
equals of algebraic sum of rotor and stator real powers (cd ,ch ).
The output power and power losses can be derived from
voltage and current equations as below:
converters with an intermediate capacitor link for DC bus Pout Ps  Pr 3real V s I s* V r I r*  (6)
voltage regularity. The converter that is connected to the rotor
called rotor side converter and the other named 'grid side Ploss 3(R r | I r |2  R s | I s |2 ) (7)
converter'. By controlling the grid and rotor converters, the
DFIG can be adjusted to achieve many capabilities versus IV. CONVENTIONAL METHOD FOR OBTAINING MAXIMUM
conventional squirrel cage induction generators. OUTPUT POWER OF WIND TURBINE
Doubly-fed induction generator delivers power from stator In conventional method, maximum point of wind turbine
to the grid. The rotor depends on its voltage can either delivers mechanical power is considered as the best point to obtain
power to the grid or absorbs it. The equivalent circuit of DFIG maximum output power of wind turbine.
is shown in fig.3. )RU D SDUWLFXODU YDOXH RI SLWFK DQJOH HJ  ZLQG
The voltage equations of machine are as follows: turbine mechanical power Pm can be depicted as functions of
V s R s  jX s jX m I s turbine speed and wind velocity, as shown in Fig. 3.
V Rr (3) Two operating modes are identified in Fig. 3: the
r s jX m  jX r I r maximum power tracking mode and the rated output power
s
mode.
When the wind velocity is low and generator output power
Then the current equations of stator and rotor are obtained
does not exceed its maximum value, the wind turbine is
from (3):

 
operated at the maximum turbine mechanical power tracking
Rr V
 jX r V s  ( jX m ) r mode. In this case, the turbine speed must change with varying
s s wind velocity in order to achieve maximum power[11].
(4)
 
Is
Rr
 R s  jX s  s  jX r  X m 2 When the wind velocity is high and the generator output
power Ps+Pr exceeds its upper limit, the DFIG must be operated
at the rated output power mode in order to avoid generator
 R s  jX s V r s  ( jX m )V s overload. Usually, blade pitch angle control is used in this mode
(5)
 
Ir to avoid generator overspeed.
Rr
 R s  jX s  s
 jX r  X m2

Fig.2. DFIG overall system in a wind turbine-generator

Fig. 4. Wind turbine mechanical power ( 

76 
   

V. OPTIMIZATION METHOD FOR OUTPUT POWER OF WIND
TURBINE
This technique consists of two sections. In first section,
output power is maximized. This optimization is formulated as
below:
Maximize Pout
Subject to the constraints:
(1) PT Pout  Plosses
(2) V r d V rmax
(3) I s d I smax
a) Generator slip of DFIG versus wind speed
(4) I r d I rmax
As shown above, the power equations are nonlinear. So; it
is impossible to obtain the optimal points from simple
analytical methods. In this paper, A Subspace, Interior, and
Conjugate Gradient Method for Large-Scale Bound-
Constrained Minimization Problems in MATLAB environment
is used for optimization.
When DFIG production reaches its rated power, second
section of optimization is begun. It is formulated as follows:
Minimize Plosses
Subject to the constraints:
(1) Pout Pout max
b) Voltage rotor magnitude of DFIG versus wind speed
(2) PT Pout  Plosses
(3) V r d V rmax
(4) I s d I smax
(5) I r d I rmax

VI. OPTIMIZATION RESULTS

The effect of rotor voltage on output power was studied in
a wind turbine coupled to a DFIG generator.
Characteristics of the generator are shown in appendix. The
results of optimization are shown in fig.4, when the wind
turbines were operated in first part of optimization to
maximize the output power of machine. After determining c) Output power of DFIG versus wind speed
generator slip and rotor excitation voltage (Fig. 4(a)-4(b)), the
powers cjlm , and cm were computed and the results are
depicted in Figs. 4(c)-4(d).
Based on the outcomes in Fig. 4, the following observation
has been come in order:
1) When wind speed is increasing, slip of the generator
is decreasing until it reaches its minimum value or magnitude
of rotor voltage reaches its maximum value. It is shown in Fig.
4(a)-4(b).
2) The mechanical power of wind turbine in this new
method is lower than when only the mechanical power of
turbine is maximized. Before voltage magnitude of rotor
reaches its maximum value, turbine mechanical powers of
d) Turbine mechanical power versus wind speed: 2XWSXWSRZHU
proposed and conventional schemes are so close and difference maximization . Turbine mechanical power maximization
between them in first case is increasing from 0.058% to
Fig. 4. Results of output power maximization versus wind speed
0.085% (Fig. 4(d)).


   
 77
 Machine parameters
Vs=220 V
Rs 
Rr=
Xs 
Xr 
Xm 
Turn ratio (Ns/Nr)=5.4:1
(Rr, Xr, and Xm are referred to the stator)

a) Output power of DFIG versus wind speed
b) Power losses of DFIG versus wind speed
Fig. 5. Results of power losses minimization as functions of wind speed
The results of second part of optimization (power losses
minimizing) is shown in Fig .5.
In this section of optimization, the machine output power is
reached to its rated value and because of this, power losses is
minimized. The powers Pout and Plosses is shown in Figs .5(a)-
5(b).
VII. CONCLUSION
Optimization of output power for a doubly-fed wind turbine
generator has been investigated. Specific conclusions are as
follows
1) When the wind speed is low and the DFIG is operated at
the maximum output power tracking mode, the optimal rotor
excitation voltage changes with rotor speed and wind speed.
When wind speed is increasing, slip of the generator is
decreasing until it reaches its minimum value or magnitude of
rotor voltage reaches its maximum value.
2) The mechanical power of wind turbine in this new
method is lower than when only the mechanical power of
turbine is maximized, but output power is maximum value in
new approach.
REFERENCES
[1] H. Patel and V. Agarwal, Maximum power point tracking scheme for
PV systems operating under partially shaded conditions, IEEE Trans.
Ind. Electron. , vol. 55, no. 4, pp. 1689-1698, Apr. 2008.
[2] M. Cirrincione, M. Pucci, and G. Vitale, A single-phase DG
generation unit with shunt active power filter capability by adaptive
neural filtering, IEEE Trans. Ind. Electron. , vol. 55, no. 5, pp. 2093-
2110, May 2008.
[3] C. Rodriguez and G. A. J. Amaratunga, Long-lifetime power inverter
for photovoltaic AC modules, IEEE Trans. Ind. Electron. , vol. 55, no.
7, pp. 2593-2601, Jul. 2008.
[4] N. Femia, G. Lisi, G. Petrone, and G. Spagnuolo, Distributed
maximum power point tracking of photovoltaic arrays: novel approach
and system analysis, IEEE Trans. Ind. Electron. , vol. 55, no. 7, pp.
2610-2621, Jul. 2008.
[5] T. Abedinzadeh, M. Ehsan, D. Talebi Dynamic Modeling and Control
of Doubly-Fed Induction Generators in Wind Turbines Using
PSCAD/EMTDC, IEEE conference, ICEEN 2011.
[6] A. K. Jain, and V. T. Ranganathan, Wound rotor induction generator
with sensorless control and integrated active filter for feeding
nonlinear loads in a stand-alone grid, IEEE Trans. Ind. Electron. , vol.
55, no. 1, pp. 218-228, Jan. 2008.
[7] R. Pena, R. Cardenas, J. Proboste, G. Asher, and J. Clare, Sensorless
control of doubly-fed induction generators using a rotor-current-based
MRAS observer, IEEE Trans. Ind. Electron. , vol. 55, no. 1, pp. 330-
339, Jan. 2008.
[8] A. G. Abo-Khalil and D. C. Lee, MPPT control of wind generation
systems based on estimated wind speed using SVR, IEEE Trans. Ind.
Electron. , vol. 55, no. 3, pp. 1489-1490, Mar. 2008.
[9] Seul-Ki Kim, Eung-Sang Kim, Jae-Young Yoon, Ho-Yong Kim,
"PSCAD/EMTDC Based Dynamic Modeling and Analysis of a
Variable Speed Wind Turbine", IEEE Transaction on Power Systems,
VOL. 42, No. 6, November/December 2004.
[10]Shuhui Li, Sitanshu Sinha, "A Simulation Analysis of Double-Fed
Induction Generators for Wind Energy Conversion Using PSpice",
IEEE Transaction on Energy Conversion Systems. VOL. 32, No. 2,
November 2006.
[11] Chien-Hung Liu, Yuan-Yih Hsu, Effect of Rotor Excitation Voltage
on Steady-State Stability and Maximum Output Power of a Doubly-
Fed Induction Generator IEEE Trans, 2009.

APPENDIX
DFIG parameters
Nameplate data
3-phase, 60 Hz, Y-connected, 6 poles.
Stator: 220 V/phase, 6.9 A.
Rotor: 70 V/phase, 6.9 A.

78 
   

 Analysis of a Wind Energy Conversion System
with Doubly-Fed Induction Generator in healthy
and Fault Condition via Simulation
Dimitrios G.Giaourakis, PhD Student, Athanasios N.Safacas, Professor Dr-Ing, Savvas Tsotoulidis, PhD Student

Electromechanical Energy Conversion Laboratory

Electrical and Computer Engineering Department,
University of Patras, Rio-Patras, Greece, 26504
Tel:(+30)-2610-997351, fax: (+30)-2610-997362
Email: dgkiaourakis@ece.upatras.gr
a.n.safacas@ece.upatras.gr
stsotoulidis@ece.upatras.gr

Abstract- In this paper, a wind energy conversion system that
uses the Cascade technology especially the structure of
Doubly-Fed Induction Generator (DFIG) is presented and
analyzed. Moreover, a simulation of this wind energy
conversion system under fault conditions in MATLAB/
SIMULINK is done and the results are evaluated. Open-circuit
faults in one IGBT or in two IGBTs are studied.
Keywords: DFIG, simulation, back-to-back converter, PWM,
fault conditions in DFIG.
I. INTRODUCTION
Since the oil crisis happened in 1973, people started
experimenting on how the wind potential could be used for
the production of electric energy. Attempts have been made
to find out ways to exploit Wind Potential (as well as the
other forms of renewable energy, for example solar energy,
hydroelectric energy, etc) in the production of electric
energy. Various types of systems for wind potential
conversion have been designed. Each one of them has
advantages and disadvantages and they could be used in
specific occasions. The most important of these systems are
the following [4]:
a) Fixed-speed wind turbine with an induction generator
Danish concept.
b) Variable speed wind turbine concept with variable
rotor resistance.
c) Variable speed concept with full-scale power.
d) Variable-speed wind turbine equipped with a doubly-
fed induction generator.
In this work the behaviour of a wind energy conversion
system with Doubly-Fed Induction Generator (DFIG) is
studied. This type of system has a lot of advantages. One of
them is that the wind generators can be controlled more
easily and additionally they can be operated in the region of
low wind speeds e.g. 4m/s, in which other wind energy
conversion systems couldnt be operated. The variable speed
range of such wind energy systems is about 30% below and
above the nominal synchronous speed. The rating of the
power converter which is a back-to-back voltage source
converter is 25-35% of the generator capacity [5, 6], which
made this an economical choice compared to the full
converter design of most other variable speed wind turbines.
For these systems monitoring and diagnosis are important to
prevent fault conditions that maybe have catastrophic
consequences. Fault diagnosis could be done by measuring
some basic quantities of the system in healthy condition,
registering data and comparing them with the corresponding
of fault condition. So, fault diagnosis and identification could
be achieved.
As it is concluded by the international literature the
fault conditions that have been investigated mostly are:
a) unbalanced stator conditions
b) voltage sags of the grid
c) faults in induction motor (e.g. broken bar)
d) DC link fluctuation.
In this work faults in the back-to-back converter are
studied, when the DFIG operates in subsynchronous mode,
so the Grid Side Converter (GSC) operates as a boost
rectifier and without using the upper IGBTs (Fig. 1). Two
types of defects can appear on IGBT-an open circuit or a
short circuit fault. It is more possible that these faults can
appear in one IGBT only or in two IGBTs either in the same
leg of converter, or in two different legs. In the first case, the
system can operate for a considerable period of time but with
degraded performance and low efficiency. Open-circuit
faults in one IGBT (in Grid Side Converter or in Rotor Side
Converter) and in two IGBTs (in Rotor Side Converter) are
investigated.
II. SYSTEM DESCRIPTION
The systems structure including the DFIG (Fig.1)
consists of an asynchronous machine that operates in
generation mode, of which the stator is connected directly to
the grid of constant voltage and frequency and the rotor
plugging in the grid via a back-to-back power electronic
converter. This system is used to achieve a wide range of
operation from subsynchronous to supersynchronous
operation, namely the generator to operate with positive and
negative slip. This can be achieved via power flow in both
directions.


   
 79
 Fig. 1. Schematic diagram of the investigated system-basic structure.
leakage inductances [H] , is the speed of the reference
That is why the bi-directional topology of AC/DC/AC frame [rad/s].
converter is used. The rotor electrical power Pr is only a By calculating the apparent power and taking the real
fraction of stator active power output Ps (Pr=-sPs). Since the parts, the following equations represent the stator-side and
electromagnetic torque Mel is negative in case of power rotor-side active and reactive power respectively:
generation and s is positive and constant for a constant
frequency grid voltage, the sign of Pr depends on the slip   (5)

sign. Pr is positive for negative slip (speed greater than
synchronous speed), or negative for positive slip (speed    (6)

lower than synchronous speed). For supersynchronous speed The mutual flux between rotor and stator produces
operation, Pr is transmitted to DC bus capacitor and tends to magnetic energy, which is stored in the magnetic field. This
raise the DC voltage. For subsynchronous speed operation, energy produces an electromagnetic torque which is
Pr is taken out of the DC bus capacitor and tends to decrease calculated as
the DC bus voltage. M el p (iq <Sd  id <Sq ) , (7)
This paper deals with the reaction of the DFIG when an
where p is the number of pole pairs.
open-cicruit fault happens in one or two IGBTs of the back-
Equation (7) describes the torque balance in DFIG:
to-back converter in subsynchronous mode.
d:
A. Doubly-Fed Induction Generator model
(J M  J L )  M el ML
dt . (8)
The analysis of the DFIG operation is based on the Parameter values for the asynchronous machine are
well-known dynamical model described by the following presented in Table 1.
equations (synchronously frame): TABLE I
d<Sd PARAMETER VALUES OF DFIG
u Sd R S i Sd  s <Sq  PN (kW) 6.5
dt VN (V) 380
d<Sq p 2
u Sq RS i Sd  Z s <Sd  RS ()
LlR (H)
0.8
0.005
dt , (1)  () 1
d<Rd  (H) 0.005
u Rd RS i Rd   ( s  R )<Rq Lm (H) 0.114
dt
, (2)
d<Rq B. Rotor side converter (RSC)
u Rq RS i Rq   ( s  R )<Rd
dt Rotor side converter acts as a PWM rectifier during the
<Sd ( LSV  Lm )i Sd  Lm i Rd machine working in super synchronous mode and as an
inverter during subsynchronous mode. The purpose of the
<Sq ( LSV  Lm )i Sq  Lm i Rq rotor side converters control is the independent control of
, (3)
the rotors active and reactive power. To succeed it, three
<Rd ( LRV  Lm )i Rd  Lm i Sd phase quantities are transformed in d-, q- synchronously
. (4)
<Rq ( L RV  Lm )i Rq  Lm i Sq reference frame, where the d-axis is aligned with the stators
flux vector (Fig. 2). This happens because the stator of the
where uSd, uSq, uRd, uRd, iSd, iSq, iRd, iRq, , , , , are
DFIG is directly connected to the grids voltage. This means
voltages [V], currents [A] and flux linkages [Wb] of the
that S is constant and Sq is equal to zero. So the following
relations are applied:
stator and rotor in d- and q-axis, RS is the resistance of the 

stator windings [], LS, LR, Lm, are the stator, rotor and  
 
 (9)
mutual inductances [H], LR, LS are the stator and rotor

80 





 Neglecting the small stators resistance Rs:
uSd=0.
The following equations are obtained solving the (3):
 , (10)
By substituting these equations to (9), (11), the active and
reactive power are given by the following equations:
, , (11)
From the equations (11) it is concluded that independent
control of active and reactive power can be succeeded.
Finally, the equation that calculates the electromagnetic
torque is given by the following equation:
(12)
Fig. 2. Block diagram of RSCs control.
C. Grid side converter (GSC)
The grid side converter acts as a PWM inverter during
the machine working in super synchronous mode and as a
rectifier during subsynchronous mode. The purpose of the
grid side converter is to produce active power and feed it in
the rotor side converter. In subsynchronous mode, the
rotors power is coming from the capacitor of the DC link so
the voltage interconnection is dropped down. This leads to
power transportation from the grid to the capacitor. In
supersynchronous mode the rotor feeds the capacitor, its
voltage increases and the grid side converter transfers active
power to the grid. Moreover, reactive power that is
exchanged between grid side converter and the grid, can be
controlled. In our case the upper IGBTs of the GSC are
always off.
The control of this converter (Fig. 3) is achieved by
transforming the three phase quantities to a d-, q-
synchronously frame, in which the d-axis is aligned with the
vector of the grids voltage, because grids voltage is
constant. So, the following relations are resulted:
 analysis for fault diagnosis and identification has been done.
 , , (13)
where ugd is the component of grid voltage in d-axis and ugq is
the component of grid voltage in q-axis.
Through the following equations the active and reactive
power can be calculated:
 ,  (14) modulation technique with variable duty cycle and fsw=10
It is concluded that independent active and reactive power
control can be achieved.


Fig. 3 Block diagram of GSCs control.
III. FAULTS STUDY IN DFIG
In Fig. 1 the rotor side converter in fault conditions can
be seen (cases F2, F3). Since the transistor T1 has now an
open-circuit fault (F2), the phase a of the rotor of the
induction machine is connected to the positive dc rail
through the diode D1. The rotor phase a voltage is then
determined by the polarity of current and the switching
pattern of transistor T4. The phase voltage (Va) will be
clamped to the negative rail if the rotor current ia is positive.
On the other hand, the phase voltage Va will be clamped to
the negative rail when transistor T4 is switched on, and then
to the positive rail when transistor T4 is off and D4 is on, if ia
is negative.
If the upper switch T1of rotor phase (a) and lower
 switch T6 of the rotor phase (b) are simultaneously opened
(case F3), the defect causes a loss of reversibility in currents
of two faulty switches. Thus, the two currents of two faulty
arms became unidirectional and nonsinusoidal. The loss of
the current alternation for each faulty phase, will involve a
deformation on the third phase current, and since the neutral
is insulated, this last is almost null when the two currents of
phases are simultaneously null.
In Fig. 1 the grid side converter in fault condition is
seen (F1). Since T2 has an open-circuit fault (F1), only its
diode (D2) is conducting in specific periods of time. In our
case that the upper IGBTs are constantly off and only the
upper diodes are conducting, when (F1) happens then diode
D5 doesnt conduct in the moment that it would conduct in
normal case.
IV. SIMULATION RESULTS
To study in detail the behaviour of a DFIG 6.5 kW,
when the open-circuit faults, that have been described before,
happen, the whole system has been simulated in
Matlab/Simulink environment and some characteristic results
are presented in each case of fault. Furthermore, FFT
A constant value of wind speed is considered. So the wind
turbine and rotor speed of the asynchronous machine is
specific and equal to 900rpm (DFIG operates in
subynchronous mode).
In the rotor side converter SPWM modulation technique
with fsw=10 kHz is used and in the grid side converter PWM
kHz is used. Parameters value of filters: a) rotor side
converter: C=1mF, L=5mH, b) grid side converter:
C=0.5mF, L=4mH (it is observed via simulation that with
lower values of GSCs LC filter a high harmonic content
occurs). Also, the value of the CDClink is equal to 1mF.
In Fig. 4-10 some characteristic simulations results
can be seen, when open-circuit fault (F1) happens.




 81
 Furthermore, the spectrum of some basic variables of the
system is evaluated in order to use them for fault diagnosis
and identification.

Fig. 6. DC link voltage (open-circuit fault happens in t=2sec).

Fig. 4. Grid line to line voltage in GSC filter input in fault condition.

In Fig. 5b normalized FFT analysis spectrum of grid

faulty phase current in GSCs filter input in healthy(blue
line) and fault condition(red line) can be seen. The harmonic
content of this variable of the system in fault condition is
presented in Table II.
TABLE II
HARMORMONIC CONTENT OF GRID FAULTY PHASE CURRENT
FREQUENCIES (Hz) NORMALIZED AMPLITUDE
(a)
0 0.85
50 1
100 0.8
106 0.05
150 0.1

The range of frequences (100 Hz-115 Hz), that appears

around of 110 Hz both in healthy condition and in fault
condition, is due to the operation of DFIG (on similar case of
fault in a boost rectifier this bandwith doesnt appear[14]). It
is observed via simulation results that normalized FFT
analysis spectrum of the other two grid phase currents in (b)
GSCs filter input is similar to this in Fig. 5b appearing a Fig. 7. (a)Rotors phase currents in healthy and fault condition.
lower DC component. (b) Normalized FFT analysis spectrum of rotor phase current in
healthy(blue line) and fault condition (red line).

(a)
Fig. 8. Electromagnetic torque(open-circuit happens in t=2sec).

As it is observed when the open-circuit fault (F1)

(b)
Fig. 5. (a) Grid phase currents in GSCs filter input in fault condition.
(b) Normalized FFT analysis spectrum of grid faulty phase current
in GSCs filter input in healthy(blue line) and fault condition(red
line).
occurs, the only part of the system that is affected is that in
GSCs side. More specifically, a specific harmonic content
occurs in grids phase currents in GSC. This harmonic
content occurs only when an open-circuit fault happens in
one IGBT of GSC and for that reason it could be used for
fault diagnosis. A possible fault diagnosis method of this
kind of fault is measuring and recording the grids phase
currents in healthy condition and comparing them with the
results of measurements that are done in regularly intervals.
Moreover, fault identification about which IGBT of the GSC
has an open-circuit fault could be done comparing the
normalized amplitude of the DC component of the grids
phase currents. The faulty phase current appears a very high
normalized amplitude and it also is higher than the one

82 
   

 appearing in the currents of other phases-approximately
twice.
In Fig. 9-12 some characteristic simulations results can
be seen, when open-circuit fault (F2) happens. Furthermore,
the spectrum of some basic variables of the system is
evaluated in order to use them for fault diagnosis and
identification. In this case the only part of the system that is
affected is the RSC and the behaviour of the asynchronous
machine. For that reason simulation results only for some
basic rotors and stators variables are presented.
Fig. 11. Electromagnetic torque(open-circuit happens in t=2sec).

Fig. 9. DC link voltage (open-circuit fault happens in t=2sec).

In Fig. 9b normalized FFT analysis spectrum of rotor

faulty phase current in healthy(blue line) and fault (a)
condition(red line) can be seen. The harmonic content of this
variable of the system in fault condition is presented in Table
III.
TABLE III
HARMORMONIC CONTENT OF ROTOR FAULTY PHASE CURRENT
FREQUENCIES (Hz) NORMALIZED AMPLITUDE
0 1
20 0.3
40 0.25
60 0.1
80 0.005
100 0.005 (b)
It is observed via simulation results that normalized Fig. 12. (a) Stators phase currents in fault condition.
FFT analysis spectrum of the other two rotor phase currents (b) Normalized FFT analysis spectrum of stators phase
current in healthy(blue line) and fault condition (red line).
is similar to this in Fig. 9b appearing a higher basic harmonic
component. As it is observed when the open-circuit fault (F2)
occurs, the only part of the system that is affected is that in
RSCs side. More specifically, a specific harmonic content
occurs in rotor phase currents. This harmonic content occurs
only when an open-circuit fault happens in one IGBT of RSC
and for that reason it could be used for fault diagnosis.
Moreover, fault identification about which IGBT of the RSC
has an open-circuit fault could be done comparing the
normalized amplitude of the basic harmonic of the rotor
phase currents. The faulty phase current appears a low
(a)
normalized amplitude and it also is lower than the one
appearing in the currents of other phases-approximately three
times.
In Fig. 13 rotor phase currents and their spectrum can
be seen and evaluated in order to use them for fault diagnosis
and identification, when open-circuit fault (F3) happens. In
this case the only part of the system that is affected is the
RSC and the behaviour of the asynchronous machine.
Electromagnetic torque and stator phase currents are similar
to those of the previous case.
In Fig. 13b normalized FFT analysis spectrum of rotor
(b)
Fig. 10. (a) Rotors phase currents in fault condition. faulty phase current in healthy (blue line) and fault condition
(b) Normalized FFT analysis spectrum of rotor phase (red line) can be seen. The harmonic content of this variable
current in healthy(blue line) and fault condition (red of the system in fault condition is presented in Table IV.
line).






 83
 TABLE IV
HARMORMONIC CONTENT OF ROTOR FAULTY PHASE CURRENT
FREQUENCIES (Hz) NORMALIZED AMPLITUDE
0 1
20 0.3
40 0.1
60 0.1
80 0.01
100 0.01
It is observed via simulation results that normalized
FFT analysis spectrum of the other rotor phase currents is
similar to this in Fig. 13b. The other faulty phase appears the
same basic harmonic component, while the healthy rotor
phase current appears a higher basic harmonic content.
(a)
(b)
Fig. 13. (a) Rotors phase currents in fault condition.
(b) Normalized FFT analysis spectrum of rotor phase
current in healthy(blue line) and fault condition (red
line).
As it is observed when the open-circuit fault (F3)
occurs, the only part of the system that is affected is that in
RSCs side. More specifically, a specific harmonic content
occurs in rotor phase currents. This harmonic content occurs
only when an open-circuit fault happens in two IGBTs in
different legs of RSC and for that reason it could be used for
fault diagnosis by measuring and recording the rotor phase
currents in healthy condition and comparing them with the
results of measurements that are done in regularly intervals.
Moreover, fault identification about which IGBTs of the
RSC have an open-circuit fault could be done comparing the
normalized amplitude of the basic harmonic of the rotor
phase currents. The faulty phase currents appear low
normalized amplitude and it also is lower than the one
appearing in the current of the other phase-approximately
three times.
V. CONCLUSION
In this paper a wind energy conversion system with
DFIG under fault condition in subsynchronous mode is
simulated. The faults that are investigated are open-circuit
faults in one IGBT (in Grid Side Converter or in Rotor Side
Converter) and in two IGBTs (in Rotor Side Converter). It is
concluded that when an open-circuit fault in one IGBT of the
84
GSC happens, the only part of the system that is affected is
that in GSC side. This fault causes a specific harmonic
content in the grid phase currents, which is unique for this
kind of fault and it could be used for fault diagnosis and
identification. More specifically, it causes a DC component
that in faulty phase has higher amplitude. The other parts of
DFIG arent affected so much when this open-circuit fault
occurs.
On the other hand, when an open-circuit fault in one or
in two IGBTs in different legs of the RSC, the only part of
the system, that is affected, is the RSCs side and the
asynchronous machines behavior. This fault causes a
specific harmonic content of the rotor and stator phase
currents, which is unique for this kind of fault and it could be
used for fault diagnosis and identification. More specifically,
a low amplitude of the basic harmonic appears in faulty
phases. A possible fault diagnosis method is the measuring
and recording the values of grids phase currents and rotors
phase currents and comparing their spectrum with the
corresponding data that have been recording in healthy
condition.
References
[1]. R. Pena, J. C. Clare, and G. M. Asher, Doubly fed induction generator
using back-to-back PWM converters and its application to variable-
speed wind-energy generation, IEE Proc. Electr. Power Appl., vol.
143, pp. 231241, May 1996.
[2]. D. G. Giaourakis, A. Safacas, Review of Wind Energy Conversion
Systems for Large Wind Turbines and Simulation of a back-to-back
Electronic Power Converter, EVER2011, Monte Carlo, Monaco, 31
March- 3 April 2011, conference proceedings.
[3]. Ned Mohan, Ted K. A. Brekken, Control of a Doubly Fed Induction
Wind Generator Under Unbalanced Grid Voltage Conditions, IEEE
Transactions Energy conversion, vol.no22. 1, march 2007 page 129-
135.
[4].Muller S, Deicke M., Rik W., Doncker D.: Doubly fed induction
generator systems for wind turbines, IEEE Trans Inds. Appl.,
Magazine 8 (3) (May/June 2002), pp. 2633.
[5]. D. Kastha, B.K. Bose, Investigation of Fault Modes of Voltage-Fed
Inverter System for Induction Motor Drive, IEEE Trans. Ind. Applic.,
vol. 30, no. 4, pp. 10281037, 1994.
[6]. R. Peuget, S.Courtine, J.P. Rognon, Fault Detection and Isolation on a
PWM Inverter by Knowledge-Based Model, IEEE Trans. Ind.
Applicat., vol. 34, No. 6, pp. 13181325, 1998.
[7]. Francisco K. A. Lima, Alvaro Luna, Pedro Rodriguez, Edson H.
Watanabe, and Frede Blaabjerg, Rotor Voltage Dynamics in the
Doubly Fed Induction Generator During Grid Faults, IEEE
TRANSACTIONS ON POWER ELECTRONICS, VOL. 25, NO. 1,
JANUARY 2010.
[8]. Lingling Fan, Subbaraya Yuvarajan and Rajesh Kavasseri, Harmonic
Analysis of a DFIG for Wind Energy Conversion System, IEEE
TRANSACTIONS ON ENERGY CONVERSION, VOL. 25, NO. 1,
MARCH 2010.
[9]. Lingling Fan, Zhixin Miao, Subbaraya Yuvarajan and Rajesh
Kavasseri,Hybrid modeling of DFIGs for wind energy conversion
systems, Simulation Modeling Practice and Theory 18 (2010),
Elsevier, pp. 1032-1045.
[10]. I. Colak, E. Kabalci, R. Bayindir and S. Sagiroglu, The design and
analysis of a 5-level cascaded voltage source inverter with low THD,
2nd PowerEng Conference, Lisbon, pp. 575-580, 18-20 March 2009,
conference proceedings.
[11]. M. Trabelsi, M. Boussak, M. Gossa, Multiple IGBTs Open Circuit
Faults Diagnosis in Voltage Source Inverter Fed Induction Motor Using
Modified Slope Method, XIX International Conference on Electrical
Machines - ICEM 2010, Rome, conference proceedings, pp. 1-6.
[12}. Tsotoulidis, S., Mitronikas, E., Safacas, A., Design of a wavelet
multiresolution controller for a fuel cell powered motor drive system,
2010 XIX International Conference on Electrical Machines (ICEM), pp
1 6, September 2010.






 Analysis of DFIG controlled by a matrix converter
Y. Soufi (1), T. Bahi (2), S. Ghoudelbourk (3), S. Lekhchine (2) and H. Merabet (4)
(1)
Electrical Engineering Department, University of Tbessa, Algeria
(2)
Electrical Department, Badji Mokhtar University, Annaba, Algeria
(3)
Electrical Department, University 20 August, Skikda, Algeria
(4)
Industrial Technology Unit Research, Annaba, Algeria
E-mail: y_soufi@yahoo.fr

Abstract- The principle of the converter topology is based on a
matrix linking each input phase to each output phase by a
bidirectional power switch. However, the symmetry of the matrix
is one of the advantages of matrix converter which allow the
power flow in both directions and thus directly operate in four
quadrants. This work presents an investigation on the
performance of a wind-based chain of a double-fed asynchronous
machine. The first analysis concerns the control strategy for the
LMSE modulation. The simulation tests are performed using
Matlab Simulink and the results are presented and analyzed.
Keywords- Matrix converter, DFIG, LMSE modulation, control.
II. SEPARATE CONTROL OF ACTIVE AND REACTIVE POWER
In this part, we will present a study on modeling and control of
active and reactive power of DFIG. We consider the operation
of the machine in all areas of possible functioning. Thus,
according to the conditions imposed on references powers, we
analyze the performance of the control for different modes of
operation and we offer different types of order. The order of
the speed of the wind machine in the chain is ensured by static
converters. Figure 1 shows the schematic of the considered
conversion chain.

I. INTRODUCTION

Assuming that the inverter is usually powered and provided by

a controlled rectifier system which may make all the currents
of highly polluted lines have an adverse effect on loads and
sensitive equipment connected nearby. The solution was to
add one or more storage elements (capacitors in general). This
device is connected upstream of the speed variations. It is used
to compensate harmonic currents absorbed by the drive Fig. 1. Schematic of the conversion chain
leading to almost sinusoidal current networks [1] [2]. In
contrast, the active filter that is added to the system To easily control the production of electricity from wind, we
controllable converters generating greater complexity of will achieve independent control of active and reactive power.
control, an increase of overall losses, a significant additional Once the stator of the machine connected to the grid, the ratio
cost and generates a consequent volume (between 30% and between electromotive forces created in the rotor and the
50% of the total volume of the variation for levels of power stator is defined by [5]:
superior than the kilowatt) because it is desirable to place
large-capacity capacitors to minimize the DC bus voltage to
obtain a higher output voltages quality in the inverter. These Er N r ws  p.Z r
. m.g (1)
capacitors also have a strong sensitivity to high temperatures, Es Ns ws
its lifetime and thus increasing maintenance problems. To
overcome these drawbacks, we propose to replace the power Considering the equations of stator and rotor voltages of the
of the DFIG rotor by a matrix converter [3] [4]. The structure
DFIG in Park frame and after a mathematical development
and the modeling of the double-fed asynchronous generator in
and orienting the stator flux along the axis "d" of the rotating
addition to the modeling of matrix converter and its control frame to ensure a decoupling between the stator and rotor [6]
strategy are presented in the following sections. The results of [7][8].
simulations are presented and discussed in the last section.


   
 85
 The equations of stator and rotor voltages and flux of the current due to a constant imposed by the network. Thus, the
DFIG in Park frame are: stator power can be controlled independently of one another.
In order to properly control the machine, it must then establish
d I sd the relationship between currents and rotor voltages which are
Vsd Rs I sd   T sI sq shown as follows:
dt
dI
Vsq Rs I sq  sq  T sI sd dI rd

dt Vrd Rr I rd  V .Lr . dt  erd
Vrd dI rd (9)
Rr I rd   T rI sq (2) dI
Vrq Rr I rq  V .Lr . q  erq  e0
dt
dI dt
Vrq Rr I rq  rd  T rI rq
dt With,
M2 M2
I sd L s I sd  M I rd erd  gws ( Lr  ) I rq , erq gws ( Lr  ) I rd ,
I Ls Ls
sq L s I sq  M I rq
(3) MVs (M 2 )
and V 1
I rd L r I rd  M I sd e0 g ( 10)
Ls Ls .Lr
I rq L r I rq  M I sq
Thus, we can build relationships between the applied
Where, the mechanical equation of the machine is given by the voltages to the machine rotor and stator power (see figure 2).
following expression:

d:
J Tem  Tr  K f .: (4)
dt
The electromagnetic torque (Tem) is given by:
M
Tem p ( I rd I sq  I rqI sd ) (5)
Ls
To this end, we will determine the equations linking the stator
active and reactive powers with rotor sizes on which we act to
control the machine.
Either: Fig. 2. Internal structure of the DFIG

I sdI and I s 0 sq
(6)
So, We note that the powers and tensions are bound by a transfer
M function of the first order. Moreover, because of the low value
Tem  p I rqI sd (7)
of slip g, it will be possible to establish a vector control
Ls
because the influences of the couplings remain weak and d
Taking into account (6) the equations of the flux (3) become: and q axes can be controlled separately with their own
I sd Ls I sd  M I rd regulators. Figure 3 shows the control block diagram deduced
(8)
from previous relationships.
0 Ls I sq  M I rq
Assuming that the electrical grid is stable, with simple voltage,
the stator flux constant and hence the equation (7) shows that
the electromagnetic torque is directly proportional to the
quadrature component of rotor current. So, if we neglect the
resistance of the stator windings and after some transformation
we show that the flow is kept constant (this condition is
ensured in the case of a stable network connected to the stator
of the DFIG, the choice of the dq reference makes the
electromagnetic torque produced and therefore the stator
power and rotor current proportional to the axis q. The stator
reactive power, in turn, is proportional to the d-axis rotor Fig. 3. Control Block diagram of DFIG

86 
   

 III. DFIG FED BY MATRIX CONVERTER The matrix converter will be designed and controlled to obtain
the desired output voltages will be defined by [10]:
Its a direct converter, totally reversible [9]. The denomination
derives directly from the fact that this converter has no storage
stage intermediate in the conversion chain. It converts a three
phase to another electric three phase network with variable U o1 U o . sin( ws .t )
frequency and amplitude. This converter allows both 2 .S
sinusoidal absorption of current networks, but also zero U o 2 U o . sin( wo .t  ) (13)
consumption of reactive power. Figure 4 shows the structure 3
U 2.S
of a converter with the input phase voltages U i1, U i2, U i3 and U o . sin( wo .t  )
output desired phase voltages Uo1, Uo2 and Uo3. The input o 2 3
phases are connected to the output phases through nine
switches four segments with forced commutation that
Similarly, the output currents, amplitude and phase Is, }s
represent the matrix elements K11, K12, ..., K33 which we
compared with output voltages are defined by:
illustrated by bidirectional switches capable of blocking
voltage in both directions.

I o1 I o . sin(wo .t  M o )
2.S
I o1 I o . sin( wo .t  M o  ) (14)
3
I 2.S
I o . sin(wo .t  M o  )
o1 3

The formulation of electrical converter quantities is based

primarily on a written matrix (connection matrix called [F])
Fig.4. Structure of Matrix Converter and the use of functions of connections (fij) of the equation
by:

The voltages at the entrance of the matrix converter are

K11 K12 K13
defined by: >F @ K K 22 K 23 (15)
21
K 31 K 32 K 33

U i1 (t ) U i . sin( w i .t ) (11)
2 .S These connection functions (logic function) describe the state
U i2 (t ) U i . sin( w i . t  )
3 of the switches constituting the converter. A connection
U 2 .S function equal to fij = 1 corresponds to a closed switch so that
(t ) U i . sin( w i .t  )
i3
3 the same connection function zero fij = 0 corresponds to an
open switch. Thus, the matrix formulation of the instantaneous
Ui:maximum amplitude of the input voltage; electrical behavior of the converter can be expressed by the
wi: pulse input voltages. following relations:
Similarly, the current reference line to be drawn on the
network input of amplitude Ii is in phase with its input voltage U s1 U e1
is defined by the equation: U >F @.U e 2 (16)
s2
U s 3 U e 3

I i1* (t ) I i . sin(wi .t ) and
* 2.S (12)
I i 2 (t ) I i . sin( wi .t  )
3 I s1 I e1
I * (t ) I . sin( w .t  2.S ) I
s2 > @
F . I e 2
T
(17)

i1 i i
3
I s 3 I e 3


   
 87
 IV. CONTROL ALGORITHM FOR THREE P HASE MATRIX if min E1, E 2, E 3, E 4 ..........E 27 E1 then K11 K 22 K 33 on
CONVERTER if min E1, E 2, E3, E 4 ..........E 27 E 2 then K 12 K 23 K 31 on
(19)
if min E1, E 2, E 3, E 4 ..........E 27 E 3 then K13 K 21 K 32 on
A matrix converter is composed of 3x3 switches. This number .......................................................
allows 512 combinations of switching states for the converter .........................................................
to operate safely we must, through its order to avoid if min E1, E 2, E 3, E 4 ..........E 27 E 27 then K 31 K 32 K 33 on
combinations that cause a short-circuit power and those that
cause open circuits of the load. For this purpose the 512
combinations of states will be reduced to only 27 states can be The relationship between input currents and output currents
realized. Table 1 shows different modes of the matrix are given by:
converter .

TABLE I i o1 ( t ) Mode 1
THREE P HASES M ATRIX CONVERTER SWITCHING STATES i (t ) Mode 2
MODE Sw 1 Sw 2 Sw 3 Us1 Us2 Us3 o2
1 K11 K22 K33 Ue1 Ue2 Ue3
i o 3 ( t ) Mode 3
2
3
K12
K13
K23
K21
K31
K32
Ue3
Ue2
Ue1
Ue3
Ue2
Ue1
>ii1 ( t ) @
4 K13 K22 K31 Ue3 Ue2 Ue1
5 K11 K23 K32 Ue1 Ue3 Ue2
6 K12 K21 K33 Ue2 Ue1 Ue3
7 K11 K22 K23 Ue1 Ue2 Ue2 0 Mode 27
8 K21 K32 K33 Ue2 Ue3 Ue3
9 K12 K13 K31 Ue3 Ue1 Ue1 i o 2 (t ) Mode 1
10 K12 K13 K21 Ue2 Ue1 Ue1 i (t ) Mode 2
11
12
K22
K11
K23
K32
K31
K33
Ue3
Ue1
Ue2
Ue3
Ue2
Ue3
o3
13 K12 K21 K23 Ue2 Ue1 Ue2 i o 4 ( t ) Mode 3
14 K22 K31 K33 Ue3 Ue2 Ue3
[ i i 2 ( t )]
15 K11 K13 K32 Ue1 Ue3 Ue1
16 K11 K13 K22 Ue1 Ue2 Ue1
17 K21 K23 K32 Ue2 Ue3 Ue2
18 K12 K31 K33 Ue3 Ue1 Ue3 0 Mode 27
19 K13 K21 K22 Ue2 Ue2 Ue1
20 K23 K31 K32 Ue3 Ue3 Ue2 io 3 (t ) Mode 1
21
22
K11
K11
K12
K12
K33
K23
Ue1
Ue1
Ue1
Ue1
Ue3
Ue2
i o1 (t ) Mode 2
23 K21 K22 K33 Ue2 Ue2 Ue3
24 K13 K31 K32 Ue3 Ue3 Ue1 io 3 (t ) Mode 3
25 K11 K12 K13 Ue1 Ue1 Ue1 [ i i 3 ( t )]
26 K21 K22 K23 Ue2 Ue2 Ue2
27 K31 K32 K33 Ue3 Ue3 Ue3

The control strategy that we apply to the three phases of the i o 1 ( t )  i o 2 ( t )  i o 3 ( t )....... Mode 27
matrix converter is based on the difference of error (Least
Mean Square Error LMSE) between measured values and
desired outputs [11].
IV. S IMULATIONS RESULTS
E1, where K11, K22, K33 are closed (Mode 1), when E2, K12, Figure 5 shows the results when the DFIG is fed by a matrix
K23, K31 are closed (Mode 2), ... ..... E27 when K31, K32, converter. For orders on active and reactive power imposed
K33 are closed (Mode 27). according to the paces shown respectively by Figures 5.a and
5.b, (where the power factor is unit), we notice that these
The equations of E1, E2, ..., E27, are represented by:
powers perfectly follow their values of respective orders. The
torque developed by the machine (see Figure 5.C) has the
E1 (t ) >U x1d (t )  U e1 (t )@2  >U x 2 d (t )  U e 2 (t )@2  >U x 3d (t )  U x 3 (t )@2
(18) same form as that of the active power, which is consistent with
E1 (t ) >U x1d (t )  U e3 (t ) @  >U x 2d (t )  U e1 (t )@  >U x3d (t )  U x 2 (t )@
2 2 2

the technique of separation of powers. Also, the zoom on the

E1 (t ) >U x1d (t )  U e3 (t ) @  >U x 2 d (t )  U e 2 (t ) @  >U x 3d (t )  U x 3 (t ) @
2 2 2

.........................................................................................................
E 27 (t ) >U x1d (t )  U e3 (t ) @2  >U x 2 d (t )  U e 3 (t )@2  >U x 3d (t )  U x 3 (t )@2
torque makes it clear that the torque has fewer oscillations.
Finally, Figure 5.d shows the phase current and its spectral
analysis. The analysis shows that the THD is 0.41% which
shows good performance tuning separate powers.

88 






a sd, sq ,rd and rq
Zoom
cb
d V. CONCLUSION
Fig. 6. Results for DFIG fed by matrix converter
V. CONCLUSION
The idea of providing a simple conversion structure based on a
matrix converter MC structure instead of a double conversion
inverter-rectifier and allowed to have a chain of high-
performance setting. To this end, the structure of separate
control active and reactive power has proved to be conclusive
and even for a unit power factor. In addition, the line current
has little harmonic distortion. Also, the advantage of this
method is the minimum loss of converter because the numbers
of switching are minimized. Therefore this converter can be
used as an active filter and show the effectiveness of the used
method and algorithm.

NOMOCLATURE
Ls, Lr Inductance of stator and rotor
M Mutual inductance
Vsd, Vsq, Vrd and Vrq Direct and quadrate voltages
Isd, Isq, Ird and Irq Direct and quadrate current
Direct and quadrate flow
Tem and Tr Electromagnetic and Load torque
Kf friction torque
 Rotational Speed
J Moment of inertia
P Number of pole pair
E r, E s Rotor and Stator Electromotive force
Nr, Ns Number of rotor and winding
 Dispersion coefficient
Kij (i=1,2,3) (j=1,2,3) Bidirectional switches
REFERENCES
[1] Fadaeinedjad, R. Moallem, M. Moschopoulos, G., Simulation
of a Wind Turbine With DFIG by FAST and Simulink, IEEE
Transactions on Energy Conversion, Vol. 23, No. 2, pp: 690
700, 2008.
[2] A. Boyette, Control and command a DFIG with storage system
for wind generation , PhD thesis, Nancy I, 2006.
[3] E. Erdem, Y. Tatar and S. Snter, Modeling and Simulation of
Matrix Converter Using Space Vector Control Algorithm,
EUROCON 2005, Serbia & Montenegro, Belgrade, November
22-24, 2005
[4] H. Altun and S. Sunter ; Matrix converter IM drive: modeling,
simulation and control, Electrical Engineering 86, pp. 2533,
2003.
[5] S. Ghoudelbourk, T. Bahi, H. Merabet, Study and control
powers of DFIG in a wind channel, ICE011, Adrar, 2011.
[6] T. Bahi, A. Dekhane, Y. Soufi, H, Hichem, Modeling and
Control of a Doubly-Fed Induction Generator, Proc. EVER11,
Monaco, 2011.
[7] S. Dendouga ,Control of active and reactive power of the
machine has dual power (DFIM), PhD thesis, University of
Batna, Algeria, 2010.
[8] P. S. Mayurappriyan , J. Jerome, M. Ramkumar and K. Rajambal
, Dynamic Modeling and Analysis of Wind Turbine Driven
Doubly Fed Induction Generator, International Journal of
Recent Trends in Engineering, Vol 2, No. 5, pp. 367-372, 2009.
[9] MM. Rezaoui, Study of output voltages of a matrix converter
feeding an five AC - induction Machine Using the strategy
calculated modulation PMW, Modern Electric Power Systems,
Wroclaw, Poland, 2010.
[10] J. Lettl and S. Fligl, Matrix Converter Control System,
Progress In Electromagnetics Research Symposium , China,
Vol. 4, 2005.
[11] S.H. Hosseini, E. Babaei, A New Control Algorithm for MC
Under Distorted and Unbalanced Conditions, Proc. IEEE
Conference on Control Application, Turkey, Vol. 2, pp. 1088-
1093, 2003.




 89
90