CONTROL DESIGN AND DYNAMIC PERFORMANCE ANALYSIS OF

A WIND TURBINE-INDUCTION GENERATOR UNIT

E.S. Abdin and W. Xu

Electrical Engineerkg Dept.,
University of Alberta,
Edmonton, Canada,
T6G 2G7

Abstract sources as an alternative to the use of synchronous

This paper presents the modeling and control design for a generators [5-61.
wind energy conversion scheme using induction generators.
The scheme consists of a three-phase induction generator In this paper, the dynamic modeling and control design
driven by a horizontal axis wind turbine and interfaced to for a horizontal axis wind turbine induction generator is
the utility thrdugh a double overhead transmission line. A presented. The effect of gating angle change of FC-TCR
static VAR compensator was connected at the induction static VAR compensator on the generator voltage is studied.
generator terminals to regulate its voltage. The mechanical For output regulation, state and output PI controllers are
power input was controlled using the blade pitch-angle. designed based on the linear model and tested under a
Both state and output feedback controllers are designed severe wind gust and large electrical disturbances.
using MATLAB software to regulate the generator output.
From the simulation results the response of closed loop 2-Wind-Generator System Configuration
system exhibited a good damping and fast recovery under Fig. 1 shows the wind-generator system configuration. The
different type of large disturbances. system consists of a horizontal axis wind turbine (HAWT)
and an induction generator (IG) interfaced to the utility grid
l-Introduction through a double overhead transmission line. A fixed-
The control of a system, that generates power fkom an capacitor thyristor-controlled reactor (FC-TCR) static VAR
unsteady input as the wind, presents a formidable problem. compensator is connected a t the generator terminals for
The wind speed varies from time to time due to gusts, and voltage regulation.
is fbrther disturbed by the effect of supporting tower
shadow. Attempts have been made to overcome this HAWT COBUS
problem for wind-synchronous generator systems by
keeping the rotor speed as nearly constant as possible using
the blade pitch angle control [l-21. A mathematical model
of a stand-alone wind energy conversion system with
battery energy storage was developed in Ref. 3. In this Ref.,
a permaqent magnet brushless synchronous generator was
used and the wind turbine was considered as the only Fig. 1 Wind-Generator System
source of power.
A study of the application of induction generators in 3-System Modeling
power system was given in Ref. 4 where the effect of 3-1 Wind turbine model
induction generator parameters and voltage support The wind turbine is characterized by nondimentional curves
requirements for stability under disturbance conditions were of the power coefficient C, as a hnction of both tip speed
studied. The advantages of induction generator versus ratio, ;1and the blade pitch angle, P. The tip speed ratio A. is
synchronous generators lie in lower cost and higher the ratio of linear speed at the tip of blades to the speed of
reliability. They have simpler control and auxiliary the wind. It can be expressed as follows [ 11:
equipment. The recent progress in the development of
controlled static VAR sources makes it possible to consider A=-aR
the induction generator coupled with controlled VAR vw
where R is the WT rotor radius, R is the mechanical angular
velocity of the WT rotor and V, is the wind velocity. For
the wind turbine used in the study, the following form
approximates Cp as a function of h and p [ 11:

0-7803-475441981$1Q.QQ 0 1998 IEEE -1198-

The output mechanical torque of the wind turbine, T, can component of the inductor current can be calculated by the
be calculated fiom the following equation [2]: following equation [ 5 ] :
1
T, = -pARC,V:
2 12.. (31
where p is the air density and A is the swept area by the
.
I, =
2(x -a)- sin[2(x -a)]
f l L
v, =- V,
XLgq
t 12)
blades
where X,, is the equivalent reactance of the TCR and a is
3-2 Induction generator model the gating firing angle. The differential equations for the d-
The wind turbine drives an induction generator whose flux and q-axis components of iL in synchonous reference
linkage model can be described in d- and q-axis as follows~
synchronous reference fiame as [7]:

Ppdr = Ob('dr - "ridr)+ (Ob - O m b q r (6) 3-ATransmission line model

The d and q equations, which describe the dynamics
Ppqr = O b q('r - k i q r )- ( O b - CL', bd? ('I introduced by the transmission line, are given as:

In the above equations, all the rotor variables are referred to

the stator side. The expression for the electromagnetic Vdr= vdb + RrLidTL - xTLiqTL + -
XTL
pioL (15)
torque in per unit can be written as: Ob

T, = P*iqs - Pqsi* (8) 5 s = 56 + RTLiqTL + X ~ ~ i d T

+ - ~PXi T
Ob
q TLL (16)

The wind turbine and induction generator rotors are 4-small signal ~ i ~ Model
~ ~ ~ i ~ ~ d
represented as a l'lmped One mass* the The design techniques used for the control systems are
equation of motion cim be written as: based on the state space linear model. The non-linear model
PO, = - :c) t9)
of the wind generator was linearized around an operating
point. The linearized model takes the following form
where H is the equivalent inertia constant of both wind
turbine and induction generator rotors. px=Ax+Bu+M

3-3 Static VAR comDensator model

In this paper, the static fixed-capacitor thyristor-controlled
reactor (FC-TCR) compensator was used to maintain a
constant voltage at the generator terminals by continuous
adjustment of the reactive power using the gating angle of
the thyristor. For the FC, the relation between current and
voltage in d- and q-iixis synchronous reference frame can
be given by [8]
5-State and OutDut Controller Design
Two paths are available for control of the wind turbine.
Static VAR control at the terminal of the induction
1 generator allows for regulation the terminal voltage to a set
v* = --pvgs + WbXciqc (1 1)
point. The second path is control of the blade pitch angle,
Ob
For the TCR, the full conduction is obtained with a gating which effects the mechanical input power to the generator.
tiring angle of 90". Tihe effect of increasing the gating firing To limit the mechanical stress in the wind turbine torsional
angle is to reduce the fundamental harmonic component of system, a rate limit C10 deg/sec should be put on the blade
the inductor current, reducing its reactive power as well as pitch actuator, which controls the mechanical power. The
its current. Using Fourier analysis, the fundamental electrical power and voltage regulations are achieved
through PI controllers. The controllers' design is based on

-1199-
the linearized model. Two integrators are applied on the wide range of operating conditions by regulating the gating
error signals (Pr,f -Pg) and (Vref-Vs). These two integrators firing angle. .A comparison between the nonlinear and
are augmented with the state space linearized model to give linearized full-order models is shown in Figures 3 and 4 for
the following augmented model: small pulse changes in the input signals p and a. A close
agreement between the two model responses is found.
pz = @Z + Tu +E& (18) Based on the linear model, the state feedback controller was
designed. After many iterations, the desired closed loop
where poles were selected as shown iri Table 1 .

Table 1

Open loop Closed loop

eigenvalues Eigenvalues wirh eigenvalues with
State feedback output feedback
controller controller
-24 4 kj2685.2 -24.4 kj2685.2
-29.4 fj1931.1 -29.4 kj 193 1.1 -32.5 fj 1933.1
-23.9 fj375.1 -23.9 fj375.1 -21 .O ij379.6
Based on the augmented model, a state feedback controller, -0 10 fj377.0 -40.0 fj377.0 -8.00 kj368.1
which takes the following form: -32.3 k j9.8 -32.3 k j9.8 -32.4 kj8.9
-0 60 -12.0 -13.1
0.0 -0.5 -0.4
0.0 -6.8 -0.9

can be designed using modem control design techniques. In By using the MATLAB software, the state feedback control
this paper, the pole-assignment technique was used. By matrix, K, was derived and given by
selecting the desired closed loop poles, K, can be obtained
using the MATLAB software.
The output feedback controller gains can be derived fiom K = - 34.75 - 1.294 76.89 11.69 - 22.06 1.024 - 0.488
0.009 0.029 1.00 0.521 0.058 0.002 0.001
the state feedback gains using Pseudoinverse matrix form.
The output vector y can be written as follows: - 11.85 - 13.45 - 6.473 - 1.623 191.5 - 241.75
-1001 -0117 0007 -0007 -7.711 -5.738
y=cz
The output vector was selected to be
If CcCis invertable, the state vector z can be obtained as a
fimction of output vector y as follows:

z = [ C T T c y = pi.v(C) y (21)
The output feedback gains were derived from K, and given
By defining the output feedback control law as by

and using Equations 19 and 2 1, &I can be derived from K,

1 1 667
0.109
- 7278
- 0.037
- 10905 - 22058 191 502 - 241 748
0.427 0058 - 7 7 1 1 - 5738 1
The closed loop poles with output controller are calculated
as follows;
and given in Table 1 . The state and output controller were
implemented in the nonlinear simulation model. A three-
KO= K , pinv(C) (23) phase short circuit at the generator terminals for 100 msec
following by switching off the faulted line was chosen to
6-Simulation Results compare the performance of the two controllers. The
Figure 2 shows 'the steady state variation of the stator response of the closed loop system under this disturbance is
voltage as a function of the thyristor gate firing angle, a for shown in Fig. 5. The terminal voltage and electrical power
different value of slips. As shown from the Figure, the were returned to its reference values in very short time and
stator voltage can be kept constant at rated value over a the deviation in rotor speed was reduced to a very small
value (about 0.7 rad./sec.). The response of the closed loop

-1200-
system with output controller is very close to that of state [7] I. Boldea and S.A. Nasar,” Electric Machine
feedback. In the practical, the output feedback is preferred Dynamics”, MACMILLAN PUBLISHMG
because all the output signals are available for measurement COMPANY, New York, 1986.
and there is no need for a full order or a reduced order [SI M.A. Rahman, A.M. Osheiba, T.S. Radwan and E.S.
observer as in the state feedback. Fig. 6 shows the response Abdin,” Modelling and Controller Design of an
of closed loop system with output controller to a severe Isolated Diesel Engine Permanent Magnet
wind gust. The initial wind speed is 28 mileh, near the Synchronous Generator’, IEEE Trans. on EC, Vol. .No.
rated value (24 mile/hr) and the peak of the gust at 42 , pp., June 1996.
milehr is near the cut out wind speed (45 mileh). From Appendix
the Figure, the closed loop response has a good damping Induction Penerator parameters
and the maximum terminal voltage and electrical power 4 poles, 6.6 Kv, 60 Hz.
excursions are sigliificantly smaller and limited to about X, =4.161 pu. , X1,=0.135 pu. , X1,=0.075 pu.
0.45% for voltage and 1.25% for power. The rotor speed R, =0.0059 pu. , R, =0.0339 pu. ,H,=1.975 sec.
deviation is a very ijmall and is equal to 0.25 rad./sec. Static VAR compensator parameters
XI- =4.0 PU. , Xc=3.8 PU.
7-Conclusions Transmission line parameters
The paper has bee? presented the modeling and controller XTL =o. 15 pu. , R~t=0.015PU. ,V ~ ~ 1 . 0 4
PU.
design for a wind turbine induction generator unit. The Wind turbine parameters
voltage regulation of induction generator has been achieved Horizontal axis wind turbine
using a fixed-capacitor thyristor controlled reactor static Rated power = 6 MW , R =200 ft. , GR -103.6
VAR compensator over a wide range of operating V, (rated) =24.0 Mile/Hr.
conditions. State and output PI controllers were designed V, (Cut in) =12.0 Mile/Hr.
and implemented in the nonlinear simulation model. A V, (Cut out) =45.0 Mile/Hr.
good damping pe>?formance has been achieved for the HT=16.72 sec.
closed loop system during severe wind gusts and large
electrical system disturbances.

Acknowledgement
E.S. Abdin thanks the Egyptian govemment for the
financial support itnd the university of Alberta for the
equipment facilities.

References
[ l ] J.R. Winkelman and S.H. Javid, ”Control Design and
-z
,# 0.96 --.-*--
*.--
~ - .-.
-1!
Performance Analysis of a 6 MW Wind Turbine-
Generator”, IEEE Trans. on PAS, Vol. 102,No. 5, pp. I

1340-1347, May 1983.

[2] 0. Wasynczuk. D.T. Man and J.P. Sullivan,”Dynamic
Behavior of 21 Class of Wind Turbine Generators
During Random Wind Fluctuations”, IEEE Trans. on
PAS, Vol. 100, No. 6, pp. 2837-2845, June 1981.
[3] B.S. Borowy and Z.M. Salameh, ”Dynamic Response
of a Stand-Alone Wind Energy Conversion System
with Battery Energy Storage to a Wind Gust”, IEEE
Trans. on EC, Vol. 12, No. 1, pp. 73-78, March 1997.
[4] F.P. de Mellc,J.W. Feltes, L.N. Hannett and J.C. -P 4,0400 ~
, --
;-
Nonlinear model

White,” Application of Induction Generators in Power -“ i

System”, IEEE Trans. on PAS, Vol. 101, No. 9, pp.
U
E !
Y -0.0800 i
3385-3393, 1982. 4
[5] T.J.E. Miller, ”Reactive Power Control in Electric
-0.1200 J

System”, A Wiley-Interscience Publication, John

Wiley &sons, 1982.
[6] Y.H. Song, A.T. Johns and R.K. Agganval, “Nonlinear Fig. 3 Comparison between nonlinear model
Thyristor-Controlled Static VAR Compensation”, The
and fill-order linearized model for
European Power Electronics Association, Brighton,
pp.56-60, Sept. 13-16, 1993. small pluse change in p

-1201 -
 Time (sec.) 0.8080 -
0 05 0 10 0 15 0 20
-
$
0.8Ow -

'a 0.7920 ~

Nonlinear mdel
4 -0002 Full order linearized model
0.78404
e
U
-0004 0.0 5.0 10.0 15.0 20.0

3 -0006 \ Time (sec.)

-0008 J

0.9940

Fig. 4 Comparison between nonlinear model

and full-order linearized model for
-
iz
0.9920

5 0.9900
small pluse change in a
0.9880 4
0.0 5.0 10.0 15.0 20.0
Time (sec.)

387.5 388.60 7
?
387.0 8 388.50 -
1 .o
00 05 1.5
3 388.40 -
Time (sec.)
2e 380.30 -
388.20 4
2.000 1
0.0 5.0 10.0 15.0 20.0
Time (sec.)

0.0 0.5 1.0 1.5

25.000 4
0.0 5.0 10.0 15.0 20.0
Time (sec.)

15.000 -

2 10.000 -
m
P
5.000 -
0.000 J
0.0 5.0 10.0 15.0 20.0
Time (sec.)
Fig. 5 Response of the closed loop system for 3-4
short circuit following by switching off
1.680
the faulted TL (Tt-100 msec.) 1.660
1.640
1.620
8 1.600
1.580

0.0 5.0 10.0 15.0 20.0

Time (sec.)

Fig. 6 Response of CLS with output controller for a gust in

vw

-1202-