Enhancement of transient stability of DFIG wind

turbine using Active Disturbance Rejection controller

M. A. Chowdhury A. H. M. Sayem K. S. Islam W. X. Shen
Faculty of Science, Engineering and Technology
Swinburne University of Technology Hawthorn, VIC 3122, Australia
achowdhury@swin.edu.au, asayem@swin.edu.au, kislam@swin.edu.au, wshen@swin.edu.au

Abstract-- This paper presents an active disturbance rejection control technique in [8], a current sensing control in [9], and
(ADR) controller for the enhancement of transient stability of an advanced control approach [10], an internal model control
doubly fed induction generators (DFIGs). In real time estimation [11], and a predictive control approach in [12]. In order to
and mitigation of the total effect of various uncertainties, ADR improve stability of DFIG wind turbines, nonlinear control
controller is particularly effective against a wide range of
strategies have also been used [13].
parameter variations, model uncertainties, and large
disturbances. The performance of the designed controller is
demonstrated by large-disturbance simulations on a test system. However, these approaches do not consider the inherent
The controller successfully limits the oscillations, the peak value nonlinearity of power systems and interactions among wind
of the rotor current, and the dc-link voltage fluctuations and thus farms. This paper specifically considers this issue and
enhances transient stability of the DFIG wind turbine. proposes a solution based on active disturbance rejection
(ADR) controller [14]. This is an emerging control technology
Keywords—Doubly-fed induction generator; ADR Controller; that estimate and mitigate uncertainties, internal and external,
Short circuit fault; Transient stability in real time, resulting in a controller that does not require
accurate model information and is inherently robust against
I. INTRODUCTION structural uncertainties. Moreover, this is simple to implement

I N the past, most wind farms got disconnected whenever

there was a fault at the PCC or in the vicinity of the
wind farm. However, with the improvement in technology and
in practice.

This paper is organized as follows: Section II provides the

the introduction of DFIG units, wind farms are no longer mathematical modelling of the DFIG wind turbine. The test
required to be disconnected as they have an inherent capability system and control objectives are presented in Section III.
to control the active and reactive powers. This capability can Section IV describes the design strategy of the ADR
be utilized in designing a robust control to meet the current controller. In Section V, the performance of the controller is
grid code requirements. demonstrated through simulation results. Concluding remarks
are given in Section VI.
The stator of a DFIG is connected through a transformer
and the rotor is connected through a back-to-back power II. DFIG MODEL
converter to the grid. Due to partial-scale power converters, The dynamic models of the DFIG wind turbine used in this
wind turbines based on DFIGs are very sensitive to grid research work are presented in this section. The nonlinear
disturbances, particularly voltage dips during grid faults [1]. model of the wind turbines, made up of the various blocks
Faults in a power system, even far away from the location of a shown in Fig. 1, is based on a static model of the
wind turbine, can cause a voltage dip at the connection point aerodynamics, a two-mass model of the drivetrain, a third-
of the generator. Such an abrupt drop in voltage will cause order model of the generator, a grid side converter (GSC) with
overcurrent in the rotor windings and overvoltage in the dc a dc-link capacitor, a pitch controller, and a rotor side
buses of the power converters. Typically, a rotor’s current converter (RSC).
limit is 2 p.u., and the dc-link voltage limit is 1.2 times its
nominal value [2]. In addition, a voltage dip will cause The aerodynamic torque applied to the rotor by the
overspeeding of the wind turbine which will threaten its safe effective wind speed passing through the rotor is given as
operation. ρ
Tae = Awt c p (λ ,θ )v w3 (1)
A great deal of research has been carried out on transient 2ω m
stability enhancement of DFIG wind turbines. Several control where ȡ is the air density, Ȧ is the rotor speed Awt is the swept
schemes that involve hardware modifications [3], such as the area, cp is the pitch angle, ș is the pitch angle and Ȝ is the tip
crowbar approach [4], virtual resistance strategy [5], dynamic speed ratio.
braking resistance [6], and DFIG converter control, have been The drivetrain attached to the wind turbine converts the
proposed in the literature: an indirect matrix converter control aerodynamic torque Tae on the rotor into the torque on the
scheme in [7], a coordinated control approach in [2], a robust low-speed shaft, which is scaled down through the gearbox to

978-1-4799-4315-9/14/$31.00 2014
c IEEE 36
the torque on the high-speed shaft. A twoo-mass drivetrain vqr X m
'
vqr = (16)
model is used in this paper as it can satisfacctorily reproduce (X m + X r )
the dynamic characteristics of the wind tuurbine generator
system. The dynamics of the shaft is representted as [15] In Eqs. (5)-(16), E’ is the traansient voltage, X’ is the
⋅ transient reactance, X is the rotor op
pen-circuit reactance and Te
ωm =
1
[Tae − K s γ − Dmω m ] (2) is the electrical torque.
2H m
⋅ The dc-link dynamics are given by
ωG =
1
[K sγ − Te − DGωG ] (3)
⋅ 2
2H G vdc
Cv dc vdc = − Pr (t ) − Pg (t ) (17)
§
⋅ 1 · Rloss
γ = 2πf ¨ ω m −
ωG ¸ (4) where C is the DC-link capacitancce, vdc is the voltage of the
¨ Ng ¸
© ¹ DC-link capacitor and Rloss is the t total conducting and
where H is the inertia constant, Ks is the torsiion stiffness, Ȗ is switching losses of the converrter. Also, Pr(t) is the
the torsion angle, D is the torsion damping, Te is the electrical instantaneous input rotor power and d Pg(t) is the instantaneous
torque, f is the grid frequency and Ng is the gear ratio. output power of the GSC, which aree given by
Pr = v rd ird + v rq irq (18)
Pg = v gd i gd + v gq i gqq (19)

Using this model of a DFIG, a control strategy has been

proposed in the next section for a single
s machine infinite bus
test system.

III. TEST SYSTEM AND CONTROL TASK

A one-line diagram of the sin ngle machine (DFIG wind
turbine) infinite bus system is show
wn in Fig. 2. Controller is
designed for the wind turbine.

Fig. 1. Block diagram of a DFIG wind turbine system.

The transient model of a DFIG is describbed in this paper

by the following equations [16]:
⋅
s=
1
[Te − Tm ] (5)
2H G Fig. 2. Single machine infinite bus test system
m
⋅
'
Eqr =
1 '
To'
[ ( ) ]
Eqr − X − X ' ids − sω s E dr
'
− ω s vdr
'
(6) The GSC and RSC control strattegies are shown in Fig. 3.
The functions of the RSC control are to limit the rotor fault
⋅
[ ) ]
current and to increase the dam mping of stator flux and,
'
E dr =
1 '
To'
(
E dr + X − X ' iqs + sω s E qr
'
+ ω s vqr
'
(7) consequently, to enhance the system m stability. Once a voltage
dip is detected, the RSC operates too control active and reactive
vds = Rs ids − X qs
'
+ Edr
'
(8) powers. The RSC requires a consttant dc voltage to operate,
which is supplied by the GSC. The T GSC supplies the real
vqs = Rs ids + X qs
'
+ Eqr
'
(9) power demand of the RSC throu ugh the dc link. As GSC
operates to maintain a relatively constant
c voltage, RSC will
vt = vds
2
+ v qs
2
(10) receive almost constant power supp ply. RSC also controls the
where real and reactive power flows betw ween the grid and the GSC.
X ' = X s + X m X r (X m + X r ) (11) The q-axis current is used to controol the dc voltage, and the d-
axis current regulates the reactiv ve power or the terminal
X = Xs + Xm (12) voltage. idg and iqg are obtained d from ǻvdc and ǻvt by
L + Lm proportional–integral (PI) controlllers as shown in Fig. 3.
To' = r (13)
Rr Always, the priority belongs to the active power, so the
Checker block checks the availaable capacity for idg with
Te = E dr ids + E qr iqs (14)
respect to iqg while the rating of GSC is 25% of the DFIG. The
vdr X m a reactive powers with iqr
RSC controller controls the active and
'
vdr = (15)
(X m + X r ) and idr. Since the dc-link dynaamics are nonlinear, the
conventional linear control canno ot properly limit the dc

2014 IEEE 9th Conference on Industrial Electronics and Applications (ICIEA) 37

 voltage under severe voltage dips. In thee proposed ride- Y (s ) bm s m + bm −1s m −1 + .... + b1s + b0
G (s ) = = , n ≥ m (21)
U (s ) an s n + a n −1s n −1 + .... + a1s + a0
through approach, the DFIG contributes tto grid stability
enhancement by controlling both active and reactive powers
taking into account the nonlinearity of the DFIIG wind turbine. … m) are the coefficients of
where ai and bj (i = 1, …, n, j = 1, …,
G(s).

Fig. 4. ADR controller configuration

Dividing both sides of Eq. (20) by b G(s), we will have

(1 / G (s ))Y (s ) = U (s ) + W ' (s ) (22)
where W’(s) = W(s) / G(s). 1/G(s) can be conducted as
1 a s n + an −1s n −1 + .... + a1s + a0
= n m ,n ≥ m
G p (s ) bm s + bm −1s m −1 + .... + b1s + b0
= cn − m s n − m + bc n − m −1s n − m −1 + ..... + c1s + c0 + Gleft (s )
Fig. 3. Structure of the converter controller

The reference signals for the dc and terminnal voltages vdcref (23)
and vtref are set to a constant value (1 p.u.), inndependent of the where ci (i = 0, …, n – m) are coefficients of the polynomial
wind turbine operation mode. In fact, those vvalues are firmly division result, and the remainder Gleft(s) is
dependent on the size of the converter, thhe stator-voltage-
d m −1s m −1 + d m − 2 s m − 2 + .... + d1s + d 0
rotor-voltage ratio, and the modulation facttor of the power Gleft (s ) = (24)
converter. The reference set point Qgref for thee RSC can be set bm s m + bm −1s m −1 + .... + b1s + b0
to either zero or a certain value depending on whether the
DFIG is required to contribute with reactivve power. In this In Eq. (24), dj (j = 0, …, m – 1) are coefficients of the
paper, the performance of the controller is tessted for both zero numerator of the remainder. Subsstituting Eq. (23) into Eq.
and nonzero values. During normal operaation, the active (22), we have
power set point Pgref is defined by maximuum power point
tracking (MPPT) control. However, during ffaulted condition,
[c n−m s
n−m
]
+ bcn − m −1s n − m −1 + .... + c1s + c0 + Gleft (s ) Y (s )
(25)
it is determined by the damping controller w which damps out = U (s ) + W (s ) '
the torsional oscillations excited in the drrivetrain. Proper
placement and selection of input–output signaals for controllers where
allow control objectives to be met much more easily.
an
cn − m = (26)
IV. DESIGN OF ADR CONTROLLEER bm

Configuration of an ADR controller is sho

own in Fig. 4. Eq. (25) can be rewritten as
cn − m s n − mY (s ) = U (s ) −
A. Derivation of Transfer Function for DFIG G system [c n − m −1s
n − m −1
]
+ .... + c1s + c0 + Gleft (s ) Y (s ) + W ' (s )
(27)

The GSC or RSC with post-fault distturbance can be

represented by
Dividing both sides of Eq. (27) by cn-m, we will have
Y (s ) = G (s )U (s ) + W (s ) (20)
s n − mY (s ) = bU (s ) + D(s ) (28)
where U(s) and Y(s) are the input and output rrespectively, G(s)
is the transfer function (TF) between U(s) annd Y(s), and W(s) where b =1/cn-m, and a modified gen neralized disturbance is
is the generalized disturbance including unnknown internal
dynamics and external disturbances [2]. In Eqq. (20), the TF of
the converter G(s) can be described as

38 2014 IEEE 9th Conference on Industrial Electronics and Applications (ICIEA)

D(s ) = −
1
cn − m
[ ]
cn − m −1s n − m −1 + .... + c1s + c0 + Gleft (s ) Y (s )
§ n − m· i
β i = ¨¨ ¸¸ ⋅ ωo i=1,2,…..n-m
(29) © i ¹
1
W ' (s )
(32)
+
cn − m Therefore the observer gains can be changed through
tuning the unique parameter Ȧo, which is also the bandwidth
Eq. (28) will be taken as the system model for controller of the observer. With a well-tuned ESO, Zi(s) will be able to
design. estimate the value of Xi(s) closely (i = 1, …, n – m). Then we
have
∧
B. Design of Extended State Observer (ESO) Z n − m (s ) = D(s ) ≈ D (s ) (33)
The effectiveness of the ADR controller is dependent on the ∧
where D(s ) represents estimated D(s).
accurate estimation of the D(s) [2]. Consequently an ESO is
developed to estimate the disturbance in real time. This can be
achieved by augmenting the state variables of the system to C. Design of ADR controller
For the control input of U (s ) = (U o (s ) − Z n − m (s )) / b , the
include D(s). In order to construct the ESO, the system model
is rewritten as
sX (s ) = AX (s ) + BU (s ) + EsD(s )
original system will be reduced to a pure integral form. This
(30) process can be demonstrated by Eq. (34), where U0(s) is the
Y (s ) = cX (s ) control law for regulating the ACE output Y(s).
U (s ) − Z n − m (s )
In Eq. (28), s n − mY (s ) = b ⋅ o + D(s )
b (34)
∧
ª X 1 (s ) º ª0 1 0 .. 0º = U o (s ) − D (s ) + D(s ) ≈ U o (s )

« X 2 (s ) »
« » « »
«0 .. 1 .. ..»
X (s ) = « .. » A = «.. .. 0 .. 0» Our control goal is to minimize the effect of disturbance in
« » « »
« .. » «.. .. .. .. 1» output. A traditional PD controller can help to reach this goal.
« X n − m (s )» «0 0 .. .. 0»¼ (n − m )×(n − m ) So the control law U0(s) is chosen as Eq. (35), where R(s) is a
¬ ¼ (n − m ) ¬
reference input.
U o (s ) = k1 (R(s ) − Z1 (s )) − k 2 Z 2 (s ) − ... − k n − m −1Z n − m −1 (s ) (35)
ª0º
« » ª 0º To further simplify the tuning process, all the closed-loop
«..» «..» poles of the PD controller are set to –Ȧc. Then the controller
B = «0» E=« » C = [1 0 .. 0](n − m ) gains have to be selected as
« » «b»
«b» « » § n − m − 1· n − m − i
«¬0» ¬1 ¼ ( n − m ) ki = ¨¨ ¸¸ ⋅ ωc i=1,2,…..n-m-1 (36)
¼ ©n − m − i¹

We assume that D(s) has the local Lipschitz continuity and

sD(S) is bounded within domain of interests. Then the ESO is V. CONTROLLER PERFORMANCE EVALUATION
§ ∧ ·
sZ (s ) = AZ (s ) + BU (s )L¨¨ Y (s ) − Y (s )¸¸ The PD controller is of the first order [KP+sKI], and it has
© ¹ (31) the same input and output signals as those of the designed
∧
Y (s ) = cZ (s ) controller shown in Fig. 3. In practice, gains of the controller
are tuned by trial-and-error approaches by changing Ȧc,
where
sometimes starting with initial values given by, for example,
Z (s ) = [Z1 (s ) Z 2 (s ) Z n − m (s )]T(n − m ) and the Ziegler–Nichols method [19].
A simulation study is performed for the single machine
L = [β1 β2 β n − m ]T(n − m ) infinite bus system in Fig. 2 with the purpose of evaluating the
transient and voltage stability limits achievable with the
For stability purpose, all the eigenvalues of the ESO proposed and conventional PI-based vector control schemes.
should be located in the –Ȧo. So, the observer gains are chosen The critical clearing time and critical speed for a three-phase
as fault are 0.375 s and 1.329 p.u., respectively, with the
designed controller as compared with 0.291 s and 1.243 p.u.
with a properly tuned PI controller.

2014 IEEE 9th Conference on Industrial Electronics and Applications (ICIEA) 39

 In order to evaluate the performance of the designed Fig. 8 shows the rotor current, from which it is clear that the
controller, in enhancing transient stability of the DFIG wind peak rotor fault current at the instants of occurring and
turbine, detailed simulations are performed for a symmetrical clearing the fault is above the RSC tolerable threshold, i.e., 2
three-phase fault at one of the lines at 1 s which is p.u., with the PI controller. The proposed approach effectively
subsequently cleared after 150 ms. Figs. 5 and 6 show the real scales the peak rotor fault current down to 1.25 p.u., which is
power and terminal voltage responses of the DFIG wind well below the RSC tolerable limit. The proposed method
turbine with the proposed ADR control and the conventional also minimizes the rotor current fluctuations.
PI-based vector controller. From Fig. 5, it can be seen that the
PI controller results in an oscillatory post-fault behavior. The
designed controller provides improved performance in terms
of damped oscillations, lower settling time, and better
damping. The proposed method also helps PCC voltage to
reach to its prefault value and thus enhances the DFIG voltage
dip behavior as shown in Fig. 6.

Fig. 7. DC-link voltage of the DFIG wind turbine for a three-phase fault

Fig. 5. Real power of the DFIG wind turbine for a three-phase fault

Fig. 8. Rotor current of the DFIG wind turbine for a three-phase fault

VI. CONCLUSIONS
In this paper, a robust active disturbance rejection (ADR)
control has been proposed to enhance transient stability of the
DFIG wind turbine. A systematic procedure to design the
Fig. 6. PCC voltage of the DFIG wind turbine for a three-phase fault controller has been discussed. The control for a GSC has been
designed to stabilize both internal and external dynamics and
Next, a three-phase fault is applied at one of the lines at 1 s limit the dc-link voltage fluctuations. Also, the control for an
and cleared after 300 ms. Figs. 7 and 8 show the responses of RSC improves the damping of transients during a voltage dip.
the dc-link voltage and rotor current of the DFIG wind Simulation results show that, the controller successfully limits
turbine. From Fig. 7, by using the proposed dc-link control the oscillations, the peak value of the rotor current, and the dc-
technique, the fluctuation range of the dc-link voltage is link voltage fluctuations and thus enhances transient stability
reduced and limited to 1.06 p.u., while the dc-link voltage of the DFIG wind turbine.
with the conventional control in Fig. 7 reaches the maximum
value of 1.96 p.u. during the fault. Thus, the dc-link capacitor
would be under a large voltage stress and may be damaged.

40 2014 IEEE 9th Conference on Industrial Electronics and Applications (ICIEA)

 VII. REFERENCES
[1] O. Abdel-Baqi and A. Nasiri, “Series voltage compensation for DFIG
wind turbine low-voltage ride-through solution,” IEEE Trans. Energy
Convers., vol. 26, no. 1, pp. 272–280, Mar. 2011.
[2] M. Rahimi and M. Parniani, “Coordinated control approaches for
lowvoltage ride-through enhancement in wind turbines with doubly fed
induction generators,” IEEE Trans. Energy Convers., vol. 25, no. 3, pp. 873–
883, Sep. 2010.
[3] N. Joshi and N. Mohan, “A novel scheme to connect wind turbines to the
power grid,” IEEE Trans. Energy Convers., vol. 24, no. 2, pp. 504–510, Jun.
2009.
[4] S. Tohidi, H. Oraee, M. R. Zolghadri, S. Shao, and P. Tavner, “Analysis
and enhancement of low-voltage ride-through capability of brushless doubly
fed induction generator,” IEEE Trans. Ind. Electron., vol. 60, no. 3,
pp. 1146–1155, Mar. 2013.
[5] S. Hu, X. Lin, Y. Kang, and X. Zou, “An improved low voltage ride
through control strategy of doubly-fed induction generator during grid faults,”
IEEE Trans. Power Electron., vol. 26, no. 12, pp. 3653–3665, Dec. 2011.
[6] A. O. Ibrahim, T. H. Nguyen, D.-C. Lee, and S.-C. Kim, “A fault ride
through technique of DFIG wind turbine systems using dynamic voltage
restorers,” IEEE Trans. Energy Convers., vol. 26, no. 3, pp. 871–882, Sep.
2011.
[7] R. Pena, R. Cardenas, E. Reyes, J. Clare, and P. Wheeler, “Control of a
doubly fed induction generator via an indirect matrix converter with changing
DC voltage,” IEEE Trans. Ind. Electron., vol. 58, no. 10,
pp. 4664–4674, Oct. 2011.
[8] J. P. da Costa, H. Pinheiro, T. Degner, and G. Arnold, “Robust controller
for DFIGs of grid-connected wind turbines,” IEEE Trans. Ind. Electron., vol.
58, no. 9, pp. 4023–4038, Sep. 2011.
[9] S. Karimi, A. Gaillard, P. Poure, and S. Saadate, “Current sensor fault
tolerant control for WECS with DFIG,” IEEE Trans. Ind. Electron., vol. 56,
no. 11, pp. 4660–4670, Nov. 2009.
[10] S. Zhang, K. J. Tseng, S. S. Choi, T. D. Nguyen, and D. L. Yao,
“Advanced control of series voltage compensation to enhance wind turbine
ride-through,” IEEE Trans. Power Electron., vol. 27, no. 2, pp. 763–772, Feb.
2012.
[11] D. Campos-Gaona, E. L. Moreno-Goytia, and O. Anaya-Lara, “Fault
ride-through improvement of DFIG-WT by integrating a two-degreesof-
freedom internal model control,” IEEE Trans. Ind. Electron., vol. 60, no. 3,
pp. 1133–1145, Mar. 2013.
[12] C. Abad,M. A. Rodriguez, and J. Poza, “Three-level NPC converter-
based predictive direct power control of the doubly fed induction machine at
low constant switching frequency,” IEEE Trans. Ind. Electron., vol. 55, no.
12, pp. 4417–4429, Dec. 2008.
[13] A. Mullane, G. Lightbody, and R. Yacamini, “Wind-turbine fault
ridethrough enhancement,” IEEE Trans. Power Syst., vol. 20, no. 4, pp. 1929–
1937, Nov. 2005.
[14] Z. Gao, “Scaling and parameterization based controller tuning,” Proc. of
the American Control Conference, pp. 4989-4996, 2003.
[15] T. Ackermann, Wind Power in Power Systems. Hoboken, NJ: Wiley,
2005.
[16] Y. Lei, A. Mullane, G. Lightbody, and R. Yacamini, “Modeling of the
wind turbine with a doubly fed induction generator for grid integration
studies,” IEEE Trans. Energy Convers., vol. 21, no. 1, pp. 257–264, Mar.
2006.
[17] B. C. Pal, A. H. Coonick, and D. C. Macdonald, “Robust damping
controller design in power systems with superconducting magnetic energy
storage devices,” IEEE Trans. Power Syst., vol. 15, no. 1, pp. 320–325, Feb.
2000.
[18] Z. Gao, “Active disturbance rejection control: a paradigm shift in
feedback control system design,” Proc. Of the American Control Conference,
pp. 2399-2405, 2006.
[19] G. Mallesham, S. Mishra, and A. N. Jha, “Maiden application of Ziegler
Nichols method to AGC of distributed generation system,” in Proc. IEEE/PES
Power Syst. Conf. Expo., Seattle,WA, Mar. 15–18, 2009, pp. 1–7.

2014 IEEE 9th Conference on Industrial Electronics and Applications (ICIEA) 41