Electrical Power and Energy Systems 79 (2016) 172–187

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

A new control strategy of WFSG-based wind turbine to enhance the LVRT

capability
Khalfallah Tahir a,⇑, Cheikh Belfedal a, Tayeb Allaoui a, Champenois Gérard b
a
Department of Electrical Engineering, L2GEGI Laboratory, Ibn Khaldoun University, Tiaret, Algeria
b
University of Poitiers, Laboratoire d’Informatique et d’Automatique pour les Systèmes, Bâtiment B25, 2, rue Pierre Brousse, 86022 Poitiers, France

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

Article history: This paper proposes a competent and effective scheme to enhance the low voltage ride through (LVRT)
Received 23 September 2014 capability of wound field synchronous generator (WFSG) based wind turbines (WTs) under unbalanced
Received in revised form 17 December 2015 voltage dip conditions. A technique for grid synchronization against voltage excursions, i.e., a PLL using
Accepted 2 January 2016
positive grid voltages with a high selectivity filters (HSFs) is utilized to extract a robust grid voltage syn-
chronization signal irrespective of the mains condition to enhance the overall system performance.
Besides, a new controller (adaptive fuzzy RST) for both the stator side converter (SSC) and grid side con-
Keywords:
verter (GSC) current regulation are employed to further improve dynamic performance. Also, a reactive
Wind energy conversion system (WECS)
Wound field synchronous generator (WFSG)
power support scheme to manage the WFSG reactive power during contingencies and fulfill the grid
Adaptive fuzzy RST controller codes obligations is presented. Moreover, an additional device such as braking chopper (BC) circuit is
Low voltage ride through (LVRT) used in the DC-link circuit for stable operation of the wind energy conversion system (WECS) under-
PLL line fault. Effectiveness of the proposed control strategy is verified by the numerical simulations.
Braking chopper (BC) Ó 2016 Elsevier Ltd. All rights reserved.

Introduction tion generator (DFIG) wind systems, a (WECS) based on wound

field synchronous generators (WFSGs) with a full power converter
Among renewable energy sources, wind energy generation has has a lot of advantages such as high power density, because it
been noted as the most rapidly growing technology; being one of employs the whole stator current for the electromagnetic torque
the most cost-effective and environmental friendly means to gen- production [5], provides more extensive speed operating range,
erate electricity from renewable sources. The increasing penetra- and full decoupling between the generator and the grid, which
tion level of wind energy can have a significant impact on the results to enhanced capability to fulfill the LVRT requirement.
grid, especially under abnormal grid voltage conditions [1]. In These properties make this implementation interesting [6].
order to avoid this problem, an increasing number of power system Quite a few studies have been carried out to improve the LVRT
operators have implemented technical standards known as grid capability for WT systems. Among the available approaches, pro-
codes that wind turbines (WTs) must meet when connecting to tection devices such as energy storage system (ESS) or flywheel
the grid [2,3]. Generally, these grid codes requirements cover many systems are required to mitigate the output power fluctuation
topics such as voltage operating range, power factor regulation, [7–9]. Using the external devices increases the cost of the whole
and frequency operating range, grid support capability, and low system. With the ESS added to wind generation systems, not only
fault ride-through (LVRT) capability. Indeed, grid codes dictate the power smoothening but also FRT can be achieved effectively
fault ride-through (FRT) requirements. LVRT capability is consid- [8,9]. However, the cost of the ESS is too high to solve this problem
ered to be the biggest challenge in (WTs) design and manufactur- practically. One of the simplest methods for an LVRT is to apply a
ing technology [4]. LVRT requires (WTs) to remain connected to the braking chopper (BC) [10,11]. However, it is impossible to improve
grid in the presence of grid voltage sags. the output power quality of the WT systems since the BC just dis-
Variable-speed wind energy systems are currently preferred sipates the power and cannot return it to the grid.
than fixed-speed (WTs) due to their superior wind power extrac- A control scheme to improve LVRT capability of the variable
tion and better efficiency. Differently from the doubly fed induc- speed WT under the unbalanced and distorted grid voltages is pre-
sented in this paper. Many kinds of methods for LVRT under the
distortedly unbalanced grid voltages have been presented. PLL
⇑ Corresponding author. using positive grid voltages to control the grid side converter
E-mail address: [email protected] (K. Tahir).

http://dx.doi.org/10.1016/j.ijepes.2016.01.008
0142-0615/Ó 2016 Elsevier Ltd. All rights reserved.
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 173

(GSC) under the unbalanced grid voltages [12], PLL using 120 Hz
100%
Notch filter to filter the negative current components [13], and
90%

Line-to-line voltage
dual currents controller which controls each positive components
and negative components of the currents [14] belong to this. But, 70%

(U/UN)
the major disadvantages of dual sequence control include its Limit line 1
extensive measurements, complicated computation for the refer- 45% Limit line 2
ence current values and the usage of low pass filters (LPFs) for
sequence component separation. These filters contribute excessive
time delays and can deteriorate the control performance [15].
In this work, PLL using positive grid voltages to control the GSC
under the unbalanced grid voltages is proposed. In this novel
Fig. 1. Voltage limits curves to allow generator disconnection.
scheme the high selectivity filters (HSFs) have been utilized
instead of the classical extraction filters, e.g., LPFs. Moreover, this
algorithm works effectively not only under balanced supply volt-
ages, but also under distorted or unbalanced conditions.
To improve control dynamic response, proportional integral (PI)
plus resonant (R) current regulator [16] and proportional resonant
(PR) current regulator [17] are employed in a stator stationary ref-
erence frame. But, this kind of control could be easily saturated
when dealing with substantial sag. Moreover, they are sensitive
to the generator parameters and other phenomena such as distur-
bances and unmodeled dynamics [18]. Therefore and in this partic-
ular context, this paper proposes the use of adaptive fuzzy RST
controller as an improved solution that handles the classical
controllers.
To check the overall control strategy ride-through performance,
simulations using MATLAB/SIMULINK on a 7.5 kW WFSG-based Fig. 2. Reactive current to be fed under a voltage dip.
WT are carried out in case of varying wind speed and unbalanced
voltage sags. Simulation results demonstrate that the presented
WFSG-WT model possesses desirable capabilities of operation at
[19,20]. E.ON introduces a voltage profile and the limiting curves
the maximum power point. The DC-link overvoltage protection
and regions defining the LVRT requirement as shown in Fig. 1.
method is effective in protecting the converter from over voltage
Accordingly, when the grid voltage is within the shaded area and
damage and thus enhancing the LVRT capacity under the grid fault
above the limit line 2 in Fig. 1, LVRT is required for remain the
condition as well. And reactive power support can also act as a
power generation plant connected to the grid. But in case of insta-
desirable means to enhance LVRT capability by enhancing the grid
bility a short term interruption (STI) is allowed. Below the limit
voltage during the grid fault event.
line 2 in Fig. 1, no LVRT is required and STI from the grid is always
permissible. Moreover, the WECS has to provide the amount of
Wind conversion system description reactive current specified in Fig. 2, to support the grid voltage
recovery during voltage dips. According to this figure, it has pro-
A simplified diagram of the power system based wind power vided 2% of the reactive current for each 1% of the voltage dip,
generation is illustrated in Fig. 3. It consists of a WT, a gearbox, a when the grid voltages are in the range of 50–90%, and has pro-
WFSG, and back-to-back (B2B) converters. The rotor winding of vided 100% of the reactive current when the voltage falls below
the WFSG is connected to DC bus by DC/DC converter, whereas 50%.
the stator winding is fed by B2B bidirectional the Pulse Width
Modulation Voltage Source Converter (PWM-VSC). The B2B con- Control of machine-side converter
verter contains a stator side converter (SSC) and a GSC, which are
connected by a DC bus. A BC is connected in parallel with the Wind turbine modeling
DC-link. The BC will be activated to dissipate the excessive power
in case of deep voltage sags. The control system of the WECS con- In the WT model, the aerodynamic power output is given as
sists of the generator-side control sub-system and a grid-side con- [21].
trol sub-system. The Maximum Power Point Tracking (MPPT)
1
algorithm is based on fuzzy logic and controls the generator side Pa ¼ qpR2 V 3w C p ðk; bÞ: ð1Þ
converter. The GSC controller maintains the dc-link voltage at the 2
desired value by exporting active power to the grid and it controls where q is the air density; R is the WT blade radius; Vw is the wind
the reactive power exchange with the grid. The SSC controls the velocity (m/s), and Cp is called the power coefficient, and is given as
power flow from the WFSG to the grid via the control of the stator a nonlinear function of the tip speed ratio (TSR) k with
currents of the direct and quadrature components of the WFSG sta-
RXt
tor current to achieve decoupled control of the active and reactive k¼ : ð2Þ
Vw
powers. The quadrature component controls the active power,
whereas the direct component controls the reactive power. where Xt is the turbine speed. Cp is a function of the TSR k and the
blade pitch angle b, and is general defined with
8
 18:4
LVRT requirement >
< C p ¼ 0:73 151 
 0:58b  0:002b2:14  13:2 e ki
ki
ð3Þ
Among the different grid codes of LVRT, the most widely >
: ki ¼ 1
1
0:003
k0:02b b3 þ1
adopted one is the E.ON code which is proposed by Germany
174 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187

Fig. 3. Configuration of a WFSG wind turbine system with the BC.

Pa
Ta ¼ ð5Þ
Xt
The WT shaft is connected to the WFSG rotor through a gearbox
which adapts the slow speed of the turbine to the WFSG speed.
This gearbox is modeled by the following equations:
Xm
Xt ¼ ð6Þ
G

Ta
Tm ¼ ð7Þ
G
From the dynamics fundamental relation, the mechanical gen-
Fig. 4. Power coefficient versus tip speed ratio. erator speed is determined as follows:
dXm
J ¼ T m  T em  f Xm ð8Þ
dt
where J and f are the total moment of inertia and the viscous friction
coefficient appearing at the generator side, Tm is the gearbox torque,
Tem is the generator torque.

WFSG modeling

The simplified scheme of the synchronous generator with dam-

pers at the rotor is shown in Fig. 6. The generator is modeled by the
following voltage equations in the rotating d–q reference frame
[22]:
(
v ds ¼ rs ids  xe uqs þ dtd uds
ð9Þ
Fig. 5. Rotor power versus rotational speed of generator. v qs ¼ rs iqs þ xe uds þ dtd uqs
8
< v f ¼ r f if þ dt uf
d
>
By using (3), the typical Cp versus k curve is shown in Fig. 4. In a WT,
there is an optimum value of tip speed ratio kopt that leads to max- 0 ¼ r D iD þ dtd uD ð10Þ
>
:
imum power coefficient C p max . When the TSR in (2) is adjusted to 0 ¼ r Q iQ þ dtd uQ
its optimum value kopt ¼ 8:1 as well as the power coefficient reach-
where:
ing C p max ¼ 0:48; the control objective of the maximum power
– iD, iQ: direct and transverse dampers currents.
extraction has been achieved. From (1) and (2), we get
– uD , uQ : direct and transverse dampers total flux.
1 – uf : main field total flux.
Pa max ¼ pqC p max R
5
X3t opt ð4Þ
2k3opt – rs: stator resistance, rf: main field resistance, rD and rQ: dampers
resistances.
This equation shows the relationship between the turbine – v ds ; v qs ; ids ; iqs ; uds ; uqs : d and q components of stator voltages,
power and turbine speed at maximum power output. Fig. 5 shows
currents and flux linkage respectively.
the power rotational speed curves of a 10 kW wind turbine consid-
– vf, if : main field excitation voltage and main field current,
ered in the present paper under different wind speeds. The dotted
respectively.
line indicates the optimal power points P a max where the Cp coeffi-
- xe : is the electrical angular speed.
cient is kept at its maximum value.
The turbine torque is expressed by the ratio of the aerodynamic xe ¼ pXm ð11Þ
power to the turbine speed as follows:
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 175

Current vector control

The current vector control of the machine-side converter con-

tains a faster current inner loop and a slower speed outer loop.
The d-axial current is only affected by vqs. Meanwhile, the q-axial
current is affected by vds [24]. Therefore, the current inner loop is
decoupled. The current inner loops of d and q axis jointly adjust
the WT speed and regulate the d-axial current to zero to enhance
the WFSG efficiency.
The decoupled control strategy of d and q current loops is
obtained by rewriting (9) and (12) as
 
did dif diD
v ds ¼  r s ids þ Ld þ msf þ msD þ Lq xe iqs  msQ xe iQ
dt dt dt
ð17Þ
 
diqs diQ
v qs ¼  r s iqs þ Lq þ msQ  Lq xe ids þ msf xe if þ msD xe iD
Fig. 6. Wound field synchronous generator with dampers. dt dt
ð18Þ

where p is the pole pair number and Xm is the mechanical speed of The dampers occur whenever the rotary field is varied (fast load
WFSG. change, pulsation in the torque of the machine). Induced currents
The WFSG magnetic equations give the flux expressions as arise in the absorber and produce a torque that dampens oscilla-
follows: tions and maintains the synchronism.
( During the steady state (constant speed, constant torque) the
uds ¼ Lds ids þ msf if þ msD iD two current iD and iQ are zero (no relative movement between
ð12Þ
uqs ¼ Lqs iqs þ msQ iQ the rotating field winding and dampers). Therefore, Eqs. (17) and
(18) can be reduced as follows:
8  
< uf ¼ Lf if þ mfD iD  msf ids
> dids
uD ¼ LD iD þ mfD if  msD ids ð13Þ
v ds ¼  r s ids þ Ld þ Lq xe iqs  msQ xe iQ ð19Þ
dt
>
:
uQ ¼ LQ iQ  msQ iqs  
diqs
where: v qs ¼  r s iqs þ Lq  Lq xe ids þ msf xe if þ msD xe iD ð20Þ
dt
– LD, LQ : inductances of the direct and quadrature damper
windings. where the item in the bracket of (19) and (20) is treated as the state
– Lf: inductance of the main field winding. equation between the voltage and current in the d and q loops
– Lds, Lqs: inductances of the d-axis stator winding and q-axis sta- respectively, and the others items are treated as compensation or
tor winding. disturbance terms. This treatment assumes that vds in (19) has no
– msf: mutual inductance between the field winding and the major influence on iqs and vqs that in (20) has no significant effect
d-axis stator winding. on ids.
– msD: mutual inductance between the d-axis stator winding and
the d-axis damper winding. The fuzzy MPPT controller
– msQ: mutual inductance between the q-axis stator winding and
the q-axis damper winding. Since that the wind profile in unknown, a simple MPPT con-
– mfD: mutual inductance between the field winding and the troller based on fuzzy logic is proposed to extract maximum power
d-axis damper winding. from the wind turbine. The fuzzy rules are based on the variation of
the wind power DP a , and the rotational speed of the WFSG ðDXm Þ.
The electromagnetic torque is expressed by: The block diagram for the power optimization by fuzzy logic is
shown in Fig. 7.
T em ¼ pðuds iqs  uqs ids Þ ð14Þ
The control rules are indicated in Table 1: with ðDP a Þ and ðDXm Þ
as inputs, while ðDXm ref Þ represent the output. The membership
Power converter model functions for inputs output variables are shown in Fig. 8. The lin-
guistic terms used for input output membership functions are
The converter model with ideal switches is expressed by the labeled as GN (grand negative), N (negative), ZR (zero), P (positive),
two following equations: GP (grand positive)). From these linguistic rules, the FLC proposes a
0 1 0 10 1 variation of the reference speed ðDXm ref Þ according to:
V an 2 1 1 Sa r 8
B C V dc B CB C < DPa ¼ Pa ½k  Pa ½k  1
>
@ V bn A ¼ @ 1 2 1 A@ Sb r A ð15Þ
3 DXm ¼ Xm ½k  Xm ½k  1 ð21Þ
V cn 1 1 2 Sc r >
:
Xm ref ½k ¼ Xm ½k  1 þ DXmref ½k
ir ¼ Sa r ias þ Sb r ibs þ Sc r ics ð16Þ
dc
where Pa[k] and Xm ½k are the wind power and rotational speed at
With Vdc and ir dc are the direct bus voltage and current respec- sampled times (k), and Xm ref ½k is the instant of reference speed
tively. The switching states ðSa r ; Sb r ; Sc r Þ are generated using the [25].
proposed control method, or alternatively using vector control/ As explained previously, the FLC optimizes the reference speed
Direct Torque Control (DTC) [23]. Xm ref for maximum power tracking. This speed represents a
176 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187

Fig. 7. Structure of MPPT fuzzy controller.

Table 1
Fuzzy rules. Design methodology of the AF-RST controller. The AF-RST controller
applies fuzzy control rules to modify RST parameters on-line using
DXm DP a
error e and change-in-error ce as inputs, which can meet the
GN N ZR P GP request of e and ce in different time for self-tuning RST parameters.
GN GN GN GN N ZR Its structure is shown in Fig. 10.
N GN N N ZR P
ZR N N ZR P P
RST regulator synthesis. RST controller is a polynomial controller
P N ZR P P GP
GP ZR P GP GP GP based on the robust pole placement theory, it has the advantage
of solving the tradeoff between rapidity and performance, com-
pared with proportional integral PI controller. The block diagram
of WFSG current control using an RST controller is shown in
Fig. 11, where KPWM is the gain of the power electronic converter,
Ts is equal to one switching period, and R, S, T are polynomials
which constitute the controller. Besides, eds and eqs as the distur-
bances are the d and q axial induced voltages of WFSG,
respectively.
The closed-loop transfer-function of the controlled system is
BT BS
idqs ¼ idqs ref þ edqs ð23Þ
AS þ BR AS þ BR
where A and B are presently defined as follows:
Fig. 8. Membership functions of e, ec and DU.
( Ldq
A ¼ a1 s þ a0 ¼ rs
sþ1
ð24Þ
positive input (reference) of the PI anti-windup controller, which B ¼ b0 ¼ 1
rs
performs the speed control in steady state. The PI anti-windup loop
S(s) and R(s) appear in the denominator of (23); therefore, their
operates with a fast rate and provides fast response and overall
parameters cab be obtained by solving the Bezout equation defined
system stability.
by

Design of machine-side controller DðsÞ ¼ AðsÞSðsÞ þ BðsÞRðsÞ ð25Þ

The choice of the polynomials orders is an important step in the
The block diagram of the d and q inner current loops of the design of the RST controller. It is a compromise between perfor-
machine-side controller is shown in Fig. 9, which has the identical mance and complex controllers. We choose a strictly proper con-
type and also has the identical corresponding controller. Consider- troller in order to achieve a good accuracy [27,28]. So, if A is a
ing the q-axis loop, the compensator is used adaptive Fuzzy RST polynomial of order n (deg(A) = n), we must have
(AF-RST) controllers. 8
The reference q-axial current is determined from the reference < degðDÞ ¼ 2n þ 1
>
torque as follows: degðSÞ ¼ degðAÞ þ 1 ð26Þ
>
:
degðRÞ ¼ degðAÞ
T em ref
iqs ref ¼ ð22Þ
pðmsf if þ msD iD Þ Thus, the polynomials D(s), R(s) and S(s) can be written as
8
< DðsÞ ¼ d3 s þ d2 s þ d1 s
> 3 2

Design of the AF-RST controller SðsÞ ¼ s2 s þ s1 s

2 ð27Þ
>
:
Why adaptive fuzzy of the RST controller?. It is well known that the RðsÞ ¼ r1 s þ r 0
WT model characteristics change dynamically. These changes drive
According to the robust pole placement strategy, the polyno-
the process in different operation points. From the standpoint of
mial D(s) is written as
control performance, the RST controller does not yield reasonable
   2
performance over a wide range of operating conditions because 1 1
of the fixed RST gains. This is the reason to use the fuzzy logic to DðsÞ ¼ CF ¼ sþ  sþ ð28Þ
Tc Tf
tune the parameters of RST controller for real-time [26].
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 177

Fig. 9. Control block diagram of the WFSG.

Fig. 10. Structure of the adaptive fuzzy RST controller.

With pc ¼  T1c pole of polynomial order C and pf ¼  T1 double of The identification between (28) and (29) makes it possible to
f

the polynomial filter F. obtain the system of four equations with four unknown terms
8
The pole pc must accelerate the system and is generally chosen >
> a1 s2 ¼ 1
>
>
2–5 times greater than the pole of A. pf is generally chosen 3–5 >
< a1 s1 þ a0 s2 ¼ T f þ T c
> 2 1
times smaller than pc.
a0 s1 þ b0 r1 ¼ T12 þ T c2T ð30Þ
By replacing A(s) and B(s) from (24) and (27) in the Bezout >
>
>
> f f
equation (25), we find >
>
: b0 r 0 ¼ T 1T 2
c f
DðsÞ ¼ a1 s2 s3 þ ða1 s1 þ a0 s2 Þs2 þ ða0 s1 þ b0 r 1 Þs þ b0 r0 ð29Þ
178 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187

Fig. 11. RST controller system.

Fig. 13. Membership functions of DS1 ; DR1 ; DT 0 .

Table 2
Fuzzy rule of adaptive fuzzy RST controller.

DS1 ce NB NM NS ZO PS PM PB
DR1
e DT 0
NB B/Z/S S/S/M M/M/Z M/B/M M/M/Z S/S/M B/Z/S
NM B/Z/S M/S/M B/B/S M/B/M B/B/S M/S/M B/Z/S
NS B/Z/S M/Z/M B/M/S B/B/M B/M/S M/Z/M B/Z/S
ZO B/Z/S M/Z/M S/M/S Z/B/M S/M/S M/Z/M B/Z/S
PS B/Z/S M/Z/M B/M/S B/B/M B/M/S M/Z/M B/Z/S
Fig. 12. Membership functions of e and ce. PM B/Z/S M/S/M B/B/S B/B/M B/B/S M/S/M B/Z/S
PB B/Z/S S/S/M M/M/Z M/B/M M/M/Z S/S/M B/Z/S

T(s) can be a constant that guarantees zero steady-state errors. By

using (23) and considering that S(s) is zero in steady state, T(s) 8 di
can be obtained as follows: < dtdg ¼ xiqg  RL f idg þ L1 v df  L1 v dgþ
ð33Þ
f f f

TðsÞ ¼ t 0 ¼ r 0 ð31Þ : diqg ¼ xi  Rf i þ 1 v  1 v

dt dg L qg L f
qf L qgþ
f f

Choosing fuzzy controller inputs and outputs. The AF-RST controller where vdf, vqf are the voltages generated at inverter terminals in the
has two inputs and three outputs. Input variables are e and ce; out- grid d–q frame; v dgþ , v qgþ are the positive sequence grid voltage in
put variables are the proportional gain DS1 , the integral gain DR1 , the grid d–q frame; idg, iqg are the grid current in the grid d–q frame;
the derivative gain DT 0 , so the equation of the final values of three Lf is the filter inductance; Rf is the filter resistance; x is the grid
RST parameters are in [29]: angular frequency.
8 The instantaneous active and reactive power delivered to the
0
< S 1 ¼ S1 þ D S 1
> grid under unbalanced conditions are expressed in terms of the
R1 ¼ R01 þ DR1 ð32Þ positive sequence component as
>
:
T 0 ¼ T 00 þ DT 0 (
Pg ¼ v dgþ idg þ v qgþ iqg
ð34Þ
where, S1, R1, T0 are the certain value of last sampling time. Three Q g ¼ v qgþ idg  v dgþ iqg
certain parameters of RST can be automatically adjusted by the
state of the controlled object. The domain of e, ce is where the superscripts ‘‘+” is the positive sequence component.
f6; 5; 4; 3; 2; 1; 0; 1; 2; 3; 4; 5; 6g and DS1 ; DR1 ; DT 0 is
f0; 1; 2; 3; 4; 5; 6g. The fuzzy subset of inputs linguistic variables Synchronization algorithm
are expressed as negative big (NB), negative medium (NM), negative
small (NS), zero (ZO), positive small (PS), positive medium (PM), The Synchronization algorithm for achieving a controllable
positive big (PB) and of outputs as zero (Z), small (S), medium power factor in the connection must detect the phase angle of
(M), big (B). the 3-phase grid voltages with optimal dynamic response. Various
The membership functions for inputs and outputs are shown in phase locked-loop (PLL) techniques have been developed for three
Figs. 12 and 13. phase system applications. The most widely used one is the syn-
All the fuzzy inference rules are summarized in Table 2. chronous rotating frame PLL (SRF PLL); however, its performance
in a case of grid voltage distortion or unbalance is not acceptable
Control of grid-side converter [30]. The improved PLL shown in Fig. 14 is presented for the pro-
posed control scheme, where a Secondary Order Generalized
Model of the grid side Integrator-Quadrature Signal Generator (SOGI-QSG) algorithm is
adopted as a SSM for extracting the positive and negative
The GSC has to deal with the grid dips. In this work, the vector sequences in the ab frame. By adding a high selectivity filter after
current controller is implemented on the positive sequence, and the block of SOGI-QSG, the impact of distorted grid voltage is
there is no need to develop the negative-sequence model for the removed.
GSC. The sequence separation method (SSM) is used only to obtain A block diagram of a SOGI-QSG algorithm is depicted in Fig. 14
the symmetrical components of the grid voltages. Therefore, the as subsystem, and its transfer functions are described by (35) and
model for the GSC is shown in (33). (36), respectively.
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 179

Fig. 14. The scheme of the improved PLL.

Fig. 15. Control block diagram of the grid-connected PWM converter in normal operation and under grid dip (LVRT fulfillment required).

where k is the gain of the SOGI-QSG and DQ is the quadrature

signals [31].
Two SOGI-QSG blocks working on the ab stationary reference
frame are combined to obtain a Double Second Order Generalized
Integrator based on Quadrature Signal Generator (DSOGI-QSG)
structure for 3-phase systems [31], as shown in Fig. 14.
The Positive Sequence Calculation (PSC) Block contains the
equations presented in (37) and (38), which supplies the positive
sequence voltage of the input voltage, by means of arithmetic
Fig. 16. Hysteresis control of the active crowbar. operation.
v0 kxs 1
DðsÞ ¼
v
ðsÞ ¼ 2
s þ k xs þ x2
ð35Þ v aþ ¼ ðv a  qv b Þ ð37Þ
2

qv 0 k x2 1
Q ðsÞ ¼ ðsÞ ¼ ð36Þ v bþ ¼ ðqv a þ v b Þ ð38Þ
v s2 þ k xs þ x2 2
180 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187

Fig. 17. Wind generation performance for normal grid condition. (a) Wind speed ([m/s]). (b) Generator speed ([rad/s]). (c) Generator torque ([N m]). (d) dq-axis generator
current ([A]). (e) Grid abc currents ([A]). (f) Stator current and voltage (phase a) ([A], [V]). (g) DC-link voltage ([V]). (h) Active and reactive grid power ([W], [VAR]). (i) Grid
current and voltage (phase a) ([A], [V]).
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 181

Fig. 17 (continued)

Fig. 18. Frequency spectra under normal conditions of (a) stator current (b) grid current.

p
where q ¼ ej2 is a phase-shift time-domain operator to obtain in- of the real and reactive power of the grid. The reference currents
quadrature version (90°-lagging) of an original waveform. of idg and iqg is
For the implementation of the PLL, the d-axis of the syn- " # "  #
chronous reference frame has been aligned with the positive

idg 1 Pg Q g v dg 
¼ ð39Þ
sequence vector of the grid voltage (v qgþ ¼ 0).

iqg v 2dg þ v qg Q g
2 
Pg v qg
During grid voltage sags, the control of the DC-link voltage by
Current control the GSC is deactivated, and the GSC injects reactive current to
the grid according to the requirements of the grid code, while
The GSC controls the DC-link voltage, vdc to be constant under the DC-link chopper resistor is operated to dissipate the active
normal conditions. Cascaded control structure with an inner cur- power surplus in it.
rent control loop and the outer DC-link voltage control loop is Like in the machine side converter control design, adaptive
applied. Thus, idg and iqg can be controlled through the regulation Fuzzy RST controller is also applied to the GSC control loop to
182 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187

Fig. 19. Simulation of the generator-side dynamic response during voltage dip. (a) Generator abc currents ([A]). (b) Generator dq-axis currents ([A]). (c) Generator torque
([N m]). (d) Generator speed ([rad/s]).

 
improve dynamic response. The control block diagram of the GSC is  k iv  
Pg ¼  kpv v dc  v dc þ v dc  v dc :v dc þ Pgen  PBC ð42Þ
shown in Fig. 15. s

DC link voltage and control where kpv and kiv are PI controller gains of the DC-link voltage
control.
Neglecting the power losses of the converters, the dynamic
equation of the DC-link voltage is expressed as [32] BC control

1 dv dc As described earlier, the DC-link voltage vdc is controlled by the

2
Pgen  PBC  Pg ¼ C ð40Þ
2 dt GSC. However, the rapidly rising DC-link voltage under system
fault is difficult to avoid by using the PI controller only. Therefore,
where C is the DC-link capacitance, Pgen is the generator power, Pg is a chopper circuit is used in the DC-link in parallel with the DC-link
the active power flowing into the grid, and PBC is the power dissi- capacitor, as shown in Fig. 3. To avoid DC-link over-voltages under
pated by the BC, which is expressed as system fault, an IGBT is used as the switching element of the chop-
per circuit. The chopper circuit makes it possible to suppress the
v 2dc
PBC ¼ Sc ð41Þ DC-link voltage fluctuations by expending the energy from the
rcc
WFSG [7].
where Sc is the duty ratio for the switch of the BC, which takes the In order to reduce the operation time of the crowbar, an
value 1 when the switch is closed and 0 when it is open. improved hysteresis control strategy is adopted, as it is shown in
From the control block diagram shown in Fig. 15, the reference Fig. 16. The maximum absolute value of DC-link voltage v dc max is
active power injected to the electrical supply network is given as compared with a threshold value v dc th and a safety value v dc sa .
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 183

Fig. 20. Simulation of the grid-side dynamic response during three-phase balanced voltage sag (75%) without reactive power support. (a) DC-link voltage ([V]). (b) Grid
voltages ([V]). (c) Grid abc currents ([A]). (d) Grid dq-axis currents ([A]). (e) Active and reactive grid power ([W], [VAR]).

If v dc max is greater than v dc th , the crowbar is activated for protect- studies have been performed using Matlab/Simulink software,
ing the power converters. And when v dc max decreases to be less where the parameters of the WECS are given in Appendix (Table 3).
than v dc sa , the crowbar will be cut off and the GSC is restarted to The power converter and the control algorithm are also imple-
control the grid. mented and included in the model. The sampling time used for
the simulation is 10 ls and the switching frequency of the B2B
Simulation results and discussion converters is 10 kHz.
A more practical wind speed profile (Fig. 17(a)) is now used to
Operation under normal condition demonstrate further the dynamic performance of the WFSG with
the proposed control.
To verify the effectiveness of the proposed control strategies of During the simulation, the d-axis command current of the
the WFSG wind system under various excursions, simulation WFSG side converter control system, ids, is set to zero; whereas,
184 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187

Fig. 21. Simulation of the grid-side dynamic response during single-phase unbalanced voltage sag (55%). (a) DC-link voltage ([V]). (b) Grid voltages ([V]). (c) Grid abc currents
([A]). (d) Grid dq-axis currents ([A]). (e) Active and reactive grid power ([W], [VAR]).

for the GSC, the q-axis command current, iqg, is set to zero. Simula- terms of the actual wind speed, the optimal WFSG speed command
tion results are given in Fig. 17(a)–(i). It can be seen that the wind is obtained by (2).
speed increases, the rotor angular velocity increases proportionally By applying the proposed control scheme, the optimal speed
too, the power coefficient will drop to maintain the rated output command is accurately tracked to extract the maximum power
power. The WECS operate under MPPT control. The initial pitch from the wind energy at any moment. In Fig. 17(c) the generated
angle b keeps the value of 0°, the tip speed ratio maintains the opti- torque reference follows the optimum mechanical torque of the
mal value 8.1, and the power coefficient Cp is the maximum around turbine quite well. The decoupling effect of the between the direct
0.48. Fig. 17(b) shows the speed tracking results of the WFSG. In and quadratic stator current of the WFSG is illustrated in Fig. 17(d).
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 185

Fig. 22. Simulation of the grid-side dynamic response during two-phase unbalanced voltage sag (55%). (a) DC-link voltage ([V]). (b) Grid voltages ([V]). (c) Grid abc currents
([A]). (d) Grid dq-axis currents ([A]). (e) Active and reactive grid power ([W], [VAR]).

The stator current and voltage waveforms and these zoom of power is kept at zero, which indicates that the unity power factor
the WFSG are presented in Fig. 17(e and f). As shown in this Fig- control is achieved. Also, the active power is proportional to the
ures, the stator currents are proportional to the wind speed. This wind speed.
is due to the reason that when the wind speed increases (not larger Fig. 17(i) shows the variation and a closer observation of three
than 9.1 m/s), there is more power generated, thus yield more cur- phase current and voltage and these zoom of GRID. The frequency
rents in the stator windings of the WFSG. imposed by the grid is 50 Hz. Similar to the WFSG stator currents,
Fig. 17(g) shows the simulation result of DC-link voltage that the grid currents also increase with the wind speed, because there
remains a constant value. Thus, proves the effectiveness of the is more power generated and injected into the grid. The harmonic
established regulators. As can be seen in Fig. 17(h), the reactive spectrum of stator and grid current obtained by using the Fast
186 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187

Fourier Transform (FFT) technique is shown in Fig. 18(a and b). The
simulation results demonstrate that the controller works very well
and shows very good dynamic and steady-state performance. The
control algorithm can be used to extract maximum power from
the variable-speed WT under fluctuating wind and effectively
eliminate the harmonics.

Operation under system fault

Symmetrical fault
In the proposed controller, a validation for stable operation of
the WECS under a system fault is necessary because the DC-link
voltage changes according to the wind-speed variation. For 75%
sag of the all grid phase voltages for 0.5 s between 3.2 s and
3.7 s, the system performance is shown in Figs. 19 and 20. Fig. 19 Fig. 23. Simulated results under unbalanced grid voltages: (a) gird frequency
(a)–(d) shows the stator abc currents, dq currents, the electromag- variation. (b) Dynamic responses of GSC voltages ([V]).
netic torque of the WFSG, the electrical speed, respectively. From
these figures, it is seen that the MPPT control and the efficient
operation of WECS are achieved. Due to the three-phase balanced power from the generator can be quickly dissipated by the protec-
fault, the voltage falls down and the current rises up. This can be tion crowbar in form of thermal heat, which prevents the DC-link
seen in Fig. 20(b) and (c) respectively. In addition, as shown in voltage from rising excessively. Moreover, the overshoot response
Fig. 20(a)–(e), the large voltage and power fluctuations (resultant of active power output at the GSC is minimized accordingly since
for the occurrence fault) are reduced significantly although the the power balance between the rectifier and inverter can be main-
chopper circuit is not used in the DC-link circuit. From these sim- tained during the transient process.
ulation results, it is evident that stable operation of the WECS is The gird frequency variation, the dynamic response of the GSC
achieved by using the proposed controller. voltages is shown in Fig. 23. Due to unsymmetrical Fault of the grid
phase voltages, the frequency reduction occurs at 3.2 s as shown in
Fig. 23(a), and then the frequency reduction is compensated by the
Unsymmetrical fault
governor, so that the frequency reduction is completely compen-
To verify the effectiveness of the proposed control strategy
sated at 3.25 s after making an overshoot of 0.1% as shown in
under unsymmetrical fault conditions, another simulation of sev-
Fig. 23(a).
ere unbalanced sag has been performed.
By this means, the LVRT capability of WFSG-WT is enhanced so
A 55% voltage drop in one single phase for the one test and a
that the system can stay in connection with the grid although the
55% voltage drop in two phases for the second test have been gen-
severe fault occurs in the power grid. After the fault clearance, the
erated during 0.5 s (between 3.2 and 3.7 s), as depicted in Figs. 21
DC-link voltage recovers by injecting more active power to the grid
(b) and 22(b) respectively.
than the generated.
Figs. 19, 21 and 22 show the simulation results if no specific
control action is done on the generator-side controller during the
dip, i.e., SSC works in the same way as in steady state, and the
GSC is controlled to meet the LVRT requirement. Table 3
Grid abc currents, Figs. 21(c) and 22(c), are unbalanced because Parameters used in the simulation models.
positive sequence and negative sequence are injected. Grid dq cur-

Parameters Value
rents show unity power factor ðidg ¼ idg ; iqg ¼ 0½AÞ in steady-state,
Turbine
see Figs. 21(d) and 22(d). Recall that the grid-side d-axis is aligned Rated power of the turbine, Pt 10 (Kw)
with the positive sequence of the grid voltage. With the current Density area, q 1.225 (kg m2)
voltage dip, as shown in Figs. 1 and 2, the LVRT requirement Radius of the turbine, R 3 (m)
demands the system to inject purely reactive current to the grid Number of blades 3
Gear ratio, G 5.4
during the dip. This is achieved by properly changing the grid Viscous friction coefficient, f 0.017 (N m s1)
active and reactive power references (i.e., the dq grid current refer-
WFSG
ences) according to the LVRT requirement. During the dip, Rated power of the generator, Sn 7.5 (KVA)
idg ¼ 0½A and iqg ¼ 7:5½A are observed in Figs. 21(d) and 22(d), Stator résistance, rs 1.19 (O)
the average active power delivered to the grid is 0 [W] and the Rotor résistance, rf 3.01 (O)
average reactive power delivered to the grid is about 2500 [VAR], Phase to phase rated voltage, Urms 400 (V)
Direct synchronous reactance, Xd 1.4 (p.u)
see Figs. 21(e) and 22(e).
Transverse synchronous reactance, Xq 0.7 (p.u)
Instantaneous ðPg ; Q g Þ active and reactive powers delivered to Open circuit transient time constant, T 00do 522 (ms)
the grid are shown in Figs. 21(e) and 22(e). Only active power is Direct transient synchronous reactance, X 0d 0.099 (p.u)
delivered to the grid in steady state. During the dip, the average Direct sub transient synchronous reactance, X 00d 0.049 (p.u)
Pg is zero and only average reactive power is injected to the grid. Direct transient time constant, T 0d 40 (p.u)
Therefore, the LVRT requirement is met. Direct sub transient time constant, T 00d 3.7 (ms)
Armature time constant, Ta 6 (ms)
The generator-side controller keeps its regulation unaltered;
therefore, there are no changes in the generator currents [Fig. 19 Grid
DC-Link voltage, vdc 600 (V)
(a) and (b)], electromagnetic torque [Fig. 19(c)], and speed
DC capacitance, C 1500 (lF)
[Fig. 19(d)]. Effective voltage, Vrms 220 (V)
With the adoption of a dc over-voltage protection in Figs. 21 Frequency, f 50 (Hz)
(a) and 22(a), the DC-link voltage can be limited below the Leakage résistance, Rf 1 (O)
600.5 V in accordance with the reference value. And the excess The leakage inductance, Lf 12 (mH)
 K. Tahir et al. / Electrical Power and Energy Systems 79 (2016) 172–187 187

Conclusion [11] Li W, Abbey C, Joos G. Control and performance of wind turbine generators
based on permanent magnet synchronous machines feeding a diode rectifier.
In: Proc IEEE PESC; 2006. p. 1–6.
A control approach has been presented to fulfill the LVRT [12] Lee SJ, Kang JK, Sul SK. A new phase detecting method for power conversion
requirements in a wind energy generation system based on a systems considering distorted conditions in power system. IEEE
1999:2167–72.
WFSG connected to the grid with a B2B full power converter. In
[13] Yazdani A, Iravani R. A unified dynamic model and control for the voltage-
the proposed method, the grid voltage phase is detected with pos- sourced converter under unbalanced grid conditions. IEEE Trans Power
itive sequence voltages with a HSF are used to enhance the overall Delivery 2006;21:1620–9.
[14] Song SH, Kim JJ. Implementation of dual current controller and real time
system performance under severe disturbances. To provide precise
power limiting algorithm in gird-connected inverter during unbalanced
control, a new controller (AF-RST) for both the SSC and GSC current voltage conditions. Trans KIEE 2006;55.
regulation are employed to further improve dynamic performance. [15] Fan L, Yin H, Miao Z. A novel control scheme for DFIG-based wind energy
In addition, a BC circuit is used in the DC link circuit, therefore, systems under unbalanced grid conditions. Electr Power Syst Res
2011;81:254–62.
stable operation of the WECS can be achieved during the system [16] Hu J, He Y, Xu L, Williams BW. Improved control of DFIG systems during
fault. Matlab/Simulink tests for a 7.5 kW WFSG-based WT demon- network unbalance using PI-R current regulators. IEEE Trans Ind Electron
strate the effectiveness of the proposed control strategy. On the 2009;56:439–51.
[17] Hu J, He Y. Reinforced control and operation of DFIG-based wind power-
other hand according to simulation results variable speed driven generation system under unbalanced grid voltage conditions. IEEE Trans
WFSG is not significantly impacted in fault conditions. Energy Convers 2009;24:905–15.
[18] Benbouzid M, Beltran B, Amirat Y, Yao G, Han J, Mangel H. Second-order sliding
mode control for DFIG-based wind turbines fault ride-through capability
Acknowledgement enhancement. ISA Trans 2014;53:827–33.
[19] E.ON Netz GmbH. Grid code-high and extra high voltage, Germany, April 2006.
Available at: <http://www.thewindpower.net>.
The authors would like to acknowledge the financial support of [20] Yang YH, Blaabjerg F. Low-voltage ride-through capability of a single-stage
the Algeria’s Ministry of Higher Education and Scientific Research. single-phase photovoltaic system connected to the low voltage grid. Int J
Photoenergy 2013.
[21] Conroy JF, Watson R. Low-voltage ride-through of a full converter wind
Appendix A turbine with permanent magnet generator. IET Renew Power Gener
2007:182–9.
[22] Abdallah B, Slim T, Gérard C, Emile M. Analysis of synchronous machine
See Table 3. modeling for simulation and industrial applications. Simul Model Pract Theory
2010;18:1382–96.
[23] Abdelli R, Rekioua D, Rekioua T, Tounzi T. Improved direct torque control of an
References induction generator used in a wind conversion system connected to the grid.
ISA Trans 2013;52:525–38.
[1] Nguyen TH, Lee DC. Ride-through technique for PMSG wind turbine using [24] Yan J, Lin H, Feng Y, Zhu ZQ. Control of a grid-connected direct-drive wind
energy storage systems. J Power Electron 2010;10:733–8. energy conversion system. Renew Energy 2014;66:371–80.
[2] Mohseni M, Islam SM. Review of international grid codes for wind power [25] Abdeddaim S, Betka A. Optimal tracking and robust power control of the DFIG
integration: diversity, technology and a case for global standard. Renew wind turbine. Int J Electr Power Energy Syst 2013;49:234–42.
Sustain Energy Rev 2012;16:3876–90. [26] Dounis AI, Kofinas P, Alafodimos C, Tseles D. Adaptive fuzzy gain scheduling
[3] Tsili M, Papathanassiou S. A review of grid code technical requirements for PID controller for maximum power point tracking of photovoltaic system.
wind farms. IET Renew Power Gener 2009;3:308–32. Renew Energy 2013;60:202–14.
[4] Yingcheng X, Nengling T. Review of contribution to frequency control through [27] Poitiers F, Bouaouiche T, Machmoum M. Advanced control of a doubly-fed
variable speed wind turbine. Renew Energy 2011;36:1671–7. induction generator for wind energy conversion. Electr Power Syst Res
[5] Topal ME, Ergene LT. Designing a wind turbine with permanent magnet 2009;79:1085–96.
synchronous machine. Istanbul Univ-J Electr Electron Eng (IU-JEEE) 2011;11 [28] Benghanem M, Bouzid AM, Bouhamida M, Draou A. Voltage control of an
(1):1311–7. isolated self-excited induction generator using static synchronous
[6] Alepuz S, Calle A, Monge SB, Kouro S, Wu B. Use of stored energy in PMSG rotor compensator”. J Renew Sustain Energy 2013;5:043118.
inertia for low-voltage ride-through in back-to-back NPC converter-based [29] Liang L, Jian X, Jianzhao H. Fuzzy adaptive PID control of large erecting system.
wind power systems. IEEE Trans Ind Electron 2013;60:1787–96. J Theor Appl Inform Technol 2013;47:412–8.
[7] Uehara A, Pratap A, Goya T, Senjyu T, Yona A, Arasaki N, et al. A coordinated [30] Hailiang Xu, Jiabing Hu, Yikang He. Integrated modeling and enhanced control
control method to smooth wind power fluctuations of a PMSG-based WECS. of DFIG under unbalanced and distorted grid voltage conditions. IEEE Trans
IEEE Trans Energy Convers 2011;26:550–8. Energy Convers 2012;27(3).
[8] Abbey C, Joos G. Supercapacitor energy storage for wind energy applications. [31] Rodriguez P, Luna A, Ciobotaru M, Teodorescu R, Blaabjerg F. Advanced grid
IEEE Trans Ind Appl 2007;43:769–76. synchronization system for power converters under unbalanced and distorted
[9] Nguyen TH, Lee DC. Improved LVRT capacity and power smoothening of DFIG operating conditions. IEEE Ind Electron IECON 2006:5173–8.
wind turbine systems. J Power Electron 2011;11:568–75. [32] Nguyen TH, Lee DC. Advanced fault ride-through technique for PMSG wind
[10] Conroy JF, Watson R. Low-voltage ride-through of a full converter wind turbine turbine systems using line-side converter as STATCOM. IEEE Trans Ind Electron
with permanent magnet generator. IET Renew Power Gener 2007;1 2013;60:2842–50.
(September 3):182–9.