Elektrotehniški vestnik 85(4): 149-154, 2018 ORIGINAL SCIENTIFIC PAPER
A New control strategy for active-and reactive-power dispatching in a wind farm based on DFIG at no speed and power over-rating
Mohamed Boutoubat1 Abdallah Zegaoui2 Lakhdar Mokrani1 Mohamed Machmoum3
1	LACoSERE Laboratory, Laghouat University, Ghardaia Street Bp 37 G, Laghouat, (03000), Algeria.
2	LGEER Laboratory, University Hassiba Benbouali of Chief, BP151, 02000 Chief, Algeria. 3IREENA, 37Boulevard de l'Université, Bp 406, 44602 Saint-Nazaire, Nantes, France. E-mail: boutoubat90@yahoo.fr
Abstract. The paper proposes a new strategy for wind farm to operate in three different active-and reactive-power dispatching modes without any over rating of the generator speed and power. Using the proposed modes, i.e. balance mode, maximum power point tracking (MPPT) mode and delta mode, enables the grid-operator to meet the active-and reactive-power demand. To avoid over-rating, the generators speed and power are controlled. In each operating mode, the power is controlled by electric components (DFIG rotor currents) and the speed by a pitch angle-device. In each operating mode, a new expression is proposed to calculate the stator reference active power and the generator power control. The obtained simulation results confirm the usefulness of the proposed three-mode power-dispatching control strategy.
Keywords: Wind farm control; Active and reactive power control; Balance control; Delta control; Pitch-angle control.
Krmiljenje skupine vetrnih elektrarn z uravnavanjem delovne in jalove moči in preprečevanjem preobremenitve
V prispevku smo predstavili metodo za krmiljenje skupine vetrnih elektrarn, ki delujejo v treh načinih z omejitvijo proizvedene moči in hitrosti vrtenja. S predlaganim načinom krmiljenja, ki vključuje ravnotežni način, sledenje točki največje moči in način delta, omogočamo operaterju omrežja da zadosti zahtevam po delovni in jalovi moči. Preobremenitev preprečujemo z nadzorom proizvedene moči in hitrostjo vrtenja. Pri vsakem načinu delovanja proizvedeno moč krmilimo z električnimi gradniki. Za vsak način delovanja predlagamo nov izračun moči generatorja. Rezultati simulacije potrjujejo uporabnost predlaganega načina krmiljenja.
1 Introduction
Today, there is an increasing demand for connecting large wind farms to the power network. Due to the active role of the wind farms in the grid, it is clear that large wind farms should be able to control their produced active and reactive power and to provide ancillary services, such as the grid frequency and voltage control without any over-rating. There are many papers which have studied the modelling and control of wind farms to supervise effectively the produced electricity in terms of the active-and reactive-power generation. In [1]-[2]-[3], a wind farm is controlled according to the utility manager requirements. A methodology to develop electromagnetic transient
simulation models of wind farms to predict their behaviour under normal operating conditions has been presented in [4]. A particle swarm method is used to optimize the reactive power dispatch in a wind farm [5]. In [6], the authors study possibilities of a coordinated control and management of different wind farm concepts to guarantee reaching the operational set point. A supervisory control of a wind farm with a short-term wind-speed prediction is presented in [7]. In [8], the authors present a scheme for a supervisory control of a wind farm by using either an external energy-storage system or a power reserve. A storage device based on a statcom and battery storage system is presented in [9]. The main goal of the control strategy presented in [10] is to mitigate the flicker emission by a wind farm. In [11], the authors assess the contribution to the frequency control by using a variable-speed wind turbine.
Following our previous work in which we investigated the generator active and reactive power over-rating, this paper proposes a new strategy to control the active-and reactive-power of a wind farm by operating in one of the three different operating modes (i.e. balance mode, maximum power point tracking mode (MPPT) and delta mode), at no generator over-rating speed and power. Three new expressions are given to calculate the stator reference active power for the three proposed modes.
Received 15 May 2018 Accepted 9 June 2018
150
BOUTOUBAT, ZEGAOUI, MOKRANI, MACHMOUM
2 Wind-farm configuration
In this study, a small wind farm operating with six wind turbines is considered. Each turbine is of 1.5 MW of the nominal power. It is equipped with DFIG and variable pitch-angle device and is connected to its own terminal voltage of about 20 kV. The wind farm is connected to a 66 kV infinite bus bar as shown in Fig. 1.
TVi
Ug-i -h-
Pinj-1 Qinj-1
TV2
Ug-2 —h
Pinj-2 Qinj-2
TVs
Pinj-3_
Qinj-S
TV4
Ug-S
H—
Ug-4
I Pinj-4 Qinj-4
TV5
Ug
g-5
H-
\Pinj-5 Qinj-5
TVs
Ug-6
H—
I Pinj-6 Qinj-6
PCC
Ti
2500kVA 20/0.69 kV 6%
T2 2 —GD-I
2500kVA 20/0.69 kV 6%
Tf
-GO-
8000kVA 66/20 kV
Z12
Ts
<EH
s
2500kVA 20/0.69 kV 6%
_r
T4
Z2s
Zs4
2500kVA 20/0.69 kV 6%
T5 5
—GO-
2500kVA 20/0.69 kV 6%
Z45
T6
J
Z56
2500kVA 20/0.69 kV 6%
Ptotal
Qtotal
Figurel. Layout of the studied wind farm. 2.1 Wind- system modelling
A classical Wind-energy Conversion System (WECS) with DFIG is illustrated in Fig. 2.
Gearbox
Ps-t lsabc-i ^ Qs-i			Pinj-i Qinj-i
Vdc-i	Igabc-i		-►
		Pg-i Qg-i	
_L ~r			
DFIG	rsc	. GSC
Figure2. Scheme of the ith WECS of the studied wind farm.
2.1.1
Turbine model
The mechanical power captured by the wind turbine is given by the following classical expression [12]:
Pt = - Pcp
2 (1)
where, p is the air density, 5 is the area of the wind wheel (m2), v is the wind speed (m/s), cp (X, ft) is the turbine power coefficient, X is the tip speed-ratio and ft is the pitch angle.
The tip-speed ratio is given by the following equation:
_ Rat
v (2) where mt is the turbine speed (rd/s) and R is the blade radius (m).
Fig. 3 shows variations in the power coefficient versus (X) for a constant value of pitch angle ft.
Cp
X.
opt
X
Figure 3. Power coefficient versus the tip-speed ratio.
2.1.2 DFIG Modeling
The DFIG voltage and flux equations, expressed in the Park reference frame, are given by [13]:
_ dw,
u, = Rj^ + —— - aswqs
dt
J,. , d^qs
u = R i H--—
qs s qs dt
Udr = Rrldr +
dWd. dt dw„
u = Ri
qr r qr dt
+
■- (as	~)Wq

(3)
and:
1
4
c
pmax
A NEW CONTROL STRATEGY FOR ACTIVE-AND REACTIVE-POWER DISPATCHING IN A WIND FARM BASED ON DFIG.
151
Wds = Lsids + Midr Wqs = LJqs + Miqr Wdr = Lridr + Mlds
w = Li + Mi
qr	r qr	qs
The electromagnetic torque is given by:
T em	3 P J ('d'rWWqs lqr^^ds )
(4)
(5)
The stator active-and reactive-power expressions are:
Ps = 3 (. s 2v
s	.ds1 ds +Uqslqs
)
Q. = 3 (.
(6)
(7)
2
The WECS classical mechanical equation is: dO
- + f O + T, dt J g L
(9.a) (9 b) (10.a) (10.b)
(11)
Where us is the stator-voltage magnitude assumed to be constant and g is the DFIG slip range.
Eqs. 9.b, 10.a and 10.b are used to rewrite the stator active and reactive power (Eqs. 6 and 7):
3u M P =--s-1„_
! 0sWs
Ws = Lslds + Mld
3 M
T em =-- P— Wslqr
2 L,
Qs = 3-r~ (.s - Mmslir )
2 Ls°s
(12)
3 Rotor-side converter control
From Eq. 12 one can see that the stator active (Ps) and reactive (Qs) power can be controlled by acting on the rotor components (iq and id). This is done by using a Fuzzy Logic Controller (FLC) seen in Fig. 4. GSC is controlled to maintain a constant DC voltage between the two converters (RSC and GSC) with a unity power factor at PCC. A more detailed control scheme of this converter is in [13].
4 DESCRIPTION of the proposed three MODE POWER-DISPATCHING STRATEGY
In this section, three different control modes of the proposed power-dispatching strategy of a wind farm are discussed. For each control mode, stator reference active power (Psref) is calculated as follows:
4.1 MPPT operating mode
In this mode, each of the six turbines is controlled so as to produce its maximum active power. For this purpose, the stator reference active power (Psref) is determined. To determine this reference value of the MPPT operating mode, the power balance sheet is used on the basis of the WECS rotor dynamic equation. By using Eqs. 11 and 12, the following equality is easily demonstrated:
(8) Tem Og = Ps (1 - g)
(13)
where Tl is the aerodynamic torque of the turbine.
To establish a decoupled control of the active and reactive power, stator flux (^s) is set aligned with the d axis and assumed to be constant (providing the grid is powerful and stable) [14]-[19]. Then, one can write Wds=ys and tyqS=0. Consequently, equations (3), (4) and (5) for a steady-state regime are:
Uj. ~ 0
By replacing the electromagnetic torque using its expression (Eq .8) in Eq .13, the stator reference active power (Psref) in the MPPT mode is calculated as follows:
jQgef (kT )-Qgref ((k -1T ),
Ogref (kT ) +f Ogef '(kT ) + Pm
P f =-
sref
1-g
re
(14)
where T is the sampling time; and PaerM is the maximum mechanical power captured by the wind turbine in the MPPT mode:
P = — nc
aerM ^ ^ P max
(15)
where gref is the DFIG slip in the MPPT mode. It is calculated by the following equation:
03s - PO
'gref
S ref
m* (16)
The generator speed reference in the MPPT mode is determined by the following equation:
Q
gref
= 8-
R
(17)
4.2 Balance operating mode (14)
In this mode, the wind farm is controlled to meet the system operator active and reactive power (Pwref, Qwref) demand at (PCC) (see Fig. 1). Each turbine (i) injects a constant value of the active and reactive power (Pinj-i, Qinj-i) at its PCC (see Figs. 1 and 2) in such a way that the values (Pwref, Qwref) are satisfied. The required values (Pinj-i, Qinj-i) of each turbine are determined by minimizing the following objective function:
Min {(Pwref - Poa)2 + (Qwref - Qtota)
}
(18)
T = J
em
3
s v
v
152
BOUTOUBAT, ZEGAOUI, MOKRANI, MACHMOUM
where Ptotai and Qtotai are the total active and reactive power, respectively, produced by the wind farm at the PCC (see Fig.1).
The goal is to find Pin— and Qinj-i (i=1, 2,..., 6) injected by the turbines in such a way that Eq.18 is minimized.
From Fig.2, for each turbine (i) and at its PCC, one can write the following equations:
P = —P (1 -g )
inj—i	s—i\ «3 i y
q =-Q - Q
Z^tnj — t	ZL--S — i	'g — i
(19)
(20)
Psref—i

pnj—i
1 — g
Qsref—i = —Qi
)
-inj—i
(21) (22)
Table 1. Power dispatching results using the balance control mode.
Index of the wind turbine (i)	Pj (MW)	Qj (MVAr)
1	0.53465	0.20124
2	0.50562	0.19533
3	0.50023	0.19702
4	0.49547	0.20907
5	0.50989	0.20644
6	0.49686	0.21545
Total	3	1.5
AP=X Api
(23)
this case, the stator reference active-power (Psf is determined by using the same expression as in the MPPT mode (Eq.14) by subtracting the power reserve (APi) as follows:
J
Psref
(24)
QSef (kT) — asef ((k —1)T) T
0„f (kT) + f 0.2(kT )+PeM —Ap
1 — gef
5 Pitch angl e control
where giis generator slip range (i); Ps-i, Qs-i are the stator active and reactive power produced by the generator (i); Qg-i is the reactive power produced from the GSC side (Qg-i=0 since GSC is controlled to ensure a unity power at the grid side). Finally and after determination of (Pinj-i and Qinj-i) (see Table1) by minimizing Eq.18, the proposed stator reference active-and reactive-power are expressed easily using Eqs.19 and 20 as follows:
The pitch angle control limits the produced power or speed. In our case, the pitch-angle control maintains the generators speeds to their nominal values, in the case of the balance and delta control. The pith angle (fi) system is shown in Fig .5.
dp
_firej	-
Qgn ,
Q,
R)
4.3 Delta control mode
In this mode, the wind farm produces less than its maximum active-power. In fact, the wind farm is controlled to produce its maximum active power with a power reserve (AP). Consequently, each wind turbine (i) is controlled to produce its maximum active power with a power reserve (APi) so that:
The wind turbine (i) will be controlled to produce its maximum active power with a power reserve (APi). In
Figure 5. Pitch angle control.
6 Simulation results and discussion
Three different modes are simulated to show how the proposed wind farm controller makes the wind farm power production runs according to the system-operator requirements. In the simulation, the six wind turbines are driven at different wind powers at the same mean speed value (8 m/s).
The wind turbines (TV1 and TV2), (TV3 and TV4) and (TV5 and TV6) are driven by three different wind profiles V1-2, V3-4 and V5-6 respectively, with a same mean value of 8m/s as shown in Fig.6. To allow for simplicity and since the turbines responses are equal, only the simulation results obtained for the first turbine (TV1) are presented and discussed in detail. Figs.7, 8 and 9 show its active and reactive power (Pmj-i and Qinj-i), its generator speed (Qgi), its power coefficient (Cpi) and its pitch angle (fii), respectively.
In the first period (0 s<t<100 s), the turbine runs at its maximum power. In the second period (100 s<t<200 s), the turbine injects a fixed active power (Pinji=- 0.5346 MW) and a fixed reactive power (Qinji=- 0.2012 MVAr) (see Table 1 and Fig. 7). In the third and last period (200 s<t<300 s), the turbine runs with a power reserve (APi =0.1MW) (see Fig.7). Since in the balance and reserve mode the turbine is controlled to provide less active power than its maximum, the turbine generator turbine (1) tends to accelerate. So, to avoid any over-rating, when the speed reaches its nominal value (204.2 rd/s), the pitch-angle system intervenes to limit the generator speed (Qgi) by acting on the pitch angle (fii) and then the speed (Qgi) is maintained at its nominal value (204.2 rd/s) (see Figs .8 and 9).
i =1
A NEW CONTROL STRATEGY FOR ACTIVE-AND REACTIVE-POWER DISPATCHING IN A WIND FARM BASED ON DFIG.
153
Turbine
Figure 4. Control scheme of the RSC.
V5-6	Vl-2	Si			
					
-3».					Y v
					t
				v	-4
					
					
					
					
					
Time (s)
Figure 6. Wind-speed profiles applied to the farm turbines (m/s).
AP,


Time (s)
Figure 7. TV1 turbine Powers: Injected power (Pinji), maximum power (Pmppti), power reserve (APi) (W) and injected reactive power (Qinji) (VAr).
					
			Qgi		
	A M			L	
		1	■z-v—		
A	V	\	\ r	r\	
/ S	/	V	V	/\J	v \
			Q	MPPTl	
					
					
Time (s)
Figure 8. TV1 turbine Speeds: Actual generator speed (Qgi) and the reference speed at the MPPT mode (0.gMPm) (rd/s).
			C C	P6	
a)			A		
		/	A		
	Cpi	s	Cp3		
					
					
					
o)					
					
	ß		ß3		
		\		ß6	
					
		M	Yv J<	■T ill nr	
				T	
					







Temps (s)

Figure 9. Turbines TV1,TV3, TV4 Characteristics: a)
Power coefficients, b) Pitch angles (°).
154
BOUTOUBAT, ZEGAOUI, MOKRANI, MACHMOUM
Finally, from Fig.10 showing the DC link voltage for the TV1 turbine, one can see that the DC voltage (vdci) is kept practically around its reference value of 2000 V at each of the studied operating modes by controlling GSC.
19600	50	100	150	200	250
FigurelO. DC link voltage (vdci) in the case of the TV1 turbine.
7 Conclusion
This paper investigates the wind farm capability to control the produced active and reactive power. A strategy of different operating modes (i.e. balance operating mode, MPPT operating mode and delta operating mode) is studied and applied to a wind farm in order to meet the grid-operator active-and reactive-power demand. Three mathematical expressions are proposed to calculate the stator reference active power for each mode. A combination of the power and control speed has permits the wind farm to operate in three different modes at no system speed and power over-rating.
References
[1]	J.L. Rodriguez-Amenedo, S. Arnaltes, J.C. Burgos, "Automatic generation control of a wind farm with variable speed wind turbines," IEEE Transactions on Energy Conversion 17 (2), pp. 279-284, 2002.
[2]	T. Ghennam, B. Francois, E.M. Berkouk, "Local supervisory algorithm for reactive power dispatching of wind farm," 13th European Conference on Power Electronics and Applications, Barchalona, Spain, 2009.
[3]	A. D. Hansen, P. Sorensen, F. Iov, F. Blaabjerd, "Centralised power control of wind farm with doubly fed induction generators,", Renewable Energy, 31(7), pp: 935-951, 2006.
[4]	I. zubia, X. Ostolaza, A. Susperregui, G. Tapia, "Complete wind farm electromagnetic transient modelling for grid integration studies, ", Energy Conversion and Management, 50(3), pp: 600-610, 2009.
[5]	M. Martenez-Rajos, A. Sumper, O. Gomis-Bellmunt, A. Sudria-Andru, "Dispatch in wind farms using particle swarm optimization technique and feasible solutions search,", Applied Energy, 88(2), pp: 4678-4686, 2011.
[6]	J.L. Rodriguez-Amenedo, S. Arnales, M.A. Rodriguez, "Operation and coordinated control of fixed and variable speed wind turbines," Renewable Energy (33, pp: 406-414, 2008.
[7]	F. Islam, A. El-Derra, S.M. Muyeen, "Smoothing of wind farm output by prediction and supervisory control-based FESS", IEEE Transactions on Sustainable Energy 4(4), pp: 925-933, 2013.
[8]	Z. Lubosny, J. W. Bialek, "Supervisory control of wind farms," IEEE Transactions on power Systems 22(3), pp: 985-994, 2007.
[9]	N.Y. Abed, S. Teleke, J.J. Castanedo, "Planning and operation of dynamic energy storage for improved integration of wind energy," IEEE Conference on Power and Energy General Meeting, San Diego, 2011.
[10]	F. Gibrau-LIistuella, A. Sumper, F. Diaz-Gonzalez, S. Galceran-Arellano, "Flicker mitigation by reactive power control in wind farm with doubly fed induction generators," Electrical Power and Energy Systems 55, pp: 285-296, 2014.
[11]	X. Ying, T. Nengling, "Review of contribution to frequency control through variable speed wind turbine," Renewable Energy 26, pp: 1671-1677, 2011.
[12]	M. Soleimanzadeh, R. Winiewski, S. Karnev, "An optimisation frame work for load and power distribution in wind farms,", Wind Engineering and Industrial Aerodynamics 107, pp: 356363, 2012.
[13]	M. Boutoubat, L. Mokrani, M. Machmoum, "Control of a wind energy conversion system equipped by DFIG for active power generation and power quality improvement," Renewable Energy 53, pp: 433-445, 2013.
Mohamed Boutoubat Has been a lecturer at the Laghouat University, Algeria, since 2015. His research interests include renewable-energy system control and power quality and management.
Abdallah Zegaoui has been a lecturer at the Chlef University, Algeria, since 2012. His research interests include renewable-energy system control and management.
Lakhdar Mokrani has been a professor at the Laghouat University, since 2013. His main research area includes Modeling and CAD of electrical machines.
Mohamed Machmoum has been a Professor at the Nantes University, France, since 2005. His main area of interest includes power electronics and power quality, wind or tidal-energy conversion systems and power line communication.