Li 2011
Li 2011
Li 2011
Recommended Citation:
Li, Shuhui; Haskew, Timothy A.; Challoo, Rajab; and Nemmers, Marty (2011) "Wind Power
Extraction from DFIG Wind Turbines Using Stator-Voltage and Stator-Flux Oriented Frames,"
International Journal of Emerging Electric Power Systems: Vol. 12: Iss. 3, Article 7.
DOI: 10.2202/1553-779X.2676
Abstract
This paper investigates and compares wind power extraction from a DFIG (doubly-fed
induction generator) wind turbine using stator-voltage and stator-flux oriented frames. The paper
analyzes how wind power extraction control of a DFIG wind turbine is converted to speed control,
and then how speed and reactive power control of the wind turbine is converted to generator
current control using the two different orientation frames. The paper also investigates what are the
differences in developing wind power extraction control strategies of a DFIG wind turbine using
the two different orientation frames. Simulation study is conducted for a 1.5MW DFIG wind
turbine, in which the turbine driving torque is modeled by considering typical wind turbine
aerodynamic characteristics. The study shows that the performance of DFIG wind power
extraction is similar using both stator-voltage and stator-flux oriented frames. But, it is found that
a conventional wind power extraction approach using the stator-flux oriented frame could
deteriorate the power quality of the DFIG system while it is more stable to estimate the position of
the stator-flux space vector by simply adding -90 degree to the stator-voltage space vector.
Author Notes: This work was supported in part by the U.S. National Science Foundation under
Grant 1059265.
1. Introduction
A doubly-fed induction generator (DFIG) is an adjustable-speed induction
machine widely utilized in the modern wind power industry [1, 2]. The reasons to
use variable speed wind turbines are fourfold: a higher energy yield, a reduction
of mechanical loads and a simpler pitch control, wide range controllability of both
active and reactive powers, and less fluctuation in output power [2, 3].
However, energy extraction from a DFIG wind turbine depends not only
on the induction generator but also on the control strategies developed using
different orientation frames. The normal orientation frames used in a DFIG
system are stator-flux-oriented (SFO) and stator-voltage-oriented (SVO) frames.
In most traditional schemes, wind power extraction from a DFIG system is
achieved through a nested speed- and current-loop control structure using the SFO
frame [4, 5]. In [5] to [7], the SFO frame is used to develop DFIG wind power
extraction mechanisms. Other approaches, such as direct-power-control strategies
for DFIG wind turbines using the SFO frame [8, 9], have also been proposed
recently. Although the SVO frame is normally not used in a DFIG design, [10]
and [11] report special approaches to improve DFIG stability under unbalanced
conditions using the SVO frame. In [12], a cascaded DFIG configuration and a
relevant control mechanism are presented, in which one DFIG (power machine) is
controlled by the power converter indirectly through the other DFIG (control
machine). Although the system configuration is a little bit different, most
techniques as reported in [12] are similar to those used in traditional DFIG wind
turbines.
As various approaches are being developed for DFIG wind power
extraction using either the SFO or SVO frame, it is important to investigate the
difference and similarity as well as the performance and power quality of DFIG
system design using the two different orientation frames. A variety of techniques
have been developed to study the wind power extraction from a DFIG wind
turbine. These techniques can be divided into two categories: 1) transient
approaches [3-12], and 2) steady-state methods [13-17]. Transient approaches are
essential to study DFIG dynamic performance in a short time period. Steady-state
methods are vital to examine DFIG characteristics in a broader spectrum.
The comparison of a doubly-fed induction generator using SVO and SFO
reference frames was initially investigated in [18], where the comparison focuses
only on the DFIG current-loop controller without maximum power extraction
consideration, making the comparison unrealistic to a practical variable-speed
wind turbine. Unlike [18], this paper compares the two reference frame control
from maximum power extraction standpoint, and investigates, through combined
steady-state and transient studies, how the wind power extraction from a DFIG
wind turbine is converted to the variable-speed control of the wind turbine and
then how the speed and reactive power control of the wind turbine is converted to
generator current control using SVO and SFO reference frames.
In the sections that follow, the paper first describes the operating
principles and the wind power extraction mechanism of a DFIG wind turbine.
Then, DFIG transient and steady-state models are developed. General control
structures for DFIG wind power extraction using SVO and SFO reference frames
are presented. Both analytical and simulation studies are conducted to compare
what are differences and similarities of the wind power extraction control
mechanisms developed using the two different orientation frames. Finally, the
paper is summarized.
DOI: 10.2202/1553-779X.2676 2
1
Pw = ρair AbladeC p ( β , λ )vw3 (1)
2
λ = Rbladeωm vw (2)
where ρ air is the air density [kg/m3], Ablade is the area covered by the rotor blades
[m2], Cp is the turbine performance coefficient, vw is wind speed [m/s], Rblade is the
radius of the rotor blades [m], ωm is the angular speed of the blades, and λ is a
ratio of the rotor tip speed over wind speed. The performance coefficient, Cp, is a
function of the tip-speed-ratio λ and the pitch angle of the rotor blades β. It is
determined by aerodynamic laws and thus may change from one wind turbine
type to another. Here β is assumed to be a one-dimensional variable that is
equally applied to all the rotor blades via the wind turbine pitch control
mechanism.
Figure 2 shows the Cp curves for a 1.5MW DFIG wind turbine [20]. The
mathematical representation of the Cp curves is obtained through curve fitting as
shown by (3) where aij coefficients are given in [20]. The curve fit is a very good
approximation for values of 2 < λ < 13. Values of λ outside this range represent
very high and low wind speeds, respectively, that are normally outside the
continuous rating of the machine [20].
4 4
C p ( β , λ ) = ∑∑ aij β i λ j (3)
i = 0 j =0
For each pitch angle, there is an optimal tip-speed-ratio λopt under which
Cp takes a maximum value. Therefore, the rotational speed of the rotor blades ωm
is normally regulated to a value of ωm_opt for maximum wind power extraction
through the variable-speed operation of the wind turbine generator so that
λ = Rbladeωm _ opt Vw = λopt while the pitch angle is fixed at a low value (Fig. 2). As
the captured wind power exceeds the rated power at a high wind speed, a power
limitation control is activated to keep the generated power at the rated value by
adjusting the pitch angle. Under typical Weibull wind distribution [21], a DFIG
wind turbine operates in the variable speed mode most of the time for maximum
wind power extraction.
1
Pw _ max = ρ air AbladeC p max vw3 , ωm _ opt = λopt vw Rblade (4)
2
Fig. 3. Maximum power extracted from the wind versus turbine rotational speed
(ρ air=1.17kg/m3)
DOI: 10.2202/1553-779X.2676 4
where Rs, Rr, Lls, and Llr are resistances and leakage inductances of DFIG stator
and rotor windings; Lm is the mutual inductance; vsd, vsq, vrd, vrq, isd, isq, ird, irq, λ sd,
λ sq, λ rd, and λ rq are d and q components of stator and rotor voltages, currents, and
fluxes in a d-q reference frame; and ωs and ωr are angular frequencies of stator
and rotor currents. (5) and (6) can also be combined together into complex
equations using space vectors
d
vs _ dq = Rs is _ dq + λs _ dq + jωs λs _ dq (9)
dt
d
vr _ dq = Rr ir _ dq + λr _ dq + jωr λr _ dq (10)
dt
Vr _ dq
I r _ dq + jωs Llr I r _ dq + jωs Lm ( I s _ dq + I r _ dq )
Rr
= (12)
s s
The steady-state equivalent circuit, obtained from (11) and (12), is shown
by Fig. 5. The relationship between stator-voltage and stator-flux space vectors in
steady state can be estimated from (9). Under the steady-state condition, (9)
becomes (13). Thus, if neglecting stator winding resistance, the stator-voltage
space vector leads the stator-flux space vector by 90°.
DOI: 10.2202/1553-779X.2676 6
Vs _ dq = Rs I s _ dq + jωs λs _ dq (13)
+ 1
+
Vr_dq
Ir_dq Lm
Vs_dq Ems_dq
2
Vr_dq(1-s)/s
- -
β-axis
qsv-axis
ωs
vr_dq dsv -axis
qsf -axis
α-axis
dsf -axis
Fig. 6. Relationship between SVO and SFO frames (neglecting stator winding resistance)
relevant position of the stator-flux space vector over the stator-voltage space
vector can be calculated from
in which the stator and rotor d-q currents can be solved from Fig. 5. It is found
that the influence of the stator winding resistance to the angle between stator-flux
and stator-voltage space vectors is very small especially for a DFIG operating
within the rated power. Table 1 gives parameters of a 1.5MW DFIG wind turbine.
Table 2 presents typical angles of the stator-flux space vector over the stator-
voltage space vector for DFIG parameters given in Table 1, in which the stator-
voltage space vector is taken as the reference. As it can be seen, the stator-voltage
space vector still leads the stator-flux space vector by about 90°.
Table 2. Angle between stator-flux (θ sλ) and stator-voltage (θ sv) space vectors
Generator slip, Rotor control voltage θ sv - θ sλ
s=0.05, Vrd_sv=20V, Vrq_sv=2V -89.89°
s=0.22, Vrd_sv=100V, Vrq_sv=5V -90.15°
s=0.362, Vrd_sv=150V, Vrq_sv=7V -90.01°
s=0.638, Vrd_sv=250V, Vrq_sv=7V -89.94°
DOI: 10.2202/1553-779X.2676 8
dωg
τ em = J eq + Baωg + τ w (15)
dt
where Jeq is the total equivalent inertia referred to the generator and Ba is the
active damping coefficient representing turbine rotational losses. The rotor
electrical speed ωr, generator speed ωg, turbine rotational speed ωm, and generator
slip s are related by (16), where p is the generator pole pairs, ngear is the gear ratio
from the high- to low-speed shaft of a wind turbine, and ωsyn is the synchronous
speed. The electromagnetic torque τem can be computed from (17). In the steady-
state condition, a wind turbine must operate at a speed point that the driving
torque Tw balances with the generator electromagnetic torque Tem if neglecting the
rotational losses, i.e., Tw = Tem.
ωg ωm ⋅ ngear
ωr = p ⋅ ω g , ω g = ngear ⋅ ωm , s = 1 − = 1− (16)
ωsyn ωsyn
According to (15), this paper converts a speed control demand of the wind
turbine into a torque control demand. Then, the torque and generator reactive
power demands are converted into d-q current control. This process is a little bit
different from Fig. 4. The conversion from torque and reactive power control to d-
q current control depends on what orientation frame is employed.
Vs _ dq
I ms _ dq = , λs _ dq = Lm I ms _ dq (18)
jωs Lm
and the d component of the magnetizing current is zero. Then, the stator d-q
current space vector I s _ dq _ sv , according to Fig. 5, is
I s _ dq _ sv = I ms _ dq _ sv − I r _ dq _ sv = − I rd _ sv − j ( I ms + I rq _ sv ) (19)
In terms of the magnetizing and rotor currents, the electromagnetic torque and the
stator reactive power according to (17) and Fig. 5 are
Hence, given torque and stator reactive power references Tem* and Qs*, the
conversion from torque and stator reactive power to rotor d and q currents is
I s _ dq _ sf = I ms _ dq _ sf − I r _ dq _ sf = ( I ms − I rd _ sf ) − jI rq _ sf (23)
In terms of rotor and magnetizing currents, torque and stator reactive power
according to (17) and Fig. 5 are
DOI: 10.2202/1553-779X.2676 10
Thus, given torque and stator reactive power references Tem* and Qs*, the
conversion from torque and stator reactive power to rotor d and q currents is
di
vrd = Rr ird + Llr rd − ωr Llr irq − ωr Lm ims
dt
(26)
dirq
vrq = Rr irq + Llr + ωr Llr ird
dt
11
acquired from (26) by replacing the bracket by d or q voltage from the current-
loop controller as shown by (28), in which the magnetizing current compensation
term is added to the rotor d-axis voltage equation in the SVO frame.
vrd* = vrd' − ωr Llr irq − ωr Lm ims , vrq* = vrq' + ωr Llr ird (28)
In the SFO frame, ims _ dq _ sf = ims + j 0 , where ims = λsd _ sf Lm is almost constant if
the stator voltage keeps unchanged. Thus, (6) can be expressed by (29).
Traditionally, the position of the stator-flux space vector θ sf is estimated through
(30), which requires measurements of both stator-voltage and stator-current space
vectors in α-β reference frame. Since the integration is involved, computational
error to estimate θ sf using (30) is normally higher than that to estimate θ sv using
(27).
di
vrd = Rr ird + Llr rd − ωr Llr irq
dt
(29)
dirq
vrq = Rr irq + Llr + ωr Llr ird + ωr Lm ims
dt
DOI: 10.2202/1553-779X.2676 12
vrd* = vrd' − ωr Llr irq , vrq* = vrq' + ωr Llr ird + ωr Lmims (31)
Fig. 8 DFIG wind power extraction control using SVO and SFO frames
It is necessary to point out that other approaches are also developed recently
for DFIG wind turbine control, such as predictive and sliding mode control
mechanisms [25-27]. For example, instead of using a standard PI current-loop
13
controller, the predictive current control (PCC) uses predictive current and
voltage equations or approaches to generate control voltages vrd* or vrq* [25].
Although a PCC controller is different from a standard PI controller, the
fundamental differences and similarities of a PCC controller design using SVO
and SFO frames are analogous to Fig. 8 according to Fig. 6.
DOI: 10.2202/1553-779X.2676 14
after the system is stable, are consistent with steady-state results that can be
calculated from the steady-state model (Fig. 5) [28], demonstrating the
effectiveness of the DFIG wind power extraction control using the SVO frame.
180
Actual speed
160
Speed (rad/s)
Reference speed
140
120
(a)
100
80
60
4 6 8 10 12 14 16
Time (s)
1000
Stator reactive power
Power (kW/KVar)
500
1500
1000
Stator Current (A)
500
0
(c)
-500
-1000
-1500
15 15.02 15.04 15.06 15.08 15.1 15.12 15.14 15.16 15.18 15.2
Time (s)
1500
1000
Rotor Current (A)
500
0
(d)
-500
-1000
-1500
15 15.02 15.04 15.06 15.08 15.1 15.12 15.14 15.16 15.18 15.2
Time (s)
Fig. 9. Wind power extraction using the SVO frame: (a) generator speed and reference
speed, (b) maximum available wind power, net turbine output power and stator reactive
power, (c) stator current, (d) rotor current referred to the stator
15
180
Actual speed
160
Reference speed
Speed (rad/s)
140
120
(a)
100
80
60
4 6 8 10 12 14 16
Time (s)
1000
500
1500
Stator Current (A)
1000
500
0
(c)
-500
-1000
-1500
15 15.02 15.04 15.06 15.08 15.1 15.12 15.14 15.16 15.18 15.2
Time (s)
1500
Rotor Current (A)
1000
500
0
(d) -500
-1000
-1500
15 15.02 15.04 15.06 15.08 15.1 15.12 15.14 15.16 15.18 15.2
Time (s)
Fig. 10. Wind power extraction using the SFO frame (Approach 1): (a) generator rotor
speed and reference speed, (b) maximum available power, net turbine output power and
stator reactive power, (c) stator current, (d) rotor current referred to the stator
DOI: 10.2202/1553-779X.2676 16
θ sv [ kT ] = tan −1 ( vα s [ kT ] vβ s [ kT ]) (32)
λα s [ kT ] = λα s ( k − 1) T + T ( vα s [ kT ] − Rsiα s [ kT ])
λβ s [ kT ] = λβ s ( k − 1) T + T ( vβ s [ kT ] − Rsiβ s [ kT ] ) (33)
θ sf [ kT ] = tan −1 ( λα s [ kT ] λβ s [ kT ])
17
is stable, the angle calculated using Method 2 is stable at nearly -90° while the
angle calculated using Method 1 oscillates around -90° (Fig. 11).
-87
-88
Angle (degree)
-89
-90
-91
-92 Method 1
Method 2
-93
3 3.05 3.1 3.15 3.2 3.25 3.3 3.35 3.4 3.45 3.5
Time (s)
Fig. 11. Comparison of angles between stator-flux and stator-voltage space vectors
computed using two different methods
DOI: 10.2202/1553-779X.2676 18
180
Actual speed
160
Reference speed
Speed (rad/s)
140
120
(a)
100
80
60
4 6 8 10 12 14 16
Time (s)
1000
Stator reactive power
Power (kW/KVar)
500
Fig. 12. Wind power extraction control using the SFO frame (Approach 2): (a) generator
rotor speed and reference speed, (b) maximum available power, net turbine output power
and stator reactive power
N
vw (t ) = Vw + ∑ Ai cos(ωi t + φi ) (34)
i =1
0.475σ 2 ( L Vw )
Svv (ωi ) = 56
(35)
1 + (ωi L Vw )2
19
VI-C, respectively. Before t=4ms, wind speed is 8m/s. At t=4ms, wind speed is
generated by using (34) to (36) with Vw=8m/s, which results in variable and short-
term gust wind as shown by Fig. 13(a). Under the variable and gust wind
condition, the maximum available power that can be extracted by the wind turbine
fluctuates sharply as the wind speed varies [Figs. 13(b) & 13(c)]. However, the
net output power of the wind turbine follows the maximum available wind power
properly, demonstrating the effectiveness of the wind power extraction in variable
and gust wind conditions. Again, as demonstrated by Figs. 13(b) and 13(c), under
the variable and gust wind condition, the wind turbine generator has a better
performance and higher power quality when the stator-flux space vector position
is estimated by using Approach 2, which is consistent with the analysis shown in
Sections VI-B and VI-C.
10
Wind Speed (m/s)
8
a)
7
4 6 8 10 12 14
Time (s)
500
Stator reactive power
Power (kW/kVar)
0
Net output
-500
power
b)
-1000
Maximum power
-1500
4 6 8 10 12 14 16
Time (s)
500
Stator reactive power
Power (kW/kVar)
0
Net output
power
-500
c)
-1000
Maximum power
-1500
4 6 8 10 12 14
Time (s)
Fig. 13. Wind power extraction under variable wind: a) wind speed, b) maximum
available power, net turbine output power and stator reactive power using Approach 1 in
the SFO frame, and c) maximum available power, net turbine output power and stator
reactive power using Approach 2 in the SFO frame
DOI: 10.2202/1553-779X.2676 20
7. Conclusions
This paper investigates wind power extraction from a DFIG wind turbine using
SVO and SFO frames. The paper first presents steady-state and transient models
of a DFIG wind turbine. Then, the steady-state model is used to obtain the general
relationship of wind power extraction from the DFIG wind turbine using SVO and
SFO frames. The transient model is used to develop DFIG wind power extraction
control mechanisms using the two different orientation frames. Analytical study
demonstrates that the overall control structures are different using SVO and SFO
frames. Another difference is the approaches that are used to estimate stator-
voltage and stator-flux space vector positions. The performance of the wind
power extraction using SVO and SFO frames is equivalent if the position of the
stator-voltage or stator-flux space vector can be estimated accurately. Regarding
the two SFO approaches used in the paper, Approach 1, a popular conventional
approach, is theoretically more accurate but computationally less stable, which
would deteriorate the power quality of a DFIG wind turbine. On the other hand,
Approach 2 estimates stator-flux space vector position by simply adding -90° to
the stator-voltage space vector position, which is computationally more stable and
therefore is beneficial to reduce the control oscillation of a DFIG system.
References
[1] R. Zavadil, N. Miller, A. Ellis, and E. Muljadi, “Making Connections: Wind
Generation Challenges and Progress,” IEEE Power & Energy Magazine,
Vol.3, No. 6, Nov. 2005.
[2] S. Muller, M. Deicke, R.W. De Doncker, “Doubly Fed Induction Generator
Systems for Wind Turbines,” IEEE Industry Applications Magazine, Vol. 8,
No. 3, 26-33, May/June 2002.
[3] W.L. Kling and J.G. Slootweg, “Wind turbines as Power Plants,” in
Proceedings of the IEEE/Cigré workshop on Wind Power and the Impacts
on Power Systems, 17 – 18 June 2002, Oslo, Norway.
[4] L. Morel, H. Godfroid, A. Mirzaian, J.M. Kauffmann, “Double-fed
induction machine: converter optimization and field oriented control without
position sensor,” IEE Proc.-Electr. Power Appl., Vol. 145, No 4, May 1998.
[5] R. Pena, J.C. Clare, and G. M. Asher, “Doubly fed induction generator using
back-to-back PWM converters and its application to variable speed wind-
energy generation,” IEE Proc.-Electr. Power Appl., Vol. 143, No 3, May
1996.
21
DOI: 10.2202/1553-779X.2676 22
[18] S. Li, R. Challoo and M.J. Nemmers, “Comparative Study of DFIG Power
Control Using Stator-Voltage and Stator-Flux Oriented Frames,”
Proceedings of 2009 IEEE PES General Meeting, Calgary Alberta, Canada,
July 26-30, 2009.
[19] L. L. Freris, Wind energy conversion system (Upper Saddle River, NJ:
Prentice Hall, 1990).
[20] N.W. Miller, W.W. Price, and J.J. Sanchez-Gasca, “Dynamic Modeling of
GE 1.5 and 3.6 Wind Turbine-Generators,” GE Power Systems, October 27,
2003.
[21] Z. Yu and A. Tuzuner, “Wind Speed Modeling and Energy Production
Simulation with Weibull Sampling,” Proceedings of 2008 IEEE PES
General Meeting, Pittsburgh, PA, July 20-24, 2008.
[22] A.D. Hansen, P. Sørensen, F. Iov and F. Blaabjerg, “Control of variable
speed wind turbines with doubly-fed induction generators,” Wind
Engineering, Vol. 28, No. 4, June 2004 , pp. 411-432.
[23] N. Mohan, T.M. Undeland, and W.P. Robbins, Power Electronics:
Converters, Applications, and Design, 3rd Ed., John Wiley & Sons Inc.,
October 2002.
[24] N. Mohan, Advanced Electric Drives – Analysis, Modeling and Control
using Simulink, MN: Minnesota Power Electronics Research & Education,
ISBN 0-9715292-0-5, 2001.
[25] M. Nemec, K. Drobniˇc, D. Nedeljkovi´c, and V. Ambrožiˇc, “Direct
Current Control of a Synchronous Machine in Field Coordinates,” IEEE
Transactions on Industrial Electronics, Vol. 56, No. 10, October 2009, pp.
4052-4061.
[26] N. Patin, W. Naouar, E. Monmasson, and J.P. Louis, “Predictive control of a
doubly-fed induction generator connected to an isolated grid,” Proceedings
of 32th Annual Conference of IEEE Industrial Electronics (IECON 2006),
Paris, Nov.6-10, 2006, pp. 873-878.
[27] B. Beltran, T. Ahmed-Ali, M. El Hachemi Benbouzid, “Sliding Mode Power
Control of Variable-Speed Wind Energy Conversion Systems,” IEEE Trans
on Energy Conversion, Vol. 23, No. 2, June 2008, pp. 551-558.
[28] S. Li and T.A. Haskew, “Simulation Analysis of DFIG Characteristics under
d-q Control Strategies in Stator-Voltage-Oriented Frame,” International
Journal of Power and Energy Systems, Vol. 29, No. 2, 2009, pp. 103-115.
23
DOI: 10.2202/1553-779X.2676 24
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: