Sdarticle 9
Sdarticle 9
Renewable Energy
journal homepage: www.elsevier.com/locate/renene
Integrated power characteristic study of DFIG and its frequency converter in wind
power generation
Shuhui Li*, Timothy A. Haskew, Jeff Jackson
Department of Electrical and Computer Engineering, The University of Alabama, 317 Houser Hall, Tuscaloosa, AL 35487, USA
a r t i c l e i n f o a b s t r a c t
Article history: A doubly fed induction generator (DFIG) is a variable speed induction machine. It is a standard, wound
Received 26 April 2008 rotor induction machine with its stator windings directly connected to the grid and its rotor windings
Accepted 22 July 2009 connected to the grid through a back-to-back AC/DC/AC PWM converter. The power generation of a DFIG
Available online 13 August 2009
includes power delivered from two paths, one from the stator to the grid and the other from the rotor,
through the frequency converter, to the grid. The power production characteristics, therefore, depend not
Keywords:
only on the induction machine but also on the two PWM converters as well as how they are controlled.
Wind power
This paper investigates power generation characteristics of a DFIG system through computer simulation.
Doubly fed inductor generator
PWM converter The specific features of the study are (1) a steady-state model of a DFIG system in d–q reference frame, (2)
d–q vector control a simulation mechanism that reflects decoupled d–q control strategies, (3) power characteristic simu-
Modeling lation for both generator and converter, and (4) an integrative study combining stator, rotor and
Simulation converter together. An extensive analysis is conducted to examine integrated power generation char-
acteristics of DFIG and its frequency converter under different wind and d–q control conditions so as to
benefit the development of advanced DFIG control technology.
Ó 2009 Elsevier Ltd. All rights reserved.
0960-1481/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2009.07.016
S. Li et al. / Renewable Energy 35 (2010) 42–51 43
! ! d! !
Is_dq Rs Lls
Ir_dq Llr Rr Rr(1-s)/s v dq ¼ R$ i dq þ L i dq þ jus L$ i dq þ !
v dq1 (15)
1 2 1 2
dt
+ + ! ! ! !
1
Vr_dq
V dq ¼ R$ I dq þ jus L$ I dq þ V dq1 (16)
Lm Ems_dq The d–q steady-state equivalent circuit of the grid-side
Vs_dq
converter system, Fig. 6, is obtained from Eq. (16). From the grid
2
Vr_dq(1-s)/s !
- - point of view, the converter injects a voltage, V dq1 , into the grid.
Using the PWM converter average model, the d and q components
Fig. 4. DFIG dq steady-state equivalent circuit. of the injected voltage, Vd1 and Vq1, are linearly proportional to the
d and q control voltages, Vd1* and Vq1*, as shown in Fig. 3 [16]. To
maintain a constant DC-link voltage, the control function should
values imply that the real or reactive power is generated by the make the real power outputted from one converter (rotor-/grid-
generator or the converters and delivered to the grid. side converter) equal the power entered in another converter
(grid-/rotor-side converter) if assuming no loss in the converters.
If the d-axis of the reference frame is aligned along the grid-
! !* ! !
Ps þ jQs ¼ V s dq Is dq (8) voltage position, V dq ¼ Vd þ j0. Assuming V dq1 ¼ Vd1 þ jVq1 ,
then, the current flowing between the grid and the converter can be
obtained by
PRCL ¼ Ir2 dq $Rr (9)
! !
! V dq V dq1 Vd Vd1 jVq1
h!
!* i I dq ¼ ¼ (17)
PAG ¼ Re E ms (10) R þ jXL R þ jXL
dq Ir dq
in which XL stand for the grid filter reactance.
! !* Using the passive sign convention, the power absorbed from the
Pr þ jQr ¼ V r dq I r dq (11)
grid can be achieved from the fundamental complex power equa-
! ! !
tion, i.e., Pgrid þ jQgrid ¼ V dq I *dq ¼ Vd I *dq . By solving this
Pconv ¼ PAG þ Protor PRCL (12) complex power equation together with Eq. (17), Eqs. (18) and (19)
are obtained.
R jX Idq
A 1 2 B
Vdq
Vdq1
Fig. 5. Grid-side converter schematic. Fig. 6. Grid-side converter d–q equivalent circuit.
S. Li et al. / Renewable Energy 35 (2010) 42–51 45
a 60 a 1.2
50
Stator Reactive Power (p.u.)
1
Vrq = 0.28p.u. 40
R e al P o wer (p.u.)
Vrq =0 p.u. 0.8
30 Vrq = 0.07p.u. Vrq = 0p.u.
20 0.6
10 0.4
0
0.2
-0.2 -0.1 0 0.1 0.2
-10
0
-20
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
-30 -0.2
Vrq = -0.16p.u.
-40 -0.4 Vrq = -0.04p.u.
Slip S l ip
30 b 4
Stator Reactive Power (p.u.)
R e al P o wer (p.u.)
10
2.5 Vrd = 0.4p.u.
0 2
-0.2 -0.1 0 0.1 0.2 Vrd = 0p.u.
-10 1.5
Vrd = 0p.u.
1
-20
Vrd = 0.28 p.u. 0.5
-30
Slip 0
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
-0.5
Fig. 8. Stator reactive power characteristics (a) Vrq ¼ 0.16 to 0.28 p.u., DVrq ¼ 0.04,
Slip
Vrd ¼ 0 p.u.; (b) Vrd ¼ 0.08 to 0.28 p.u., DVrd ¼ 0.04, Vrq ¼ 0 p.u.
Vrq =0 to 0.4p.u., ΔVrd =0.05p.u., Vrq =0p.u
that the amplitude of the converter injected voltage to the grid must Fig. 9. Rotor real power over speed characteristics.
a 0.8 a
0.6
2
Vrq_sv = 0p.u.
b
b 5
4 3
3 Vrd_s v = 0.4p.u.
2.5
Vrd_s v = 0p.u. 2
2
1
1.5
0
-0 .8 -0 .6 -0 .4 -0 .2 00 .2 0.40 .6 0.8 1
-1
0.5
-2
0
-3
1
-0.5 0.5
-4 1
0.5 0
0 -0.5
-5 -0.5 -1
-1
Slip Vd1
Vq1
Vrd =0 to 0.4p.u., ΔVrd =0.05p.u., Vrq =0 p.u.
1 1.5
2 2
Pconv ¼ RV d1 RV q1 þ RVd V d1 X L Vd Vq1 (22)
R2 þ XL2
1 2
1
Qgrid ¼ 2 XL V X L Vd V d1 þ RVd V q1 (23)
(pu)
d
R þ XL2
0.5
Under the primary and secondary control goals defined above,
the solution of Vd1 and Vq1, that satisfies the requirements of (1)
balancing active power at rotor- and grid-side converter, (2) mini- 0
mizing the resultant reactive power sending to the grid, and (3) 1
being within the converter rated current and linear modulation 0.5 1
0.5
constraints, becomes a nonlinear programming problem illustrated 0
0
below. -0.5 -0.5
Minimize: -1 -1
Vq1 Vd1
1 2
jQgrid Qs j ¼ XL V d XL Vd V d1 þ RVd V q1 Qs Fig. 11. (a) Grid-side converter real power characteristics; (b) characteristics of reactive
R2 þ XL2
power absorbed from the grid; and (c) amplitude of the injected voltage to the grid by
grid-side converter.
Subject to:
1
2 RV 2 þ RV V
RVd1 modulation limit. The grid filter resistance and inductance are the
R2 þXL2 q1 d d1 XL Vd Vq1 ¼ Pr
qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
ffi same as those used in Fig. 11. The DC-link voltages are 3, 4, and 5 p.u.
Id2 þIq2 Vd1 þVq1
3 Irated ; Vconv ¼ 3 pVffiffidcffi (grid phase voltage as the base), respectively, meaning that the
2 2Vtri
maximum allowable per-phase injected voltage to the grid by the
The integrated DFIG power characteristic study considers the grid-side converter are 1.06, 1.41, and 1.77 p.u.
nonlinear programming configuration. Fig. 13 presents the resultant From the integrated power simulation study, the following
DFIG real and reactive power characteristics compared to the stator regularities are obtained. (1) When the real power absorbed or
real and reactive power characteristics for a typical Vrd and Vrq generated by the DFIG rotor is high and the DC-link voltage is low,
control condition applied to the rotor. The voltage Vd1 and Vq1 of the the control of the grid-side converter is unable to balance the real
grid-side converter are obtained according to the nonlinear power needed by the rotor (Fig. 13a). (2) When the rotor absorbs
programming formulation. Data are only presented for those DFIG real power from the grid, the stator power includes the captured
operating slips that do not cause the converter going over the linear wind power plus the power passed to the stator from the rotor.
48 S. Li et al. / Renewable Energy 35 (2010) 42–51
a 0.5
0
Specify an injected DFIG rotor voltage (V q & Vd) -0.6 -0.4 -0.2 -0.5 0 0.2 0.4 0.6
Fig. 13. Integrated DFIG active (a) and reactive (b) power characteristics under the
Therefore, the resultant generating real power is smaller than the
nonlinear programming formulation (rotor-side converter control: Vrd ¼ 0.4 p.u. and
real power outputted from the stator path (Fig. 13a). (3) Within the Vrq ¼ 0 p.u.).
converter linear modulation limit, if the stator absorbs reactive
power, it would be harder for the grid-side converter to compen-
sate the stator reactive power; if the stator generates reactive rotor blades [m2], Cp is the turbine performance coefficient, vw is
power, however, it would be easier for the grid-side converter to wind speed [m/s], Rblade is the radius of the rotor blades [m], and um
compensate the stator reactive power if the real power transferred is the angular speed of the blades. The performance coefficient, Cp,
on the converter path is moderate and the stator generating is a function of the tip-speed-ratio l, Eq. (25), and the pitch angle of
reactive power is not too high (Fig. 13b). (4) The higher the DC-link the rotor blades b. It is determined by aerodynamic laws and thus
voltage is, the wider is the slip range in which the stator and the may change from one wind turbine type to another. The relation-
grid-side converter reactive power can be cancelled by each other ship between ur, the generator rotor electrical speed, and um
under the nonlinear programming formulation condition follows from
(Fig. 13b). ur ¼ p$ngear $um (26)
where p is the generator pole pairs and ngear is the gear ratio from
8. DFIG system study in an integrative environment the high- to low-speed shaft of a wind turbine.
Fig. 14 shows the Cp curves for a 1.5 MW DFIG wind turbine [6].
As it can be seen from above sections, a DFIG for wind power The mathematical representation of the Cp curves is obtained
generation is affected by the following factors simultaneously: (i)
input power extracted by a DFIG from the wind, (ii) the d–q voltage
applied to the rotor at the generator level, and (iii) the d–q voltage 0.5
1
injected to the grid by the grid-side converter. Thus, for a compre- 3
hensive and accurate study, it is important to (1) examine DFIG 0.4
parametric data concurrently through an integrative approach, and 5
(2) consider various practical constraints in the integrative study,
0.3 7
such as the rated stator and rotor power, acceptable slip and
Cp
9
converter linear modulation limit.
0.2 11
The mechanical power extracted by a wind turbine from the 13
wind is expressed by the well-known cube law equation [20]:
0.1 15
1
Pw ¼ rair Ablade Cp ðb; lÞv3w (24)
2 0.0
0 2 4 6 8 10 12 14 16 18 20
l ¼ Rblade um =vw (25) tip-s p eed-ratio
where rair is the air density [kg/m3], Ablade is the area covered by the Fig. 14. A 1.5 MW DFIG wind turbine Cp curves.
S. Li et al. / Renewable Energy 35 (2010) 42–51 49
Computation of Generator
Protor Re Vr_dq Ir_dq Protor 157.216 kW Qrotor Im Vr_dq Ir_dq Qrotor 65.661 kW
Computation of Converter
Vdq Vdq1
Idq Idq 228.76 A Idq 228.76 0.004i A
Zf
Real and reactive power absorbed from the grid by the converter
Pgrid Re Vdq Idq Pgrid 157.844 kW Qgrid Im Vdq I dq Qgrid 0.003 kVar
Ptotal Pstator Pgrid Ptotal 1744.457 kW Qtotal Qstator Qgrid Qtotal 0 kVar
Efficiency
Ptotal
Pin P0 Cp 98.657 %
Pin
Fig. 15. Using Mathcad for integrated DFIG study (neglect rotational losses).
50 S. Li et al. / Renewable Energy 35 (2010) 42–51
through curve fitting as shown by Eq. (27) where aij coefficients are Table 2
given in [6]. The curve fit is a very good approximation for values of Maximum wind power extraction study for integrated DFIG generator and
converter.
2 < l < 13. Values of l outside this range represent very high and
low wind speeds, respectively, that are normally outside the Wind speed
continuous rating of the machine [6]. For each pitch angle, there is 7 m/s 8 m/s 9 m/s
an optimal tip-speed-ratio lopt under which Cp takes a maximum
Qs ¼ 0 Qs < 0 Qs ¼ 0 Qs < 0 Qs ¼ 0 Qs < 0
value, i.e., maximum power extraction from the wind for that pitch
Pmax (pu) 0.304 0.304 0.453 0.453 0.646 0.646
angle. sopt 0.412 0.412 0.328 0.328 0.244 0.244
4 X
X 4 Vr_dq (pu) 0.428 0.441 0.343 0.355 0.258 0.274
i j
Cp ðb; lÞ ¼ aij b l (27) Ps (pu) 0.516 0.515 0.673 0.673 0.852 0.850
i¼0 j¼0 Qs (pu) 0 0.202 0 0.232 0 0.402
Pr (pu) 0.216 0.215 0.224 0.225 0.213 0.215
Qr (pu) 0.142 0.236 0.124 0.210 0.104 0.218
The general procedures of the integrative approach are the ma 1.0 0.68 1.0 0.651 1.0 0.85
following. (1) For a wind speed and a blade pitch angle under Pgrid (pu) 0.217 0.219 0.224 0.229 0.213 0.223
a combined Vrd and Vrq control applied to the rotor, find generator Qgrid (pu) 0.025 0.202 0.03 0.232 0.023 0.402
slip by solving a nonlinear equation, i.e., power balance equation Ptotal (pu) 0.299 0.297 0.447 0.444 0.637 0.628
Qtotal (pu) 0.025 0 0.03 0 0.023 0
between the converted power, Eq. (12), and the turbine driving h 98.46% 97.64% 98.55% 97.84% 98.62% 97.24%
power Eq. (24). (2) Compute stator and rotor real and reactive
power. (3) Calculate Vd1 and Vq1 according to the nonlinear
programming formulation as shown in Section 7. (4) Compute
resultant real and reactive powers transferred to the grid by the consistent with the analysis shown in Section 7. In other words, the
integrated generator and converter system. For a demonstration integrated study suggests that a smart and optimal DFIG control
purpose, Fig. 15 shows a partial computation given by Mathcad for design is appropriate for enhanced and reliable DFIG system
a condition that makes the wind turbine operate at the maximum performance.
power extraction condition and zero stator reactive power. The
wind speed is 11 m/s, air density is 1.17 kg/m3, pitch angle is 1, the 9. Conclusions
rotor diameter of the turbine blade is 77 m, the gear ratio is 35, the
grid frequency is 60 Hz, the generator pole pairs is 4, and the DC- This paper presents a simulation study on the integrated power
link voltage is 1200 V. generation characteristics of a DFIG and its back-to-back PWM AC/
Table 2 presents the integrated DFIG system study for three DC/AC converter.
different wind speeds at 7 m/s, 8 m/s, and 9 m/s, respectively, The power characteristics of a DFIG depend on the d–q voltage
under the maximum wind power extraction condition. For all Vrq and Vrd applied to the rotor. Both d and q components of the
the three wind speed conditions, two different stator reactive rotor voltage can control DFIG speed and regulate the reactive
power regulation mechanisms are considered, i.e., zero and power. Below usyn, the rotor absorbs real power and Vrd is more
nonzero reactive power generated from the stator path. The effective for speed control in the stator-voltage-oriented frame
system parameters are the same as those used in Fig. 15. In the while Vrq is needed to compensate the generating reactive power
table, Pmax stands for the maximum available wind power, sopt caused by Vrd. Above usyn, the rotor generates real power and Vrq is
represents the generator slip for maximum power extraction, more effective for speed control in the stator-voltage-oriented
Vr_dq is the rotor d–q voltage amplitude, Ps, Qs, Pr, Qr, Pgrid and frame while Vrd is necessary to compensate the absorbing reactive
Qgrid are defined in Sections 3 and 4, Vdq1 is the amplitude of power resulted from Vrq.
the d–q voltage injected into the grid by the grid-side converter, The goal the grid-side converter control is to keep a constant
ma is the modulation ratio of the grid-side converter, Ptotal and DC-link voltage and to minimize the reactive power taken from the
Qtotal are real and reactive powers of the integrated DFIG grid. Due to the constraints of converter linear modulation and
generator and converter system, and h is the generator rated power, a nonlinear programming formulation results. The
efficiency. control performance of the grid-side converter depends on the real
The integrative study presents an important issue regarding the power transferred between the rotor- and grid-side converters and
behavior of the integrated DFIG generator and converter system as the DC-link voltage. The smaller the real power transferred
explained below. Traditionally, control of both the generator and the between the two converters and the higher the DC-link voltage, the
grid-side converter is designed to have zero reactive power absor- more effective it is for the grid-side converter to maintain
bed from the grid. However, Table 2 shows that, when the stator a constant DC-link voltage and minimize the resultant reactive
reactive power is zero, the grid-side converter may be unable to power absorbed from the grid.
completely compensate the reactive power transferred between the The integrated DFIG system study suggests that there is
converter and the grid due to the converter linear modulation a need for a smart and optimal DFIG control design that is
limitation. Hence, under the conventional zero reactive power different from conventional DFIG control strategies. In other
control rule, the grid-side controller would generate a control words, for optimal performance and stability of the integrated
voltage that is over the converter linear modulation limit, which DFIG generator and converter systems, the stator should
normally results in high unbalance and oscillation of the system generate a certain amount of reactive power and the grid-side
although the control signal is processed by a saturation function converter should operate by absorbing reactive power from the
[21]. Instead, for optimal performance and stability of the integrated grid.
DFIG generator and converter systems, the stator should generate
a certain amount of reactive power and the grid-side converter References
should operate by absorbing reactive power from the grid, which
makes the converter far below the linear modulation limit while [1] Zavadil R, Miller N, Ellis A, Muljadi E. Making connections: wind generation
challenges and progress. IEEE Power Energy Mag November 2005;3(No. 6).
maintaining a zero resultant reactive power with a small increase of [2] Muller S, Deicke M, De Doncker RW. Doubly fed induction generator systems
the voltage amplitude applied to the rotor (Table 2). This result is for wind turbines. IEEE Ind Appl Mag May/June 2002;8(No. 3):26–33.
S. Li et al. / Renewable Energy 35 (2010) 42–51 51
[3] Kling WL, Slootweg JG. Wind turbines as power plants. In: Proceedings of [12] Hopfensperger B, Atkinson D, Lakin RA. Stator flux oriented control of
the IEEE/Cigré workshop on wind power and the impacts on power systems, a cascaded doubly-fed induction machine. IEEE Proc Electr Power Appl
17–18 June 2002, Oslo, Norway. November 1999;146(No. 6):597–605.
[4] Bonnet F, Vidal PE, Pietrzak-David M. Dual direct torque control of doubly fed [13] Morel L, Godfroid H, Mirzaian A, Kauffmann J. Double-fed induction machine:
induction machine. IEEE Trans Ind Electron October 2007;54(No. 5):2482–90. converter optimization and field oriented control without position sensor.
[5] Slootweg JG, de Haan SWH, Polinder H, King WL. General model for repre- IEEE Proc Electr Power Appl July 1998;145(No. 4):360–8.
senting variable speed wind turbines in power system dynamics simulations. [14] Leonhard W. Control of electrical drives. 2nd ed. Berlin, Germany: Springer-
IEEE Trans Power Syst February 2003;18(No. 1). Verlag; 1996.
[6] Miller Nicholas W, Price William W, Sanchez-Gasca Juan J. Dynamic modeling [15] Mullane Alan, Lightbody Gordon, Yacamini R. Wind-turbine fault ride-through
of GE 1.5 and 3.6 wind turbine-generators. GE Power Syst October 27, 2003. enhancement. IEEE Trans Power Syst November 2005;20(No. 4).
[7] Smith S, Todd R, Barnes M, Tavner PJ. Improved energy conversion for doubly- [16] Mohan N. Advanced electric drives – analysis, modeling and control using
fed wind generators. IEEE Trans Ind Appl November 2006;42(No. 6):1421–8. Simulink. MN: Minnesota Power Electronics Research & Education, ISBN
[8] Cadirci I, Ermis M. Double-output induction generator operating at sub- 0-9715292-0-5; 2001.
synchronous and supersynchronous speeds: steady-state performance opti- [17] Li S, Haskew TA. Simulation and analysis of DFIG characteristics under general
misation and wind-energy recovery. IEEE Proc Electr Power Appl September d–q control strategies in stator-voltage-oriented frame. Int J Power Energy Syst
1992;139(No. 5):429–42. 2009;29(No. 2).
[9] Vicatos MS, Tegopoulos JA. Steady state analysis of a doubly-fed induction [18] Mohan N, Undeland TM, Robbins WP. Power electronics: converters, appli-
generator under synchronous operation. IEEE Trans Energy Convers cations, and design. 3rd ed. John Wiley & Sons Inc.; October 2002.
September 1989;4(No. 3):495–501. [19] Xu L, Wang Y. Dynamic modeling and control of DFIG based wind turbines
[10] Pena R, Clare JC, Asher GM. Doubly fed induction generator using back-to-back under unbalanced network conditions. IEEE Trans Power Syst
PWM converters and its application to variable speed wind-energy genera- 2007;22(1):314–23.
tion. IEEE Proc Electr Power Appl May 1996;143(No 3). [20] Freris LL. Wind energy conversion system. Upper Saddle River, NJ: Prentice
[11] Hansen AD, Iov F, Sørensen P, Blaabjerg F. Overall control strategy of variable Hall; 1990.
speed doubly-fed induction generator wind turbine. In: Grid integration and [21] Li S, Haskew TA, Jackson J. Comparative study of PWM converter controls for
electrical systems of wind turbines and wind farms (CD-ROM). Nordic wind grid integration of renewable energy. In: Proceedings of IEEE international
power conference 2004 (NWPC 04), Göteborg (SE), 1–2 Mar 2004. Göteborg: conference on sustainable energy technologies (ICSET 2008), Singapore,
Chalmers University of Technology; 2004. p. 7. November 24–27, 2008.