Pitch Regulation
Pitch Regulation
N
simulations are presented. OWDAYS the totally installed wind power capacity is
constantly increasing. Wind has become a competitive player
Index Terms—Doubly fed induction generator (DFIG), in the power market. Large wind parks have been installed
variable speed wind systems, low voltage ride-through (LVRT), around Europe and around the world, while the average power of
crowbar. today’s wind parks has reached the magnitude of conventional power
plants. Therefore, the effect of the wind turbines, their control and
I. NOMENCLATURE the protection system is of vital importance. This fact shows the need
A Rotor area. for suitable models and detailed analysis of the dynamic stability of
β Wind turbine blade pitch angle. wind parks.
Cp Wind turbine power coefficient.
As the penetration of wind energy within the power systems
H Inertia constant.
increases, the requirements set by the Transmission System
i Current.
Operators (TSOs), are being constantly reviewed and expanded.
KP, KI Coefficients for the proportional-integral controller.
Power quality, reactive power control, and most recently fault ride-
λ Wind turbine tip-speed ratio.
through are considered as most significant requirements. Especially
L Inductance.
the fault ride-through capability is considered crucial for the system
ω s, ω m , Synchronous, wind turbine shaft,
stability. The interaction between the wind turbines and the grid
ωr, ωb generator rotor angle and base angular speed.
during voltage dips is being carefully investigated. Detailed and well
Ψ Flux linkage.
documented models of the wind turbines equipped with the so called
P, Q Active and reactive power.
Low Voltage Ride-Through capability should therefore be developed
ρ Air density.
to ensure accurate conclusions regarding the behavior of the wind
r Resistance.
turbine during grid faults.
P Number of poles.
R Wind turbine blade radius. Although many wind parks are equipped with fixed speed stall
s Rotor slip. regulated wind turbines, the variable speed concept has become
Tem Electromagnetic torque. today’s standard. Both synchronous and induction generators are
Tm Mechanical torque act on the generator rotor. being used with each concept having pos and cons depending on the
Tw Wind turbine prime torque from wind. issues being investigated. One of the most often used variable speed
U Voltage. configurations is the pitch controlled wind turbine equipped with
Vw W ind speed. doubly fed induction generator. In DFIGs the stator is directly
X Reactance. connected to the grid while the rotor is connected via a back-to-back
Symbols PWM power converter. The DFIG concept has been described in
1/s Integral operator. many publications ([1]-[5]). However, the wind turbine equipped
MPT Maximum power tracking logic. with DFIG appears to have very special features that need thorough
PF Power factor. investigation. The sensitivity of the power converters towards high
GSC Grid side converter. currents following voltage dips require detailed modeling of the Low
RSC Rotor side converter. Voltage Ride-Through protection system.
2
18
Ψqr = X rr ⋅ iqr + X m ⋅ iqs (15)
P
1.4
Power factor
or
1.2 Terminal
Us
Voltage
Control mode
1
i*
dr
) Udr
u 0.8 +
p( PI
-
m
u
mi 0.6
pt
o
P idr
0.4
Fig. 6. Block diagrams of the rotor side converter – (a) Speed
control block (b) Reactive power control block.
0.2
When designing the grid side converter, the main objectives are
0
8 9 10 11 12 13 14 15 16 17 18 the control of the reactive power exchanged between the GSC and
Rotor speed (rpm) the grid and mainly the control of the dc voltage connecting the two
Fig. 5. Optimum active power production (pu) versus rotor speed PWM converters. The dc voltage is kept to nominal value and, in
(r.p.m.). most cases, the reactive power is kept to zero.
In this case, the reference frame attached to the stator voltage
The solid line is computed from equation (5) keeping Cp=Cp,max for
vector was chosen. The equations giving the active and reactive
all wind speeds below the nominal. The dashed line was, however,
power exchanged between the GSC and the grid are:
used in the model to avoid computing problems in the min and max
rotor speed. Pgs = U qs ⋅ iq gs + U ds id gs ⇒ Pgs = U qs ⋅ iq gs (23)
Respectively, the reference value for the d-coordinate of the rotor
current is computed from equation (22), where the reference value of Q gs = U qs ⋅ id gs − U ds i q gs ⇒ Q gs = U qs ⋅ id gs (24)
the reactive power Qs* depends on the preferred control strategy.
Usually the power factor is controlled to unity. As mentioned before, the common case for the GSC is to operate in
unity power factor:
U s ω b X ss Qs Q gs = 0 ⇒ i d gs = 0
i dr = − (22) (25)
X M ωe X M U s
The dc voltage is kept to equal its nominal value (1 pu)
This variable speed configuration gives the opportunity to control the through the q-coordinate of the GSC, iq gs. The differential
reactive power exchanged with the grid and, therefore, to contribute describing the dynamics of the capacitor in the dc-link of the
in the terminal voltage control. As the penetration of wind energy converter system was added to the model and is given in
increases, the trend in the wind turbine design is to ensure voltage equation (26):
control, following the principles of conventional power production.
Of course, this is possible provided that the converter rating allows
handling of the reactive current even in nominal active current dU dc ∆Pconv
condition. The reactive power controller implemented can be used C dc = (26)
either in power factor or terminal voltage control mode. dt U dc
The errors ( (i dr
*
− idr ) and
*
(iqr − iqr ) are driven to where ∆Pconv = Pgs − Pr , the difference between the active
proportional-integral controllers and the dq-coordinates of the rotor power supplied from the rotor circuit and the active power
voltage are derived. exchanged between the GSC and the grid. The parameters of the dc-
link, as well as the values of the KP and KI parameters of the PI
(6a) controllers are given in the Appendix.
ωr
Speed Control F. Initialization of the model
i*
qr One of the tasks which had to be implemented during the design
of this model was the initialization, thus the computing of the initial
Uqr values of the state variables of the model. In the literature, many
+
-
PI methods are proposed. For the solution of the initialization problem
in this case, an initial wind speed was assumed. Using the steady-
state equations of the generator, as well as the control equations,
iqr those derived in Section E, the initial values for all the state variables
of the model were derived. The results of the initialization algorithm
which was developed are given in the following Table:
TABLE I
(6b) RESULTS OF THE INITIALIZATION ALGORITHM
5
Uqs0 1 Te0 -0.8403 the last part comparative results with the 3rd order model are studied.
The wind turbine operates with unity power factor.
Uds0 0 Ψds0 1.0083
Qtot0 0 Ψqs0 2.6881e-005 1) System response in step change of the wind speed
A step reduction was applied at the wind speed, from the initial
Ptot0 -1 Ψdr0 1.0347 value 13 m/s to 7 m/s and the response of several system variables is
shown.
ωr0 1.2 Ψqr0 0.1522
5
s0 -0.2 Vdr0 0.0338 wr
4.5
Tw
ids0 0.0027 Vqr0 -0.1983 4 pitch angle
)
g
e
iqs0 -0.8333 Pgs0 -0.1667 d(
e
3.5
gl
n 3
a
idr0 0.3333 Ps0 -0.8333 h
ct
i
P, 2.5
)
iqr0 0.8611 u
p( 2
w
T,
) 1.5
u
p(
G. Low Voltage Ride Through - Crowbar g 1
w
and till now, the basic idea of the regulations remains the same. The
minimum voltage dip level and the maximum time duration of the Fig. 8. Rotor speed (pu), wind turbine prime torque (pu) and
event, for which the generator is obliged to stay ‘on-line’ are pitch angle when the wind speed decreases.
specified. Sometimes it must also contribute with reactive power In this case, when the wind speed decreases, the aerodynamic torque
injection. falls and the rotor is decelerated. The pitch control takes over control
A lot of novel ideas have been proposed in the bibliography. An and forces the pitch angle to zero, in order to maximize the extracted
effective way for a successful ride-through is the use of the so-called power. As mentioned before, for wind speeds below the nominal (12
crowbar, which is connected between the rotor circuit and the RSC m/s) the wind turbine operates in MPT model and the corresponding
[8]. High rotor currents are a serious problem for the semiconductor value of the pitch angle is zero. Parameters that are crucial in such
switches of the converters. The crowbar short-circuits the rotor and simulations are the rotor inertia H, the time constant of the
provides the transient currents with an alternative way, bypassing the servomechanism, as well as the maximum possible rate dβ/dt for the
RSC. When the transient phenomenon has passed, the RSC can regulation of the pitch angle. The response of electromagnetic torque
resume normal operation controlling the DFIG. In the next Sections is shown in Fig. 9.
results from voltage dips simulations will be analyzed. The circuit of
-0.2
the crowbar is shown in Fig. 7. Te
DFIG -0.3
-0.4
-0.5
IGBT )
u
p
(
e
Resistor T -0.6
-0.7
-0.8
Machine
side -0.9
20 30 40 50 60 70
converter time (sec)
H. Simulation results
In this Section the simulation results from the simulation of a 4-pole
2 MW DFIG wind turbine will be presented. The control strategy
analyzed in Section E was applied. Results from simulations under
wind speed variations and voltage dip simulations are presented. In
6
1.3
Us
1.2 0.9
1.1
0.8
1
)
u
p( 0.9 )
w pu( 0.7
T,
) 0.8 s
u
p( U
r 0.7
w 0.6
0.6
0.5
wr 0.5
0.4 Tw
0.3 0.4
15 20 25 30 35 40 45 50 19 20 21 22 23 24 25
time (sec) time (sec)
Fig. 10. Rotor speed (pu), wind turbine prime torque (pu) for an . Fig. 12. Voltage dip at the generator’s terminals.
increase in the wind speed.
6
30
ir
pitch angle
5
25
4
20
)
g
e )
d( u 3
el 15 p(
g ri
n
a
h
ct 2
pi 10
1
5
0
19 19.5 20 20.5 21 21.5 22 22.5 23
0
15 20 25 30 35 40 45 50 55 60 time (sec)
time sec)
. Fig. 11. Pitch angle response for an increase in the wind speed. Fig. 13. Response of the rotor current.
As the wind speed increases, the rotor is accelerated and the pitch 1.35
angle increases. The control ensures that the torque is limited close wr
to the nominal value, even if the wind speed is way above the 1.3
nominal. 1.25
terminal falls at the level of 0.4 pu for 200 ms. Before the evolution time (sec)
in the area of LVRT the WT would normally trip off to protect its Fig. 14. Response of the rotor speed.
power electronics from the high inrush currents. In this case,
however, the crowbar short circuits the rotor bypassing the RSC. The 5
Qs
generator has quite big thermal constants and can withstand the high
4
transients. During the operation of the crowbar, the RSC losses
control of the generator and the DFIG behaves as a typical wound 3
rotor induction machine with short-circuited rotor, thus absorbs
reactive power from the grid. )
u
p 2
(
s
Q
-1
19 19.5 20 20.5 21 21.5 22 22.5 23
time (sec)
Right after the fault, the high rotor current is detected and the
5
control closes the crowbar circuit. When the transient has expired, 3rd order
the IGBT of the crowbar opens, and the RSC operates as a typical 3- 5th order
4
phase diode rectifier, thus the IGBTs are still blocked. The rotor
current goes to zero and the machine is ready to go back to normal
operation. 3
1.1
) 3
u
p
( 1
ri
2 5th order
0.9
3rd order
1 0.8
)
u
p 0.7
(
0
19.8 20 20.2 20.4 20.6 20.8 21 s
d
time (sec) Z
0.6
Fig. 16. Response of the rotor current for various values of the 0.5
crowbar resistance.
0.4
3) Comparative results of the 5th and 3rd order model.
Neglecting the stator transients in the DFIG concept, where the 19.8 20 20.2 20.4 20.6 20.8 21 21.2 21.4 21.6 21.8
time (sec)
active and reactive power are controlled independently from each
other, results in neglect of the fundamental frequency transients in Fig. 19. Response of the stator flux in the 5th order (solid line)
all the electrical variables of the machine. This simplification is and the 3rd order model (dashed line).
generally accepted for small signal analysis. However, when
investigating the sensitivity of power converters in high currents, the IV. CONCLUSIONS
values for the currents extracted from the model must be precise. In rd
A 3 and a 5th order model of a variable speed, pitch controlled
the following Figures the same voltage dip has been applied to the wind turbine equipped with a DFIG has been implemented, using
generator’s terminal both for the 3rd and the 5th order model and the
Mathlab/Simulink. The various subsystems have been
comparative results are presented.
presented, including the aerodynamic system, the control
6 system and the generator. The dynamic model has been
3rd order
5th order developed with a vector controller for each converter,
5
decoupling the active and reactive power produced. The
operation of the model under MPT conditions has been
verified with the desired results. The LVRT capability of the
4
)
u 3
system has also been tested with satisfactory results using the
crowbar for short circuiting the rotor. The response of both
p
(
ri
Fig. 17. Response of the rotor current in the 5th order (solid
line) and the 3rd order model (dashed line).
8
VI. REFERENCES
[1] S. Heier: "Grid intergration of wind energy conversion
system", (John Wiley, Chisester 1998).
[2] CIGRE: "CIGRE Technical brochure on Modelling new forms
of generation and storage", TF 38.01.10, 2000, N. Hatziargyriou, M.
Donnelly, S. Papathanassiou, et al.
[3] J. G. Slootweg, H. Polinder, and W. L. Kling, "Dynamic
modeling of a wind turbine with doubly fed induction generator".
Presented at the IEEE Power Engineering Society Summer Meeting,
Vancouver, Canada, 15th-19th July 2001.
[4] A. Feijoo, J. Cidras, and C. Carrillo: "A third order model for
the doubly fed induction machine", Electr. Power Syst. Res., 2000,
56, (2), pp. 121-126.
[5] R. Pena, J. C. Clare, and H. Knudsen: "Modelling of windmill
induction generators in dynamic simulation programs". Proceedings
of the IEEE power technology ’99 conference, Budapest, Hungary,
29th August-2nd September 1999.
[6] E.ON Netz, Ergazende Netzanaschlussregeln fur
windenergieanlangen, Dec. 2001, Germany.