You are on page 1of 29

Chapter 8

Wind Turbine Generators


in Power Systems

8.1 Introduction
Wind turbine generators are increasingly being employed world-wide to
harness wind energy for the purpose of generating electricity. Worldwide
initiatives in smart grids have led to increasing interest in wind energy
conversion systems, not only because the primary resource is wind
which is free, but also because the environmental impact of conventional
generation systems is unsustainable. But the introduction of wind power
into the existing grids at bulk power levels has opened up new challenges
in system operation and control. Therefore, it is important to understand
the dynamics of wind turbine generators so as to enable high penetration
levels of wind power to be achieved in existing grids.
Most popular designs of wind turbine generators employ induction
generators or permanent magnet synchronous generators (PMSG).
The PMSG requires a power electronic converter interface, through
which the generated power flows into the network. The rating of the
converter, must match that of the machine, making these designs more
expensive. The induction generators are relatively cheaper owing to the
absence of power electronic converters, and more popular. For greater
controllability, at higher ratings the induction generators used are of
doubly fed type i.e. both stator and rotor are connected to a voltage
source. While the stator is directly connected to a three-phase network
(typically at 500-700 V), the rotor is also connected to the network
(usually through a transformer) via a back-to-back power electronic
converter with a capacitor connected in between as shown in Figure 8.1.
Such generators are called doubly fed induction generators (DFIG). This
chapter is concerned with the dynamics of DFIG systems.
274 Wind Turbine Generators in Power Systems

Figure 8.1: Grid connected DFIG based wind turbine generator

The amount of electrical energy that can be harvested from the


wind depends primarily on its speed; the power generated is a function
of the cubic root of the wind speed. For instance, increasing the wind
speed two times can boost the power output by eight times. The power
generated also depends on the cross-sectional area of the wind swept
by the rotating turbine blades, i.e. it varies as a square of the turbine
blade length. Physically the generators are located on top of a tower,
within a protective housing termed as nacelle. The nacelle also houses
the turbine rotor, mechanical controllers and drive-train for pitch and
yaw control, and the gear box. The pitch control regulates the angle
of the turbine blades with respect to the wind direction, while the yaw
control adjusts the direction in which the turbine faces. The gear box is
required to allow power transfer from the turbine rotating at 15-30 RPM
to the electrical generator which is locked to the grid frequency of 50/60
Hz. The nacelle is located on top of a tower, and its physical design
along with weight is critical in determining the height and stability of
the tower. The turbines have usually three-blades for maximum energy
extraction. Increasing the number of blades to four or higher increases
the energy extraction, but the corresponding increase in rotor weight is
too much to allow such designs to be practical. The taller the hub height,
the higher wind speeds the hub is likely to experience and therefore
annual energy production increases with increase in hub height. The
rate at which wind speed increases with height, depends primarily on
the slowing effect of the nearby geographical terrain on the prevailing
winds. Thus, in forested areas the wind speed slows down rapidly as the
height decreases. In such a location, taller towers are required to be able
to produce a desirable annual energy generation profile. Wind turbines
may also be located in the middle of the sea, in which case they are
referred to as offshore wind turbines. The advantage of offshore wind
turbines is that high power ratings may be achieved with relatively lower
Introduction 275

hub-height. The reason is due to the lubricating effect of the sea-surface;


wind speeds experienced at sea-level are comparatively unchanged as the
tower height increases. However, considerable infrastructure for power
transmission is required for evacuation of the generated power from
offshore wind turbines to terrestrial power grids.
There is a limit to the amount of electrical energy that can be
harvested from the wind. If all the kinetic energy contained within
the wind, were converted into electricity, the wind would stop at
the blades, and no further wind would be allowed to blow. Such a
situation is impossible, and in fact it can be proved using simple
physics that maximum blade efficiency is possible if the drop in wind
speed from upstream of the turbine to downstream approaches one
third. Theoretically, this allows converting upto 60% of the wind kinetic
energy into electricity, a limit termed as Betz efficiency. In modern
wind turbines, the maximum rotor blade efficiency is about 45-50%,
which is achieved by a combination of physical design of the turbine
rotor and blades, as well as control of the turbine speed to maintain
optimum tip-speed-ratio (TSR). The TSR is defined as the ratio of the
blade tip speed and the wind speed at just upsteam of the turbine,
l = wturbine R/vwind, where R is the radius of the turbine.
At very low wind speeds, below the cut-in speed, wind turbines
are switched off as the energy extracted is not enough to rotate the
turbines. Most wind turbines are designed to extract maximum energy
from the wind, during wind speeds up to the rated wind speed. This is
achieved by maintaining the rotor blade efficiency at its maximum value
by increasing the rotor speed in tandem with the wind speed. When
experiencing very high wind speeds (e.g. during storms), the turbine
is prevented from rotating, as rotation can produce severe mechanical
torques that can damage the turbine assembly. At extremely high wind
speeds, there are two ways of controlling the rotor speed. Stall controlled
turbines have their blades designed such that the aerodynamic torque
automatically drops as the wind speed exceeds a designed threshold
value. Without the need for any intervention, the rotor efficiency worsens
as the wind speed increases, until beyond the cut-out speed when the
turbine stops rotating. Active pitch controlled turbines on the other
hand have controllers that sense the wind speed/rotor speed and act
to pitch the blade (around its axis) away from the wind so that the
aerodynamic torque is reduced. The power developed in both these
two types of turbines show similar profiles for low wind speeds upto
the rated speed. Beyond the rated speed and upto the cut-out speed,
the pitch controlled turbines develop constant power, while the stall
controlled turbines develop reduced power. Beyond the cut-out speed,
all turbines are switched off.
276 Wind Turbine Generators in Power Systems

8.2 Turbine Characteristics


The coefficient of power extraction is a measure of the ability of the
wind turbine to extract power from the wind and is a mathematical
measure of the rotor efficiency. It is a nonlinear function of the TSR,
l, and the pitch angle, b, in degrees, with values ranging from zero to
one. The pitch angle, b, is defined as the angle of attack of the wind
on the axis of the turbine hub. When the turbine faces the wind speed
squarely, the pitch angle is said to be zero. Keeping the pitch angle
fixed, and varying the turbine angular speed can modify the coefficient
of power extraction. Specially designed look-up tables may be obtained
based on a turbine design, for finding the value of Cp(l, b) for particular
values of TSR and pitch angle. For the purpose of illustration, a generic
equation for the calculation of Cp(l, b) is used as
 
1 0.035
   – 21  l + 0.08 b – 3 
Cp (l, b) = 0.5176 116  1 0.035  b +1
– 3 – 0.4 b – 5  e  
  l + 0.08 b b + 1  
   
+ 0.0068l (8.1)
Maximum power extraction occurs at an optimum TSR, lnom,
corresponding to a maximum value for Cp(l, b) = CpMAX. While the
actual values of both CpMAX and lnom depend on the mechanical design
of the turbine blades, as per (1) they are 0.48 and 8.1 respectively, at
a pitch angle of 0° as shown in Figure 8.2.

Figure 8.2: Cp – l characteristics for different pitch angles


Turbine Characteristics 277

The aerodynamic power extracted by the turbine in watts is given by


1
Paero = ​ _ ​ rA​v3​wind​Cp(l, b) (8.2)
2
where r is the density of air (approx. 1.225 kg/m3), vwind is the wind
velocity in m/s, A is the area swept by the wind turbine in m2. Cp(l, b)
is a unit-less quantity indicating the coefficient of power extraction,
with a maximum value of CpMAX (typically 0.4 – 0.5 in modern three
blade wind turbines). For instance, a 1.5 MW wind turbine generator
with 26.2 m turbine blades can extract a maximum of 0.42 p.u. and
0.73 p.u. of power at 10 m/s and 12 m/s respectively. For a base wind
speed of 12 m/s, kopt is therefore 0.73.
A better way to understand (2) is to convert it into its per-unit
form. Dividing both sides by the turbine power base, SBASE
1
Pturb = rA​v3​wind​Cp(l, b) (8.3)
2S BASE
The wind speed may converted into per-unit (vw) by dividing by the
base wind speed, vBASE, and the value of Cp(l, b) may be normalised
against its maximum value, CpMAX.
Pturb = kopt ​v3​w​ ​​ C​ppu​ (8.4)

 rAv 3 C pMAX 
where kopt  = BASE
 indicates the maximum power (p.u.) that
 2S BASE 
 
can be extracted by the turbine at the base wind speed. For optimum
l v
power extraction, the turbine is rotated at a speed, wturbine = n om wind .
R
The normalised tip-speed ratio is defined as the ratio of the per-unit
turbine speed and the per-unit wind speed.
l wturb
lp.u. = = (8.5)
ln om vw

At steady state the per-unit electrical power output, equal to


the turbine power (neglecting losses), may be related to the per-unit
rotor speed (equal to the per-unit turbine speed at steady state) by
substituting for vw in (4) to get

Pe REF = PMPPT = kopt ​w​3r​ ​ (8.6)

Equation (6) assumes that the base rotor speed is the speed that
enables optimum power extraction at base wind speed. A generator
278 Wind Turbine Generators in Power Systems

controlled to follow the characteristics of eqn. (6) is said to be operating


under maximum power point tracking (MPPT) control. In Figure. 8.3
the grey lines indicate the aerodynamic characteristics of the turbine at
different wind speeds. The solid line is the MPPT characteristic of the
generator, achieved using various control strategies described later in
the chapter. Under MPPT control scheme, the generator is controlled
to always operate on the solid line. If the WTG is allowed to start
rotating under MPPT control from zero speed while being subjected to
a constant 12 m/s wind, the evolution of the generator power output
and the turbine power are described by the solid and the dashed lines
from zero to the point of operation – A. It may be observed that
the turbine power is always more than the generator power, therefore
the shaft accelerates till the two powers balance out each other at A.
If the wind speed further increases to 12.5 m/s, the turbine power,
jumps from A to C due to the change in the wind characteristics,
without any immediate change in the generator power ouput. Since the
turbine shaft is now subjected to a higher driving torque, it accelerates;
thereby increasing the turbine power from point C to B. During this
acceleration, the generator power output also increases from point A to
B simultaneously. At point B, the two powers are balanced, i.e. turbine
power equals the generator power output and hence the shaft speed
stabilizes at this point.

Figure 8.3: Turbine power characteristics


Modelling of DFIG 279

The operating point for the wind turbine generator is defined by


the intersection of the power-slip characteristics of the generator and
that of the turbine. In Figure 8.4, three turbine characteristics are
drawn for three different wind speeds – 11 m/s, 12 m/s, and 13 m/s.
The peak of the three curves occur at different slips i.e. 0.08, 0 and
– 0.08 respectively. The generator (considered as a doubly fed induction
generator (DFIG)) when operated under MPPT strategy produces
three different torque-slip characteristics in Figure 8.4, that result in
the curves C, B and A respectively. This is made possible due to the
variation in the rotor injected voltage, which results in MPPT operation.
The operating points are the intersection of the generator characteristics
for a given rotor voltage input and turbine characteristics for a certain
wind speed.

Figure 8.4: Intersection of generator and turbine characteristics

8.3 Modelling of DFIG
The equivalent circuit of a DFIG is given in Figure 8.5. The rest of the
discussion concerns with two pole machines where the electrical speed
equals the mechanical speed. In general, the electrical speed and the
mechanical speed are related through the number of pole pairs, np, as
w
wmech = elec . Figure 5 shows the rotor circuit referred to the stator
np
which is stationary while in reality the rotor circuit rotates at slip speed,
wr = s ws, where ws is the stator synchronous frequency.
280 Wind Turbine Generators in Power Systems

8.4 Equivalent Circuit
The electrical equivalent circuit of a DFIG consists of the stator
resistance, rs, in series with the stator and rotor leakage reactance,
xr and xr. The shunt branch reactance, xm, is significant in magnitude
and represents the magnetising inductance caused due to the presence of
the low-reluctance air gap within the machine. The rotor reactance, rr,
is also included in the rotor circuit in series with a controlled voltage
V
source, r , indicating the rotor injected voltage.
s

Figure 8.5: Equivalent circuit of a DFIG

By shifting the magnetising branch, containing current Im, to the


terminals of the circuit in Figure 8.5, the rotor current is approximately
calculated as
Vr
Vs –
Ir = s (8.7)
 r 
 rs + r  + j (x s + x r )
 s 
r
The rotor resistance, r , can be broken into two parts: rr and
1– s x
rr . rr is used to represent the actual power loss in the rotor
s
circuit (= ​I 2​r​ ​rr) due to the rotor current. Similarly, the injected voltage
1– s
can also be represented as a sum of two components: Vr and Vr .
s
The power absorbed/injected by Vr refers to the actual physical power
1– s
flow through the rotor side converter. The power developed in rr
s
Ê 1– sˆ 1– s
2
Á = I r rr ˜ and Vr together indicate the mechanical power
ÁË s ˜¯ s
developed by the DFIG. The power loss in the rotor resistance, rr, The
rotor active power (in p.u.) developed by the controllable source in the
rotor circuit is given by
Torque Slip Characteristics 281

Pm 1  (1 – s ) (1 – s ) 
Tm = = Vr I r cos q + I r2rr  (8.8)
wr wr  s s 
 
The torque-slip characteristics of the DFIG is given at different rotor
injected voltage in Figure 8.6. With respect to the stator voltage, the
injected rotor voltage, Vr, may be resolved into the in-phase component,
Vqr, and the out-of-phase component, Vdr . Independent control of these
two components can significantly affect the speed of the DFIG, as
evident from Figure 8.6. In general, the operation of the DFIG may be
summarized as follows: At sub-synchronous speeds, (s > 0), the DFIG
absorbs power through the rotor circuit, but generates power through
the stator. At super-synchronous speeds, the rotor circuit generates
power along with the stator.

8.5 Torque Slip Characteristics


By controlling the magnitude of the injected rotor voltage, the rotor
current can be modulated, thereby controlling the active torque and the
speed of the machine. On the other hand, controlling the phase of the
injected rotor voltage can change the power factor of the rotor circuit,
and therefore the entire machine. Thus, the active and reactive power
developed by the DFIG can be independently controlled by controlling
the magnitude and phase of the injected rotor voltage.

Figure 8.6: Torque-slip characteristics of a DFIG under different rotor


injected voltage.
282 Wind Turbine Generators in Power Systems

8.6 Dynamic Model of DFIG


The turbine and drive-train equations are as follows:

d wt 1
= (T – Tshaft ) (8.9)
dt 2H t turb

dqtw
= w B (wturb – wr ) (8.10)
dt

dwr 1
= (T – Te ) (8.11)
dt 2H g shaft

Pturb ¢ iqs + e ¢ds ids


e qs
where Tturb = , Tshaft = kshqtw, Te = . The rotational
wt ws
rigidity of the shaft is represented by the shaft stiffness constant, ksh,
and any mechanical damping is neglected.
The generator is represented by four electrical differential equations.
The stator voltage equations are as follows:

¢
deqs Ê ¢
e qs ˆ
= ws w B Á K 2rr ids – + s e ds
¢ – Kvdr ˜ (8.12)
dt Ë ws t r ¯

¢
deds Ê e¢ ˆ
= ws w B Á – K 2 rriqs – s eqs¢ – ds + Kvqr ˜ (8.13)
dt Ë ws t r ¯

The stator current equations are


diqs wB Ê 2 wr e¢ ˆ
= Á – (rs + K rr ) iqs + ws Lsc ids + ¢ – ds – vqs + Kvqr ˜
eqs
dt LSC Ë ws ws t r ¯
(8.14)

dids wB  2
e ′qs w 
=  − ws LSC iqs − (rs + K rr ) ids + + r eds
′ + Kvdr − vds 
dt LSC  ws t r ws 
(8.15)

1 Lr + Lm w
where K = , tr = , LSC = Ls + (Lr ||Lm ), and s = 1 – r .
Lr rr ws
1+
Lm
Open Loop Operation 283

8.7 Open Loop Operation


The DFIG is considered connected to an infinite bus, through a
transmission line of reactance, Xe (p.u.). With the infinite bus voltage
(V•) taken as reference, the active and reactive power flow through the
line together determine the magnitude and phase of the terminal voltage
of the generator, Vs –Ye. Therefore, two more algebraic equations are
required to describe these two unknowns.
P P
VsV•  
rotor
  
stator

Pgrid = sin g e = vdr idr + vqr iqr + vds ids + vqs iqs (8.16)
Xe

Q
Vs2 – VsV∞ cos g e  
stator

Qgrid = = vds iqs – vqs ids (8.17)
Xe

In (17), the reactive power exchange through the rotor circuit is


assumed to be zero. The rotor currents required in (16)-(17) are obtained
from the states as
e′qs
idr = – Kids (8.18)
Xm

e ′ds
iqr = – – Kiqs (8.19)
Xm
The complete model of the DFIG connected to an infinite bus consists
of differential equations (9)-(15) and algebraic equations (16)-(17) and
is termed as system of differential algebraic equations (DAE). The
parameters used for simulation are given in Table 9.1. The equations
may be written in the form
X = F(X, Z, U)

0 = G(X, Z, U) (8.20)
where X = [wturb, qtw, wr, eqs¢ , e¢ds, iqs, ids]T is the vector of system states,
Z = [Vs, ge]T is the vector of algebraic variables, and U = [vw, vqr, vdr]T
is the vector of inputs (wind speed and rotor injected voltage). If the
DFIG is to be connected to the infinite bus directly i.e. Xe = 0, its
terminal voltage magnitude and phase no longer depend on the power
injected into the infinite bus. In that case, (16)-(17) become irrelevant
and the system of DAE of (20) reduces to a set of seven differential
equations only.
284 Wind Turbine Generators in Power Systems

Table 8.1: Parameters for simulation of DFIG connected to infinite bus

wB ksh Ht Hg Lm Ls Lr Rs Rr Xe
314.16 0.3 4 0.4 5.419 0.167 0.132 0.0084 0.0083 0.06

The operation of a DFIG is compared for different values of rotor


injected voltage. In Figure 8.7 the wind speed is 12 m/s from 0-10 s,
13 m/s from 10-20 s and 11 m/s from 20-30 s. The first curve obtained
for vdr = vqr = 0, corresponds to a squirrel cage induction generator
(SCIG) running at super-synchronous speed. The other two curves are
for two different values of rotor injected voltage. For the SCIG, the
turbine speed remains relatively constant despite significant variations
in the wind speed. The SCIG always consumes reactive power, as a
result the terminal voltage is typically depressed. For case vdr = – 0.01,
vqr = 0.15, the active power is considerably reduced at higher wind
speeds, however the reactive power injections is high at different wind
speeds.

Figure 8.7: Open loop operation of DFIG. Wind speed is 12 m/s from
0-10 s, 13 m/s from 10-20 s and 11 m/s from 20-30 s.
Linearization of DFIG – Open Loop 285

The variation of C ​ ​ppu​ with changes in wind speed for the same
rotor voltage injections as in Figure 8.7 Open loop operation of DFIG.
is shown in Figure 8.8. It may be observed that the SCIG develops
optimum power (​C​ppu​ = 1) at 12 m/s, but the reduced value at other
wind speeds indicate that the MPPT operation is not achieved. On the
other hand, vdr = – 0.01, vqr = 0.15, helps the DFIG achieve MPPT
at 11 m/s, while vdr = – 0.01, vqr = – 0.03 helps attain MPPT at
13 m/s.

Figure 8.8: Variation of Cp(p.u.) for open loop operation.

8.8 Linearization of DFIG – Open Loop


In order to carry out a small signal stability analysis, the model of (20)
must be linearized around an operating condition. The partial derivative
of F and G with respect to the states and the algebraic variables
is calculated and the system matrix is computed at an operating
condition as
 –1 
∂F ∂F  ∂G  ∂G 
A= –   (8.21)
 ∂X ∂Z  ∂Z  ∂X 
  X 0 ,Z 0 ,U 0
The eigenvalues of A yield the natural modes of the system.
In open loop operation, the operating conditions depend on the rotor
injected voltage. The DFIG is assumed to be connected to the grid
286 Wind Turbine Generators in Power Systems

via a transmission line of reactance, Xe = 0.06 p.u. At wind velocity


11 m/s, the operating conditions are summarised for different rotor
voltages in Table 8.2.

Table 8.2: DFIG operating conditions for different rotor injected


voltage

vdr = 0 vdr = – 0.01 vdr = – 0.01


vqr = 0 vqr = – 0.03 vqr = 0.15
iqs 0.554 0.503 0.502
ids 0.279 – 0.491 – 2.204
e¢qs 0.905 1.179 1.789
e¢ds 0.166 0.145 0.130
Vs 0.983 1.029 1.132
Ye 0.033 0.031 0.025
wturb = wr 1.005 1.028 0.920
qtw 1.824 1.738 2.036

The eigenvalues of the DFIG at 11 m/s are shown in Table 8.3.

Table 8.3: Modal analysis of a DFIG in open loop at wind speed of


11 m/s.

Participating iqs, ids e¢qs, e¢ds, wr e¢qs, wtrub, qtw


states Æ
Rotor voltage Freq Damping Freq Damping Freq Damping

vdr = 0 49.96 Hz, 0.028 5.70 Hz, 0.114 0.52 Hz, 0.148
vqr = 0
vdr = – 0.01 49.96 Hz, 0.028 6.71 Hz, 0.115 0.55 Hz, 0.213
vqr = – 0.03
vdr = – 0.01 49.96 Hz, 0.028 9.30 Hz, 0.061 0.59 Hz, 0.365
vqr = 0.15

8.9 Control of Doubly Fed Induction Generators


The DFIG allows independent control of the machine power output
and the terminal voltage. Any control strategy must therefore have two
Control of Doubly Fed Induction Generators 287

different control loops, with minimal interaction between them. The


rotor flux magnitude and angle control (FMAC) is one such strategy,
in which the DFIG active power output is controlled by the rotor
flux angle, while the terminal voltage is controlled by the rotor flux
magnitude. The injected rotor voltage is obtained from the rotor side
converter, which is made to behave like a voltage source using pulse
width modulation switching. In the phasor diagram of Figure 8.9, the
internally induced voltage E¢ depends on the rotor flux yr, which may in
turn be modulated by the injected rotor voltage Vr. Thus, the terminal
voltage magnitude reference, VsREF, is used control the rotor injected
voltage magnitude, |Vr |, and the active power reference, PeREF, is allowed
to control the rotor voltage phase, dr, as shown in Figure 8.10. The grid
side converter is behaves like a current source, responsible to maintain
the dc link voltage, Vdc, to its reference value, VdcREF (taken as 1 p.u.)
and the power factor to unity, i.e. QcREF = 0.

Figure 8.9: Phasor diagram for DFIG operation.

Figure 8.10: Flux-magnitude angle control – rotor side converter control


288 Wind Turbine Generators in Power Systems

Figure 8.11: Flux-magnitude angle control – grid side converter control

The reference stator terminal voltage, VsREF, is assumed to be unity.


With the q-axis assumed to be aligned with the stator terminal voltage,
Vs –0° = vqs + jvds = vqs. The consolidated controller equations are
obtained from the figures above. For the rotor side converter controller,
the equations are
du1
= VsREF – Vs (8.22)
dt
du1
|E ′|REF = k p1u1 + ki1 (8.23)
dt
|E ′|

du2 (8.24)
′2 + e ds
= |E ′|REF – eqs ′2
dt
du2
|Vr | = kp 2u2 + ki2 (8.25)
dt
P Pe
  
ref

du3 3 (8.26)
= kopt wr – wrTe
dt
du3
dREF = ki3 u3 + kp3 (8.27)
dt
d 

du4 – 1 e′ (8.28)
= dREF – tan ds
dt eqs′

du4
d r = ki 4u 4 + k p 4 (8.29)
dt
Closed Loop Operation 289

′ iqs + eds
eqs ′ ids
The electric torque, Te, is calculated as Te = . The rotor
ws
voltage thus obtained, may be resolved into the d-q components as
vqr = Vr cos dr; vdr = Vr sin dr (8.30)
The dc-link voltage is governed by the energy balance between the
rotor side converter and the grid side converter.

dVdc  Protor   Pc 
CVdc = –   (8.31)
dt  vdr idr + vqr iqr   vd g idg + vqg iqg 
   
The rotor side and grid side converter currents are calculated as
eqs′ eds

idr = – Kids ; iqr = – – Kiqs (8.32)
ws Lm ws Lm
(vqg – Vs ) vdg
idg = – ; iqg = (8.33)
xTFR xTFR
1
where K = . The GSC controller regulates the dc-link voltage
Lr
1+
Lm
and the reactive power exchanged through the GSC.
du5
= VdcREF – Vdc (8.34)
dt
 
vdg = – xTFR  k u + k du5  (8.35)
 i5 5 p5
dt
 
  Qc
 
du6  
= QcREF –  vdg iqg – vqg idg  (8.36)
dt  
 

 du 
vqg = – xTFR  k16u6 + k p 6 6  – (VsREF – VS ) (8.37)
 dt 

8.10 Closed Loop Operation


In order to simulate the closed loop operation of the DFIG, along
with the FMAC controllers the states of the system are defined as
X = [wturb, qtw, wr, e¢qs, e¢ds, iqs, ids, u1, u2, u3, u4, Vdc, u5, u6]T.
The corresponding differential equations are (9)-(15), (22), (24), (26),
(28), (31), (34) and (36). The algebraic variables of the system are
290 Wind Turbine Generators in Power Systems

Z = [Vs, ge, vdg, vqg]T, corresponding to the equations (35) and (37)
and the following two equations modified from equations (16) and (17)
to include the effect of the converter and the dc link dynamics.
P
VV  
stator

s ∞
Pgrid = sin g e = vds idg + vqs iqg + vds ids + vqs iqs (8.38)
Ve
Q
Vs2 – VsV∞ cos g e  
stator

Qgrid = = vds iqs – vqs ids + vds iqg – vqs idg (8.39)
Xe

In order to solve for the behaviour of the DFIG, all the differential
and algebraic equations have to be solved simultaneously using an
appropriate solver. The plot of Cp in Figure 8.13 shows that the action
of the controllers restores the DFIG operation to its MPPT level after
every change in wind speed occurs.
The tuning of the controllers is key to the performance of the
FMAC strategy. There are six proportional-integral (PI) controllers,
and they affect the DFIG performance in different ways. The
sub-optimal values selected for the simulations in this chapter are
presented in Table 8.4.

Table 8.4: Controller parameters selected for simulation of DFIG


kp1 ki1 kp2 ki2 kp3 ki3 kp4 ki4 kp5 ki5 kp6 ki6
50 50 10 50 5 50 17.4 40 3.1 0.1 2.1 0.1

Figure 8.12: Closed loop operation of DFIG.


Linearization of DFIG – Closed Loop 291

Figure 8.13: Variation of Cp(p.u.) for closed loop operation of DFIG.

8.11 Linearization of DFIG – Closed Loop


The linearization of the DFIG under closed loop conditions follow a
similar methodology as used for the open loop system. Selected initial
conditions (Vs = 1–0° used as reference) at two different wind speeds
are shown in Table 8.5.

Table 8.5: Initial conditions of a DFIG in closed loop, at different wind


speeds.

vw = 11 m/s vw = 13 m/s
Pgrid 0.49 0.85
Qgrid 0.01 0.02
wr 0.92 1.08
E¢ 1.55–5.8° 1.57–8.4°
Vr 0.14–– 1.7° 0.13–– 163.6°
Is 1.91–– 72.2° 2.02–– 65.9°
Ir 2.15–106.2° 2.25–112.1°
ys 1.01–89.1° 1.00–89.1°
yr 1.59–95.8° 1.61–98.4°
292 Wind Turbine Generators in Power Systems

The key modes of the system are shown in Table 8.6 at 11 m/s.
The electromechanical mode for this system has a frequency of 1.81 Hz,
and this does not change appreciably with the wind speed.

Table 8.6: Modal analysis of a DFIG under closed loop control.

Participating states Frequency (Hz) Damping


iqs, ids, e¢qs, e¢ds 51.98 0.004
qtw, wr 1.81 0.067

8.12 Primary Frequency Response from Dfig-Wind Turbines


The ability of a generator to respond to network frequency changes so as
to aid its recovery in the event of load – generation imbalance is termed
as the primary frequency response of the generator. In synchronous
generators employed in conventional thermal and hydro power plants,
the turbine speed is locked to the grid frequency. Changes in the grid
frequency immediately manifest themselves in turbine speed deviations.
The kinetic energy corresponding to the turbine speed change, is injected
into or drawn from the network thereby temporarily mitigating the
load-generation imbalance. This kind of inertial response is key to the
frequency stability of grids comprising of large units rated at hundreds
of megawatts. Further, the turbine speed deviations are sensed by
speed governors which are usually armed with the ability to respond
by modulating the input steam (or water in the case of hydro plants)
so as to prevent further deviations. The control is proportional in
nature, without any reset action; the gain of the controller termed as
droop of the governor. The lower the droop the more is the response
of the governor to a frequency deviation, and consequently the better
is the frequency recovery following a disturbance. As per the latest grid
codes in most countries, conventional generating plants participating in
primary frequency regulation are required to set their droop between
3% – 6%.
In the case of DFIG based wind turbine generators, the rotor speed
is inherently decoupled from the grid frequency making it difficult to
obtain inertial response from these machines. In order to enable DFIGs
to provide network frequency support, the grid frequency is electrically
measured, and both the electrical power of the generator and the
mechanical power of the wind turbine are modulated through control
loops as shown in Figure 8.14. In order for this to be possible, it is
necessary that the wind turbine should initially have a positive power
margin, to be able to increase its generation as part of its primary
Primary Frequency Response from Dfig-Wind Turbines 293

frequency response. A positive pitch angle (bREF) at steady-state creates


this margin, although it results in some wind being wasted during
steady state operation. The benefit of this pitch angle based deloading
is that the percentage of deloading is independent of the prevailing wind
speed; rather it depends only on the value of bREF. In Figure 8.15 the
dashed lines refer to the turbine characteristics at different wind speeds
when deloaded, as compared to the solid lines indicating operation at
full pitch. When there is a load-generation imbalance event creating a
positive network frequency deviation due to excess generation in the
system, the pitch angle is increased to further decrease the aerodynamic
power harvested from the wind. Simultaneously, the power reference
(provided to the RSC) decreases from its MPPT value due to the action
of the droop controller. The result is that the DFIG power output drops
and settles to a lower value to allow the system frequency to recover.
In a similar manner, a negative frequency deviation indicative of lack
of generation in the system, results in a decrease of the pitch angle
and an increase of the RSC power reference. When due to a negative
frequency deviation in the network, the WTG power output has to be
increased by decreasing the pitch angle, the new turbine characteristic
becomes operational somewhere between the solid and the dashed lines
of Figure 8.15 for a particular wind speed. The operating point must
shift slightly to the right to accommodate this response i.e. the speed of
the turbine has to be increased marginally to allow primary frequency
response.

Figure 8.14: Primary frequency controller

The initial steady-state pitch angle, bREF, is carefully chosen to


achieve the desired deloading of the wind turbine. Figure 8.15 shows
the turbine characteristics for two different wind speeds, VA and VB, at
two different pitch angles – zero and bREF . The steady-state operating
point is the intersection of the turbine characteristics and the generator
RSC MPPT controller characteristic curve. It may be observed from
Figure 8.15 that the steady-state operating point shifts to the left due
294 Wind Turbine Generators in Power Systems

to deloading i.e. speed reduces, where the operating values of Cp(l, b)


operating point shifts to the left due to deloading i.e. speed reduces,
where the operating values of Cp(l, b) and l are C ​ ​ppu​ = CpREF and
lpu = lREF, both of which are less than one. At this operating condition,
the turbine torque is
koptC p vw3
pu
Tturb =
wturb
In steady-state both CpREF and lREF are independent of the wind speed,
and depend only on the steady-state pitch angle bREF. The sensitivity of​
 ∂C p   ∂C p 
C​ppu​ to changes in l  i.e. K(C , l ) =  , and b  i.e. K(C p , b ) =  , may
 p ∂l   ∂b 
be calculated from the turbine Cp(l, b) characteristics, as the slope of
the Cp – l and Cp – b curve at the operating point.

Figure 8.15: Pitch angle based deloading

In order to understand the dynamics of the primary frequency


response, a linearized model of the wind turbine generator is considered,
that neglects the detailed dynamics of the generator. The power reference
given to the RSC is considered to be the output of the generator. The
electromechanical equations are adapted from (9)-(11) as
 3 
d wt 1  koptC ppu vw
F1 = = – ksh qtw  (8.41)
dt 2H t  wturb 
 
Primary Frequency Response from Dfig-Wind Turbines 295

dqtw
F2 = = w B (wt – wr ) (8.42)
dt
 
Te
 
  K f fD  
dwr 1  2
F3 = = k q – k w –  (8.43)
dt 2H g  sh tw  opt r wr  
 
 

For small changes in the wind speed and the turbine/generator


speeds, F1, F2 and F3 may be linearized around the operating point
characterised by [​v​w0​, ​w​turb0​, ​w​r0​, ​C​ppu​], as

  k C v3 
 ∂F1  ∂  1 opt ppu w
  =  – ksh qtw 

 ∂wturb  ∂wturb  2H t  wturb




 k C v3 kopt vw3 ∂C p 
1 opt pREF w 0
= – + 0 pu

2H t  w 2 w ∂ w turb 
 turb0 turb0 

kopt  C pREF vw0 vw3 ∂C p ∂lpu 


3

= – + 0 pu

2H t  w 2 wturb ∂lpu ∂wturb 
 turb0 0 

kopt  C pREF vw0 


3
vw2
= – + 0
K (C p , l )
 (8.44)
2H t  w 2 w 
 turb0 turb 0

∂F1 ksh ∂F1


= ; = 0; (8.45)
∂qtw 2H t ∂w r

∂F2 ∂F2 ∂F2


= wB ; =0 = – wB ; (8.46)
∂wturb ∂qtw ∂w r

∂F3 ∂F3 ksh ∂F3 kopt wr


= 0; = ; =– 0
; (8.47)
∂wturb ∂qtw 2H g ∂w r Hg

The inputs to the linearized model are the deviations in the wind
speed and the network frequency from their initial operating conditions.
Accordingly, the sensitivity of F1, F2 and F3 to these inputs is
296 Wind Turbine Generators in Power Systems

∂F1  ∂C p 
1
=  3koptC pREF vw2 + kopt vw3 pu

∂vw 2H t wturb  0 0 ∂vw 
0

 ∂C p ∂l 
1 pu
=  3koptC pREF vw2 + kopt vw3 pu

2H t wturb  0 0 ∂lpu ∂vw 
0

1
= (3koptC pREF vw2 – kopt vw K(C , l ) wturb ) (8.48)
2H t wturb 0 0 p 0
0

 3 
kopt vw3 ∂C p kopt vw3 ∂C p ∂b
∂F1 1 ∂  koptC ppu vw 
= = 0 pu
= 0 pu

∂f D 2H t ∂fD  wturb  2H t wturb ∂fD 2H t wturb ∂b ∂f D


  0 0

kopt vw3
=
0
K(C Kb
, b) (8.49)
2H t wturb p
0

∂F2 ∂F2 ∂F3 ∂F3 K f fD


= 0; = 0; = 0; = ; (8.50)
∂vw ∂fD ∂vw ∂fD 2H g wr0

The state-space model of the controlled wind turbine generator is


written as

 ∂F1 ∂F1 ∂F1   ∂F1 ∂F1 


   
w   ∂wturb ∂qtw ∂w r  w   ∂vw ∂fD 
 turbD   ∂F ∂F2 ∂F2   turbD   ∂F ∂F2  vw 
 qtw  =  2
  qtw  +  2   D  (8.51)
 D ∂ w
  turb ∂qtw ∂w r   D   ∂vw ∂fD   fD 
 wrD   ∂F ∂F3 ∂F3   wrD   ∂F ∂F3 
 3   3 
 ∂wturb ∂qtw ∂wr   ∂vw ∂fD 

The output of the system, being the deviation in the electrical power
output of the wind turbine generator, may be considered equal to the
change in PeREF if the generator dynamics are ignored.
​P​eD​ = 3kopt w​r 2​0​ ​​ w​rD​ – Kf fD (8.52)

The initial values of the states and algebraic variables in (51) may
be obtained by setting the left hand side of (41)-(43) to zero and
solving simultaneously. At 10 m/s, the values obtained for 10% deloading
are v0 = 0.833, ​wturb
​ 0​ = w​ r​ 0​ = 0.8054, kopt = 0.73, b = 1°, Kf = 15,​
Primary Frequency Response from Dfig-Wind Turbines 297

C​pREF​ = 0.9029, lREF = 0.96648, Kb = 600. The sensitivity of ​C​ppu​ to


both l and b calculated from the turbine Cp(l, b) characteristics are​
K​(C , l)​ = 0.0771 and K
​ ​(C , b)​ = – 0.1421. The inertia constant of the
p p
high speed and low speed shafts are 0.4s and 4s respectively, while the
spring constant is ksh = 0.3.
The eigenvalues of (42) are – 0.195 and – 0.671 ± 11.357. The
oscillatory mode corresponds to an electromechanical mode of frequency
1.81 Hz and damping – 0.06. These values may be compared with the
eigenvalues obtained for the detailed model. The step response to a
0.12 m/s rise in wind speed and a 0.5 Hz rise in grid frequency is
shown in Figure 8.16. The change in the power output of the wind
turbine resembles a first-order system, and this may be verified by
calculating the transfer function between the wind speed and the
power output, from (49)-(50). On the other hand, the response of
power to a rise in grid frequency shows oscillations that quickly die
out. Another important feature of this response is the availability of
reserve immediately after a disturbance has occurred. This feature has
been employed to design power oscillation damping controllers from such
wind turbine generators.

Figure 8.16 Step response of linearized model of wind turbine generator


An alternative generator deloading scheme is possible, wherein the
deloading is towards the right of the MPPT curve.
298 Wind Turbine Generators in Power Systems

P1 = kdel w3 (8.53)
3
lnom
where kdel = kopt . The value of ldel should be such that the
3
ldel
corresponding Cp(l, b) obtained from equation (1) is deloaded as per
the requirements. For instance, setting ldel = 9.5909 instead of its
nominal value of 8.1 yields Cp(l, b) = 0.432, kdel = 0.44. Consequently,
the power is deloaded by 10% at every wind speed, when deloaded to
the right. Figure 8.17 shows the deloaded power reference curve along
with the MPPT curve, superimposed upon the turbine characteristics
at a wind speed of 10m/s. The operating point shifts to the right due
to deloading i.e. instead of operating at w0 to generate P0, it operates
at a higher w1 to generate P1, which is 90% of P0. During primary
frequency response, the operating point is shifted along the straight line
to {w, Pdel} as indicated; the new power reference is calculated as

P0 – P1
Pdel = P1 + (w1 – w ) (8.54)
w1 – w0
At high wind speeds, and non-zero pitch (owing to pitch based
deloading), the speed increase associated with deloading can become
a limiting factor in deciding the extent of deloading because the rotor
side converter rating depends on the maximum slip.

Figure 8.17 Deloading to the right


Remarks 299

8.13 Remarks
Increasing the penetration of wind generation into bulk power grids,
can deteriorate the system frequency response in many ways. Firstly,
replacing conventional synchronous generators (employed in thermal,
hydro and nuclear power plants) with induction generators (associated
with wind generation systems) can lead to significant reduction in system
inertia. This is because the speed of induction generators interfaced to
the grid via power electronic converters, is de-linked from the system
frequency. Hence, any small disturbance can lead to increased frequency
excursions. Further, speed governors and automatic voltage regulators
employed with conventional generators provide primary and secondary
frequency response as well as power oscillation damping, and aid the
recovery of the system during disturbances. These controller technologies
have not matured enough in the context of induction generators, so their
impact is reduced. Moreover, an inherent uncertainty is associated with
wind as a resource, and the spinning reserve available at any point of
time due to deloading of wind generators is unreliable. Thus, the overall
system stiffness decreases with increase in wind generation.
The annual capacity factor (CF) of any generator is defined as the
total annual energy produced divided by the product of the rating of
the generator and the number of hours in a year (= 8760). For a wind
farm, the CF is typically in the range of 20% – 30%. Considering that
all the wind turbines in a wind farm are deloaded by 10%, the CF
may reduce to 27% from 30%. Due to various frequency events, and
opportunities for the wind turbine to contribute its primary frequency
reserve, this CF may ultimately improve from 27% to somewhere
close to 30%. Therefore, the loss of revenue due to deloading, must
be compensated by appropriate ancillary service pricing mechanism to
actively encourage wind farm operators to opt for deloading strategies.
The selection of the frequency droop factor, Kf, also has an important
impact on the CF of the wind turbine. Generally, at locations with
high prevailing wind speeds, the spinning reserve is likely to higher and
therefore Kf should be chosen to enable maximum frequency response.
The CF accordingly would improve substantially.

8.14 Multi-Machine Simulation
The need for simulation of multiple wind turbine generators may
arise when wind farms connected to a grid have to be studied. If the
dynamic behaviour occurring at the high-voltage grid is of interest, it
is frequently convenient to represent the entire farm as a dynamically
equivalent machine with the aggregated injection. This is because the
300 Wind Turbine Generators in Power Systems

individual dynamics of the relatively small wind turbines have hardly


any impact on the grid dynamics. Rather, the grid is affected by the
variations in voltage and injection at the point-of-common coupling,
which justifies reduced order modelling of wind farms.
Sometimes, particularly in the case of offshore wind farms employing
dc grids, the dynamics of individual wind turbines assumes significance
especially if the grid connection is weak. In such a case the entire
low-voltage or dc network has to be modelled in an appropriate manner
considering high R/X ratio of the feeders and lines interconnecting the
network as well as voltage unbalance within. The impact of modelling
various generators located on the high-voltage grid has reduced
importance in that case.

8.15 Conclusion
In this chapter detailed modelling of DFIG based wind turbine generators
is discussed and its linearization presented. Small signal analysis of the
DFIG was carried out at different wind speeds. The ability to respond
to network frequency disturbances was also analysed using mathematical
models.

8.16 References
1. O. Anaya-Lara, N. Jenkins, J. Ekanayake, P. Cartwright, and M.
Hughes, Wind Energy Generation – Modelling and Control, John
Wiley and Sons Ltd., United Kingdom, 2009
2. B. Wu, Y. Lang, N. Zargari, and Z. Kuo, Power Conversion and
Control of Wind Energy Systems, Wiley-IEEE Press, New Jersey,
2011
3. Y. Mishra, S. Mishra, M. Tripathy, N. Senroy, Z. Y. Dong, “Improving
stability of a DFIG-based wind power system with tuned damping
controller,” IEEE Trans. Energy Conv., vol. 24, No. 3, pp. 650-660,
2009
4. S. Ghosh and N. Senroy, “Electromechanical dynamics of controlled
variable speed wind turbines,” IEEE Sys. J., vol. 9, No., pp. 639-
646, 2015
5. F. M. Hughes, O. Anaya-Lara, N. Jenkins, and G. Strbac, “A power
system stabilizer for DFIG-based wind generation,” IEEE Trans.
Power Sys., vol. 21, No. 2, pp. 763-772, May 2006
6. R. Datta and V. T. Ranganathan, “A method of tracking the peak
power points for a variable speed wind energy conversion system,”
IEEE Trans. Energy Conv., vol. 18, No. 1, pp. 163-168, 2003
References 301

7. E. Muljadi and C. P. Butterfield, “Pitch controlled variable speed


wind turbine generation,” IEEE Trans. Ind. Appln., vol. 37, No. 1,
pp. 240-246, 2001
8. K. V. Vidyanandan and N. Senroy, “Primary Frequency Regulation
by Deloaded Wind Turbines Using Variable Droop,” IEEE Trans.
Power Sys., vol. 28, No. 2 pp. 837-846, 2013.

You might also like