You are on page 1of 12

Electrical Power and Energy Systems 125 (2021) 106439

Contents lists available at ScienceDirect

Electrical Power and Energy Systems


journal homepage: www.elsevier.com/locate/ijepes

A novel model of the lower active power limit (LAPL) of DFIG under T
overspeed mode

Shenghu Lia, , Min Wangb
a
School of Electrical Engineering and Automation, Hefei University of Technology, Hefei 230009, China
b
College of Energy and Electrical Engineering, Hohai University, Nanjing 211100, China

A R T I C LE I N FO A B S T R A C T

Keywords: The downward dispatch to the doubly-fed induction generators (DFIGs) may prevent the synchronous generators
Doubly-fed induction generator (DFIG) (SGs) from frequent shut-down during light loading. There are papers using power reserve of the DFIG under the
Lower active power limit (LAPL) overspeed mode for power dispatch or frequency control, and other papers quantify the upper active power limit
Rotor-side converter (RSC) of the DFIG, but none quantifies the lower active power limit (LAPL) of the DFIG.
Overspeed mode
In this paper, it is newly found that the capacity of the rotor-side converter (RSC) limits the LAPL of the DFIG
Sensitivity
Power dispatch
under the overspeed mode. By including this constraint, the analytical solution to the LAPL of the DFIG is newly
proposed. A sensitivity model is newly proposed to quantify the dependency of the LAPL on the DFIG’s para-
meters. The ratios of the reserved power and the reserved kinetic energy with respect to the maximum output of
the DFIG, i.e. RRP, RRE, and the ratios of the reserved power and the reserved kinetic energy with respect to the
rated capacity of the RSC, i.e. RRPC and RREC, are newly defined to evaluate the benefit of the LAPL of the DFIG.
The LAPL of the DFIG is newly included in the optimal power flow (OPF) to quantify its benefit to the SGs and
the power systems.

1. Introduction of power systems with the wind power, in the hourly or daily time
scale [4-6].
The synchronous generators (SGs) are often expected to provide the (ii) It is expected that the WTGs be controlled for the system purposes.
power reserve to maintain active power balance and frequency stability The most common application is frequency regulation of the time
against the random load changes. Recently, the increasing fluctuating scale of seconds [7-12]. With the input of the frequency deviation
wind power intensifies the need of power dispatch capability [1-3]. For or its change rate, the droop control, the virtual inertia control [13-
the power systems with the wind turbine generator (WTG), e.g. the 15], or combination of both [16], is designed to increase the active
doubly-fed induction generator (DFIG), the existing researches about power or release the kinetic energy of the DFIG, similar to the SGs
the dispatchability and controllability of the wind power may be [17]. To keep the reserve, the underspeed control, the overspeed
roughly classified by, control, or the pitch control, may be applied to drive the wind
turbine (WT) from the optimal point. Among them, the overspeed
(i) It is assumed that the DFIGs operate at the maximum power point control has the merit of storing more kinetic energy on the rotors.
tracking (MPPT) mode, hence are not dispatchable. Imbalance of
the active power of the system is counteracted by the SGs. These If the output of the DFIGs is dispatchable [18], power dispatch to
studies often concern more on the probabilistic nature of wind the system will be more flexible, and burden on the SGs will be alle-
power than on detailed configuration and control of the DFIG. viated. To fulfill it, a challenging task is that the DFIGs be dispatchable
Many papers are about the unit commitment, the optimal power not only upward, but also downward, since with light loading, the SGs
flow (OPF), the probabilistic power flow, or the reliability analysis have to operate at the lower active power limit (LAPL) or shut-down.

Abbreviations: DFIG, Doubly-fed induction generator; GSC, Grid-side converter; LAPL, Lower active power limit; MPPT, Maximum power point tracking; OPF,
Optimal power flow; RRP, Ratio of reserved power with respect to PDFIG,max; RRE, Ratio of reserved kinetic energy with respect to EDFIG,max; RRPC, Ratio of reserved
power with respect to SRSC,N; RREC, Ratio of reserved kinetic energy with respect to SRSC,N; RSC, Rotor side converter; SG, Synchronous generator; WT, Wind turbine;
WTG, Wind turbine generator

Corresponding author.
E-mail address: shenghuli@hfut.edu.cn (S. Li).

https://doi.org/10.1016/j.ijepes.2020.106439
Received 25 April 2020; Received in revised form 23 July 2020; Accepted 8 August 2020
Available online 17 August 2020
0142-0615/ © 2020 Elsevier Ltd. All rights reserved.
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

Nomenclature Pem Electromagnetic power.


J Jacobian matrix.
Notations x, u Unknowns and contributing factors.
ΔF Steady-state constraints of DFIG.
υw, ρ Wind speed and air density. ξ Sensitivity.
η, p Gear ratio and pole pairs. f, g Equality and inequality constraints in the OPF.
ω, ω0 Rotor speed and synchronous speed.
s, β Slip and pitch angle. Superscripts
D, A Diameter and sweeping area.
λ, λi Tip speed ratio, intermittent variable. * Conjugate.
Cp Power utilization coefficient of WT.
c8, c9 Coefficients of Cp function. Subscripts
T, H, E Torque, inertia, and kinetic energy.
R, X Resistance and reactance. s, m, r, g Stator, magnetizing circuit, rotor (or RSC), and GSC.
V = V ∠θ Voltage, its magnitude and angle. op, cr, set The optimal, critical, and set values.
I, I Current and its magnitude. max, min The maximum and the minimum values.
P, Q, S Active, reactive, and apparent powers. N, * Rated value, per unit value.

The importance of the downward dispatchability with wind power has (i) The DFIG at the LAPL has its active power output less than that
been recognized by the researchers and the power utilities [19-21]. under the MPPT mode. Hence to quantify the LAPL, the analytical
The spinning reserve of the SGs is common in power systems. The solution model for the DFIG under the MPPT mode and the dis-
power reserve of the DFIG is similar, with the purpose of providing fast patched mode should be derived.
response to power imbalance and frequency drop after disturbances and (ii) Among the factors influencing the active power of the DFIG, e.g.
faults. Shut down of SGs and DFIGs may also decrease the generation, the wind speed, the rotor speed, the pitch angle, and the active and
but its re-start needs time, hence can not provide fast response to dy- the reactive power settings to the DFIG, etc, which one is critical to
namic events. the LAPL of the DFIG?
Another concerns of the power reserve is about the security of the (iii) If the main constraint to the LAPL is known, we may repetitively
DFIG itself. If the rotor speed is larger than the rated value, the me- solve the DFIG until the main constraint is violated, then the active
chanical stress will shorten the lifetime of the WT. But “overspeed” here power before violation is the LAPL. But this process is computa-
means that the rotor speed is higher than the optimal speed instead of tionally expensive. An analytical solution to the LAPL is expected.
the rated value; it is still below the rated speed, hence the adverse (iv) For power dispatch, the operational range of the generators should
impact on the WT can be ignored. be relatively fixed. Hence it is to be verified if the LAPL of the DFIG
To get power reserve with the downward dispatch, the operational changes obviously with the stator voltage. If so, it is not suggested
region (or range) of the generators is to be known. For the SGs, the to include the LAPL of the DFIG into power system dispatch model.
upper active power limit is decided by the rated capacity, and the LAPL
is about 30–40% of the rated capacity to maintain the stable operation It may be expected to quantify the LAPL from experimental test,
of the boiler and the steamer. For the DFIG, the upper/lower reactive which is however expensive and sometimes destructive. Quantifying
power limits are decided by the thermal ratings of the stator and the the LAPL with the analytical model is more desirable, and common for
rotor and capacity of the converters, which have been well studied. The power system calculations, e.g. the OPF, frequency regulation, fault
upper active power limit of the DFIG is the maximum captured power analysis, dynamic control, and stability control.
by the WT minus the loss of DFIG, as seen in [22] for a matrix con- In this paper, the LAPL of the DFIG is newly proposed to enhance
verter-fed DFIG. It is shown in [23] that the maximum active power the downward dispatch capability of power systems. It is newly found
capability of the DFIG integrating to weak grid is dependent more on that the capacity of the rotor- side converter (RSC) is a constraint to the
the power angle limit instead of the rotor current limit. As to the lower LAPL of the DFIG under the overspeed mode. By including this con-
active power limit of the DFIG, we research the digital libraries of IEEE, straint, the analytical solution to the LAPL of the DFIG is newly pro-
IET, Elsevier, Springer, John Wiley & Sons, Taylor & Francis, China posed. A sensitivity model is newly proposed to quantify the de-
National Knowledge Infrastructure (CNKI), etc. It is found that there are pendency of the LAPL to system conditions. Four indices are newly
many papers about the reactive power limit of the DFIG, some papers proposed to evaluate the benefit of the LAPL. The LAPL of the DFIG is
about the upper active power limit of the DFIG, but no paper about the newly included in the optimal power flow (OPF) model to quantify its
lower active power limit of the DFIG. benefit to the SGs and power system. Numerical results are given to
The difficulties of quantifying the LAPL of the DFIG lie in, validate the feasibility and accuracy of the proposed model.

Fig. 1. Configuration of DFIG.

2
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

D
2. Solution of DFIG under MPPT mode and power reserve ω0 (1 − s )
2
ηp
const(λ ) = λ − =0
In Section 2.1, the currents and the powers along the windings and υw (16)
the converters of the DFIGs are defined, which follow the law of the
where ω0 is the synchronous speed, p is the pole pairs, s is the rotor slip,
electrical circuits and power capture characteristics of the DFIG, as seen
so ω0(1-s)/p is the rotor speed of the induction generator. η is the gear
in many early literatures [24-34]. In Section 2.2, solution to the DFIG
ratio, D is the diameter of the WT, so ω0(1-s)D/2/η/p is the linear ve-
under the MPPT and the power reserve modes are given, which are the
locity at the tip of the WT, whose ratio to υw is the tip speed ratio.
previous work of the 1st author [35]. These equations provide the basis
The intermittent variable λi is constrained by (17), where c8 and c9
to derive the lower active power limit of the DFIG in Section 3.
are the coefficients in the Cp function.
1 1 c
2.1. Power and torque balances of DFIG const(λi ) = − + 3 9 =0
λi λ + c8 β β +1 (17)
The configuration and the equivalent circuit of the DFIG are shown The shaft dynamics is constrained by the torque balance (18), where
in Fig. 1. Some notations are defined for the DFIG. The voltage V is T is the torque.
given by the magnitude V and the angle θ. P, Q, and S are the active,
PWT
reactive, and apparent powers respectively. R and X are the resistance ΔT = − − Pem = 0
1−s (18)
and the reactance respectively. s is the slip of the rotor. Re denotes the
real part. The superscript * denotes the conjugate. The subscripts s, m, r, When solving the parameters of the DFIG, including all the con-
and g denote the stator, the magnetizing path, the rotor (or the RSC), straints or not is decided by the operation modes of the DFIG.
and the GSC. The subscript set denotes the reference setting. The sub-
script N denotes the rated value. The subscript op denotes the optimal 2.2. Solution to DFIG under the MPPT and the power reserve modes
value.
The currents I of the stator, the rotor, and the filter of the DFIG are For the DFIG under the MPPT mode, s and the optimal output of the
given by, WT (PWT,op) are known, so (16)(17) are unnecessary. Since the active
power loss PDFIG,loss is related to Vs which changes with system condi-
Vs − Vm
Is = tion, the maximum active power output of the DFIG, i.e. PDFIG,max, is
Rs + jXs (1) unknown. Hence ΔPs = 0 is deleted, and the effective constraints are
given by (19) [35], where JFx is the Jacobian matrix.
Vr s − Vm
Ir =
Rr s + jXr (2) ⎡ Δθm ⎤
⎡ ΔT ⎤ 0
⎢ ΔQs ⎥ ⎢ ΔVm ⎥ ⎡ ⎤
⎢ ⎥ ⎢ 0⎥
Vg − Vs ⎢ ΔPm ⎥ Δθr
Ig = ⎢ ⎥ + JFx ⎢ ⎥ = ⎢0⎥
Rg + jXg (3) ⎢ ΔQm ⎥ ⎢ ΔVr ⎥ ⎢ 0 ⎥
⎢ ΔPg ⎥ ⎢ Δθg ⎥ ⎢ 0 ⎥
The corresponding branch powers are given by (4)–(9), where the ⎢ ΔQ ⎥ ⎢ ⎥ ⎢ ⎣0⎦

⎢ ΔV
subscripts denote the power direction, e.g. s,m denotes power from s to ⎣ g ⎦ ⎣ g⎥ ⎦ (19)
m. After iterative solution, PDFIG is given by (20), which is actually
Ps,m + jQs,m = Vs I s* PDFIG,max under the MPPT mode.
(4)
PDFIG = −Ps,m − Ps,g (20)
Ps,g + jQs,g = −Vs I g* (5)
It is clear that due to dependency of power loss on the stator vol-
Pm,s + jQm,s = −Vm I s*, Pem = Pm,s tage, the active power output of the WT and the DFIG, i.e. PWT and
(6)
PDFIG, can not be set simultaneously. Hence the constraints under the
Pm,r + jQm,r = −Vm I r* reserve differ from those under the MPPT mode. With the active power
(7)
setting PDFIG,set < PDFIG,max, power reserve is possible.
Pr,m = Re(Vr I r*) (8) Estimate the rotor speed ω from the power curve for the overspeed
mode, and keep ΔPs as a constraint. The effective constraints for the
Pg,s = Re(Vg I g*) DFIG with power reserve under the overspeed mode are given by [35],
(9)

Theoretically, the buses s and m are constrained by the active and ⎡ ΔPs ⎤ ⎡ Δθm ⎤
reactive power balances,
⎢ ΔQs ⎥ ⎢ ΔVm ⎥ ⎡ 0 ⎤
⎢ ΔPm ⎥ ⎢ Δθ ⎥ ⎢ 0 ⎥
ΔPs = −Ps,m − Ps,g − PDFIG,set = 0
⎢ ⎥ ⎢ r ⎥ ⎢0⎥
(10) ⎢ ΔQ m ⎥ ⎢ ΔVr ⎥ ⎢ 0 ⎥
⎢ ΔPg ⎥ + JFx ⎢ Δθ ⎥ = ⎢ 0 ⎥
g
ΔQs = −Qs,m − Qs,g − QDFIG,set = 0 (11) ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ ΔQg ⎥ ⎢ ΔVg ⎥ ⎢ 0 ⎥
⎢ ΔT ⎥ ⎢ Δs ⎥ ⎢ 0 ⎥
ΔPm = −Pm,s − Pm,r = 0 (12) ⎢ ⎥ ⎢ ⎥ ⎢0⎥
const(λ )
⎢ i ⎥ ⎢ Δλi ⎥ ⎣ 0 ⎦
ΔQm = −Qm,s − Qm,m − Qm,r = 0 (13) ⎢
⎣ const(λ ) ⎥⎦ ⎣ Δλ ⎦ (21)

The RSC and the GSC are constrained by active power balance and It should be noted that the DFIG may operate at the super- or sub-
control setting (14(15). The reactive power setting to the RSC is synchronous speed. The sign of s is troublesome to the convergence of
QDFIG,set–Qg,set. (21). By setting the initial value of θr according to the sign of s, a
modified flat start is proposed in [35] to get the solution with 5 to 7
ΔPg = −Pr,m − Pg,s = 0 (14)
iterations.
ΔQg = Qs,g + Qg,set = 0 The active power loss of the DFIG is the sum of those of the stator,
(15)
the rotor, and the transformer (or inductance) (22). The reactive power
The tip-speed ratio λ of the wind turbine (WT) is constrained by, loss is the sum of those of the stator, the rotor, the transformer (or

3
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

inductance), and the magnetizing circuit. It should be noted that when ΔEWT = EWT − EWT,op = HWT (ω*2 − ωop*
2
) PDFIG,N (24)
the DFIG is set for power reserve, usually the power loss and energy
conversion efficiency are not the most important. If one wants less loss, where ρ is the air density, A is the sweep area, Cp is the power utili-
it is desirable to regulate the pitch angle instead of the rotor speed. But zation coefficient, λop is the optimal value of the tip-speed ratio λ, β is
the pitch angle regulation has the drawbacks of more mechanical stress the pitch angle, and HWT is the inertia of the WT. The subscript * de-
and slower response, thus not recommended. notes per unit value. Eqs. (23)(24) are existing expressions, but with
them, we will find a new conclusion about the constraint of the rotor
⎧ P loss = Is2 Rs + Ir2 Rr + Ig2 Rg speed on the lower active power limit of the DFIG.
2
Vm
For power reserve, the power setting to the WT, i.e. PWT,set, is
⎨Q loss = Is2 Xs + Ir2 Xr + Ig2 Xg +
⎩ Xm (22) constrained by,
PWT,set ⩽ min{PWT,N, PWT,op (υ w )} (25)
The control strategy of the DFIG under power reserve is shown in
Fig. 2 [18,36]. Under the MPPT mode, the rotor speed is decided by the For υW1, to releasing the reserved power, the operation point moves
upper/lower limits and the wind speed. The active and the reactive from B to A (the red curve), but PWT exceeds PWT,N, and the overspeed
powers of the DFIG are within the rated value. The current/power mode is not allowed. Hence the practical operation point should be
limits of the stator/rotor and the converters will not be violated. Only below the rated output of the WT, i.e. υW should be less than υW,cr,
under certain circumstances will these limits be violated, for example, otherwise it is suggested to adjust β for power reserve, which however
can not release the kinetic energy.
(a) Fault-ride through. With sudden drop of the stator voltage, the To compare the adjustable range of power reserve, adopt υw2 and
stator/rotor currents increase remarkably. To avoid the over- υw3 less than υW,cr, and υw2 > υw3, hence PWT,op < PWT,N. Power re-
current, the limiters are introduced for dynamic analysis and con- serve under the overspeed mode is possible by moving from G to F, or
trol for the ride through of the DFIG. from I to H. Compared with that of υW3, PWT,op of υW2 is obviously lager.
(b) Overspeed mode for active power reserve. The rotor speed is set One may have the impression that the larger υW yields more power
larger than the optimal speed. The slip power may be large enough reserve capability. But it does not consider the capacity constraint of
to violate the capacity of the rotor (or RSC), as shown in the Section the converters. To show the constraint of the converters on the LAPL,
3. PDFIG,loss is temporarily ignored. The active slip power Pr,m is propor-
tional to the electromagnetic power Pem (26), where s = 1–ω*, the
The rotor speed of the WT is usually controlled by the hydraulic subscript * denote the per unit.
motor in one of the manners, (1) during normal operation, the intense
|Pr,m| = |s||Pem| (26)
change of the rotor speed due to the torque imbalance will activate the
pitch control to increase the pitch angle, restore the rotor speed, and Under the MPPT mode, both |s| and |Pem| are not very large, so Pr,m
smooth the output of the DFIG, (2) during the power reserve with the is limited. But for power reserve under overspeed mode, ω is much
overspeed control, the rotor speed is regulated to the preset value, and higher. The slip power may be large enough to violate the RSC’s ca-
the pitch control is not required to restore the rotor speed. pacity. With the higher υW or the lower PDFIG,set, the converter is easier
It should be noted that the overspeed control is one among many to be overloaded. To avoid it, the LAPL, i.e. the minimum PDFIG,set, is to
choices to keep the power reserve. For primary frequency regulation be decided.
during longer period, the pitch control with slower response may also
be used. The fast response of the overspeed control is preferred to 3.2. Solution to the LAPL of DFIG constrained by SRSC,N
provide the inertia during short period. But when the rotor speed drops
below the optimal value, the captured output of the WT reduces, which With the converter loss ignored, active power outputs of the RSC
will counteract the released inertia, possibly yielding further frequency and the GSC are same (with opposite directions). But their reactive
drop. powers are different, even the stator receives the same reactive powers
from them. The reason is that reactive power loss of the induction
3. The LAPL of DFIG constrained by RSC capacity generator is larger than that of the transformer (or the filter). Hence
even with the balanced design to the converters, the RSC is more
3.1. Critical constraint to the LAPL of DFIG under overspeed mode

The WT operates under the MPPT mode with the optimal speed ωop,
yielding the maximum output PWT,op at the points A, D, F, and H for
different wind speeds (υw), as shown in Fig. 3, where υw1 denotes the
wind speeds with the output larger that that of the WT, υw2 and υw3
denote the wind speeds with the output less that that of the wind tur-
bine or the DFIG, and the critical wind speed υW,cr yields the rated
output of the WT.
Compared under the overspeed mode, the underspeed control and
pitch angle control have less pressure on the converter capacity, thus
not discussed. Under the overspeed mode, ω is higher than ωop but
within the upper limit ωmax. Its output PWT is within the rated capacity
PWT,N. Considering active power loss of the DFIG (PDFIG,loss), PWT,N is
higher than the rated output of the DFIG (PDFIG,N).
To maintain the power balance after the disturbance, ω is reduced to
increase PWT to PWT,op at most (23), and reduce the kinetic energy from
EWT at ω to EWT,op at ωop (24) [7-17],

1
ΔPWT = PWT,op − PWT = ρAυw3 (Cp (λ op , β ) − Cp (λ, β ))
2 (23) Fig. 2. Including the LAPL in the control to DFIG.

4
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

PDFIG,N υw1 A and the GSC (QDFIG,set, Qg,set), and SRSC,N, (28) is given in the matrix
PWT,N D B form and extended to,
1
υw,cr C Δx
ΔF + [ JF , x JF , u ] ⎡ ⎤ = 0
0.8 E ⎣ Δu ⎦ (35)
υw2 F
where ΔF denotes the mismatches. The derivatives of ΔF to u, i.e. JFu,
PWT (p.u.)

0.6 are given by,


PWT,set G
Power reserve ∂ΔQs
H
0.4 with overspeed control ∂Vs
υw3 Xs cos θs,m + Rs sin θs,m 2X V
I = Vm − 2 s s 2 + Vg
0.2 Rs2 + Xs2 Rs + Xs
Xg cos θs,g + Rg sin θs,g 2Xg Vs
− 2
0 Rg2 + Xg2 Rg + Xg2 (36)
10 15 20 25 30
∂ΔQs
ω (rpm) = −1
∂QDFIG,set (37)
Fig. 3. Power reserve of WT under overspeed mode.
∂ΔPm Rs cos θm,s − Xs sin θm,s
= Vm
vulnerable to overloading with low PDFIG,set. To avoid it, ΔPs = 0 is ∂Vs Rs2 + Xs2 (38)
replaced by the RSC constraint (27), where SRSC,N is the rated value of Xs cos θm,s + Rs sin θm,s
∂ΔQm
SRSC. = Vm
∂Vs Rs2 + Xs2 (39)
2 2
ΔSRSC = SRSC − SRSC,N = Pr,m + Qr,m − SRSC,N
∂ΔPg Rg cos θg,s − Xg sin θg,s
= Vg
∂Vs Rg2 + Xg2 (40)
Vr4 − 2sVm Vr3 cos θr,m + s 2Vm2 Vr2
= − SRSC,N
Rr2 + s 2Xr2 (27) ∂ΔQg 2Xg Vs Xg cos θs,g + Rg sin θs,g
= − Vg
Finally, the effective constraints of the DFIG under the power re- ∂Vs Rg2 + Xg2 Rg2 + Xg2 (41)
serve and with RSC constraint are given by (28), where the derivatives
∂ΔQg
of SRSC to the unknown × are given in (29)-(34). =1
∂Qg,set (42)
⎡ ΔSRSC ⎤ ⎡ Δθm ⎤
⎢ ΔQs ⎥ ⎢ ΔVm ⎥ ⎡ 0 ⎤ ∂ΔT Rs cos θm,s − Xs sin θm,s
⎢ ΔPm ⎥ ⎢ Δθ ⎥ ⎢ 0 ⎥ = −Vm
∂Vs Rs2 + Xs2 (43)
⎢ ⎥ ⎢ r ⎥ ⎢0⎥
⎢ ΔQ m ⎥ ⎢ ΔVr ⎥ ⎢ 0 ⎥
⎢ ΔPg ⎥ + JFx ⎢ Δθ ⎥ = ⎢ 0 ⎥ Define y = PDFIG,min, which is linearized as,
g
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ ΔQg ⎥ ⎢ ΔVg ⎥ ⎢ 0 ⎥ Δx
Δy = [ Jy, x Jy, u ] ⎡ ⎤
⎢ ΔT ⎥ ⎢ Δs ⎥ ⎢ 0 ⎥ ⎣ Δu ⎦ (44)

const(λ )
⎥ ⎢ ⎥ ⎢0⎥
⎢ i ⎥ ⎢ Δλi ⎥ ⎣ 0 ⎦ With the convergence of (28), ΔF = 0. Combining (35) and (44), the
⎣ const(λ ) ⎥
⎢ ⎦ ⎣ Δλ ⎦ (28) sensitivity ξyu between y and u is derived,

∂SRSC sVm V 3 sin θr,m Δy = (Jy, u − Jy, x Jx , x −1Jx , u )Δu = ξ yu Δu (45)


= − 2 r2 2
∂θm (Rr + s Xr ) SRSC (29)

∂SRSC sVm − Vr cos θr,m 3.4. Evaluation to the reserved power and reserved kinetic energy with the
= sVr2 2 LAPL
∂Vm (Rr + s 2Xr2 ) SRSC (30)

∂SRSC sVm Vr3 sin θr,m A DFIG operating at the LAPL has the reserved power and the re-
=
∂θr (Rr2 + s 2Xr2 ) SRSC (31) served kinetic energy. The reserved power may be released for power
dispatch and primary frequency control. The reserved kinetic energy
∂SRSC 2V 2 − 3sVm Vr cos θr,m + s 2Vm2 may be released to improve the inertia during the frequency drop.
= Vr r
∂Vr (Rr2 + s 2Xr2 ) SRSC (32) Four indices are newly defined to describe the benefit of the reserve.
The 1st index is the ratio of the reserved power with respect to the
∂SRSC Vr2 [sRr2 Vm2 − sXr2 Vr2 − (Rr2 − s 2Xr2 ) Vr Vm cos θr,m ] maximum output of the DFIG (RRP),
=
∂s (Rr2 + s 2Xr2 )2SRSC (33) PDFIG,max − PDFIG,min
RRP =
∂ΔSRSC PDFIG,max (46)
= −1
∂SRSC,N (34) The 2nd index is the Ratio of the reserved kinetic energy with re-
With the solution, PDFIG,min under the overspeed mode and the ca- spect to the kinetic energy at PDFIG,max (RRE),
pacity constraint of the RSC is calculated with (20). EDFIG |P DFIG,min − EDFIG |P DFIG,max
RRE =
EDFIG |P DFIG,max (47)
3.3. Sensitivity analysis to the LAPL of DFIG
The lower active power limit of DFIG may be further reduced with
The sensitivity analysis helps to quantify the factors influencing the the RSC of larger capacity, however with more investment to the con-
LAPL, and find measures to further reduce the LAPL. By setting the verters. Hence there will be a balance between the power dispatch-
variable parameters u, e.g. Vs, the reactive power setting to the DFIG ability and the RSC's investment. To quantify the balance, the ratios of

5
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

the reserved power and the reserved kinetic energy with respect to the the windings of the electrical machines (SGs, DFIGs, and PMSGs)
rated capacity of the RSC, i.e. RRPC and RREC, are newly defined, are rarely included in the OPF model of the power systems.
PDFIG,max − PDFIG,min
RRPC =
SRSC,N (48)
5. Numerical analysis
EDFIG |P DFIG,min − EDFIG |P DFIG,max
RREC = The proposed model is implemented with MATLAB R2007a lan-
SRSC,N (49)
guage compiled by the authors, and run on the computer with the
processor of Intel Core i7-4770 CPU 3.4 GHz and the memory of 4 G.
4. The OPF with dispatchable DFIGs The related parameters are given by ρ = 1.225 kg/m3, D = 71 m,
c8 = –0.02, c9 = –0.003, η = 94, p = 2, PDFIG,N = 2 MW. From the
The LAPL of the DFIG is valuable for the downward dispatch of the power curve of the WT it is found that υw,cr = 12.44 m/s, i.e. the
power systems. To show its impact, it is newly included in the OPF overspeed mode is valid with υw, < υw,cr. Referred to PDFIG,N,
model with the least generation cost of the SGs and the DFIG-based Rs = 0.0078p.u., Xs = 0.0794p.u., Rr = 0.025p.u., Xr = 0.4 p.u.,
wind farms, Xm = 4.1039p.u., Rg = 0.03p.u., Xg = 0.05p.u.,
SRSC,N = SGSC,N = 0.35p.u., QDFIG,set = 0.2 p.u., Qg,set = 0.1 p.u., and
min Cost = ∑ Cost(PSG,i) + ∑ Cost(PDFIG,i)
i ∈ SG i ∈ DFIG (50) HDFIG = 5 s. Vs is set at 1.0 p.u.
There is no standard for the convergence criterion when solving the
where × includes the bus voltages, the active and reactive outputs (PSG DFIG or system power flow. Usually 10-5 or 10-6 is accurate enough. By
and QSG) of the SGs, and PDFIG. f denotes the equality constraints of the using the modified flat start, this paper chooses more rigid criterion 10-
bus power balance. g denotes the inequality constraints to the power 12
with only 4–7 iterations.
loadings of the lines and the transformers. The upper and lower limits In the following, Sections 5.1, 5.2, and 5.3 studies the LAPL in per
are set for V, PG, QG, and PDFIG, where PDFIG is constrained by one of the unit of the DFIG with the base power of the DFIG. In Section 5.4, the
following, LAPL of the DFIG is included in the OPF analysis of the RTS test system,
PDFIG = PDFIG,max , and the base power is 100 MVA.

0 ⩽ PDFIG ⩽ PDFIG,max , or

⎩ PDFIG,min ⩽ PDFIG ⩽ PDFIG,max (51)
where the 1st row shows the DFIG operating at the MPPT mode hence υW Power curve
not dispatchable, the 2nd row shows DFIG has the LAPL of zero, i.e.
completely dispatchable, and the 3rd row is the based on LAPL model Solve DFIG under MPPT,
proposed in this paper. ωop
find PDFIG,max (≤ PDFIG,N)
As to the originality, (1)-(18) are the basic definitions to the DFIG,
which are easily got from the equivalent circuit as seen in many papers,
so the references are not given. (19) and (21) solve the DFIG under the Set PDFIG,set=PDFIG,max)
MPPT and the reserve modes, proposed by 1st author in [35]. (22)
quantifies the power loss of the DFIG. (23)-(26) are quite simple, but Regulate
Solve DFIG
newly explain how the rotor capacity limits the power reserve of the ω
DFIG under the overspeed mode. (27)-(51) give the original idea of this
no
paper. Reduce PDFIG,set Converge?
Fig. 4 shows the solution procedure of this paper. For a given υW, the yes
DFIG under the MPPT mode is solved to find PDFIG,op within the rated PDFIG,min=
find s, SRSC
speed of the WT and the rated capacities of the WT and DFIG. If the RSC previous PDFIG,set
is not overloaded, the power reserve with the setting PDFIG,set is effec-
tive, but it may be higher than the lower limit PDFIG,min. To quantify
no
RSC overload?
PDFIG,min, PDFIG,set is reduced again until the solution diverges or the
RSC is overloaded. For the latter, the capacity constraint of RSC is yes
added to (28) to find PDFIG,min. Then some indices are defined to eval- Regulate Solve DFIG with SRSC,N,
uate the benefit of the LAPL. Finally, the LAPL is included in the OPF to ω find PDFIG and PDFIG,loss
show its contribution to the SGs and the power system.
Some comments are given to the proposed model,
PDFIG,min=PDFIG
(i) In actual operations, some DFIGs may be dispatched to shut down.
So the LAPL of the wind farm is given by the discrete number of the
DFIGs and the continuous range of each DFIG, which may be Sensitivity of PDFIG,min, PDFIG,loss
quantified with the proposed model, and not discussed here. to Vs, QDFIG,set, Qg,set, SRSC,N
(ii) To show the impact of the LAPL, QDFIG is assumed fixed here. But if
necessary, the operational range of QDFIG may be easily included
[37]. Evaluation indices to PDFIG,min
(iii) It is theoretically possible to find the LAPL of the DFIG with the
optimization method, but due to the extreme nonlinearity of the Cp
OPF analysis with PDFIG,min
function, the convergence is very poor. Often no feasible solution is
found. Even the optimization converges, it gives only one solution.
The proposed model also quantifies the sensitivity of the lower End
limit to the operational parameters.
(iv) Considering the problem size and the convergence, constraints of Fig. 4. Solution procedure of this paper.

6
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

5.1. Solving the LAPL of DFIG Table 1


Solution to DFIG with RSC constraint.
Different from fault ride-through with the time of seconds, the Parameter Case a Case b Case c
power reserve of the DFIG is for long time operation. The steady-state MPPT mode PDFIG,set = 0.6p.u. SRSC,N = 0.35p.u.
loading limit of the RSC is strictly satisfied, and overloading is not al-
s −0.3361 −0.6291 −0.5512
lowed.
Pwt (p.u.) 0.7120 0.6096 0.6554
With υw = 11 m/s, the DFIG under power reserve is solved. The PDFIG (p.u.) 0.6983 0.6 0.6446
voltages, the slip, the active power loss, and the converter loadings are SRSC (p.u.) 0.2474 0.3723 0.35
shown in Fig. 5, and compared with those under the MPPT mode SGSC (p.u.) 0.1972 0.2509 0.2477
(shown with the blue circle, PDFIG,max = 0.6983p.u.). With smaller
PDFIG,set, Vm and PDFIG,loss decrease, but ω, Vg, Vr , SRSC, and SGSC in-
crease. If PDFIG is less than PDFIG,min, the rotor voltage, the rotor slip, Under the MPPT mode, the rotor speed limit of the DFIG is decided
and the loading of the RSC will too large to be acceptable. It is also by the capacity of the converter circuit only. With the capacity of the
found that although with the same capacity, the GSC is less easily RSC of 40% of that of the DFIG, the rotor speed range 0.6–1.4p.u. is safe
overloaded, since its reactive power is less than that of the RSC. for all wind speeds. In Fig. 6 (b), with the wind speed of 10 m/s, the
To find PDFIG,min directly instead of the repeated calculation, SRSC,N maximum output is a little more than 50% of the rated value, hence the
is set as constraint and the DFIG is solved (case c). The solution is rotor speed may be set much higher than 1.4p.u. for more kinetic en-
compared with those under the MPPT mode (case a) and the fixed ergy while not violating the capacity of the RSC. For the DFIG operating
power setting (case b). Due to the extreme nonlinearity of the Cp at the overspeed mode for power reserve, if the wind speed is low, the
function, the convergence is more dependent on the initial value. The captured power is small. A high rotor speed with more kinetic energy
solution may converge slowly or diverge when the initial value is far will not cause overloading of the rotor. But for a high wind speed, the
from the exact solution. Hence to test the robustness of the proposed power is large, hence the overspeed is limited to a smaller range. That is
model, set PDFIG,set = 0.6p.u., which is far from PDFIG,min. The results why the rotor speed is decreasing for increasing wind speed for
are given in Table 1. In case b, SRSC > SRSC,N, thus PDFIG,set should be PDFIG,min.
increased. In case c, PDFIG,min = 0.6446p.u. It is clear that the LAPL of As shown in Fig. 6 (a) and (c), with lower wind speed, the difference
the DFIG is limited by its RSC capacity. of PDFIG,min and PDFIG,max is larger than that with higher wind speed.
With υw < 10 m/s, PDFIG is too low to violate SRSC,N. With υw close Similarly, with lower wind speed, the difference of EDFIG,min and
to or higher than υw,cr, there is little or no power reserve available with EDFIG,max is larger than that with the higher wind speed. Hence the DFIG
the overspeed mode. Hence the dispatchable range of the DFIG is stu- with the lower wind speed has wider range to adjust its active power
died, as shown in Fig. 6. With higher wind speed, PDFIG,min approaches and release its kinetic energy. The active and reactive power losses of
PDFIG,max very quickly, hence the power reserve is smaller. Hence for the DFIG under the overspeed mode are less than those under the MPPT
high wind speed, the overspeed control with negligible reserve is mode, increasing with υw.
technically possible but not worthwhile, hence the results of wind speed To avoid instability of the DFIG under the overspeed mode, the
between 11.2 and υw,cr are not shown in the figures. highest rotor speed in Fig. 6 (b) at υw = 10 m/s is applied to quantify

Fig. 5. DFIG parameters with different settings to PDFIG.

7
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

Fig. 6. DFIG parameters within the dispatchable range.

the eigenvalues, as shown in Table 2. The mode 19, 20 deciding the sensitivity model as shown in Fig. 7, where CC and SM denote the
stability margin of 0.0503p.u. shows that the DFIG is stable. The tor-
sional mode 15,16 has the lowest damping ratio of 0.0678, which is Table 2
larger than the acceptable criterion (usually 0.03–0.05). Small-signal stability of DFIG under overspeed mode.
No. Eigenvalue (p.u.) Frequency (Hz) Damping ratio
5.2. Sensitivity analysis to the LAPL
1 −9613.4 1
For cases a to c, the sensitivities of the solution to the variable 2 −4406.6 1
3,4 −2.28 ± 310.9i 49.4822 0.0073
parameters are given in Table 3, where PDFIG,min is found from Case c.
5 −179.14 1
As seen in the 10th row, the sensitivity of PDFIG,min to Vs is 0.2382. It 6 −179.23 1
sensitivity to QDFIG,set is 0.7991. Its sensitivity to Qg,set is −0.8053. Its 7 −330.28 1
sensitivity to SRSC,N is −1.4817. By comparing the absolute value of the 8 −314.11 1
9 −217.25 1
sensitivities it is fund that the sensitivity to Vs is much less, hence
10 −71.98 1
PDFIG,min is more sensitive to QDFIG,set and Qg,set, and less sensitive to Vs 11 −108.97 1
and the system conditions, i.e. the LAPL is relatively fixed with the 12,13 −0.496 ± 3.92 0.6245 0.1254
system condition and suitable for system dispatch. It is also found that 14 −64.1349 1
to reduce the LAPL while not increasing the RSC capacity, QDFIG,set , or 15,16 −0.786 ± 11.57i 1.8413 0.0678
17 −19.6307 1
more exactly, the reactive power setting to the RSC, should be reduced.
18 −19.7323 1
To validate the accuracy of the sensitivities, SRSC,N in case c is 19,20 −0.0503 ± 0.65i 0.1036 0.0771
changed within ± 0.1p.u. The results are compared with those with the

8
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

Table 3 at bus 21 is replaced by a wind farm of 200 2-MW DFIGs. The lower
Sensitivity of DFIG’s solution to parameter setting. limits of other SGs are set by 0.3 PSG,N and −0.2 QSG,N respectively. The
Sensitivity (p.u.) active and the reactive loads are reduced by 50% and 10% respectively
from the peak load for light loading condition.
Case a Vs QDFIG,set Qg,set At first, the cost of the SGs and the DFIGs are given by,
2
PDFIG,max 0.0187 −0.0189 0.0130 Cost(PSG) = 0.2 + 0.7PSG + 0.01PSG (10 4$ p. u. ) (52)
SRSC 0.0691 0.3162 −0.3162
SGSC 0.0085 −0.0152 0.5372 Cost(PDFIG) = 0.1 + 0.8PDFIG (10 4$ p. u. ) (53)
Case b Vs QDFIG,set Qg,set PDFIG,set
SRSC 0.1992 0.6452 −0.6474 −0.3721 The DFIG output is shown in Fig. 10(a). In case d, the DFIG is under
SGSC 0.0016 −0.0158 0.4308 0.0208 the MPPT mode, PDFIG = PDFIG,max, so 4 SGs (bus 13, 18, 22, 23) reach
Case c Vs QDFIG,set Qg,set SRSC,N
their LAPLs respectively (shown with the red circle). In case e, PDFIG is
PDFIG,min 0.2382 0.7991 −0.8053 −1.4817
SGSC −0.0402 −0.1756 0.5955 0.2886 dispatchable between 0 and PDFIG,max, hence only one SG (bus 23)
reaches the LAPL (blue circle), which is too optimistic. In case f, PDFIG is
dispatchable between PDFIG,min and PDFIG,max. With υw ≤ 10.5 m/s, only
continuous calculation and the sensitivity model. For PDFIG,min, the SG at bus 23 reaches the limit. With υw between 10.6 and 10.7 m/s,
PDFIG,loss, QDFIG,loss, and SGSC, the results from the sensitivity model are two more SGs at 18 and 22 reach the limit (shown with the magenta
tangent to those from the continuous calculation at the current opera- circle). With higher υw, one more SG at bus 13 reaches the limit. Hence
tion point shown with the circle. The relative errors are quite small. the LAPL of the DFIGs helps to the SGs not often operating at their
Hence the proposed LAPL and its sensitivity model is correct. LAPLs.
The total system costs of three cases are shown in Fig. 10(b). With
5.3. Evaluation to the LAPL υw close to υw,cr, the dispatchable range of PDFIG is much narrower, and
the system cost of cases d and f are closer. The costs of the SGs and the
The indices to evaluate the LAPL, i.e. RRP, RRE, RRPC, and RREC, DFIG of case f are shown in Fig. 10(c). Since the wind power is not free,
are shown in Fig. 8. With the lower υw, the DFIG has more reserve its MPPT mode yields more wind power and more cost.
power and kinetic energy reserved. Hence, increasing the capacity of It is found for case f, PDFIG increases from 0.5p.u. to about 2.5p.u. As
the RSC to reduce PDFIG,min is more effective for low υw, which however shown in Fig. 6 (a), with higher wind speed, the dispatchable range is
needs more investment and is impractical for the installed DFIG. So the narrower. So with the wind speed increase of 1.2 m/s, the dispatched
lower active power limit may be reduced by adjusting the reactive output of the DFIG changes about 2p.u. The concept is different from
power. that under the MPPT mode.
Then the ramping cost of the SGs is added to the total generation
5.4. Economic dispatch considering the LAPL of the DFIG cost of the system,
(0) 2
RampCost(PSG) = 2×(PSG − PSG ) (54)
To verify the impact of LAPL of the DFIG on power dispatch, the
(0)
RTS test system is applied (Fig. 9) [38]. The nuclear unit rated 400 MW where the PSG is the active output of the SG before dispatch.

Fig. 7. Validation of the sensitivity model with respect to SRSC,N.

9
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

Fig. 8. Benefit of the LAPL of DFIG.

6. Conclusions
18 21 22

17 This paper newly quantifies the lower active power limit (LAPL) of
23 the DFIG under overspeed mode, which will enhance the downward
dispatch capability of DFIG and the power system. It is a fundamental
19 20
16 problem to the wind power systems, but has not been studied in the
existing literatures.
15 1 13 The analytical derivation and numerical simulation show that the
4 active power setting and the RSC capacity constraint may not be sa-
tisfied simultaneously for DFIGs under the overspeed mode. Under the
24 11 12 overspeed mode, the DFIG operating at the LAPL has more kinetic en-
230 kV
ergy. It is also found that the LAPL of the DFIG is not sensitive to the
stator voltage, hence suitable to be incorporated in the OPF model of
138 kV
9 10 6 the power systems. With the downward dispatch of DFIG, the SGs will
3
be less often operating at the lower limit during the light loading, and
4
the ramping cost of the SGs will be reduced.
5 8 Author Statement
This paper is the original work of the authors.
It does not copy or unsuitably cite any exiting literatures. It is not
1 2 7
published or under review by any other journal or any conference.

Fig. 9. IEEE RTS test system.


Declaration of Competing Interest
With the DFIG operating at the MPPT mode or power dispatch based
on the LAPL, the costs of the SGs and the wind farm, and the total The authors declare that they have no known competing financial
system cost are given in Table 4. interests or personal relationships that could have appeared to influ-
With the higher wind speed, the dispatchable range of the DFIG ence the work reported in this paper.
reduces, hence the ramping cost of the SGs and the total cost of the
system increase. Compared with that with the DFIG under the MPPT
mode, the system cost with the dispatchable DFIG is reduced. The re- Acknowledgment
duction is smaller with the higher wind speed.
This work was supported by National Natural Science Foundation of
China (NSFC) under Grant 51877061.

10
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

Fig.10. Dispatch results within the adjustable range of the DFIG.

Table 4 regulation in electrical power systems. Energy Procedia 2017;142:2183–8.


Reduction of generation cost with downward dispatch of DFIG. [9] Xue Y, Tai N. System frequency regulation in doubly fed induction generators. Int. J.
Electr. Power Energy Syst. 2012;43(1):977–83.
υw (m/s) Mode of Generation cost (104 $/p.u.) Reduction of the [10] Bhatt P, Roy R, Ghoshal SP. Dynamic participation of doubly fed induction gen-
DFIG system cost (%) erator in automatic generation control. Renew. Energy 2011;36(4):1203–13.
SG DFIG Total [11] Amadou MMD, Mehrjerdi H, Saad M, Asber D. Improving participation of doubly
fed induction generator in frequency regulation in an isolated power system. Int. J.
10.0 MPPT 44.25 1.78 46.03 / Electr. Power Energy Syst. 2018;100:550–8.
Dispatch 37.91 0.56 38.47 16.43 [12] Attya AB, Ademi S, Jovanović M, Anaya-Lara O. Frequency support using doubly
10.2 MPPT 44.86 1.88 46.74 / fed induction and reluctance wind turbine generators. Int. J. Electr. Power Energy
Syst. 2018;101:403–14.
Dispatch 39.11 0.81 39.92 14.60
[13] Kayikci M, Milanovic JV. Dynamic contribution of DFIG-based wind plants to
10.4 MPPT 45.51 1.99 47.50 /
system frequency disturbances. IEEE Trans. Power Syst. 2009;24(2):859–67.
Dispatch 40.80 1.15 41.95 11.67
[14] Liao K, Xu Y, Wang Y, Lin P. Hybrid control of DFIGs for short-term and long-term
10.6 MPPT 46.19 2.10 48.29 / frequency regulation support in power systems. IET Renew. Power Gener.
Dispatch 42.95 1.55 44.50 7.86 2019;13(8):1271–9.
10.8 MPPT 46.95 2.22 49.17 / [15] Zhang X, Wang Y, Fu Y, Xu L. A novel method for obtaining virtual inertial response
Dispatch 44.95 1.90 46.84 4.73 of DFIG-based wind turbines. Wind Energy 2016;19(2):313–28.
11.0 MPPT 47.72 2.33 50.05 / [16] Fu Y, Zhang X, Hei Y, Wang H. Active participation of variable speed wind turbine
Dispatch 46.62 2.16 48.78 2.53 in inertial and primary frequency regulations. Electr. Power Syst. Res.
11.2 MPPT 48.53 2.46 50.98 / 2017;147:174–84.
Dispatch 47.99 2.38 50.37 1.21 [17] Chang-Chien LR, Yin YC. Strategies for operating wind power in a similar manner of
conventional power plant. IEEE Trans. Energy Convers. 2009;24(4):926–34.
[18] Hansen AD, Sørensen P, Iov F, Blaabjerg F. Centralised power control of wind farm
with doubly fed induction generators. Renew. Energy 2006;31(7):935–51.
References [19] Nock D, Krishnan V, McCalley DJ. Dispatching intermittent wind resources for
ancillary services via wind control and its impact on power system economics.
Renew. Energy 2014;71:396–400.
[1] Saiz-Marin E, Garcia-Gonzalez J, Barquin J, Lobato E. Economic assessment of the
[20] Dovorkin Y, Ortega-Vazquez MA, Kirschen SD. Wind generation as a reserve pro-
participation of wind generation in the secondary regulation market. IEEE Trans.
vider. IET Gener. Transm. Distrib. 2015;9(8):779–87.
Power Syst. 2012;27(2):866–76.
[21] Durreachter HL, Looney SK. Integration of wind generation into the ERCOT market.
[2] Cardell JB, Anderson CL. A flexible dispatch margin for wind integration. IEEE
IEEE Trans. Sustain. Energy 2012;4(3):862–7.
Trans. Power Syst. 2015;30(3):1501–10.
[22] Mondal S, Kastha D. Maximum active and reactive power capability of a matrix
[3] Hedayati-Mehdiabadi M, Zhang J, Hedman KW. Wind power dispatch margin for
converter-fed DFIG-based wind energy conversion system, IEEE. J. Emerging
flexible energy and reserve scheduling with increased wind generation. IEEE Trans.
Selected Topics Power Electron. 2017;5(3):1322–33.
Sustain. Energy 2015;6(4):1543–52.
[23] Guo X, Zou X, Jiang C, Zhu D, Yang Y, Li P, et al. Active power limit for DFIG-based
[4] Jabr RA. Adjustable robust OPF with renewable energy sources. IEEE Trans. Power
wind turbine under weak grid. IEEE Energy Convers. Congress Exposition (ECCE)
Sys. 2013;28(4):4742–51.
2019. https://doi.org/10.1109/ecce.2019.8912202.
[5] Das K, Litong-Palima M, Maule P, Altin M, Hansen AD, Sørensen PE, et al. Adequacy
[24] Feijóo AE, Cidrás J. Modeling of wind farms in the load flow analysis. IEEE Trans.
of frequency reserves for high wind power generation. IET Renew. Power Gener.
Power Syst. 2000;15(1):110–5.
2017;11(8):1286–94.
[25] Ruiz-Vega D, Olivares TIA, Salinas DO. An approach to the initialization of dynamic
[6] Vos KD, Morbee J, Driesen J, Belmans R. Impact of wind power on sizing and al-
induction motor models. IEEE Trans. Power Syst. 2002;17(3):747–51.
location of reserve requirements. IET Renew. Power Gener. 2013;7(1):1–9.
[26] Divya KC, Rao PSN. Models for wind turbine generating systems and their appli-
[7] Díaz G. Optimal primary reserve in DFIGs for frequency support. Int. J. Electr.
cation in load flow studies. Electr. Power Syst. Res. 2006;76(9–10):844–56.
Power Energy Sys. 2012;43(1):1193–5.
[27] Eminoglu U, Dursun B, Hocaoglu MH. Incorporation of a new wind turbine gen-
[8] Bhatt P, Long C, Wu J, Mehta B. Dynamic participation of DFIG for frequency
erating system model into distribution systems load flow analysis. Wind Energy

11
S. Li and M. Wang Electrical Power and Energy Systems 125 (2021) 106439

2009;12(4):375–90. 2011;128(5):174–80.
[28] Zhao M, Chen Z, Blaabjerg F. Load flow analysis for variable speed offshore wind [34] Aguglia D, Viarouge P, Wamkeue R, Cros J. Analytical determination of steady-state
farms. IET Renew. Power Gen. 2009;3(2):120–32. converter control laws for wind turbines equipped with doubly fed induction gen-
[29] Ekanayake JB, Holdsworth L, Wu XG, Jenkins N. Dynamic modeling of doubly fed erators. IET Renew. Power Gener. 2008;2(1):16–25.
induction generator wind turbines. IEEE Trans. Power Syst. 2003;18(2):803–9. [35] Li S. Power flow modeling to doubly-fed induction generators (DFIGs) under power
[30] Holdsworth L, Wu XG, Ekanayake JB, Jenkins N. Direct solution method for in- regulation. IEEE Trans. Power Sys. 2013;28(3):3292–301.
duction wind turbines in models initialising doubly-fed power system dynamic. IEE [36] Li S. Low-frequency oscillations of wind power systems caused by doubly-fed in-
Proc. Gener., Transm. & Distrib. 2003;150(3):334–42. duction generators. Renew. Energy 2017;104:129–38.
[31] Padrón JFM, Lorenzo AEF. Calculating steady-state operating conditions for doubly- [37] Li S. Operation region of doubly-fed induction generators based on rotor slip under
fed induction generator wind turbines. IEEE Trans. Power Syst. 2010;25(2):922–8. MPPT control and power dispatch. Electr. Power Compon. Syst.
[32] Senjyu T, Sakamoto R, Urasaki N, Funabashi T, Fujita H, Sekine H. Output power 2014;42(8):808–17.
leveling of wind turbine generator for all operating regions by pitch angle control. [38] Reliability Test System Task Force of the IEEE Subcommittee on the Application of
IEEE Trans. Energy Convers. 2006;21(2):467–75. Probability Methods, IEEE reliability test system, IEEE Trans. Power App. Syst.
[33] Ofner G, Handgruber P, Seebacher R. Doubly fed induction generators for wind 98(6) (1979) 2047-2054.
turbines-modeling the steady state. Elektrotechnik & Informationstechnik

12

You might also like