You are on page 1of 12

2094 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 32, NO.

3, MAY 2017

Research on the Impact of DFIG Virtual Inertia


Control on Power System Small-Signal Stability
Considering the Phase-Locked Loop
Jing Ma, Member, IEEE, Yang Qiu, Yinan Li, Weibo Zhang, Zhanxiang Song, and James S. Thorp, Life Fellow, IEEE

Abstract—d induction generator (DFIG) wind turbines with vir- NOMENCLATURE


tual inertia control are coupled to power system in dynamic char-
acteristics, and the control input of virtual inertia control is directly xPLL , δPLL State variables of PLL.
affected by the tracking ability of phase-locked loop (PLL). Thus, KP PLL , KI PLL PI controller parameters of PLL.
it is urgent to study the impact of DFIG wind turbines with virtual
inertia control on power system small-signal stability considering
xω State variable of virtual inertia control.
the effects of PLL. First, based on DFIG operation characteristic Kω Virtual inertia control gain.
and control strategy, a small-signal model of interconnected system δ Power angle of synchronous generator.
with DFIG integration considering PLL and virtual inertial control ω Angular speed of synchronous generator.
is established. Second, the attenuation time constants of DFIG state Ed ,Eq Two-axis transient emf.
variables are calculated, and according to the attenuation speeds of ω s , ωr Stator and rotor speed.
different state variables and the coupling between them, it is found  
Efd , Efq Intermediate variables of DFIG rotor
out that PLL and virtual inertia are the main factors that affect voltage.
the coupling between DFIG and synchronous generators. And
then, considering that both PLL and virtual inertia control will x1 , x2 , x3 , x4 Intermediate variables of rotor converter.
affect the oscillation modes of synchronous generators, analytical usd , usq , urd , urq Two-axis voltages.
method is used to reveal system small-signal stability under the isd , isq , ird , irq Two-axis currents.
joint effects of the two factors quantitatively. Analysis results show ψsd , ψsq , ψrd , ψrq Two-axis flux linkages.
that, for DFIG wind turbines with virtual inertial control, PLL Ls , Lr Stator and rotor inductances.
affects system damping mainly by affecting the participation of Lm Mutual inductances.
virtual inertia in the system. The smaller the PI parameters of PLL s Rotor slip.
are, the smaller the participation factor of virtual inertia control
β Wind turbine pitch angle.
state variables in the interarea oscillation mode is, and the bigger
the electromechanical oscillation mode damping ratio is. Simu- H Inertia constant.
lation results verify the reasonableness of the established model SUBSCRIPTS
and the possibility that virtual inertia control may cause system d, q Synchronously rotating d, q axis.
small-signal stability to deteriorate in multimachine system. s, r Stator, rotor.
Index Terms—DFIG wind turbines, virtual inertia, phase-locked
SG, G Synchronous generator.
loop, coupling characteristics, small-signal stability, damping 0 Initial value.
characteristics. I. INTRODUCTION
IND power generation has become an ever bigger pro-
Manuscript received January 31, 2016; revised May 5, 2016 and July 17,
2016; accepted July 23, 2016. Date of publication July 29, 2016; date of current
version April 17, 2017. This work was supported in part by the National Natural
W portion in power system due to its cleanness and wide
source. The impact of wind generation on the safe and stable
Science Foundation of China under Grant 51277193, in part by the National operation of power system is gaining more and more attention,
Basic Research Program of China (973 program) under Grant 2012CB215206, especially its impact on power system small-signal stability [1]–
and in part by the Chinese Universities Scientific Fund under Grant 2014ZZD02.
Paper no. TPWRS-00167-2016. [2].
J. Ma is with the State Key Laboratory of Alternate Electrical Power Sys- Doubly-fed induction generator (DFIG) is the most widely
tem with Renewable Energy Sources, North China Electric Power University, used wind turbine type due to its excellent operation charac-
Beijing 102206, China, and also with the Bradley Department of Electrical teristics [3]–[4]. In recent years, many scholars have conducted
and Computer Engineering, Virginia Polytechnic Institute and State University,
Blacksburg, VA 24061 USA (e-mail: hdmajing@163.com). relatively systematic research on the small-signal stability of
Y. Qiu, Y. Li, W. Zhang, and Z. Song are with the State Key Labora- power system with large scale wind power integration. Refer-
tory of Alternate Electrical Power System with Renewable Energy Sources, ence [5] discusses how different integration locations of wind
North China Electric Power University, Beijing 102206, China (e-mail: farm affect the system oscillation modes, improve or deprave.
qiuyang1011@163.com; liyinanmuzi@sina.com; 1152201153@ncepu.edu.cn;
songzhanxiang887@163.com). Reference [6] analyzes how the dynamic characteristics of DFIG
J. S. Thorp is with the Bradley Department of Electrical and Computer affect power system small-signal stability, and comes to the con-
Engineering, Virginia Polytechnic Institute and State University, Blacksburg, clusion that the dynamic characteristics of DFIG and power sys-
VA 24061 USA (e-mail: jsthorp@163.com). tem are nearly decoupled. Reference [7] points out that when
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. the wind turbines apply the stator voltage phase-locked loop
Digital Object Identifier 10.1109/TPWRS.2016.2594781 (PLL), improper parameter setting will have adverse effects on
0885-8950 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
MA et al.: RESEARCH ON THE IMPACT OF DFIG VIRTUAL INERTIA CONTROL ON POWER SYSTEM SMALL-SIGNAL STABILITY 2095

the electromechanical oscillation modes of power system. The


above research shows that, with the increase of permeability,
DFIG wind turbines will exert significant influence on power
system small-signal stability.
With the proposal of ‘grid-friendly’ wind farms, many coun-
tries have raised requirements on the frequency of wind farm
integration point in their wind power integration guidelines [8].
Considerable studies have proved that the inertia constants of
wind turbine generators are comparable to synchronous genera-
tors [9]–[10]. Thus, it is necessary to add inertia control in wind
turbines or wind farms to improve the integration performance
Fig. 1. Model of PLL (phase-locked loop).
of wind turbines. In reference [11], there are two methods pre-
sented to implementing virtual inertia and the inertia control
performance under different operating conditions is studied in control of the stator flux [15]. The model of stator voltage PLL
detail using theoretical analysis and simulation. However, the is shown in Fig. 1, the dynamic equations of which are:
dynamic characteristic of DFIG with inertia will be coupled
with that of power system, thus affecting system small-signal ẋPLL = −usd (1)
stability. Reference [12] points out that virtual inertial control
will have adverse effects on the small-signal stability of weakly δ̇PLL = ωs PLL − ωn (2)
interconnected systems. ωs PLL = KI PLL xPLL − KP P L L usd (3)
Existing virtual inertia control uses the system frequency as
control input to regulate the output power of wind turbines [13], where xPLL is the accumulated error in tracking the stator d-
and system frequency is usually measured by the PLL [14]. axis voltage. usd is DFIG stator d-axis voltage. δPLL is the angle
Thus, it is of practical value to analyze the impact of PLL on by which the stator voltage vector is observed ahead of the x-
virtual inertial control, and the small-signal stability of power axis in the xy coordinate system. ωs PLL is the rotating speed of
system under the joint influence of PLL and virtual inertial con- the dq coordinate system measured by PLL. ωn is power grid
trol. Currently little literature has any touch on this topic. In this frequency.
paper, a small-signal model of interconnected system with DFIG
integration considering the PLL and virtual inertial control is es- B. Model of Virtual Inertia Control
tablished first. Second, the attenuation time constants of DFIG Due to control effects of the rotor-side inverter, DFIG could
state variables are calculated, and the coupling between them operate at varying speed in a relatively wide range of wind
is revealed. And then, take the two-machine infinite-bus system speed, thus realizing maximum wind power tracking. However,
for example, the small-signal stability of power system under when the speed of DFIG varies, the rotating speed of DFIG
the joint effects of PLL and virtual inertial control is analyzed will be decoupled from the grid frequency, which means the
using analytical method. According to the analysis results, for mechanical inertia of DFIG wind turbines is not fully utilized
DFIG wind turbines with virtual inertial control, the impact of [16]. In fact, the inertia of DFIG wind turbines is quite consid-
the PLL on system damping characteristics is contrary to that erable. The total mechanical inertia time constant of a typical
in wind turbines without virtual inertial control. For DFIG wind 1.5 MW DFIG is usually 2–6 s [9], [17], i.e. Kω is about 4–12,
turbines with virtual inertial control, the PLL affects system depending on the wind speed, inverter capacity, etc. Most of the
damping mainly by affecting the participation of virtual iner- existing megawatt wind turbines have no inertia response func-
tia in the system. The smaller the PLL PI parameters are, the tion, the wind power inertia model based on frequency feedback
smaller the participation factor of virtual inertia control state proposed by the American General Electric Company (GE) pro-
variables in the inter-area oscillation mode is, and the bigger vides wind turbines with inertia response ability [18].
the electromechanical oscillation mode damping ratio is, which The overall picture of the virtual inertia control, PLL and
is beneficial for system stability. Finally, simulation tests on DFIG rotor side converter is as shown in Fig. A1 in Appendix A.
the IEEE four-generator two-area system verify the reasonable- The typical virtual inertia control is shown in Fig. 2. When
ness of the established model and correctness of the analysis system frequency drops, the active power reference value at the
results. input of the rotor inverter will increase, thus by regulating the
rotor current, the output active power will increase. The rotor
II. DYNAMIC MODELS then could release part of the kinetic energy as a dynamic support
of system frequency variation.
A. Model of PLL In the virtual inertia control, the system frequency is
In stator flux directional control, d-axis of the synchronous introduced and the model can be expressed as:
rotating coordinate system is set to coincide with the stator flux 1
vector. In steady-state operation, neglecting the stator resistance, ẋω = (fm eas − xω ) (4)

the stator voltage vector is 90° forward the stator flux vector in
phase. Due to the complexity of accurate flux observation algo- where xω is the introduced intermediate variable. fm eas is the
rithms, wind turbines commonly used in industry all apply the measured value of system frequency, which can be measured by
stator voltage directional control method, thus realizing indirect PLL, i.e. fm eas = ωs PLL /2π. Tω is the time constant.
2096 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 32, NO. 3, MAY 2017

be obtained by linearization at a system equilibrium point:


⎡ ⎤ ⎡ ⎤
ΔẋSG ΔxSG
⎣ ΔẋDFIG ⎦ = As ⎣ ΔxDFIG ⎦
ΔẋM ECH ΔxM ECH
⎡ ⎤⎡ ⎤
A11 A12 A13 ΔxSG
= ⎣ A21 A22 A23 ⎦ ⎣ ΔxDFIG ⎦ (6)
A31 A32 A33 ΔxM ECH

where
 T
xSG = δ ω E d E q 4(n −1)×1 (7)
  

xDFIG = δPLL xPLL xω Efd Efq x1 x2 x3 x4 T9×1 (8)
Fig. A1. Diagram of DFIG stator voltage vector directional control strategy.
xM ECH = [ ωr β ] T
2×1 (9)

A11 , A22 and A33 characterize the internal dynamic char-


acteristics of synchronous generator, DFIG and the mechani-
cal part of DFIG respectively. The other variables character-
ize the coupling relationships in different dynamic processes.
Consider that the expressions of elements in the characteris-
tic matrices are complex, only the expressions of the main
elements in A11 and A12 are given, and the non-zero ele-
Fig. 2. Typical block diagram of virtual inertia control.
ments in the other characteristic matrices are all represented
by ‘∗’.

Considering the effect of virtual inertia, the reference value


of the input active power of rotor inverter is:

Pref = Popt − PJ = Popt − Kω ẋω (5)

where Popt is the active power reference value at the output of


the maximum power tracking module. PJ is the power output
signal of virtual inertia control. Kω is the proportion coefficient
of the virtual inertia control link used to imitate the virtual
inertia, Kω > 0.

C. Modeling and Analysis of Inter-Connected Power Systems


Suppose the type and operation state of all the wind turbines
in the wind farm are the same, then the wind farm could be
equalized to a large-capacity wind turbine, the parameters of
which could be calculated according to parameters of single
wind turbine [19]. Therefore, the equivalent model of wind farm
could be expressed with the DFIG wind turbine model composed
of (A2)–(A3), (A9)–(A12), (A15) and (A17) in Appendix A, and
the above PLL and virtual inertia models composed of (1)–(2)
and (4).
In this paper, the equivalent wind farm model composed
of multiple DFIG wind turbines is introduced to the multi-
generator power system for small-signal analysis. For a multi-
generator system with n generators, suppose that the first n −1
are synchronous generators which apply the four-order model
[20], and the nth generator is an equivalent DFIG considering
PLL and virtual inertia control. The system state equations can
MA et al.: RESEARCH ON THE IMPACT OF DFIG VIRTUAL INERTIA CONTROL ON POWER SYSTEM SMALL-SIGNAL STABILITY 2097

where
Ξ Ξ
K ℵ = diag(Ix,i ) · cℵ
Ξ Ξ Ξ
+ diag(Ix,i ) · dℵ + diag(Ux,i ) · eℵ + diag(Ux,i ) · f ℵ ,

xq ,i − xq ,i
J ℵ = diag ·[diag(cos δi ) · êℵ +diag(sin δi )· fˆℵ ],
Tq 0,i
 Fig. 3. Two-machine infinite-bus system.
xd,i − xd,i
Qℵ = −diag ˆ
·[diag(sin δi )· êℵ −diag(cos δi )· f ℵ ],

Td0,i TABLE I
TIME CONSTANTS OF DFIG STATE VARIABLES
ℵ corresponds to subscript δ, d and q, and Ξ corresponds to top-
script ∼ and ∧. The meaning of the above variables and other
Number Time constant (s) State variables
elements can be seen in Appendix B.
Existing research [6]–[7] demonstrates that, the matrix of cou- 1 0.2231 δP L L , xP L L
pling between synchronous generator and DFIG without virtual 2 0.9995 xω
3 1.8783 ωr
inertia control is zero matrix, thus the synchronous generator 4 1.3827 β
system is decoupled with the wind generator system, i.e. the 5 0.0600 E f d , E f q
state variables of wind generator do not participate in the elec- 6 0.01511 x1 , x3
tromechanical oscillation between synchronous generators [6]. 7 0.01265 x2
8 0.01332 x4
However, for DFIG with virtual inertia control, according to (11)
and (12), matrix A12 and A21 which characterize the coupling
between synchronous generator and DFIG are both non-zero
matrices, thus DFIG is coupled with synchronous generator in converter are much smaller than that of virtual inertia control,
dynamic characteristics. in tens of milliseconds.
Meanwhile, it can be seen from (11) and (13) that, matrix Keep the other parameters constant, and change the virtual
A13 and A23 which characterize the coupling between the me- inertia gain from 4 to 12, calculate the time constants of state
chanical part of DFIG and electromagnetic part are both zero variables, the results shown in Fig. 4(a). Similarly, change the
matrices. This means that, in the study on small-signal stability, power output of a single wind turbine from 1.5 MW to 1.0 MW
whether rotor speed ωr or pitch angle β changes, the disturbance (the number of wind turbines in parallel operation is n = 100),
will not pass on from the wind turbine to DFIG itself, neither the time constants of state variables are calculated and shown in
will it affect the small-signal stability of synchronous generators Fig. 4(b). It can be seen from Fig. 4 that, although the variation
in the system. Therefore, it can be taken that the mechanical part of virtual inertia gain and wind power output, etc. will affect the
and electromagnetic part of DFIG are approximately decoupled time constants of some state variables, the order of magnitude
with each other on small signal stability. of the time constant of each state variable remains the same, and
As for the other state variables, i.e. the rotor voltage, rotor- the relationship between the attenuation speeds of different state
side converter, PLL and virtual inertia control, the coupling variables also remains unchanged, i.e. the dynamic response of
characteristics of them are analyzed below according to the virtual inertia control and the mechanical part is slightly slower
time constant of oscillation mode of each variable. than that of PLL, and the dynamic response of rotor voltage and
rotor-side converter is much faster than that of virtual inertia
III. TIME CONSTANTS AND COUPLING CHARACTERISTICS control and the mechanical part.
OF DFIG STATE VARIABLES According to the attenuation speeds of the above state vari-
ables, rotor voltage and rotor-side converter have high response
This paper uses the Lyapunov first method to calculate the speeds, thus on the time scale they are approximately decoupled
time constants of state variables of the high-order system [20]. with the other state variables and will not exert big influence on
First, calculate system eigenvlaues and the corresponding par- the small-signal stability of DFIG-synchronous generator inter-
ticipation factors; then derive the relationship between state connected system. Besides, the time constant of virtual inertia
variables and eigenvalues according to the participation factors; is close to that of electromechanical transient process, and the
and then, calculate the time constant of a certain state variable measurement of PLL directly affects the output of virtual inertia
using the eigenvalue corresponding to the state variable. The control, thus PLL and virtual inertia are the main factors that
two-machine infinite-bus system as shown in Fig. 3 is used to affect the coupling between DFIG and synchronous generators.
calculate the time constants of DFIG state variables, the system The effects of the two factors on system small-signal stability
parameters shown in Appendix C. When DFIG operates in fully are analyzed below.
loaded state, the calculated time constant of each state variable
is shown in Table I.
IV. IMPACT OF PLL AND VIRTUAL INERTIA CONTROL
It can be seen from Table I that, the time constant of virtual
ON SYSTEM DAMPING CHARACTERISTICS
inertia control is close to that of the mechanical part of DFIG,
both in seconds. The time constant of PLL is in hundreds of mil- The impact of PLL and virtual inertia control on system
liseconds and the time constants of rotor voltage and rotor-side damping characteristics is studied quantitatively below using
2098 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 32, NO. 3, MAY 2017

In order to obtain the analytical expression, the variation of


DFIG output voltage amplitude is neglected and the small-signal
perturbation of (16) is:
UG U cos (δ0 − θ0 ) E  UG cos θ0
(Δδ − Δθ) = Δθ + ΔPD
z2 z1
(19)
Under conventional control mode, DFIG operates in maxi-
mum power point tracking state, where the rotor speed is de-
coupled with grid frequency, and the active power output does
not vary with small disturbance on the grid side (e.g. load fluc-
tuation). However, for DFIG wind turbines with virtual inertia
control, the active power output will fluctuate when grid fre-
quency varies.
Take the time constant of PLL as infinitesimal, then the vir-
tual inertia controller input frequency is the system frequency.
Meanwhile, consider that in the two-machine infinite-bus sys-
tem in Fig. 3, the angular frequency variation measured at PCC
is approximately equal to the angular frequency variation of
the generator, thus the active power output variation of DFIG
without/with virtual inertia control is:
ΔPD =

0 without virtual inertia control


− K2πω Δω̇c = − K2πω Δδ̈ with virtual inertia control
(20)
where Δωc is the reference value of angular frequency variation
Fig. 4. Variation of time constants of DFIG state variables with virtual in- at PCC, i.e. the angular frequency value measured by PLL under
ertia coefficient and wind power output. (a) Variation of virtual inertia gain. ideal conditions.
(b) Variation of wind power output. Apply (19) and (20) to (15), so that the system equations
of DFIG without/with virtual inertia control can be obtained,
shown in (21) and (22) respectively.
the small-signal analysis method. To facilitate analysis, the two-
machine infinite-bus system shown in Fig. 3 is taken for exam- 2HG Δδ̈ + DΔδ̇ + a0 Δδ = 0 (21)
ple. G1 is synchronous generator, and DFIG is double-fed wind (2HG + Hω ) Δδ̈ + DΔδ̇ + a0 Δδ = 0 (22)
turbine. E  is the q-axis transient emf of G1. UG is the voltage
at the wind farm integration point. U is the voltage at the infinite where
bus. δ is the phase angle difference between E  and U , and θ is Kω a0 z2
the phase angle difference between E  and UG . z1 and z2 are Hω = >0 (23)
2πUG U cos (δ0 − θ0 )
line impedance parameters.
The synchronous generator applies the classical two-order E  UG U cos θ0 cos (δ0 − θ0 )
a0 = (24)
model, which can be expressed as: E  z2 cos θ0 + U z1 cos (δ0 − θ0 )

δ̇ = ω − 1 For the system shown in 2HΔδ̈ + DΔδ̇ + AΔδ = 0, the


(14)
2HG ω̇ = Pm − PG − D (ω − 1) eigenvalues and damping ratio are as follows:

Suppose the mechanical power is constant, linearization of −D ± D2 − 8AH
p1,2 = (25)
(14) yields: 4H

2HG Δδ̈ + DΔδ̇ + ΔPG = 0 (15) D 1
ξ= (26)
2 2A0 H
According to the power balance relationship,
It can be seen from (25) and (26) that, as inertia coefficient H
PΣ = PG + P D (16)
increases, system damping ratio will decrease, which is unfa-
where vorable to system power angle stability. Comparison with (21)
UG U and (22) shows that, when virtual inertia control is applied, since
PΣ = sin (δ − θ) (17) the supplementary inertia coefficient Hω > 0, the system eigen-
z2 values will move to the right and damping ratio will decrease.
E  UG Therefore, compared to when virtual inertia control is not ap-
PG = sin θ (18) plied, system small-signal stability will deteriorate. Meanwhile,
z1
MA et al.: RESEARCH ON THE IMPACT OF DFIG VIRTUAL INERTIA CONTROL ON POWER SYSTEM SMALL-SIGNAL STABILITY 2099

oscillation between synchronous generators. Thus this paper is


based on the hypothesis that the wind speed/wind farm output
power is constant.
Two groups of PI parameters are used in this paper, i.e. typical
values KP PLL = 73.6 and KI PLL = 333.3, and weak damping
values KP PLL = 0.5 and KI PLL = 17. The setting process and
performance test of the typical values can be found in reference
[23]–[24]. According to reference [7], for DFIG wind turbines
without virtual inertia control, when PLL applies typical val-
Fig. 5. Diagram of four-generator two-area system. ues KP PLL = 73.6 and KI PLL = 333.3, the wind turbines do
not participate in system electromechanical oscillation modes.
according to (23), the supplementary inertia coefficient Hω in- When PLL applies weak damping values KP PLL = 0.5 and
creases as DFIG virtual inertia control coefficient Kω increases, KI PLL = 17, the modes dominated by PLL are close to the
thus system damping ratio will decrease as Kω increases, which frequencies of system electromechanical oscillation modes, and
is unfavorable to the suppression of grid power oscillation. the PLL modes will be reflected in the wind power, which will
Considering the influence of PLL on virtual inertia control, further strongly affect the system electromechanical oscillation
the perturbation of DFIG output active power can be expressed modes, causing system stability to deteriorate. While according
as: to the above analysis, for DFIG wind turbines with virtual inertia
control, PLL affects the system damping mainly by affecting the
Kω K
ΔPg = − Δω̇s PLL = − ω Δδ̈ (27) participation degree of virtual inertia in the system oscillation
2π 2π modes.
where Δωs PLL is the value of angular frequency variation at In order to have a more intuitional comparison of the ef-
PCC measured by PLL. fects of PLL on system small-signal stability in DFIG wind tur-
Consider the communication delay in real measuring device, bines with/without virtual inertia control, and reveal the function
the measuring speed of PLL cannot track the variation of system mechanism of PLL in DFIG wind turbines with virtual inertia
frequency, it can be seen that, the angular frequency variation control, simulation verification is conducted below using the
measured by PLL Δω̇s PLL < Δω̇c . Comparing (20) and (27), above two groups of values for PLL PI parameters.
when the system is under the same intensity of small-signal
perturbation, Δδ remains unchanged. Since Kω < Kω , Hω will
decrease, and the system damping ratio will increase, i.e. PLL B. Impact of PLL and Virtual Inertia Control on System
weakens the participation of virtual inertia in system oscillation. Damping Characteristics
Therefore, the smaller the PI parameters of PLL are, the bigger When the PI parameters of PLL apply the above two groups of
the inertia time constant is, the smaller Δω̇s PLL is, and the values respectively, the variation of inter-area oscillation mode
smaller the participation degree of virtual inertia in the system damping ratio with the virtual inertia is analyzed, and the sys-
is, the bigger system damping ratio is. tem inter-area oscillation modes obtained with the modeling
In this section, the expressions of system eigenvalue and method and simulation method are shown in Fig. 6. The mod-
damping ratio under the effect of virtual inertia are derived based eling method obtains system inter-area oscillation modes by
on two-machine infinite-bus system, in an attempt to reveal the solving the state coefficient matrix As of the small-signal sta-
effect of inertia on small-signal stability with analytical method bility system derived above, and the results are shown in the red
(many other literatures also apply this method, see [20]–[21]), dashed line. The simulation method obtains system inter-area
and then to explore the possibility that ‘virtual inertia control oscillation modes by means of Digsilent/Power Factory, and the
may cause system small-signal stability to deteriorate’ in multi- results are shown in the black solid line.
machine system through modeling and simulation analysis in It can be seen that, the results obtained with the modeling
the following section. method and simulation method are close to each other. The
variation of the damping ratio of inter-area oscillation mode
V. SIMULATION ANALYSIS with the virtual inertia is shown in Fig. 7, where the maximum
damping ratio deviation is 4.350%, within acceptable limits,
A. Test System
which further verifies the correctness of the small-signal model
Take the IEEE four-generator two-area system shown in Fig. 5 established in this paper. It can also be seen from Fig. 7 that,
for example [22], the inter-area oscillation modes of the system the variation trend of damping ratio in two cases is the same,
are analyzed. G1, G3 and G4 are synchronous generators with i.e. the damping ratio corresponding to the inter-area oscillation
output power of 700 MW. G3 is the balancing machine. A wind mode gradually decreases as virtual inertia control gain Kω
farm equalized by 500 DFIG wind turbines with stator-side increases. Meanwhile, when the PI parameters of PLL apply the
capacity of 1.5 MW is integrated at Bus 2. Each wind turbine weak damping values, the values of damping ratio are bigger
operates with 1.4 MW power output, thus the total power output than when the PI parameters apply the typical values. This is
of wind farm is 700 MW. The total system load is 2734 MW. contrary to the case without virtual inertia control.
This paper is mainly focused on analyzing the impact of DFIG Fig. 8 shows the participation factors of DFIG state variables
integration on system small-signal stability considering PLL in system inter-area oscillation modes when the PI parameters
and virtual inertia control, i.e. how the integration of DFIG (as a of PLL and the virtual inertia apply different values. It can
dynamical system) to power grid affects the electromechanical be seen that, the participation factors of virtual inertia control
2100 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 32, NO. 3, MAY 2017

Fig. 8. Participation factors of DFIG state variables with different PI


parameters. (a) K P P L L = 73.6, K I P L L = 333.3. (b) K P P L L = 0.5,
K I P L L = 17.

state variable xω and PLL state variable δPLL in system inter-


area oscillation modes are much bigger than those of other
state variables, and the contribution of virtual inertia control
state variable to system inter-area oscillation modes is dominant,
compared with that of PLL.
According to the variation of inter-area oscillation damping
ratio and participation factors with the PI parameters of PLL,
Fig. 6. Variation of inter-area oscillation mode damping ratio with vir- it can be see that the effects of PLL on system stability for
tual inertia. (a) K P P L L = 73.6, K I P L L = 333.3. (b) K P P L L = 0.5, traditional DFIG and DFIG with virtual inertia control are dif-
K I P L L = 17. ferent. Traditional DFIG modifies PLL parameters to make the
PLL modes manifest weak damping and be close to the system
electromechanical oscillation modes in frequency, thus the PLL
modes could affect system electromechanical oscillation modes
significantly. For DFIG with virtual inertia control, PLL affects
the system damping mainly by affecting the participation degree
of virtual inertia in system oscillation modes. The smaller the
PI parameters of PLL are, the less accurately PLL could track
the variation of system frequency, thus the weaker the participa-
tion of virtual inertia control state variable in system inter-area
oscillation modes, and the stronger the system stability.

C. Verification of the Impact of PLL on System Small-Signal


Stability
Fig. 7. Variation of inter-area oscillation damping ratio with virtual
inertia gain.
To verify the correctness of the above conclusion, the rela-
tionship between the participation factor of virtual inertia con-
trol state variable and the damping ratio of inter-area oscillation
MA et al.: RESEARCH ON THE IMPACT OF DFIG VIRTUAL INERTIA CONTROL ON POWER SYSTEM SMALL-SIGNAL STABILITY 2101

Fig. 9. Variation of damping ratio with PI parameters of PLL. (a) Typical


values. (b) Weak damping values.

Fig. 11. Dynamic responses of system with different PI parameters. (a) The
variation part of wind farm active power output (p.u.). (b) G1-G4 relative power
angle.

mode is further examined through eigenvalue analysis and time-


domain simulation.
1) Eigenvalue Analysis: Fig. 9 shows the variation of damp-
ing ratio with PLL PI parameters. Keep Kω = 10 and increase
KI PLL from 314 to 350 in the step of 2 and increase KP PLL
from 40 to 110 in the step of 5, the results are shown in Fig. 9(a).
Keep Kω = 10 and increase KI PLL from 10 to 19 in the step
of 0.5 and increase KP PLL from 0.1 to 1 in the step of 0.05, the
results are shown in Fig. 9(b).
Fig. 10 shows the variation of participation factors of DFIG
state variables with integral gain of PLL. Keep KP PLL = 73.6
and Kω = 10, increase KI PLL from 300 to 360 in the step of
10, the results are shown in Fig. 10(a). Keep KP PLL = 0.5 and
Kω = 10, increase KI PLL from 14 to 20 in the step of 1, the
results are shown in Fig. 10(b).
There is similar trend in variation participation factor with
the increase of KP PLL and thus is not mentioned here. It can
be concluded that, as KP PLL and KI PLL increase, the partic-
ipation factor of virtual inertia control state variable increases,
while the damping ratio of inter-area oscillation mode decreases.
Thus it is verified that PLL affects the system damping mainly
by affecting the participation degree of virtual inertia in the
system.
2) Time-Domain Simulation: Set Kω = 10, and conduct
time-domain simulation tests on the system when the PI pa-
Fig. 10. Variation of participation factor of DFIG state variables with rameters of PLL apply the typical values and weak damping
K I P L L . (a) K P P L L = 73.6. (b) K P P L L = 0.5. values respectively. Suppose a power disturbance occurs in the
2102 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 32, NO. 3, MAY 2017

system, i.e. the load suddenly grows by 200 MW at t = 3s. The can be expressed as:
dynamic response of system is shown in Fig. 11. ⎧
Due to the introduction of virtual inertia control, DFIG is ⎪
⎪ usd = ψ̇sd − ωs ψsq


similar to synchronous generator in inertia characteristics, and ⎪


⎪ usq = ψ̇sq + ωs ψsd
could support system active power within a short time after ⎪


⎪ urd = ψ̇rd − (ωs − ωr ) ψrq + Rr ird
the frequency fluctuation. It can be seen that, the bigger the ⎪

PI parameters of PLL are, the bigger the inertia support wind ⎪

urq = ψ̇rq + (ωs − ωr ) ψrd + Rr irq
turbines provide for the system is, and the poorer the system (A1)

⎪ ψsd = Ls isd + Lm ird
power angle stability is. This conclusion is consistent with the ⎪



above eigenvalue analysis. ⎪
⎪ ψsq = Ls isq + Lm irq





⎪ ψrd = Lr ird + Lm isd


VI. CONCLUSION ψrq = Lr irq + Lm isq
A small-signal model of interconnected system with DFIG The transient behavior of DFIG rotor voltage can be expressed
integration considering the PLL and virtual inertial control is as:
established in this paper, and the small-signal stability of power
system under the joint influence of PLL and virtual inertial  Xs − Xs 1 
Ėfd = − isq −  Efd
control is analyzed. The following conclusions are reached: T0 T0
1) When PLL and virtual inertia control are considered,
DFIG becomes coupled with synchronous generators in  ωs Lm
+ (ωs − ωr ) Efq − urq (A2)
dynamic characteristics, thus DFIG will participate in the Lr
electromechanical oscillation of system, the participation
 Xs − Xs 1  
degree depending on the structure parameters of PLL and Ėfq =  isd −  Efq − (ωs − ωr ) Efd
T0 T0
virtual inertia.
2) When the PI parameters of PLL apply the weak damping ωs Lm
+ urd (A3)
values, the damping ratio remains almost unchanged as the Lr
virtual inertia varies. However, when the PI parameters of
PLL apply typical values, the damping ratio decreases where
obviously as the virtual inertia increases. ωs Lm

3) For DFIG wind turbines with virtual inertial control, the Efd = − ψrq (A4)
PLL affects system damping mainly by affecting the par- Lr
ticipation of virtual inertia in the system. The smaller the  ωs Lm
Efq = ψrd (A5)
PI parameters of PLL are, the less accurately PLL could Lr
track system frequency variation, thus the participation ωs  
of virtual inertia control state variables in the inter-area Xs = Lm Lr − L2m (A6)
Lr
oscillation mode will be weaker, which will improve sys-
tem power angle stability. Therefore, for wind turbines Xs = ωs Ls (A7)
containing virtual inertia, the cooperation between PLL T0 = Lr / Rr (A8)
and virtual inertia should be taken into account when
designing the parameters of PLL. The rotor-side inverter applies the stator flux directional rotor
Since how change of machine inertia affects power system current control method to realize the decoupling of DFIG active
electromechanical oscillation modes is truly complicated [25],
power and reactive power. The state equations of the inverter
the conclusions in this paper cannot be simply generalized to
control system are:
all large-scale inter-connected systems and operating conditions
and the effect of mechanical inertia on system stability needs ẋ1 = Pref − Pm eas (A9)
further research in depth. However, the research in this paper
reveals that, the introduction of DFIG virtual inertia control may ẋ2 = irq ref − irq = KP 1 (Pref − Pm eas ) + KI 1 x1 − irq
to some extent cause system small-signal stability to deteriorate,
while PLL parameters could affect the small-signal stability by (A10)
affecting the participation degree of virtual inertia, thus the joint ẋ3 = Qref − Qm eas (A11)
effect of the two factors should be considered in the design of
control strategy and parameters. ẋ4 = ird ref − ird = KP 3 (Qref − Qm eas ) + KI 3 x3 − ird
(A12)

APPENDIX A urq = KP 2 (irq ref − irq ) + KI 2 x2 + sωs Lm isd + sωs Lr ird

In the selected dq rotating coordinate system, neglect the (A13)


flux saturation effect and stator resistance, and apply the motor urd = KP 2 (ird ref − ird ) + KI 2 x4 − sωs Lm isq − sωs Lr irq
convention, so that the stator/rotor voltage and flux equations
(A14)
MA et al.: RESEARCH ON THE IMPACT OF DFIG VIRTUAL INERTIA CONTROL ON POWER SYSTEM SMALL-SIGNAL STABILITY 2103

The shaft model with single mass module is applied in this


paper to represent the mechanical part of DFIG. 1
BFF = − (xd sin2 δ + xq cos2 δ),
xd xq
2Hr ω̇r = Tm − Te (A15) xd − xq
  
 GFF =
2xd xq
sin 2δ;
Te = isd Efd + isq Efq ωs (A16)
aq = GF cos δ − BF sin δ,
where Hr is the inertia time constant of generator. Tm is the bq = BFF cos δ + GFF sin δ;
mechanical torque of wind turbine and Te is the electromagnetic
torque of generator. ad = GF sin δ + BF cos δ,
The pitch angle control system can be described as follows: bd = BFF sin δ − GFF cos δ;
aδ = Ed [(GF − BF ) sin δ + (GF + BF ) cos δ]
dβ 1
= (β0 − β) (A17) + Eq [(GF − BF ) cos δ − (GF + BF ) sin δ]
dt Tβ
− (GF Ux − BF Uy ),
where β0 is the initial pitch angle, Tβ is the inertia time constant bδ = Ed [(BFF

+ GFF ) sin δ + (BFF − GFF ) cos δ]
of pitch angle control system.
+ Eq [(BFF

+ GFF ) cos δ + (GFF − BFF ) sin δ]

− (BFF Ux + GFF Uy );
APPENDIX B
dGF x − xq
First, in the model of synchronous generator, the stator voltage GF = = − d   cos 2δ, BF , BFF

, GFF
dδ xd xq
equation is expressed in the admittance form:

     are respectively the derivative of BF , BFF , GFF along δ.


id 1 0 X Ed − ud Similarly, in the equivalent wind farm model, the stator volt-
=   (B1)
iq Xd Xq −Xd 0 Eq − uq age equation of DFIG wind farm shown in (A-1) is expressed
in the admittance form:
Pre-multiplying the left
 side and post-multiplying
 the right     

sin δ cos δ isd 1 0 −1 Efd − usd
side of (18) by matrix , so that (B1) is trans- =   (B4)
− cos δ sin δ isq Xs 1 0 Efq − usq
formed from dq − xy coordinate system:

     Pre-multiply
 both sides of the above equation by ma-
Ix GF sin δ · Ed + cos δ · Eq − Ux
−BF sin δPLL cos δPLL
= trix , so that it is transformed from
Iy BFF − cos δ · Ed + sin δ · Eq − Uy
GFF − cos δPLL sin δPLL
(B2) dq − xy coordinate system:
where Ix and Iy are the x-axis and y-axis generator terminal
currents. Ux and Uy are the x-axis and y-axis generator terminal    
Ifx 1 0 −1
voltages. =
Ify Xs 1 0
The increment expression of the above equation is:
   
sin δPLL · Efd + cos δPLL · Efq − Ufx
       ×   (B5)
ΔIx GF −BF ΔUx a − cos δPLL · Efd + sin δPLL · Efq − Ufy
= − + q ΔEq
ΔIy BFF GFF ΔUy by
    It can be seen from (B5) that, the admittance form of
ad a
+ ΔEd + δ Δδ (B3) DFIG stator voltage in the xy coordinate system is similar
bd bδ
to that of synchronous generator shown in (B2). Thus, the
increment expression of DFIG is also similar to that of syn-
where chronous generator shown in (B3). In this case, GF = GFF = 0,
BF = BFF = X1  .
s
xd − xq Combine the system network admittance matrices, and elim-
GF = − sin 2δ, inate all the load nodes, so that the network admittance matrix
2xd xq
equation containing only generator terminal nodes is gained.
1 Apply this equation to (B3), so that the increment matrix
BF = − (xd cos2 δ + xq sin2 δ),
xd xq expression of generator terminal voltages expressed in state
2104 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 32, NO. 3, MAY 2017

variables can be gained: elements in (10) can be seen as follows:



 ⎤ ⎡  c̃
⎡ ĉδ
⎤ xq,i − xq,i
  Δ Ũ x
δ
  a31 = diag  · [diag (cos δi ) · ẽδ
⎢ ΔUx,n ⎥ ⎢ cδ,n n ⎥

c Tq0,i
ΔU x
= ⎢   ⎥=⎢ ⎢  δ ⎥

Δδ
⎣ ΔŨ ⎦ ⎣ d˜δ dˆδ ⎦ ΔδPLL
ΔU y y

+ diag (sin δi ) · f˜δ + diag(cos δi · Iy,i − sin δi · Ix,i )],
ΔUy ,n d δ dδ,n n 
⎡ ⎤ xq ,i − xq ,i
c̃d ĉd a33 = diag · [diag(cos δi ) · ẽd
Tq 0,i
⎢ c d cd,n n ⎥ 
 
⎢  ⎥ ΔE  d 
+⎢ ˜ ⎥
⎣ dd dˆd ⎦ ΔE  1
fd + diag(sin δi ) · f˜d ] − diag ,

dd dd,n n Tq 0,i
⎡ ⎤ 
c̃q ĉq xq ,i − xq ,i
⎢    a34 = diag · [diag(cos δi ) · ẽq
cq ,n n ⎥
⎢ cq ΔE  q Tq 0,i
+ ⎢ ˜ ⎥
⎥ (B6)
dˆq
⎣ dq ⎦ ΔEfq 
 + diag(sin δi ) · f˜q ],
d q dq ,n n 
      xd,i − xd,i
cδ cd c a41 = −diag · [diag(sin δi ) · ẽδ
= Δδ z + ΔE zd + q ΔE  zq
 
Td0,i
dδ dd dq
+ diag(cos δi ) · f˜δ + diag(cos δi · Ix,i sin δi · Iy,i )],
Since the increment expression of DFIG stator voltage is 
similar to that of synchronous generator, DFIG can be taken as xd,i − xd,i
the nth synchronous generator when solving the node voltages a43 = −diag  · [diag(sin δi ) · ẽd
Td0,i
and currents in the network. Therefore, the increment matrix
expression of generator terminal currents in the whole system − diag(cos δi ) · f˜d ] ,
can be further gained: 
xd,i − xd,i
        a44 = −diag  · [diag(sin δi ) · ẽq
ΔI x eδ ed eq Td0,i
= Δδ z + ΔE zd + ΔE zq
ΔI y fδ fd fq 
⎡ ⎤ ⎡ ⎤ 1
− diag(cos δi ) · f˜q ] − diag 
ẽδ êδ ẽd êd Td0,i
⎢  ⎥  ⎢  ⎥ 
⎢ eδ eδ,n n ⎥ Δδ ⎢ e d ed,n n ⎥ ΔE d
=⎢ 
⎢ f˜
⎥
⎥ ΔδPLL + ⎢
⎢ f˜
⎥
⎥ ΔE  Subscript i represents the number of synchronous generators,
⎣ δ fˆδ ⎦ ⎣ d fˆd ⎦ fd i = 1, 2, · · · , n − 1.
 
fδ fδ,n n f d fd,n n Parameters of DFIG wind turbines in Fig. 3 are: stator re-
⎡ ⎤ ⎡ ⎤ sistance Rs = 0 pu, rotor resistance Rr = 0.05631 p.u., sta-
ẽq êq ΔI˜x tor leakage inductance Ls = 0.1 p.u., rotor leakage inductance
⎢  ⎥  ⎢ ⎥
⎢ eq eq ,n n ⎥ ΔE  q ⎢ ΔIx,n ⎥ Lr = 0.03129 p.u., mutual inductance Lm = 0.13129 p.u.,
+⎢ 
⎢ f˜
⎥
⎥ ΔE  = ⎢
⎢ ΔI˜
⎥
⎥ (B7)
⎣ q fˆq ⎦ fq ⎣ y ⎦
DFIG inertia time constant Hr = 5.29 s, virtual iner-
 tia Kω = 10 p.u. Control parameters of rotor-side inverter
fq fq ,n n ΔIy ,n are: KP 1 = 0.3 p.u, KI 1 = 8 s−1 ; KP 2 = 0.55 p.u, KI 2 =
100 s−1 ; KP 3 = 0.3 p.u., KI 3 = 8 s−1 ; KP 4 = 1.25 p.u,
where KI 4 = 300 s−1 . The PLL PI controller parameters are
⎧    KP PLL = 73.6, KI PLL = 333.3.
⎪ GF ,i −BF ,i cℵ,ij
  ⎪ ⎪−
⎨ B (i = j)
eℵ,ij F F ,i GF F ,i dℵ,ij
=      REFERENCES
fℵ,ij ⎪
⎪ GF ,i −BF ,i cℵ,ij aℵ,ij
⎪−
⎩ + (i = j) [1] Y. Mishra, S. Mishra, F. Li, Z. Y. Dong, and R. C. Bansal, “Small-
BF F ,i GF F ,i dℵ,ij bℵ,ij signal stability analysis of a DFIG-based wind power system under dif-
ferent modes of operation,” IEEE Trans. Power Convers., vol. 24, no. 4,
(i, j = 1, 2, · · · , n) pp. 972–982, Dec. 2009.
[2] S. Q. Bu, W. Du, and H. F. Wang, “Investigation on probabilistic small-
ℵ corresponds to subscripts δ, d and q. signal stability of power systems as affected by offshore wind generation,”
IEEE Trans. Power Syst., vol. 30, no. 5, pp. 2479–2486, Sep. 2015.
Apply (B6) and (B7) to the linearized equations of DFIG and [3] F. Mei and B. Pal, “Modal analysis of grid-connected doubly fed induction
synchronous generator respectively, so that the elements in the generators,” IEEE Trans. Energy Convers., vol. 22, no. 3, pp. 728–736,
characteristic matrix of the inter-connected system As can be Sep. 2007.
[4] R. Cárdenas, R. Peña, S. Alepuz, and G. Asher, “Overview of control
obtained, shown in (10)–(13). The detailed expression of the systems for the operation of DFIGs in wind energy applications,” IEEE
Trans. Ind. Electron., vol. 60, no. 7, pp. 2776–2798, Jul. 2013.
MA et al.: RESEARCH ON THE IMPACT OF DFIG VIRTUAL INERTIA CONTROL ON POWER SYSTEM SMALL-SIGNAL STABILITY 2105

[5] J. G. Slootweg and W. L. Kling, “The impact of large scale wind power Jing Ma (S’06–M’08) was born in Hebei, China, on Feburary 25, 1981. He
generation on power system oscillations,” Elect. Power Syst. Res., vol. 67, received the B.S. and Ph.D. degrees from North China Electric Power University,
no. 1, pp. 9–20, Oct. 2003. Beijing, China, in 2003 and 2008, respectively. He is currently an Associate
[6] S. Jiaying and S. Chen, “Impact of DFIG wind power on power system Professor in the School of Electrical and Electronic Engineering, North China
small signal stability,” in Proc. IEEE PES Innovative Smart Grid Technol., Electric Power University. He was a visiting scholar in the Bradley Department
Washington, DC, USA, 2013, pp. 1–6. of Electrical and Computer Engineering, Virginia Polytechnic Institute and
[7] Z. Wang, C. Shen, and F. Liu, “Impact of DFIG with phase lock loop State University, Blacksburg, VA, USA, from 2008 to 2009. His major interests
dynamics on power systems small signal stability,” in Proc. IEEE PES include power system modeling, wide-area monitoring, and small signal stability
General Meeting Conf. Expo., Jul. 2014, pp. 1–5. and control.
[8] J. Morren, S. W. H. de Haan, W. L. Kling, and J. A. Ferreira, “Wind turbines
emulating inertia and supporting primary frequency control,” IEEE Trans.
Power Syst., vol. 21, no. 1, pp. 433–434, Feb. 2006.
Yang Qiu was born in Shandong, China. She received the B.S. degree from
[9] K. Ping-Kwan, L. Pei, H. Banakar, and B. T. Ooi, “Kinetic energy of wind-
turbine generators for system frequency support,” IEEE Trans. Power North China Electric Power University, Beijing, China, in 2014. She is currently
working toward the Master’s degree in the School of Electrical and Electronic
Syst., vol. 24, no. 1, pp. 279–287, Feb. 2009.
Engineering, North China Electric Power University. Her interests mainly in-
[10] M. Kayikçi and J. V. Milanovic, “Dynamic contribution of DFIG based
clude power system stability analysis and control.
wind plants to system frequency disturbances,” IEEE Trans. Power Syst.,
vol. 24, no. 2, pp. 859–867, May 2009.
[11] M. F. M. Arani and E. F. El-Saadany, “Implementing virtual inertia in
DFIG-based wind power generation,” IEEE Trans. Power Syst., vol. 28,
no. 2, pp. 1373–1384, May 2013. Yinan Li was born in Fujian, China. She received the B.S. degree from North
[12] X. Xi, H. Geng, and G. Yang, “Small signal stability of weak power China Electric Power University, Beijing, China, in 2012. She is currently
system integrated with inertia tuned large scale wind farm,” in Proc. IEEE working toward the Master’s degree in the School of Electrical and Electronic
PES Innovative Smart Grid Technol., Kuala Lumpur, Malaysia, 2014, Engineering, North China Electric Power University. Her interests mainly in-
pp. 514–518. clude power system stability analysis and control.
[13] G. Lalor, A. Mullane, and M. O’Malley, “Frequency control and
wind turbine technologies,” IEEE Trans. Power Syst., vol. 20, no. 4,
pp. 1905–1913, Nov. 2005.
[14] S. Wang, J. Hu, X. Yuan, and L. Sun, “On inertial dynamics of virtual-
synchronous-controlled DFIG-based wind turbines,” IEEE Trans. Energy Weibo Zhang was born in Zhejiang, China. He received the B.S. degree from
Convers., vol. 30, no. 4, pp. 1691–1702, Dec. 2015. North China Electric Power University, Beijing, China, in 2015. He is currently
[15] S. Müller, M. Deicke, and R. W. D. De Doncker, “Doubly fed induction working toward the Master’s degree in the School of Electrical and Electronic
generator system for wind turbines,” IEEE Ind. Appl. Mag., vol. 8, no. 3, Engineering, North China Electric Power University. His interest mainly include
pp. 26–33, May 2002. power system protection.
[16] J. Ekanayake and N. Jenkins, “Comparison of the response of doubly fed
and fixed-speed induction generator wind turbines to changes in network
frequency,” IEEE Trans. Energy Convers., vol. 19, no. 4, pp. 800–802,
Dec. 2004.
[17] X. Zhu, Y. Wang, L. Xu, X. Zhang, and H. Li, “Virtual inertia control of
DFIG-based wind turbines for dynamic grid frequency support,” in Proc.
IET Conf. Renewable Power Generation, Sep. 2011, pp. 1–6. Zhanxiang Song was born in Hebei, China. He received the B.S. degree from
[18] N. W. Miller, K. Clark, M. E. Cardinal, and R. W. Delmerico, “GE wind North China Electric Power University, Beijing, China, in 2015. He is currently
plant dynamic performance for grid and wind events,” in Proc. AEE Tech- working toward the Master’s degree in the School of Electrical and Electronic
windgrid, Madrid, Spain, Apr. 2009. Engineering, North China Electric Power University. His interests mainly in-
[19] G. Tapia, A. Tapia, and J. X. Ostolaza, “Two alternative modeling ap- clude power system stability analysis and control.
proaches for the evaluation of wind farm active and reactive power per-
formances,” IEEE Trans. Energy Convers., vol. 21, no. 4, pp. 909–920,
Dec. 2006.
[20] P. Kundur, Power System Stability and Control. New York, NY, USA:
McGraw-Hill, 1994. James S. Thorp (S’58–M’63–SM’80–F’89–LF’03) is the Hugh P. and Ethel C.
[21] Y. Ni, S. Chen, and B. Zhang, Power System Dynamics: Theory and Kelley Professor of Electrical and Computer Engineering and the Department
Analysis. Beijing, China: Tsinghua Univ. Press, 2001. Head of the Bradley Department of Electrical and Computer Engineering at
[22] G. Rogers, Power System Oscillations. Norwell, MA, USA: Kluwer, 2000. Virginia Tech, Blacksburg, VA, USA. He was the Charles N. Mellowes Professor
[23] L. Hadjidemetriou, E. Kyriakides, and F. Blaabjerg, “A new hybrid PLL in Engineering at Cornell University from 1994 to 2004. He was the Director
for interconnecting renewable energy systems to the grid,” IEEE Trans. of the Cornell School of Electrical and Computer Engineering from 1994 to
Ind. Appl., vol. 49, no. 6, pp. 2709–2719, Nov./Dec. 2013. 2001, a Faculty Intern, American Electric Power Service Corporation in 1976–
[24] G. F. Franklin, J. D. Powell, and A. Emami-Naeini, Feedback Control 1977, and an Overseas Fellow, Churchill College, Cambridge University in
of Dynamic Systems, 4th ed. Englewood Cliffs, NJ, USA: Prentice Hall, 1988. He was an Alfred P. Sloan Foundation National Scholar and a Member
2002. of the National Academy of Engineering in 1996. He received the 2001 Power
[25] D. Gautam, V. Vittal, and T. Harbour, “Impact of increased penetration of Engineering Society Career Service award, the 2006 IEEE Outstanding Power
DFIG-based wind turbine generators on transient and small signal stability Engineering Educator Award, and shared the 2007 Benjamin Franklin Medal
of power systems,” IEEE Trans. Power Syst., vol. 24, no. 3, pp. 1426–1434, with A. G. Phadke.
Aug. 2009.