IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO.

12, DECEMBER 2024 15881

Improving Transient Stability of Grid-Forming

DFIG Based on Enhanced Hybrid
Synchronization Control
Zhe Li , Zhen Xie , Member, IEEE, Shang Xu , and Xing Zhang , Senior Member, IEEE

Abstract—Grid-forming (GFM) technique is attractive for Ws,d, Ws,q dq-axis components of the stator flux.
doubly fed induction generator (DFIG)-based wind turbines Ig,d, Ig,q dq-axis components of the current output by grid-side
(WTs) to realize its stable operation and grid-support ability converter (GSC).
under weak ac grid. Meanwhile, GFM-DFIG suffers from Vdc, C DC-link voltage and DC-link capacitance.
instability under grid voltage sag due to its poor ability to LF, CF Filter inductance and filter capacitance.
resist disturbed grid voltage. In this article, focusing on its
transient stability under symmetrical grid voltage fault, the Subscripts
large-signal model of GFM-DFIG with voltage-orientation ref Reference value.
control (VOC) considered is established, based on which its
transient behavior is studied. It is found that two types of I. INTRODUCTION
transient instability exist during a symmetrical grid voltage
fault. The first is induced by the inexistence of the postfault
equilibrium point (EP). The second is induced by the exces-
sive dynamic deviation between synchronous angle and
W IND power generation is one kind of essential sources in
power system in view of its contribution to low-carbon
economy. With the sustained capacity growth of wind power
point of common coupling (PCC) voltage angle. Given that
the postfault EP and transient damping are vital for tran-
generation, the renewable energy generation (REG)-dominated
sient stability, an enhanced hybrid-synchronization control power system has formed in some local area located far away
(EHSC) strategy based on adaptive Integrator is proposed, from load center, causing low short-circuit ratio (SCR) and great
through which both the postfault EP and the transient necessity of grid-supporting wind power generation [1].
damping can be improved. Finally, theoretical findings and Among various types of wind power generation, doubly fed
effectiveness of proposed strategy are verified with a 5 kW
DFIG-based experimental platform.
induction generator (DFIG)-based wind turbines (WTs) are used
widely for its lower costs of construction and maintenance. The
Index Terms—Doubly fed induction generator (DFIG), grid-supporting ability of DFIG-based WTs can be realized
grid-forming (GFM), transient stability. through grid-following (GFL) control or grid-forming (GFM)
control. The GFL-DFIG synchronizes with grid through phase-
NOMENCLATURE locked loop (PLL), the output of which does not respond to the
variation of grid frequency and voltage [2], so virtual inertia
hpcc, href PCC voltage angle and synchronous angle. (VI) and automatic voltage regulator (AVR) are needed to real-
Vpcc, Vref Amplitudes of PCC voltage and command.
hsl, hr Slip angle and electrical angle of rotor. ized its grid-supporting ability [3], [4]. As a contrast, the GFM-
kslip, kSCR Slip rate of DFIG and short-circuit ratio (SCR). DFIG synchronizes with grid via its power controller (PC), and
hpll Output angle of phase-locked loop (PLL). it can support grid frequency and voltage without additional
Ls, Lr, Lm Stator, rotor and mutual inductances of DFIG. control loops [5], [6]. In 2023, the Australian energy market
Lg, Xg Grid inductance and grid reactance. operator (AEMO) established an expert group on the specifica-
Vpcc,d,Vpcc,q dq-axis components of the PCC voltage.
Ipcc,d, Ipcc,q dq-axis components of the PCC current. tion for GFM devices, and it pointed out that a GFM device
Vs,d, Vs,q dq-axis components of the stator voltage. should behave like a voltage source while in normal operation,
Is,d, Is,q dq-axis components of the stator current. thereby allowing immediate response to a change in the external
grid [7].
Regarding as the choice between GFL and GFM control, it is
Manuscript received 26 October 2023; revised 22 January 2024 and found that GFL control is more suitable in stiff grid (i.e., SCR > 3)
12 March 2024; accepted 12 April 2024. Date of publication 23 May and GFM control is more suitable in weak grid (i.e., SCR < 3),
2024; date of current version 11 October 2024. This work was due to their small-signal stability relevant to grid impedance [8], [9].
supported by the National Natural Science Foundation of China under
Grant 52377177. (Corresponding author: Zhen Xie.) Therefore, it is suggested for DFIG-based WTs to operate with
The authors are with the School of Electrical Engineering and GFM control under weak grid condition. However, this choice is
Automation, Hefei University of Technology, Hefei 230009, China hard to trust when a grid voltage fault occurs. Limited by the
(e-mail: [email protected]; [email protected]; poor ability to provide short-circuit current, GFM-REGs tend to be
[email protected]; [email protected]).
unstable if it tries to maintain its GFM ability under a grid voltage
Color versions of one or more figures in this article are available at
https://doi.org/10.1109/TIE.2024.3393134. fault [10], which aggravates the imbalance of active power control
Digital Object Identifier 10.1109/TIE.2024.3393134 and the uncontrollability of grid voltage. To avoid this problem,

0278-0046 © 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
 15882 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 12, DECEMBER 2024

switching the control algorithm from GFM to GFL is considered its transient stability, but GFM-DFIG should switch its reactive
under lager-signal disturbances [11], [12], which seems to be a fea- power loop to avoid the fault current caused by the excessive
sible scheme. In fact, GFL-DFIG is also vulnerable to small-signal reactive power [27]. Therefore, the reliability of those transient
instability during a weak grid fault [13], [14], [15]. In [13], the stability analysis and methods for GFM-VSC are uncertain for a
small-signal behavior of a GFL-DFIG attached to weak grid is GFM-DFIG. Given that less researches focus on the transient
investigated, and it is found that the negative damping due to the stability of GFM-DFIG, additional research on its transient
interaction between PLL and rotor current control will be increased behavior and transient stability is necessary.
under deeper voltage sag. Similar conclusions are also obtained in In this article, focusing the transient behavior of GFM-DFIG
[14], [15], where the stability of GFL-DFIG during grid fault is in electromechanical time scale, both the dynamic of inner volt-
investigated based on the impedance method and the singular per- age control and the switched reactive power control is consid-
turbation theory, respectively. Thus, new schemes should be pro- ered during the transient modeling process. The impacts of
vided to ensure the transient stability of GFM-DFIG during the control parameter in APC and inner voltage loop on transient
weak grid fault. stability is studied with the state trajectory, and an enhanced
There are many researches on the transient stability of GFM hybrid-synchronization control (EHSC) strategy is proposed to
control under a symmetrical grid voltage fault, most of which improving the transient stability boundary of GFM-DFIG under
are relevant to a voltage-source converter (VSC) system [16], a symmetrical grid voltage sag. The contributions of this article
[17], [18], [19], [20]. The transient stability of GFM-VSC is can be concluded as follows.
studied in [16], where noninertia GFM control and inertia GFM 1) Two types of transient instability for GFM-DFIG during
control are investigated comprehensively with the phase por- a symmetrical grid voltage fault is revealed based on the
trait. In [17], the damping in GFM control is quantitatively large-signal model and state trajectory analysis. The first
assessed via the damping energy visualization and geometry is induced by the inexistence of the postfault EP. The
approximation, which has critical impact on its transient stability second is induced by the excessive dynamic deviation
boundary and critical clearing angle (CCA). In [18], the large- between synchronous angle and point of common cou-
signal stability of GFM control and GFL control is compara- pling (PCC) voltage angle.
tively studied based on the energy function model, and it is 2) It is found that the postfault EP of GFM-DFIG tends to
shown that GFL-VSC will lose stability by exhibiting a varying be nonexistent with lower grid voltage, SCR and higher
damping coefficient. In [19], both the CCA and critical clearing postfault active power command, which is similar to that
time (CCT) is analyzed to quantitatively describe the transient of a GFM-VSC.
stability boundary of virtual synchronous generator (VSG). In 3) Voltage-orientation control (VOC) plays considerable
[20], the transient stability of hybrid power systems dominated roles on the resynchronization of GFM-DFIG, which
by different GFM devices is investigated, which contributes the regulates the active rotor current based on dynamic devi-
further appliance of GFM control in modern power system. In ation between synchronous angle and PCC voltage
those researches, the active and reactive power control loops are angle, so the second types of transient instability can be
focused, so many methods on enhancing transient stability of improved with smaller P-f droop gain and larger propor-
GFM control are proposed on this basis [21], [22], [23], [24], tional and integral gain of VOC.
[25]. The poor damping of VSG during a grid voltage fault is 4) The EHSC strategy is proposed to address the transient
usually treated as a main factor causing instability, so a transient instability of GFM-DFIG. The transient damping can be
damping method (TDM) is introduced in [21] to enhance the improved by the conventional HSC strategy. Specially,
transient stability of GFM-VSC. An enhanced VSG controller is synchronous angle when postfault EP does not exist can
proposed in [22], which also focuses the optimization of VSG be limited by an adaptive integrator, which generates an
control. A mode- adaptive power-angel control method is pro- adaptive compensation based on the cumulative error
posed in [23] to restrain the divergent power angle when there is between synchronous angle and PCC voltage angle.
no postfault equilibrium point (EP). In addition, the impacts of The rest of this article is organized as follows. The structure
current limitation on the transient stability of GFM-VSC have of GFM-DFIG is introduced in Section II. The large-signal
been paid more attention [24], [25], since steady-state overcur- modeling and transient stability analysis are given in Section III.
rent during a grid voltage fault will be inevitable for the In Section IV, the concept of HSC is introduced first, then
unchanged power control loops. EHSC strategy is proposed to improve the transient stability of
Although those researches are valuable for solving transient GFM-DFIG, and its effectiveness is validated by comparing the
instability of a GFM-DFIG, there still exist differences between stable area of EHSC and HSC. Experimental validation is given
a GFM-VSC and a GFM-DFIG. First, the dynamic of inner volt- in Section V. Finally, Section VI concludes the article.
age control is neglected in most transient stability analysis of
GFM-VSC, since the bandwidth of inner voltage control can be II. GFM CONTROL OF DFIG-BASED WTS
designed high to realize a faster power response. However, lim-
A. Basic Principle of GFM-DFIG
ited by the large electromagnetic time constant, dynamic of
inner voltage control is usually slower and cannot be neglected Usually, a GFM-DFIG is the DFIG-based WT realizing its
[26]. Second, the reactive power control loop of a GFM-VSC GFM capacities by rotor-side control [28], and its structure is
remain unchanged during the grid voltage fault when analyzing given in Fig. 1. It can be seen that the GFM algorithm is used in

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
LI et al.: TRANSIENT STABILITY OF GRID-FORMING DFIG 15883

rotor-side control to realize the amplitude and phase control to

the voltage at PCC, and vector current control (VCC) is used in
grid-side converter to stabilize dc voltage. The detailed princi-
ples are given as follows.
In rotor-side control, active power control (APC) can provide
a synchronous angle href based on the balance between active
power command Pref and total active power Pout in grid-side,
there is
ð
href ¼ ½KP ðPref  Pout Þ þ x0  (1)

where KP is P-f droop gain; x0 equals to 100p (rad/s). Fig. 1. Basic structure of GFM-DFIG.
The synchronous angle in (1) serves as the oriented angle of
synchronous rotating frame (SRF) in rotor-side control, then
VOC can ensure the alignment of PCC voltage vector with the
d-axis of SRF, that is,
(
ðK p;q þKi,q =sÞð0  Vpcc,q Þ ¼ Ir,d
ref
: (2)
Vpcc,q ¼ xs Lm Ir,d þ Rs Is,q þ xs Ls Is,d

In (2), the first equation represents the regulation process to

the q-axis components of PCC voltage (i.e., Vpcc,q); Kp,q, Ki,q are
proportional and integral gains of VOC. The second equation is
Fig. 2. GFM scheme realized by rotor-side control.
obtained from the voltage and flux equations of DFIG, which
provides the basis of adjusting Vpcc,q with the d-axis component
of rotor current (i.e., Ir,d). current. A closed loop for rotor current control is provided to
In view of the function of VOC, the steady-state decoupling ensure the limitation of fault current. Besides, the reactive-current
of power control can be realized based on control instead of AVR is applied to realize the fast reactive
power support during a grid voltage fault, via which the fault cur-
Vpcc,q ¼ Vs,q ¼ 0, Vpcc,d ¼ Vs,d ¼ Vpcc : (3) rent caused by the excessive reactive power can be also avoided.
Then, the active and reactive power of GFM-DFIG in steady As for grid-side control, PLL is used to acquire the phase of
state can be controlled by the d-axis and q-axis components of PCC voltage, and the dc voltage is stabilized based on
rotor current, respectively, 8 Cd
>
< Pslip  PGSC ¼ kslip Pm  PGSC ¼ ðVdc 2 Þ
3 2 dt  (6)
Pout ¼ ð1  kslip Þ Pm ¼ KðVs,d Is,d þ Vs,q Is,q Þ > 3 3 Kidc
|fflfflfflfflfflffl{zfflfflfflfflfflffl} 2 : PGSC  Vpcc Ig,d ref
¼ Vpcc ðVdcref
 Vdc Þ Kpdc þ
K 2 2 s
 
3 Ws,d  Lm Ir,d Ws,q  Lm Ir,q where Pslip ¼ kslipPm denotes the slip power transferred from
¼ K Vs,d  Vs,q
2 Ls Ls rotor windings of DFIG; PGSC is the active power output by
  
3 Vs,d Vs,q 3KLm grid-side converter; KpDC and KiDC are proportional gain and
 K   Lm Ir,d  Vpcc Ir,d (4)
2 Ls xs 2Ls integral gain of dc voltage controller.
It is noted that both the GFM control and the GFL control
3 exist in a GFM-DFIG system, but the characteristic of whole
Qout ¼ ðVs,q Is,d  Vs,d Is,q Þ system in grid-side is determined by the rotor-side control,
2 
3 Vs,q Vs,d which regulates the total grid-connected power and realizes the
¼  ðWs,d  Lm Ir,d Þ þ ðWs,q  Lm Ir,q Þ GFM capacities of DFIG.
2 Ls Ls
   
3 Vs,d Vs,d 3Vpcc Vpcc B. The Participation of VOC in Grid Synchronization
   Lm Ir,q   þ Lm Ir,q
2 Ls xs 2Ls xs
Grid synchronization is the critical for GFM-REGs to ensure
(5)
its stability under different working conditions [29]. Usually, the
where K ¼ 1  kslip represents the conversion coefficient grid synchronization of a GFM-VSC is determined by its outer
from the active power output by stator windings (i.e., Pm) to power control loops, which produces the synchronous angle on
total grid-connected active power Pout. basis of power balance, and the inner control loops including
Therefore, a GFM scheme as Fig. 2 can be constructed based voltage control and current control are neglected. A typical syn-
on (1)–(5). The functions of APC and AVR form the basic GFM chronous loop of GFM-VSC is given in Fig. 3(a).
capacity of the unit. VOC serves as the mediate link for the volt- As shown in Fig. 3(a), the droop-controlled GFM-VSC is a
age orientation and the reference adjustment of d-axis rotor first-order system, and the synchronous angle can be treated as

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
 15884 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 12, DECEMBER 2024

First, treating stator current as the total grid-side current, the

relationship between rotor current and grid-side current is given as

Ipcc,d ¼ Is,d ¼ ðWs,d =Ls  Lm Ir,d =Ls Þ
(8)
Ipcc,q ¼ Is,q ¼ ðWs,q =Ls  Lm Ir,q =Ls Þ

where all variables are expressed in SRF. The positive direction of

stator current and grid-side current is flowing into the grid.
Then, the stator flux in (8) can be calculated based on the volt-
age model, neglecting stator resistor, there is

Ws,d ¼ ðVpcc,q þ Rs Is,q Þ=xs  Vpcc,q =xs
(9)
Ws,q ¼ ðVpcc,d þ Rs Is,d Þ=xs  Vpcc,d =xs :

According to (8) and (9), the relationship between PCC voltage

and rotor current can be expressed as

Fig. 3. Synchronous loops: (a) droop-controlled GFM-VSC; and Ipcc,d ¼ Vpcc,q =ðxs Ls Þ þ ðLm =Ls Þ  Ir,d
(b) droop-controlled GFM-DFIG. (10)
Ipcc,q ¼ Vpcc,d =ðxs Ls Þ þ ðLm =Ls Þ  Ir,q :

the power angle when the voltage phase of infinite grid bus is The voltage equation at PCC is expressed as
taken as a reference. Therefore, power angle stability is used to
describe the transient characteristics of GFM-VSC during its Vpcc ejhpcc ¼ Vpcc ejðhpcc href Þ ejhref ¼ Vg þ jXg Is ejðhpcc þuÞ (11)
grid synchronization. In Fig. 3(a), active power feedback is cal- where u denotes the power factor angle at PCC.
culated based on Equation (11) can be repressed in another SRF, the d-axis of
Pout ¼ 1:5Eref Vg  sin href =Xg : (7) which is aligned with href
However, the synchronous loop of GFM-DFIG as shown in Vpcc ejðhpcc href Þ ¼ Vg ejhref þ jXg Is ejðhpcc href þuÞ : (12)
Fig. 2 is different with that of a droop-controlled GFM-VSC,
and it is given in Fig. 3(b) for illustration. It can be seen that That is
VOC is a mediate link between the forward loop of APC and (
G
Vpcc,d ¼ Vg coshref  Xg Ipcc,q
G
¼ Vpcc cosðhpcc  href Þ
the feedback loop of active power. In Fig. 3(b), (18) denotes the (13)
calculation of active power feedback, which will be given in the Vpcc,q ¼ Vg sinhref þ Xg Ipcc,d ¼ Vpcc sinðhpcc  href Þ
G G

next section. Given that VOC regulates the active rotor current
based on the error between PCC voltage angle and the synchro- where superscript “G” denotes the SRF oriented based on GFM
nous angle, it is nonnegligible when analyzing the transient control.
characteristics of GFM-DFIG during its grid synchronization. Assuming the rotor current can always track to its reference
During a symmetrical grid voltage sag, the resynchronization in GFM-based SFR, there are
of GFM-DFIG is closely related to its transient stability. Based G
Ir,d ¼ Ir,d
ref
¼ ðKp,q þ Ki,q =sÞð0  Vpcc,q
G
Þ (14)
on the aforementioned information, a high-order large-signal
model should be considered when analyzing the transient stabil- G
Ir,q ¼ Ir,q
ref
¼ Ir,q0 þ DIr,q
lvrt
(15)
ity of GFM-DFIG, which will be mainly introduced in the
Section III of our article. where Kp,q and Ki,q are proportional and integral coefficients
of VOC; Ir,q0 is initial rotor current before grid voltage dips,
and DIr,q
lvrt
is dynamic reactive rotor current to support grid
III. LARGE-SIGNAL MODEL AND TRANSIENT STABILITY OF G G
voltage.Vpcc,d , Vpcc,q can be calculated based on (10) and
GFM-DFIG UNDER SYMMETRICAL GRID VOLTAGE DIP
(13), there is
A. Large-Signal Reduced-Order Model (
G
Vpcc,d ¼ k0 ðVg coshref  Ir,q
G
 Lm Xg =Ls Þ
The purpose of deducing the large-signal model is to analyze (16)
the transient behavior of nonlinear system, so that quantitative Vpcc,q ¼ k0 ðIr,d  Lm Xg =Ls  Vg sinhref Þ
G G

judgments can be made for the transient stability of system. As

where k0 ¼ xsLs /(xsLs þXg).
mentioned before, the grid-side characteristics of GFM-DFIG
Then, the active power can be calculated as
are determined by its rotor-side control, which regulates the total
grid-connected power and realizes the GFM capacities of DFIG, Pout ¼ 1:5ðVpcc,d
G G
Ipcc,d þ Vpcc,q
G G
Ipcc,q Þ: (17)
so the dynamic of rotor-side control system will be mainly con-
sidered when deducing the large-signal model. Considering the Substituting (10) and (16) into (17), there is
transient instability is related to the movement of synchronous
angle href in electromechanical time scale, the dynamic of rotor Lm  G
Pout ¼ 1:5  k0 Vg I coshref Ir,q
G
sinhref : (18)
current loop can be neglected. Ls r,d

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
LI et al.: TRANSIENT STABILITY OF GRID-FORMING DFIG 15885

The angle and amplitude of PCC voltage can be calculated as

!
G
Vpcc,q
hpcc ¼ arctan G þ href (19)
Vpcc,d
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vpcc ¼ G 2
Vpcc,d þ Vpcc,qG 2: (20)

Based on (1), (2) and (8)–(20), the large-signal reduced-order

model of GFM-DFIG can be finally established. Before the tran-
sient stability analysis, some mathematical work should be done
as follows:
First, the VOC dynamic is rewritten based on (14) and (16)
G
Vpcc,q ¼ k0 ½ðLm Xg =Ls ÞðKp,q þ Ki,q =sÞVpcc,q
G
þ Vg sinhref :
(21)
When letting x_ pcc,q ¼ G
Vpcc,q ,
(21) is expressed as
 
Lm Xg
x_ pcc,q ¼ k0 k1 Ki,q xpcc,q þ Vg sinhref (22)
Ls Fig. 4. State trajectories under grid voltage sag:(a) Vg ¼ 0.7 p.u., SCR ¼ 5,
Pref ¼ 0.6 p.u.; (b) Vg ¼ 0.5 p.u., SCR ¼ 5, Pref ¼ 0.6 p.u; (c) Vg ¼ 0.7 p.u.,
1
where k1 ¼ ð1 þ k0 Kp,q Xg Lm =Ls Þ : SCR ¼ 2, Pref ¼ 0.6 p.u; and (d) Vg ¼ 0.5 p.u., SCR ¼ 5, Pref ¼ 0.4 p.u.
Thus, the d-axis rotor current can be rewritten as
deviation between the PCC voltage angle hpcc and the synchro-
G
Ir,d ¼ ðKp,q x_ pcc,q þ Ki,q xpcc,q Þ ¼ f1 ðhref , xpcc,q Þ nous angle href. Therefore, conventional power angle stability
 
k0 k1 Kp,q Lm Xg defined in 2-D plane (i.e., [x, href]) is not suitable for evaluating
¼ Ki,q  1 xpcc,q þ k0 k1 Kp,q Vg sinhref :
Ls the transient stability of GFM-DFIG again. According to the
(23) established reduced-order large-signal model in (25), four state
variables (i.e., href, xpcc,q, Vpcc, and hpcc) are used to the describe
The active power in (18) can be rewritten as the transient behavior of GFM-DFIG. In the large-signal analy-
Lm h i sis later, three of the four state variables above are selected to
Pout ¼ 1:5k0 Vg f1 ðhref , xpcc,q Þcoshref  Ir,q G
sinhref : (24)
Ls |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} form a 3-D state trajectory (i.e., [hpcc, href, Vpcc]). By solving
f2 ðhref , xpcc,q Þ (25) with initial conditions, transient stability of GFM-DFIG
can be evaluated by observing its state trajectories. A conver-
Finally, the large-signal reduced-order model of a GFM- gent trajectory denotes transient stability and a divergent trajec-
DFIG can be described as tory denotes transient instability. Considering the advantage of
8  
> Lm compensating reactive current on enhancing transient stability is
>
> _
h ref ¼ KP Pref  1:5k0 Vg f2 ðhref , xpcc,q Þ
>
> Ls obvious due to its function of lifting grid voltage [15], a constant
>
>  
>
> reactive rotor current is assumed for simplification in the large-
>
> L X
m g
>
> x_ pcc,q ¼ k0 k1 Ki,q þ Vg xpcc,q sinhref signal analysis later.
>
< Ls 1) Impacts of Grid Voltage Sag, SCR and Postfault Active
> Vpcc ¼ k0 ½ðVg coshref  Ir,q
G
 Lm Xg =Ls Þ2 þ Power Command: To investigate the impacts of grid voltage
>
>
>
> sag, SCR and postfault active power command, the state trajec-
>
> ðVg sinhref  f1 ðhref ,xpcc,q Þ  Lm Xg =Ls Þ2 0:5
>
> ! tories in a 3-D space (i.e., [Vpcc,d, Vpcc,q, href]) are depicted in
>
>
>
> Lm Xg f1 ðhref , xpcc,q Þ  Ls Vg sinhref Fig. 4. It is noted that all cases above have the same initial state
>
: hpcc ¼ arctan þ href :
L V cosh  L X I G
s g ref m g r,q
[VN, 0, href0]. VN is the rated PCC voltage, and href0 is the initial
power angle calculated based on the prefault active power.
(25) In Fig. 4(a), A1 is a convergence point and B1 is its projection
on plane [Vpcc,d, Vpcc,q], which means the system is stable under
this condition (SCR ¼ 5, Vg ¼ 0.7 p.u.). It can be found that B1
B. Transient Stability Analysis
is on the line “Vpcc,q ¼ 0,” so the PCC voltage angle can track to
The impact of grid condition, APC, and VOC on the transient its reference href for the existence of VOC. Besides, the maxi-
stability of GFM-DFIG will be revealed in this subsection. Usu- mum synchronous angle href_max exceeds its steady-state value
ally, power angle stability is defined to evaluate the transient sta- href1 corresponding to A1, suggesting that overshoot of synchro-
bility of GFM-VSC, since its synchronous angle href can be nous angle occurs even though the P-f droop control is used in
treated as power angle when the voltage phase of infinite grid APC. Given that P-f droop control is a first-order element featur-
bus is taken as a reference. However, before a new stable state ing overdamped characteristics [16], this overshoot of synchro-
of system is achieved, the synchronous angle href of GFM-DFIG nous angle is caused by VOC, the impacts of which on transient
cannot be treated as power angle in view of the dynamic stability will be discussed later.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
 15886 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 12, DECEMBER 2024

The initial conditions of in Fig. 4(b) and 4(c) are changed

comparing with Fig. 4(a). The postfault grid voltage of Fig. 4(b)
decreases from 0.7 to 0.5 p.u., and the SCR of Fig. 4(c) decreases
from 5 to 2. It is shown that trajectories in Fig. 4(b) and 4(c) are
divergent, where d-axis and q-axis components of PCC voltage
are alternating. Meanwhile, the synchronous angle href monotoni-
cally increases, suggesting that GFM-DFIG tends to be unstable
with lower SCR or grid voltage during voltage sag.
In Fig. 4(d), the divergent trajectory of Fig. 4(b) is replotted
in a 3-D space [hpcc, href, Vpcc] as a comparison (i.e., the red Fig. 5. Transient trajectories under grid voltage sag: (a) increasing Kp,q;
line), and a new state trajectory corresponding to a decreased and (b) increasing Ki,q.
postfault active power command (i.e., 0.4 p.u) is plotted by the
blue line. It is shown that the state trajectory depicted by red line
is divergent (SCR ¼ 5, Vg ¼ 0.5 p.u.), and this result turns to
stable when decreasing the active power reference from 0.6 to
0.4 p.u. (i.e., the blue line), suggesting that GFM-DFIG can
keep stable by decreasing its active power command when grid
voltage sags.
In conclusion, the impacts of grid voltage sag, SCR and post-
fault active power command on the transient stability of GFM-
DFIG is still important as other GFM-based ERGs, since these
factors determine the existence of postfault EP when a grid volt-
age fault occurs. The existence of postfault EP represents the Fig. 6. Transient trajectories under grid voltage sag: increasing P-f droop
balance between the postfault active power command and the coefficient.
active power output by REGs, which can be judged by
8 fault
< Pref ¼ Pout ¼ 3Vpcc Vg =ð2Xg Þsinhref  Pmax
> fault fault
proportional gain of VOC will only influence the convergency
Pmax ¼ 3Vpcc Vg =ð2Xg Þ ¼ 3Vpcc SN =ð2VN Þ  Vg  kSCR
fault 2 speed of the resynchronization process.
>
: In Fig. 5(b), impact of integral gain Kiq is also investigated. It
kSCR ¼ VN 2 =ðXg SN Þ
is found that the convergency of trajectory is also accelerated
(26)
with a larger Kiq, and the value of integral gain imposes more
where Pfault fault
ref is the postfault active power command; Pout and significant impacts on the resynchronization process than that of
fault
Pmax represent the postfault active power output by REGs and proportional gain. Especially, a weak integral gain will lead to
its maximum values; VN and SN represent the nominal value of the failure of resynchronization [i.e., the red line in Fig. 5(b)],
PCC line voltage (rms) and the unit capacity. which is caused by the unacceptable dynamic error between
According to (26), the maximum postfault active power is synchronous angle href and PCC voltage angle hpcc.
positive to the terminal grid voltage Vg and SCR, so the balance The results of Fig. 5 can be understood together with that of
between Pfault fault
ref and Pout is more difficult to realize with lower Fig. 6, where state trajectories with different droop coefficient
SCR or grid voltage. On the other hand, decreasing Pfault ref will are depicted. It is shown that the position of convergence point
definitely help to the realization of active power balance when will not be influenced by droop coefficient, while the dynamic
the maximum postfault active power is limited. error between href and hpcc during resynchronization is changed.
These similar conclusions obtained from Fig. 4 and (26) In fact, the function of VOC is to force the PCC voltage angle
prove the correctness of the reduced-order large-signal in (25). trace to the synchronous angle output by APC, so the dynamic
Except the description for the balanced relationship of active error between href and hpcc will be enlarged as the bandwidth of
power in steady state, the dynamic of inner voltage control loop APC is increased or the bandwidth of VOC is decreased.
(i.e., VOC) during grid synchronization is also considered in Based on the results above, resynchronization process of a
(25), which is discussed in the next. droop-controlled GFM-DFIG will exhibit characteristics of a
2) Impacts of VOC: During the resynchronization process of second-order system due to the existence of VOC, the damping
GFM-DFIG when grid voltage sags, VOC also plays a critical of which can be enhanced with stronger VOC parameters or
part since it serves as a mediate link to control the angle of PCC weaker P-f droop gain. Besides, inevitable dynamic errors exist
voltage, so the impacts of VOC parameter on transient stability between the synchronous angle and the PCC voltage angle. The
are analyzed. overshoot of synchronous angle href is caused by the delayed
The precondition of Fig. 5 is as same as that of Fig. 4(a), so tracking process of PCC voltage angle to the synchronous angle,
the existence of postfault EP is guaranteed in theory. As shown which is undesirable for ensuring the power angle stability of
in Fig. 5(a), the convergency speed of trajectory is accelerated system. To better describe those conclusions obtained in this
by increasing the proportional gain Kp,q. Besides, all trajectories section, a conclusive table is given as follows. In Table I, risks
in Fig. 5(a) are convergent to a same point A1, proving that the of transient instability in GFM-DFIG are classified into two

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
LI et al.: TRANSIENT STABILITY OF GRID-FORMING DFIG 15887

TABLE I
IMPACTS OF DIFFERENT FACTORS ON TRANSIENT STABILITY

Risks of Transient Instability Factors Correlation

in GFM-DFIG
1) Inexistence of postfault Depth of grid Positive
EP voltage sag
SCR Negative
Postfault active Positive
power command
Fig. 8. VI-based GFM control.
2) Excessive dynamic error Proportional Negative
between href and hpcc gain of VOC
Integral gain of Negative where J and D are inertia coefficient and damping coefficient.
VOC Then, (27) can be rewritten as
P-f droop gain Positive dx
J ¼ ðPref  Pout Þ þ Dðx0  xÞ: (28)
dt
As for HSC, its synchronization equation can be expressed as
ð 
Pref  Pout
href ¼ þ KP,VBS Vpcc sinðhpcc  href Þ þ x0 :
Js þ D
(29)
Comparing with (28), (29) can be also rewritten as
dx
J ¼ ½Pref  DKP,VBS Vpcc sinðhref  hpcc Þ Pout
Fig. 7. Basic topology of HSC. dt |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
Pref ,eq

þ ½JKI,VBS Vpcc cosðhref  hpcc Þ þ D ðx0  xÞ: (30)

types, and term “correlation” denotes the relationship between |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
certain factor and the risk of transient instability. Deq

It is concluded that the equivalent active power instruction

IV. ENHANCING TRANSIENT STABILITY OF GFM-DFIG Pref,eq is decreased and the equivalent damping Deq is increased
BASED ON HYBRID SYNCHRONIZATION CONTROL when HSC is applied in APC. The first conclusion helps to pro-
A. HSC and Its Benefits on Improving Transient vide the convergency point when grid voltage sags, and the
Stability second conclusion helps to slow down the divergency of syn-
chronous angle href, so the transient stability is enhanced.
The basic topology of HSC is introduced firstly, which can be
treated as special controller in APC. As shown in Fig. 7, HSC
consists of two parts: power-based synchronization (PBS) and B. Enhancing HSC Based on Adaptive Integrator
voltage-based synchronization (VBS). PBS can be general con- Although the conclusions of (30) have been verified in [31],
trollers used by GFM control, such as P-f droop control and the application of the HSC method is still limited. First, it is
VSG control. The function of VBS is same as a PLL, while its assumed that a positive error between href and hpcc exists, so the
output frequency x2 is added to the frequency x1 output by decreased equivalent active power command and the increased
PBS. The synchronous angle href is obtained by integrating the equivalent damping can be realized. It means virtual impedance
superposition of x1, x2, and the fixed feedforward frequency x0,
(VI) control must be used before the inner voltage loop, the typi-
which serves as the orientation angle of SRF in GFM control.
cal structure of which is given in Fig. 8.
As for the advantages of HSC, on the one hand, HSC can be
In Fig. 9, subgraphs (a) and (c) denote two different steady
treated as a kind of synchronization technique to improve the
states of GFM control, and subgraph (b) denotes an intermediate
small-signal stability of GFL-REG under weak grid [30]. On the
state between (a) and (c). As shown in Fig. 9(a) and 9(c), the
other hand, the added VBS serves as a damper winding during
SRF oriented to synchronous angle href is marked with red line,
the resynchronization of GFM-REG when grid voltage sags, so
the d-axis of which is align with the PCC voltage vector of
the transient stability can be improved [31].
GFM-DFIG (i.e., Vpcc2) in steady state. Besides, there always
The mechanism of improving the transient stability of GFM-
exists a nonzero error between the PCC voltage vector con-
REG based on HSC can be explained as follows.
trolled by VI-based GFM strategy (i.e., Vpcc1) and the synchro-
First, the synchronization equation of a VSG-based GFM-
nous angle href, which provides the basis for realizing the
REG is given as
ð function of HSC method as (30). However, VI control increases
href ¼ ½ðPref  Pout Þ=ðJs þ DÞ þ x0  (27) the equivalent impedance of GFM-based REG, so the equivalent
SCR is decreased, which is disadvantage for transient stability

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
 15888 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 12, DECEMBER 2024

Fig. 9. Vector diagram: (a) steady state 1; (b) transient process; and
(c) steady state 2.
Fig. 10. Basic topology of EHSC.

in view of the conclusion in Section III-B. Besides, a steady-

state control error will exist between synchronous angle href and
PCC voltage angle hpcc, which cannot be limited effectively Given that the change of hpcc lags behind that of href due to
when grid voltage sags. When applying HSC method to enhance the hysteresis characteristic of VOC, so the integral term DPeq,2
transient stability of a GFM-DFIG system as shown in Fig. 1, it is positive, which contributes to the decrease of equivalent
will be found that this premise of nonzero angle error does not active power command even though href equals to hpcc finally.
exist due to VOC, who always force PCC voltage vector (i.e., Assuming href equals to hpcc, (32) can be rewritten as
Vpcc2) to align with the d-axis of SRF. Therefore, modification ð
must be considered when HSC is applied in a GFM-DFIG with Dx ¼ KP ðPfault  P fault
Þ þ ½KI ,VBS Vpcc sinðhpcc  href Þ :
|fflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl}
ref out
PI-based VOC. x1 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl
ffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
Although the steady-state nonzero error between synchronous x2

angle href and PCC voltage angle hpcc does not exist for GFM- (33)
DFIG with PI-based VOC, it can be found from Fig. 9(b) that a Usually, the active power command is calculated based on
dynamic angle error exists due to the hysteresis characteristic of the real-time torque command determined by maximum power
VOC. Therefore, the enhanced HSC method can be proposed as point tracking (MPPT) control, and the maximum active power
shown in Fig. 10, which utilizes the dynamic error between syn- transmission capability of DFIG is limited by its rated current,
chronous angle href and PCC voltage angle hpcc to enhance the SCR and grid voltage [32]. Thus, the active power unbalance
transient stability of GFM-DFIG. would exist between its command and feedback when grid volt-
As shown in Fig. 10, there are two modifications comparing age sags, which generates a positive angular velocity x1 acceler-
with the original HSC method in Fig. 7. The first modification is ating the divergence of href and inducing synchronous
a low-pass filter (LPF) after coordinate transformation, and its instability, and this process can be understood with the time-
function is to eliminate the impact of voltage jump caused by domain curves in Fig. 11(a). Now, this positive angular velocity
the electromagnetic process in DFIG. The second modification x1 can be offset by the negative integral term x2 provided by
is the adaptive integral gain of VBS, and it is given as DPeq,2. As shown in (33), Dx can equal to 0 by compensating
KI,VBS ¼ KI,0 ½1 þ Kt ð1  Vpcc =VN Þ (31) x2 when the active power unbalance exists, and this process can
be demonstrated by the time-domain curves in Fig. 11(b).
where KI,0 is the fiducial value of integration coefficient and Kt Besides, x2 is calculated automatically based on the dynamic
is a variable gain related to active power instruction and SCR. variation of q-axis PCC voltage, and the adaptive integral gain
The principles of proposed EHSC method can be explained KI,VBS is regulated with the change of grid voltage. It also pro-
as follows. First, the synchronization equation of droop-based vides additional degrees of freedom for regulating x2 (i.e., Kt).
APC with EHSC method is given as Given that the active power unbalance will be aggravated at
Dx high wind speed and low SCR grid, Kt can be further designed
¼ Pfault
ref  ðKP,VBS Vpcc =KP Þsinðhref  hpcc Þ as (34) to enhance its transient stability under different scenarios
KP |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
DPeq,1
Kt ¼ Kt0 ½1 þ kSCR 1 ePref =PN
fault
ð  (34)
 ½ðKI,VBS Vpcc =KP Þsinðhref  hpcc Þ Pfault out (32) where Kt0 is the fiducial value of Kt.
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
DPeq,2 C. Parameter Design of Adaptive Integral Gain
According to (32), there are two terms (i.e., DPeq,1 and It should be noted that the task of providing frequency com-
DPeq,2) after the postfault active power command. DPeq,1 is pensation is time-starved because synchronous angle href will
related to the real-time deviation between synchronous angle exceed its stable margin if a positive Dx exists for a long time,
href and PCC voltage angle hpcc, which equals to zero in steady which means frequency compensation x2 should be provided
state. DPeq,2 is related to the cumulative deviation between correctly in a short time. Usually, area “href > p/2” denotes a
them, so it is a nonzero term helping to adjust the equivalent dangerous signal of transient instability, so a permissive accel-
active power command. eration time Tþ exists when neglecting the compensation term

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
LI et al.: TRANSIENT STABILITY OF GRID-FORMING DFIG 15889

Fig. 12. Transient trajectories under grid voltage sag: (a) increasing
KP,VBS; and (b) effect of EHSC on transient stability.

where p denotes the depth of grid voltage sag.

When the needed compensation x2 is fixed, it can be seen
from (37) and (38) that a larger KI,VBS is needed under a smaller
jVpcc,q
eq
j, so the parameters KI,0 and Kt0 can be chosen on the basis
of different estimated jVpcc,q
eq
j.
Parameter KI,0 is the fiducial value of KI,VBS when adaptive
gain is not considered. Setting jVpcc,q
eq
j ¼ VN =3, KI,0 can be cho-
sen based on
KI,0 ðVN =3ÞTþmin ¼ KP PN  KP ðPfault
ref  Pout Þ:
fault
(40)
Kt0 is an adjustable parameter influencing adaptive gain.
Fig. 11. Time-domain curves of href and Dx: (a) without EHSC method; Setting jVpcc,q
eq
j ¼ VN =6, Kt0 can be chosen based on
and (b) with EHSC method.    
PN 0 VN KP
1 þ Kt0 1 þ ð1Þ1 ePN 1   Tþmin ¼ PN :
x2, that is, VN 6 KI,0
(41)
hmax  h0 p=2  h0
Tþ ¼ ¼ (35)
x1 KP ðPfault
ref  Pout Þ
fault
D. Effectiveness of EHSC for GFM-DFIG
where hmax ¼ p/2; h0 is the prefault synchronous angle. To verify the effectiveness of EHSC method, the large-signal
According to (35), Tþ is an uncertain time changing with dif- reduced-order model should be modified, so its state trajectory
ferent working conditions, and a shortest Tþmin exists with the can be obtained. The main modification located at the equation
prefault working condition “Pout ¼ PN”, that is, of APC, and it is rewritten as
 
Tþmin ¼ ðp=2  hN Þ=ðKP PN Þ (36) Lm
h_ ref ¼ KP Pref  1:5k0 Vg f2 ðhref , xpcc,q Þ þ
Tþmin denotes the shortest time for synchronous angle to cross its Ls
  
stable margin when no limitation of synchronous angular frequency Vpcc
KP,VBS  x_ pcc,q þ KI,0 1 þ Kt 1   xpcc,q :
exists. To ensure the transient stability of GFM-DFIG when the VN
postfault EP does not exist, (37) should be satisfied as follows: (42)
ð t0 þDt
The rest of the large-signal reduced-order model are as same
x1 þ ½KI,VBS Vpcc sinðhpcc  href Þ ¼ 0, Dt < Tþmin : (37)
t0
as (25). By setting the initial state of (42), the transient dynamic
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} of GFM-DFIG with EHSC method can be investigated.
x2
1) Impacts of Proportional Gain in EHSC: Considering the
Since “Vpccsin (hpcc  href)” denotes the q-axis component of case where only proportional gain exists in EHSC (i.e., KI,VBS ¼
PCC voltage, which is a variable varying with time. Usually, 0), the EHSC method can be treated as HSC method. Assuming
when the synchronous angle satisfies href < p/2, the dynamic that the prefault active power is 0.6 p.u. and the grid voltage
angle between hpcc and href is still under control. Assuming the dips to from 1 to 0.7 p.u. symmetrically, the impact of propor-
eq
equivalent value of Vpcc,q is Vpcc,q , there is tional gain KP,VBS on the resynchronization of GFM-DFIG are
ð t0 þDt investigated.
½Vpcc sinðhpcc  href Þ ¼ Vpcc,q
eq
Dt: (38) As shown in Fig. 12(a), the swing amplitude of state trajectory
t0 decreases when enlarging proportional gain KP,VBS, while the
Before synchronous angle crosses its stability threshold, final convergency point of state trajectory remain unchanged,
Vpcc,q satisfies the following: which suggests that the postfault EP cannot be changed by HSC,
and the enhanced transient damping can still be realized in view
pVN < Vpcc,q
eq
0 (39)
of the dynamic error between href and hpcc.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
 15890 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 12, DECEMBER 2024

Fig. 13. Transient stability boundary: (a) Pout ¼ 0.3 p.u.; and (b) Pout ¼
0.9 p.u. Fig. 14. Experimental platform.

2) Impacts of Adaptive Integral Gain in EHSC: Since the

transient stability is related to the postfault EP, the impact of TABLE II
adaptive integral gain KI,VBS on providing postfault EP is inves- SYSTEM AND CONTROL PARAMETERS
tigated first. In Fig. 12(b), EP0 is the convergency point of state
trajectory when Vg drops from 1 to 0.7 p.u. (green line). How- Symbol Quantity Value
ever, the state trajectory is divergent when Vg drops to 0.5 p.u. Vg Rated voltage amplitude (rms) 200 V
(red line), which is caused by the absence of postfault EP. When Vdc DC-link voltage 300 V
SN Rated power 5 kVA
the adaptive integral gain is applied, the divergent state trajec- Rs Stator resistance 0.7 X
tory converges to a new EP (i.e., EP1), which means system is Rr Rotor resistance 0.59 X
stable again. Lls Stator leakage inductance 2.1 mH
Lm Mutual inductance 72.6 mH
For clearer presentation of the enhanced transient stability, Llr Rotor leakage inductance 4.1 mH
the stability boundary of GFM-DFIG is calculated based on the Lg Line inductance 4 mL/24 mH
large-signal reduced-order model, and two power points are set
(i.e., Pout ¼ 0.3 p.u., Pout ¼ 0.9 p.u.). It is noted that 8100 points
are calculated since 90 points of Vg are chosen from 0.1 to 1 p.u.
and 90 points of SCR are chosen from 1 to 10.
As shown in Fig. 13, the green area denotes the stable area of
GFM-DFIG with no strategy applied. The blue and red areas
denote the additional stable area extended by HSC and EHSC
strategies. In general, the stable area of three cases (i.e., without
strategy, with HSC strategy and with EHSC strategy) are incre-
mental, which verified the advantage of EHSC method on
improving transient stability. Besides, more details can be read
from the comparison between Fig. 13(a) and 13(b). First, the
unstable area is larger at high active power condition, which cor-
Fig. 15. Regulation of PCC voltage realized by AVR.
responds to a high wind speed condition. This result is caused
by the enlarged imbalance between the active power instruction
(calculated from the input torque command) and the real active
power output of DFIG, which is restricted by SCR and the grid A. Transient Switching Control of GFM-DFIG
voltage level. Second, the contribution of HSC is weak at low
Figure 15 shows the function of PCC voltage control when
active power condition, which corresponds to a low wind speed
the active power reference Pref steps from 0 to 0.7 p.u., and the
condition. Actually, for the state trajectory in this condition, the
line inductance Lg is 24 mL (i.e., SCR ¼ 1.06). It can be seen
distance that the system moves from initial point to the postfault
that the PCC voltage will slowly drop with the increasement of
EP is relatively short, so the impact of transient damping pro-
active power Pe, due to the considerable impedance voltage
vided by HSC is not obvious.
drop in low SCR grid. However, the PCC voltage can return to
nominal value finally for the function of PCC voltage control.
V. EXPERIMENTAL VALIDATION Figure 16 shows the process where GFM-DFIG switches
To verify the correctness of the large-signal modeling and from PCC voltage control to reactive-current control under grid
the effectiveness of the proposed strategy, a 5-kW DFIG-based voltage fault. The prefault active power is 0.7 p.u., and the line
WT platform is established as shown in Fig. 14. The control inductance Lg is 4 mL (i.e., SCR ¼ 6.36). It can be seen that the
algorithm is realized based on TI TMS320f28335 DSP, and the reactive rotor current of DFIG is limited to its reference when
symmetrical grid voltage fault is simulated by a set of three grid voltage drops from 1 to 0.5 p.u., and the system is stable
phase regulator. Detailed experimental parameters are given during the grid voltage fault. It is noted that the result of Fig. 15
in Table II. will serves as a control group for Figs. 16 and 17.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
LI et al.: TRANSIENT STABILITY OF GRID-FORMING DFIG 15891

Fig. 16. Reactive-current-controlled mode under grid voltage sag. Fig. 19. Transient stability when Pref decreases from 0.7 to 0.35 p.u.

Fig. 17. Transient instability when Vg drops from 1 to 0.2 p.u. Fig. 20. Dynamic error between href and hpcc: (a) increasing Kp,q; and
(b) increasing Ki,q.

Fig. 18. Transient instability when Lg increases from 4 to 24 mL.

Fig. 21. Dynamic error influenced by KP.

B. Transient Stability Problem of GFM-DFIG However, this dynamic angle error can never be eliminated even
Figure 17 shows the transient instability of GFM-DFIG due to pushing the VOC parameter to its limitation. In fact, the PCC
the increasing depth of grid voltage sag. Comparing with Fig. 16, voltage angle is changed by the flowing current on the line
the only change is that the postfault grid voltage drops from 0.5 impedance, and it is easier to be changed in low SCR grid. Con-
to 0.2 p.u. This instability is also shown in Fig. 18, comparing sidering the rate of change of PCC voltage angle is limited, it
with Fig. 16, the line inductance increases from 4 to 24 mL. Both cannot track to its command href as fast as it is desired. Besides,
the transient instability of Figs. 17 and 18 are manifested in the the flowing current on the line impedance is controlled by the
enlarged active power imbalance and the uncontrolled synchro- regulation of this dynamic angle error realized in VOC, so the
nous angle href, which suggests that a lower SCR and a severer inner voltage loop cannot be neglected in the large-signal model
voltage fault will induce the transient instability. of GFM-DFIG. This dynamic angle error will also be influenced
As a contrast to Fig. 18, the result of Fig. 19 is stable and the by the P-f droop gain. In Fig. 21, the dynamic angle error is
active power reference reduces to half the original, suggesting enlarged when increasing KP from 0.7 to 1.5 p.u. Specially, the
that GFM-DFIG is tend to be stable under a lower active power system will be unstable due to the overlarge dynamic angle error
reference when grid voltage dips. when KP is 1.5 p.u.
The impacts of VOC parameters and P-f droop gain are given
as Figs. 20 and 21, where the dynamic error between href and C. Comparison of HSC and EHSC Methods
hpcc is observed during the active power synchronization (Pref ¼ The result of Fig. 18 is used here to illustrate the function of HSC
0.7 p.u., Lg ¼ 24 mL). As shown in Fig. 20(a) and 20(b), the method and EHSC method, where Pref ¼ 0.7 p.u., Lg ¼ 24 mL,
dynamic angle error will decrease with larger Kpq or Kiq. and the grid voltage drops from 1 to 0.5 p.u.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
 15892 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 12, DECEMBER 2024

Fig. 22. Effect of HSC method: Vg ¼ 0.5 p.u., Pref ¼ 0.7 p.u., Lg ¼ 24 mL.

Fig. 25. Effect of EHSC method (Vg ¼ 0.2 p.u., Pref ¼ 0.7 p.u., Lg ¼
24 mL): (a) results of active power and synchronous angle. (b) result of
PCC voltage.

As shown in Fig. 23, the synchronous angle and active power

converge to a new postfault state when EHSC method is applied.
Contributed by the feedforward frequency provided by the adap-
tive integrator, the droop-controlled APC can allow a fixed
Fig. 23. Effect of EHSC method: Vg ¼ 0.5 p.u., Pref ¼ 0.7 p.u., Lg ¼ imbalance between the active power feedback and its reference,
24 mL. preventing further divergence of the power angle.
The effectiveness of EHSC method is also validated under
different working conditions, both of which are done under the
terminal grid voltage of 0.2 p.u. In Fig. 24, synchronous angle is
restricted, and the PCC voltage amplitude can be maintained to
a stable value (0.2 p.u.) with EHSC method. Comparing with
the divergent synchronous angle in Fig. 24 (i.e., the red line),
the divergent speed of synchronous angle in Fig. 25 is faster,
which verifies the negative relationship between SCR and the
risk of transient instability. Fig. 25 corresponds to a terrible
working condition, where EHSC method still works well to
ensure the stability of GFM-DFIG, which proves its superiority.
Besides, the PCC voltage amplitude can be maintained to a
higher value due to the contribution of reactive-current support
under high-grid impedance condition.

Fig. 24. Effect of EHSC method (Vg ¼ 0.2 p.u., Pref ¼ 0.7 p.u., Lg ¼ VI. CONCLUSION
4 mL): (a) results of active power and synchronous angle and (b) result of
PCC voltage. To investigate the transient behavior of the DFIG-based WT
with GFM control under a symmetrical grid voltage fault, the
In Figs. 22 and 23, the uncontrolled synchronous angle reduced-order large-signal model is established in this article,
depicted by red line and the imbalanced active power depicted through which two types of transient instability for GFM-DFIG
by purple line are the unstable results. The results of angle refer- is revealed. The first is induced by the inexistence of postfault
ence and active power with HSC/EHSC method applied are EP. The second is induced by the excessive dynamic deviation
denoted by blue and green line, respectively. between synchronous angle and PCC voltage angle. An EHSC
In Fig. 22, the synchronous angle and active power are still method based on adaptive integrator is proposed to improve its
divergent when HSC method is applied, but the divergence transient stability during a symmetrical grid voltage fault. The
speed is slowed down. In the previous analysis, when HSC main conclusions can be summarized as follows.
method is applied on a GFM-DFIG with VOC, it cannot provide 1) The transient behavior of GFM-DFIG conforms to the
a new postfault EP if the original postfault EP does not exist, general rules of GFM control. During a symmetrical grid
while its only contribution on enhancing transient stability is voltage sag, the active power imbalance will be enlarged
improving the transient damping. This conclusion can be well with lower SCR and grid voltage, or higher active power
validated by the result of Fig. 22. instruction, causing the absence of postfault EP.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
LI et al.: TRANSIENT STABILITY OF GRID-FORMING DFIG 15893

2) Differing with GFM-VSC, whose inner voltage dynamic [16] D. Pan, X. Wang, F. Liu, and R. Shi, “Transient stability of voltage-
can be neglected during transient. A GFM-DFIG using source converters with grid-forming control: A design-oriented study,”
IEEE J. Emerg. Sel. Topics Power Electron., vol. 8, no. 2,
P-f droop controller exhibits underdamped feature due to pp. 1019–1033, Jun. 2020.
the lag characteristic introduced by VOC, which is not [17] J. Lei, X. Xiang, B. Liu, W. Li, and X. He, “Transient stability analysis
consistent with the overdamped P-f droop control. of grid forming converters based on damping energy visualization and
geometry approximation,” IEEE Trans. Ind. Electron., vol. 71, no. 3,
3) The HSC method can enhance the transient stability of pp. 2510–2521, Mar. 2024.
GFM control by improving the transient damping and [18] X. Fu et al., “Large-signal stability of grid-forming and grid-following
the postfault EP. However, it is unsuitable for a GFM- controls in voltage source converter: A comparative study,” IEEE
Trans. Power Electron., vol. 36, no. 7, pp. 7832–7840, Jul. 2021.
DFIG, whose VOC forces PCC voltage angle tracking [19] P. Ge, C. Tu, F. Xiao, Q. Guo, and J. Gao, “Design-oriented analysis and
its angle command. The proposed EHSC method can transient stability enhancement control for a virtual synchronous generator,”
resolve this problem by introducing a compensation fre- IEEE Trans. Ind. Electron., vol. 70, no. 3, pp. 2675–2684, Mar. 2023.
[20] X. He, S. Pan, and H. Geng, “Transient stability of hybrid power
quency to its synchronous angle, which is adaptive to systems dominated by different types of grid-forming devices,” IEEE
grid voltage, SCR and postfault active power instruction. Trans. Energy Convers., vol. 37, no. 2, pp. 868–879, Jun. 2022.
[21] X. Xiong, C. Wu, P. Cheng, and F. Blaabjerg, “An optimal damping
design of virtual synchronous generators for transient stability
REFERENCES enhancement,” IEEE Trans. Power Electron., vol. 36, no. 10,
pp. 11026–11030, Oct. 2021.
[1] Q. Peng, Q. Jiang, Y. Yang, T. Liu, H. Wang, and F. Blaabjerg, “On
[22] M. Chen, D. Zhou, and F. Blaabjerg, “Enhanced transient angle stability
the stability of power electronics-dominated systems: Challenges
control of grid-forming converter based on virtual synchronous generator,”
and potential solutions,” IEEE Trans. Ind. Appl., vol. 55, no. 6,
IEEE Trans. Ind. Electron., vol. 69, no. 9, pp. 9133–9144, Sep. 2022.
pp. 7657–7670, Nov. 2019.
[23] H. Wu and X. Wang, “A mode-adaptive power-angle control method
[2] J. Ma, Y. Shen, and A. G. Phadke, “DFIG active damping control
for transient stability enhancement of virtual synchronous generators,”
strategy based on remodeling of multiple energy branches,” IEEE
IEEE J. Emerg. Sel. Topics Power Electron., vol. 8, no. 2,
Trans. Power Electron., vol. 36, no. 4, pp. 4169–4186, Apr. 2021.
pp. 1034–1049, Jun. 2020.
[3] Y. Zhou, D. Zhu, X. Zou, C. He, J. Hu, and Y. Kang, “Adaptive
[24] E. Rokrok, T. Qoria, A. Bruyere, B. Francois, and X. Guillaud, “Transient
temporary frequency support for DFIG-based wind turbines,” IEEE
stability assessment and enhancement of grid-forming converters
Trans. Energy Convers., vol. 38, no. 3, pp. 1937–1949, Sep. 2023.
embedding current reference saturation as current limiting strategy,” IEEE
[4] Z. Dong, Z. Li, L. Du, Y. Liu, and Z. Ding, “Coordination strategy of
Trans. Power Syst., vol. 37, no. 2, pp. 1519–1531, Mar. 2022.
large-scale DFIG-based wind farm for voltage support with high
[25] S. P. Me et al., “Transient stability analysis of virtual synchronous
converter capacity utilization,” IEEE Trans. Sustain. Energy, vol. 12,
generator equipped with quadrature-prioritized current limiter,” IEEE
no. 2, pp. 1416–1425, Apr. 2021.
Trans. Power Electron., vol. 38, no. 9, pp. 10547–10553, Sep. 2023.
[5] L. Huang, H. Xin, L. Zhang, Z. Wang, K. Wu, and H. Wang,
[26] Z. Li, Z. Xie, and X. Zhang, “An improved strategy of grid-forming
“Synchronization and frequency regulation of DFIG-based wind turbine
DFIG based on disturbance rejection stator flux control,” IEEE Trans.
generators with synchronized control,” IEEE Trans. Energy Convers.,
Ind. Electron., vol. 71, no. 3, pp. 2498–2509, Mar. 2024.
vol. 32, no. 3, pp. 1251–1262, Sep. 2017.
[27] S. Wang and L. Shang, “Fault ride through strategy of virtual-
[6] H. Nian and Y. Jiao, “Improved virtual synchronous generator control
synchronous-controlled DFIG-based wind turbines under symmetrical
of DFIG to ride-through symmetrical voltage fault,” IEEE Trans.
grid faults,” IEEE Trans. Energy Convers., vol. 35, no. 3,
Energy Convers., vol. 35, no. 2, pp. 672–683, Jun. 2020.
pp. 1360–1371, Sep. 2020.
[7] “Voluntary specification for grid-forming inverters,” AEMO Tech.
[28] I. Oraa, J. Samanes, J. Lopez, and E. Gubia, “Single-loop droop control
Rep., 2023. [Online]. Available: https://aemo.com.au/-/media/files/
strategy for a grid-connected DFIG wind turbine,” IEEE Trans. Ind.
initiatives/primary-frequency-response/2023/gfm-voluntary-spec.pdf
Electron., vol. 71, no. 8, pp. 8819–8830, Aug. 2024.
[8] W. Wu et al., “Sequence impedance modeling and stability comparative
[29] R. Rosso, X. Wang, M. Liserre, X. Lu, and S. Engelken, “Grid-forming
analysis of voltage-controlled VSGs and current-controlled VSGs,”
converters: Control approaches, grid-synchronization, and future
IEEE Trans. Ind. Electron., vol. 66, no. 8, pp. 6460–6472, Aug. 2019.
trends—A review,” IEEE Open J. Ind. Appl., vol. 2, pp. 93–109, 2021.
[9] X. Zhang, D. Xia, Z. Fu, G. Wang, and D. Xu, “An improved
[30] L. Harnefors, J. Kukkola, M. Routimo, M. Hinkkanen, and X. Wang,
feedforward control method considering PLL dynamics to improve
“A universal controller for grid-connected voltage-source converters,”
weak grid stability of grid-connected inverters,” IEEE Trans. Ind. Appl.,
IEEE J. Emerg. Sel. Topics Power Electron., vol. 9, no. 5,
vol. 54, no. 5, pp. 5143–5151, Sep. 2018.
pp. 5761–5770, Oct. 2021.
[10] M. G. Taul, X. Wang, P. Davari, and F. Blaabjerg, “Current limiting
[31] T. Liu and X. Wang, “Physical insight into hybrid-synchronization-
control with enhanced dynamics of grid-forming converters during fault
controlled grid-forming inverters under large disturbances,” IEEE Trans.
conditions,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 8, no. 2,
Power Electron., vol. 37, no. 10, pp. 11475–11480, Oct. 2022.
pp. 1062–1073, Jun. 2020.
[32] X. Guo et al., “Analysis and enhancement of active power transfer
[11] K. Shi, W. Song, P. Xu, R. Liu, Z. Fang, and Y. Ji, “Low-voltage ride-
capability for DFIG-based WTs in very weak grid,” IEEE J. Emerg. Sel.
through control strategy for a virtual synchronous generator based on
Topics Power Electron., vol. 10, no. 4, pp. 3895–3906, Aug. 2022.
smooth switching,” IEEE Access, vol. 6, pp. 2703–2711, 2018.
[12] Z. Shuai, W. Huang, C. Shen, J. Ge, and Z. J. Shen, “Characteristics
and restraining method of fast transient inrush fault currents in
synchronverters,” IEEE Trans. Ind. Electron., vol. 64, no. 9,
pp. 7487–7497, Sep. 2017.
[13] J. Hu, B. Wang, W. Wang, H. Tang, Y. Chi, and Q. Hu, “Small signal
dynamics of DFIG-based wind turbines during riding through
symmetrical faults in weak AC grid,” IEEE Trans. Energy Convers., Zhe Li received the B.S degree in electrical
vol. 32, no. 2, pp. 720–730, Jun. 2017. engineering and automation in 2019 from Hefei
[14] Y. Ma, D. Zhu, Z. Zhang, X. Zou, J. Hu, and Y. Kang, “Modeling and University of Technology, Hefei, China, where
transient stability analysis for type-3 wind turbines using singular he is currently working toward the Ph.D. degree
perturbation and lyapunov methods,” IEEE Trans. Ind. Electron., in electrical engineering.
vol. 70, no. 8, pp. 8075–8086, Aug. 2023. His research interests include modeling and
[15] P. Sun, J. Yao, R. Liu, J. Pei, H. Zhang, and Y. Liu, “Virtual control of wind turbine generation system,
capacitance control for improving dynamic stability of the DFIG-based renewable energy generation, and the power
wind turbines during a symmetrical fault in a weak ac grid,” IEEE electronics application in power systems.
Trans. Ind. Electron., vol. 68, no. 1, pp. 333–346, Jan. 2021.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.
 15894 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 12, DECEMBER 2024

Zhen Xie (Member, IEEE) was born in Anhui, Xing Zhang (Senior Member, IEEE) was born
China, in 1976. He received the Ph.D. degree in in Shanghai, China, in 1963. He received the
electrical engineering from Hefei University of B.S., M.S., and Ph.D. degrees in electrical engi-
Technology, Hefei, China, in 2005. neering and automation from Hefei University of
From 2005 to 2008, he has been a Lecturer Technology, Hefei, China, in 1984, 1990, and
with the School of Electrical Engineering and 2003, respectively.
Automation, Hefei University of Technology, Since 1984, he has been a Faculty Member
where he is currently a Professor. His current with the School of Electric Engineering and
research interests include the control of doubly Automation, Hefei University of Technology,
fed induction generators for wind power genera- where he is currently a Professor. He is also
tion and renewable energy conversion systems. with the Photovoltaic Engineering Research
Center of the Ministry of Education. His research interests include photo-
voltaic generation technologies, wind power generation technologies,
and distributed generation systems.
Shang Xu received the B.S degree in electrical
engineering and automation in 2020 from Hefei
University of Technology, Hefei, China, where
he is currently working toward the Ph.D. degree
in electrical engineering.
His research interests include control and
modeling of wind power generation systems,
system stability analysis, renewable energy gen-
eration, and motor control with power electronics
devices.

Authorized licensed use limited to: State Grid Corporation of China (SGCC). Downloaded on November 27,2024 at 08:50:38 UTC from IEEE Xplore. Restrictions apply.