7832 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 36, NO.

7, JULY 2021

Large-Signal Stability of Grid-Forming and

Grid-Following Controls in Voltage Source
Converter: A Comparative Study
Xikun Fu , Jianjun Sun , Member, IEEE, Meng Huang , Member, IEEE, Zhen Tian , Member, IEEE,
Han Yan , Herbert Ho-Ching Iu , Senior Member, IEEE, Pan Hu , Member, IEEE,
and Xiaoming Zha , Member, IEEE

Abstract—The grid-forming and grid-following controls are So far, most of the VSCs are regarded as the controlled current
adopted in three-phase voltage source converters according to source, which follow the voltage and frequency at the point of
different grid conditions. However, the basic operation principle of common coupling (PCC). This control strategy is considered as
the two kinds of control is different and will lead to the instability
of the grid-connected system through various paths. In this article, the grid-following control, which is suitable for connecting to a
the energy function model of the grid-forming and grid-following strong grid [3]. However, it is difficult to guarantee the stability
controlled converters are established and compared in detail. The of the VSC when the grid impedance is very high. In order to
influence of system parameters is studied, and the large-signal improve the stability of the VSC under this condition, the VSC
stability boundaries are derived through a general method. More- is required to form the system voltage and frequency, which is
over, compared to the grid-forming controlled converter, the grid-
following converter will lose stability by exhibiting a varying damp- regarded as the grid-forming control. Unlike the grid-following
ing coefficient transient. The equivalent damping coefficient can be control, the grid-forming control is treated as controlled voltage
used as a criterion of whether the grid-forming/following control is sources, which can enhance synchronize stability in weak grid
suitable for the weak/strong grid condition. The simulation and case [4].
experiment results show the difference of the transient process
In fact, the two control modes suffer from stability issue in
and validate the control criterion for different grid strength under
large-signal disturbance. different conditions. Most of current researches are devoted
to the stability of the VSCs by using small-signal modeling
Index Terms—Energy function, grid-following control, grid- method or other linear theory, which is not suitable for the
forming control, La Salle’s invariance principle, large-signal
stability, voltage source converters (VSCs).
stability analysis under the large-signal disturbance [5], [6].
The linear methods indicate that the VSCs are stable when it
exists stable equilibrium point (SEP) after the disturbance [7],
I. INTRODUCTION
[8]. Nevertheless, a transient instability phenomenon may occur
S THE structure of energy production changes, the pro-
A portion of renewable energy source (RES) in the power
system is rapidly increasing [1], [2]. For the RES connected
even though SEP exists in some special cases [9], [10]. In [11],
[12], the method of impedance model shows the effect of the
system bandwidth to the stability of the VSC systems.
to the power grid, voltage source converters (VSCs) have been With the consideration of nonlinear characteristic, the equal
widely employed as grid interfaces. area criterion (EAC) is proposed to analyze the large-signal
stability of the VSCs [13], [14]. However, the EAC is not suitable
Manuscript received September 2, 2020; revised November 3, 2020; accepted for system containing negative damping. Phase portraits are
December 19, 2020. Date of publication December 25, 2020; date of current
version March 5, 2021. This work was supported in part by the Science and Tech- presented to solve the restriction which appears in the EAC [15],
nology Project of State Grid Corporation under Grant 52153218000B and in part [16]. In [17], they express the transient process of the VSCs by
by the National Natural Science Foundation of China under Grants 51877159, using a reduced order nonlinear model.
U1866601, and 51637007. Recommended for publication by Associate Editor
M. Su. (Corresponding author: Meng Huang.) Meanwhile, the methods in the time-domain directly show the
Xikun Fu, Jianjun Sun, Meng Huang, Zhen Tian, Han Yan, and Xiaoming Zha results of transient process, but solving the nonlinear differential
are with the School of Electrical Engineering and Automation, Wuhan Univer- equations of the system is a challenge [18], [19]. In addition, the
sity, Wuhan 430072, China (e-mail: [email protected]; [email protected];
meng.huang@whu. edu.cn; [email protected]; [email protected]; time domain method requires a lot of time and provides little
[email protected]). physical insight into the stability mechanism of the VSCs.
Herbert Ho-Ching Iu is with the School of Electrical, Electronic and Computer Because of the deficiency of current analysis methods, a
Engineering, The University of Western Australia, Crawley, WA 6009, Australia
(e-mail: [email protected]). method is urgently needed to analyze the physical mechanism
Pan Hu is with the State Grid Hubei Electric Power Co., Ltd. Electric Power of the VSCs when the system losing stability under large-signal
Research Institute, Wuhan 430077, China (e-mail: [email protected]). disturbance. To fill the research gap, the transient behavior of the
Color versions of one or more figures in this article are available at https:
//10.1109/TPEL.2020.3047480. VSC system is studied from the energy perspective. The energy
Digital Object Identifier 10.1109/TPEL.2020.3047480 functions of the systems can be got by comparing the energy

0885-8993 © 2020 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: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 FU et al.: LARGE-SIGNAL STABILITY OF GRID-FORMING AND GRID-FOLLOWING CONTROLS IN VOLTAGE SOURCE CONVERTER 7833

Fig. 1. Three-phase grid-connected VSC system, where its output current is within the current limit. (a) Grid-forming VSC. (b) Grid-following VSC.

function of the synchronous generator (SG) and the physical TABLE I

SYSTEM MAIN PARAMETERS
mechanism of the system losing stability is discussed.
In fact, negative damping equivalently increases the accel-
eration area in the EAC. When the system contains negative
damping, using the EAC will affect the judgment of system
stability, so system damping has to be considered. This article
obtains the distribution of system damping by constructing the
energy function, and then considers the large-signal stability of
the system including positive and negative damping in two parts.
Part 1: In order to give an analytic and conservative large-signal
stability boundary in the positive damping region, the La Salle’s
Invariance Principle is used. Part 2: For the purpose of analyz-
ing the large-signal stability of the system in the positive and
negative damping region, the indicator of equivalent damping is
proposed. When the equivalent damping is negative, the energy
of the system gradually increases to be greater than the energy at
the unstable equilibrium point (UEP), and eventually the system
loses stability. Based on (1) and (2),transforming
 
II. SYSTEM DESCRIPTION AND CONTROL PRINCIPLE 3 VPCC Vg
δ̈ = −ωp δ̇ + ωp Kp · Pref − sin δ . (3)
2 Xg
This article pays attention to typical VSCs with grid-following
control and grid-forming control. The topology is shown as
Fig. 1 and the parameters are listed in Table I. The VSCs are B. Grid-Following Control
connected to the power grid through two paralleled transmission Fig. 1(b) shows a common block diagram of the synchronous
lines Xg1 and Xg2 . The grid impedance Xg cannot be ignored reference frame-phase locked loop (PLL) [20]. The three-phase
when the VSC connected to the weak grid, as shown in Fig. 1. voltages VPCC are transformed into the dq0-frame, with the
q-axis voltage as the input of PI controller.
A. Grid-Forming Control Thereby, the output angle of the PLL can be expressed as
In order to reduce the noise in measured active power P, a    
low-pass filter (LPF) is added into the droop control, which is θPLL = ωn + (Kp + Ki )vPCCq . (4)
displayed in Fig. 1(a). The ω p named cut-off frequency is applied
in LPF. The PCC voltage is affected by the grid voltage, current, and
The active power P from the PCC can be derived as grid impedance. And, the PCC voltage equation in the q-frame
can be written as
3 VPCC Vg
P = sin δ. (1)
2 Xg vPCCq = VPCC sin(θg − θPLL ) + Ig Zg sin(θZ + ϕ) (5)
The phase difference δ can be obtained as where ϕ = tan−1 (iref ref
q /id )is the angle of the injected current
Kp ω p reference relative to the PLL phase angle. When the reference
δ= · (Pref − P ) . (2)
s s + ωp reactive current is 0, there is ϕ = 0.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 7834 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 36, NO. 7, JULY 2021

Fig. 2. Comparison of three dynamics nonlinear control diagram. (a) SG is connected to weak grid. (b) Grid-forming VSC is connected to weak grid. (c)
Grid-following VSC is connected to weak grid.

Therefore, combining (4) and (5), new expression of the The energy function of the SG is derived as follows [21]:
output angle could be rewritten as follows:
     3VPCC Vg 1
V (δ  , δ̇  ) = −Pm δ  − cos δ  + H δ̇ 2 + E0 (12)
θPLL = ω n + Kp + K i [VPCC sin(θg − θPLL ) 2Xg 2

 where E0 is a constant.
+Ig Zg sin(θZ )] . (6) Based on (11) and (12), the derivative of V (δ  , δ̇  )can be
written as [22]
Different angle between the grid phase angle and the output ∂V (δ  , δ̇  ) ˙ ∂V (δ  , δ̇  ) 
phase angle of the PLL is introduced as δ V̇ (δ  , δ̇  ) = δ + δ̈
∂δ  ∂ δ̇ 
δ = θPLL − θg . (7)  
3VPCC Vg
= −Pm + sin δ + H δ̈ δ˙ = −Dδ̇ 2 .
 
Substituting (7) into (6). Hence, the different angle δ can be 2Xg
expressed as (13)
    
To compare the difference between the VSCs and the SG, the
δ= Kp + K i [Ig Zg sin(θZ ) − Vg sin(δ)] . (8)
nonlinear differential (3), (10), and (11) are transformed into
Ignoring the grid resistance, the grid impedance can be written control block diagrams, as displayed in Fig. 2.
as
A. Energy Function Model of the Grid-Forming Control
Zg = (ωn + δ̇)Lg . (9)
As shown in Fig. 2(a) and (b), by comparing the VSC with
Based on (8) and (9), the second-derivative equation of δ can grid-forming control and the SG, reference active power P1ref ,
be obtained as actual active power P1 , the equivalent inertial time constant H1 ,
(1 − Kp Ig Lg )δ̈ and damping coefficient D1 in the VSC system can be obtained
  ⎧
Kp Vg cos δ ⎪
⎪ P1ref = Pref
= Ki · Ig ωn Lg + Ig δ̇Lg − Vg sin δ − ⎨
Ki
δ̇ . P1 = 3VPCC Vg /2Xg
(14)
⎪
⎪ H1 = 1/(ωp · KP )
(10) ⎩
D1 = 1/KP .
III. ENERGY FUNCTION MODELS FOR LARGE-SIGNAL And, (3) can be rewritten as follows:
STABILITY ANALYSIS
H1 δ̈ = P1ref − P1 sin δ − D1 · δ̇. (15)
In order to obtain the energy functions of the grid-connected
VSCs, the nonlinear characteristic equation of SG is introduced. As a result, the energy function of the control system can be
The equation of motion of the generator rotor, i.e., the swing presented as follows:
equation, can be expressed as
1
3VPCC Vg V (δ, δ̇) = −P1ref δ − P1 cos δ + H1 δ̇ 2 + E0 . (16)

H δ̈ = Pm − sin δ  − Dδ̇  (11) 2
2Xg
In the light of (15) and (16), the derivative of V (δ, δ̇)is
where δ  represents the power angle, H is the inertial time
constant of the SG, and D is the damping coefficient of the SG. V̇ (δ, δ̇) = −P1ref δ̇ + P1 sin δ · δ̇ + H1 δ̇ δ̈ = −D1 δ̇ 2 . (17)

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 FU et al.: LARGE-SIGNAL STABILITY OF GRID-FORMING AND GRID-FOLLOWING CONTROLS IN VOLTAGE SOURCE CONVERTER 7835

The equivalent potential energy Ep and kinetic energy Ek of

the control system are
Ep = −P1ref δ − P1 cos δ + E0 (18)
1
Ek = H1 δ̇ 2 . (19)
2

B. Energy Function Model of the Grid-Following Control

As displayed in Fig. 2(a) and (c), it can be found that the
nonlinear characteristics of the grid-following VSC is similar
Fig. 3. Schematic diagram of system damping division. (a) Grid-forming VSC.
with the SG. Further, the energy function of the control system (b) Grid-following VSC.
can be got. By comparing the two systems, reference active
power P2ref , actual active power P2 , the equivalent inertial time
constant H2 , and damping coefficient D2 in the VSC system can
be obtained.
⎧
⎪
⎪ P2ref = Ki Ig ωn Lg
⎨
P 2 = Ki V g
(20)
⎪
⎪ H2 = (1 − Kp Ig Lg )
⎩
D2 (δ) = −Ki Ig Lg + Kp Vg cos δ.
And, (10) can be rewritten as follows:
H2 δ̈ = P2ref − P2 sin δ − D2 (δ) · δ̇. (21)
As a result, the energy function of the control system can be
presented as follows:
1
V (δ, δ̇) = −P2ref δ − P2 cos δ + H2 δ̇ 2 + E0 . (22)
2
According to (21) and (22), the derivative of V (δ, δ̇)can be
expressed as
V̇ (δ, δ̇) = −P2ref δ̇ + P2 sin δ · δ̇ + H2 δ̇ δ̈ = −D2 (δ)δ̇ 2 . (23)
It should be noted that D2 (δ)>0 cannot to be guaranteed.
The equivalent potential energy Ep and kinetic energy Ek of
the control system are
Ep = −P2ref δ − P2 cos δ + E0 (24)
1 Fig. 4. Comparison chart of system characteristic parameters between the
Ek = H2 δ̇ 2 . (25) grid-following and the grid-forming VSCs (red line is increased impact and
2 blue line is reduced impact). (a) Grid-forming VSC. (b) Grid-following VSC.
From the EAC perspective, the effect of positive damping
increases the deceleration area, which will not affect the analysis
C. Parameter Influence
of system stability. However, the influence of negative damping
is to increase the system kinetic energy, which can be equivalent Although the grid-following and the grid-forming VSCs have
to increasing a part of the acceleration area. The acceleration similar nonlinear characteristics, there are still some differences
area will affect the analysis of system stability. Nevertheless, between the two controls. From the four system characteris-
it is related to the actual trajectory and cannot be calculated. tic parameters, the equivalent reference active power Pref , the
Therefore, the EAC is not suitable for system including negative equivalent actual active power P, the inertial time constant H,
damping. and the damping coefficient D, the difference can be explored.
By constructing the energy functions of the VSCs, the damp- Fig. 4(a) reveals that P is affected by the uncontrollable
ing distribution of the system can be obtained from (14) and parameters (Vpcc , Lg ). Its means that grid voltage fluctuations
(20). As shown in Fig. 3(a), the damping distribution of the will cause power fluctuations and produce large surge currents
grid-forming VSC is globally positive. As displayed in Fig. 3(b), in the grid-forming VSC. And, Fig. 4(b) shows that P is mainly
under the influence of phase difference δ, the grid-following affected by the control gain Ki and is not affected by the PCC
VSC contains positive and negative damping. Combine the voltage. It means that the power fluctuation is not appearing in
previous discussion about the EAC, the EAC cannot be used the grid-following VSC.
as a general method to analyze the large signal stability of the Fig. 4(a) demonstrates that D and H are only affected by the
two types of VSC. controllable parameters (Kp , ω p ) in the grid-forming VSC. It

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 7836 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 36, NO. 7, JULY 2021

indicates that the characteristics of grid-forming VSC will not

change with grid current or impedance. Meanwhile, Fig. 4(b)
indicates that the uncontrollable parameters (Ig , Lg , δ) will affect
D and H. Moreover, the conditions, which are D>0 and H>0,
must be satisfied. It means that the high grid current or grid
impedance cannot be satisfied and a suitable Kp and Ki need
to be selected in the grid-following VSC. Otherwise the SG-
like characteristics of the grid-following VSCs will change. It
reveals that the grid-forming VSC is more suitable than the grid-
following VSCs in the case of high impedance.
In addition, in the case of low grid impedance, the damping
D2 (δ) of the grid-following VSC is considered to be greater Fig. 5. Schematic diagram of the local La Salle’s invariance principle.
than 0. In other words, the system is a dissipated energy system.
When the grid impedance is high, the damping may be negative
damping. It can also explain that grid-following VSC is likely
to lose stability under high grid impedance.

IV. STABILITY BOUNDARIES

For large-signal stability analysis, it is generally assumed
that the system is operating in stable states before the fault,
and the initial stability power angle is δ 0 . A short circuit fault
occurs on a line of the paralleled transmission ones. When the
protection device has detected the fault, the fault is removed.
The whole transient process is divided into: 1) before fault, 2)
during fault, and 3) after fault. Generally, the condition (26) will
be established.
PI > PIII > PII . (26)
In view of the problem that the VSC systems contain neg-
ative damping, the system damping has to be considered. The
distribution of system damping is obtained by constructing the
energy function. In order to analyzing the large-signal stability
of the two types of VSCs in the positive damping region, a novel
and general method based on Lyapunov stability criterion and Fig. 6. Large-signal stability analysis about grid-forming VSCs.(a) Equivalent
actual physical system. (b) Stability boundary and trajectory in different states:
the La Salle’s Invariance Principle is proposed in the article. stable, critically stable, and unstable.

A. Application of Local La Salle’s Invariance Principle

The energy functions of the VSC systems have been selected Fig. 5. When the largest invariant set M moves out of the region
in the Section III, then the local Lyapunov function can be Ωl , the stability of VSCs cannot be judged under the restriction
found. According to the Local La Salle’s Invariance Principle, of local La Salle’s invariance principle.
the global energy function V(y) > 0 is not satisfied and the local
stability of the VSCs will be obtained [22]. The Principle will B. Large-Signal Stability Analysis of the
be introduced as follows: Grid-Forming Control

The (14) reveals that the system damping D1 is only affected
Ωl (y) = y ∈ R2 |V (y) < l, l > 0, V̇ (y) ≤ 0
by control parameter Kp , which guarantees D1 >0. Therefore,
 any finite region Ωl can be found to ensure that the condition
R(y) = y ∈ Ωl (y)|V̇ (y) = 0 (27)
V̇ (y) ≤ 0 is satisfied.
As displayed in Fig. 6(b), when the UEPs are included in the
where R represent some points when V̇ (y) = 0 and M is the
region Ωl , the region Ωl will become an infinite area (red line).
largest invariant set in the region R. In fact, M is the SEP in the
In order to avoid this situation, an additional constraint, which
region Ωl .
is only an SEP in the region Ωl , needs to be added to ensure the
The system state y0 (y10 , y20 ) is the initial state after disturbance.
region Ωl is a finite region.
The system state ye (y1e , 0) is the SEP after disturbance. It is
The energy Emax1 of the system at the UEP can be expressed
worth noting that Ωl is an equipotential surface.
as
Any points that start from the region Ωl will converge to this
largest invariant set M when time goes to infinity, as shown in Emax 1 = −(P1ref δU + P1III cos δU ). (28)

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 FU et al.: LARGE-SIGNAL STABILITY OF GRID-FORMING AND GRID-FOLLOWING CONTROLS IN VOLTAGE SOURCE CONVERTER 7837

boundary under positive damping condition, the system will

restore stability.
Based on the above studying, it leads to two constraints for
system stability: 1) there exists an SEP, and 2) the total energy
E after fault is lower than the energy Emax2 at the δ max2 .

V. VERIFICATION
In this section, the grid-tied VSCs are built in MATLAB/
Simulink platform and RT-LAB platform. The topology is shown
as Fig. 1(a)–(b) and the parameters are listed in Table I. This
article mainly verifies two stability boundaries of the grid-tied
VSCs.

A. Stability Boundary of the Grid-Forming Control

By increasing the item P1, the system states (δ, δ̇) change
during grid fault is simulated. In Fig. 8(a), the states of the system
after the slight fault are still within the stability boundary, so the
system will restore stability.
In Fig. 8(b), the system states cross the stability boundary and
Fig. 7. Large-signal stability analysis about the grid-following VSCs. (a) returns to the interior of the stability boundary under medium
Equivalent actual physical system. (b) Stability boundary. grid fault. During the movement, the total energy of the system
gradually decreases to be lower than the energy of the UEP, so the
system cannot cross the UEP, and the system restores stability.
Therefore, the equipotential surface E = Emax1 can be re- The system states need to cross the stability boundary enough
garded as the biggest large-signal stability boundary (blue line), distance to overcome the system damping effect. The system
as shown in Fig. 6(b). When the system states (δ, δ̇) are in states have crossed the stability boundary after severe grid fault,
the boundary after fault, the system will restore stability. Or as displayed in Fig. 8(c). After the fault, the system is outside
the system states return to the inside of the stability boundary the stability boundary, so the system states will stride over the
under the effect of damping, and the system restores stability. UEP and lose stability. The results verify the reliability of the
Moreover, a smaller stability boundary can be found by selecting boundary.
an equipotential surface with lower energy (green line).
Through the above analysis, it leads to two constraints for B. Stability Boundary of the Grid-Following Control
system stability: 1) there exists an SEP, and 2) the total energy
E after fault is lower than the energy Emax1 at the UEP. By increasing the item Ig , the system states (δ, δ̇) change
during slight grid fault is simulated. In Fig. 9(a), the states of
the system after the fault enter into the stability boundary, so the
C. Large-Signal Stability Analysis of the
system will restore stability.
Grid-Following Control
In Fig. 9(b), the system trajectory state δ, δ̇ is a limit cycle. To
The (20) shows that the system damping D2 (δ) is affected by explain this phenomenon, an indicator of the equivalent damping
the phase difference δ. It means that the region Ωl cannot be is proposed. Its physical meaning represents the average energy
chosen arbitrarily. consumed by the extended trajectory in a period. The indicator
The phase difference δ max2 of the system at D(δmax 2 ) = 0 ¯
I(δ)can be written as
can be written as  
¯ ΔE −D2 (δ)δ̇dδ
Ki I g L g I(δ) =  =  (31)
δmax 2 = arccos . (29) Δδ Δδ
Kp V g  
where ΔEis the reduced energy and Δδ is the length of
The energy Emax2 of the system at the δ max2 is the track running in one cycle.
Due to the equivalent damping I(δ) ¯ ≈ 0, even if the system
Emax 2 = −(P2ref δmax 2 + P2III cos δmax 2 ). (30)
does not in the boundary, it can remain temporarily stable. Under
Therefore, an additional constraint, which is the total energy medium grid fault, the total energy of the system gradually
E ≤ Emax 2 , needs to be added to ensure the system damping ¯ ≥ 0, so the system states enter
decreases by the indicator I(δ)
D2 (δ) >0 in region Ωl . into the boundary, and the system restores stability.
Based on the above analysis, the large-signal stability bound- The system states are not inside the stability boundary after
ary can be got in Fig. 7(b). The system states will converge to severe grid fault, as displayed in Fig. 9(c). And the system total
SEP when the system’s initial states after fault in the boundary. energy gradually increases under the indicator I(δ) ¯ < 0. So that
And the trajectory of system states after fault enters the stability the system total energy is greater than the energy at UEP, that is,

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 7838 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 36, NO. 7, JULY 2021

Fig. 8. Transient process of the grid-forming VSCs after fault. (a) After slight grid fault. (b) After medium grid fault. (c) After severe grid fault.

Fig. 9. Transient process of the grid-following VSCs after fault. (a) After slight grid fault. (b) After medium grid fault. (c) After severe grid fault.

the system will lose stability. The results verify the reliability of
the boundaries.
In addition, the quantitative indicator can be used as a crite-
rion of which grid-forming/grid-following VSC is suitable for
¯ > 0 must be established in the
weak/strong grid condition. I(δ)
grid-forming VSC, because its damping is globally positive.
¯ > 0 cannot to be guaranteed in the grid-following
However, I(δ)
VSC due to the influence of phase difference. When I(δ) ¯ > 0,
the stability of the grid-following VSC is guaranteed, and it
can be employed to deal with the Vpcc voltage fluctuations.
¯ < 0, compared with the grid-following VSC, the
When I(δ)
application of the grid-forming VSC improves the stability of
the system. However, the characteristics of resisting voltage
fluctuations are sacrificed.
Fig. 10. Configuration of the experimental setup.

C. Experimental Verification in Fig. 11(c), the system states have not entered the boundary
To further verify the reliability of the boundaries, hardware- under the effect of damping, the system loses stability in this
in-the-loop (HIL) experiments with grid-connected VSCs are cycle. Obviously, the experimental results are consistent with
performed. The circuit of the VSCs is established in the RT-LAB the theoretical analysis.
platform and the control part is achieved by an ARM-FPGA Fig. 12 shows experimental results of the phase difference δ
digital control platform, as displayed in Fig. 10. by changing reference active current in grid-following VSC.
Fig. 11 shows experimental results of the phase difference δ After slight grid fault, the system states are still in stability
by changing reference active power in grid-forming VSC. After boundary, the system restore stability, as shown in Fig. 12(a). As
slight grid fault, the system states are still in stability boundary, demonstrated in Fig. 12(b), the system states enter the boundary
the system restore stability, as shown in Fig. 11(a). As exhibited when the equivalent damping I(δ)¯ is positive, the system restores
in Fig. 11(b), the system states enter the boundary under the stability. Meanwhile, the phenomenon of limit cycles appeared
influence of damping, the system restore stability. As displayed ¯ of the system
in the process of operation. It reveals that the I(δ)

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 FU et al.: LARGE-SIGNAL STABILITY OF GRID-FORMING AND GRID-FOLLOWING CONTROLS IN VOLTAGE SOURCE CONVERTER 7839

Fig. 11. Response of the grid voltage, current, and phase difference δ by changing the reference active power in grid-forming VSC. (a) System states after slight
grid fault is in the boundary. (b) System states after medium grid fault return to the boundary under the influence of damping. (c) System states after severe grid
fault does not enter the boundary even under the influence of damping.

Fig. 12. Response of the grid voltage, current, and phase difference δ by changing the reference active current in grid-following VSC. (a) System states after
slight grid fault is in the boundary. (b) System states after medium grid fault return to the boundary under the influence of positive damping. (c) System states after
severe grid fault does not enter the boundary under the influence of negative damping.

operating is approximately 0 in the limit cycle, that is, the energy the method. Finally, the stability boundaries are verified by
of the system is almost not consumed. And this phenomenon simulations of VSCs built in MATLAB/Simulink platform and
cannot appear in grid-forming VSC. As displayed in Fig. 12(c), HIL experiments.
the system states have not entered the boundary and the I(δ) ¯
is negative. Then, the energy of the system gradually increases REFERENCES
to be greater than the energy at UEP. Finally, the system loses [1] C. K. Tse, M. Huang, X. Zhang, D. Liu, and X. L. Li, “Circuits and systems
stability. Obviously, the experimental results verify the effec- issues in power electronics penetrated power grid,” IEEE Open J. Circuits
Syst., vol. 1, pp. 140–156, 2020.
tiveness of the theoretical analysis. [2] M. Huang, Y. Peng, C. K. Tse, Y. Liu, J. Sun, and X. Zha, “Bifurcation
and large-signal stability analysis of three-phase voltage source converter
under grid voltage dips,” IEEE Trans. Power Electron., vol. 32, no. 11,
VI. CONCLUSION pp. 8868–8879, Nov. 2017.
[3] B. K. Poolla, D. Gross, and F. Dorfler, “Placement and implementation
In this article, considering the nonlinear characteristics of of grid-forming and grid-following virtual inertia and fast frequency re-
the grid-forming and the grid-following VSCs, the large-signal sponse,” IEEE Trans. Power Syst., vol. 34, no. 4, pp. 3035–3046, Jul. 2019.
[4] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodriguez, “Control of power
stability is investigated for the two types of VSCs connected to converters in AC microgrids,” IEEE Trans. Power Electron., vol. 27, no. 11,
power grid. Comparing the two VSC systems with the swing pp. 4734–4749, Nov. 2012.
equation of the SG, the energy functions of two systems (the [5] G. Ö, R. Teodorescu, C. L. Bak, F. Iov, and P. C. Kjær, “Instability of wind
turbine converters during current injection to low voltage grid faults and
grid-forming and the grid-following VSCs) are derived. The PLL frequency based stability solution,” IEEE Trans. Power Syst., vol. 29,
four system characteristic parameters of the two types of VSC no. 4, pp. 1683–1691, Jul. 2014.
systems are introduced in this article. They reveal the similarities [6] L. Zhang, L. Harnefors, and H. Nee, “Power-synchronization control
of grid-connected voltage-source converters,” IEEE Trans. Power Syst.,
and differences between the two systems. vol. 25, no. 2, pp. 809–820, May 2010.
A Lyapunov-based stability criterion and the La Salle’s invari- [7] D. Dong, B. Wen, P. Mattavelli, D. Boroyevich, and Y. S. Xue, “Grid-
ance principle are employed to analyze the large-signal stability synchronization modeling and its stability analysis for multi-paralleled
three-phase inverter systems,” in Proc. 28th Annu. IEEE Appl. Power
and derive the stability boundaries of the two types of VSCs in Electron. Conf. Expo., 2013, pp. 439–446.
the positive damping region. Unlike the grid-forming VSC, the [8] F. Gao et al., “Comparative stability analysis of droop control approaches
grid-following VSC will lose stability by exhibiting a varying in voltage-source-converter-based DC microgrids,” IEEE Trans. Power
Electron., vol. 32, no. 3, pp. 2395–2415, Mar. 2017.
damping coefficient. The indicator of the equivalent damping [9] H. H. Xin, L. B. Huang, L. Q. Zhang, Z. Wang, and J. B. Hu, “Synchronous
can be used as a criterion of whether the grid-forming/-following instability mechanism of P-f droop-controlled voltage source converter
control is suitable for the weak/strong grid condition. caused by current saturation,” IEEE Trans. Power Syst., vol. 31, no. 6,
pp. 5206–5207, Nov. 2016.
It is worth noting that a unified large-signal stability analysis [10] C. Zhang, X. Cai, and Z. Li, “Transient stability analysis of wind turbines
method is proposed for two types of VSCs and the physical with Full-scale voltage source converter,” Proc. Chin. Soc. Elect. Eng.,
mechanism of the system losing stability can be studied in vol. 37, no. 14, pp. 4018–4026, 2017.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.
 7840 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 36, NO. 7, JULY 2021

[11] H. Wu and X. F. Wang, “Transient angle stability analysis of grid- Zhen Tian (Member, IEEE) received the B.S. de-
connected converters with the first-order active power loop,” in Proc. 33rd gree in electrical engineering from Wuhan University,
Annu. IEEE Appl. Power Electron. Conf. Expo., 2018, pp. 3011–3016. Wuhan, China, in 2014, and the Ph.D. degree in
[12] X. F. Wang, L. Harnefors, and F. Blaabjerg, “Unified impedance model of control science and engineering from Shanghai Jiao
grid-connected voltage-source converters,” IEEE Trans. Power Electron., Tong University, Shanghai, China, in 2019.
vol. 33, no. 2, pp. 1775–1787, Feb. 2018. From 2017 to 2019, he was a Visiting Scholar
[13] H. Geng, L. Liu, and R. Q. Li, “Synchronization and reactive current with the Department of Electrical and Computer En-
support of PMSG-based wind farm during severe grid fault,” IEEE Trans. gineering, Illinois Institute of Technology, Chicago,
Sustain. Energy, vol. 9, no. 4, pp. 1596–1604, Oct. 2018. IL, USA. He is currently a Postdoctoral Fellow with
[14] H. Wu and X. Wang, “Design-oriented transient stability analysis of PLL- the School of Electrical Engineering and Automa-
synchronized voltage-source converters,” IEEE Trans. Power Electron., tion, Wuhan University. His research interests include
vol. 35, no. 4, pp. 3573–3589, Apr. 2020. modeling, control, and stability analysis of renewable energy generation, micro-
[15] S. H. Strogatz, Nonlinear Dynamics and Chaos: With Applications to grid and power-electronics-enabled power systems.
Physics, Biology, Chemistry, and Engineering. New York, NY, USA:
Peruses Books, 1994, pp. 145–163. Han Yan was born in Bozhou, Anhui Province,
[16] H. Wu and X. F. Wang, “Design-oriented transient stability analysis of grid-
China, in 1995. She received the B.Eng. degree in
connected converters with power synchronization control,” IEEE Trans.
electrical engineering in 2019 from Wuhan Univer-
Ind. Electron., vol. 66, no. 8, pp. 6473–6482, Aug. 2019.
sity, Wuhan, China, where she is currently working
[17] P. Hart and B. Lesieutre, “Energy function for a grid-tied, droop-controlled
toward the M.Eng. degree.
inverter,” in Proc. North Amer. Power Symp., 2014, pp. 1–6.
Her research interests include the modeling and
[18] P. Kundur, Power System Stability and Control. New York, NY, USA:
stability analysis of power electronics.
McGraw-Hill, 1994.
[19] J. Zhao, M. Huang, H. Yan, C. K. Tse, and X. Zha, “Nonlinear and transient
stability analysis of phase-locked loops in grid-connected converters,”
IEEE Trans. Power Electron., vol. 36, no. 1, pp. 1018–1029, Jan. 2021.
[20] S. K. Chung, “A phase tracking system for three phase utility interface Herbert Ho-Ching Iu (Senior Member, IEEE) re-
inverters,” IEEE Trans. Power Electron., vol. 15, no. 3, pp. 431–438, ceived the B.Eng. (hons.) degree in electrical and
May 2000. electronic engineering from the University of Hong
[21] N. G. Bretas and L. F. C. Alberto, “Lyapunov function for power system’s Kong, Hong Kong, in 1997, and the Ph.D. degree in
with transfer conductances: Extension of the invariance principle,” IEEE electronic and information engineering from Hong
Trans. Power Syst., vol. 18, no. 2, pp. 769–777, May 2003. Kong Polytechnic University, Hong Kong, in 2000.
[22] H. K. Khalil, Nonlinear Systems. New Jersey, NJ, USA: Prentice Hall, In 2002, he was with the School of Electrical, Elec-
2002, pp. 111–116. tronic and Computer Engineering, The University of
Western Australia, Perth, WA, Australia, where he is
Xikun Fu was born in Fengcheng, Jiangxi Province, currently a Professor. His research interests include
China. He received the B.S. degree in electrical engi- power electronics, renewable energy, nonlinear dy-
neering from the Central South University, Changsha, namics, current sensing techniques, and memristive systems.
China, in 2019. He is currently working toward the Prof. Iu currently serves as an Associate Editor for the IEEE TRANSACTIONS
M.Eng. degree with the School of Electrical Engi- ON POWER ELECTRONICS, the IEEE TRANSACTIONS ON SMART GRIDS, the
neering and Automation, Wuhan University, Wuhan, IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING, the IEEE
China. TRANSACTIONS ON CIRCUITS AND SYSTEMS-II, and the IEEE ACCESS.
His current research interests include nonlinear
behavior and transient stability of power electronic
converters. Pan Hu (Member, IEEE) was born in Wuhan, Hubei
Province, P.R. China, on August 18, 1989. He re-
ceived the B.E degree in electrical engineering from
Jianjun Sun (Member, IEEE) was born in 1975. Wuhan Textile University, Wuhan, China, in 2012,
He received the B.Eng., M.Eng., and Ph.D. degrees and the Ph.D. degree from the School of Electri-
in electrical engineering from Wuhan University, cal Engineering and Automation, Wuhan University,
Wuhan, China, in 1997, 2000, and 2007, respectively. Wuhan, China, in 2018.
He is currently an Associate Professor with the In 2017, he was a Visiting Ph.D. Student with
School of Electrical Engineering, Wuhan University, Energy System Department, University of Alberta,
where he is also the Deputy Director of the Motor Edmonton, AB, Canada. Since 2018, he has been a
and Power Electronics Center. His research interests Senior Engineer and Research Staff with the State
include modeling and analysis of power electronic Grid Hubei Electric Power Research Institute, China. He has authored more
systems and microgrids. than 30 articles, and more than four inventions. His research interests include
renewable energy planning, power system stability analysis and control, and
power quality mitigation.
Meng Huang (Member, IEEE) received the B.Eng.
and M.Eng. degrees from the Huazhong University
of Science and Technology, Wuhan, China, in 2006 Xiaoming Zha (Member, IEEE) was born in Huain-
and 2008, respectively, and the Ph.D. degree from ing, Anhui Province, China, in 1967. He received the
Hong Kong Polytechnic University, Hong Kong, in B.S., M.S., and Ph.D. degrees in electrical engineer-
2013. ing from Wuhan University, Wuhan, China, in 1989,
He is currently an Associate Professor with the 1992, and 2001, respectively.
School of Electrical Engineering and Automation, He was a Postdoctoral Fellow with the University
Wuhan University, Wuhan, China. His research inter- of Alberta, Edmonton, AB, Canada, from 2001 to
ests include nonlinear analysis of power converters 2003. Since 1992, he has been a Faculty Member
and power electronics reliability. with Wuhan University, where he became a Professor
Dr. Huang was the recipient of the Best Paper Award of the IEEE TRANS- in 2003, and is currently the Deputy Dean with the
ACTIONS ON POWER ELECTRONICS, in 2016. He currently serves as the Guest School of Electrical Engineering. His research inter-
Associate Editor for the IEEE JOURNAL OF EMERGING AND SELECTED TOPICS ests include power electronic converter, the application of power electronics
OF POWER ELECTRONICS, the IEEE JOURNAL OF EMERGING AND SELECTED in smart grid and renewable energy generation, the analysis and control of
TOPICS OF CIRCUITS AND SYSTEMS, and the Associate Editor of IEEE ACCESS. microgrid, the analysis and control of power quality, and frequency control of
high-voltage high-power electric motors.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on July 11,2024 at 17:34:58 UTC from IEEE Xplore. Restrictions apply.