This article has been accepted for publication in IEEE Transactions on Power Delivery.

This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

Positive-Sequence Modeling of Droop-

Controlled Grid-Forming Inverters for Transient
Stability Simulation of Transmission Systems
Wei Du, Senior Member, IEEE, Quan H. Nguyen, Member, IEEE, Song Wang, Senior Member, IEEE,
Jinho Kim, Member, IEEE, Yuan Liu, Member, IEEE, Songzhe Zhu, Senior Member, IEEE, Francis
K. Tuffner, Senior Member, IEEE, Zhenyu Huang, Fellow, IEEE
Numerous research and field demonstration projects have
Abstract—This paper describes a positive-sequence model to shown that such voltage source characteristic of grid-forming
represent two widely reported droop-controlled grid-forming inverters can bring significant benefits to power systems at
inverters for transmission system transient stability simulation both the transmission and distribution levels, which include
studies. Methods of how to develop the equivalent voltage source
improving the system frequency response [4-6], increasing the
behind impedance to represent inverters with and without inner
control loops, modeling of P-f and Q-V droop controls, active and system damping [7, 8], being able to operate in very weak
reactive power limiting modeling, and modeling of fault current systems [9], providing the black-start capability [9, 10], and
limiting controls have been described in detail. The model was enabling the system to operate in the islanded mode [11-13].
implemented in commercially available positive-sequence Current research work primarily focuses on developing
simulation tools and compared with detailed electromagnetic innovative controls to further improve the dynamic
transient models in both a single-grid-forming-inverter infinite-
performance of grid-forming inverters under various fault
bus system and a modified IEEE 39-bus system. Finally, the
model was tested on the U.S. Western Interconnection system. events and analyzing the transient stability of power grids
The study results showed that the model has a good level of dominated by grid-forming inverters. The work in [14]
accuracy compared to the detailed EMT models, and at the same analyzed the transient stability of droop-controlled grid-
time also achieves a high computational efficiency, making it forming inverters when the output currents are saturated and
suitable for large-scale transmission system simulation studies. proposed a simple feedback loop to improve the system
transient stability by modifying the inverter output frequency.
Index Terms—Droop control, grid-forming inverter,
modeling, positive-sequence, transmission system, simulation
The work in [15] and [16] investigated the transient stability
and critical clearing time of droop-controlled grid-forming
I. INTRODUCTION inverters using the virtual impedance and current reference
saturation as the current limiting algorithms, respectively. The
T HE grid-forming inverter (GFM) is considered a critical
asset for power grids with a high penetration of inverter-
based resources [1]. This technology was first used at low-
work in [17] outlined a stationary-frame control architecture
for grid-forming inverters to better handle unbalanced faults.
Reference [18] proposed a DQ-frame asymmetrical virtual
voltage levels such as microgrids [2], but has drawn a lot of impedance control to enhance the transient stability of
attention at transmission levels in recent years because of the droop-controlled grid-forming inverters. The work in [19]
rapid integration of renewables to the grid around the world proposed a system-level supervisory control that uses low-
[3]. bandwidth communication to pre-emptively assign a set of
Because a grid-forming inverter approximately behaves as local ride-through control parameters to individual grid-
a voltage source behind impedance during normal operations, forming inverters to guarantee the system transient stability.
its dynamic characteristic is fundamentally different from the Although many innovative controls are currently being
conventional current-controlled grid-following inverters. proposed for grid-forming inverters, less attention is paid to
the modeling of grid-forming inverters for the bulk power
The Pacific Northwest National Laboratory (PNNL) is operated by system simulation studies. Electromagnetic transient (EMT)
Battelle for the U.S. Department of Energy under Contract DE-AC05- modeling and positive-sequence modeling are the two most
76RL01830. This material is based upon work supported by the U.S. popular methods to simulate the power system dynamic
Department of Energy’s Office of Energy Efficiency and Renewable Energy
(EERE) under Solar Energy Technologies Office (SETO) Award Number behaviors. Theoretically, the EMT model can capture all the
38637. The views expressed herein do not necessarily represent the views of dynamics of an inverter, but it may not be suitable to
the U.S. Department of Energy or the United States Government. simulate large-scale power systems because of the high
W. Du, Q. H. Nguyen, J. Kim, Y. Liu, F. K. Tuffner, and Z. Huang are
with Pacific Northwest National Laboratory, Richland, WA 99354 USA (e-
computational burden. In contrast, the positive-sequence
mail: [email protected], [email protected], [email protected], model is based on the phasor assumption and can
[email protected], [email protected], [email protected]). significantly improve the computational efficiency. In
Song Wang is with Portland General Electric, Portland, OR 97204 USA
today’s utility industry, power system engineers still heavily
(e-mail: [email protected]).
Songzhe Zhu is with GridBright, Alamo, CA 94507 USA (e-mail: rely on the positive-sequence simulation tools to study the
[email protected]).

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

dynamic behaviors of transmission systems due to the scale models, and at the same time also achieves a high
of practical systems. Therefore, it is important to develop the computational efficiency, making it suitable for large-scale
positive-sequence model of grid-forming inverters and transmission system simulation studies. Most parts of the
investigate how accurately the model can reflect the dynamic model described in the paper have been included in a generic
behaviors of grid-forming inverters in the time frame of library positive-sequence grid-forming inverter model,
interest. The work in [20] proposed to use the Thevenin REGFM_A1 [27], which was recently adopted by the US
voltage source instead of an ideal current source to represent Western Electricity Coordinating Council (WECC) and
grid-following inverters in the positive-sequence simulation included in the model libraries of major commercial positive-
tool, which improved the convergence of the network sequence simulation tools, including PTI-PSS®E [28], GE-
solution. Reference [21] performed a comprehensive PSLF [29], PowerWorld Simulator [30], and TSAT [31],
comparative study of the EMT and positive-sequence models marking the first publicly available grid-forming inverter
of grid-following inverters and provided simulation model that is included in all major commercial transient
guidelines of grid-following inverters for grid-integration stability simulation tools used in North America.
studies. The work in [6] developed three-phase phasor The rest of the paper is organized as follows. Section II
models of grid-forming and grid-following inverters to study introduces the basic concept of grid-forming inverters.
the dynamic behaviors of large-scale distribution systems, Section III presents detailed EMT models of two widely
but the fault response of grid-forming inverters was not reported grid-forming droop controls. Section IV introduces
modeled. References [22] and [23] developed positive- the developed positive-sequence model and the interface to
sequence models of grid-forming inverters with and without the network solver. Section V and VI present the comparison
considering the dynamics of the dc bus respectively, but the between the positive-sequence model and the two detailed
comparison with detailed EMT models was not presented. EMT models in a single-GFM infinite-bus system and a
Reference [24] modeled the virtual impedance control of a modified IEEE 39-bus system, respectively. Section VII
grid-forming inverter in the phasor domain. Reference [25] presents the case study on the U.S. Western Interconnection
developed reduced-order models for grid-forming and grid- system, and Section VIII draws the conclusion of this paper.
following inverters to study the small signal stability and
compared the results with the full-order models. The work in II. GRID-FORMING CONCEPT
[26] developed positive-sequence and negative-sequence Unlike a grid-following inverter, which approximately
models of grid-forming inverters that were mainly used for behaves as a controllable current source, a grid-forming
the power system protection scheme design instead of inverter operates as a controllable voltage source behind
dynamic simulation. Literature review revealed that impedance during normal operations, as shown in Fig. 1.
significant work in the domain of positive-sequence
modeling of grid-forming inverters for transmission system E δE Vg δg
XL
transient stability simulation studies is needed. E
It is worth mentioning that although numerous recent ω
studies reported detailed EMT simulations [14-16, 19] to Grid
P,Q
examine the system transient stability, the analytical Fig. 1 Basic model of a grid-forming inverter
approaches reported in those studies were still based on the
phasor assumption, which ignored the dynamics of the If the coupling reactance XL is well designed, for example
inverter fast inner control loops and the dynamics of the between 0.05 pu and 0.2 pu on an inverter rating base, the
network. This indicates that it might be feasible to use the inverter output active power P is approximately linear with the
positive-sequence models to accurately simulate the transient phase angle δP, and the inverter output reactive power Q is
stability of power systems dominated by grid-forming approximately linear with the inverter internal voltage
inverters. magnitude E, according to (1)–(3). This decoupling
This paper describes a positive-sequence model to characteristic simplifies the controller design.
represent two widely reported droop-controlled grid-forming  p = E - g (1)
inverters for the transmission system transient stability
EVg EVg
simulation studies. Methods of how to develop the P= sin  p  p (2)
equivalent voltage source behind impedance to represent XL XL
inverters with and without inner control loops, modeling of E 2 − EVg cos  p E ( E − Vg )
P-f and Q-V droop controls, active and reactive power Q=  (3)
XL XL
limiting modeling, and modeling of fault current limiting
controls are described in detail. The positive-sequence model
III. EMT MODELS OF TWO WIDELY USED DROOP CONTROLS
was validated against detailed EMT models in both a single-
GFM infinite-bus system and a modified IEEE 39-bus Droop control is one of the most popular approaches to
system. Finally, the model was tested on the U.S. Western realize the parallel operation of multiple grid-forming
Interconnection system. Study results showed that the model inverters [32]. A grid-forming inverter does not require the use
has a good level of accuracy compared to the detailed EMT of a phase-locked loop (PLL) in comparison to its counterpart,

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

einv il L1 vo L2 io vt einv il L1 vo io L2 vt
PWM
Comm.3-4 C
C

Comm.3-4
abc θo
dq

PWM ω0pu Pset ed eq

io vt ω0pu Pset
+ +
E θ 0 ω + - Pf vo io + Pf vo io
mp il θ o 0 ω + + -
s P, Q mp
Δω + Calculation s Δω +
P, Q
i*ld Calculation
& Current Voltage
ΔV + Low Pass i*lq ΔV + &
- Qf Loop Loop v*od - Qf Low Pass
mq Filter mq
+ θo v*oq=0 + Filter
Vset Vset

(a) (b)

E δE Vo δo
XL1 XL2 XL2
E Vo
ω ω
(c) (d)

Fig. 2 Dierct and multi-loop droop GFM controls. (a) Direct droop control. (b) Multi-loop droop control. (c) Equivalent circuit of the direct droop control in
the fundamental frequency. (d) Equivalent circuit of the multi-loop droop control in the fundamental frequency.

the grid-following inverter. This section introduces the variables and parameters are listed in Table I in this section,
detailed EMT models of two widely reported grid-forming and Table II in Section V, respectively.
droop controls. A detailed description of the EMT models in
1 1
this section makes it more convenient for readers to Pf = P, Q f = Q (4)
1 + Ts 1 + Ts
understand how the positive-sequence model is developed in
Section IV by making appropriate assumptions. In addition, as
TABLE I
shown in Fig. 2 (a) and (b), this paper assumes the DC bus Electrical and control variables of grid-forming inverters
voltage is constant and ignores the dynamics at the DC side. Common electrical and control variables of direct and multi-
loop droop controls
A. Two Widely Used Grid-Forming Droop Controls Symbol Description
L1 Inverter bridge side filter inductance
Historically, two types of grid-forming droop controls are C Inverter filter capacitor
widely reported in the literature [33], which are defined as the L2 Inverter grid side filter inductance
direct droop control and the multi-loop droop control in this XL1 Inverter bridge side filter reactance
XL2 Inverter grid side filter reactance
paper, as shown in Fig. 2 (a) and (b), respectively. As shown einv Inverter bridge side voltage
in Fig. 2 (a), the direct droop control directly regulates the vo Filter capacitor voltage
angular frequency ω and magnitude E of the modulation vt Inverter terminal voltage
il Inverter output current at the bridge side
waveform according to the P-f droop control and Q-V droop io Inverter output current at the grid side
control, respectively, where Pf and Qf are the active power and Pf Active power filtered by a first-order low-pass filter
reactive power filtered by a first-order low-pass filter as Qf Reactive power filtered by a first-order low-pass filter
shown in (4). In contrast, for the multi-loop droop control, Δω Output from the active power limiting
ΔV Output from the reactive power limiting
there are additional cascaded inner voltage and current loops Electrical and control variables for the direct droop control
other than the P-f and Q-V droop controls, as shown in Fig. 2 Symbol Description
(b). The purpose of the cascaded inner voltage and current E Magnitude of the modulation waveform
θ Rotating phase angle of the modulation waveform
loops is to achieve fast control of the waveform of the filter
Electrical and control variables for the multi-loop droop control
capacitor voltage vo. To achieve the fast control of vo, the Symbol Description
bandwidths of the inner voltage and current loops are typically ed, eq dq frame components of the modulation waveform
designed to be very high. For example, the bandwidths of the θo Rotating phase angle of the filter capacitor voltage
v*od, v*oq Input references of the voltage loop
voltage and current loops used in [34] are 400 Hz and 1.6 kHz, i*ld, i*lq Current references before the current limiting block
respectively. With the waveform of vo well controlled by the i*lds, i*lqs Current references after the current limiting block
inner voltage and current loops, the P-f and Q-V droop vod, voq dq frame components of the filter capacitor voltage
ild, ilq dq frame components of the inverter bridge side current
controls then regulate the angular frequency and magnitude of ild, ilq dq frame components of the inverter grid side current
vo in a relatively slow manner. The explanations of all the

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

Fig. 2 (c) and (d) show the equivalent circuits of the two insulated-gate bipolar transistors (IGBTs), the overcurrent
controls in the fundamental frequency. For the direct droop needs to be limited rapidly, typically within 1 ms as reported
control, because it directly regulates the magnitude and by the inverter manufacturer [36]. This subsection introduces
angular frequency of the modulation waveform, the equivalent two current limiting controls that are typically used by the
circuit can be represented by a voltage source behind direct droop and multi-loop droop controls.
impedances XL1 and XL2, as shown in Fig. 2 (c). The pulse For the direct droop control, since it does not have an inner
width modulation (PWM) control can be ignored in the current loop, the current limiting is implemented at the PWM
fundamental frequency. In addition, the filter capacitor C can control layer. The typical approach is to block the associated
be ignored because of its negligible value in the fundamental IGBTs once detecting the overcurrent, and then release the
frequency. For the multi-loop droop control, because it blocking signal once the current drops below the threshold Imax
regulates the magnitude and angular frequency of the filter using a hysteresis loop. Fig. 4 shows the current limiting
capacitor voltage vo, the equivalent circuit in the fundamental control block for phase A. Once the output current ila exceeds
frequency can be represented by a voltage source behind Imax, the hysteresis loop sends the blocking signal blka to IGBT
impedance XL2. The cascaded inner voltage and current loops, 1 and IGBT 4 of the phase A bridge, where g1 and g4 are the
the inverter side inductor L1 and filter capacitor C, and the original gate signals generated by the PWM control, and g’1
PWM control can be ignored in the fundamental frequency. and g’4 are the new gate signals modulated by the blocking
The work in [33] pointed out the above difference between the signal blka. The currents on phase B and C can be limited in
two controls and compared their small signal performances the same manner. This approach was briefly mentioned in
using detailed EMT models. [36], and explained in detail in [37]. The advantage of this
approach is that it can limit the overcurrent very quickly, for
B. Active and Reactive Power Limiting
example within 1–2 PWM cycles. However, this approach
When connecting a GFM to the bulk power system, power results in a flat-topped current waveform during faults, which
limiting controls are needed to prevent the output P and Q is shown in Section V.
from exceeding the limits under disturbances like loss of
generation units and the step change in voltage. Fig. 3 shows + blka
Imax g'1
the active and reactive power limiting control blocks. - g1
|ila|
0 0 g'4
Pmax Qmax g4
kpP kpQ
+- + +- +
0 + ωLLim 0 + VLLim
kiP/s kiQ/s Fig. 4 The current limiting used by the direct droop control.
Pf ωLLim + Δω Qf VLLim + ΔV
ωHLim + VHLim +
For the multi-loop droop control, since it has an inner
kiP/s kiQ/s current loop, the overcurrent can be limited by saturating the
0 ωHLim 0 VHLim current references. Fig. 5 shows the cascaded inner voltage
Pmin+- ++ Qmin+- ++
kpP kpQ and current loops together with the current limiting block.
0 0
(a) (b) * einvd
v*od + + + i ld i*lds + +
kpv+kiv/s kpi+kii/s
Fig. 3 Power Limiting. (a) Active power limiting. (b) Reactive power limiting. - + + - +
vod -ωC1 F -ωL1
ild
As shown in Fig. 3 (a), when Pf exceeds Pmax or drops iod Current
below Pmin, the anti-windup proportional-integral (PI) ioq Limiting
voq ωC1 ilq ωL1
controller is activated to limit Pf by changing the inverter F
- + + i*lq * - + einvq
v*oq + i lqs+
output angular frequency ω. This control was proposed by the kpv+kiv/s + + kpi+kii/s +
Consortium for Electric Reliability Technology Solutions
Fig. 5 Inner voltage and current loops of the multi-loop droop control.
(CERTS) Microgrid project and tested in the field [2, 11].
Similarly, as shown in Fig. 3 (b), once Qf exceeds Qmax or The current limiting algorithm used in this paper is shown
drops below Qmin, the reactive power limiting is activated to in (5). If the magnitude of the current reference exceeds Imax,
limit Qf by changing the inverter output voltage [35]. The both current references i*lds and i*lqs are reduced proportionally
active power and reactive power limiting controls can be according to (5). This assures that the magnitude of the current
added either on the direct droop control or on the multi-loop is limited at Imax during the fault. Such method is called
droop control. circular current limiting control according to [38]. It should be
noted that when using this approach, the saturation limits of
C. Current Limiting during Faults
the voltage loop need to be carefully designed to avoid windup
Because a GFM approximately behaves as a voltage source when the output current is saturated [10].
behind impedance during normal operations, during a severe
short-circuit fault it can generate high currents with the i* = i* , i* = i* if i* 2 + i* 2  I
possibility of damaging the inverter hardware. Therefore,  lds ld lqs lq ld lq max

* * *
current limiting controls are needed to prevent GFMs from ild I max ilq I max (5)
ilds = * 2 * 2 , ilqs = * 2 * 2 if ild + ilq  I max
* *2 *2

overcurrent during faults. Considering the characteristic of  ild + ilq ild + ilq

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

Compared to the current limiting method used by the direct reactance XL2, as expressed in (6) and (7). The filter capacitor
droop control, this approach does not result in the distortion of C can be ignored because its value in the fundamental
current waveforms under balanced faults. However, its frequency is typically very small.
response speed is constrained by the bandwidth of the current
EGFM GFM = E E (6)
loop, so it is not as fast as the approach shown in Fig. 4. In
addition, this approach also results in distorted current X L = X L1 + X L 2 (7)
waveforms under unbalanced faults.
When representing the multi-loop droop control, the
It is important to note that although the two current limiting
voltage phasor EGFMδGFM refers to the filter capacitor voltage
controls described earlier can effectively limit the overcurrent
Voδo, and XL refers to the inverter grid side filter reactance
during faults, they only work for short-term faults that are
XL2 only, as expressed in (8) and (9). The cascaded inner
quickly cleared by the power system protection devices (e.g., a
voltage and current loops together with the inverter bridge side
0.1 s fault). If the fault lasts too long, these two methods can inductor L1 and filter capacitor C are ignored in the phasor
result in the loss of synchronism to the grid after the fault, as domain. As explained in Section III A, the bandwidths of the
reported in [14, 16]. inner voltage and current loops are at least in the order of a
few hundred Hz, so it is reasonable to assume those inner
IV. POSITIVE-SEQUENCE MODELING loops track ideally in the positive-sequence tool and ignore
After describing detailed EMT models of the two types of their dynamics.
GFM droop controls, this section introduces how to develop EGFM GFM = Vo o (8)
the equivalent positive-sequence model to represent them in
X L = X L2 (9)
the phasor domain. The positive-sequence modeling is based
on the phasor assumption and ignores the network dynamics. B. Droop Control and Power Limiting Modeling
In addition, the positive-sequence simulation typically runs at
In the positive sequence tool, the low-pass filter, P-f and Q-
a time step of a quarter of a cycle (e.g., 4.1667 ms for a 60 Hz
V droop control, and active and reactive power limiting of
power system), which is much larger than the time step used
GFMs can be modeled as the same as modeled in the EMT
by EMT simulations. Therefore, appropriate simplifications
simulation tool. This is because the bandwidths of those
must be made to model GFMs in the positive-sequence tool.
control loops are relatively low, so that they can be modeled in
A. Voltage Phasor behind Impedance Representation the phasor domain without causing numerical stability issues.
According to the explanation in Section III A, a GFM can Fig. 7 shows the overall control scheme. Detailed control
be represented by a voltage phasor EGFM∠δGFM behind blocks were shown in Fig. 2 and Fig. 3, and explained in
impedance XL in the positive-sequence tool, as shown in Fig. 6 Section III. In Fig. 7 (a) and (b), δdroop and Edroop are the static
(a), and the GFM controller changes the voltage magnitude phase angle and voltage magnitude obtained from the control
and phase angle of the voltage phasor. The dynamic of filter blocks. Note that in the positive-sequence tool, the static phase
inductance is ignored and is represented by an algebraic term angle δdroop should be used by the network solver instead of
“jXL” instead of a differential term. This is significantly the rotating phase angle θ. By comparing Fig. 7 (a) and Fig. 2
different from how a traditional grid-following inverter is (b), it can be seen that θ=δdroop+ω0t. Assuming the GFM
modeled in positive-sequence tools. As a grid-following output frequency is 60 Hz, then δdroop is a fixed value while θ
inverter is typically modeled as a controllable current phasor, changes from 0 to 2 periodically.
and the controller changes the active and reactive current Pset
references [39]. ω 0 δdroop
P-f Droop ++ +-
Pf s
P Low-Pass ω0pu
XL V t δt V t δt Filter
Active Power Δω
Limiting
P, Q, I φ P, Q, I φ
Pmax Pmin
EGFM δGFM IN φ N YL (a)
Vset
Edroop
Q-V Droop ++
(a) (b) Q Low-Pass Qf
Filter
Fig. 6 GFM representation. (a) Voltage phasor behind impedance Reactive Power ΔV
representation. (b) Norton equivalent circuit. Limiting

It is important to note that the value of XL is different when Qmax Qmin

(b)
representing the two different GFM droop controls described
in Section III. When representing the direct droop control, the Fig. 7 Overall control scheme. (a) P-f droop and active power limiting (b) Q-V
voltage phasor EGFMδGFM refers to the inverter bridge side droop and reactive power limiting.
voltage EδE, and the reactance XL is the summation of the C. Fault Current Limiting Modeling
inverter bridge side filter reactance XL1 and grid side filter Neither the PWM blocking method, nor the circular current

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

limiting method described in Section III C can be directly model. Equation (12) and (13) are used as a criterion to
modeled in the positive-sequence tool because of their high determine if the GFM will go into the current limiting or not.
bandwidths. Therefore, an equivalent modeling method must At each simulation time step, the GFM output current phasor
be developed to accurately represent their control effect and at Iφ is calculated using (12). If the current magnitude I is
the same time allows a larger simulation time step for the smaller than Imax, then the GFM will not go into the current
positive-sequence simulation. Since the current limiting limiting mode, and EGFM and δGFM are governed by the P-f and
control typically responds very fast, it is reasonable ignore its Q-V droop controls, as shown in (13-a). Once I exceed Imax
dynamics and model it algebraically in the positive-sequence caused by a fault, for example Vt drops to 0 pu, the GFM will
tool. The work in [38] pointed out that the circular current go into the current limiting mode and the voltage phasor
limiting control used by the multi-loop droop-controlled EGFMδGFM is calculated using (13-b). In the current limiting
GFMs as described in (5) is equivalent to a virtual resistor. mode, the GFM output current phasor ImaxφLim is dependent
Similarly, simulation studies show that the control effect of on the virtual resistor Re, as shown in Fig. 9. Equation (13-b)
PWM blocking method used by the direct-droop-controlled assures that the magnitude of the GFM output current is
GFMs as described in Fig. 4 is also like a virtual resistor. limited at Imax in the current limiting mode.
Therefore, those two current limiting controls described in
Section III C can be equivalently modeled as a virtual resistor Edroop  droop − Vt  t
Re in the positive-sequence tool, as shown in Fig. 8. I  = (12)
jX L

Re XL Vt δt  EGFM  GFM =Edroop  droop , if I  I max

 (13-a)

 EGFM  GFM =Vt  t + jX L I max  Lim , if I  I max
 (13-b)
P, Q, I φ
Edroop δdroop D. Interface with the Network Solver
The flow chart in Fig. 10 describes how the developed
positive-sequence GFM model interacts with the network
Fig. 8 Equivalent circuit of the GFM in the current limiting mode. solver.
The function of the virtual resistor Re is to limit the
P, Q
magnitude of output current at Imax during a fault as described
Droop Controls
in (10), and its value can be calculated using (11). (Fig. 7)

Edroop droop − Vt  t Edroop δdroop

= I max (10)
Re + jX L V t δt Current Limiting
(Fig. 9)
E 2 droop + V 2t − 2 EdroopVt cos( droop −  t )
Re = 2
− X L2 (11) EGFM δGFM
I max Internal Voltage Source
(Fig. 6(a))

Eq. (14)-(15)
Edroop δdroop, Vt δt
Norton Equivalent
Edroop  droop − Vt  t Circuit (Fig. 6(b))
I  =
jX L IN φ N , Y L

Network Solver
No
I>I max ? EGFM  GFM =Edroop  droop
No
Yes Converged ?

Yes
E 2 droop + V 2t − 2 EdroopVt cos( droop −  t )
Re = 2
− X L2
I max Fig. 10 Flow chart describing the interaction between the grid-forming
inverter model and the network solver.

At each simulation time step, the P-f and Q-V droop

Edroop  droop − Vt  t
I max  Lim = controls as shown in Fig. 7 first generate the voltage phasor
Re + jX L
Edroopδdroop based on the P and Q returned by the network
solver, and then the current limiting algorithm shown in Fig. 9
EGFM  GFM =Vt  t + jX L I max Lim checks if the inverter output current exceeds Imax or not and
determines the actual inverter internal voltage phasor
Fig. 9 Flow chart of the current limiting algorithm developed for positive-
sequence simulation tools.
EGFMδGFM. Because most positive-sequence tools used in the
United States use the current-injection-based methods to solve
The flow chart in Fig. 9 describes the virtual-resistor-based the network equations, the voltage phasor EGFMδGFM behind
current limiting algorithm developed for the positive-sequence impedance XL shown in Fig. 6 (a) is converted to its Norton

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

equivalent circuit, as shown in Fig. 6 (b), using (14) and (15). TABLE III
The values of L1, L2, and equivalent XL
Once the current phasor INφN and admittance YL are
Equivalent XL in the
obtained, they are sent to the network solver to solve the Control Mode L1 L2
positive-sequence model
network algebraic equations for the next simulation time step. Direct droop 0.08 pu 0.07 pu 0.15 pu
Multi-loop droop 0.05 pu 0.15 pu 0.15 pu
EGFM  GFM
I N  N = (14)
jX L Table II shows the controller parameters of the direct droop
1 control and the multi-loop droop control. The parameters of
YL = (15) the low-pass filter, P-f and Q-V droop, and power limiting are
XL
the same for both GFM controls. The line impedance is Rg +
jXg = 0.02 + j0.15 pu. According to Fig. 2 (c) and (d), if the
V. COMPARISON IN A SINGLE-GFM INFINITE-BUS SYSTEM
values of L1 and L2 remain the same when using two GFM
To investigate how accurately the developed positive- droop controls, they will have different values of XL in the
sequence model can represent the two GFM droop controls, a fundamental frequency, resulting in different simulation
rigorous comparison between EMT and positive-sequence results. Therefore, to achieve a reasonable comparison, the
simulations was performed on a single-GFM infinite bus values of L1 and L2 were designed to assure that the two GFM
system as shown in Fig. 11. The EMT models of the two GFM controls have the same value of XL in the fundamental
controls were developed in the PSCAD simulation tool [40]. frequency, as listed in Table III. The filter capacitor C was
The IGBT-based model was used to represent the inverter in designed to have a cutoff frequency of 600 Hz. With the
PSCAD, and the PWM switching frequency was set at 10 values of XL and other major parameters remaining the same,
kHz. The positive-sequence model was developed in the it is expected that the direct droop control and the multi-loop
commercially available positive-sequence simulation tool droop control will have similar dynamic performances.
PSS/E. The time step for the PSCAD simulation was 5 μs, and
the time step for the PSS/E simulation was 4.167 ms. A. Step Change in Voltage Magnitude
In this scenario, the magnitude of grid voltage Vg dropped
vt vg to 0.8 pu at 1.2 s, resulting in the GFM terminal voltage Vt to
il L1 vo L2 io Rg Lg
drop to 0.86 pu.
C

Fig. 11 A single-GFM infinite-bus system.

TABLE II
Controller parameters of the two GFM droop controls
Common controller parameters for the direct and multi-loop droop
controls
Symbol Description Value Qmax=0.44 pu
Pset Active power setpoint 0.6 pu
Vset Voltage setpoint 1 pu
mp P-f droop gain 0.01 pu
mq Q-V droop gain 0.01 pu
T Time constant of low-pass filter 0.01 s
ω0pu Rated angular frequency in per unit value 1 pu
ω0 Rated angular frequency 377 rad/s
Pmax Upper limit of active power 0.9 pu
Pmin Lower limit of active power 0 pu
kpP Proportional gain of active power limiting 0.01 pu
kiP Integral gain of active power limiting 0.1 pu/s
ωLLim Lower output limit of active power limiting 0.1 pu
ωHLim Upper output limit of active power limiting -0.1 pu
Qmax Upper limit of reactive power 0.44 pu
Qmin Lower limit of reactive power -0.44 pu (a) (b)
kpQ Proportional gain of reactive power limiting 0.1 pu Fig. 12 GFM response to a step change in the grid voltage magnitude. (a)
kiQ Integral gain of reactive power limiting 20 pu/s EMT simulations of two GFM controls. (b) Positive-sequence simulation.
VLLim Lower output limit of reactive power limiting -1 pu
VHLim Upper output limit of reactive power limiting 1 pu Because the internal voltage magnitude Edroop still held
Imax Inverter allowable maximum output current 2 pu approximately constant, the voltage difference between Edroop
Controller parameters for the direct droop control only and Vt resulted in a sudden increase in Q. Once Q reached the
Symbol Description Value
H Bandwidth of the hysteresis current loop 0.001 pu limit Qmax, the reactive power limiting control shown in Fig. 3
Controller parameters for the multi-loop droop control only (b) was activated to limit Q at 0.44 pu by reducing Edroop. The
Symbol Description Value system reached a steady state after about 0.1 s. The output
kpv Proportional gain of inner volage loop 20 pu frequency of the GFM deviated from 60 Hz during the
kiv Integral gain of inner voltage loop 400 pu/s
F Gain of feedforward term 0.75 transient because of the slight change in P, and it returned to
kpi Proportional gain of inner current loop 2 pu 60 Hz in the steady state because the grid frequency was fixed
kii Integral gain of inner current loop 1000 pu/s at 60 Hz. Fig. 12 (a) and Fig. 12 (b) show the EMT and
positive-sequence simulation results. Note that in the EMT

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

simulation, there were some fast oscillations in the output P,

Q, and f of the GFM in the first 0.06 s after the disturbance.
Because the positive-sequence simulation approach ignores Imax=2 pu
the dynamics of the network and inverter filter inductor, it
cannot capture those fast oscillations as the EMT simulation
does. However, the positive-sequence simulation results
overall aligned well with the EMT simulation results in the
studied 0.5 s time frame.
B. Typical Fault
In this scenario, a 0.1 s bolted fault was applied at the
terminal of the GFM. As shown in Fig. 13, because the (a) (b)
terminal voltage Vt dropped to 0, the GFM output P and Q Fig. 14 GFM output current and phase angle during the 0.1 s bolted fault. (a)
EMT simulations of two GFM controls. (b) Positive-sequence simulation.
became 0 during the fault. After the fault was cleared, the
system remained stable. Fig. 14 shows the GFM output current
Io and the phase angle from the P-f droop control δdroop. During
the fault, the output current was immediately limited at 2 pu.
For the EMT simulations, this is because the inner current
limiting controls respond very fast, so the dynamics are
invisible in the selected time frame. For the positive-sequence
simulation, this is because the current limiting is implemented
algebraically, so its dynamics are ignored. The inverter output (a) (b)
currents from the EMT simulations were slightly higher than 2 Fig. 15 EMT simulations of the GFM bridge side current waveform during the
pu. This is because the output current in Fig. 14 refers to the 0.1 s bolted fault. (a) Direct droop control. (b) Multi-loop droop control.
grid side current so that the filter capacitor has a contribution Note that for the positive-sequence simulation, the P, Q,
to it. Fig. 15 shows the inverter bridge side current waveforms and Vt shown in the simulation results were returned by the
obtained from the EMT simulations. For the direct droop network solver. They were not the outputs of the low-pass
control, the inverter output currents were limited at 2 pu, but filters. Therefore, they had a “step change” response during
the waveforms were distorted due to the current limiting the disturbance. In contrast, for the EMT simulations, the P,
control algorithm. This distortion also resulted in a slightly Q, and Vt shown in the simulation results were passed through
higher root-mean square value of the current, as shown in Fig. low-pass filters to filter out high-frequency noises. Therefore,
14. For the multi-loop droop control, the currents were limited they did not have the “step change” response during the
at 2 pu without distortions, but there was a slight overshoot in disturbance.
the first cycle during the fault. The phase angle δdroop increased
during the fault according to the P-f droop control and C. Delayed Clearing Fault
returned to the pre-fault value after the fault was cleared. Both Fig. 16 and Fig. 17 show the responses of the GFM to a
the EMT simulations and the positive-sequence simulation 0.45 s delayed clearing fault. Because the fault lasted a longer
reflected the change of the phase angle. duration, the phase angle δdroop lost synchronism to the grid
and could not return to the pre-fault value. Although δdroop
stabilized at another value after the fault was cleared, as
shown in Fig. 17, the system experienced large transients on
P, Q, Vt, and f during and after the fault as shown in Fig. 16.
Such large transients can trigger the inverter internal
protections in a real system. Therefore, the current limiting
controls described in this paper cannot help the GFM survive
such severe faults. The loss of synchronism phenomenon
caused by similar current limiting algorithms has been
explained previously in references [14, 16]. Fig. 16 (b) and
Fig. 17 (b) show that such phenomenon can also be captured
by the positive-sequence model. Note that many innovative
controls are currently being proposed to help GFMs survive
such large disturbances. However, developing new fault ride-
through controls of GFMs is not the focus of this paper. After
the fault was cleared, the output current Io experienced some
transients in the EMT simulations but not in the positive-
(a) (b) sequence simulation. Again, this is because the filter capacitor
Fig. 13 GFM response to a 0.1 s bolted fault. (a) EMT simulations of two exists in the EMT models but not in the positive-sequence
GFM controls. (b) Positive-sequence simulation. model.

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

As shown in Fig. 18, five out of ten synchronous machines

were replaced with GFMs, out of which three of them used the
direct droop control and the other two used the multi-loop
droop control. The parameters of the GFMs were the same as
those described in Section V, except for that Pmax was changed
to 1 pu and Qmax and Qmin were changed to 0.6 pu and -0.6 pu
for each GFM to ensure a flat initialization of the positive-
sequence simulation. The time steps used in the simulations
are the same as used in Section V.
A. Fault-Induced Loss of Generation Event
In this scenario, a three-phase bolted fault was applied at
Bus 19 which is close to Synchronous Machine 33, and the
fault was cleared after 0.1 s. Because of the substantial voltage
drop, Synchronous Machine 33 tripped off after the fault. Fig.
19 (a) and (b) show the comparison of EMT and positive-
sequence simulation results for GFM 34 and GFM 30,
(a) (b) respectively. The GFM 34 uses the direct droop control and
Fig. 16 GFM response to a delayed clearing fault. (a) EMT simulations of two GFM 30 uses the multi-loop droop control. Because GFM 34
GFM controls. (b) Positive-sequence simulation. was close to the fault location, its terminal voltage V dropped
to 0.37 pu during the fault, resulting in the output current Io
reaching the limit Imax = 2 pu, as shown in Fig. 19 (a). The
GFM 34 quickly exited the current limiting mode after the
Imax=2 pu fault was cleared. As for GFM 30, because it was far away
from the fault location, its terminal voltage V only dropped to
0.8 pu and the output current Io did not reach the limit Imax
during the fault. Despite that some fast electromagnetic
transients were not captured by the positive-sequence models,
the positive-sequence simulation results aligned very well with
the EMT simulation results, as shown in Fig. 19 (a) and (b).

(a) (b)
Fig. 17 GFM output current and phase angle during the delayed clearing fault.
(a) EMT simulations of two GFM controls. (b) Positive-sequence simulation.

VI. COMPARISON IN A MODIFIED IEEE 39-BUS SYSTEM

To further validate the positive-sequence model in a bulk
power system environment, a rigorous comparison between
EMT and positive-sequence simulations was performed on a
modified IEEE 39-Bus System.

GFM37 Direct Droop GFM

37 Multi-Loop Droop GFM
GFM30
30 25
2 26 28 29 38
24
1 18 27
17
GFM35
16
39 3
15
4 35
14
5 12 21 22
6
11 13
19 23

7 31 20
36
8 10 34 33 (a) (b)
9 32
GFM32 GFM34 Fig. 19 Comparision of EMT and positive-sequence simulation results for the
modified IEEE 39-Bus System. (a) Responses of GFM 34. (b) Responses of
Fig. 18 The modified IEEE 39-bus system with 5 GFMs. GFM 30.

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

10

The EMT and positive-sequence simulations for each GFM The system model includes 22,000+ buses, 5,000+
and synchronous machine were compared, and all synchronous machines, and 1,000+ grid-following inverters.
comparisons achieved the same level of alignment as shown in The 2021 Heavy Summer case was selected, and the total load
Fig. 19. was around 150,000 MW. Because of its size, it would be
extremely difficult to conduct an EMT simulation for such a
B. Computational Time Summary
large system. Therefore, only positive-sequence simulations
Table IV summarizes the computational times. The were performed. A Pacific DC Intertie (PDCI) bi-pole outage
simulation runtime was 20 s. Because five IGBT-based GFM was selected as the contingency to examine the system
models were used and simulated at a time step of 5 μs, it took transient stability. Two cases were simulated. The first case
18,634 s to complete the EMT simulation. In contrast, it only simulated the original system, where synchronous machines
took 0.2 s to complete the positive-sequence simulation, which dominated the grid. For the second case, all the synchronous
shows the advantage of using positive-sequence simulations to machines were replaced by GFMs, creating a 100% inverter
study the transient stability of transmission systems. Note that case with high penetrations of GFMs. Simulations were
this paper aims to achieve a very rigorous comparison between performed in PSS/E with a time step of 4.167 ms, and the
EMT and positive-sequence simulations so that the IGBT- simulation runtime was 35 s. 25 generation units across the
based EMT models were used. The computational time of system were monitored. It took around 12.3 minutes to
EMT simulations is expected to be significantly reduced by complete Case 1, and around 18.3 minutes to complete Case 2.
using simplified models such as switching function models. Fig. 20 (a) and (b) present the simulation results of the 25
However, such studies are out of the scope of this paper. generations units for the two cases. By comparing these
TABLE IV results, it becomes evident that replacing all synchronous
Computational time (Simulation runtime: 20 s) machines with GFMs significantly improves the system
EMT simulation (IGBT-based model) Positive-sequence simulation damping. This case study shows that a) the developed
18,634 s 0.2 s
positive-sequence GFM model is numerically stable in a
practical bulk power system environment, and b) the high
VII. STUDY IN THE WESTERN INTERCONNECTION SYSTEM
penetration of GFMs can significantly enhance the system
With rigorous model validations performed in Section V damping compared to traditional synchronous machine-
and VI, the positive-sequence GFM model was finally studied dominated systems. Future work will explore additional
in the U.S. Western Interconnection system in this section. scenarios with varying levels of GFM penetration based on the
developed the GFM model.

VIII. CONCLUSIONS
As GFMs are being rapidly connected to the power grids
around the world, it is important to understand how GFMs
should be modeled in conventional positive-sequence
simulation tools for studying their impacts on the transient
stability of large-scale transmission systems. This paper
describes a positive-sequence model to represent to two
widely reported droop-controlled GFMs. Methods of how to
develop the equivalent voltage source behind impedance to
represent inverters with and without inner control loops,
modeling of P-f and Q-V droop controls, active and reactive
power limiting modeling, and modeling of fault current
limiting controls have been described in detail. The positive-
sequence model was compared against detailed EMT models
in both a single-GFM infinite-bus system and a modified
IEEE 39-bus system. Finally, the model has been tested on
the U.S. Western Interconnection system. The study results
show that the model has a good level of accuracy compared
to the detailed EMT models, and at the same time also
achieves a high computational efficiency, making it suitable
for large-scale transmission system simulation studies. Most
parts of the model described in the paper have been included
in a generic library positive-sequence grid-forming inverter
model, REGFM_A1, which was recently adopted by the US
(a) (b)
WECC and included in the model libraries of major
Fig. 20 Simulation results of the Western Interconnection system. (a) Case 1: commercial positive-sequence simulation tools, including PTI-
The original case. (b) Case 2: All synchronous machines repalced by GFMs.
PSS®E, GE-PSLF, PowerWorld Simulator, and TSAT,

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

11

marking the first publicly available grid-forming inverter Large Scale Grid Simulations," IEEE Transactions on Power Systems,
vol. 32, no. 1, pp. 765-773, 2017.
model that is included in all major commercial transient [21] V. A. Lacerda, E. P. Araujo, M. Cheah-Mañe, and O. Gomis-Bellmunt,
stability simulation tools used in North America. "Phasor Modeling Approaches and Simulation Guidelines of Voltage-
Source Converters in Grid-Integration Studies," IEEE Access, vol. 10, pp.
51826-51838, 2022.
IX. REFERENCES
[22] W. Du, Y. Liu, R. Huang, F. K. Tuffner, J. Xie, and Z. Huang, "Positive-
[1] R. Lasseter, Z. Chen, and D. Pattabiraman, "Grid-Forming Inverters: A Sequence Phasor Modeling of Droop-Controlled, Grid-Forming Inverters
Critical Asset for the Power Grid," IEEE Journal of Emerging and with Fault Current Limiting Function," in 2022 IEEE Power & Energy
Selected Topics in Power Electronics, pp. 1-1, 2019. Society Innovative Smart Grid Technologies Conference (ISGT), 2022,
[2] R. H. Lasseter et al., "CERTS Microgrid Laboratory Test Bed," IEEE pp. 1-5.
Transactions on Power Delivery, vol. 26, no. 1, pp. 325-332, 2011. [23] S. Roy and H. N. V. Pico, "Transient Stability and Active Protection of
[3] J. Matevosyan et al., "Grid-forming inverters: Are they the key for high Power Systems with Grid-Forming PV Power Plants," IEEE Transactions
renewable penetration?," IEEE Power and Energy Magazine, vol. 17, no. on Power Systems, pp. 1-1, 2022.
6, pp. 89-98, 2019. [24] D. Pattabiraman, R. H. Lasseter, and T. M. Jahns, "Transient Stability
[4] M. E. Elkhatib, W. Du, and R. H. Lasseter, "Evaluation of Inverter-based Modeling of Droop-Controlled Grid-Forming Inverters with Fault Current
Grid Frequency Support using Frequency-Watt and Grid-Forming PV Limiting," in 2020 IEEE Power & Energy Society General Meeting
Inverters," in 2018 IEEE Power & Energy Society General Meeting (PESGM), 2-6 Aug. 2020 2020, pp. 1-5, doi:
(PESGM), 2018, pp. 1-5. 10.1109/PESGM41954.2020.9281712.
[5] A. Crivellaro et al., "Beyond low-inertia systems: Massive integration of [25] Y. Gu, N. Bottrell, and T. C. Green, "Reduced-Order Models for
grid-forming power converters in transmission grids," in 2020 IEEE Representing Converters in Power System Studies," IEEE Transactions
Power & Energy Society General Meeting (PESGM), 2020, pp. 1-5. on Power Electronics, vol. 33, no. 4, pp. 3644-3654, 2018.
[6] W. Du et al., "Modeling of Grid-Forming and Grid-Following Inverters [26] B. Mahamedi and J. E. Fletcher, "The Equivalent Models of Grid-
for Dynamic Simulation of Large-Scale Distribution Systems," IEEE Forming Inverters in the Sequence Domain for the Steady-State Analysis
Transactions on Power Delivery, 2020. of Power Systems," IEEE Transactions on Power Systems, vol. 35, no. 4,
[7] P. F. Mayer, M. Gordon, W. C. Huang, and C. Hardt, "Improving grid pp. 2876-2887, 2020.
strength in a wide ‐ area transmission system with grid forming [27] W. Du, "Model Specification of Droop-Controlled, Grid-Forming
inverters," IET Generation, Transmission & Distribution, 2022. Inverters (REGFM_A1)," Pacific Northwest National Laboratory,
[8] D. Pattabiraman, R. H. Lasseter, and T. M. Jahns, "Comparison of Grid Richland, WA, United States, No. PNNL-35110, 2023.
Following and Grid Forming Control for a High Inverter Penetration [28] Siemens, PSS/E Software, 2022 [Online] Available:
Power System," in 2018 IEEE Power & Energy Society General Meeting https://new.siemens.com/global/en/products/energy/energy-automation-
(PESGM), 2018, pp. 1-5. and-smart-grid/pss-software/pss-e.html
[9] P. Brogan, T. Knueppel, D. Elliott, and N. Goldenbaum, "Experience of [29] GE Energy Consulting, PSLF Software, 2022 [Online] Available:
grid forming power converter control," in 17th Wind Integration https://www.geenergyconsulting.com/practice-area/software-products/pslf
Workshop, Stockholm, Sweden, 2018. [30] PowerWorld Corporation, PowerWorld Software, 2022 [Online]
[10] H. N. V. Pico and V. Gevorgian, "Blackstart Capability and Survivability Available: https://www.powerworld.com/
of Wind Turbines With Fully Rated Converters," IEEE Transactions on [31] Powertech Labs Inc, TSAT Software, 2022 [Online] Available:
Energy Conversion, vol. 37, no. 4, pp. 2482-2497, 2022. https://www.dsatools.com/tsat/
[11] W. Du, R. H. Lasseter, and A. S. Khalsa, "Survivability of Autonomous [32] M. C. Chandorkar, D. M. Divan, and R. Adapa, "Control of parallel
Microgrid During Overload Events," IEEE Transactions on Smart Grid, connected inverters in stand-alone AC supply systems," in 1991 IEEE
vol. 10, no. 4, pp. 3515-3524, 2019. Industry Applications Society Annual Meeting, 1991, pp. 1003-1009
[12] O. Schomann, "Experiences with large grid-forming inverters on various vol.1.
island and microgrid projects," in Hybrid Power Systems Workshop, [33] W. Du et al., "A Comparative Study of Two Widely Used Grid-Forming
2019. Droop Controls on Microgrid Small-Signal Stability," IEEE Journal of
[13] A. Roscoe et al., "Practical experience of providing enhanced grid Emerging and Selected Topics in Power Electronics, vol. 8, no. 2, pp.
forming services from an onshore wind park," in Proc. 19th Wind 963-975, 2019.
Integration Workshop, 2020, pp. 1-4. [34] N. Pogaku, M. Prodanovic, and T. C. Green, "Modeling, Analysis and
[14] L. Huang, H. Xin, Z. Wang, L. Zhang, K. Wu, and J. Hu, "Transient Testing of Autonomous Operation of an Inverter-Based Microgrid," IEEE
Stability Analysis and Control Design of Droop-Controlled Voltage Transactions on Power Electronics, vol. 22, no. 2, pp. 613-625, 2007.
Source Converters Considering Current Limitation," IEEE Transactions [35] A. Knobloch et al., "Synchronous energy storage system with inertia
on Smart Grid, vol. 10, no. 1, pp. 578-591, 2019. capabilities for angle, voltage and frequency stabilization in power grids,"
[15] T. Qoria, F. Gruson, F. Colas, G. Denis, T. Prevost, and X. Guillaud, in 11th Solar & Storage Power System Integration Workshop (SIW 2021),
"Critical Clearing Time Determination and Enhancement of Grid- 2021, pp. 71-78.
Forming Converters Embedding Virtual Impedance as Current Limitation [36] D. Duckwitz, A. Knobloch, F. Welck, T. Becker, C. Gloeckler, and T.
Algorithm," IEEE Journal of Emerging and Selected Topics in Power Buelo, "Experimental Short-Circuit Testing of Grid-Forming Inverters in
Electronics, vol. 8, no. 2, pp. 1050-1061, 2020. Microgrid and Interconnected Mode," in NEIS 2018; Conference on
[16] E. Rokrok, T. Qoria, A. Bruyere, B. Francois, and X. Guillaud, "Transient Sustainable Energy Supply and Energy Storage Systems, 20-21 Sept.
Stability Assessment and Enhancement of Grid-Forming Converters 2018 2018, pp. 1-6.
Embedding Current Reference Saturation as Current Limiting Strategy," [37] W. Du, Q. Nguyen, Y. Liu, and S. M. Mohiuddin, "A Current Limiting
IEEE Transactions on Power Systems, vol. 37, no. 2, pp. 1519-1531, Control Strategy for Single-Loop Droop-Controlled Grid-Forming
2022. Inverters Under Balanced and Unbalanced Faults," in 2022 IEEE Energy
[17] N. Baeckeland, D. Venkatramanan, M. Kleemann, and S. Dhople, Conversion Congress and Exposition (ECCE), 2022, pp. 1-7.
"Stationary-Frame Grid-Forming Inverter Control Architectures for [38] B. Fan and X. Wang, "Equivalent Circuit Model of Grid-Forming
Unbalanced Fault-Current Limiting," IEEE Transactions on Energy Converters With Circular Current Limiter for Transient Stability
Conversion, vol. 37, no. 4, pp. 2813-2825, 2022. Analysis," IEEE Transactions on Power Systems, vol. 37, no. 4, pp. 3141-
[18] Z. Jin and X. Wang, "A DQ-Frame Asymmetrical Virtual Impedance 3144, 2022, doi: 10.1109/TPWRS.2022.3173160.
Control for Enhancing Transient Stability of Grid-Forming Inverters," [39] P. Pourbeik et al., "Generic Dynamic Models for Modeling Wind Power
IEEE Transactions on Power Electronics, vol. 37, no. 4, pp. 4535-4544, Plants and Other Renewable Technologies in Large-Scale Power System
2022. Studies," IEEE Transactions on Energy Conversion, vol. 32, no. 3, pp.
[19] P. J. Hart, M. Gong, H. Liu, Z. Chen, Y. Zhang, and Y. Wang, "Provably- 1108-1116, 2017, doi: 10.1109/TEC.2016.2639050.
Stable Overload Ride-Through Control for Grid-Forming Inverters Using [40] Manitoba HVDC Research Centre, PSCAD Software, 2019 [Online]
System-Wide Lyapunov Function Analysis," IEEE Transactions on Available: https://hvdc.ca/pscad/
Energy Conversion, vol. 37, no. 4, pp. 2761-2776, 2022.
[20] D. Ramasubramanian, Z. Yu, R. Ayyanar, V. Vittal, and J. Undrill,
"Converter Model for Representing Converter Interfaced Generation in

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Power Delivery. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2024.3376245

12

Wei Du (S’12–M’14-SM’2022) received the Ph.D.

in Electrical Engineering from Tsinghua University, Songzhe Zhu received her Ph. D in Electrical
Beijing, China in 2014. He is currently a staff Engineering from Iowa State University in 2000. She
research engineer at Pacific Northwest National is currently a senior principal consultant with
Laboratory. He was a Visiting Student with the GridBright, a Qualus Company. She was previously a
University of Wisconsin-Madison from 2012 to senior advisor at California ISO. She worked
2013, where he also worked as a post-doctoral extensively in the areas of generation interconnection
researcher from 2016-2018. He worked as a and transmission planning at California ISO. She has
Research Engineer in the key Real Time Digital been actively involved in NERC and WECC
Simulation (RTDS) lab of power system in China workgroups to develop and enhance modeling of
Southern Power Grid Company from 2014 to 2016. His main areas of research inverter-based resources.
are control design, modeling, and simulation of power systems with high
penetration of power electronics. He currently serves as the Principal
Investigator of multiple projects funded by the U.S. Department of Energy
that investigate the impacts of inverter-based resources on the stability of
power systems. He is the co-lead of the Modeling and Simulation Area of the
Yuan Liu (Member, IEEE) received a MS and PhD
Universal Interoperability for Grid-Forming Inverters (UNIFI) Consortium.
in electrical engineering from Arizona State
He is the lead developer of the first generation of the US WECC generic grid-
University in 2012 and 2016. He is currently a staff
forming inverter model, REGFM_A1. The REGFM_A1 model has been
engineer at Pacific Northwest National Laboratory.
integrated into major commercial transient stability simulation tools including
His research interests include renewable energy
Siemens PTI-PSS®E, GE-PSLF, PowerWorld Simulator, and TSAT for
integration, load modeling, and power system
industry use.
simulation and control.

Quan Nguyen (S’15–M’19) received the B.E. degree

from the Hanoi University of Science and
Technology, Hanoi, Vietnam, in 2012, and the M.S.
and Ph.D. degrees from The University of Texas at
Austin, Austin, TX, USA, in 2016 and 2019, Frank Tuffner (M'08-SM'20) received the Ph.D.
respectively, all in electrical engineering. He has degree in electrical engineering from the University
been a Power System Engineer with The Pacific of Wyoming, Laramie, WY, USA in 2008. He is
Northwest National Laboratory since 2019. His currently a research engineer with the Pacific
research interests include production cost modeling, Northwest National Laboratory. His research
control, optimization, and simulation of power interests include signal processing applied to power
systems with high penetrations of inverter-based resources and flexible systems, EV-Grid integration, embedded control
HVDC and low-frequency transmission. devices, microgrids for resilience, and distribution
level modeling.

Song Wang (M’08-SM’19) obtained his master’s

degree in Electrical Engineering from Shenyang
University of Technology, China. With over 30 years Dr. Zhenyu (Henry) Huang (FIEEE-2017) received
of experience in transmission system planning, his B. Eng. from Huazhong University of Science
design, modeling, and compliance, he currently and Technology, Wuhan, China, in 1994, and Ph.D.
holds the position of Sr. Principal Engineer at degree from Tsinghua University, Beijing, China, in
Portland General Electric, USA. As a transmission 1999, both in Electrical Engineering. He is
planning expert, he provides consultation and Laboratory Fellow at Pacific Northwest National
recommendations to senior management. He is an Laboratory (PNNL), Richland, WA and holds a joint
Adjunct Professor at Portland State University, appointment of Research Professor at Washington
teaching Power System Stability and Power System I State University, Pullman, WA. He is also a Policy
classes. He has also been Chairman of the US Western Electricity Advisor on grid modernization for the US
Coordinating Council’s Modeling and Validation Subcommittee since 2016. Department of Energy (DOE)’s Undersecretary for
He leads the review, recommendation, development, and validation of power Science and Innovation. He was a Technical Advisor
system models to support reliability assessments to advance the industry. In at the DOE EERE Solar Energy Technologies Office (SETO) in 2019 – 2020.
the past, he has been a Point of Contact for multiple DOE projects. He is the At PNNL, Dr. Huang is leading the power electronics and renewable
Principal Investigator of a US DOE Grid Services Demonstration project. The integration portfolios. His research interests include high performance
project aims to demonstrate that grid-forming technologies enhance grid computing, data analytics, and optimization and control for inverter- and
stability and promote the utility adoption of grid-forming technologies. renewable-dominant power and energy systems. Dr. Huang has over 200 peer-
reviewed publications. Prior to joining PNNL in 2003, Dr. Huang conducted
extensive research on power system stability and harmonics at the University
of Alberta (Canada), McGill University (Canada), and the University of Hong
Jinho Kim (S’14−M’18) received the Ph.D. in Kong.
Electrical Engineering Department from Jeonbuk
National University, Jeonju, South Korea in 2018.
During his Ph.D., he was an assistant researcher with
the Wind Energy Grid-Adaptive Technologies
Research Center, Jeonju, South Korea, supported by
the Ministry of Science, ICT, and Future Planning,
South Korea. From 2018 to 2022, he worked as a
Research Fellow I in Electrical and Computer Engineering Department from
Auburn University, Auburn, AL, USA. He is currently a power system
research engineer at Pacific Northwest National Laboratory. His main areas of
research are control coordination, modeling, multi-time scale simulation for
power systems with high penetration levels of inverter-based resources. He’s
currently engaged in multiple DOE-funded projects.

Authorized licensed use limited to: Jorge Vega. Downloaded on April 14,2024 at 15:40:28 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.