Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

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.2024.3382843

Power Swing in Systems with Inverter-Based


Resources—Part II: Impact on Protection Systems
Mohamad-Amin Nasr and Ali Hooshyar , Senior Member, IEEE

Abstract—After introducing and analytically proving the dis- guidance for future IBR interconnection standards, parameter
tinctive dynamics associated with power swing of inverter-based tuning of inverter controllers, determining appropriate proce-
resources (IBRs) in Part I, Part II of this paper investigates the dures for the power swing detection component of relays, and
implications of such dynamics for power system protection. The
paper sheds light on some of the major consequences of IBRs’ the potential need for modifying these elements.
unique power swing patterns. The study employs theoretical The power swing frequency is directly reflected in the rate of
analysis and detailed PSCAD/EMTDC simulation models to change of apparent impedance, which is the primary param-
unveil the shortcomings of the conventional wisdom regarding eter examined by practical power swing detection methods
the power swing characteristics of IBRs. The paper offers new
discussed in [7], [8], [9], [10]. This poses several critical
insights into the operation of transmission line relays during
power swing conditions. This part of the study is supported by questions:
extensive testing of commercial relays. The findings of this paper 1) How can the dynamic model developed in Part I of the
will provide guidance for future IBR interconnection standards,
paper be utilized to derive analytical expressions that
the design of inverter control schemes, the power swing detection
methods used by relays, and the appropriate relay settings. describe the rate of change of impedance measured by
relays in systems with IBRs?
Index Terms—Inverter-based resources (IBRs), large-signal
disturbances, power swing, transmission line relays.
2) How can these expressions be employed to analytically
investigate the performance of power swing blocking
schemes of transmission line relays?
3) Do the behaviors of real-world, commercial relays repli-
I. I NTRODUCTION cate the findings from this paper?
4) What would be the system-wide implications of any
P ART I of this paper developed a novel state-space model
to analyze the power swing characteristics of systems
with inverter-based resources (IBRs) and synchronous ma-
potential relay malfunction caused by the IBRs’ uncon-
ventional power swing characteristics?
chines (SMs) [1]. The paper identified and mathematically 5) How can the power swing-related issues in systems with
represented the specific elements of an IBR control system IBRs be mitigated?
and the corresponding inverter parameters that affect power Each of the following five sections presents an answer to
swings after a fault. As a consequence, the developed model one of these questions.
enables assessing the combined impact of IBRs and SMs
on active power oscillations following a fault. This impact
analysis includes the range of frequency components in the II. R ATE OF C HANGE OF I MPEDANCE IN
voltage angle variations under different system conditions, G RIDS WITH IBR S
along with the origin of each component.
The first principle used by transmission line relays to
To ensure system stability after large-signal disturbances,
detect power swings is that the slow rotor dynamics of SMs
it is crucial to accurately distinguish between stable power
result in relatively small rates of impedance change for power
swings and faults [2]. For this purpose, transmission line relays
swings, whereas faults cause sudden changes in the apparent
primarily assess the swing frequency. The prevailing notion in
impedance seen by the relay [7], [8], [9], [10], [11]. To
the available literature is that the decrease in system inertia
determine the impact of IBRs on this approach, this section
caused by IBRs can raise the swing frequency and potentially
investigates the rate of change of impedance in grids with
hinder the reliable detection of power swing conditions by
IBRs. The following analysis shows how the dynamics of
relays [3], [4], [5], [6]. Part II of this paper will investigate if
voltage angle at IBR and SM buses—which were determined
the heightened swing frequency resulting from reduced system
by the state-space model in [1]—can be used to derive closed-
inertia poses the primary challenge for power swing detection
form relations for the rate of impedance change at any location
of line relays in systems with IBRs, or if there are other
in the system. These relations will form the basis for analyzing
aspects of the IBRs’ dynamics that could adversely affect these
the impact of IBRs on power swing detection techniques.
relays in a more significant manner. The findings will provide
To determine the rate of change of impedance, consider the
This work was supported by the Khalifa University of Science and transmission line in Fig. 1, which connects two subgrids that
Technology under Grant CIRA-2020-013. include a mixture of IBRs and SMs. In this figure, |Vx |̸ δx
The authors are with the Department of Electrical and Computer En-
gineering, University of Toronto, Toronto, ON M5S 3G4, Canada (email: and |Vy |̸ δy denote the voltages at buses x and y, respectively;
ma.nasr@mail.utoronto.ca; hooshyar@ece.utoronto.ca). Ix−y is the current flowing through the line; and Zx−y is

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

|V y | δ y |V x | δ x of impedance can be determined analytically if the voltage


angle dynamics at the two ends of the line in Fig. 1, i.e.,
Ix-y
Zx-y ∆δ(t) = δx (t) − δy (t), are known. The analytical expression
for ∆δ(t) is not straightforward to derive for grids with IBRs
Relay because the voltage angle dynamics are not solely determined
Subgrid 2 Subgrid 1
by SMs. Thus, the following demonstrates how the dynamic
Fig. 1. A transmission line connecting two subgrids, both of which include model developed in [1] can be adopted to find ∆δ(t) for any
a mixture of IBRs and SMs. given pair of bus voltages.
Consider the power system in Fig. 9 of [1]. This system can
be described by the set of nodal equations in (8).
the line’s impedance. During a balanced fault, the apparent     
impedance measured by the relay at bus x is In Ynn Ynm Ynk Vn
 SM   T  ′ 
Vx (t) Vx (t)  Im  =  Ynm Ymm Ymk   Em  . (8)
Zapp (t) = =  . (1) T T poc
Ix−y (t) Vx (t) − Vy (t) /Zx−y
IBR
Ik Ynk Ymk Ykk Ek
In (8), In denotes the vector of the net currents injected to
Denoting the difference between the bus voltage angles in Fig.
the first n buses of the system, which correspond to the non-
1, δx (t) − δy (t), by ∆δ(t) and considering the voltage of bus
generation buses. The voltages at these buses are expressed by
y as the reference phasor, Zapp (t) can be written as
  the vector Vn . The vector ISMm represents the currents injected
|Vx | cos(∆δ(t)) + j sin(∆δ(t)) by SMs to buses n+1 to n+m of the system. The voltages at

Zapp (t) =   Zx−y . these buses are expressed by the vector Em . Finally, the vector
|Vx | cos(∆δ(t)) + j sin(∆δ(t)) − |Vy | IIBR
k shows the currents injected by IBRs to buses n + m + 1
(2) to n + m + k of the system. The voltages at these buses are
Considering that |Vx | can be approximated by |Vy | [12], expressed by the vector Epoc k . Note that the bus admittance
Zapp (t) can be simplified to the following equation. This matrix in (8) is partitioned into nine submatrices. A submatrix
approximation upholds a conservative approach, which, as with i rows and j columns is represented by Yij . (·)T is the
substantiated in the following sections, maintains the integrity transpose of (·).
of the problem identification process. Since each load is modeled by an admittance embedded in
the bus admittance matrix, the net currents injected into the
cos(∆δ(t)) + j sin(∆δ(t))
Zapp (t) = Zx−y . (3) non-generation buses of the power system in Fig. 9 of [1] are
cos(∆δ(t)) − 1 + j sin(∆δ(t)) zero, i.e., In = 0 [13]. Therefore, multiplying the first row of
Multiplying this equation by the complex conjugate of the the bus admittance matrix in (8) by the vector of voltages in
denominator and considering the cos2 (·)+sin2 (·) = 1 identity, (8) element-wise yields the sum in (9).
(3) is re-written as ′
0 = Ynn Vn + Ynm Em + Ynk Epoc
k . (9)
1 − cos(∆δ(t)) − j sin(∆δ(t))
Zapp (t) = Zx−y . (4) Solving for Vn in the above equation results in
2 − 2 cos(∆δ(t))

Since transmission line impedances are mainly inductive, −1
Vn = −Ynn −1
Ynm Em − Ynn Ynk Epoc
k . (10)
Zx−y ≈ jXx−y , and (4) can be written as
In this equation, the admittance submatrices are known
1  sin(∆δ(t))  1 based on the grid configuration. The voltage magnitudes
Zapp (t) = Xx−y + j Xx−y . (5) ′
2 1 − cos(∆δ(t)) 2 in vectors Em = [|E1′ |̸ δ1′ . . . |Em ′ ̸
| δm ] and Epoc
′ T
k =
poc ̸ poc poc ̸ poc T
The real part of Zapp (t) in (5) varies with ∆δ(t), which, in [|E1 | δ1 . . . |Ek | δk ] are calculated using power
turn, changes with respect to time. Thus, the rate of change of flow analysis [1]. The voltage angles in these two vectors are
ℜ{Zapp (t)} is equivalent to the rate of change in the impedance state variables x1 = [δ1′ . . . δm ] and x3 = [δ1poc . . . δkpoc ]T
′ T

seen by the relay, as in (6). in [1] and can be obtained using (38) and (40) in [1].
Therefore, all the parameters on the right side of (10) are
d h ∂ ih d i
known. The voltage angles of all the buses are in vector
ℜ{Zapp (t)} = ℜ{Zapp (t)} ∆δ(t) . (6)
dt ∂∆δ(t) dt Vn = [|V1 |̸ δ1 . . . |Vn |̸ δn ]T on the left side of (10). The
Calculating the partial derivative of the real part of Zapp (t) voltage angles calculated using (38) and (40) in [1] and
with respect to ∆δ(t) in (5) and substituting the result into (10) can be plugged into (7) to derive the rate of change of
(6) yield impedance measured by a relay.
" # It is worth noting that the oscillations associated with the
d −Xx−y 1 hd i
inverters’ outer control loops, including the dc-link voltage
ℜ{Zapp (t)} = ·   ∆δ(t) . (7)
dt 4 sin2 ∆δ(t) dt control loop discussed in [1], fall under the category of
2
low-frequency oscillations [14]. During such oscillations, the
The equation presented above is of a generic nature, thereby voltage and current waveforms are still dominated by the
making it applicable to any grid, regardless of the generation fundamental-frequency component. Therefore, modeling the
type involved. This equation shows that the rate of change electrical network using quasi-static phasors, as in (8) to

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

(10), remains a justified practice for understanding the power 1 7 8 3


600 km
swing dynamics in systems with IBRs. This is particularly
relevant considering the significantly broader time scale of 2 5 4 100 km
9 200 km 300 km 300 km
the phenomenon examined in this study relative to the fast 100 km
switching frequency of power electronic inverters. 200 km 10
IBR 1
SM 1

III. P ROBLEM F ORMULATION SM 2 IBR 3 IBR 2


6
100 km 300 km
This section investigates how the rate of change of
12 14a
impedance derived in Section II is affected by IBRs. To this
150 MW
end, we will define different scenarios for the test system 100 km SM 4a
14b
in [1], which is shown in Fig. 2 for ease of reference. The R12-6
11 150 MW
focus is on the specified area, where SM 4 supplies 600 MW SM 4b
14c
through four transmission lines. During normal operation, lines
12-6 and 12-13 in Fig. 2 each transmit 250 MW, totaling SM 3 100 km 13 150 MW
50 km SM 4c
100 km 14d
500 MW. These lines are prone to power swings caused by
150 MW
SM 4. In the base scenario depicted in Fig. 2, SM 4 is an SM 4d
aggregated 600 MVA plant that consists of four 150 MVA SMs 500 MW 100 MW
PF = 0.95 PF = 0.95
[1], all operating at unity power factor. To define scenarios
that serve the objective of this section, we gradually change
Fig. 2. Single-line diagram of the study system highlighting an area prone
the source at bus 12 from SM to IBR. In Scenario 1, as to power swings.
illustrated in Fig. 3, we replace 25% of SM 4 with an IBR.
Each subsequent scenario depicted in this figure introduces an
by the SMs’ slow rotor dynamics. Thus, the STFT result in
additional 25% replacement of SM 4 with IBRs, culminating
Fig. 5(a) indicates that the dominant frequency of the curve
in Scenario 4, where the source at bus 12 is entirely based on
in Fig. 4(a) is about 1 Hz, which is a typical swing frequency
inverters. The per-unit specifications of the added IBRs match
for SMs [15]. This can also be visually verified in Fig. 4(a),
those described in Section II of [1]. The following analysis
where the time difference between the two consecutive peaks
investigates the frequency of power swings after a fault for
is 0.961 s, corresponding to a frequency of 1.04 Hz.
each of these scenarios and examines how this frequency is
By contrast, in Scenario 4, where the generation behind bus
reflected in the rate of change of impedance.
12 is entirely inverter-based, the voltage angle oscillations at
bus 12 are solely influenced by the IBRs. Thus, the frequency
A. Swing Frequency of oscillations in Fig. 4(e) is determined primarily by the IBRs’
Equation (7) indicates the dependence of the rate of change dc-link voltage control dynamics [1]. This is evident in Fig.
d
of impedance on dt ∆δ(t) and, consequently, the frequency 5(e), where the dominant frequency component of the curve in
of ∆δ(t) oscillations. To investigate the impacts of IBRs on Fig. 4(e) measures approximately 13 Hz. This swing frequency
the frequency of the voltage angle oscillations, consider a is significantly higher than the typical frequency range of 0.2
bolted balanced fault close to bus 13 on line 12-13 in Fig. ∼ 2.5 Hz for stable swings in conventional power systems
2 for the scenarios defined in Fig. 3. The fault happens at [15]. The frequency of oscillations in Fig. 4(e) is significantly
t = 2 s and is cleared 75 ms later at t = 2.075 s when line higher than even the maximum power swing frequency of 7
12-13 is tripped. This 75 ms window represents the critical Hz that can be accommodated by the power swing detection
clearing time, beyond which the SM 4 units undergo transient elements of commercial relays [8], [9], [16].
instability in the base scenario. After line 12-13 is tripped, line In Figs. 4(b) to 4(d) for Scenarios 1 to 3, which involve
12-6 transmits the entire 500 MW of power, and therefore, it a combination of IBRs and SMs behind bus 12, the voltage
becomes prone to power swings with large magnitudes. angle oscillations at bus 12 exhibit a mix of the characteristics
The difference between the angles of the voltages at buses shown in Figs. 4(a) and 4(e). Since the parameters of the dc-
agg agg
12 and 6 (i.e., δ12 − δ6 ) is calculated using (10) based on link voltage control (specifically kprp and kint in (16) of [1])
(38) and (40) of [1] and plotted in Fig. 4 for the above- are identical for all of these scenarios, the frequency of the
mentioned scenarios following the fault clearance (i.e., the component introduced by the IBRs to the curves in Figs. 4(b)-
analysis that follows is grounded in theoretical derivations in (d) remains equal to the dominant frequency of the curve in
[1] and Section II, rather than simulations). Unlike conven- Fig. 4(e). Only the magnitude of this component increases
tional systems for which the frequency of oscillations during commensurate with the share of IBRs. Both of these features
stable power swings varies in a narrow range of 0.2 ∼ 2.5 are corroborated by the STFT results shown in Fig. 5, where
Hz [15], the oscillations in Fig. 4(b)-(d) contain multiple the 13 Hz band is the dominant high-frequency component
frequencies, which change over time. Such multi-component and exhibits growth in magnitude from Scenario 1 in Fig.
non-stationary signals require time-localized spectral analysis, 5(b) to Scenario 3 in Fig. 5(d) within the [2.075 s, 2.250 s]
which is provided by the short-time Fourier transform (STFT) window. This consistency is also visually confirmed in Figs.
in Fig. 5. In the base scenario, where all the four sources at bus 4(b)-(d), where the time difference between the first two peaks
12 are SMs, the voltage angle oscillations at bus 12 are dictated in Scenarios 1 to 3 remains constant at 0.078 s, matching the

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

6 12 14a 6 12 14a
150 MW 150 MW
SM 4a 100 km SM 4a
100 km
14b
R12-6 150 MW R12-6 150 MW

Rest of SM 4b Rest of IBR 4a


the grid 14c the grid
13 13 150 MW
100 km 150 MW 100 km
50 km SM 4c 50 km IBR 4b
100 km 100 km
150 MW 150 MW
IBR 4a IBR 4c
500 MW 100 MW 500 MW 100 MW
PF = 0.95 PF = 0.95 PF = 0.95 PF = 0.95
(a) (c)
6 12 14a 6 12
150 MW 150 MW
100 km SM 4a 100 km IBR 4a
14b
R12-6 150 MW R12-6 150 MW
Rest of SM 4b Rest of IBR 4b
the grid the grid
13 13
100 km 150 MW 100 km 150 MW
50 km IBR 4a 50 km IBR 4c
100 km 100 km
150 MW 150 MW
IBR 4b IBR 4d
500 MW 100 MW 500 MW 100 MW
PF = 0.95 PF = 0.95 PF = 0.95 PF = 0.95
(b) (d)

Fig. 3. Area highlighted in Fig. 2 when IBRs replace the SM 4 units in: (a) Scenario 1; (b) Scenario 2; (c) Scenario 3; and (d) Scenario 4.

time difference shown in Fig. 4(e) for Scenario 4. Although the frequency of the SMs’ rotor oscillations intensi-
Figs. 5(b)-(d) indicate that once the 13 Hz swings subside fies due to the reduced inertia resulting from the replacement
around t = 2.25 s, the dominant frequency components of SMs with IBRs, the frequency range of the SMs’ rotor
transition to lower-frequency swings driven by the slower rotor oscillations in Figs. 4(b)-(d) remains within the commonly
dynamics of the SMs. Since the total inertia of the generation recognized 0.2 ∼ 2.5 Hz band for stable swing frequency of
behind bus 12 is determined based on the sum of the inertia SMs [15]. More importantly, this frequency range stays well
constants of the individual machines connected to this bus below the 7 Hz maximum frequency supported by the power
[17], reducing the number of SMs connected to bus 12 results swing detection elements of commercial relays [8], [9], [16].
in a decrease in the total inertia of the generation at this bus.
Reducing the inertia amplifies the swing frequency of the SMs
[18]. As a result, the frequency of the SMs’ rotor oscillations B. Rate of Change of Impedance
increases from 1/0.961 = 1.04 Hz in Fig. 4(a) (where the This subsection investigates the impact of the increase in
total inertia behind bus 12 reaches its maximum value of 8 the swing frequency discussed above on the rate of change of
s) to 1/0.731 = 1.37 Hz in Fig. 4(d) (where the total inertia apparent impedance seen by relay R12−6 in Fig. 3 after line
behind bus 12 is only 2 s). The potential increase in the swing 12-13 is tripped. Zones 1 and 2 of the phase distance element
frequency discussed in [3], [4], [5], [6] corresponds to this of this relay protect 80% and 120% of line 12-6, respectively.
specific increase, not the one explained in the previous two This relay uses the concentric characteristics delay to detect
paragraphs. Even if 95% of the generation at bus 12 becomes power swings [7], [8], [9], [10]. This technique measures
inverter-based, and the inertia of the SMs for the remaining the time it takes for the impedance locus to pass through
5% of the generation is H = 1 s (which is the minimum a zone delimited by two impedance characteristics, referred
practical value for SM inertia [19]), the swing frequency will to as the outer and middle characteristics [7]. If the elapsed
be only 2.5 Hz after the initial 13 Hz oscillations die out. time exceeds a certain threshold, such as 30 ms [10], the
The above finding also aligns with the simulation results in relay detects a power swing condition and blocks the distance
[3], where the swing frequency of SMs increases from 0.63 elements. The settings of the concentric technique for R12−6
Hz to only 0.77 Hz when the inertia of all the SMs is halved. are presented in Appendix A.

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

Normalized amplitude Normalized amplitude Normalized amplitude Normalized amplitude Normalized amplitude
1

of oscillation
0.5
∆ t = 0.961 s
0

(a) (a)
1

of oscillation
0.5

∆ t = 0.078 s ∆ t = 0.882 s
0

(b) (b)
1

of oscillation
0.5
∆ t = 0.078 s ∆ t = 0.775 s
0

(c) (c)
1

of oscillation
0.5
∆ t = 0.078 s ∆ t = 0.731 s
0

(d) (d)
1

of oscillation
0.5
∆ t = 0.078 s
0

(e) (e)

Fig. 4. Phase angle difference between the voltages of buses 12 and 6, Fig. 5. STFT of δ12 − δ6 curves in Fig. 4 for: (a) the base scenario; (b)
calculated using (10) based on (38) and (40) in [1] for: (a) the base scenario; Scenario 1; (c) Scenario 2; (d) Scenario 3; and (e) Scenario 4.
(b) Scenario 1; (c) Scenario 2; (d) Scenario 3; and (e) Scenario 4.

The difference between the right blinders of the outer and rates depicted in Fig. 6, the following two-step methodology
middle characteristics of the concentric method used by R12−6 is pursued: (i) the vertical axes in Fig. 6 are scaled to convert
is about 7 Ω in Appendix A. Therefore, the objective is to the unit of the rate of change of impedance from Ω/s, as given
determine if the rate at which the impedance changes during by (7), to Ω/(30 ms). This rescaling aligns the measurement
the discussed power swing scenarios is high enough to cause scale with the time window pertinent to this analysis; and (ii)
the impedance to traverse the 7 Ω distance between the two the moving average of the solid-line curves in Fig. 6 is calcu-
blinders in less than 30 ms, leading to erroneous detection of a lated over a 30 ms window to calculate the above-mentioned
fault by R12−6 . To calculate the instantaneous rate of change integral. The average curves are shown as dashed lines in Fig.
of impedance, the voltage angles shown in Fig. 4 are plugged 6. This two-step process allows for easy determination of the
into (7), yielding the solid-line curves depicted in Fig. 6. To impedance variation over the preceding 30 ms at any given
determine whether the impedance trajectory crosses the critical point in time. For instance, for the base scenario, Fig. 6(a)
distance of 7 Ω between the power swing blinders in less illustrates that within the first 30 ms following the fault (i.e.,
than or more than 30 ms, the integral of these instantaneous from t = 2.075 to t = 2.105 s), the impedance measured
rates should be computed over a 30 ms period; i.e., the solid- by R12−6 decreases by only 2.14 Ω. This value is reliably
line curves in Fig. 6, in their original form, do not directly below 7 Ω, enabling the successful detection of a power swing
determine whether the impedance locus passes through the condition by the relay in the base scenario.
power swing region within the 30 ms time delay. For Scenarios 1 and 2, the amplitude of the 13 Hz oscil-
To enhance the visual comprehension of the instantaneous lations added by the IBRs to the curves in Figs. 4(b)-(c) is

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

To assess the accuracy of the analytical approach presented


above, the scenarios depicted in Figs. 2 and 3 were simulated
using PSCAD/EMTDC. The simulations incorporated detailed
−2.14 models of the IBRs, SMs, and relay R12−6 . The settings of the
relay are the same as those mentioned in Appendices A and
B. Fig. 7 depicts the impedance plane of the relay obtained
from the simulations for all the aforementioned scenarios. The
(a) impedance trajectories in this figure were derived from voltage
and current phasors computed using a full-cycle discrete
Fourier transform, resulting in a one-cycle measurement delay.
It is worth noting that the inner characteristic, grayed out in
Fig. 7, is used for out-of-step tripping and falls outside the
−2.47
scope of this paper.
For the distance zones depicted in Fig. 7, the choice of
a quadrilateral characteristic primarily serves as an illustrative
(b) example to demonstrate the potential implications of the power
swing blocking misoperations revealed in the study. The core
findings of this study remain independent of the distance
characteristic used by the relay. In addition, the resistive
−6.09 reaches of the quadrilateral distance zones can be established
using various strategies, such as the zone overlapping practice
-7 Ω/(30 ms)
[9] (i.e., setting the resistive reaches of zones 1 and 2 the same)
or the coordination method detailed in [20]. For the examples
(c) that follow, we proceed with the zone overlapping strategy, as
recommended in [9], [10]. The detailed procedure for setting
the resistive reaches is discussed in Appendix B.
In Fig. 7, the impedance is time-labeled when it crosses
the outer and middle characteristics. White labels indicate
-7 Ω/(30 ms)
−12.49 the impedance entering the outer blinder without reaching
the middle characteristic. Orange labels represent the traversal
of the area between the blinders in more than 30 ms, while
(d) red labels indicate a passage faster than 30 ms. Figs. 7(a)-
(c) show that the impedance locus takes more than 80 ms to
cross the region between the outer and middle characteristics
in the base scenario, Scenario 1, and Scenario 2. This interval
-7 Ω/(30 ms) is significantly longer than the 30 ms threshold set for the
−21.66 relay in Appendix A. Thus, the relay successfully detects a
power swing condition in these three scenarios, verifying the
theoretical analysis in Figs. 6(a)-(c).
(e) In Scenario 3, the impedance trajectory passes through the
region between the outer and middle characteristics twice in
Fig. 6. Rate of change of impedance given by (7) using (10) for: (a) the base Fig. 7(d). This outcome is expected due to the relatively large
scenario; (b) Scenario 1; (c) Scenario 2; (d) Scenario 3; and (e) Scenario 4. amplitude of the first two peaks of the 13 Hz oscillations
in Fig. 4(d). The impedance enters the outer characteristic at
2.104 s and swiftly crosses the middle characteristic in less
insignificant. Therefore, the dashed curves in Figs. 6(b)-(c) than 13 ms at 2.117 s. Subsequently, the impedance locus exits
remain above −7 Ω/(30 ms), allowing the relay to successfully both the outer and middle characteristics before re-entering
detect a power swing condition. Conversely, Scenarios 3 and them again at 2.174 s and 2.189 s, respectively. In both cases,
4 in Figs. 4(d)-(e) exhibit relatively larger 13 Hz oscillations, the time required for the impedance locus to traverse between
as indicated by the STFT results in Figs. 5(d)-(e). Thus, the the outer and middle characteristics falls well below the 30
rate of impedance change measured by R12−6 is noticeably ms threshold set for the relay in Appendix A. Therefore, the
higher for these two scenarios in Figs. 6(d)-(e). In the first relay fails to detect the power swing condition in this scenario,
30 ms after the fault clearance, the change in the impedance consistent with the findings depicted in Fig. 6(d).
measured by R12−6 exceeds −12 Ω and −21 Ω in Scenarios In Scenario 4, this problem exacerbates, and the impedance
3 and 4, respectively. These changes are substantially larger obtained from the PSCAD/EMTDC simulation swiftly crosses
than −7 Ω, causing the impedance to cross the area between the area between the outer and middle characteristics in Fig.
the two blinders quite sooner than 30 ms. As a result, the relay 7(e) within just 8 ms, significantly below the 30 ms threshold
mistakes these power swings for a fault. specified in Appendix A. This outcome is also consistent

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

Impedance trajectory Impedance trajectory Impedance trajectory


50 50 50

tic

tic
tic

aracteris

aracteris
aracteris

40 40 40
istic

istic

istic
tic

tic

tic
character

character

character
30 30 30
aracteris

aracteris

aracteris
Outer ch

Outer ch
Outer ch

20 2.196 s 20 2.219 s 20 2.255 s


Inner ch

Zone 2 Zone 2

Inner ch

Inner ch
Zone 2 2.165 s
Middle

Middle

Middle
10 2.114 s 10 2.097 s 10
Zone 1 Zone 1 Zone 1 2.098 s

-30 -20 -10 10 20 30 -30 -20 -10 10 20 30 -30 -20 -10 10 20 30


-10 -10 -10 2.129 s

-20 -20 -20


-30 -30 -30

(a) (b) (c)

Impedance trajectory Impedance trajectory


50 50

tic
tic

aracteris
aracteris

40 40
istic

istic
tic

tic
character

character
30 30
aracteris

aracteris
Outer ch
Outer ch

20 2.117 s 2.122 s 2.114 s


20
2.104 s
Inner ch

Zone 2 Zone 2

Inner ch
Middle

Middle
10 Zone 1 10 Zone 1

-30 -20 -10 10 20 30 2.174 s -30 -20 -10 10 20 30


-10 -10
2.189 s
-20 -20
-30 -30
(d) (e)

Fig. 7. Impedance measured by the distance element of R12−6 in: (a) the base scenario; (b) Scenario 1; (c) Scenario 2; (d) Scenario 3; and (e) Scenario 4.

with the results obtained using theoretical analysis in Fig. IV. P ERFORMANCE E VALUATION OF
5(e), which indicates the maximum amplitude for the 13 Hz R EAL -W ORLD R ELAYS
oscillations, and Fig. 6(e), which illustrates an impedance
change rate exceeding −21 Ω/(30 ms). The 8 ms travel time is We conducted extensive laboratory experiments using com-
less than the minimum half-a-cycle setting allowed by certain mercial relays from various manufacturers to validate the
commercial relays, such as [8], for the power swing blocking aforementioned findings. Consistent results were obtained
time delay of the concentric technique. Although some other across all of these relays, irrespective of the manufacturer. This
manufacturers’ relays allow for a shorter time delay, they still section presents the outcomes obtained from two off-the-shelf
advise against reducing this delay below 30 ms [10]. relays, referred to as Commercial Relay 1 and Commercial
Note that the increased swing frequency resulting from Relay 2. These relays were updated with the latest firmware
reduced inertia, as discussed in [3], [4], [5], [6], and the last as of February 2024. The voltage and current waveforms of
paragraph of Section III-A, does not contribute to the failure PSCAD/EMTDC simulations, captured at the R12−6 location
to detect power swings in Scenarios 3 and 4. This increase in Figs. 2 and 3, were recorded as common format for transient
in swing frequency merely requires considering the actual data exchange (COMTRADE) files [21]. The relay test set
system inertia when determining the settings of the concentric detailed in [22] was used to supply these input COMTRADE
method. Moreover, as shown in Section III-A, this increase files to the relays.
is small enough to keep the required changes in the relay Fig. 8 illustrates the relay recordings, which are distinct
settings within the setting ranges commonly used in conven- from the recordings at the terminals of IBRs and SMs; this
tional systems. By contrast, the high-frequency component distinction is elaborated at the end of this section. These
introduced to the voltage angle oscillations by the inverter recordings show the currents on the secondary side of the
control system renders the first principle of the relays’ power current transformer (CT), voltages on the secondary side of the
swing detection methods ineffective. Consequently, addressing voltage transformer (VT), and the relay word bits, for the base
this issue requires more than just minor adjustments to the scenario. Word bits POWER SWING OUTER and X7ABC
relay settings. in Figs. 8(c)-(d) pick up when the impedance locus enters

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

the outer characteristic; word bits POWER SWING MIDDLE


and X6ABC in Figs. 8(c)-(d) pick up when the impedance

Current (A)
locus enters the middle characteristic; and word bits POWER
2
SWING BLOCK and OSB in Figs. 8(c)-(d) are asserted to 0
logical 1 and latched for 0.5 s if the relay detects a power -2
swing condition. PH DIST Z1 OP and Z1P in Figs. 8(c)-(d)
denote the operation of the zone-1 phase distance elements of 2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
Time (s)
these relays (the former stays latched until the reset condition (a)

Voltage (V)
is met or the element is blocked by the power swing blocking 100
function of the relay); and TRIPBUS 1 OP and TRIP word 0
bits in Figs. 8(c)-(d) show the relays’ trip signals.
-100
The difference between the pick-up times of the outer and
middle characteristics in Figs. 8(c)-(d) is in the range of 80- 2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
Time (s)
100 ms. Thus, both of these relays successfully detect a power (b)
POWER SWING OUTER
swing condition in the base scenario. This prevents the zone-1 POWER SWING MIDDLE
POWER SWING BLOCK
distance element from operating, and neither of these relays PH DIST Z1 OP
trip in this scenario. This outcome aligns with the theoretical TRIPBUS 1 OP
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
analysis depicted in Fig. 6(a) and the simulation results shown Time (s)
in Fig. 7(a). A similar consistency between the results of (c)
X7ABC
Section III and the operation of commercial relays is observed X6ABC
for Scenarios 1 and 2 as well. Figs. 9 and 10 show that OSB
Z1P
both relays can effectively detect a power swing condition, TRIP
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
preventing the operation of their distance elements.
Time (s)
In Scenario 3, Fig. 7(d) demonstrates that the impedance (d)
traverses the area between the outer and middle characteristics
Fig. 8. COMTRADE recordings of commercial relays for the base scenario
twice, both within a cycle, before entering the relay’s zone 1. of Fig. 2: (a) CT secondary currents; (b) VT secondary voltages; (c) Word
Testing of commercial relays produced the same result. Figs. bits of Commercial Relay 1; and (d) Word bits of Commercial Relay 2.
11(c)-(d) show that the middle characteristic picks up twice,
both less than a cycle after the outer characteristic is activated.
Consequently, both relays in Figs. 11(c)-(d) fail to detect the a violation of the IBRs’ 1.5 pu current limit considered in
power swing condition. As a result, the impedance enters √ (which is equivalent to a combined total of 1.5 ×
this study
zone 1, causing both PH DIST Z1 OP and Z1P to activate 600/( 3 × 230) = 2260 A). This can be further verified in
in Figs. 11(c)-(d). Consequently, both relays erroneously trip Fig. 14, where the current of each individual inverter within the
line 12-6. Figs. 12(c)-(d) illustrate a similar relay malfunction IBR plants connected to bus 12 remains below the inverters’
in Scenario 4, where the middle characteristic picks up only 1.5 pu current limit not only during the fault (i.e., from t = 2
8 ms after the outer characteristic is activated. The relay fails to t = 2.075 s) but also during the power swing condition
to detect the power swing condition, while PH DIST Z1 OP from t = 2.075 s onwards.
and Z1P are asserted, resulting in the tripping of line 12-6. In addition, the temporary overvoltage condition immedi-
As mentioned at the beginning of this section, the currents ately after the fault clearance in Fig. 13(b) is due to the reactive
shown in Fig. 12(a) are the currents measured at the location current injection of the IBRs during the fault, as per [23].
of relay R12−6 , not at the terminal of the IBRs connected to This overvoltage is normal in IBR systems during post-fault
bus 12. In addition, the currents in Fig. 12(a) are the currents conditions [24]. This temporary overvoltage condition does not
on the secondary side of the CT connected to relay R12−6 . As cause any overvoltage tripping since it is mitigated by surge
discussed in Appendix A, the CT ratio for this relay is 600:1. arrestors within the IBR plant facility [25].
The CT nominal tap for this relay is determined based on the
rated current of line 12-6. Due to the load connected to bus V. S YSTEM -W IDE I MPACTS
12 and the presence of line 12-13 during normal conditions, The last two sections demonstrated that the control system
the rated current of line 12-6 is less than half of the IBRs’ dynamics of the IBRs prevented relay R12−6 from detecting
total current injected to bus 12. Before the fault, line 12-6 stable power swings. As a result, 500 MW of generation
transmits only half of the total power exported from bus 12. transmitted through line 12-6 was lost. This section examines
Therefore, since the total rated current
√ of the IBR generation the significance of this issue and its system-wide impacts for
outflowing from bus 12 is 500/( 3 × 230) = 1255 A, the the grid shown in Figs. 2 and 3. This system is based on the
nominal current through line 12-6 is about 600 A, which is 1 CIGRE high-voltage benchmark, derived from an actual North
A on the secondary side of the CT connected to relay R12−6 . American system [26]. The significant imbalance between
The above indicates that the increased current magnitude in the generation and load caused by the aforementioned relay
Fig. 12(a), whose zoomed-in view is displayed in Fig. 13(a), malfunction has a severe impact on the frequencies of SMs 1-3
is due to the overloading of line 12-6 after the tripping of in the system. Fig. 15(a) demonstrates that following the fault,
line 12-13 at t = 2.075 s. This increase does not indicate the frequencies of these SMs decrease to 57.33 Hz before their

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

Current (A)

Current (A)
2 2
0 0
-2 -2

2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
Time (s) Time (s)
(a) (a)

Voltage (V)
Voltage (V)

100 100
0 0
-100 -100

2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
Time (s) Time (s)
(b) (b)
POWER SWING OUTER POWER SWING OUTER
POWER SWING MIDDLE POWER SWING MIDDLE
POWER SWING BLOCK POWER SWING BLOCK
PH DIST Z1 OP PH DIST Z1 OP
TRIPBUS 1 OP TRIPBUS 1 OP
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 Time (s)
Time (s) (c)
(c)
X7ABC
X7ABC X6ABC
X6ABC OSB
OSB Z1P
Z1P TRIP
TRIP
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 Time (s)
Time (s) (d)
(d)

Fig. 9. COMTRADE recordings of commercial relays for Scenario 1 of Fig. Fig. 11. COMTRADE recordings of commercial relays for Scenario 3 of Fig.
3(a): (a) CT secondary currents; (b) VT secondary voltages; (c) Word bits of 3(c): (a) CT secondary currents; (b) VT secondary voltages; (c) Word bits of
Commercial Relay 1; and (d) Word bits of Commercial Relay 2. Commercial Relay 1; and (d) Word bits of Commercial Relay 2.
Current (A)

Current (A)

2 2
0 0
-2 -2

2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
Time (s) Time (s)
(a) (a)
Voltage (V)

Voltage (V)

100 100
0 0
-100 -100

2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
Time (s) Time (s)
(b) (b)
POWER SWING OUTER POWER SWING OUTER
POWER SWING MIDDLE POWER SWING MIDDLE
POWER SWING BLOCK POWER SWING BLOCK
PH DIST Z1 OP PH DIST Z1 OP
TRIPBUS 1 OP TRIPBUS 1 OP
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 Time (s)
Time (s) (c)
(c)
X7ABC X7ABC
X6ABC X6ABC
OSB OSB
Z1P Z1P
TRIP TRIP
2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675 2.075 2.275 2.475 2.675 2.875 3.075 3.275 3.475 3.675
Time (s) Time (s)
(d) (d)
Magnified in Fig. 13

Fig. 10. COMTRADE recordings of commercial relays for Scenario 2 of Fig. Fig. 12. COMTRADE recordings of commercial relays for Scenario 4 of Fig.
3(b): (a) CT secondary currents; (b) VT secondary voltages; (c) Word bits of 3(d): (a) CT secondary currents; (b) VT secondary voltages; (c) Word bits of
Commercial Relay 1; and (d) Word bits of Commercial Relay 2. Commercial Relay 1; and (d) Word bits of Commercial Relay 2.

speed governors begin to respond. This frequency falls well all SMs in the grid may trip. To prevent cascade tripping, Fig.
below the under-frequency trip curve of the NERC Standard 15(b) illustrates that shedding 500 MW of the load at bus 6 in
PRC-006-2 [27]. Consequently, without appropriate measures, Fig. 2—which is equivalent to approximately 25% of the total

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

10

Current (A)
4
2
0
-2
-4
2.025 2.075 2.125 2.175 2.225 2.275 2.325
Time (s)
(a)
Voltage (V)

200
100
0
-100 Fig. 14. Instantaneous currents of each individual inverter within the IBR
-200 plants connected to bus 12 before, during, and after the fault of Fig. 12.
2.025 2.075 2.125 2.175 2.225 2.275 2.325
Time (s)
(b)
POWER SWING OUTER 57.33 Hz
POWER SWING MIDDLE
POWER SWING BLOCK
PH DIST Z1 OP
TRIPBUS 1 OP
2.025 2.075 2.125 2.175 2.225 2.275 2.325
Time (s)
(c)
X7ABC (a)
X6ABC
OSB
Z1P 58.53 Hz
TRIP
2.025 2.075 2.125 2.175 2.225 2.275 2.325
Time (s)
(d)
Under-frequency load shedding begins
Fig. 13. Zoomed-in view of Fig. 12: (a) CT secondary currents; (b) VT
secondary voltages; (c) Word bits of Commercial Relay 1; and (d) Word bits
of Commercial Relay 2.
(b)

system load—can mitigate the decline in frequency and limit Fig. 15. Frequency of SMs 1-3 after the loss of 500 MW of power transmitted
by line 12-6: (a) if no under-frequency load shedding scheme is implemented;
it to 58.53 Hz. The prevention of such system-wide major and (b) if an under-frequency load shedding scheme is implemented.
disturbances requires adequate awareness of the power swing
characteristics of IBRs.

VI. D ISCUSSION The above demonstrates that the primary focus in [23]
A. Regulatory Frameworks and Fault Recovery of IBRs is to mitigate the oscillations of IBRs in weak grids by
imposing a ramp limit on IBRs’ active power. By contrast, the
The phenomenon investigated in this paper occurs following priority in [28] is given to the swift recovery of IBRs’ active
the clearance of a fault. The requirements for the operation power, explicitly prohibiting ramp limitations. The NERC
of IBRs after the fault clearance are not consistent across recommendation in [28] is consistent with Section 10.2.3.3
all recent regulatory frameworks, such as the IEEE Standard of the VDE grid code as well [29]. This section discusses
2800-2022 [23], and the more recent NERC recommendations how adherence to each of these requirements influences the
in [28], most notably in power swing behavior of IBRs and explores how to combine
1) Subclause 7.2.2.6 of [23], where it is mentioned “The the advantages of both approaches while avoiding the power
active power recovery time shall be configurable within swing-related issues identified in this paper.
a range between 1 s and 10 s. The default active power Consider Scenario 4, illustrated in Fig. 3(d), as the worst-
recovery time is 1 s; however, in weak grids, in order case scenario, where the sources connected to bus 12 are
to reduce oscillatory behavior of the IBR plant upon entirely inverter-based. Fig. 16(a) depicts the active power
fault recovery and maintain system stability, it may be transmitted through line 12-6 before, during, and after the fault
desirable to reduce the average rate of active power in this scenario. As discussed in [1], the IBRs’ active power is
recovery in consultation with the TS owner;” and regulated by the dc-link voltage control loop shown in Fig. 2
2) Recommendation 4.d of [28], where it is mentioned of [1] before and after the fault. This structure, including the
“Coordinate with inverter manufacturers and power proportional-integral (PI) compensator, is well established in
plant controller manufacturers to ensure that facility the inverter control literature [30], [31], [32]. The post-fault
control modes, fault ride through modes and parameters, oscillations of the power curve in Fig. 16(a) are consistent with
and protections are set and coordinated according to those of the voltage angle curve in Fig. 4(e), leading to the
the following principles: Facility output returns to pre- relay malfunction and incorrect tripping of line 12-6. Having
disturbance active power levels as soon as possible said that, the power curve in Fig. 16(a) does comply with the
without any artificial ramp rate limit or delay imposed recommendations in [28] regarding the fast recovery of active
by the power plant controller.” power following a fault. Fig. 16(b) shows the active power

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

11

Fig. 17 illustrates that all the closed-loop poles of the system


will be on the real axis, and thus, the control loop’s transient
response will exhibit no oscillations. Since τi = 2 × 10−4
s in [1], the gain crossover frequency of Gop (s) should be
adequately smaller than 5000 rad/s, e.g., 5000/10 = 500
rad/s. Figs. 18(a) and 18(b) indicate that at the gain crossover
(a)
frequency of 500 rad/s, the compensated loop has a phase
margin of 73.5°, which is adequately large for a well-damped
response. The same bode plot in the absence of the proposed
lead filter for the control loop in Fig. 2 of [1] is depicted in
Fig. 19. The magnitude plot in Fig. 19(a) indicates that at zero
frequency, the magnitude of the open-loop transfer function is
large enough to obtain a zero steady-state error for this loop.
Thus, the design of the PI compensator of the loop in Fig. 2 of
(b) [1] meets the performance characteristic of a PI compensator.
However, the phase plot in Fig. 19(b) shows that the phase
margin of this loop is only 19.2°. This small phase margin is
the root cause of the oscillatory behavior of the inverters upon
fault recovery discussed earlier in the paper.
If the proposed compensator is utilized in Scenario 4 of Fig.
3(d), the IBRs’ active power can be restored immediately after
the fault clearance without any oscillations in Fig. 16(c). It is
(c) worth noting that the short-circuit ratio (SCR) at bus 12 of
Fig. 3(d) is only about 1.2 when line 12-13 is out of service
Fig. 16. Active power of the IBRs obtained from the PSCAD/EMTDC
simulation of Scenario 4 of Fig. 3(d) if the IBRs: (a) quickly restore active (i.e., after the fault clearance). In other words, the quick and
power using a typical PI compensator without power ramp limitation; (b) oscillation-free active power recovery is realized despite the
gradually restore active power using a typical PI compensator with power weak connection to the grid. This result indicates that, in future
ramp limitation; and (c) quickly restore active power using the proposed
compensator. revisions of [23], the requirement for the 1 pu/s ramp limit on
the IBRs’ active power recovery in Subclause 7.2.2.6 can be
eliminated and replaced with a threshold for the active power’s
curve for the same scenario when Subclause 7.2.2.6 of [23] damping ratio upon fault recovery.
is followed and a 1 pu/s ramp limit is imposed on the active
power recovery of the IBRs. This approach prevents power
B. Measures for Enhancing the Relay Security
swing but at the expense of an excessively prolonged delay in
the active power recovery of the IBRs. This poses a significant Section VI-A discussed how an inverter’s control system
problem for systems with high IBR penetration and can have can be modified to prevent the problems revealed in this paper
system-wide stability implications—the primary reason for the for power swing detection. An alternative approach to solving
updated requirements in [28]. this problem would be modifying the common practices for
The ideal objective should be the fast active power recovery power swing detection in today’s protection systems. This
of IBRs while power oscillations are prevented. To this end, could entail either (i) modifying the existing power swing
consider the control loop depicted in Fig. 2 of [1]. The PI blocking algorithms; or (ii) adding extra layers of security
compensator of this loop is intended to achieve a zero steady- to distance elements. The former necessitates new research
state error, rather than a pre-determined transient performance and development of new power swing detection methods. The
[33]. Consequently, a more elaborate compensator can be latter could involve introducing time delays to enhance the
employed to expedite the response of this loop without any security of distance elements in the vicinity of IBRs [34]. The
oscillations while still achieving a zero steady-state error. An earlier sections of the paper demonstrated that the oscillations
effective and straightforward approach is to substitute the PI exhibited by IBRs tend to be rapid but brief. Thus, the addition
compensator of Fig. 2 in [1] with an integrator and a lead of time delays to the operation of zone-1 distance elements can
filter, as in (11). prevent their inadvertent tripping during power swings in IBR
systems. This added delay can be achieved either by directly
−2
    
s+z 1 1
Gop (s) = −kc . (11) extending the pickup delay of the distance elements or by
s+p s Cdc s employing alternative methods, such as the capacitive voltage
Gop (s) is the modified open-loop transfer function of the transformers (CVT) transient detection schemes, which inher-
control loop; kc is the compensator gain; and z and p denote ently slow down the relay operation [35].
the zero and pole of the lead filter, respectively, added to However, it is important to note that such time delays
improve the control loop’s transient response. To meet the are not the ultimate solutions for all scenarios, and they
above objective, these parameters must be tuned properly. For come with potentially significant caveats depending on the
example, if kc = 5467.9 Ω−1 , z = 34.06, and p = 2226, system condition. The findings of this paper are not limited

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

12

Closed-loop pole {s} for the zone-1 distance element, cannot dynamically adapt
Open-loop pole 40 to weather variations. As a consequence, the five scenarios
Open-loop zero
20 depicted in Figs. 2 and 3 should not be necessarily seen as
{s} five different systems each with its own settings for R12−6 .
-2000 -1500 -1000 -500 Instead, the single-line diagrams in these figures can represent
-20
five potential generation mixes for a single system with one
-40 set of settings for R12−6 .
For the condition shown in Fig. 3(d), a time delay for zone 1
Fig. 17. Root locus plot of the open-loop transfer function of the dc-link
voltage control loop with the proposed lead filter.
of R12−6 does not impact the stability of the generation units
connected to bus 12. However, consider a scenario where the
power of the IBRs reduces by 25%, necessitating the dispatch
Magnitude (dB)

100 of an SM, such as a natural gas plant, similar to the case


50 illustrated in Fig. 3(c). In this figure, the sources connected to
bus 14a include three 150 MW IBRs and one 150 MW SM
0
operating at its full rated capacity (i.e., the input mechanical
-50
torque is near its maximum value).
-100 To asses the impact of the delayed tripping, consider a fault
10 0 10 1 10 2 10 3 10 4 10 5
Frequency (rad/s) close to bus 13 on line 12-13 and a 50 ms time delay for
zone 1 of relays R12−6 and R12−13 . Considering the breakers’
Phase (degree)

(a)
-90
operation time and the time it takes for the impedance to
-120 enter zone 1 after the fault inception, suppose that the faulted
-150 Phase margin = 73.5 line, i.e., line 12-13, is removed from the system in 100
ms (which does not represent the worst-case scenario for the
-180 breaker opening time). Fig. 20 shows the measurements at the
10 2 3 4 5
10 0 10 1 10 10 10
Frequency (rad/s) terminal of the SM connected to bus 14a in Fig. 3(c). The SM
(b) undergoes transient instability as a result of the additional time
delay added to the zone-1 distance element. However, if the
Fig. 18. Bode plot of the open-loop transfer function of the dc-link voltage
control loop with the proposed lead filter: (a) Magnitude and (b) Phase.
fault is cleared in 75 ms (as demonstrated in the case studies
presented earlier), the SM can successfully recover from the
fault, as shown in Fig. 21.
The above shows the need for caution when introducing
Magnitude (dB)

100
time delays to the operation of distance elements: Depending
50 on the system configuration, this delay may solve one problem
0 while creating another. It should be noted that the scenario
-50 illustrated in Fig. 20 is only one example of problematic
-100 scenarios for delayed operation of zone 1. Even if all sources
10 0 10 1 10 2 10 3 10 4 10 5 behind the relay are inverter-based, delaying zone 1 could
Frequency (rad/s) potentially affect the stability of SMs in other parts of the
Phase (degree)

(a) system if the relay is part of a pilot protection scheme.


-90
The adverse effects of IBRs on relays can be mitigated by
-120
Phase margin = 19.2 adjusting either the IBR control systems or the protection
-150 systems. For the specific problem revealed in this paper,
-180 comparing the above-mentioned challenges with the relatively
1 2 3 4 5
10 0 10 10 10 10 10 clear-cut solution outlined in Section VI-A indicates that
Frequency (rad/s) altering the inverter’s control system presents a more straight-
(b) forward solution. This adjustment can be mandated through
Fig. 19. Bode plot of the open-loop transfer function of the dc-link voltage technical regulations governing IBR interconnections, such as
control loop without the proposed lead filter: (a) Magnitude and (b) Phase. grid codes or IBR interconnection agreements.

VII. C ONCLUSION
to direct tie-lines of IBRs; these findings are actually mostly
applicable to large transmission corridors of systems with Using the dynamic model developed in Part I, Part II of
high penetration of IBRs. Consequently, reducing the speed this paper analyzed the rate of change of impedance measured
of distance relays can compromise the system stability when by transmission line relays during power swings in systems
the source on one end of a transmission corridor is a mix of with IBRs. The findings revealed that the dc-link voltage
SMs and IBRs. For such systems, the generation mix varies control dynamics of IBRs can significantly amplify the rate
with weather conditions and the availability of renewable of impedance change, surpassing the high rate of impedance
energy sources. However, relay settings, including time delays change caused by the low inertia of IBR systems. These rapid

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

13

settings for the impedances presented below are expressed in


terms of secondary ohms. The relay’s voltage transformer ratio
(VTR) and the current transformer ratio (CTR) are 2090.90
and 600, respectively.

A. Outer Characteristic
(a) The blinders of the outer characteristic should be set below
the impedance associated with the maximum line loading,
multiplied by a safety factor [10]. To this end, first consider
the 600 MW total generation behind bus 12 in Fig. 2. Given
the 100 MW load connected to the same bus, the total power
transmitted through lines 12-6 and 12-13 is 500 MW. The
maximum line loading for line 12-6 occurs when line 12-13
(b)
is out of service, i.e., when the entire 500 MW is transmitted
through line 12-6. For this emergency loading condition, the
Fig. 20. Measurements at bus 14a of the system of Fig. 3(c) when the
operation of the zone-1 distance element of the relay that protects line 12-13
impedance measured by relay R12−6 is calculated using the
is delayed by 50 ms and the fault is cleared in 100 ms: (a) Instantaneous relation in (12) [11].
currents; and (b) Instantaneous voltages.   2  
Sbase Vbase CTR
Zload =
Pline Sbase VTR
 2
100 [MVA] 230 [kV] 600
= . .
500 [MW] 100 [MVA] 2090.90
= 30.36 Ω. (12)
Based on the above load impedance and considering a 0.9
(a) safety factor [10], the right blinder of the outer characteristic
is set to
Right blinder = (0.9)(30.36)
= 27.32 Ω. (13)
The left blinder is set to the negative of the right blinder,
as recommended in [10]. In addition, the settings for the top
(i.e., forward reach) and bottom (i.e., reverse reach) of the
(b)
outer characteristic are, respectively, chosen as 45.91 Ω and
Fig. 21. Measurements at bus 14a of the system of Fig. 3(c) when the fault is −28.26 Ω, which fall within the range recommended in [37],
cleared in 75 ms: (a) Instantaneous currents; and (b) Instantaneous voltages.
that is, at least two to three times the distance element’s reach.

variations in impedance pose a potential risk of relay mal- B. Middle Characteristic


functions. The theoretical derivations presented in the paper To set the blinders of the middle characteristic, it is rec-
were substantiated through detailed PSCAD/EMTDC simula- ommended by [36] to set these blinders outside the largest
tions and laboratory experiments involving commercial relays. applicable tripping zone. However, in [10], it is recommended
Additionally, the study highlighted the substantial system-wide to set these blinders to about 80% of the outer characteristic
impacts associated with power swings of IBRs. To address blinders. To reconcile these objectives, we set the right blinder
these challenges, the paper proposed modifications to the of the middle characteristic to 22.38 Ω, which is 120% of the
existing IBR standards to mitigate active power oscillations resistive reach of the distance zones in Fig. 7 and 81% of
while achieving prompt power recovery. The paper concluded the right blinder of the outer characteristic calculated above.
by presenting practical recommendations for enhancing the The left blinder of the middle characteristic is similarly set to
relay security against the unique challenges posed by power the negative of its right blinder. To ensure that the middle
swings in systems with IBRs. characteristic is inside the outer characteristic, the settings
for the top and bottom of the middle characteristic are,
A PPENDIX A respectively, chosen as 37.3 Ω and −18.65 Ω.
S ETTINGS C ALCULATIONS FOR THE P OWER S WING
B LOCKING E LEMENT OF R12−6 C. Power Swing Blocking Time Delay
This appendix details the procedure for configuring the Faster power swings can be detected more effectively by
power swing detection element of relay R12−6 following the reducing the power swing blocking time delay. The minimum
industry practices and guidelines in [10], [36], [37]. The recommended threshold for this time delay, as per [10], is 30

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

14

ms. However, as discussed in [10] and elaborated further in the reactance reach of zone 1 is set to 80% of the reactance
[37], this time delay is system-specific and depends largely of line 12-6 in Fig. 2; i.e.,
on the oscillation frequency during power swings. Therefore,  
the recommended 30 ms threshold should be checked with rch 600
X1 = (0.8)(47.87)
the system’s oscillation frequency to ensure it is neither too 2090.90
short nor too long. The relation for determining this time delay = 10.99 Ω. (16)
based on the swing frequency is given by (14) [37].
Therefore, based on the guideline in [10], R1rch can be as high
δmid − δout
Time delay = fnom , (14) as 3 × 10.99 = 32.97 Ω. In addition, as shown in (12), the
360fs load impedance associated with the maximum line loading for
where δmid and δout are the middle and outer blinder power line 12-6 in Fig. 2 is 30.36 Ω. Therefore, to avoid interference
angles, respectively, in degrees; fnom is the system’s nominal with the load impedance, the upper limit for R1rch should be
frequency in Hz; and fs is the swing frequency in Hz. The 30.36 Ω. Setting R1rch to this upper limit, however, brings
time delay obtained using (14) is expressed in cycles. The about two challenges: (i) the power swing blinders cannot be
method for obtaining δmid and δout is outlined in [37]. Based placed between the distance zone characteristic and the load
on this method and the blinder values calculated in the last two region; and (ii) the phase angle errors in the polarizing current
subsections, δmid and δout are 72.31° and 61.81°, respectively. of the distance elements are not accounted for.
Extensive transient stability analysis performed on the sys- To address the challenges discussed above, we follow the
tem of Fig. 2 reveals that the highest oscillation frequency of guideline in [38] and recalculate R1rch . To make the relay
the voltage angle at bus 12 relative to bus 6 is approximately robust against phase angle errors in the polarizing current,
1 Hz during the most severe stable power swing for this R1rch should satisfy the inequality in (17) [38].
system. This frequency increases to about 5 Hz during unstable
power swings. Therefore, since the triple-blinder scheme used R1rch,pu < [(1 − X1rch,pu )/(θerr × π/180)]. (17)
to detect power swings in Fig. 7 decouples the power swing
blocking and the out-of-step tripping functions of the relay In this equation, R1rch,pu and X1rch,pu are, respectively, the
[36], the frequency of the most severe stable power swing is resistive and reactance reaches of the zone-1 quadrilateral
used to determine the power swing blocking time delay [10]. distance element, expressed in per-unit of line reactance;
Thus, based on (14), the power swing blocking time delay is and θerr is the phase angle error in the polarizing current.
calculated as Considering an error as large as θerr = 8° [38] and plugging
X1rch,pu = 0.8 pu in (17) result in an upper limit of 1.43 pu for
72.31 − 61.81
Time delay = 60 = 1.75 cycles. (15) R1rch,pu , which is equivalent to 1.43 × (10.99/0.8) = 19.65 Ω.
360 (1) Considering an additional 1 Ω safety margin, we set the
This time delay corresponds to 29.16 ms, which is approx- resistive reach of zone 1 of the distance elements in Fig. 7
imately the same as the 30 ms threshold recommended in to 18.65 Ω. This number is about 60% of the 30.36 Ω upper
[10]. Therefore, the power swing blocking time delay of relay limit suggested by [10], thereby addressing the two challenges
R12−6 is set to 30 ms. It should be noted that although the discussed above.
procedure outlined above closely follows multiple industry
practices and guidelines, reducing the chosen time delay any
further is not always effective for improving the power swing B. Zone-2 Resistive Reach
detection near IBRs, as discussed in Section III-B.
To configure the resistive reach for zone 2 of the distance
characteristics illustrated in Fig. 7, one strategy involves
A PPENDIX B adhering to the guideline presented in [20], which recommends
S ETTINGS C ALCULATIONS FOR THE coordinating the resistive reaches of zones 1 and 2 considering
R ESISTIVE R EACH OF R12−6 the infeed currents, hence different resistive reach settings
across these zones. Meanwhile, since zone 2 is also considered
This appendix outlines the procedure adopted to set the susceptible to inadvertent operation during power swings [11],
resistive reaches of the distance zones depicted in Fig. 7. we follow the zone overlapping practice recommended in [9],
in which the resistive reach parameters are set uniformly across
all zones. This would reduce the resistive reach of zone 2
A. Zone-1 Resistive Reach
for a given zone-1 characteristic, thereby making the passage
The setting guideline in [10] recommends to set the resistive of apparent impedance through zone 2 during power swings
reach for both the ground and phase elements of the relay as less likely. This approach is also adopted in [10], where the
high as possible without interfering with the load impedance. default settings for the resistive reaches of zones 1 and 2 of the
Meanwhile, this guideline recommends that the resistive reach quadrilateral characteristics are identical (whereas the default
for the zone-1 phase distance element, denoted in this paper settings for the reactance reaches of zone 1 and 2 in [10]
by R1rch , should be limited to three times the reactance reach are different). Therefore, the resistive reach for zone 2 of the
of zone 1, denoted by X1rch . As discussed in Section III-B, distance characteristics in Fig. 7 is similarly set to 18.65 Ω.

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.2024.3382843

15

ACKNOWLEDGMENT [24] Odessa Disturbance, Joint NERC and Texas RE Staff Report. NERC, At-
lanta, United States, 2021. [Online]. Available: https://bit.ly/46hHw45.
The authors gratefully acknowledge the contributions of [25] Reliability Guideline: BPS-Connected Inverter-Based Resource Perfor-
Dr. Manish Patel of Southern Company, Atlanta, GA, for his mance. NERC, Atlanta, United States, 2018. [Online]. Available:
valuable insights and discussions throughout this study. https://bit.ly/3tl4gS4.
[26] K. Strunz et al., Benchmark Systems for Network Integration of Renew-
able and Distributed Energy Resources. CIGRE, TF C6.04, Apr. 2014.
R EFERENCES [27] Standard PRC-006-2 — Automatic Underfrequency Load Shedding.
NERC Std. PRC-006-2, 2015.
[1] M. A. Nasr and A. Hooshyar, “Power swing in systems with inverter- [28] Industry Recommendation: Inverter-Based Resource Performance Is-
based resources—Part I: Dynamic model development,” IEEE Trans. sues. NERC, Atlanta, United States, 2023. [Online]. Available:
Power Del., In press, 2024. https://bit.ly/42OhK5k.
[2] Protection System Response to Power Swings. System Protection and [29] Technical requirements for the connection and operation of customer
Control Subcommittee, NERC, Atlanta, United States, Aug. 2013. installations to the high voltage network (TAR high voltage). VDE-AR-
[3] Impact of Inverter Based Generation on Bulk Power System Dynamics N 4120:2018-11, 2018.
and Short-Circuit Performance. IEEE/NERC Task Force on Short- [30] R. Teodorescu, M. Liserre, and P. Rodriguez, “Ac voltage and dc voltage
Circuit and System Performance Impact of Inverter Based Generation, control,” in Grid Converters for Photovoltaic and Wind Power Systems,
Tech. Rep. PES-TR68, Jul. 2018. [Online]. Available: https://bit.ly/ ch. 9, pp. 210–219, John Wiley & Sons, Ltd, 1st ed., 2011.
3PKE8ZS. [31] A. Sangwongwanich, A. Abdelhakim, Y. Yang, and K. Zhou, “Control
[4] A. Haddadi, I. Kocar, U. Karaagac, H. Gras, and E. Farantatos, “Impact of single-phase and three-phase dc/ac converters,” in Control of Power
of wind generation on power swing protection,” IEEE Trans. Power Del., Electronic Converters and Systems, ch. 6, pp. 153–173, Academic Press,
vol. 34, pp. 1118–1128, Jun. 2019. 1st ed., 2018.
[5] Impact of High Penetration of Inverter-based Generation on System [32] S. Bacha, I. Munteanu, and A. Iuliana Bratcu, “Design of the outer
Inertia of Networks. CIGRE, Working Group C2/C4, Oct. 2021. loop (voltage) controller,” in Power Electronic Converters Modeling and
[6] Electromagnetic transient simulation models for large-scale system Control, ch. 9, pp. 283–284, Springer, 1st ed., 2014.
impact studies in power systems having a high penetration of inverter- [33] K. Ogata, “Transient and steady-state response analyses,” in Modern
connected generation. CIGRE, Working Group C4.56, Sep. 2022. Control Engineering, ch. 5, pp. 218–225, Prentice Hall, 5th ed., 2010.
[Online]. Available: https://bit.ly/3FjsXRu. [34] R. Chowdhury and N. Fischer, “Transmission line protection for systems
[7] D60 Line Distance Protection System: Instruction Manual. GE, with inverter-based resources—Part II: Solutions,” IEEE Trans. Power
Markham, Canada, Jun. 2023. [Online]. Available: https://bit.ly/ Del., vol. 36, pp. 2426–2433, Aug. 2021.
3mucHXR. [35] B. Kasztenny and R. Chowdhury, “Security criterion for distance zone
[8] Instruction Manual for SEL-411L Advanced Line Differential Protection, 1 applications in high SIR systems with CCVTs,” in Proceedings of
Automation, and Control System. SEL, Pullman, United States, Dec. the 76th Annual Georgia Tech Protective Relaying Conference,, Atlanta,
2022. [Online]. Available: https://selinc.com/products/411L/docs/. GA, USA, May 2023.
[9] SIPROTEC 5, Distance Protection, 7SA82-V9.5: Manual. Siemens, [36] P. Annamdevula, A. Thurman, M. Thompson, B. Smyth, and D. Tziou-
Germany, Apr. 2023. [Online]. Available: https://sie.ag/44MKLPU. varas, “Simplifying PRC-026 compliance with practical solutions you
[10] Line Distance Protection REL670 Application Manual. ABB, Vasteras, least expect,” in 2021 74th Annual Georgia Tech Protective Relaying
Sweden, Jul. 2022. [Online]. Available: https://bit.ly/3Q1sIRH. Conference, Aug. 2021.
[11] Power swing and out-of-step considerations on transmission lines. IEEE [37] J. Mooney and N. Fischer, “Application guidelines for power swing
Power System Relaying and Control Committee (PSRC) Working Group detection on transmission systems,” in 42nd Annual Minnesota Power
WG-D6, Jul. 2005. [Online]. Available: https:bit.ly/3kQx8xI. Systems Conference, Saint Paul, MN, USA, Nov. 2006.
[12] J. Machowski, Z. Lubosny, J. W. Bialek, and J. R. Bumby, “Electrome- [38] B. Kasztenny, “Settings considerations for distance elements in line
chanical dynamics - Large disturbances,” in Power System Dynamics: protection applications,” 74th Annual Conference for Protective Relay
Stability and Control, ch. 6, pp. 229–283, John Wiley & Sons Ltd., Engineers, Texas, TX, USA, Mar. 2021.
3rd ed., 2020.
[13] H. Saadat, “Bus admittance matrix,” in Power System Analysis, ch. 6, Mohamad-Amin Nasr is currently working toward
pp. 190–195, PSA Publishing LLC, 1st ed., 1999. the Ph.D. degree in electrical engineering with the
[14] N. Hatziargyriou et al., “Definition and classification of power system Department of Electrical and Computer Engineering,
stability – Revisited & extended,” IEEE Trans. Power Syst., vol. 36, University of Toronto, Toronto, ON, Canada. He
pp. 3271–3281, Jul. 2021. received the M.Sc. (Hons.) degree in electrical engi-
[15] J. Machowski, Z. Lubosny, J. W. Bialek, and J. R. Bumby, “Electrome- neering from Amirkabir University of Technology
chanical dynamics - Small disturbances,” in Power System Dynamics: (Tehran Polytechnic), Tehran, Iran, in 2019. Mr.
Stability and Control, ch. 5, pp. 195–228, John Wiley & Sons Ltd., Nasr was recognized as an outstanding reviewer of
3rd ed., 2020. the IEEE T RANSACTIONS ON P OWER D ELIVERY
[16] G. Ziegler, “Mode of operation,” in Numerical Distance Protection: in 2021. His current research interests include power
Principles and Applications, ch. 3, p. 66, Erlangen, Germany: Publicis system dynamics and protection in the presence of
Publishing, 4th ed., 2011. inverter-based resources.
[17] E. W. Kimbark, “Solution of networks,” in Power System Stability,
Volume I, ch. 3, pp. 111–112, IEEE Press, 1st ed., 1995. Ali Hooshyar (Senior Member, IEEE) is currently
[18] J. D. Glover, M. S. Sarma, and T. J. Overbye, “Numerical integration a Canada Research Chair in Electric Power Systems
of the swing equation,” in Power System Analysis and Design, ch. 11, with the Department of Electrical and Computer
pp. 608–613, Cengage Learning, 5th ed., 2012. Engineering at the University of Toronto, Toronto,
[19] N. Tleis, “Typical data of power system equipment,” in Power Systems ON, Canada. His research interests include dynam-
Modelling and Fault Analysis: Theory and Practice, Appendix A.2, ics, control, and protection of renewable energy
pp. 608–618, Newnes, 1st ed., 2008. systems. Dr. Hooshyar is an Editor of the IEEE
[20] C. Henville and R. Chowdhury, “Coordination of resistive reach of T RANSACTIONS ON S MART G RID, IEEE T RANS -
phase and ground distance elements,” in Proceedings of the 48th Annual ACTIONS ON P OWER D ELIVERY , and IEEE P OWER
Western Protective Relay Conference, Spokane, WA, USA, Oct. 2021. E NGINEERING L ETTERS. He was the Guest Editor-
[21] IEEE/IEC Measuring relays and protection equipment – Part 24: in-Chief of the Special Issue of the IEEE T RANS -
Common format for transient data exchange (COMTRADE) for power ACTIONS ON P OWER D ELIVERY on “Resilience-Oriented Protection, Control,
systems. IEEE Std. C37.111-2013 (IEC 60255-24 Edition 2.0 2013-04), and Monitoring Systems for Power Grids.” Dr. Hooshyar was the Guest
2013. Editor of the IEEE E LECTRIFICATION M AGAZINE for the Special Issue on
[22] CMC 356 User Manual. Omicron Electronics Corporation, 2020. “Microgrid Protection and Control.” He is also an Associate Editor for the
[Online]. Available: https://bit.ly/3NVZdjo. Wiley Encyclopedia of Electrical and Electronics Engineering. Dr. Hooshyar
[23] IEEE Standard for Interconnection and Interoperability of Inverter- chairs Working Group C45 of the IEEE Power Systems Relaying and Control
Based Resources (IBR) Interconnecting with Associated Transmission Committee on short-circuit modeling and protection of power systems with
Electric Power Systems. IEEE Std. 2800-2022, 2022. high penetration of inverter-based resources.

Authorized licensed use limited to: Universität Leipzig. Downloaded on March 29,2024 at 07:24:26 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.

You might also like