Received December 22, 2020, accepted January 7, 2021, date of publication January 13, 2021, date of current version

January 22, 2021.

Digital Object Identifier 10.1109/ACCESS.2021.3051283

Battery Energy Storage to Mitigate Rapid

Voltage/Power Fluctuations in Power Grids Due
to Fast Variations of Solar/Wind Outputs
M. JAN E. ALAM 1,2 , (Senior Member, IEEE),
KASHEM M. MUTTAQI 2 , (Senior Member, IEEE),
AND DANNY SUTANTO 2 , (Senior Member, IEEE)
1 Pacific Northwest National Laboratory (PNNL), Richland, WA 99354, USA
2 School of Electrical Computer and Telecommunications Engineering, University of Wollongong, Wollongong, NSW 2522, Australia
Corresponding author: Kashem M. Muttaqi ([email protected])
This work was supported by the Australian Research Council (ARC).

ABSTRACT Passing clouds and wind gusts can create unacceptable rapid voltage/power variations in power
networks. Simulation results using a real Australian distribution feeder with real load demand and PV output
profile show that with a high penetration of PV the voltage variations can increase beyond the allowable limit
given by the standards. Similarly, wind gusts can create wind power output variations. This paper addresses
the rapid voltage/power variations caused by solar or wind power outputs and presents a control strategy
using the energy buffer in energy storage for their impact mitigation. By controlling the discharging/charging
operation of the energy storage based on the available energy buffer in the storage unit, not only the rate
of power output variations can be maintained at the desired level, but also the voltage variations can be
controlled within acceptable limits. An experimental setup, consisting of a PV emulator, inverter, and grid
simulator, demonstrates the potential threat of unacceptable voltage variations. The proposed mitigation
strategy using energy buffer in the storage for smoothing the power outputs of solar/wind is tested and
validated through simulation. Simulation results show that with the proposed control, the fast variations of
the voltage/power variations instigated by solar/wind can be maintained within the acceptable limits.

INDEX TERMS Solar PV, wind power, rapid voltage change, rapid voltage fluctuation, rapid power
variation, energy storage, control strategy, impact mitigation, renewable energy, power grids.

I. INTRODUCTION bution systems. In a weak distribution feeder with a high

The impacts of the steep variations in PV output on electric penetration of PV systems, there is a possibility of violat-
utility operation due to cloud passing has been reported in ing this limit. Rapid voltage changes in distribution net-
previous literature [1], [2]. The wind power output variation works can cause unacceptable flickers, spurious tripping of
can also cause significant rapid and steep voltage fluctuation relays and sensitive electronic equipment, stalling of induc-
as reported in the literature. A rapid voltage fluctuation [3] tion motors [7], and may also create hunting of voltage
may create unacceptable power quality disturbances [4]. Volt- regulating devices in the upstream networks [8]. Therefore,
age fluctuation is a significant problem in a power distribution a solution for smoothing the power outputs of PV and wind is
system, especially in the low voltage (LV) feeders which necessary.
are normally weak, located far away from the main sub- Previously reported research works have addressed the
station, and have a high R/X ratio [5]. The IEC standard PV/wind output smoothing problem using various technolo-
61000-2-2 [6] suggests that a rapid voltage variation over gies including the electric double-layer capacitor [9], battery
3% of the nominal value is not acceptable for LV distri- energy storage [10]–[15], or a combination of both [16], [17].
The drawback of the traditional moving average based
The associate editor coordinating the review of this manuscript and strategy for controlling the rate of the output power vari-
approving it for publication was Yunfeng Wen . ations [9], [18], as mentioned in [19], necessitates the

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 9, 2021 12191
 M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

development of a more direct approach. In [19], such con- developed and tested. Finally, the energy storage devices
trol is proposed to help in smoothing out the variations of integrated with PV/wind systems are intelligently controlled
voltage during cloud passing by injecting or consuming the using an appropriate control strategy to reduce the rate of the
appropriate amount of power using the energy storage device power output variations, and its effect on the voltage smooth-
integrated with the PV system. ing to meet the allowable rapid variation limit is assessed and
The authors in [30] have experimentally tested the perfor- the results presented.
mance of the battery system to smooth the short-term fluc- The key contributions of the paper are as below:
tuations in the wind power output along with withstanding a (a) This paper has investigated the rapid voltage fluctua-
low-voltage transient. A fuzzy-based discrete Kalman filter tions caused by the solar and wind power output variations in
approach has been proposed in [31] to smooth the output the context of IEEE and IEC standards.
power of hybrid wind and PV systems by considering the (b) A strategy to use the rule-based ramp control strategy
state of health of a battery. In [32], a strategy is proposed to charge/discharge the energy storage to compensate for the
based on the Savitzky Golay filter coupled with a control difference between actual power and desired power has been
technique, to smooth out the highly fluctuating solar power presented.
output and to reduce the ramp rate to an acceptable level. (c) The use of the energy buffer of the integrated energy
A state of charge (SOC)-based control strategy for smoothing storage, as a value-added benefit, for a new control strategy
the output fluctuation of a wind/PV/battery storage hybrid is proposed for the mitigation of unacceptable rapid voltage
generation system has been proposed in [15] for reducing variations instigated by the PV/wind power.
wind/PV hybrid output power fluctuations and regulating bat- The rest of the paper is organized as follows: Section II
tery SOC. In [33], a two-stage low-pass filter control strategy presents a detailed discussion about voltage fluctuations
with variable filter time constant has been designed for the caused by the solar PV output variations. Section III provides
use of hybrid energy storage to smooth DC power fluctua- experimental evidence of the voltage variations in a weak grid
tions in microgrids. In [34], a monotonic strategy following caused by the solar PV fluctuations. Section IV introduces the
a consistent charging/discharging direction for each battery mitigation of rapid voltage variations using energy storage.
connected in parallel to form a large-scale battery energy Section V provides examples of the voltage fluctuation mit-
storage system (BESS) for the optimal capacity of battery igation in a real distribution network. Section VI highlights
energy storage supporting wind farms is proposed. However, the mitigation of the rapid power variations at the wind power
all these methods cannot specify the required ramp rate to output using power conditioning and energy storage control.
smooth out the solar/wind power fluctuations. A mitigation Finally, Section VII concludes the paper.
approach for rapid voltage/power variations due to cloud
passing has been introduced in an earlier work and some ini- II. VOLTAGE FLUCTUATIONS CAUSED BY PV OUTPUT
tial results have been reported in a conference paper [10]. This VARIATIONS
has been extended further with a detailed investigation in this A. INVESTIGATION OF VOLTAGE VARIATION CAUSED BY PV
paper. This paper focuses on mitigating power and voltage A PV system connected to a distribution network, as shown
fluctuation. It also demonstrates the application of storage in in Fig. 1(a), is considered to determine the voltage variation
mitigating power/voltage fluctuations in support of achieving caused by steep changes in the PV inverter output. For the
compliance with distributed energy resource (DER) intercon- discussion in this section, it is assumed that the energy storage
nection criteria. is not yet integrated with the PV system.
This paper proposes an additional value-added benefit of The amount of change in the voltage magnitude and angle
the energy buffer available in the energy storage for control- in a distribution network caused by the change of the active
ling the rapid voltage/power variations instigated by the PV and reactive power can be related as given below:
and the wind output variations, especially in the networks
σ Pδ σ P|V| 1δ 1P
    
with sensitive installations, where the limits of the rapid = (1)
voltage/power variation should be strictly adhered to. For σ Qδ σ Q|V| 1 |V| 1Q
the experimental investigation of rapid voltage fluctuations, where, σ Pδ , σ P|V| , σ Qδ , σ Q|V| are the sensitivity matrices of
a measured irradiance profile is imposed on a PV emula- active and reactive power with respect to the magnitude and
tor to produce a realistic PV output, which is then fed to angle of voltage, respectively (similar to the Jacobian matrix
the grid through a PV inverter. The analysis of the voltage used in the Newton-Raphson power flow analysis) [20], 1δ is
at the inverter connection point provides the experimental the vector of changes in voltage angles 1δ, 1|V| is the vector
evidence of the risk of unacceptable voltage variation in a of changes in voltage magnitudes |1V |, 1P and 1Q are the
weak network. The real irradiance profile is also used to vary vectors of changes in net active and reactive powers, 1P and
the power output of the inverters connected to a simulated 1Q, respectively.
weak Australian LV distribution feeder to analyze the impact The expression in (1) can be treated as a system of linear
of the steep variations in PV output on voltage fluctuation. equations of the form given below:
A compensation strategy using the energy buffer in the stor-
age unit for smoothing the PV/wind power output is also Ax = b (2)

12192 VOLUME 9, 2021

M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

investigate the potential variation of 1|V| caused by a given

change of PINV (1PINV ) due to cloud passing, and taking into
account the variation of the line impedance in a weak radial
feeder.

B. ALLOWABLE NUMBER OF VOLTAGE VARIATIONS WITH

THE SHORT CIRCUIT CAPACITY AT PoC
The IEEE standard 1547.2-2008 [8] stipulates based on the
IEC standard 61000.3.7:2012 [22] that the connection of a
fluctuating installation can be accepted without further anal-
ysis if the ratio of the apparent power fluctuation (1S) to the
system short circuit capacity (SCC) at the PoC, expressed
as percentages, are within the limits specified in (5). These
limits depend on the number ‘r’ of voltage changes per
minute [8]:
1S
 
if ≤ 0.15% then, r > 200
SCC
1S
 
FIGURE 1. Distribution system model for voltage variation analysis
caused by solar PV output fluctuation. (a) A PV system in a household if 0.15% < ≤ 0.23% then, 10 ≤ r ≤ 200
connected with the grid. (b) A model of typical rural Australian LV feeder. SCC
1S
 
if 0.23% < ≤ 0.46% then, r < 10 (5)
SCC
σ Pδ σ P|V| 1δ 1P
     
where, A = ,x = ,b = The quantity ‘‘r’’ in (5) indicates the number of
σ Qδ σ Q|V| 1 |V| 1Q
changes in voltage. This quantity is defined based on
Using (2), the change in the voltage magnitude at the point IEEE 1547-2008 [8], which was the applicable DER inter-
of connection (PoC) of a PV system in the house connected connection standard during the time of conducting this work.
to the p-th phase of the i-th bus (1|V ip |) can be determined Therefore, the results are analyzed in the context of the
by applying Cramer’s rule [21] as given in (3). IEEE 1547-2008 [8]. Recently, IEEE standards committee
σ Pδ σ P|V|m has approved IEEE 1547-2018 [35] that defines 3% (when
 
det
  σ Qδ σ Q|V|m PCC at MV) and 5% (when PCC at LV) of nominal voltage
1 V ip  = (3)
 
change per second averaged over one second as the accept-
σ Pδ σ P|V|
 
det able rapid voltage. According to IEEE Std 1453-2015 [37],
σ Qδ σ Q|V|
voltage changes due to events such as motor starting, capaci-
where, ‘det’ means determinant, and σ P|V|m and σ QP|V|m are tor switching, and voltage regulator switching are categorized
the matrices obtained by replacing the columns correspond- as rapid voltage changes (RVC) as the voltage changes are
ing to the p-th phase and i-th bus of σ P|V| and σ P|V| by 1P sustained over several cycles. The perception of the RVC may
and 1Q, respectively. For clarity, the full expression of (3) vary depending upon the following [37]:
for a simple two bus system is given in the Appendix. (a) The duration of a steady-state condition between two
In (2), the amount of the active and reactive power change voltage changes;
 , 1Q ) will
at any i-th bus and p-th phase (1P ip ip
 depend on (b) The rate of change of the voltage (dv/dt); and
ip ip
the variation in the load demand 1PL , 1QL and the PV (c) The magnitude of the voltage change.

ip ip
 The IEEE standard 1547 criterion is used in this paper
inverter output 1PINV , 1QINV as given below:
to assess whether the PV output fluctuations in a weak grid

ip ip
 caused by cloud passing can exceed the allowable number
1Pip = 1PL − 1PINV
of voltage variations per minute. A 400m long LV feeder,
constructed using 4 × 95mm2 LV ABC conductors that char-
 
ip ip
1Qip = 1QL − 1QINV (4)
acterize a typical rural Australian LV feeder, is used for this
As the PV inverters in LV feeders mainly operate at unity purpose, as shown in Fig. 1(b). The transformer impedance is
power factor and the load demand does not change as abruptly (1.65 + j3.64)% based on its rating. The LV feeder supplies
as the steep variations in the PV output caused by the passing 30 customers at 10 buses with one customer per phase per
clouds, the rapid change in the voltage at the PoC caused bus. Each customer is assumed to have a solar PV system
by cloud passing can be attributed mainly to the change of installed with a 4 kVA inverter. The peak load demand is
PINV for a given network equilibrium. For a given 1PINV , 2.7 kW per customer at 0.95 power factor lagging. It is worth
the amount of 1|V| and 1δ in (2) will depend on the network to note that a typical Australian household might have its
sensitivity matrices and therefore is a function of the network yearly peak demand of around 5 kW [36], either on a hot sum-
impedance. A voltage sensitivity analysis is performed to mer afternoon when air conditioning is used, or on a winter

VOLUME 9, 2021 12193

 M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

FIGURE 3. (i) Waveform generator (ii) PV emulator (iii) inverter with PV

emulator.

FIGURE 2. The allowable number of voltage changes with the SCC at PoC.
(a) SCC along the feeder. (b) (1S/SCC) along the feeder. may arise and may lead to confusion. A Rapid Voltage Varia-
tion Index (RVVI) is proposed in this paper, which enables
the voltage variation to be expressed using a numerical
evening when electric heating is used. Therefore, the assump- value derived from a given limit of voltage variation and the
tion for the average demand of 2.7 kW per customer is measured voltage variation in per second, as 1 second is
reasonable. so far the lowest interval for PV output measurement. The
An approximate estimation of the SCC of the feeder in kVA expression for RVVI is given below.
is calculated using (6). |V (k)|−|V (k−1)|
1 VNom
Vnom RVVI = × (7)
SCCkVA = Vnom × × 1e−3 1 |V |lim 1meas
|ZPoC |
where, |ZPoC | = |ZTrafo + ZLVsub-PoC | (6) where V (k) is the voltage in Volt at PoC at any given k-th
instant, Vnom is the nominal voltage in Volt, 1|V |lim is the
where, ZPoC is the total series impedance in Ohm at the PoC acceptable voltage variation limit in pu per second, and 1meas
from the LV substation which is calculated as the sum of is the time interval of the measurement interval in seconds
the impedance of the transformer ZTrafo and the impedance (which is not limited to 1 second). Therefore, the comparison
of the line segment between the PoC and the LV substation, is between similar quantities, i.e., rate of change of measured
ZLVsub−PoC . The nominal voltage Vnom is considered to be voltage versus the acceptable rate of change of voltage. The
230V for a single-phase line. value of RVVI will be 1 if the voltage variation per second is
The values of SCC obtained using (6), which a standard equal to 1|V |lim . RVVI value greater than 1 will indicate that
definition of Short Circuit Capacity as V2 /Z, are plotted an unacceptable rapid voltage variation has occurred. It would
in Fig. 2(a) which shows that the SCC can be as low as be expected to limit the value of RVVI to less than 1 during
300 kVA at the remote end of the feeder. Using a real normal network operation. An example is given to illustrate
irradiance profile measured at a one-second interval by the this. Let us assume a scenario where measured voltage devia-
National Renewable Energy Laboratory, USA, the calculated tion over a 5-second measurement period is 10 V. Therefore,
maximum value of output power fluctuation (1S) of a 4 kW with a nominal voltage of 230V, pu change in measured
inverter (which is a typical size for a residential PV system) voltage per second is 0.0087. If 1|V |lim is 0.05 pu per second,
was found to be more than 1 kW in a one-second interval. then RVVI is 0.174 which is significantly lower than 1, and
Based on the values of SCC and 1S (1 kW), the ratio of 1S therefore, the voltage fluctuation is not characterized as a
to SCC along the feeder is plotted in Fig. 2(b) and shows that rapid voltage deviation.
it can increase up to 0.33% at the remote end of the feeder,
which is higher than 0.23% specified in the standard. For such III. EXPERIMENTAL EVIDENCE OF VOLTAGE VARIATION
a ratio, the voltage change at the remote end of the feeder IN A WEAK GRID CAUSED BY PV FLUCTUATIONS
should be limited below 10 times per minute, as specified in An experimental setup is developed to test the voltage varia-
IEEE 1547 [8]. tions caused by the solar PV output fluctuations in a weak
grid. The test setup consists of a California Instruments
C. QUANTIFICATION OF RAPID VOLTAGE VARIATIONS MX30 waveform generator, shown in Fig. 3(i), connected
Despite international standards, the acceptable value of to an Omni 3-75 impedance network that is used to sim-
rapid voltage variation may vary depending on the local ulate the grid, an Elgar TerraSAS ETS 1000/10 PV emu-
utility regulations. The limit specified for the maxi- lator, (the screen output from the PV emulator is shown
mum voltage variation in an LV feeder according to the in Fig. 3(ii)), and a Fronius IG Plus 25V-1 Inverter. The con-
IEC-61000-2-2 is 6.9V (i.e. 3% of the nominal voltage, which nection between the PV emulator and the inverter is shown
is 230V in Australia). However, no specific measurement of in Fig. 3(iii).
‘‘rapidity’’ is provided in the IEC standard [6], i.e. whether it The Elgar TerraSAS ETS 1000/10 PV simulator consists
is in second, minute, etc. The PV output profiles are captured of programmable DC power supplies, a rack-mounted con-
at different time intervals (e.g., 1 sec, 5 sec, 10 sec, 1 min) troller, keyboard, and an LCD display with control software
depending on the data logging device and the purpose of the and GUI interface, output isolation, and polarity reversing
data collection. Therefore, differences in the quantification relays and a unique PV simulation engine that controls the

12194 VOLUME 9, 2021

M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

power output. This combination of hardware can simulate

most test protocols or a combination of events that a solar
installation will be subjected to. The included software allows
the modeling of a PV panel without extensive knowledge
of the solar array parameters. The only parameters required
for the simulation are the open-circuit voltage and the short
circuit current at 25◦ C and 1000W/m2 . The slope of the I-V
curve can then be modified by the peak power parameters,
the maximum power point voltage, and the current. Changes FIGURE 4. A real irradiance profile for experimental verification of
to these parameters will allow the shape of the I-V curve unacceptable voltage variation.
to be adapted to any fill factor between 0.5 and 1. Once an
I-V curve has been generated, the changes to the irra-
diation level or temperature can be changed on the fly,
so that the behavior of a grid-tied inverter can be tested
under realistic conditions for cloud shadowing and panel
temperature rise. The inverters can be optimized for real
Maximum Power Point (MPP) search modes because the
shadowing and the temperature changes can be simulated
realistically.
The output of the PV emulator is connected to the Fronius
IG Plus 25V-1 Inverter. The inverter is connected to the Cal-
ifornia Instruments MX30 waveform generator through the FIGURE 5. Experimental results corresponding to the ramp-up and
Omni 3-75 impedance box and the additional series resistance ramp-down events. (a-i) PV current ramp-down. (a-ii) Captured
instantaneous voltage waveforms. (a-iii) Extracted RMS voltage. (b-i) PV
of 0.4 Ohm. The MX30 waveform generator can emulate current ramp-up. (b-ii) Captured instantaneous voltage waveforms.
the behavior of a real AC grid under a wide variety of grid (b-iii) Extracted RMS voltage.
conditions including harmonics and transients. The Omni
3-75 impedance box has an impedance value of (0.15+j0.15) Simulink software
R and LabVIEW. The Simulation Interface
ohm per phase to represent a reference impedance for testing Toolkit automatically generates LabVIEW code to interface
the voltage flicker according to IEC/TR 60725 [23]. The with a Simulink module
R resulting in a flexible and easy-
additional series resistance of 0.4 Ohm is added to make to-use user interface. The experimental results are presented
the R/X ratio higher than 3 to reflect the highly resistive in Fig. 5.
nature of LV networks. It is to be noted that the R/X ratio During the highest negative ramp of the irradiance profile
has been exaggerated to demonstrate the effect of a high in Fig. 4, the irradiance level is reduced by 400 W/m2 in
R/X ratio as studied in [38], [39]. The bidirectional feature 1 second period. This causes the current output of the
of the waveform generator is used to simulate the ‘‘sink’’ PV simulator to ramp down as shown in Fig. 5(a-i).
and ‘‘source’’ behavior of the grid to test the behavior of the This decrease of the PV simulator DC current causes
grid-tied inverters when injecting power to or drawing power a reduction in the PoC voltage waveform as shown
from the grid. in Fig. 5(a-ii). This is better understood from the RMS volt-
A real PV irradiance profile [24] captured at a one-second age profile in Fig. 5(a-iii) extracted from the instantaneous
interval by the National Renewable Energy Laboratory’s voltage waveform in Fig. 5(a-ii). It is found that the volt-
irradiance monitoring facility in Oahu Island, Hawaii, USA age variation is approximately 2 V/sec which is approx-
is used as the input to the PV simulator. The irradiance imately 30% of the rapid voltage variation limit of 6.9V
profile used for this experiment was captured on the 20th according to the IEC 61000-2-2 standard. As discussed in
of March, 2010, and contains fluctuations created by pass- Section II-C, unfortunately, IEC did not specify how the
ing clouds. The highest negative and positive rates of rapid voltage variation of 6.9V is calculated in terms of
output power changes from the irradiance profile, shown time (second, minute, etc.). During the highest positive ramp
in Fig. 4, are selected for the rapid voltage variation anal- event of the irradiance profile in Fig. 4(b), the irradiance
ysis. The changes in the irradiance level change the I-V level is increased by 500 W/m2 in 1 second period. This
curve and the MPP point, and hence the PV output power, causes the current from the PV emulator to ramp up as
which is reflected on the PoC voltage (i.e. the voltage at shown in Fig. 5(b-i). Therefore, the PoC instantaneous volt-
the connection point) of the inverter with the grid. The age increases as shown in Fig. 5(b-ii), and the corresponding
variations of the PoC voltage are captured using the Lab- RMS voltage profile is shown in Fig. 5(b-iii). For the ramp-up
View data acquisition system at 10,000 samples per sec- event, the increase in voltage variation is greater than 2 V/sec
ond and processed using the MATLAB/Simulink to extract because of the larger change in the irradiance level. From the
the RMS values. The Simulation Interface Toolkit (SIT) results presented in Fig. 5, it is apparent that a significant
provides seamless integration between The MathWorks Inc. voltage fluctuation may appear in a real network where the

VOLUME 9, 2021 12195

 M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

fault level can be even lower than that used in the experiment,
and the excursion in PV output can also be higher, given a
larger size of the PV inverter. Therefore, the experimental
results suggest that the threat of violating the rapid voltage
variation limit in a weak network is not negligible.

IV. MITIGATION OF RAPID VOLTAGE VARIATION USING

THE PROPOSED CONTROL
Renewable energy generation induced voltage flicker [25], FIGURE 6. An illustration of the proposed control algorithm.
voltage volatility [26], and power quality problems [27] have
been investigated in the literature. However, rapid voltage
variation caused by renewables was not thoroughly investi- performed for the voltage fluctuation mitigation. If the PV
gated and no solution for this is available that can guarantee output is suddenly reduced due to cloud passing, the storage
the voltage fluctuation can be maintained at an acceptable device will discharge 1PES amount of power, which will
level. This section discusses an approach for smoothing out be added to the PV power, which will minimize the sharp
the sudden changes in PV output using the control of the decrease in voltage. If the PV output increases suddenly
output of the energy storage and assesses the value-added due to the removal of the cloud shadow, then the storage
benefit of the energy buffer in mitigating the rapid volt- device will be charged with 1PES amount of power, which
age variations, especially in weak distribution networks. The will be subtracted from the PV output to mitigate a sharp
operational impact of the rapid voltage deviation may vary increase in voltage. The proposed control algorithm given
depending on the network characteristics, and the conse- in (8) is illustrated in Fig. 6 using a fictitious PV output PPV
quence may relate to the control/operation and performance continuing for 4 consecutive time instants from (k − 1) to
assessment philosophy of the utility under consideration. (k + 2).
However, the most direct consequence would be the risk of During the first time instant (k − 1), PINV is equal to
not complying with DER interconnection criteria and reduced PPV (assuming ηINV is 100% for simplicity of the descrip-
DER hosting capacity which may affect the utility in various tion). At the k-th instant, PPV decreases sharply due to the
form and fashion [8]. cloud shadow by such an amount that exceeds the desired
rate of 1PINV−des . To control the rate at the desired value,
A. THE PROPOSED CONTROL ALGORITHM the storage power PES is changed by +1PES (k), which
The proposed control to control the PV output is essentially is determined using (8). At the (k + 1)-th instant, PPV
the action of limiting the rate of 1PINV in (4) to the desired increases sharply as the cloud passing period ends. To limit
value. This can be realized by integrating an energy storage the change of PINV at the desired value, PES is changed
device with the PV system, as shown in Fig. 1(a). The energy by −1PES (k + 1), determined using (4). At the (k + 2)-
storage device will be controlled in such a manner that it th instant, PPV does not change; however, PES is changed
delivers (discharges) or consumes (charges) an appropriate by 1PES (k + 2) to increase PINV to attain the level of
amount of power, PES so that 1PINV can be controlled at the PPV (k + 2). This change is performed in such a way that
desired value. the rate of PINV variation does not exceed 1PINV−des .
The necessary change in storage power 1PES at a given This suggests an important consideration for developing a
k-th instant for controlling the rate of PINV changes to new control strategy. The value of 1PINV−des during the
the desired value (1PINV−des ) can be determined using the events of a fluctuation needs to be selected in such a way
expression below: that controlling PES itself does not create any fluctuation.
It is to be noted that 1PINV−des is a user-defined param-
1PINV-des (k)
1PES (k) = − 1PPV (k) (8) eter and can change with time, i.e., a function of the time
ηINV instant k.
where ηINV is the inverter efficiency. This change in PES It is worth to mention that battery life-time could be
will, in effect, contribute to mitigating the sudden voltage impacted by unnecessary charge/discharge actions. While the
fluctuation. Mathematically, the effect of the change in PES illustration of the proposed strategy in Fig. 6 appears to be
can be reflected on the system by modifying the diagonal a constant charge/discharge operation, but it is a graphical
element corresponding to the i-th bus and p-th phase of the representation of the charge/discharge operation over each
ip
matrix σP|V|m in (3), denoted by σP|V |m (k) at any k-th instant interval of time (e.g., k, k+1, k + 2). Considering the speed
as below: of operation needed for such applications, the intervals are
ip ip likely to be fast (e.g., 1 sec) and hence, the charge/discharge
ip ip 1PPV (k) ± 1PES (k)
σP|V |m (k) = 1PL (k) − (9) set-points will be updated at a high speed (e.g., 1 sec). There-
ηINV fore, it should not be a reason for any additional concern
The ± sign in (9) indicates that either the charging or other than the typical wear and tear expected from such an
the discharging operation of the storage device needs to be operation.

12196 VOLUME 9, 2021

M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

FIGURE 8. Fluctuations in a real PV output profile. (a) Normalized PPV .

(b) Rate of PV output variation.

FIGURE 7. The proposed control flowchart.

B. CONTROL FLOWCHART
A flowchart of the proposed control approach is given
in Fig. 7. The PV panel output PPV is continuously moni-
tored to detect any step-change higher than the 1PINV−des
for the given instant. If such a step change is detected, the
controller determines the appropriate 1PINV−des depending
on the event and then the required change in PES is obtained
using (9). If the determined value of PES at the given k-th
instant is positive, then SoC is checked to determine if it is
higher than SoCmin . If ‘‘Yes’’, then the discharge operation
is performed; if ‘‘No’’, then the controller transits to the next
time instant. On the other hand, if PES is negative, the SoC is
checked to determine if it is lower than SoCmax . The charging
operation is performed if the result of the SoC status-check is
FIGURE 9. Effect of the proposed control on PV inverter output and PoC
‘‘Yes’’, or the controller transits to the next time instant if the voltage. (a) Inverter power PINV . (b) Storage power PES . (c) PoC voltage.
result is ‘‘No’’.
highest negative ramp caused by passing cloud is found to
V. EXAMPLE OF VOLTAGE FLUCTUATION MITIGATION IN be −30%/second. A 5-minute interval containing the highest
A REAL DISTRIBUTION NETWORK negative ramp event is considered to analyze the effect of the
The benefit of the PV inverter control in mitigating the proposed control on voltage fluctuation.
voltage fluctuation is investigated in this section using the For illustration, the desired rate of the PV inverter vari-
same test feeder shown in Fig. 1(b). For the investigations ations (1PINV−des ) is set to two values: 50 W/sec and
in this section, all the households connected to the feeder 10 W/sec. The effects of the proposed control are observ-
are modeled to have rooftop PV systems with integrated able from Fig. 9. The PINV profiles with and without the
battery storage devices. Valve Regulated Lead Acid bat- proposed control are shown in Fig. 9(a). The smoothing
teries rated at 12 kWh are used for the investigations in of the PV output fluctuation is visible in Fig. 9(a), where
this paper; SoCmin and SoCmax values are set to 40% and the degree of smoothness increases with the decrease of
100% respectively. The modeling of the storage devices has the set value of 1PINV−des (from 50 W/sec to 10 W/sec).
incorporated the non-linear voltage versus the SoC charac- The storage power (PES ) profiles that provide this smooth-
teristics and the effect of internal resistance, as described ing effect are shown in Fig. 9(b) where the discharging
in [28]. (+ve PES ) and charging (−ve PES ) operations required for the
The results shown in this paper correspond to household proposed control are identified. Also, the simulation results
no. 28 (HH 28) connected to phase a, at bus 10 of the feeder. in Fig. 9(b) illustrates how the battery power will be changing
The load demand profile is derived from the data captured over time to mitigate fluctuations. Finally, the mitigation
by an Australian distribution utility and the PV output data of the voltage fluctuation is shown in Fig. 9(c). Without
corresponding to the irradiance profile [24] used for the the proposed control, the voltage follows a similar trend
experiment described in Section III. The normalized values of fluctuations as the PV output, and likewise, the voltage
of the PV output profile (PPV ) in a 1-second resolution are profile is smoothened with the application of the proposed
shown in Fig. 8(a), which indicates severe power fluctuations control.
occurred between 09:00 hours to 14:00 hours. The rate of PPV As this paper focuses on the benefit of the rapid voltage
changes during this interval is shown in Fig. 8(b), where the fluctuation mitigation, further investigations are performed

VOLUME 9, 2021 12197

 M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

FIGURE 11. The number of rapid voltage changes with and without the
proposed control.

FIGURE 10. Quantification of the rapid voltage change with and without
the proposed control. (a) |1V| in percentage per sec over 5-min interval
showing a violation of IEC 61000-2-2. (b) RVVI over 5-min interval.

on the results in Fig. 9(c). The amounts of changes (as

a percentage of the nominal value) in the voltage profiles
in Fig. 9(c) are plotted in Fig. 10(a) that shows, without
energy storage control, the 1|V | per sec can be lower than
−3% and therefore it will violate the limits of maximum
3% change specified in the IEC 61000-2-2 standard. With FIGURE 12. The dynamic performance of the proposed control.
(a) Transient response of PINV and PES . (b) Transient response of PoC
the proposed control strategy, 1|V | is significantly reduced voltage.
to −0.16 % with 1PINV−des of 50 W/sec and to −0.03%
with 1PINV−des of 10 W/sec. The Rapid Voltage Variation response of PES resulting from a step-change command by
Index (RVVI) values corresponding to the voltage variations the storage controller is also observable in Fig. 12(a). The
in Fig. 9(c) are plotted in 10(b) using a value of 1|V|lim transient response of the PoC voltage is given in Fig. 12(b)
of 0.03 (3%). Without any control, the RVVI can be greater which shows how the rapid voltage change is minimized
than ‘1’ which indicates that the allowable limit has been with the help of the proposed control by the storage
violated. With the proposed control, the value of RVVI is device.
significantly reduced.
According to the 1S to SCC ratio at the remote end of VI. WIND POWER OUTPUT SMOOTHING USING THE
the test LV feeder presented in Fig. 1, the number of volt- PROPOSED CONTROL
age changes per minute should be limited to below 10. The The mechanical power output of the wind turbine depends
threshold for detecting a voltage change is set to 0.5% of the on the wind speed and the parameters of the wind turbine as
nominal value according to the discussions presented in [29], given below:
where a change of less than 0.5% has been used to define
the steady-state condition of the voltage. A 1-minute sliding PMech = 0.5Cp (λ, θ) ρπR2 v3Wind (10)
window is used to extract the number of voltage changes over where CP is the power coefficient that depends on the tip
0.5% in the voltage profile in Fig. 9(c). It is found that the speed ratio λ and the pitch angle θ as given below:
number of voltage changes over 0.5% can increase beyond  
15 per minute during the PV output fluctuations, as shown 116 − 12.5
CP (λ, θ) = 0.22 − 0.4θ − 5 e λi
in Fig. 11. With the proposed control strategy, this number λi
is reduced to zero because the changes in voltage are always 
1 0.035 −1

2kωR R
below 0.5%. λi = − 3 , λ=
λ + 0.08θ θ +1 pvWind
The dynamic performance of the proposed control strategy
in voltage fluctuation mitigation is investigated in Fig. 12. ρ is the air density in kg/m3 ; Awt = πR2 is the area swept
A 500-millisecond snapshot around the highest negative ramp by the wind turbine, R is the radius of the wind turbine, and
event of the PV output profile is used for this investiga- vWind is the velocity of the wind turbine in m/sec.
tion. The transient response of PINV with and without the The pitch angle controller limits the active power output,
proposed control is shown in Fig. 12(a) that shows how the and the value of θ is zero when the wind speed is within the
desired value of PINV is maintained by controlling the amount operational limits of the wind turbine. The rotor speed ωR
of power discharged by the battery storage. The transient of the turbine is controlled in such a way that the maximum

12198 VOLUME 9, 2021

M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

mechanical power output of the turbine can be obtained.

Considering this, the steady-state output of the wind turbine
can be determined using (10) based on the value of ωR for
which the value of PMech is the maximum.
In the proposed control strategy that uses energy buffer
of the energy storage, a compensation power will be added
with the wind generator output to control the rate of the
final output of the wind generation system at the desired
level. The basic concept of the proposed control strategy is
explained in Fig. 13. Based on Fig. 13, the smoothened wind FIGURE 13. Application of the proposed control to wind power.
generation system output PWind−sm can be obtained from the
wind generator output PWind and the compensation power
PCOMP using (11).
PWind-sm = PWind + PCOMP (11)
Taking the time derivative of (11) on both sides, the rate of
PWind−sm can now be related to the rate of PCOMP and PWind ,
as given in (12).
dPWind-sm dPWind dPCOMP
= + (12)
dt dt dt
where dPWind-sm
dt is the rate of the output variations, dPdt Wind

is the rate of the wind generator power output variations,

and dPCOMP
dt is the rate of the compensation power output
variations.
To control the rate of the wind generator output variations FIGURE 14. Wind power output smoothing using the proposed control.
within the desired level, the rate of PCOMP variations can be
controlled using the expression derived from (12), as given
below: of PCOMP would not create an unacceptable fluctuation in
   PWind−sm , PCOMP does not need to be changed anymore.
dPCOMP dPWind-sm  dPWind
=  − dt (13) To accommodate this, a switching function S, as given below,
dt dt des is incorporated with (13).

where, dPWind-sm
  
dPINV  dPWind

 is the desired rate of the smoothened wind

dt des
 0, if |P
COMP | < and f =0
power variations, which will be negative during a ramp-down S= dt des dt
event and positive during a ramp-up event. 1, otherwise

The expression given in (13) is the basic control equation (15)
that will be used to control the rate of the wind power output
variations. However, the following modifications are made to The modified control equation can be described using the
accommodate practical situations: expression below:
   
(a) To allow for the slow variations of PWind , a dead-band dPCOMP dPWind-sm  dPWind
function ‘f ’ is incorporated with (13) that will force the rate =S×  −f (16)
dt dt des dt
of PWind variations to be zero if it is within an acceptable
Considering a discrete-time representation, the rate of PWind
limit, defined using the desired rate of PWind−sm variations,
variations at the k-th instant can be determined using (17).
as shown below:
   dPWind [PWind (k) − PWind (k − 1)]
 dPWind 
 ≤ dPWind-sm  (k) =
  (17)

dPWind
  0, if 
 dt  dt t (k) − t (k − 1)
f = dt 
des
dt  dPWind , otherwise
 where t(k) is the time at the k-th instant.
dt The value of the switching function is also incorporated
(14) by forcing PCOMP to zero when the conditions of S = 0 are
satisfied, as given in (15). PCOMP at the k-th instant can be
(b) Observing (13) it is apparent that when the rate of obtained using (18).
PWind variation is zero, considering the dead-band function 
mentioned in modification (a), the rate of PCOMP variations dPCOMP
PCOMP (k) = S × PCOMP (k − 1) + (k)
will be governed by the desired rate of PWind−sm variation. dt

However, when the rate of PWind variation is zero, then if
× {t (k) − t (k − 1)} (18)
PCOMP has already attained such a level that the removal

VOLUME 9, 2021 12199

 M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

The solution of any j-th unknown xj in x can be obtained

using Cramer’s rule [21] as given below.


det Aj
xj = (A.2)
det (A)
FIGURE 15. A fictitious 2-bus system. where ‘det’ means determinant, Aj is a matrix obtained by
replacing the j-th column of A by b.
For illustration, the full expression of (A.2) for a fictitious
Fig. 14 shows the wind power output smoothing using the
2-bus system, shown in Fig. 15, is given below in (A.3) where
proposed control of the energy buffer in energy storage.
the change in PoC voltage at phase b of Bus-2 caused by
the change in net power (1P2b , 1Q2b ) at the same PoC is
VII. CONCLUSION determined.
This paper has investigated the rapid voltage/power fluctua- 
σPδ σP|V |m 
tions caused by the solar and wind power output variations z }| { z }| {
in the context of IEEE and IEC standards. A rule-based ramp  σPδ σPδ
 aa ab
σPδ
ac σP|V
aa
| 0 σP|V
ac 
|
 σPδ σPδ σPδ σP|V | 1P
ba 2b
σP|V | 
bc 
 ba bb bc
control strategy has been used to charge/discharge the energy
 σPδ σPδ σPδ σP|V |
ca
σP|V
cc 
 ca cb cc
storage to compensate for the difference between actual 0 |

det 
 σQδ σQ|V |m
power and desired power. Also, the use of the energy buffer z }| { z }|

{
of the integrated energy storage, as a value-added benefit,  σQδ σQδ σQδ σQ|V
aa
σQ|V
ac 
 aa ab ac
| 0 |
for a new control strategy is proposed for the mitigation of  σQδ σQδ
 ba bb
σQδ
bc
σQ|V
ba
1Q 2b
σ bc 
| Q|V | 
unacceptable rapid voltage/power variations instigated by the cc σ ca σ cc
  σ ca
Qδ σ cb
Qδ σ Qδ Q|V | 0 Q|V |
1 V  =
PV/wind power. The amount of power delivered or stored  2b 
 σPδ σP|V | 
by the energy storage device is appropriately controlled at z }| { z }| {
ac σ aa
P|V | σP|V | σP|V | 
ab ac 
 σPδ σPδ σPδ
each instant of time to maintain the rate of the inverter output  aa ab
bc σ ba
P|V | σP|V | σP|V | 
bb bc 
changes at the desired level. In effect, the steep voltage varia-  σPδ σPδ σPδ
 ba bb
tions at the inverter connection point are mitigated. Analysis  σPδ σPδ
 ca cb
σPδ
cc
σP|V | σP|V | σP|V | 
ca cb cc  
performed on a typical weak grid in Australia using a realistic det 
 σQδ σQ|V | 
PV output and load demand profile has revealed that the volt- z }| { z }| {
 σQδ σQδ σQδ σQ|V | σQ|V | σQ|V | 
aa ab ac 
 aa ab ac
age variations without any support can exceed the allowable
 σQδ σQδ σQδ σQ|V
ba
| σQ|V | σQ|V |
bb bc 
 ba bb bc
limit as specified in the international standards. An indication 
of an unacceptable voltage variation is also found from the σQδ σQδ σQδ σQ|V | σQ|V | σQ|V |
ca cb cc ca cb cc

experimental data using a real solar irradiance profile. The (A.3)

proposed compensation strategy using the energy buffer in
the energy storage system for smoothing the PV/wind power
REFERENCES
output has been tested and validated through simulation.
[1] E. C. Kern, E. M. Gulachenski, and G. A. Kern, ‘‘Cloud effects on
With the proposed strategy, the number of variations and distributed photovoltaic generation: Slow transients at the gardner, mas-
the number of variations per minute can be maintained well sachusetts photovoltaic experiment,’’ IEEE Trans. Energy Convers., vol. 4,
below the limits and the voltage/power variations instigated no. 2, pp. 184–190, Jun. 1989.
[2] A. Mills, M. Ahlstrom, M. Brower, A. Ellis, R. George, T. Hoff, and
by the solar/wind can be maintained within the acceptable B. Kroposki, ‘‘Dark Shadows,’’ IEEE Power Energy Mag., vol. 9, no. 3,
limits. pp. 33–41, May/Jun. 2011.
The scope of this paper is limited to the investigation of the [3] G. Wiczynski, ‘‘Analysis of voltage fluctuations in power networks,’’
IEEE Trans. Instrum. Meas., vol. 57, no. 11, pp. 2655–2664,
use of energy buffer of the energy storage that is integrated Nov. 2008.
with renewable energy such as solar or wind in a particu- [4] P. K. Ray, S. R. Mohanty, and N. Kishor, ‘‘Classification of power quality
lar node or house in a typical rural Australian low voltage disturbances due to environmental characteristics in distributed generation
system,’’ IEEE Trans. Sustain. Energy, vol. 4, no. 2, pp. 302–313,
feeder. Therefore, the results validate the performance of Apr. 2013.
the proposed control strategy when the network contains the [5] H. Sugihara, K. Yokoyama, O. Saeki, K. Tsuji, and T. Funaki, ‘‘Economic
combined generation of solar/wind. and efficient voltage management using customer-owned energy storage
systems in a distribution network with high penetration of photovoltaic
systems,’’ IEEE Trans. Power Syst., vol. 28, no. 1, pp. 102–111,
APPENDIX Feb. 2013.
[6] Electromagnetic Compatibility (EMC)—Part 2-2: Environment—
Consider a system of linear equations of the form given Compatibility Levels for Low-Frequency Conducted Disturbances and
below. Signalling in Public Low-Voltage Power Supply Systems, Standard 61000-
2-2, Mar. 2002.
[7] Voltage Fluctuations in the Electric Supply System.
Accessed: Nov. 2014. [Online]. Available: http://
Ax = b (A.1)
www.elec.uow.edu.au/apqrc/content/technotes/technote7.pdf
[8] IEEE Application Guide for IEEE Std 1547, IEEE Standard for
Interconnecting Distributed Resources with Electric Power Systems,
σ Pδ σ P|V| 1δ 1P
     
where, A = ,x = ,b = Standard 1547.2-2008, 2009, pp. 1–207.
σ Qδ σ Q|V| 1 |V| 1Q
12200 VOLUME 9, 2021
M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

[9] N. Kakimoto, H. Satoh, S. Takayama, and K. Nakamura, ‘‘Ramp-rate [29] K. Brekke, H. Seljeseth, and O. Mogstad, ‘‘Rapid voltage changes–
control of photovoltaic generator with electric double-layer capacitor,’’ definition and minimum requirements,’’ in Proc. IET Conf., Prague,
IEEE Trans. Energy Convers., vol. 24, no. 2, pp. 465–473, Jun. 2009. Czech Republic, 2009, pp. 1–4.
[10] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, ‘‘Mitigation of rapid [30] A. Sattar, A. Al-Durra, C. Caruana, M. Debouza, and S. M. Muyeen,
voltage variations caused by passing clouds in distribution networks with ‘‘Testing the performance of battery energy storage in a wind energy
solar PV using energy storage,’’ in Proc. 8th Int. Conf. Electr. Comput. conversion system,’’ IEEE Trans. Ind. Appl., vol. 56, no. 3, pp. 3196–3206,
Eng., Dhaka, Bangladesh, Dec. 2014, pp. 305–308. Jun. 2020.
[11] W. Guishi, M. Ciobotaru, and V. G. Agelidis, ‘‘Power smoothing of large [31] D. Lamsal, V. Sreeram, Y. Mishra, and D. Kumar, ‘‘Smoothing control
solar PV plant using hybrid energy storage,’’ IEEE Trans. Sustain. Energy, strategy of wind and photovoltaic output power fluctuation by considering
vol. 5, no. 3, pp. 834–842, Jul. 2014. the state of health of battery energy storage system,’’ IET Renew. Power
[12] W. A. Omran, M. Kazerani, and M. M. A. Salama, ‘‘Investigation of Gener., vol. 13, no. 4, pp. 578–586, Mar. 2019.
methods for reduction of power fluctuations generated from large grid- [32] A. Atif and M. Khalid, ‘‘Saviztky–Golay filtering for solar power
connected photovoltaic systems,’’ IEEE Trans. Energy Convers., vol. 26, smoothing and ramp rate reduction based on controlled battery energy
no. 1, pp. 318–327, Mar. 2011. storage,’’ IEEE Access, vol. 8, pp. 33806–33817, 2020.
[33] T. Wu, W. Yu, and L. Guo, ‘‘A study on use of hybrid energy storage
[13] J. Traube, F. Lu, D. Maksimovic, J. Mossoba, M. Kromer, P. Faill,
system along with variable filter time constant to smooth DC power
S. Katz, B. Borowy, S. Nichols, and L. Casey, ‘‘Mitigation of solar
fluctuation in microgrid,’’ IEEE Access, vol. 7, pp. 175377–175385,
irradiance intermittency in photovoltaic power systems with integrated
2019.
electric-vehicle charging functionality,’’ IEEE Trans. Power Electron.,
[34] A. V. Savkin, M. Khalid, and V. G. Agelidis, ‘‘A constrained monotonic
vol. 28, no. 6, pp. 3058–3067, Jun. 2013.
charging/discharging strategy for optimal capacity of battery energy
[14] T. Wang, H. Kamath, and S. Willard, ‘‘Control and optimization storage supporting wind farms,’’ IEEE Trans. Sustain. Energy, vol. 7,
of grid-tied photovoltaic storage systems using model predictive no. 3, pp. 1224–1231, Jul. 2016.
control,’’ IEEE Trans. Smart Grid, vol. 5, no. 2, pp. 1010–1017, [35] IEEE Standards Coordinating Committee 21: IEEE Std 1547–2018
Mar. 2014. (Revision of IEEE Std 1547-2003) IEEE Standard for Interconnection
[15] X. Li, D. Hui, and X. Lai, ‘‘Battery energy storage station (BESS)-based and Interoperability of Distributed Energy Resources with Associated
smoothing control of photovoltaic (PV) and wind power generation Electric Power Systems Interface, Standard 1547–2018, Feb. 2018,
fluctuations,’’ IEEE Trans. Sustain. Energy, vol. 4, no. 2, pp. 464–473, pp. 1–136.
Apr. 2013. [36] Australian Energy Regulator. (Feb. 2017). Tariff Structure Statements—
[16] T. Kinjo, T. Senjyu, N. Urasaki, and H. Fujita, ‘‘Output levelling of Ausgrid, Endeavour and Essential Energy. Accessed: Jul. 17, 2020.
renewable energy by electric double-layer capacitor applied for energy [Online]. Available: https://www.aer.gov.au/system/files/nsw%20final
storage system,’’ IEEE Trans. Energy Convers., vol. 21, no. 1, pp. 221–227, %20decision%20-%20consolidated%20version%20%28D2016-
Mar. 2006. 00169120%29_1.pdf
[17] T. Monai, I. Takano, H. Nishikawa, and Y. Sawada, ‘‘A collaborative [37] Revision of IEEE Std 1453-2011: IEEE Recommended Practice
operation method between new energy-type dispersed power supply for the Analysis of Fluctuating Installations on Power Systems,
and EDLC,’’ IEEE Trans. Energy Convers., vol. 19, no. 3, pp. 590–598, Standard 1453-2015, Oct. 2015, pp. 1–74.
Sep. 2004. [38] M. A. Kashem and G. Ledwich, ‘‘Distributed generation as voltage
[18] T. D. Hund, S. Gonzalez, and K. Barrett, ‘‘Grid-tied PV system energy support for single wire Earth return systems,’’ IEEE Trans. Power Del.,
smoothing,’’ in Proc. 35th IEEE Photovoltaic Spec. Conf., Honolulu, HI, vol. 19, no. 3, pp. 1002–1011, Jul. 2004.
USA, Jun. 2010, pp. 2762–2766. [39] M. A. Kashem and G. Ledwich, ‘‘Multiple distributed generators for
[19] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, ‘‘A novel approach for distribution feeder voltage support,’’ IEEE Trans. Energy Convers.,
ramp-rate control of solar PV using energy storage to mitigate output vol. 20, no. 3, pp. 676–684, Sep. 2005.
fluctuations caused by cloud passing,’’ IEEE Trans. Energy Convers.,
vol. 29, no. 2, pp. 507–518, Jun. 2014.
[20] E. Acha, C. R. Fuerte-Esquivel, H. Ambriz-Peárez, and
C. S. Angeles-Camacho, FACTS Modelling and Simulation in Power
Networks. West Sussex, U.K.: Wiley, 2004.
[21] R. C. Mittal and A. Al-Kurdi, ‘‘Application of the cramer rule in the
solution of sparse systems of linear algebraic equations,’’ J. Comput. Appl.
Math., vol. 136, nos. 1–2, pp. 1–15, Nov. 2001.
[22] Electromagnetic Compatibility (EMC)—Limits—Assessment of Emission
Limits for the Connection of Fluctuating Installations to MV, HV and EHV
Power Systems, document TR IEC vol. 61000.3.7,2012.
[23] Consideration of Reference Impedances and Public Supply Network
Impedances for Use in Determining the Disturbance Characteristics
of Electrical Equipment Having a Rated Current 75 A Per Phase,
document IEC/TR 607253, 2012.
[24] NREL. 1-Second Global Horizontal Irradiance Oahu, Hawaii. Accessed:
M. JAN E. ALAM (Senior Member, IEEE) is cur-
Oct. 2013. [Online]. Available: http://www.nrel.gov/midc/oahu_archive/
rently a Power Systems Engineer with the Pacific
[25] S. K. Sharma, A. Chandra, M. Saad, S. Lefebvre, D. Asber, and L. Lenoir,
Northwest National Laboratory (PNNL), where
‘‘Voltage flicker mitigation employing smart loads with high penetration
of renewable energy in distribution systems,’’ IEEE Trans. Sustain.
he has been working, since October 2016. He is
Energy, vol. 8, no. 1, pp. 414–424, Jan. 2017. a thrust-area lead within Energy Storage Indus-
[26] A. Parchure, S. J. Tyler, M. A. Peskin, K. Rahimi, R. P. Broadwater, and try Acceptance program at PNNL, sponsored by
M. Dilek, ‘‘Investigating PV generation induced voltage volatility for the U.S. DOE. He is also a Key-Contributor with
customers sharing a distribution service transformer,’’ IEEE Trans. Ind. PNNL transactive energy systems and grid ser-
Appl., vol. 53, no. 1, pp. 71–79, Jan. 2017. vices valuation domains. He is managing an effort
[27] X. Liang, ‘‘Emerging power quality challenges due to integration of to develop capabilities at PNNL for grid integra-
renewable energy sources,’’ IEEE Trans. Ind. Appl., vol. 53, no. 2, tion of marine renewable energy resources. Before joining PNNL, he was
pp. 855–866, Mar. 2017. engaged in solar PV and energy storage research in Australia and collabo-
[28] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, ‘‘Mitigation of rooftop solar rated with multiple Australian utilities. From 2005 to 2010, he also worked
PV impacts and evening peak support by managing available capacity of as an Electric Power Industry Professional in Bangladesh. He held various
distributed energy storage systems,’’ IEEE Trans. Power Syst., vol. 28, positions from plant maintenance engineer to consulting engineer working
no. 4, pp. 3874–3884, Nov. 2013. with the government agencies responsible for power sector development.

VOLUME 9, 2021 12201

 M. J. E. Alam et al.: Battery Energy Storage to Mitigate Rapid Voltage/Power Fluctuations in Power Grids

KASHEM M. MUTTAQI (Senior Member, IEEE) DANNY SUTANTO (Senior Member, IEEE)
received the B.Sc. degree in electrical and elec- received the B.Eng. (Hons.) and Ph.D. degrees in
tronic engineering from the Bangladesh University electrical engineering from the University of West-
of Engineering and Technology (BUET), Dhaka, ern Australia, Perth, WA, Australia, in 1978 and
Bangladesh, in 1993, the M.Eng.Sc. degree in elec- 1981, respectively. He is currently a Professor of
trical engineering from the University of Malaya, power engineering with the University of Wollon-
Kuala Lumpur, Malaysia, in 1996, and the Ph.D. gong, Wollongong, NSW, Australia. His research
degree in electrical engineering from Multime- interests include power system planning, power
dia University, Selangor, Malaysia, in 2001. From system emergency, analysis and harmonics, flex-
2000 to 2002, he was as a Research Fellow with ible alternating current transmission systems, and
the Queensland University of Technology, Brisbane, QLD, Australia. From battery energy storage systems. From 2014 to 2017, he was the IEEE Industry
2002 to 2007, he was associated as a Research Fellow/Lecturer/Senior Lec- Applications Society Area Chair for Region 10 (Asia Pacific).
turer with the University of Tasmania, Hobart, TAS, Australia. He is currently
a Professor with the School of Electrical, Computer, and Telecommunica-
tions Engineering, University of Wollongong, Wollongong, NSW, Australia.
He also worked as a Lecturer with Multimedia University for a period of
three years. He has more than 22 years of academic experience and authored
or coauthored 385 papers in international journals and conference proceed-
ings. His research interests include distributed generation, renewable energy,
electrical vehicles, smart-grid, and power system planning and control.

12202 VOLUME 9, 2021