Electric Power Systems Research 228 (2024) 110015

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Analytical voltage sensitivity-based distributed Volt/Var control for

mitigating voltage-violations in low-voltage distribution networks
K. Rushikesh Babu ∗, Dheeraj K. Khatod
Department of Electrical Engineering, Indian Institute of Technology Roorkee, Roorkee, 247667, Uttarakhand, India

ARTICLE INFO ABSTRACT

Keywords: Integrating solar photovoltaic system in low-voltage distribution networks leads to significant voltage vio-
Smart inverter lations. This issue can be alleviated using the cutting-edge control techniques (Volt/Var control, Volt/Watt
Volt/Var control control, etc.) of smart inverters. Moreover, the smart inverters absorb/inject the necessary amount of reactive
Voltage sensitivity analysis
power from/into the substation to alleviate the over/under voltages. This increases the reactive power burden
Hosting capacity
on the substation and deteriorates its power factor. Therefore, this paper proposes a novel distributed Volt/Var
Utilization factor of smart inverter
Substation power factor improvement
control technique to reduce the dependency on substations for reactive power support and improve its power
factor. The proposed technique is based on analytical voltage sensitivity analysis. By developing suitable
relations, the feeder nodes are categorized into two groups, one operating in lagging mode, while the other in
leading mode. This ensures proper utilization of reactive power capability of the inverters. An IEEE 13-node
low voltage distribution network with the solar photovoltaic system at different load conditions is used to test
the proposed and other existing Volt/Var control techniques. Compared to other distributed Volt/Var control
techniques, the proposed method requires a minimum reactive power support to regulate the feeder voltage.
Additionally, the obtained results by the proposed method are closer to those by the optimal reactive power
support approach using GAMS software and are validated on a real-time digital simulator platform.

1. Introduction make them less effective for voltage regulation with the fast-changing
SPV generation. Therefore, the SPV inverters, nowadays, are featured
1.1. Background with more sophisticated capabilities, such as Volt/Var control (VVC),
Volt/Watt control, constant power factor mode, voltage/frequency ride-
Large-scale penetration of solar photovoltaic (SPV) systems into the through capabilities, etc., to support the grid [9]. Such inverters are
existing low-voltage distribution networks (LVDNs) has given rise to referred to as Smart Inverters (SIs). As per the IEEE 1547.8 standard
several operating challenges and issues. Among various issues, the volt- and Rule 21 guidelines [10], reactive power support (RPS) from SIs is
age violation is the most frequent and challenging problem encountered prioritized over other voltage regulating techniques to optimize active
in LVDNs [1–3]. In [4], the impact of length and impedance of the power generation and to improve the hosting capacity of the LVDNs
feeder on voltage rise in LVDNs with high SPV proliferation has been with high penetration of SPV.
studied. According to the ANSIC84.1 [5], the voltage at LVDNs should
lie between 0.95 to 1.05 pu.
1.2. Bibliographic review on reactive power support
The methods to regulate voltage in LVDNs are entirely different
from those in transmission grids because of the different R/X ratios of
distribution and transmission lines. The traditional reactive power com- Based on the implementation of the control approach, various volt-
pensation devices, such as on-load tap changing transformers, voltage age regulation techniques employing RPS from SIs can be broadly
regulators, capacitor banks, etc., have no control over voltage regula- categorized as constant reactive power, constant power factor and
tion at the electric point of connection of LVDNs with high penetration variable voltage regulation technique. In the constant reactive power
of SPV [6]. The inverters of SPV systems coordinate with legacy de- method, each SI provides a predetermined value of lagging or leading
vices to improve voltage regulation by properly utilizing their reactive RPS based on over or under voltage, respectively, in the feeder. Due to
power capacity [7,8]. However, certain limitations (particularly, slug- this, it demands higher reactive power from the substation even at low
gish response and frequent maintenance/replacement) of these devices SPV generation levels, which increases network losses and deteriorates

∗ Corresponding author.
E-mail address: [email protected] (K.R. Babu).

https://doi.org/10.1016/j.epsr.2023.110015
Received 3 June 2023; Received in revised form 24 October 2023; Accepted 13 November 2023
Available online 17 November 2023
0378-7796/© 2023 Elsevier B.V. All rights reserved.
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

the constant reactive power approach. However, due to the non-optimal

Nomenclature control approaches, both methods are ineffective on weak LVDNs with
high SPV penetration [11,12]. In order to overcome the limitations of
Indices and Sets constant reactive power and constant power factor techniques, VVC, an
𝑖, 𝑗 Index for the node advanced variable voltage regulation method, has been widely adapted
in networks with significant SPV penetration. Under this approach, the
𝑖𝑗 Index for the branch
deployment of control schemes determines the necessary RPS from each
𝛬𝑖 ∕𝛬𝑆𝑆 Set of nodes connected to node 𝑖/substation
SI.
𝛬𝐵 ∕𝛬𝑆𝑃 𝑉 Set of network/SPV unit nodes Based on the communication infrastructure, VVC schemes are clas-
Variables sified as (i) centralized control and (ii) decentralized control [13].
Under centralized voltage control, the optimal setting of all voltage
𝑁 Number of nodes (excluding the substation regulating devices (including SIs) is decided/coordinated centrally.
bus) and branches Therefore, this control is usually formulated as an optimization problem
𝑉𝑖 Magnitude of node voltage at node 𝑖 with minimization of voltage deviations as primary objective along
𝛥𝑉𝑖 Change in node voltage magnitude at node with minimization of network losses and reactive power procurement,
𝑖 etc. [14]. Though this control is adequate for relatively long-term
𝛼𝑖 Phase angle at node 𝑖 operations (e.g., hourly), it is hard to deal with real-time operations
𝑃𝑖 , 𝑄𝑖 Active, reactive power injection at node 𝑖 due to its complex communication and control schemes [15]. Thus, the
𝑃𝑖𝑗 , 𝑄𝑖𝑗 Active, reactive power flow through branch practical implementation of centralized voltage control scheme is diffi-
𝑖𝑗 cult at LVDNs with fast-moving SPV generation. Under decentralized
voltage control, the optimal setting of all voltage regulating devices
𝑃𝑆 , 𝑄𝑆 Active, reactive power intake from main
is coordinated locally rather than centrally. The decentralized voltage
grid through substation
control can be further classified as local and distributed control based
𝑄𝑔,𝑖 Reactive power output of SI at node 𝑖
on coordination level. Under local control, the optimal setting of each
Parameters voltage regulating device is decided individually based on the local
information available at its own electric point of connection. Though
𝑔𝑖𝑗 , 𝑏𝑖𝑗 Conductance, susceptance on branch 𝑖𝑗 this control has a quick response time and does not require any commu-
𝑅, 𝑋 Equivalent resistance and reactance matrix nication infrastructure, it often results in globally non-optimal control
𝑆𝑖𝑆𝑃 𝑉 , 𝑆𝑖𝑆𝐼 Rated capacity of SPV, SI at node 𝑖 solutions due to the lack of coordination [13,15]. Unlike centralized
𝑆𝑖𝑗𝐵 Rated capacity of branch 𝑖𝑗 control, distributed control requires only the local information from the
𝑃𝑔,𝑖 Active power generation of SPV at node 𝑖 neighboring SIs to take the necessary actions. The optimal setting of all
𝑃𝑑,𝑖 , 𝑄𝑑,𝑖 Active, reactive power demand at node 𝑖 voltage regulating devices is coordinated among each other to achieve
𝑉 𝑈 𝐿 , 𝑉 𝐿𝐿 Upper, lower voltage magnitude limit a global solution. In terms of implementation, distributed control is
𝛼 𝑈 𝐿 , 𝛼 𝐿𝐿 Upper, lower phase angle limit further divided into offline and online. Under offline control, each SI is
given a new operating point after the convergence of an optimization
𝑆𝑆𝑙𝑖𝑚 Maximum apparent power limit of substa-
problem. However, offline control may lead to inappropriate operating
tion transformer
points, if the convergence time of optimization problem does not match
𝑉𝑖𝑛 , 𝑉𝑖𝑜 New, old node voltage magnitude at node 𝑖
𝑑𝑠𝑟
the fast-moving SPV generation. On the other hand, in online control,
𝑉𝑙𝑖𝑚 , 𝑉𝑙𝑖𝑚 Standard, desired Node voltage magnitude
the RPS of each SI is updated after a single-step calculation using
limit
an analytical approach [13]. Due to quick response and advances in
𝑉𝑐𝑟𝑖 Critical node voltage magnitude metering infrastructure, it is becoming increasingly popular nowadays.
𝑄𝑚𝑎𝑥
𝑔,𝑖 Maximum available Vars of SI Voltage sensitivity analysis (VSA) is the most common tool for
𝛥𝑃𝑖 , 𝛥𝑄𝑖 Change in active, reactive power injection various voltage control schemes on LVDNs. Based on the computation
at node 𝑖 method, it is divided into two primary categories, i.e. numerical and
𝑄𝑛𝑖 , 𝑄𝑜𝑖 New, old reactive power injection at node analytical methods. Numerical VSA methods, such as the Newton–
𝑖 Raphson load flow and perturb-and-observe-based methods, use iter-
cos𝜙𝑠 Substation power factor ative algorithms to achieve the solution. Due to iterative algorithms,
these methods have a high computational cost. On the other hand,
Abbreviations analytical VSA methods provide a closed-form solution from suitable
SPV Solar photovoltaic VSA expressions and, thus, overcome the shortcomings of numerical
LVDNs Low-voltage distribution networks methods [16,17].
In [11,18,19], the required RPS from each SI is locally computed
SIs Smart Inverters
using the Newton–Raphson load flow method. The impact of short
RPS Reactive power support
circuit reactance of distribution transformer on voltage deviations is
RPD Reactive power demand discussed in [11]. However, the constraint on RPS from each SI is
RPC Reactive power compensation not taken into consideration. As a result, the SIs at the far end of
RPO Reactive power optimization the feeder operate at a lower power factor, which increases the line
VVC Volt/Var control losses. The distributed VVC is implemented using a piece-wise linear
VSA Voltage sensitivity analysis droop control in coordination with legacy voltage regulating devices
RTDS Real-time digital simulator in [18]. However, based on the SPV location, the droop parameters
need to be adjusted with the variable power injection at other nodes.
In order to regulate the voltage at a node, the unused reactive power
capacity of nearby SIs is used when the SI of that node does not have
the substation power factor. In the constant power factor method, the enough reactive power capacity [19]. A two-stage distributed VVC-
RPS of SI is proportional to the active power generation from SPV. As a based reactive power management is addressed in [20]. In the first
result, the network losses at low SPV generation are less as compared to stage, each SI provides its own RPS to regulate the voltage. If the

2
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

voltage does not reach its desired level, the RPS from other SIs is 2.1. Voltage expression
considered in the second stage. In the methods presented in [19,20],
the kVA rating of SIs is considered equal to the kWp rating of the A typical 𝑁-node radial LVDNs with the SPV unit and load demand
SPV system. Therefore, these methods may not be helpful when most at each node is shown in Fig. 1. Here, each SPV is connected to the
SIs operate at their full kWp rating. Hence, to optimize the utilization LVDNs through its SI and the entire system is referred to as the SPV
of available SPV generation, the kVA rating of SIs is considered 1.1 unit. Each SPV unit can provide the RPS during over/under voltage
times the kWp rating of the SPV system [21]. This ensures around 44% conditions. In this figure, 𝑧𝑖 (= 𝑟𝑖 + 𝑗 𝑥𝑖 ) represents the impedance of
RPS even under the maximum active power generation condition of 𝑖th branch. All the parameters are represented in pu.
SPV. Moreover, the selection of appropriate droop control parameters Considering the power flow from the SPV node to the main grid
is not addressed in the above literature. In contrast, a uniform active node, as shown in Fig. 1, the column vector (𝑉 ) representing the
power curtailment and RPS are advised to design the droop control approximate node voltage can be expressed as [22,23]:
parameters in [22]. However, the effect of voltage sensitivity on active
𝑉 ≅ 𝑉0 + 𝑅.𝑃 + 𝑋.𝑄 (1)
power curtailment and RPS has not been considered.
where, 𝑉 , 𝑉0 , 𝑃 (= 𝑃𝑔 − 𝑃𝑑 ) and 𝑄 (= 𝑄𝑔 − 𝑄𝑑 ) are the column vectors
1.3. Contributions and paper framework of size 𝑁; and 𝑅 and 𝑋 are the symmetrical square matrices of size
𝑁. 𝑅 and 𝑋 are obtained by multiplying branch-current to bus-voltage
matrix with bus-injection to branch-current matrix [24].
Though various voltage regulation techniques based on RPS through
Due to the high 𝑅∕𝑋 value of LVDNs, a significant rise (or drop)
SIs have been discussed in the above literature, several issues still
in 𝑉 may be observed with high SPV generation and low demand con-
need to be addressed in the near future. These issues include the
dition (or low SPV generation and high demand condition) according
proper setting of the VVC (leading or lagging) based on the voltage
to (1). This causes over/under voltage violations and may lead to the
sensitivity, adequate setting of control parameters of SI, cost-effective
unintentional islanding of the SPV unit from the LVDNs. Therefore, to
coordination among SIs, voltage regulation with minimum RPS from avoid this situation, each SI must absorb/inject the required reactive
the substation, etc. To address some of the existing limitations of SI- power to support LVDNs locally [10]. The voltage sensitivity helps to
RPS based voltage regulation techniques, this paper proposes a novel calculate the required Vars on the basis of distance of the SPV from the
distributed VVC technique using the analytical VSA. The proposed main grid node.
technique is tested on an IEEE 13-node LVDN with SPV system at
different load conditions. It requires minimum RPS from the substation
2.2. Voltage sensitivity
to alleviate the voltage violation compared to the existing distributed
VVC techniques. The performance of the proposed VVC technique is
Voltage sensitivity is defined as the change in voltage magnitude
validated by the real-time digital simulation (RTDS) platform for over-
at a particular node due to the change in net active/reactive power
voltage and under-voltage conditions. The main contributions of this
injection at other nodes of the LVDNs [25]. Considering the main
paper are:
grid voltage, 𝑉0 to be constant and using (1), the change in voltage
magnitude due to change in net active and reactive power injection
• Using an analytical VSA approach, this paper presents suitable
can be expressed as:
mathematical expression to identify the critical nodes causing
voltage violation. 𝛥𝑉 ≅ 𝑅.𝛥𝑃 + 𝑋.𝛥𝑄 (2)
• This paper develops suitable mathematical relations to identify
and to separate the feeder nodes into two groups. One group of where, 𝛥𝑃 , 𝛥𝑄, and 𝛥𝑉 are the column vectors, each of size 𝑁,
SI-connected nodes is operated in lagging mode, while the other representing the change in net active power injection, the change in
net reactive power injection and the change in node voltage magnitude,
group is operated in leading mode. This minimizes the reactive
respectively.
power from substation, and hence, improves its power factor.
From (2), the sensitivities of node voltage with respect to 𝛥𝑃 and
• To ensure RPS even under the maximum generation from SPV,
𝛥𝑄 can be computed as [22,23]:
SIs are intentionally overrated. Therefore, this work adequately
utilize the reactive power capability of SIs and improves their 𝛥𝑉 .𝛥𝑃 −1 = 𝑅; 𝛥𝑉 .𝛥𝑄−1 = 𝑋 (3)
utilization factor.
• For ensuring no voltage violation at any node, the proposed Thus, matrices 𝑅 and 𝑋 represent the sensitivity of voltage with respect
method first identifies the critical node at which maximum or to (w.r.t.) the active power injection and the sensitivity of voltage
minimum voltage is observed, and then, calculates the desired w.r.t. the reactive power injection, respectively. For a given topology
voltage limit. Thus, the proposed method dynamically adjusts the of LVDNs, 𝑅 and 𝑋 are the constant matrices.
desired voltage limit depending up on voltage violation in the
feeder. 2.3. Reactive power calculation

The remaining portions of this paper are organized as follows. Keeping the net active power injection constant (i.e. 𝛥𝑃 = 0), and
The preliminary topics are given in Section 2. Section 3 discusses considering the change in only reactive power generation, the change in
the existing Volt/Var control methods. A detailed description of the voltage magnitude at any 𝑖th node with the help of (2) can be expressed
proposed VVC method is given in Section 4. The simulation results as:
and comparative study are provided in Section 5. Finally, Section 6 ∑
𝑁
concludes the work presented in this paper and its future prospects. 𝛥𝑉𝑖 = 𝑉𝑖𝑛 − 𝑉𝑖𝑜 = 𝑋𝑖,𝑗 .(𝑄𝑛𝑔,𝑗 − 𝑄𝑜𝑔,𝑗 ) (4)
𝑗=1

2. Preliminaries In order to maintain 𝑉𝑖𝑛 with the pre-specified limits, the following
conditions should be satisfied:

This section discusses the expressions of node voltage, voltage sen- ∑

𝑁
𝑉 𝐿𝐿 − 𝑉𝑖𝑜 ≤ 𝑋𝑖,𝑗 .(𝑄𝑛𝑔,𝑗 − 𝑄𝑜𝑔,𝑗 ) (5)
sitivity and the utilization factor of SI in LVDNs. 𝑗=1

3
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

Fig. 1. A typical N -node radial distribution feeder.

3.1.1. Uniform RPS at all nodes

∑
𝑁
𝑉𝑖𝑜 − 𝑉 𝑈 𝐿 ≤ − 𝑋𝑖,𝑗 .(𝑄𝑛𝑔,𝑗 − 𝑄𝑜𝑔,𝑗 ) (6) In this case, all the SPV-connected nodes of LVDNs are provided
𝑗=1 with the same RPS. Using (7) and considering the value of 𝑄𝑛𝑔,𝑖 ,
The above two equations can be combined by taking a common ∀𝑖 = 1, 2 … 𝑁 to be same at all the nodes, the required RPS, 𝑄𝑛𝑔,𝑖 is
voltage limit, 𝑉𝑙𝑖𝑚 for the lower and upper voltage cases as: obtained as:
∑
±(𝑉𝑙𝑖𝑚 − 𝑉𝑖𝑜 ) ± 𝑁 𝑜
𝑗=1 𝑋𝑖,𝑗 .𝑄𝑔,𝑗
∑
𝑁
𝑄𝑛𝑔,𝑖 = (10)
± (𝑉𝑙𝑖𝑚 − 𝑉𝑖𝑜 ) ≤ ± 𝑋𝑖,𝑗 .(𝑄𝑛𝑔,𝑗 − 𝑄𝑜𝑔,𝑗 ) (7) ∑𝑁
± 𝑗=1 𝑋𝑖,𝑗
𝑗=1
Using (10), the required RPS at each SPV-connected node is calcu-
where, +𝑣𝑒 and −𝑣𝑒 signs are for the under-and over-voltage cases,
lated, and among all, the maximum absolute value of RPS is set for each
respectively.
SI as:
The required RPS at 𝑖th node can be obtained from (7), while
satisfying the following constraint: 𝑄𝑔 = 𝑚𝑎𝑥(|𝑄𝑛𝑔,𝑖 |, ∀𝑖 = 1, 2 … 𝑁) (11)
√ √
( 𝑆𝐼 )2 ( )2 ( 𝑆𝐼 )2 ( )2 In this case, the set point of RPS for each SI corresponds to the
− 𝑆𝑖 − 𝑃𝑔,𝑖 ≤ 𝑄𝑛𝑔,𝑖 ≤ 𝑆𝑖 − 𝑃𝑔,𝑖 (8)
RPS required at the most voltage-violated node regardless of voltage
2.4. Utilization factor of SI violation at individual nodes.

The kVA rating of SIs is usually kept slightly higher than the kWp 3.1.2. Uniform RPS at voltage violated nodes
rating of the SPV system to ensure RPS even under maximum SPV In this case, the nodes experiencing voltage violation are first iden-
active power generation condition [10]. In this case, the performance tified (say, 𝑖 = 𝑚 to 𝑁), and then the uniform RPS for such nodes
of a SI can be assessed by its utilization factor as [26]: are computed. For voltage violated nodes, the required RPS, 𝑄𝑛𝑔,𝑖 is
obtained as:
Operating kVA of SI ∑
Utilization factor of SI = (9) ±(𝑉𝑙𝑖𝑚 − 𝑉𝑖𝑜 ) ± 𝑁 𝑜
𝑗=𝑚 𝑋𝑖,𝑗 .𝑄𝑔,𝑗
kVA rating of SI 𝑄𝑛𝑔,𝑖 = (12)
∑
± 𝑁 𝑗=𝑚 𝑋𝑖,𝑗
3. Existing Volt/Var control methods
Using (12), the required RPS at each voltage-violated node is cal-
This section briefly describes various existing methods to implement culated, and among all, the maximum absolute value of RPS is set for
VVC. The VVC uses the information about voltage deviation at a node each SI of the voltage-violated nodes as:
to decide the required RPS (absorption/injection) from each SI. Fig. 2
𝑄𝑔 = 𝑚𝑎𝑥(|𝑄𝑛𝑔,𝑖 |, ∀𝑖 = 𝑚 … 𝑁) (13)
depicts a generalized volt/var curve for a specific 𝑖th node. It indicates
the process of changing RPS at a node in response to the change in its In this case, the SIs only at the voltage-violated nodes provide the
voltage magnitude. The SI counteracts the deviations in node voltage uniform RPS, and the rest of the SIs operate at the unity power factor.
and brings it to its normal operating region by providing RPS. The Hence, compared to the uniform RPS at all nodes, this case requires
SI can be operated in three regions, as shown in Fig. 2. In a normal less RPS from the main grid [27].
operating region, the voltage at a node lies within its permitted limits;
therefore, the SI provides no RPS (i.e. 𝑄𝑔 = 0). In the over-voltage 3.2. Optimal RPS
region, the SI absorbs the 𝑄𝑔 Vars from the substation while operating
at point A to bring the 𝑉𝑖 to the normal operating region. In the under- In this case, the over/under voltage mitigation is achieved by op-
voltage region, the SI injects 𝑄𝑔 Vars while operating at point B to timally distributing the RPS from the SIs of the LVDNs. Based on the
return the 𝑉𝑖 to the normal operating region. The operating point varies objective function, the amount of RPS at 𝑖th node, 𝑄𝑔,𝑖 is calculated
within the available Vars of each SI. Hence, such SI with 𝑄𝑔 capability using reactive power optimization (RPO). There are several possible
is called a reactive power compensator. The available Vars from an SI formulations for RPO. For the purpose of comparison, an RPO is
is the amount of RPS that can be injected/absorbed without affecting formulated in this paper and is discussed in the following section:
its real power output, and is calculated using (8).
Based on the amount of RPS, the existing VVC methods can be 3.2.1. Objective function
categorized as: The objective of RPO is to determine the decision vector, 𝜒, to
minimize the RPS from the substation. The objective function, 𝑓 can
3.1. Uniform RPS be expressed with the help of (14) as:
∑
𝑓 = min |𝑄𝑆,𝑖 | (14)
This method achieves over/under voltage mitigation by evenly dis- 𝜒
𝑖∈𝛬𝑆𝑆
tributing the RPS from the SIs connected to the LVDNs. Based on the
where,
number of SIs involved in implementing VVC, uniform RPS VVC can
[ ]
further be subdivided as: 𝜒 = 𝑉𝑖 , 𝛼𝑖 , 𝑃𝑖𝑗 , 𝑄𝑖𝑗 , 𝑃𝑆,𝑖 , 𝑄𝑆,𝑖 , 𝑄𝑔,𝑖 (15)

4
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

Fig. 2. Volt/Var control method.

3.2.2. Constraints To ensure the voltage to be within its permissible limits, the voltage
The following constraints have been considered to develop the RPO magnitude, as given by (1), at any 𝑖th node must satisfy the following
model: relation,
( ) ( ) }
𝑃𝑖𝑗 = 𝑔𝑖𝑗 𝑉𝑖2 − 𝑔𝑖𝑗 𝑉𝑖 𝑉𝑗 cos 𝛼𝑖 − 𝛼𝑗 − 𝑏𝑖𝑗 𝑉𝑖 𝑉𝑗 sin 𝛼𝑖 − 𝛼𝑗 ∑
𝑁 ∑
𝑁
( ) ( ) ∀𝑖𝑗 (16)
2
𝑄𝑖𝑗 = −𝑏𝑖𝑗 𝑉𝑖 + 𝑏𝑖𝑗 𝑉𝑖 𝑉𝑗 cos 𝛼𝑖 − 𝛼𝑗 − 𝑔𝑖𝑗 𝑉𝑖 𝑉𝑗 sin 𝛼𝑖 − 𝛼𝑗 𝑉 𝐿𝐿 ≤ 𝑉0 + 𝑅𝑖,𝑗 .𝑃𝑗 + 𝑋𝑖,𝑗 .𝑄𝑗 ≤ 𝑉 𝑈 𝐿 (23)
√ 𝑗=1 𝑗=1
( )2 ( )2
𝑃𝑖𝑗 + 𝑄𝑖𝑗 ≤ 𝑆𝑖𝑗𝐵 ∀𝑖𝑗 (17) Taking a voltage limit, 𝑉𝑙𝑖𝑚 for the lower and upper voltage cases,
∑ } the above equation can also be expressed as
𝑃𝑆,𝑖 + 𝑃𝑔,𝑖 − 𝑃𝑑,𝑖 = 𝑗∈𝛬𝑖 𝑃𝑖𝑗
∑ ∀𝑖 ∈ 𝛬𝐵 (18) ∑
𝑁 ∑
𝑁
𝑄𝑆,𝑖 + 𝑄𝑔,𝑖 − 𝑄𝑑,𝑖 = 𝑗∈𝛬𝑖 𝑄𝑖𝑗 ±( 𝑅𝑖,𝑗 .𝑃𝑗 + 𝑋𝑖,𝑗 .𝑄𝑗 ) ≤ ±(𝑉𝑙𝑖𝑚 − 𝑉0 ) (24)
∑ ⎫ 𝑗=1 𝑗=1
𝑃𝑆,𝑖 = 𝑗∈𝛬𝑖 𝑃𝑖𝑗
∑ ⎪ 𝑆𝑆 where, +𝑣𝑒 and −𝑣𝑒 signs are for the over- and under-voltage cases,
𝑄𝑆,𝑖 = 𝑗∈𝛬𝑖 𝑄𝑖𝑗 ⎬ ∀𝑖 ∈ 𝛬 (19)
𝑉𝑖 = 1 𝑎𝑛𝑑 𝛼𝑖 = 0⎭ ⎪ respectively.
} Now, rearranging the terms of (24) with the considered assumptions
𝑉 𝐿𝐿 ≤ 𝑉𝑖 ≤ 𝑉 𝑈 𝐿 leads to the following VSA expression:
∀𝑖 ∈ 𝛬𝐵 ∕𝛬𝑆𝑆 (20)
𝛼 𝐿𝐿 ≤ 𝛼𝑖 ≤ 𝛼 𝑈 𝐿 ∑𝑁 ∑
√ 𝑗=1 𝑅𝑖,𝑗 (𝑉𝑙𝑖𝑚 − 𝑉0 − 𝑁 𝑗=1 𝑋𝑖,𝑗 .𝑄𝑗 )
( )2 ( )2 | ∑𝑁 |≤| ∑𝑁 | (25)
𝑃𝑆,𝑖 + 𝑄𝑆,𝑖 ≤ 𝑆𝑆𝑙𝑖𝑚 ∀𝑖 ∈ 𝛬𝑆𝑆 (21)
𝑗=1 𝑋𝑖,𝑗 𝑗=1 𝑋𝑖,𝑗 .𝑃𝑗
√ √
( 𝑆𝐼 )2 ( )2 ( 𝑆𝐼 )2 ( )2 The above relation represents a condition which can be used to
− 𝑆𝑖 − 𝑃𝑔,𝑖 ≤ 𝑄𝑔,𝑖 ≤ 𝑆𝑖 − 𝑃𝑔,𝑖 ∀𝑖 ∈ 𝛬𝑆𝑃 𝑉 (22) check the voltage volition at a given node.
The flow of active and reactive power through any 𝑖𝑗th branch is
given in (16), and bounded by (17). The active and reactive power 4.1. Modes of VVC implementation
balance at any 𝑖th node is expressed in (18). The intake of active and
reactive power from the main grid and the bounds on the voltage Usually, shunt capacitors are installed in LVDNs to provide RPS
variable of the main grid through substation is given by (19). The at normal voltage levels. However, during over/under voltage, the
bounds on voltage variables are given in (20). The bounds on apparent SIs are given priority to provide the RPS locally through VVC to
power intake from the main grid through the substation transformer mitigate voltage violation [28]. Under uniform RPS methods, the SIs
are provided in (21). The reactive power output from SI is limited by are operated uniformly in either an inductive (lagging) or a capaci-
(22). The RPO model, given by (14)–(22), is a nonlinear programming tive (leading) mode to absorb/inject the reactive power to alleviate
problem, which is coded in GAMS and solved with a CONOPT solver. over/under voltages. This uniform RPS throughout the LVDNs increases
the reactive power burden on the substation and deteriorates its power
4. Proposed methodology factor. Therefore, based on the voltage magnitudes, the SIs must be
operated in an appropriate VVC mode (i.e. inductive/capacitive) to
This section presents a detailed description of the proposed VVC improve the substation power factor. To achieve the same, this paper
method. The proposed method is based on the analytical VSA [27], proposes a novel VSA based approach to identify and operate the
which helps separate the normal and voltage-violated nodes from the appropriate nodes under either reactive power compensation (RPC)
LVDNs. The following assumptions are made to derive the expressions mode or reactive power demand (RPD) mode, if there is a voltage
for the proposed method: (1) There is a uniform active and reactive violation.
power injection at all the nodes, (2) The node voltage of the main Let the nodes from 𝑖 = 𝑚 to 𝑁 violate the permissible voltage limit.
grid node, 𝑉0 is constant, and (3) The considered LVDNs has a similar To bring the voltage of such nodes within their permissible limit, the
daily load and generation patterns at all the nodes, therefore, the nodes from 𝑖 = 𝑚 to 𝑁 are set to operate under RPD mode, and the SIs
over and under-voltage conditions cannot occur simultaneously in such at these nodes absorb the reactive power during over-voltage and inject
networks. the reactive power during under-voltage. At the same time, to minimize

5
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

the reactive power burden on the substation, the nodes from 𝑖 = 1 to Table 1
System parameters.
𝑚 − 1 are set to operate under RPC mode, and the SIs at these nodes
inject the reactive power during over-voltage and absorb the reactive Parameters Value

power during under-voltage. Node to node distance 100 m

Line impedance 0.549 + j0.072 Ω/km
Thus, the proposed VVC method sets the SIs of voltage-violated
Distribution transformer 100 kVA, 11/0.4 kV, X = 8%, R = 1%
nodes to operate in RPD mode and the SIs of normal nodes (without SPV size 9 kWp
voltage-violation) to operate in RPC mode. In contrast, the uniform RPS Smart inverter rating 10 kVA
methods set the SIs to operate only in RPD mode. Due to this, there Base load demand 2 kVA, 0.95 lag
is a need to revisit the voltage limit, 𝑉𝑙𝑖𝑚 considered for uniform RPS Peak load demand 4 kVA, 0.95 lag

methods. This is because the necessary RPS computed using (12) with
the given 𝑉𝑙𝑖𝑚 may bring the voltage of the nodes under RPD mode
within the permissible range. However, when the SIs under RPC mode 4.1.1. RPD mode
provide the counter-reactive power to the nodes under RPD mode, there In this mode, the required uniform RPS, 𝑄𝑛𝑔,𝑖 at voltage violated
is a high possibility of violation of 𝑉𝑙𝑖𝑚 at these nodes again. Hence, to nodes is computed as:
ensure all the node voltages are within 𝑉𝑙𝑖𝑚 , the proposed VVC method 𝑑𝑠𝑟 − 𝑉 𝑜 ) ± ∑𝑁 𝑋 .𝑄𝑜
±(𝑉𝑙𝑖𝑚
first observes the critical node voltage and then calculates the desired 𝑖 𝑗=𝑚 𝑖,𝑗 𝑔,𝑗
𝑄𝑛𝑔,𝑖 = ∑ (29)
voltage limit as: ± 𝑁 𝑋
𝑗=𝑚 𝑖,𝑗
𝑑𝑠𝑟
𝑉𝑙𝑖𝑚 = 𝑉𝑙𝑖𝑚 − (|𝑉𝑐𝑟𝑖 − 𝑉𝑙𝑖𝑚 |) (26) Using (29), the required RPS at each voltage violated node is
calculated, and among all, the maximum absolute value of RPS is set
where, 𝑉𝑐𝑟𝑖 is the voltage of the critical node at which maximum or for each SI of the voltage-violated nodes similar to (13).
minimum voltage is observed, and 𝑉𝑙𝑖𝑚 can be replaced with 𝑉𝑈 𝐿 for
over-voltage and 𝑉𝐿𝐿 for under-voltage scenario. 4.1.2. RPC mode
𝑑𝑠𝑟 is greater than and
The relation (26) ensures that the value of 𝑉𝑙𝑖𝑚 In this mode, the required uniform RPS, 𝑄𝑛𝑔,𝑖 at each normal node
less than the value of 𝑉𝑙𝑖𝑚 during under-voltage and over-voltage, re- (without voltage violation) is computed as:
𝑑𝑠𝑟 − 𝑉 𝑜 ) ± ∑𝑚−1 𝑋 .𝑄𝑜
𝑑𝑠𝑟 is calculated
spectively. In the proposed method, first, the value of 𝑉𝑙𝑖𝑚
𝑑𝑠𝑟 ±(𝑉𝑙𝑖𝑚 𝑖 𝑗=1 𝑖,𝑗 𝑔,𝑗
from (26) based on the value of 𝑉𝑐𝑟𝑖 , and then, replacing 𝑉𝑙𝑖𝑚 by 𝑉𝑙𝑖𝑚 𝑛
𝑄𝑔,𝑖 = (30)
∑𝑚−1
in (25) leads to the proposed VSA expression: ± 𝑗=1 𝑋𝑖,𝑗
∑𝑁 𝑑𝑠𝑟 − 𝑉 − ∑𝑁 𝑋 .𝑄 )
𝑗=1 𝑅𝑖,𝑗 (𝑉𝑙𝑖𝑚 0 𝑗=1 𝑖,𝑗 𝑗
Using (30), the required RPS at each normal node is calculated, and
| ∑𝑁 |≤| ∑𝑁 | (27) among all, the minimum absolute value of RPS is set for each SI of the
𝑗=1 𝑋𝑖,𝑗 𝑗=1 𝑋𝑖,𝑗 .𝑃𝑗 normal nodes as,
Now, for further analysis and discussion, the LHS and RHS parts of 𝑄𝑔 = 𝑚𝑖𝑛(|𝑄𝑛𝑔,𝑖 |, ∀𝑖 = 1 … 𝑚 − 1) (31)
(27) can be represented by variables 𝐾𝑖 and 𝐾𝑐,𝑖 , respectively, as:
The calculated RPS at each node under RPC or RPD mode must
∑𝑁 ⎫ satisfy (8). If the calculated RPS violates (8), then it is set equal to the
𝑗=1 𝑅𝑖,𝑗 ⎪
𝐾𝑖 = | ∑𝑁 |
𝑗=1 𝑋𝑖,𝑗 ⎪ violating limit.
𝑑𝑠𝑟 −𝑉 −∑𝑁 𝑋 .𝑄 ) ⎬
∀𝑖 = 1 … 𝑁 (28)
(𝑉𝑙𝑖𝑚 0 𝑗=1 𝑖,𝑗 𝑗 ⎪
𝐾𝑐,𝑖 = | ∑𝑁 | 4.2. Algorithm of proposed VVC technique
𝑗=1 𝑋𝑖,𝑗 .𝑃𝑗
⎪
⎭
The voltage sensitivity ratio, 𝐾𝑖 of any 𝑖th node, is the ratio of the In the proposed VVC technique, the following broads steps are
sensitivity of its voltage magnitude w.r.t. net active power injection involved to regulate the voltage of LVDNs:
to that w.r.t. net reactive power injection. It is a constant term and
• Monitor the node voltage magnitude, 𝑉𝑖 at all nodes and find the
depends on the network topology and line parameters. The presence
node with maximum voltage violation to set as the critical node
of a distribution transformer in LVDNs makes 𝐾𝑖 < 1 for the nodes voltage magnitude, 𝑉𝑐𝑟𝑖
closer to the main grid. Thus, the voltages at such nodes are more
• Calculate 𝑉𝑙𝑖𝑚
𝑑𝑠𝑟 , 𝐾 and 𝐾
𝑖 𝑐,𝑖 using (26) and (28), respectively
sensitive w.r.t. reactive power injection than active power injection.
• Set the nodes for RPD or RPC modes based on 𝐾𝑖 and 𝐾𝑐,𝑖 values
As the distance of a node from the main grid increases, the value
𝐾𝑖 also increases due to the dominance of the resistance term over • Calculate the required RPS, 𝑄𝑔,𝑖 as discussed in previous sub-
the reactance term. This makes 𝐾𝑖 > 1 for the nodes far from the section
main grid. The voltages at such nodes are more sensitive w.r.t. active • Update the RPS set point for each SI
power injection than reactive power injection. On the other hand, the
5. Results and discussion
voltage violation ratio, 𝐾𝑐,𝑖 , of any 𝑖th node varies in real-time with
the net active and reactive power injection. Usually, in LVDNs, the
This section presents the simulation results of proposed and existing
net active power injection is much higher than the net reactive power VVC methods and their comparative study. A modified 13-node residen-
injection. Due to this, the denominator component of 𝐾𝑐,𝑖 increases with tial LVDNs has been considered to test the effectiveness of the proposed
an increase in the distance of a node from the main grid in contrast to VVC method, which is a part of the IEEE 34-bus system. With three
the numerator component, which is the difference between a constant laterals, the test system has seven identical SPV units at nodes 2, 4, 5,
𝑑𝑠𝑟 − 𝑉 ) and the product of 𝑋 times the net reactive power
value (i.e. 𝑉𝑙𝑖𝑚 0 7, 8, 11, and 13, as depicted in Fig. 4 [27,29]. A balanced three-phase
injection. Therefore, the value 𝐾𝑐,𝑖 is higher at nodes closer to the main identical R-L load with a 0.95 power factor has been taken at all the
grid than at nodes away from the main grid. As the distance of a node nodes. For simulation purposes, the parameters of the LVDNs are given
from the main grid increases, the value 𝐾𝑖 rises, while the value of 𝐾𝑐,𝑖 in Table 1. At each node, the same ratings of SPV unit and load demand
falls. are considered as given in Table 1. The value of 𝑉𝑙𝑖𝑚 is considered 0.95
The nodes not satisfying (27) violate the voltage limit. Thus, the pu and 1.05 pu for under- and over-voltage, respectively, complying
set of nodes under RPC or RPD mode is identified. Later, the re- ANSIC84.1 [5].
quired RPS at various nodes is computed by (12). Different modes of The results obtained by the proposed technique have been compared
implementation of the proposed VVC method are shown in Fig. 3. with those by the following existing VVC methods/techniques:

6
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

Fig. 3. Modes of proposed VVC implementation.

Fig. 4. Modified 13-node residential LVDNs.

• Method A: Uniform RPS at all nodes [22]

• Method B: Uniform RPS at voltage violated nodes [27]
• Method C: Optimal RPS [28]

Various VVC methods are implemented under MATLAB R2020b and

RTDS platform using a personal computer with an Intel(R) Core(TM)
i7-8700 CPU @ 3.20 GHz and 16 GB internal memory. Further, to test
the robustness of the proposed technique over the existing VVC tech-
niques, both over- and under-voltage scenarios/conditions in the test
system are considered. The over-voltage occurs during maximum SPV
generation (daylight hours) with the base load demand. In contrast, the
under-voltage occurs during minimal SPV generation (evening hours)
with the peak load demand. The test system is simulated in RTDS,
and the under or over-voltage condition is created in the test system.
The node voltage profile obtained from RTDS is given as input to the
MATLAB environment to execute the VVC method and, thereby, to get
the RPS set points for SIs. These RPS set points are given as input to
the RTDS platform to check their impact on the voltage profile of the
test system. The interfacing between MATLAB and RTDS platforms is
shown in Fig. 5. The scenario-wise discussion of the results is presented
in the following sub-sections.
Fig. 5. Interfacing between MATLAB and RTDS platforms.

5.1. Over-voltage scenario

The over-voltage scenario is created with base load and 100% SPV in capacitive mode. It is observed from Fig. 6 that 𝐾𝑖 > 𝐾𝑐,𝑖 for node 7
generation. Under this scenario, different VVC techniques are applied onwards, and among different nodes, the maximum voltage violation is
to the test system to mitigate the over-voltage, and the obtained results observed at node 11. Hence, considering node 11 as the critical node,
are presented in Figs. 6–8, and Tables 2–3. the SIs at nodes 7 to 13 are operated at the RPD (inductive) mode, and
In method A, a uniform lagging RPS is set at all nodes having the SIs at nodes 1 to 6 are operated at the RPC (capacitive) mode in
SPV irrespective of voltage violation at individual nodes (Fig. 7). It is the proposed VVC method (Fig. 7).
observed from Fig. 6 that the voltage at nodes 8 to 13 exceeds its upper After implementing various VVC methods, the obtained voltage pro-
limit (i.e. 1.05 pu). Therefore, in method B, the SIs at nodes 8 to 13 are files are shown in Fig. 8, which indicates that all methods considered
operated at the inductive (lagging) mode, and the SIs of the remaining can mitigate the over-voltage in the test system and that the node
nodes are operated at the unity power factor (Fig. 7). In method C, voltages are within their maximum limit.
the required RPS at different nodes is computed after solving the RPO, Further, the net reactive power required from the substation to
which aims to minimize the net RPS from the substation. As a result, mitigate over-voltage in the test system in method A, method B, method
some nodes are operated in inductive mode, and others are operated C and the proposed method is 7.0573 kVAR, 6.1666 kVAR, 2.5077

7
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

Fig. 6. Voltage profile without RPS and voltage sensitivity indices for over-voltage scenario.

Fig. 7. Reactive power support by SIs in various VVC methods for over-voltage scenario.

Fig. 8. Voltage profiles in various VVC methods for over-voltage scenario.

kVAR, and 2.7289 kVAR, respectively (Table 2). Out of the different substation, resulting in the best power factor of the substation. Method
methods considered, method C requires the net minimum RPS from the C also best utilizes the SIs (Table 3). The results of the proposed method

8
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

Fig. 9. Voltage profile without RPS and voltage sensitivity indices for under-voltage scenario.

Table 2 Table 4
Summary of results for over-voltage scenario. Summary of results for under-voltage scenario.
Parameters Method A Method B Method C Proposed Parameters Method A Method B Method C Proposed
Total 𝑄𝑔 (kVAR) 7.0573 6.1666 2.5077 2.7289 Total 𝑄𝑔 (kVAR) −52.6865 −50.7031 −47.9455 −48.1002
Net 𝑃𝑠 (kW) −36.0227 −35.882 −36.0469 −35.7621 Net 𝑃𝑠 (kW) 47.3839 47.872 48.3047 48.633
Net 𝑄𝑠 (kVAR) 16.177 15.239 11.6332 11.7362 Net 𝑄𝑠 (kVAR) −34.2665 −32.2866 −29.439 −29.059
cos𝜙𝑠 0.9122 0.9204 0.9516 0.9501 cos𝜙𝑠 0.8103 0.829 0.8539 0.8584

Table 5
Table 3 Utilization factor of SIs for under-voltage scenario.
Utilization factor of SIs for over-voltage scenario. Node Method A (%) Method B (%) Method C (%) Proposed (%)
Node Method A (%) Method B (%) Method C (%) Proposed (%)
2 75.80 9 100 100
2 90.56 90 100 100 4 75.80 84.98 100 97.18
4 90.56 90 100 100 5 75.80 84.98 81.57 97.18
5 90.56 90 100 100 7 75.80 84.98 100 97.18
7 90.56 90 93.43 98.29 8 75.80 84.98 100 97.18
8 90.56 92.32 100 98.29 11 75.80 84.98 100 97.18
11 90.56 92.32 100 98.29 13 75.80 84.98 100 97.18
13 90.56 92.32 100 98.29

Further, the net reactive power required from the substation to

mitigate under-voltage in the test system in method A, method B,
are closer to method C than those of other distributed uniform RPS
methods (method A and method B). method C and the proposed method is 52.6865 kVAR, 50.7031 kVAR,
47.9455 kVAR, and 48.1002 kVAR, respectively (Table 4). Out of the
5.2. Under-voltage scenario different methods considered, method C requires the net minimum
RPS from the substation, resulting in the best power factor of the
The under-voltage scenario is created with peak load and 10% SPV substation. Method C also best utilizes the SIs (Table 5). The results
generation. Under this scenario, different VVC techniques are applied to of the proposed distributed method are closer to method C than those
the test system to mitigate the under-voltage, and the obtained results of other distributed uniform RPS methods (method A and method B).
are presented in Figs. 9–11, and Tables 4–5. Summarizing the results for over- and under-voltage scenarios
In method A, a uniform leading RPS is set at all nodes having among the distributed VVC methods (method A, method B, and pro-
SPV (Fig. 10). In method B, the SIs at nodes 4 to 13 are operated in posed method), the proposed VVC technique requires a minimum RPS
the capacitive (leading) mode, and the SIs of the remaining nodes are to mitigate voltage-violation in the rest system and hence, results in
operated at the unity power factor (Fig. 10) as the voltage at nodes an improved power factor of the substation. Further, as observed from
4 to 13 exceeds its lower limit (i.e. 0.95 pu) (Fig. 9). In method C, Tables 3 and 5, the proposed method can utilize the reactive power
some nodes are operated in inductive mode, and others are operated in
capability of SIs in a better way than method A and method B. Among
capacitive mode as per the results of RPO. From Fig. 9, it is observed
the different methods considered, method C sets the optimum RPS for
that 𝐾𝑖 > 𝐾𝑐,𝑖 for node 4 onwards, and among different nodes, the
the SIs to alleviate the voltage violation. Hence, this results in the
maximum voltage violation is observed at node 11. Hence, considering
optimum power factor at the substation as observed from Tables 2 and
node 11 as the critical node, the SIs at nodes 4 to 13 are operated at
the RPD (capacitive) mode, and the SIs at nodes 1 to 3 are operated at 4. The results of the proposed VVC method and method C are close
the RPC (inductive) mode in the proposed VVC method (Fig. 10). to each other. However, to implement method C, there is a mandatory
After implementing various VVC methods, the obtained voltage pro- requirement for appropriate solver/software to solve the RPO model,
files are shown in Fig. 11, which indicates that all methods considered which increases the cost and complexity of the system. On the other
can mitigate the under-voltage in the test system, and those node hand, the proposed method does not require any such software, and
voltages are above their minimum limit. hence, its implementation is simple and straightforward.

9
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

Fig. 10. Reactive power support by SIs in various VVC methods for under-voltage scenario.

Fig. 11. Voltage profiles in various VVC methods for under-voltage scenario.

6. Conclusion and under-voltage conditions simultaneously. Furthermore, this paper

considers only RPS from SIs to regulate the voltage in LVDNs. However,
This paper proposes a distributed VVC technique to mitigate the due to the high R/X ratio of LVDNs and regulatory limits on power
voltage violation in LVDNs with large SPV penetration. The existing factor of SIs, RPS alone from SIs may not be sufficient to mitigate the
distributed VVC methods set the SIs to operate in uniform RPS with voltage violations. Therefore, the present work can also be extended
either inductive (lagging) or capacitive (leading) modes throughout by incorporating both active and reactive power support from SIs as its
the LVDNs to mitigate over- or under-voltage violations. However, this future prospect.
increases the reactive power burden on the local distribution substa-
tions and hence, deteriorates its power factor. Therefore, the proposed
VVC technique aims to improve the power factor of substation while CRediT authorship contribution statement
minimizing its reactive power burden. This technique uses an analytical
VSA to identify and set the SIs to their most appropriate mode of oper- K. Rushikesh Babu: Data curation, Simulations, Writing – original
ation (i.e. inductive/capacitive). A modified 13-node residential LVDNs draft, Methodology, Visualization. Dheeraj K. Khatod: Conceptualiza-
accommodating SPV units has been considered for implementing differ- tion, Supervision, Resources, Writing – review & editing, Validation.
ent VVC techniques. The over- and under-voltage scenarios have been
created to test various existing and proposed VVC methods. When com-
pared to the existing distributed VVC techniques, it has been observed Declaration of competing interest
that the proposed technique requires the least amount of RPS from
the substation. Additionally, the proposed technique provides results The authors declare that they have no known competing finan-
closer to the optimal RPS method, which shows its effectiveness. The cial interests or personal relationships that could have appeared to
simulation results have also been validated using RTDS. The proposed influence the work reported in this paper.
VVC technique can be easily implemented on LVDNs to mitigate voltage
violations. The proposed approach assumes that the over- and under-
voltage conditions cannot occur simultaneously and hence, presents Data availability
the results for these conditions separately. Therefore, the proposed
approach may be extended by suitable modification to deal with over- Data will be made available on request.

10
K.R. Babu and D.K. Khatod Electric Power Systems Research 228 (2024) 110015

References [15] K.E. Antoniadou-Plytaria, I.N. Kouveliotis-Lysikatos, P.S. Georgilakis, N.D.

Hatziargyriou, Distributed and decentralized voltage control of smart distribution
[1] H. Sun, et al., Review of challenges and research opportunities for voltage control networks: Models, methods, and future research, IEEE Trans. Smart Grid 8 (6)
in smart grids, IEEE Trans. Power Syst. 34 (4) (2019) 2790–2801. (2017) 2999–3008.
[2] S. Pujan Jaiswal, V. Shrivastava, D. Palwalia, Opportunities and challenges of [16] K. Jhala, B. Natarajan, A. Pahwa, Probabilistic voltage sensitivity analysis
PV technology in power system, Mater. Today 34 (2021) 593–597. (PVSA)—A novel approach to quantify impact of active consumers, IEEE Trans.
[3] T. Alquthami, R. Sreerama Kumar, Shaikh, A Mitigation of voltage rise due to Power Syst. 33 (3) (2018) 2518–2527.
high solar PV penetration in Saudi distribution network, Electr. Eng. 102 (2020) [17] G. Valverde, T. Zufferey, S. Karagiannopoulos, G. Hug, Estimation of voltage sen-
881–890. sitivities to power injections using smart meter data, in: 2018 IEEE International
[4] R. Tonkoski, D. Turcotte, T.H.M. EL-Fouly, Impact of high PV penetration on Energy Conference (ENERGYCON), 2018, pp. 1–6.
voltage profiles in residential neighborhoods, IEEE Trans. Sustain. Energy 3 (3) [18] P. Jahangiri, D.C. Aliprantis, Distributed volt/VAr control by PV inverters, IEEE
(2012) 518–527. Trans. Power Syst. 28 (3) (2013) 3429–3439.
[5] American national standard for electric power systems and equipment-voltage [19] A. Safayet, P. Fajri, I. Husain, Reactive power management for overvoltage
ratings (60 hz), 2020, https://webstore.ansi.org/standards/nema/ansic842020. prevention at high PV penetration in a low-voltage distribution system, IEEE
[6] N. Mahmud, A. Zahedi, Review of control strategies for voltage regulation of Trans. Ind. Appl. 53 (6) (2017) 5786–5794.
the smart distribution network with high penetration of renewable distributed [20] B.A. Robbins, C.N. Hadjicostis, A.D. Domínguez-García, A two-stage distributed
generation, Renew. Sustain. Energy Rev. 64 (2016) 582–595. architecture for voltage control in power distribution systems, IEEE Trans. Power
[7] M. Ahmed, R. Bhattarai, S.J. Hossain, S. Abdelrazek, S. Kamalasadan, Co- Syst. 28 (2) (2013) 1470–1482.
ordinated voltage control strategy for voltage regulators and voltage source [21] R.A. Jabr, Robust volt/VAr control with photovoltaics, IEEE Trans. Power Syst.
converters integrated distribution system, IEEE Trans. Ind. Appl. 55 (4) (2019) 34 (3) (2019) 2401–2408.
4235–4246. [22] M.G. Kashani, M. Mobarrez, S. Bhattacharya, Smart inverter volt-watt control
[8] T.-T. Ku, C.-H. Lin, C.-S. Chen, C.-T. Hsu, Coordination of transformer on-load design in high PV-penetrated distribution systems, IEEE Trans. Ind. Appl. 55 (2)
tap changer and PV smart inverters for voltage control of distribution feeders, (2019) 1147–1156.
IEEE Trans. Ind. Appl. 55 (1) (2019) 256–264. [23] S. Yoshizawa, Y. Yanagiya, H. Ishii, Y. Hayashi, T. Matsuura, H. Hamada,
[9] Recommendations for updating the technical requirements for Inverters in K. Mori, Voltage-sensitivity-based volt-VAR-watt settings of smart inverters for
Distributed Energy Resources, 2014, https://www.cpuc.ca.gov/-/media/cpuc- mitigating voltage rise in distribution systems, IEEE Open Access J. Power Energy
website/divisions/energy-division/documents/rule21/smart-inverter-working- 8 (2021) 584–595.
group/siwgworkingdocinrecord.pdf. [24] J.-H. Teng, A direct approach for distribution system load flow solutions, IEEE
[10] IEEE standard for interconnection and interoperability of distributed energy re- Trans. Power Deliv. 18 (3) (2003) 882–887.
sources with associated electric power systems interfaces, in: IEEE Std 1547-2018 [25] D. Khatod, V. Pant, J. Sharma, A novel approach for sensitivity calculations in the
(Revision of IEEE Std 1547-2003), 2018, pp. 1–138. radial distribution system, IEEE Trans. Power Deliv. 21 (4) (2006) 2048–2057.
[11] E. Demirok, P.C. González, K.H.B. Frederiksen, D. Sera, P. Rodriguez, R. [26] Y. Long, D.S. Kirschen, Bi-level volt/VAR optimization in distribution networks
Teodorescu, Local reactive power control methods for overvoltage prevention of with smart PV inverters, IEEE Trans. Power Syst. 37 (5) (2022) 3604–3613.
distributed solar inverters in low-voltage grids, IEEE J. Photovolt. 1 (2) (2011) [27] K.R. Babu, D.K. Khatod, Improved volt/var control technique for over- voltage
174–182. mitigation, in: 2022 Second International Conference on Power, Control and
[12] A.Y. Fard, M.B. Shadmand, Multitimescale three-tiered voltage control framework Computing Technologies (ICPC2T), 2022, pp. 1–6.
for dispersed smart inverters at the grid edge, IEEE Trans. Ind. Appl. 57 (1) [28] C.L. Dewangan, S. Chakrabarti, S.N. Singh, M. Sharma, A fair incentive scheme
(2021) 824–834. for participation of smart inverters in voltage control, IEEE Trans. Ind. Inform.
[13] J. Li, Z. Xu, J. Zhao, C. Zhang, Distributed online voltage control in active 18 (1) (2022) 656–665.
distribution networks considering PV curtailment, IEEE Trans. Ind. Inform. 15 [29] Y. Wang, K.T. Tan, X.Y. Peng, P.L. So, Coordinated control of distributed energy-
(10) (2019) 5519–5530. storage systems for voltage regulation in distribution networks, IEEE Trans.
[14] W. Ma, W. Wang, Z. Chen, R. Hu, A centralized voltage regulation method for Power Deliv. 31 (3) (2016) 1132–1141.
distribution networks containing high penetrations of photovoltaic power, Int. J.
Electr. Power Energy Syst. 129 (2021) 106852.

11