Electric Power Components and Systems

ISSN: 1532-5008 (Print) 1532-5016 (Online) Journal homepage: https://www.tandfonline.com/loi/uemp20

Integrated Volt-VAr Optimization with Distributed

Energy Sources to Minimize Substation Energy in
Distribution System

Saran Satsangi & Ganesh Balu Kumbhar

To cite this article: Saran Satsangi & Ganesh Balu Kumbhar (2019): Integrated Volt-VAr
Optimization with Distributed Energy Sources to Minimize Substation Energy in Distribution System,
Electric Power Components and Systems, DOI: 10.1080/15325008.2018.1511004

To link to this article: https://doi.org/10.1080/15325008.2018.1511004

Published online: 20 Feb 2019.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=uemp20
Electric Power Components and Systems, 0(0): 1–18, 2019
Copyright # Taylor & Francis Group, LLC
ISSN: 1532-5008 print / 1532-5016 online
DOI: 10.1080/15325008.2018.1511004

Integrated Volt-VAr Optimization with Distributed

Energy Sources to Minimize Substation Energy in
Distribution System
Saran Satsangi and Ganesh Balu Kumbhar
Department of Electrical Engineering Indian Institute of Technology Roorkee, Roorkee, Uttarakhand, India

CONTENTS
Abstract—Volt-VAr optimization (VVO) is the tool which
1. Introduction improves the efficiency of the distribution system by coordinating
the voltage and reactive power. Due to voltage-dependence of
2. Problem Formulation
loads, both real and reactive power will be affected by variation in
3. Results and Discussion voltage. Consequently, rather than minimizing active energy of the
4. Conclusion system, the focus of this article to reduce the daily apparent
References energy of the substation using VVO. In this, on load tap changers
(OLTC), and voltage regulators (VRs) are used for voltage
optimization, whereas capacitor banks (CBs) inject reactive power
(VAr) into the system provided operational and system constraints
satisfied. This study focuses on minimization of substation
apparent energy (kVAh) by the optimal scheduling of OLTC, VRs,
and CBs. The photo-voltaic (PV) and wind turbine (WT)
generators are also assumed to be connected in the distribution
system. Moreover, this study is extended to consider the effect of
harmonics in the load current for associated loads. All the
simulations are performed on modified IEEE 33-bus and IEEE
123-bus radial distribution systems. More savings in terms of
substation kVAh are accomplished when DGs are connected into
the system. Reductions up to 20% in substation kVAh and 37% in
energy losses have been achieved.

Keywords: Volt/VAr optimization, conservation voltage reduction, energy
savings, substation energy, total harmonic distortion, voltage control, load
models, Volt-VAr control
Received 21 November 2017; accepted 29 July 2018
Address correspondence to Saran Satsangi, Department of Electrical
Engineering Indian Institute of Technology Roorkee, Uttarakhand 247
667, India. E-mail: anujsat2006@gmail.com
Color versions of one or more of the figures in the article can be found
online at www.tandfonline.com/uemp.
1. INTRODUCTION
Volt-VAr optimization (VVO) is used to reduce peak
power demand, energy losses, energy consumption, etc.
[1]. Conservation voltage reduction (CVR) is utilized for
optimal control of voltages. This concept work on the fact
that the majority of distribution loads is a constant imped-
ance type. Thus, by reducing the system voltage to its min-
imum permissible range (ANSI C84.1 Range-A) [2], the
energy or power consumption can be reduced. Worldwide,
many utilities such as American Electric Power (AEP) [3],
BC Hydro (BCH) [4, 5], Avista [6], Siemens [7], Taiwan
Power Company (TPC) [8] have utilized VVO to decrease
power consumption and energy losses.
In [9], network losses were minimized using VVO
incorporating residential (RE), commercial (CO), and
industrial (IN) customers in the distribution system. In

1
2 Electric Power Components and Systems, Vol. 0 (2019), No. 0

NOMENCLATURE
Pli;t =Qli;t Real/Reactive power consumption of ith node at kip ; kiq Voltage exponents for real, reactive power at
time t ith node
Gstd Standard solar irradiance, i.e., 1000w=m2 Ppv
m;t Real power output of PV generator connected at
kmTemp Temperature coefficient of PV-plate connected at node m over time t
mth node Ppv
m;max Maximum power capacity of PV connected at
c
Tm;t Cell temperature of mth PV generator over time t node m
r
T Reference cell temperature Gt Solar irradiance over time t
Pwt
n;t Real power output of WT generator connected at ks Steinmetz coefficient
node n over time t Tapcap:max
k Maximum tap setting of kth capacitor
Pwt
n;max Maximum power capacity of WT generator con- THDVi,t Total harmonic distortion of voltages of ith bus for
nected at node n time t
vci =vco =vr Cut-in/cut-off/rated speed of wind turbine THDVi,maxt Maximum allowable voltage THD of ith bus for
wt Actual speed of wind turbine over time t time t
Npv =Nwt Total number of PV/WT generators Ir RMS fundamental current under rated frequency and
cos /pv m Power factor of mth PV generator rated load conditions
Pni;t =Qni;t Nominal real/reactive power of ith node at time t Nh Highest harmonic order
cos /wt n Power factor of nth WT generator Pcu
loss Transformer Cu loss considering harmonics
rij Resistance of line between node i and j Prcu
loss Transformer Cu loss at rated condition
xij Reactance of line between node i and j Pec
loss Transformer winding eddy current loss consider-
zij Complex impedance of line between node i and j ing harmonics
Vimin =Vimax Minimum/maximum allowable voltage at ith node Prec
loss Transformer winding eddy current loss at
Vttap Substation transformer secondary side voltage at rated condition
t hour PrLL Total load loss at rated condition
Taptrt Transformer tap position at t hour PNLL No-load loss of transformer
DVstep Voltage change per tap position Pos
loss Other stray current loss considering harmonics
Tap
MS /MScap Maximum allowable transformer/capacitors tap oper- Pros
loss Other stray current loss at rated condition
ations in a day Vr RMS fundamental voltage under rated frequency and
Dqcap
k kVAr change per capacitor tap step at kth node rated load conditions
Vi;t Actual complex voltage of ith node at time t Vi;t ðfh Þ Actual complex voltage of ith node at time t and
Tapcapk;t Tap position of kth capacitor at time t harmonic frequency fh
Qrk;t Rated kVAr of kth capacitor at time t
Vi;n Nominal voltage of ith node

[10], the impact of DGs on Volt/VAr Control (VVC) has
been studied first time on peak-loading conditions, where
genetic algorithm (GA) is used as an optimization tool. It
is observed that loss reduction is higher when DGs are
connected to the system. VVC in presence of induction
machine based DGs has been considered in [11].
In [12–16], the VVO/CVR has been developed as a
multi-objective problem and solved with a non-sorting GA
(NSGA) and its variant. The percentage reduction in peak-
load was up to 7% and energy loss reduction was up to
33.87%. A real-time VVO/CVR method using a multi-
agent system has been proposed in [17]. In this improved
GA has been utilized for scheduling taps of the transformer
and capacitors so that apparent energy losses can be mini-
mized. In [14, 15, 18, 19], the harmonics are considered
into the loads by incorporating nonlinear loads and there-
fore solved to minimize total harmonic distortion (THD)
along with the energy losses. The problem of [14] (in
which photo-voltaic [PV] generators along with the react-
ive power capability are considered) has been modified by
considering wind turbine (WT) generator into the network
and results were reported in [15]. Here, up to 5% reduction
in THD and 28.75% reduction in peak hour losses have
been reported. In [20], Chinese distribution network has
been considered with limited switching operations of VVC
devices to minimize daily energy losses.
In [21], branch-and-bound and sequential constrained
programing has been applied to solve problems of VVC,
where reactive kVAr are supplied by DGs while minimiz-
ing injected reactive power. A VVC problem considering
renewable energy resources (RES), uncertainties of loads,
and output power of RES has been solved with modified
Teaching Learning Algorithm and results in-terms of mini-
mization of energy losses, energy demand, voltage devi-
ation are reported in [22]. MPC-based VVO have been
presented in [23], where 77% reduction in power losses
has been reported. A VVO/CVR framework was presented
in [24] to reduce the load demand, and up to 4.8% reduc-
tions in demand and 40% in losses have been reported.
Manbachi et al. [17, 25–28] focused on real-time simu-
lation platform of VVO in order to achieve various objec-
tives such as minimization of energy losses, the cost
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
Distribution System 3

incurred due to switching operations of control devices, the harmonics are considered on the load-side along with
cost of CVR, etc. In the literature, the VVO problem has the distinctive voltage-dependency of active and reactive
been formulated to achieve a different set of objectives loads of various customers. The main contributions of the
with various load models such as constant power (PQ), article are as follow:
constant current (I), constant impedance (Z), exponential, 1. Due to different voltage dependence of active-reactive
and ZIP load models. The PQ model is widely used, where powers of the loads and losses, minimization of
loss minimization was the prime objective. But, in actual apparent energy of the distribution system is considered
practice, the active and reactive components of a load acts instead of minimization of active energy or losses.
differently with variation in the terminal voltage [29, 30]. 2. DG is considered to increase the savings in-terms
reduction of substation apparent energy.
Some of the studies have considered this fact and utilized
3. The harmonics are considered on the load-side while
exponential [16, 23, 31, 32], and ZIP model [17, 24–28,
maintaining the voltage-dependence of different loads.
32] in the Volt/VAr optimization. In order to obtain differ- 4. The impact of harmonics with and without local
ent VVO objectives, control devices used so far are on generation has been studied in order to minimize
load tap changer (OLTC), voltage regulators (VR), and substation apparent energy.
capacitor banks (CBs). Moreover, some advanced devices,
Therefore, this study focuses on the VVO model with and
such as Static VAR Compensator [33, 34], DGs (PV, WT,
without DGs in order to minimize daily substation apparent
and other generators) [14, 15, 35–37], RES [22, 38–41],
energy. This study is conducted into two scenarios, i.e., with
and electric vehicles [25] have also been included in
and without DGs. Here, nondispatchable DGs are considered in
the VVO.
Most of the studies in the literature have considered the study. Each scenario is having three cases, such as (1) base
minimization of real power (kW) and energy losses (kWh) case, (2) voltage control case, and (3) Volt/VAr control case.
The rest of article is organized as follows. Section 2
for optimal coordination of VVC devices. However, mini-
mization of apparent energy losses (kVAh) have also been describes problem formulation by describing load model,
studied in [17, 27]. But as reported in [3], 5% energy DG model, loss model, objective function, and constraints.
losses have occurred in the distribution system. This has Results and discussion along with modified systems and
been focused in many studies [6, 7, 12, 16, 24, 31, 42–44], case studies are described in Section 3, whereas complete
where the prime objective was the minimization of active study is concluded in Section 4.
energy demand.
But the characteristics of active and reactive loads are 2. PROBLEM FORMULATION
distinct and varies with terminal voltage. Each class of cus-
2.1. Load Models
tomer has different voltage sensitivities [30]. Therefore, con-
trol of voltage and VAr flow in the distribution system In this study, voltage-dependent load models are considered.
affects both real and reactive powers in a different way. These are of two types [46], i.e., exponential load model
Moreover, the presence of a shunt capacitor provides a lead- (ELM) and polynomial load model (ZIP). Here, ELM is
ing current, and hence, both active and reactive power losses used for analysis and can be defined by (1) and (2) [16],
will also be affected. This gives motivation to the authors of

k p


n
Vi;t


i
this article to focus on minimization of apparent energy Pi;t ¼ Pi;t
n

l
(1)
Vi
demand or apparent energy losses instead of active energy

k q


i
n
Vi;t

demand and losses. There is no literature on minimization Qi;t ¼ Qi;t
n

l
(2)
of apparent energy demand. Apart from energy demand, the Vi
impact of distributed generation (DG) and CVR has been The values of kp and kq are different for RE, CO, and
studied [45] for the perspective of voltage profile on peak- IN loads and these are shown in Table 1 [30].
loading condition. However, its impact on energy savings
under the VVO framework needs to be studied.
Load Type kp kq
Due to complexity, the harmonics have received less
attention in VVO problem formulation. References [14, 15, RE 1.04 4.19
18, 19] deal with the harmonics considering nonlinear CO 1.50 3.15
IN 0.18 6.00
loads. However, voltage-dependence of real and reactive
components of the loads has been neglected. In this article, TABLE 1. kp and kq in ELM model.
4 Electric Power Components and Systems, Vol. 0 (2019), No. 0

2.2. Distributed Generator Models 2.3. Capacitor kVAr Model

In this study, PV, and WT generators are assumed to be The capacitive VAr at rated voltage (Vr) can be written as
connected in the distribution system. 2
Qrcap ¼ ðV r Þ xC (8)
2.2.1. PV Generator. Power generated by PV generator is
The VAr of a capacitor for voltage (let V) other than
dependent on two factors, i.e., solar irradiance (G) and cell
rated is
temperature (Tc). These two parameters will vary over  2
time. The mathematical model for output power can be V
Qcap ¼ Qrcap (9)
defined by (3). Vr

pv Gt h Temp  c r
i In this study, the load is continuously varying through-
Pm;t ¼ Pm;max
pv
1 þ km Tm;t
T (3)
Gstd out a day. In this manner, the voltage at the nodes where
capacitors are associated will likewise change, and subse-
The total power available is directly fed to the distribu-
quently, a voltage-dependent capacitor model is utilized as
tion system. a part of the analysis. Accordingly, total capacitive VAr
injected into the distribution system at time t can be math-
2.2.2. WT Generator. The output of WT generator is a
ematically written as
function of wind speed. Wind speed varies throughout the

2
X
NCB


day. Therefore, the output power of WT generator will also r
Vk;t

Qt ¼
CBs
Qk;t
r
(10)
vary. This variation can be represented by following equa- k¼1
Vk
tion [22].

n;t ¼ Pn;max
Pwt wt
2.4. Loss Models
The power from a substation to different loads will flow
8 through distribution lines, transformers, and VR (if any).
>
> r1 ðwt
vci Þ; vci  wt  v1
>
> Mathematical models of losses in these components are
< r1 ðv1
vci Þ þ r2 ðwt
v1 Þ; v1  wt  v2
 r1 ðv1
vci Þ þ r2 ðv2
v1 Þ þ r3 ðwt
v2 Þ; v2  wt  vr discussed in the following sections.
>
>
>
> 1 vr  wt  vco
: 2.4.1. Line Losses. The power losses in i – j section of a
0 otherwise
(4) line can be mathematically expressed as
Ploss ¼ jIij j2  rij (11)
where vci, v1, v2, vr, vco, r1, r2, and r3 having typical values
and can be obtained from [22]. and current in i – j section can be evaluated using follow-
Total real power generated by PV and WT generators ing relationship
can be mathematically written as Vi
Vj
Iij ¼ ¼ ðVi
Vj Þ  yij (12)
zij
X
Npv
pv
X
Nwt
PDG
t ¼ Pm;t þ wt
Pn;t (5) Here, zij ¼ rij þ jxij and yij ¼ z1ij . Therefore, Eq. (11) can
m¼1 n¼1
be re-arranged according to current evaluated in (12).
Based on power factor of each generator, total reactive Based on this, the total real and reactive power losses in
the lines at time t can be defined by (13) and (14).
power generation can also be formulated using Eq. (6)
1X Nb XNb
 2
X
Npv X
Nwt Ploss;t ¼ rij  jVi;t
Vj;t j  jyij j (13)
¼ pv
tan /pv 2 i¼1 j¼1
m þ n;t tan /n
wt
QDG
t Pm;t Pwt (6)
1X Nb X
m¼1 n¼1 Nb
 2
Qloss;t ¼ xij  jVi;t
Vj;t j  jyij j (14)
Therefore, total complex power of all DGs at time t can 2 i¼1 j¼1
be defined by (7)
Similarly, complex power losses at time t can also be
StDG ¼ PDG
t þ jQDG
t (7) written as
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
Distribution System 5

1X Nb XNb
 2 Hence, total complex load-losses of a transformer at
Sloss;t ¼ ðrij þ jxij Þ  jVi;t
Vj;t j  jyij j (15) time t can be written as
2 i¼1 j¼1
SLL;t ¼ PLL;t þ jQLL;t (21)

2.4.2. Transformer Losses. Transformer losses are com- Here, SNL;t and SLL;t are complex no-load and load
posed of no-load losses (PNLL) and load losses (PLL). In a losses of the transformer over time t.
practical distribution system, the load varies with time.
Therefore, transformer secondary side voltage will also 2.5. Inclusion of Harmonics
vary to meet the requirement of a change in voltage at the
2.5.1. Line Parameters and Admittance Matrix in Presence of
load end. This will lead to a change in PNLL as well as
Harmonics. At higher frequencies, the distribution net-
PLL. If the frequency and other parameters of a transformer
work is modeled by a harmonic frequency admittance
are assumed to be constant then PNLL at the applied voltage
matrix (Y ðfh Þ ) and harmonic current sources (Iðfh Þ ). In
(V) can be defined by (16).
this, Y ðfh Þ is evaluated by modifying the conventional
 ks  2
V r V
admittance according to the harmonic order. The distri-
PNLL ðV Þ ¼ Phr þ Pe (16)
Vr Vr bution lines or cables are considered as lumped parame-
ters connected in a nominal-p model. The longitudinal
As ks is varying from 1.4 to 1.8, approximate function ðhÞ ðhÞ
(ylongitudinal ) and shunt (yshunt ) parameters of the overhead
can be written as
lines or underground cables are incorporated into off-
  ðhÞ ðhÞ
 r  V 2 diagonal (yij ) and diagonal (yii ) elements of the modi-
PNLL  Ph þ Pe
r
fied matrix Y ðhÞ , respectively:
Vr
ð Þ 1
yijh ¼ (22)
No-load losses at time t can be re-written as rij þ j2pfh Lij
 2
Vt The harmonic frequency (fh) can also be written as h  f.
PNLL;t ¼ PNLL r
r
(17)
V Where, f is fundamental frequency and h is harmonic
order. Hence, Eq. (22) can be rewritten as follows:
Thus, complex no-load losses of a transformer at time t
can be represented as ð Þ 1 1
yijh ¼ ¼ (23)
rij þ jhð2pfLij Þ rij þ jhxij
SNLL;t ¼ PNLL;t þ jQNLL;t (18)
ð Þ
yiih ¼ j2pfh Cii (24)
QNLL;t are reactive no-load losses of the transformer,
and this is very less as compared to PNLL;t and can be Apart from this, the capacitors connected into the net-
neglected in calculations. work are also shunt element. Therefore, shunt CBs con-
Solution obtained from Eq. (17) represents approximate nected at node k can also be modeled similar to (24) and it
no-load losses. The percentage error is varying from
2.5 can be written as in (25)
to 2.5% for voltage variation in the range of 0.9 to 1.1 p.u. ð Þ
h
ycap;k ¼ hycap;k (25)
This error is negligible if it compared with overall distribu-
tion system losses. Here, ycap;k is evaluated based on mathematical Eq. (8).
Load losses (PLL) of a transformer at time t can be eval-
uated using (19) 2.5.2. Load Modeling in Presence of Harmonics. When
higher order of harmonics frequencies (fh) are consid-
PLL;t ¼ jIht;t j2 Rht þ jIlt;t j2 Rlt (19) ered in the distribution network, then loads are modeled
Here Iht;t ; Ilt;t are current of HT and LT sides of the as admittance to take the effect of harmonics into the
transformer at time t, whereas Rht and Rlt are resistances of loads. The load connected at node i can be modeled
HT and LT side and each is 50% of specified resistance. using (26)
Similarly, reactive load losses of a transformer can also Pil ðV ðfh ÞÞ Qli ðV ðfh ÞÞ
ð Þ
be written as
h
yload;i ¼ þj (26)
QLL;t ¼ jIht;t j2 Xht þ jIlt;t j2 Xlt (20) jVi ðf1 Þj2 hjVi ðf1 Þj2

In (20), Xht and Xlt are reactances of HT and LT sides Here, Pli ðV ðfh ÞÞ and Qli ðV ðfh ÞÞ are determined using
of the transformer and each is 50% of specified reactance. Eqs. (27) and (28).
6 Electric Power Components and Systems, Vol. 0 (2019), No. 0

p

Vi ðfh Þ
ki 2.5.4. Transformer Losses in Presence of Harmonics. In
Pli ðV ðfh ÞÞ ¼ Pin


(27)
Vin ðf1 Þ
the harmonic environment, load, no-load, and stray losses


q in a transformer get affected differently. According to

Vi ðfh Þ
ki
Qli ðV ðfh ÞÞ ¼ Qni

n
(28) IEEE standard C57.110-2008 [47], total load losses can be
Vi ðf1 Þ

divided into three components, i.e., copper loss (Pcu
loss ),
The nonlinear loads are modeled as current sources that winding eddy current loss (Pec
loss ) and other stray losses
inject harmonic current into the system. The fundamental (Pos
loss ). However, the percentage of Pec
loss and Pos
loss
and the h order harmonic current of the nonlinear load con- is 33 and 67% [48], respectively, in total stray load losses
nected at bus i are modeled as [14, 18, 19] of the transformer (for oil-immersed distribution trans-
 
Pi þ jQi  former). Once all these losses at rated conditions are
Ii ðf1 Þ ¼ (29) known, losses in the harmonic environment are determined
Vi ðf1 Þ
using (37)–(39).
Ii ðfh Þ ¼ Ii ðf1 Þ  C ðhÞ (30)
r
Pcu
loss X
Nh
The voltage at harmonic frequency V ðfh Þ can be deter- Pcu
loss;t ¼ It2 ðfh Þ (37)
mined using following equation ðI r Þ2 h¼1

½Y ðfh Þ½V ðfh Þ ¼ ½I ðfh Þ (31) Prec
loss X

Nh
Pec
loss;t ¼ h2  It2 ðfh Þ (38)
ðI r Þ2 h¼1
2.5.3. Calculation of Line Losses and Power Consumed by r
Pos
loss X
Nh

Loads. The line-losses will also be affected when consider- Pos
loss;t ¼ h0:8  It2 ðfh Þ (39)
ð I r Þ2 h¼1
ing harmonics in the system. Therefore, total real power loss
occurred at time instant t can be mathematically written as Therefore, total load losses at time t can be written as
1X Nb XNb XNh  2 PLL;t ¼ Pcu
loss;t þ Pec
loss;t þ Pos
loss (40)
ð Þ
Ploss;t ¼ rij  jVi;t ðfh Þ
Vj;t ðfh Þjjyijh j (32)
2 i¼1 j¼1 h¼1
Similarly, reactive power losses can also be written as The effect of harmonics can also be considered for no-
load losses of the transformer by utilizing mathematical
1X Nb XNb XNh  2
ð Þ expression (17). This equation will be modified in har-
Qloss;t ¼ hxij  jVi;t ðfh Þ
Vj;t ðfh Þj  jyijh j
2 i¼1 j¼1 h¼1 monic environment and can be written as
(33) XNh  2
Vt ðfh Þ
PNLL;t ¼ PNLL
r
(41)
However, complex power losses can be determined h¼1
Vr
using (34)

1X Nb XNb XNh  2
ðhÞ
Sloss;t ¼ ðrij þjhxij Þ  jVi;t ðfh Þ
Vj;t ðfh Þjjyij j 2.6. Objective Function
2 i¼1 j¼1 h¼1
Volt-VAr optimization affects real as well as reactive
(34)
power consumption. Thus, the significance of Volt/VAr
Apart from losses, the power consumed by loads gets optimization will be reflected, when minimization of appar-
affected due to voltage dependency of the loads. The load ent energy (kVAh) is considered. This objective function
power of ith node at time t considering harmonics effect can be mathematically defined by (42).
can be expressed by following mathematical relations
!
X 24
XNb h i


k p

XNh

i min f ¼ SEkvah ¼
Pi;t þ jQi;t
l l
n
Vi;t ðfh Þ


Pi;t ¼
l
Pi;t
n (35)
Vi ðf1 Þ

t¼1 i¼1
h¼1


k q !

X

i


Nh
n
Vi;t ðfh Þ


Qi;t ¼
l
Qi;t
n (36)
StDGs þ jQt CBs
þ ½Sloss;t  þ ½SNLL;t þ SLL;t 
(42)
h¼1
Vi ðf1 Þ

Similarly, the mathematical expressions representing l

Here, Pi;t ; Qli;t ; StDGs ; QCBs
t ; Sloss;t ; SNLL;t , and SLL;t are eval-
output of DGs (7) and capacitor kVAr (10) will also uated using (1), (2), (7), (10), (15), (18), and (21), respect-
be modified. ively. In the case of harmonics considerations, the objective
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
Distribution System 7

function will be modified using mathematical formulation

presented in previous subsection (from (22) to (41)).

2.7. System and Operational Constraints

There are various operational and system constraints, which
are defined by (43)–(51)

2.7.1. Bus Voltage Magnitude. The voltages at each bus

should be constrained between Vimin and Vimax
Vimin  jVi;t j  Vimax (43)
Here V min ¼ 0:95 p.u. and V max ¼ 1:05 p.u. at every FIGURE 1. Summer load curves for various customers in
node at each time t. However, in case of harmonics consid- Gujarat State of India.
eration, the r.m.s. voltage can be evaluated using (44)
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u Nh 2.7.4. Capacitor Bank Tap Settings. The reactive power
uX
Vi;t ¼ t
rms
jVi;t ðfh Þj2 (44) supplied by the capacitor can be defined as
h¼1 Qrk;t ¼ Tapcap cap
k;t 䉭qk (50)
and this is also constrained between the same min–max
Here Tapcap cap:max
k;t ¼ f0; 1; ::::; Tapk g and 䉭qcap
k ¼ 50
bounds. This constrained can be defined using (45)
kVAr. Switching operation limit for CB can be defined as
Vimin  Vi;trms  Vimax (45)
X
24
jTapcap cap cap
k;t
Tapk;t
1 j  MSk (51)
t¼2
2.7.2. Total Harmonic Distortion of Voltage. The corre-
sponding THD of voltage at time t can also be evaluated
using following mathematical relation
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3. RESULTS AND DISCUSSION
PNh 2
h6¼1 jVi;t ðfh Þj In this study, modified IEEE-33 bus radial distribution sys-
THDVi;t ¼  100% (46)
Vi;t ðf1 Þ tem [50] and IEEE-123 node test system [51] are consid-
ered. The daily load-shapes are adopted from a distribution
There are some limitations on allowable voltage THD
utility based in Gujarat (India). The per unit (p.u.) load
as per IEEE Standard 519-2014 [49]. Hence, THDVi;t
curves are shown in Figure 1 for three types of customers,
should satisfy this limitation. This can be defined using
i.e., RE, CO, and IN consumers [52, 53]. These per unit
(47)
load curves are used in both the systems. In this study,
THDVi;t  THDmax
Vi;t (47) Matlab COM Interfacing of OpenDSS [54] has been uti-
lized for hourly power-flow solutions and an optimization
2.7.3. Transformer Tap Settings. The tap voltage can be method is required for optimal scheduling of VVC devices.
defined as Solar irradiance, temperature and wind speed data are
adopted from Suthari, Gujarat (India) and shown in
䉭Vstep
Vttap ¼ 1 þ Taptrt (48) Figure 2.
100
Here Taptrt f
16; ::::;
1; 0; 1; ::::; þ16g and 䉭Vstep ¼
0:3125% and transformer taps operation limit can be 3.1. Selection of Optimization Technique
defined as The objective function considered in the problem formula-
X
24 tion is a nonlinear function along with nonlinear con-
jTaptrt
Taptrt
1 j  MS Tap (49) straints. The control variables are transformer taps, VR
t¼2 taps, and capacitor taps. All these variables are of integer
where MS Tap ¼ 50 is assumed for time-series simula- type. Therefore, the fitness function cannot be differenti-
tion purpose. ated with respect to control variables. Hence, heuristic
8 Electric Power Components and Systems, Vol. 0 (2019), No. 0

with this type of problems. However, other methods like

TLA, SA, TS, SFLA, HBMO, BFA, ACO, and BSO have
also been used. Due to popularity of GA and PSO, they
are selected for the optimization. It was observed that PSO
converges faster than GA, but sometimes it converges to
sub-optimal solution. In other words, GA is a time-con-
suming optimization technique but gives near-global solu-
tions in each studied case. Therefore, results presented here
are obtained with GA. However, a comparison of both GA
and PSO is also presented in sub-Section 3.5.

3.2. Case Study

FIGURE 2. Input parameters for PV and WT generator.
(a) Solar irradiance and temperature, (b) Wind speed.
techniques are more suitable for this type of problem.
There are various heuristic methods available in the litera-
ture and can be seen in Table 2. Here, it can be noticed
that GA and PSO methods are widely accepted to deal
The complete study has been carried out with two scen-
arios, i.e., Without DGs, and With DGs. Each scenario
consists of three cases. Therefore, total six cases are con-
sidered and are listed in Table 3. Case 1 is the base case.
In this, the hourly power-flow solution in presence of loads
has been obtained without violating voltage limits
(0.95–1.05 p.u.). Case 2 is a voltage control case, where
optimal tap positions of OLTC/VRs has been obtained in
order to minimize substation apparent energy. Whereas,
Case 3 deals with the optimal coordination of taps of
OLTC/VRs and CBs so that energy consumption can be
minimized and this case named as Volt-VAr control case.

Method Reference

Genetic Algorithm (GA) and its variants [10, 12–19, 25, 26, 28, 31–34, 55]
Particle Swarm Optimization (PSO) [27, 36, 37, 56, 57]
Teaching Learning Algorithm (TLA) [22]
Simulated Annealing (SA) [58]
Tabu Search (TS) [59]
Shuffled Frog Leaping Algorithm (SFLA) [38, 39]
Honey Bee Mating Optimization (HBMO) [35]
Bacterial Foraging Algorithm (BFA) [40]
Ant Colony Optimization (ACO) [60]
Bee Swarm Optimization (BSO) [41]
TABLE 2. Literature review on heuristic optimization techniques used for Volt/VAr optimization.

System IEEE 33-Bus System IEEE-123 Bus System

Scenarios VVC Devices OLTC/VRs CBs DGs OLTC VR CBs DGs Optimal Solution

Without DGs Case 1        

Case 2        
Case 3        
With DGs Case 4        
Case 5        
Case 6        
TABLE 3. Different scenario along with its cases and assumptions.
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
Distribution System 9

of voltage-dependent, which are of constant power type

in original system. Moreover, RE, CO, and IN loads
(represented by voltage-dependent model) connected at
various locations in the distribution system are shown in
Figure 4.

3.3.1. Variation in Tap Position of Transformer and

FIGURE 3. Current harmonics spectrum of various load
connected in the systems.
Case 4, Case 5, and Case 6 are similar to Case 1, Case 2,
and Case 3, respectively. In these cases, DGs are connected
in the distribution system. The VVC devices considered in
various cases are shown in Table 3. All the results of all
cases have been compared with Case 1 (base case).
All these cases have also been studied when current har-
monics are considered in the loads. To add the harmonics effect
to load current, the harmonic spectrum of RE, CO, and IN
loads has been measured using a power quality analyzer. In this
study, the load connected to a node of the network is varying
according to given load-shape (Figure 1), where load category
remains the same. Therefore, it is assumed that the measured
current harmonics spectrum will not vary in hourly power flow
calculations. The harmonics spectrum of various loads (RE,
CO, and IN) used in the analysis is shown in Figure 3.
Capacitor Banks. Transformer tap positions obtained in all
six cases are displayed in Figure 5. In this, Figure 5a repre-
sents an hourly variation of taps when DGs are not partici-
pating. In Case 1, variation can be seen from þ4 to þ9
and these positions are higher among other cases. When
optimization is considered (Case 2), these have been settled
down on lower positions. This is because of voltage-
dependencies of the loads. As soon as CBs are active in
the distribution system along with transformer taps (Case
3), these positions again settled down on the lower position
in order to optimize the system voltage from the sub-
station node.
Impacts on transformer tap position in presence of DGs
can be seen in Figure 5b. Tap positions in Case 4 have set-
tled down on lower positions (
1 to þ6) as compared to
Case 1 as local power has been injected by the DGs. And
in Case 5, when optimization is considered, these tap posi-
tions again reduced (
8 to þ4). In Case 6, when all the
VVC devices are considered to be active in the network,

3.3. Modified IEEE-33 Bus Radial Distribution System

The original IEEE-33 bus system does not have any
transformer, VR, and CBs. Therefore, the system is suit-
ably modified to consider these VVC devices. An OLTC
transformer is assumed to be connected at the substation
(start bus). This modified system also contains variable
CBs and DGs at different locations as given in Table 4.
Each capacitor can inject reactive power in 50 kVAR/ FIGURE 4. IEEE-33 bus radial distribution system with
step. The loads of the modified distribution network are residential, commercial, and industrial loads.

Test system Locations of CBs (capacity in kVAR) DG location (each of 250 kW)

IEEE 33-Bus 12 (250), 21 (100), 24 (500), 26 (900), 30 (400) WT: 14, 17 PV: 31, 32
IEEE 123-Bus 83a (200), 83b (200), 83c (200), 88a (50), 90b (50), 92c (50) WT: 115, 117, PV: 95, 122
TABLE 4. Location of CBs and DGs in the distribution networks.
10 Electric Power Components and Systems, Vol. 0 (2019), No. 0
FIGURE 5. Transformer tap positions for IEEE 33-bus
distribution system. (a) Without DGs (no harmonics), (b)
With DGs (no harmonics), (c) Without DGs (with harmon-
ics), (d) With DGs (with harmonics).
these taps are at its lowest position (
11 to þ2) among all
the cases. Considering harmonics into the system influences
planning of taps of transformer and can be effortlessly seen
in Figure 5c,d. For this situation, a totally different sched-
uling is obtained from Case 1 to Case 6.
Capacitors banks are included into the system only in
Case 3 and Case 6. Hourly tap positions of all the CBs are
displayed in Figure 6. In Case 6, switching operations of
all the CBs are less (except Cap#26) as compared with
Case 3. It is also observed that reactive VArs (through
CBs) provided here is 29.15 MVArh, whereas, it was
more (31.98 MVArh) when DGs were not connected to
the system. This is because of voltage profile, and capaci-
tors are under-performing due to its voltage-dependent
behavior. The CBs taps are likewise influenced when har-
monics are there in the system. This effect can also be
seen on the tap scheduling of capacitors (shown with dot-
ted pink lines).
3.3.2. Effect on Distribution System Voltage. Hourly vari-
ation in the system voltage can be seen in Figure 7. In
Case 1, the voltage is only controlled with the help of
OLTC, therefore higher voltages (i.e., > 1.0 p.u.) are
observed on the buses near to substation bus and can be
seen in Figure 7a. Whereas in Case 2 (Figure 7b), voltage
is relatively low because of optimal settings of substation
transformer taps. Here, more reductions in the voltage have
been observed for Case 3 and shown Figure 7c. The over-
all voltage is quite low (almost area is covered with
0.98–0.95 p.u.). This is because of simultaneous control of
CBs and OLTC taps.
Once DGs are connected in the system, there will be
more room for the operator to optimize the tap positions of
the transformer and hence overall voltage in Case 4 is rela-
tively low (Figure 7d) as compared with Case 1. Further
reduction in the system voltage can also be observed in
Figure 7e for optimal scheduling of transformer taps as in
case 5. In case 6, CVR is effectively applied as mostly vol-
tages are near to 0.95 p.u. and can be seen in Figure 7f.
This is because of optimal coordination of taps of trans-
former and CBs in the presence of DGs.
3.3.3. Energy and Volt-Amperes Supplied from Substation.
Minimized substation apparent energy is shown in Table 5
along with energy consumed by loads and losses. Up to
14.48% savings in-terms of substation kVAh has been
achieved when DGs are not connected in the distribution net-
work and harmonics are not considered. Apart from this,
major reductions in losses has also been observed. These sav-
ings are increased when DGs are connected in the system.
By connecting DGs, 5.55% reduction in substation kVAh and
13.01% in losses have been reported. These savings further
improved in subsequent cases (Case 5 and Case 6) and reduc-
tion up to 20.31% in substation kVAh and 37.63% in loss
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
FIGURE 6. Capacitor tap positions of IEEE-33 bus system in case of Volt/VAr Control (with and without harmonics). (a) Case
3, (b) Case 6.
FIGURE 7. Voltage surface plot of IEEE 33-bus distribu-
tion system for different control cases. (a) Case 1, (b) Case
2, (c) Case 3, (d) Case 4, (e) Case 5, (f) Case 6.
kWh have been obtained. The substation kVAh is slightly
increased when considering harmonics in the network.
However, savings (compared to Case 1) are similar. The
Distribution System 11
energy consumed by loads is slightly decreased (because of
the predominance of admittance type of load), however,
losses are drastically increased. Further, reduction in losses is
similar to the situation when no harmonics are there in the
system. Indeed, up to 31% reductions in energy losses have
also been acquired for this case.
Due to voltage dependency of loads, hourly substation
kVAs are also affected and can be seen in Figure 8.
Hourly profile of substation kVAs when DGs are not con-
nected in the system is shown in Figure 8a, where peak
occurs at 1500 hr. There is marginal reduction in kVAs for
Case 2. This effect is more in the lightly loading period
as compared with other time. Moreover, a major reduction
in hourly substation kVAs is observed in Case 3. Here,
reduction in peak-time is higher as compared with other
periods of time and can be seen with the red curve in
Figure 8a.
When local power through DG is available, good reduc-
tion in the substation kVAs is observed (Case 4) during the
lightly loaded period (i.e., from 0100 to 1000 hr) and
shown in Figure 8b. In Case 5, significant reductions on an
entire day are observed and shown with a green curve.
These reductions are highest for Case 6 and the behavior is
exactly same as in Case 3 but with major energy savings.
The substation kVA profile is almost similar when harmon-
ics are there in the system (therefore, not shown). But, the
magnitude is slightly more, which results in increased sub-
station kVAh.
12 Electric Power Components and Systems, Vol. 0 (2019), No. 0

Substation Apparent energy (kVAh) Energy consumed by loads (kWh) Energy losses (kWh)
Cases W/o Har. With Har. W/o Har. With Har. W/o Har. With Har.

Case 1 72494 72950 61165 59434 1702 3600

Case 2 70994 (2.07%) 71112 (2.51%) 60649 (0.84%) 58842 (1.00%) 1677 (1.51%) 3532 (1.89%)
Case 3 61997 (14.48%) 62266 (14.65%) 60519 (1.06%) 58935 (0.84%) 1303 (23.48%) 3185 (11.53%)
Case 4 68470 (5.55%) 69383 (4.89%) 60928 (0.39%) 59353 (0.14%) 1481 (13.01%) 2939 (18.36%)
Case 5 65800 (9.23%) 67500 (7.47%) 60557 (0.99%) 58767 (1.12%) 1374 (19.29%) 2865 (20.42%)
Case 6 57767 (20.31%) 57802 (20.76%) 60432 (1.20%) 58769 (1.12%) 1062 (37.63%) 2481 (31.08%)
TABLE 5. Energy consumption for IEEE 33-Bus system in different cases (% savings w.r.t. Case 1).

FIGURE 8. Hourly substation kVAs of IEEE 33-bus distri-

bution system for different cases. (a) Without DGs, (b)
With DGs.

3.3.4. Peak Shaving. Asides from reduction in substation

energy, peak-hour kVA (SSkVA) is additionally reduced
(Figure 8). The peak shaving is maximum in Case 6.
Percentage savings (when no harmonics in the system) in
terms of load and losses during the peak hour period are
shown in Figure 9a. Here, up to 21% reductions have been
observed in substation kVA. Both real (PLoss) and reactive
power (QLoss) losses having a higher reduction as compared
to other parameters. Reduction up to 41% in PLoss and
44% in QLoss have been achieved. It is also clear that major
savings/reductions have been obtained for Case 3 and Case
6 as compared with other cases, and highest for Case 6.
However, peak hour savings are different in the case of
harmonics considerations and shown in Figure 9b. Here,
the percentage savings in-terms of loads (PLoad and QLoad)
increases as compared to previous case. Unlike previous
FIGURE 9. Reduction of various parameters at peak-hour
for IEEE 33-bus system. (a) Without harmonics, (b)
With harmonics.
operating conditions, the reductions in SSkVA, PLoss, and
Qloss are highest in Case 5.
3.4. IEEE-123 Node Test System
The topology of modified IEEE-123 node feeder is shown
in Figure 10. The loads connected at various locations in
the network are assumed to be voltage-dependent (as in
IEEE-33 bus system), which are of constant power, con-
stant current, and constant impedance type in the original
system. Here, different loads (RE, CO, and IN) are shown
with different color coding. This modified system consists
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
Distribution System 13

FIGURE 10. Modified IEEE-123 bus radial distribution

test system.

of variable CBs and DGs at different locations as shown in

Table 4. The savings obtained from this system are tabu-
lated in Table 6. The trends are similar to IEEE 33-bus
system. The maximum savings are achieved in case of
Volt/VAr control.
Hourly substation kVAs is also following the same trend
as in the IEEE 33-bus system, where less kVAs are sup-
FIGURE 11. Hourly substation kVAs for IEEE-123 bus
plied for voltage control (Case 2 and Case 5) and the min- system in different control cases. (a) Without DGs, (b)
imum of these two were supplied when volt/var control With DGs.
(Case 3 and Case 6) is applied (Figure 11). Daily voltage
profile of both the scenarios is displayed in Figure 12.
Case 3 (i.e., 32%), whereas 23% reduction has been
Because of a large distribution network, these profiles are
obtained for Case 6.
shown with respect to IN, CO, and RE load connected
nodes. Here, effective voltage reduction has been observed
for CO and RE loads. Therefore, it can be stated that these
3.5. Comparison of GA and PSO
are the potential customers where CVR can be effectively
applied. For the scenario, when DGs are connected in the All the studied cases have been solved with both GA and
system, savings in terms of load kWh are almost same PSO. The convergence graphs of GA and PSO are dis-
(Table 6) for Case 5 and Case 6. This is happening played in Figure 14 for Case 3 and Case 6. Here it is
because of voltage profile and can be seen in Figure 12. observed that PSO is converging faster than that of GA,
This distribution network also follows the same trend in but solution obtained using PSO is a suboptimal solution
reduction of parameters at peak hour (except QLoss) as in and can be easily observed in Figure 14. Although, the per-
the IEEE 33-bus system and it can be seen in Figure 13. centage error in both the solution is less still GA is giving
Here, substation kVAs at peak hour is reduced by 16% for a global solution. If one can compromise on global solution
Case 6. Major reduction in QLoss has been observed for then PSO can be a good choice. Apart from this, when

Substation apparent energy (kVAh) Energy consumed by loads (kWh) Energy losses (kWh)
Cases W/o Har. With Har. W/o Har. With Har. W/o Har. With Har.

Case 1 72352 72609 58720 58551 1296 3028

Case 2 69437 (4.03%) 69618 (4.29%) 57546 (2.00%) 56949 (2.73%) 1222 (5.71%) 2990 (1.25%)
Case 3 62530 (13.57%) 63368 (12.72%) 56632 (3.56%) 56658 (3.23%) 1060 (18.22%) 2923 (3.47%)
Case 4 69409 (4.07%) 69665 (4.05%) 58745 (
0.04%) 58561 (
0.01%) 1202 (7.19%) 2931 (3.20%)
Case 5 64705 (10.57%) 64849 (10.68%) 56797 (3.28%) 56546 (3.42%) 1089 (15.92%) 2906 (4.03%)
Case 6 60893 (15.84%) 61194 (15.72%) 56841 (3.20%) 57444 (1.89%) 998 (23.00%) 2797 (7.63%)
TABLE 6. Energy consumption for IEEE123-bus system in different cases (% savings w.r.t. Case 1).
14 Electric Power Components and Systems, Vol. 0 (2019), No. 0
FIGURE 12. Voltage surface plot of various customers connected in IEEE-123 bus system. (a) Without DGs (U-Case 1, M-Case
2, L-Case 3), (b) With DGs (U-Case 4, M-Case 5, L-Case 6).
FIGURE 13. Reduction in various parameters at peak-hour
for IEEE-123 bus system. (a) Without harmonics, (b)
With harmonics.
harmonics are induced in the system, the time taken by
both the techniques is drastically increased and shown in
Table 7, but solution obtained with GA is still better than
PSO. Once the number of control variables increase (as in
IEEE-123 Bus), the time taken by both optimization techni-
ques is increased and the same has also been noticed in
this study. Here, time taken to solve cases of IEEE-123
Bus system is higher than what in IEEE-33 Bus system.
3.6. Impact of Harmonics
The energy savings (as compared to base case) are almost
similar when harmonics are considered in the load currents.
However, substation kVAh are somewhat increased compar-
ing to the case when there are no harmonics. In the case
when harmonics are considered, all loads consists of har-
monic spectrum. Therefore, the r.m.s. load current will be
higher than what in case of no harmonics. Apart from load
currents, transformer losses are likewise increased. This
ultimately increases the losses of the distribution network.
Therefore, energy losses (kWh) are increased. The power
consumed by loads is decreased because of dominating
admittance type of load. The daily profile of THD of volt-
age is shown in Figure 15. Here, it is observed that THD
at any point of time is within allowable limits (i.e.,  5%)
as per IEEE Std. 519-2014. The daily profile of THD is
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
Distribution System 15

FIGURE 14. Convergence graph of GA and PSO for Case 3 and Case 6 (IEEE-33 Bus). (a) Without harmonics, and (b)
With harmonics.

Without harmonics With harmonics

GA PSO GA PSO

IEEE-33 Bus Case 2 299 90 2843 2519

Case 3 270 267 4197 3889
Case 5 651 128 3283 2717
Case 6 394 289 5574 3920
IEEE-123 Bus Case 2 2499 2938 76714 61233
Case 3 3964 1801 77105 63079
Case 5 2544 3749 103092 47575
Case 6 2262 2306 134880 80677
TABLE 7. Time taken by GA and PSO (in seconds) to solve different cases of the studied systems (All computations are done on Intel
Core i7 3.40 GHz processor).

FIGURE 15. Daily total harmonic distortion of voltages in different cases (a) For IEEE-33 Bus System, and (b) For IEEE-123
Bus System.
16 Electric Power Components and Systems, Vol. 0 (2019), No. 0
more in the case of Volt/VAr control (Case 3 and Case 6).
However, the overall THD profile having higher magni-
tudes for IEEE-123 bus system as compared to IEEE-33
bus system. From the perspective of optimization, the prob-
lem formulation is complex and evaluating each component
of the objective function (42) is very time consuming.
Therefore, the time required for solving any of the case in
this situation (see Table 7) is drastically increased as com-
pared with no harmonics in the system.
4. CONCLUSION
In this paper, Volt/VAr control in order to minimize the
substation energy has been solved for two different scen-
arios, i.e., without DGs and with DGs. These two scen-
arios have also been formulated with harmonics in the
load currents. Results obtained have been compared with
base case and voltage control case. It is observed that
when DGs are connected to the system, the operator has
more chances to optimize the voltage profile. Therefore,
more energy savings was observed when DGs are con-
nected. Drastic reductions in substation kVAh and losses
(kWh) were observed in Volt/VAr control case. It is also
observed that voltage reduction is not so effective in volt-
age control case but proper CVR has been applied in
Volt/VAr control case. With the inclusion of harmonics,
the substation kVAh is slightly increased as compared to
no harmonics. However, there is a drastic increase in
losses of the network.
The switching operations of VVC devices are less in the
base case, but as we move towards an optimal solution,
then switching operations increases. Therefore, it can be
stated that more energy savings can be obtained if VVC
devices are allowed to operate frequently as much as
required. But this will reduce the life of VVC devices as
every device has fixed number of operations. Therefore,
there is a tradeoff between energy savings and number of
switching operations.
REFERENCES
[1] S. Borlase, “Smart Grid Technologies,” in Smart Grids
Infrastructure, Technology, and Solutions, S. Borlase, Ed.
Boca Raton, FL: CRC Press, Taylor and Francis Group,
2012, Ch. 3, pp. 61–496.
[2] American National Standard for Electrical Power Systems
and Equipment Voltage Ratings (60Hz), American National
Standard Institute, C84.1-2006 (December 2006).
[3] K. Schneider and T. Weaver, “A method for evaluating
Volt-VAR optimization field demonstrations,” IEEE Trans.
Smart Grid, vol. 5, no. 4, pp. 1696–1703, 2014.
[4] V. Dabic, C. Siew, J. Peralta, and D. Acebedo, “BC
Hydro’s experience on Voltage VAR optimization in distri-
bution system,” in: Transmission and Distribution
Conference and Exposition, 2010, New Orleans, LA: IEEE
PES, 2010, pp. 17.
[5] V. Dabic and D. Atanackovic, “Voltage VAR optimization
real time closed loop deployment - BC Hydro challenges
and opportunities,” in: Power Energy Society General
Meeting, 2015, Denver, CO: IEEE, 2015, pp. 15.
[6] S. Chanda, F. Shariatzadeh, A Srivastava, E. Lee, W. Stone,
and J. Ham, “Implementation of non-intrusive energy saving
estimation for volt/VAr control of smart distribution sys-
tem,” Electric Power Syst. Res., vol. 120, pp. 39–46, 2015.
[7] I. Roytelman, B. Wee, R. Lugtu, T. Kulas, and T. Brossart,
“Pilot project to estimate the centralized volt/VAr control
effectiveness,” IEEE Trans. Power Syst., vol. 13, no. 3, pp.
864–869, 1998.
[8] R.-H. Liang, Y.-K. Chen, and Y.-T. Chen, “Volt/var control
in a distribution system by a fuzzy optimization approach,”
Int. J. Electr. Power Energy Syst., vol. 33, no. 2, pp.
278–287, 2011.
 c, “Multi-objective integration of
[9] A. T. Saric and R. M. Ciri
real-time functions in distribution management system,” Int.
J. Electr. Power Energy Syst., vol. 25, no. 3, pp. 247–256,
2003.
[10] T. Niknam, A. Ranjbar, and A. Shirani, “Impact of distrib-
uted generation on volt/var control in distribution networks,”
in: Power Tech Conference Proceedings, 2003 IEEE
Bologna, Vol. 3, IEEE, pp. 7, 2003.
[11] F. Viawan and D. Karlsson, “Combined local and remote
voltage and reactive power control in the presence of induc-
tion machine distributed generation,” IEEE Trans. Power
Syst., vol. 22, no. 4, pp. 2003–2012, 2007.
[12] B. Milosevic and M. Begovic, “Capacitor placement for
conservative voltage reduction on distribution feeders,”
IEEE Trans. Power Deliv., vol. 19, no. 3, pp. 1360–1367,
2004.
[13] B. A. De Souza and A. M. F. De Almeida, “Multiobjective
optimization and fuzzy logic applied to planning of the volt/
var problem in distributions systems,” IEEE Trans. Power
Syst., vol. 25, no. 3, pp. 1274–1281, 2010.
[14] S. Jashfar and S. Esmaeili, “Volt/var/THD control in distri-
bution networks considering reactive power capability of
solar energy conversion,” Int. J. Electr. Power Energy Syst.,
vol. 60, pp. 221–233, 2014.
[15] S. Jashfar, M. M. Hosseini-Biyouki, and S. Esmaeili, “A
stochastic programming to volt/VAR/total harmonic distor-
tion control in distribution networks including wind
turbines,” Electric Power Compon. Syst., vol. 43, no. 7, pp.
733–746, 2015.
[16] A. Padilha-Feltrin, D. Quijano Rodezno, and J. Sanches
Mantovani, “Volt-VAR multiobjective optimization to peak-
load relief and energy efficiency in distribution networks,”
IEEE Trans. Power Deliv., vol. 30, no. 2, pp. 618–626,
2015.
[17] M. Manbachi et al., “Real-time adaptive VVO/CVR top-
ology using multi-agent system and IEC 61850-based
 Satsangi and Kumbhar: Integrated Volt-VAr Optimization with Distributed Energy Sources to Minimize Substation Energy in
communication protocol,” IEEE Trans. Sustain. Energy, vol.
5, no. 2, pp. 587–597, 2014.
[18] A. Ulinuha, M. A. S. Masoum, and S. M. Islam, “Optimal
scheduling of LTC and shunt capacitors in large distorted
distribution systems using Evolutionary-Based algorithms,”
IEEE Trans. Power Deliv., vol. 23, no. 1, pp. 434–441,
2008.
[19] A. Ulinuha, M. Masoum, and S. Islam, “Hybrid genetic-
fuzzy algorithm for volt/var/total harmonic distortion control
of distribution systems with high penetration of non-linear
loads,” IET Gener. Transm. Distrib., vol. 5, no. 4, pp. 425,
2011.
[20] M. Liu, C. Canizares, and W. Huang, “Reactive power and
voltage control in distribution systems with limited switch-
ing operations,” IEEE Trans. Power Syst., vol. 24, no. 2,
pp. 889–899, 2009.
[21] S. Deshmukh, B. Natarajan, and A. Pahwa, “Voltage/var
control in distribution networks via reactive power injection
through distributed generators,” IEEE Trans. Smart Grid,
vol. 3, no. 3, pp. 1226–1234, 2012.
[22] T. Niknam, M. Zare, and J. Aghaei, “Scenario-based multi-
objective volt/var control in distribution networks including
renewable energy sources,” IEEE Trans. Power Deliv., vol.
27, no. 4, pp. 2004–2019, 2012.
[23] Z. Wang, J. Wang, B. Chen, M. Begovic, and Y. He,
“MPC-Based voltage/var optimization for distribution cir-
cuits with distributed generators and exponential load mod-
els,” IEEE Trans. Smart Grid, vol. 5, no. 5, pp. 2412–2420,
2014.
[24] H. Ahmadi, J. Marti, and H. Dommel, “A framework for
Volt-VAR optimization in distribution systems,” IEEE
Trans. Smart Grid, vol. 6, no. 3, pp. 1473–1483, 2015.
[25] M. Manbachi, H. Farhangi, A. Palizban, and S. Arzanpour,
“A novel Volt-VAR optimization engine for smart distribu-
tion networks utilizing vehicle to grid dispatch,” Int. J.
Electr. Power Energy Syst., vol. 74, pp. 238–251, 2016.
[26] M. Manbachi et al., “Real-time co-simulation platform for
smart grid volt-var optimization using iec 61850,” IEEE
Trans. Ind. Inform., vol. 12, no. 4, pp. 1392–1402, 2016.
[27] M. Manbachi, H. Farhangi, A. Palizban, and S. Arzanpour,
“Smart grid adaptive energy conservation and optimization
engine utilizing particle swarm optimization and
fuzzification,” Appl. Energy, vol. 174, pp. 69–79, 2016.
[28] M. Manbachi et al., “Impact of EV penetration on Volt-
VAR optimization of distribution networks using real-time
co-simulation monitoring platform,” Appl. Energy, vol. 169,
pp. 28–39, 2016.
[29] T. Frantz, T. Gentile, S. Ihara, N. Simons, and M. Waldron,
“Load behavior observed in LILCO and RG E systems,”
IEEE Trans. Power Appar. Syst. PAS, vol. 103, no. 4, pp.
819–831, 1984.
[30] IEEE Task Force, “Bibliography on load models for power
flow and dynamic performance simulation,” IEEE Trans.
Power Syst., vol. 10, no. 1, pp. 523–538, 1995.
[31] S. Satsangi and G. B. Kumbhar, “Analysis of substation
energy using conservation voltage reduction in distribution
system,” in: 2016 International Conference on Electrical
Power and Energy Systems (ICEPES), Maulana Azad
Distribution System 17
National Institute of Technology, Bhopal, India, pp.
188–193, 2016.
[32] S. Satsangi and G. B. Kumbhar, “Effect of load models on
energy loss reduction using volt-var optimization,” in: 2016
National Power Systems Conference (NPSC), Indian
Institute of Technology Bhubaneswar, India, pp. 1–6, 2016.
[33] T. Senjyu, Y. Miyazato, a Yona, N. Urasaki, and T.
Funabashi, “Optimal distribution voltage control and coord-
ination with distributed generation,” IEEE Trans. Power
Deliv., vol. 23, no. 2, pp. 1236–1242, 2008.
[34] Y. Y. Hong and Y. F. Luo, “Optimal VAR control consider-
ing wind farms using probabilistic load-flow and gray-based
genetic algorithms,” IEEE Trans. Power Deliv., vol. 24, no.
3, pp. 1441–1449, 2009.
[35] T. Niknam, “A new HBMO algorithm for multiobjective
daily volt/var control in distribution systems considering dis-
tributed generators,” Appl. Energy, vol. 88, no. 3, pp.
778–788, 2011.
[36] A. Anwar, A. N. Mahmood, J. Taheri, Z. Tari, and A. Y.
Zomaya, “HPC-based intelligent volt/VAr control of unbal-
anced distribution smart grid in the presence of noise,”
IEEE Trans. Smart Grid, vol. 8, no. 3, pp. 1446–1459,
2017.
[37] F. Sayadi, S. Esmaeili, and F. Keynia, “Two-layer volt/var/
total harmonic distortion control in distribution network
based on PVs output and load forecast errors,” IET Gener.
Transm. Distrib., vol. 11, no. 8, pp. 2130–2137, 2017.
[38] A. R. Malekpour and T. Niknam, “A probabilistic multi-
objective daily volt/var control at distribution networks
including renewable energy sources,” Energy, vol. 36, no. 5,
pp. 3477–3488, 2011.
[39] A. R. Malekpour, S. Tabatabaei, and T. Niknam,
“Probabilistic approach to multi-objective volt/var control of
distribution system considering hybrid fuel cell and wind
energy sources using improved shuffled frog leaping algo-
rithm,” Renew. Energy, vol. 39, no. 1, pp. 228–240, 2012.
[40] M. Zare and T. Niknam, “A new multi-objective for envir-
onmental and economic management of volt/var control
considering renewable energy resources,” Energy, vol. 55,
pp. 236–252, 2013.
[41] M. Zare, T. Niknam, R. Azizipanah-Abarghooee, and B.
Amiri, “Multi-objective probabilistic reactive power and
voltage control with wind site correlations,” Energy, vol. 66,
pp. 810–822, 2014.
[42] F.-C. Lu and Y. Hsu, “Reactive power/voltage control in a
distribution substation using dynamic programming,” IEE
Proc. Gener. Transm. Distrib., vol. 142, no. 6, pp. 639–645,
1995.
[43] A. Borghetti, “Using mixed integer programming for the
volt/var optimization in distribution feeders,” Electric Power
Syst. Res., vol. 98, pp. 39–50, 2013.
[44] A. Borghetti, F. Napolitano, and C. A. Nucci, “Volt/var
optimization of unbalanced distribution feeders via mixed
integer linear programming,” Int. J. Electr. Power Energy
Syst., vol. 72, pp. 40–47, 2015.
[45] A. Bokhari et al., “Combined effect of cvr and dg penetra-
tion in the voltage profile of low-voltage secondary
18 Electric Power Components and Systems, Vol. 0 (2019), No. 0
distribution networks,” IEEE Trans. Power Deliv., vol. 31,
no. 1, pp. 286–293, 2016.
[46] S. Satsangi and G. B. Kumbhar, “Review on volt/VAr opti-
mization and control in electric distribution system,” in:
2016 IEEE 1st International Conference on Power
Electronics, Intelligent Control and Energy Systems
(ICPEICES 2016), DTU, New Delhi, India, 2016.
[47] IEEE Recommended Practice for Establishing Liquid-Filled
and Dry-Type Power and Distribution Transformer
Capability When Supplying Nonsinusoidal Load Currents,
IEEE Std C57.110-2008 (Revision of IEEE Std C57.110-
1998), pp. 1–52, 2008.
[48] IEEE Approved Draft Standard for General Requirements
for Liquid-Immersed Distribution, Power, and Regulating
Transformers, IEEE PC57.12.00/D2.3, October 2015
(Revision of IEEE Std C57.12.00-2010), pp. 1–71, 2015.
[49] IEEE Recommended Practice and Requirements for
Harmonic Control in Electric Power Systems, IEEE Std
519-2014 (Revision of IEEE Std 519-1992), pp. 1–29, 2014.
[50] M. E. Baran, and F. F. Wu, “Network reconfiguration in dis-
tribution systems for loss reduction and load balancing,”
IEEE Trans. Power Deliv., vol. 4, no. 2, pp. 1401–1407,
1989.
[51] IEEE-123 Bus Feeder, IEEE PES Distribution System
Analysis Subcommittee’s Distribution Test Feeder Working
Group.
[52] A. Garg, J. Maheshwari, and J. Upadhyay, Load Research
for Residential and Commercial Establishments in Gujarat,
2010. [Online]. Available: http://www.beeindia.gov.in/sites/
default/files/ctools/
Load%20Research%20Report%20for%20Gujarat-1024.pdf,
Accessed: June 04, 2017.
[53] Day Ahead Demand - (SLDC), Gujarat, SLDC Gujarat.
[Online]. Available: http://www.sldcguj.com/Operation/
Dayaheaddemand.asp
[54] R. C. Dugan and T. E. McDermott, “An open source plat-
form for collaborating on smart grid research,” in 2011
IEEE Power and Energy Society General Meeting, Detroit,
MI, USA, pp. 1–7, 2011.
[55] Z. Hu, X. Wang, H. Chen, and G. Taylor, “Volt/var control
in distribution systems using a time-interval based
approach,” IEE Proc. Gener. Transm. Distrib., vol. 150, no.
5, pp. 548–554, 2003.
[56] T. Niknam, B. B. Firouzi, and A. Ostadi, “A new fuzzy
adaptive particle swarm optimization for daily volt/var con-
trol in distribution networks considering distributed gener-
ators,” Appl. Energy, vol. 87, no. 6, pp. 1919–1928, 2010.
[57] L. Mokgonyana, J. Zhang, L. Zhang, and X. Xia,
“Coordinated two-stage volt/var management in distribution
networks,” Electric Power Syst. Res., vol. 141, pp. 157–164,
2016.
[58] R-h Liang and Y-s Wang, “Fuzzy-based reactive
power and voltage control in a distribution system,”
IEEE Trans. Power Deliv., vol. 18, no. 2, pp. 610–618,
2003.
[59] D. F. Pires, A. Martins, and C. Antunes, “A multiobjective
model for VAR planning in radial distribution networks
based on tabu search,” IEEE Trans. Power Syst., vol. 20,
no. 2, pp. 1089–1094, 2005.
[60] T. Niknam, “A new approach based on ant colony optimiza-
tion for daily volt/var control in distribution networks con-
sidering distributed generators,” Energy Convers. Manag.,
vol. 49, no. 12, pp. 3417–3424, 2008.