Applied Energy 124 (2014) 62–72

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

An optimal investment planning framework for multiple distributed

generation units in industrial distribution systems
Duong Quoc Hung a, N. Mithulananthan a,⇑, R.C. Bansal b
a
School of Information Technology and Electrical Engineering, The University of Queensland, Brisbane, Qld 4072, Australia
b
Department of Electrical, Electronic and Computer Engineering, University of Pretoria, South Africa

h i g h l i g h t s

 DG allocation for minimizing energy loss and enhancing voltage stability.

 Expressions to find the optimal power factor of DG with commercial standard size.
 A methodology for DG planning to recover investment for DG owners.
 Impact of technical and environmental benefits on DG investment decisions.
 Benefit-cost analysis to specify the optimal location, size and number of DG units.

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents new analytical expressions to efficiently capture the optimal power factor of each
Received 4 January 2014 Distributed Generation (DG) unit for reducing energy losses and enhancing voltage stability over a given
Received in revised form 17 February 2014 planning horizon. These expressions are based on the derivation of a multi-objective index (IMO), which
Accepted 1 March 2014
is formulated as a combination of active and reactive power loss indices. The decision for the optimal
Available online 22 March 2014
location, size and number of DG units is then obtained through a benefit–cost analysis. Here, the total
benefit includes energy sales and additional benefits, namely energy loss reduction, network upgrade
Keywords:
deferral and emission reduction. The total cost is a sum of capital, operation and maintenance costs.
Distributed generation
Emission reduction
The methodology was applied to a 69-bus industrial distribution system. The results showed that the
Loss reduction additional benefits are imperative. Inclusion of these in the analysis would yield faster DG investment
Network upgrade deferral recovery.
Optimal power factor Ó 2014 Elsevier Ltd. All rights reserved.
Voltage stability

1. Introduction improvement, voltage stability enhancement, network upgrade

deferral and reliability while supplying energy sales as a primary
For the reasons of energy security and economical and environ- purpose [3–13]. In addition, DG units can participate into the com-
mental benefits, there has been increased interest in the usage of petitive market to provide ancillary services such as spinning re-
Distributed Generation (DG) worldwide. DG can be defined as serve, voltage regulation, reactive power support and frequency
small-scale generating units located close to the loads that are control [14–16]. However, inappropriate allocation and operations
being served [1]. It is possible to classify DG technologies into of these resources may lead to high losses, voltage rise and system
two broad categories: non-renewable and renewable energy re- instability as a result of reverse power flow [17,18].
sources [2]. The former comprises reciprocating engines, combus- DG planning by considering various technical issues has been
tion gas turbines, micro-turbines, fuel cells, and micro-Combined discussed considerably over the last decade. Several approaches
Heat and Power (CHP) plants. The latter includes biomass, wind, have been developed to place and size DG units for loss reduction
solar photovoltaic (PV) and ocean-based power plants. From the due to its impact on the utilities’ revenue. Typical examples are
utilities’ perspective, DG units can bring multiple technical benefits analytical methods [19–21], numerical approaches [22–24] and a
to distribution systems such as loss reduction, voltage profile wide range of heuristic algorithms such as Genetic Algorithm
(GA) [25], Particle Swarm Optimization (PSO) [26] and artificial
⇑ Corresponding author. Tel.: +61 7 3365 4194; fax: +61 7 3365 4999. bee colony algorithm [27]. Moreover, in recent years, due to shar-
E-mail address: [email protected] (N. Mithulananthan). ply increased loads and the demand for higher system security, DG

http://dx.doi.org/10.1016/j.apenergy.2014.03.005
0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 63

Nomenclature

AEy actual annual emission of the system with DG units NPV net present value
(Ton CO2) OMy annual operation, maintenance and fuel costs ($/year)
ALossy actual annual energy loss of the system with DG units pfDGi power factor of DG unit at bus i
(MW h) PDGi, QDGi, SDGi active, reactive and apparent power sizes of DG
AVSM average voltage stability margin of the system unit, respectively at bus i
B present value benefit over a planning horizon ($) PDi, QDi active and reactive power of load, respectively at bus i
BCR benefit and cost ratio Pi, Qi net active and reactive power injections, respectively at
C present value cost over a planning horizon ($) bus i
CDG capital cost of DG ($/kW) PLDG, QLDG total system active and reactive power losses with DG
CEy cost of each ton of generated CO2 ($/Ton CO2) unit (MW), respectively
CLossy loss value ($/MW h) PL, QL total system active and reactive power losses without
d discount rate DG unit (MW), respectively
EIy emission incentive ($/year) Ry annual energy sales ($/year)
LF load factor or average load level of the system over a TEy annual emission target level of the system without DG
planning horizon (Ton CO2)
LFbase load factor or average load level of the system over the TLossy annual energy loss target level of the system without
base year DG (MW h)
ILP, ILQ active and reactive power loss indices, respectively VSM voltage stability margin
IMO multi-objective index |Vi|, di voltage magnitude and angle, respectively at bus i
IRR internal rate of return Zij ijth element of impedance matrix (Zij = rij + jxij)
LIy loss incentive ($/year) d growth rate of demand a year
N number of buses kmax maximum loading
ND network deferral benefit ($/kW) DAVSM an increase in the average voltage stability margin
Ny planning horizon (years)

allocation for voltage stability at the distribution system level has and power factor of DG for minimizing power losses [26]. In [21],
attracted the interest of some recent research efforts. For instance, three different analytical approaches were presented to determine
DG units are located and sized using different methods: iterative the location, size and power factor of renewable DG (i.e., biomass,
techniques based on Continuous Power Flow (CPF) [8] and a hybrid wind and solar PV) for minimizing energy losses. A dual index-
of model analysis and CPF [28], power stability index-based meth- based analytical approach was proposed to find the location, size
od [29], numerical approach [30,31], simulated annealing algo- and power factor of DG for minimizing power loss and improving
rithm [32] and PSO [33–35]. However, the cost–benefit analyses loadability [38]. Finally, a self-correction algorithm was proposed
of DG planning have been ignored in the works presented above. to specify the size and power factor of PV and battery energy
Furthermore, a few recent studies have indicated that network storage units for minimizing energy losses and enhancing voltage
investment deferral and emission reduction are other attractive stability [39]. The above review shows that a few works have dis-
options for DG planning. For instance, an optimal power flow- cussed the optimal power factor of DG units. However, the size of
based method was successfully developed to place and size DG DG units obtained from the existing studies may not match the
units for postponing network upgrade [4]. An immune-GA method standard sizes available in the market. Furthermore, a comprehen-
was presented for placing and sizing DG units to reduce the total sive benefit–cost study on multiple DG allocation with optimal
emission while minimizing the total cost as a sum of electricity power factor while considering the issues of energy loss and volt-
purchased from the grid, installation, operation and network rein- age stability has not been reported in the literature.
forcement costs [36]. An improved honey bee mating optimization This paper aims at expending the previous preliminary study in
approach was also proposed for locating and sizing DG units to re- [40] where analytical expressions were developed based on a sin-
duce the total emission while minimizing the capital, fuel, opera- gle objective to identify the optimal power factor of each DG unit
tion and maintenance costs, voltage deviation and energy loss for minimizing energy losses. In this paper, analytical expressions
[5]. In addition, an planning framework was also developed for are presented based on a multi-objective index (IMO) to determine
PV integration by reducing the installation, operation and mainte- the optimal power factor for reducing energy losses and enhancing
nance costs and the energy imported from the grid [37]. It is obvi- voltage stability in industrial distribution systems over a given
ous from the above review that numerous methodologies have planning horizon. Here, new analytical expressions are developed
developed for DG allocation in distribution systems with different to efficiently determine the optimal power factor of each DG unit
applications. However, most of them have assumed that DG units with a commercial standard size to ease the computational burden.
operate at a pre-defined power factor. Depending on the nature In this study, it is assumed that DG units are owned and operated
of loads served, DG operation at optimum power factor may have by distribution utilities. To make the work comprehensive, in addi-
positive impacts on system losses, voltage stability, and system tion to the analytical expressions presented to specify the optimal
capacity release. power factor, a benefit–cost analysis is carried out in the paper to
Recently, a few studies have presented DG allocation while con- determine the optimal location, size and number of DG units. The
sidering the optimal power factor, to which the active and reactive total benefit as a sum of energy sales, energy loss reduction, net-
power injections of each DG are optimized simultaneously. For in- work upgrade deferral and emission reduction is compared to
stance, a rule of thumb for DG operation was developed for mini- the total cost including capital, operation and maintenance costs.
mizing power losses [20]. For this rule, it is recommended that The rest of the paper is structured as follows: Section 2 de-
the power factor of DG should be equal to the system load factor. scribes the modeling of loads and DG units. Section 3 presents ac-
A PSO-based method was presented to identify the location, size tive and reactive power loss indices and a combination of both
64 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72

known as the IMO. Methods of assessing the energy loss and volt- voltage regulation, but they are not fast enough to compensate
age stability, and analyzing the benefit and cost with DG units are for transient events [43,44]. As fast response devices, synchronous
also presented in this section. Section 4 presents analytical expres- machines-based DG technologies (e.g., gas turbine engine) are al-
sions to capture the optimal power factor of DG units and a com- lowed to control reactive power for voltage regulation.
putational procedure for DG allocation. Section 5 presents and As reported in [45], gas turbine engines use a turbine spun by
discusses a case study on a 69-bus test distribution system. Finally, the gases of combustion to rotate an electric generator. For DG
the key contributions and conclusions of the work are summarized application, gas turbine engines have smaller sizes than any other
in Section 6. source of rotating power and provide higher reliability than recip-
rocating engines. They also have superior response to load varia-
2. Load and DG modeling tions and excellent steady state frequency regulation when
compared to steam turbines or reciprocating engines. Moreover,
2.1. Load modeling they can operate on a wide range of fuels such as natural gas, waste
gas, methane, propane and diesel. In addition, gas turbine engines
The system considered under the study is assumed to follow the require lower maintenance and produce lower emissions than
industrial load duration curve as shown in Fig. 1, including four reciprocating engines.
discrete load bands (maximum, normal, medium and minimum) For the above reasons, gas turbine engine-based DG units are
that change as the load grows over a planning horizon. The load adopted in this study. As synchronous machines, the DG units
factor or average load level of the system over the base year, LFbase are capable of delivering active power and injecting or absorbing
can be defined as the ratio of the area under load curve to the total reactive power. It is assumed that the units offer a constant energy
duration (four load bands: 8760 h). That means LF base ¼ supply at rated capacity. Given PDGi and QDGi values which corre-
P8760 p:u: loadðtÞ spond to the active and reactive power of DG unit injected at bus
t¼1 8760
, where p.u. load (t) is the demand in p.u. at period t.
i, the power factor of DG unit at bus i (pfDGi) can be expressed as
Assuming the growth rate of demand a year (d), the load factor
follows [21]:
or average load level of the system over a given planning horizon
(Ny), LF can be calculated as: PDGi
pfDGi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3Þ
1 X8X
Ny
760
p:u: loadðtÞ PDGi þ Q 2DGi
2

LF ¼  ð1 þ dÞy ð1Þ
Ny y¼1 t¼1 8760

The dependence of loads on the voltage and time at period t can 3. Problem formulation
be expressed as [41]:
3.1. Active power loss index
n n
Pi ðtÞ ¼ Poi ðtÞ  V i p ðtÞ; Q i ðtÞ ¼ Q oi ðtÞ  V i q ðtÞ ð2Þ
The total active power loss (PL) in a system with N buses can be
where Pi and Qi are respectively the active and reactive power injec-
expressed as [46]:
tions at bus i, Poi and Qoi are respectively the active and reactive
loads at bus i at nominal voltage; Vi is the voltage at bus i; X
N X
N

np = 0.18 and nq = 6.0 are respectively the active and reactive indus- PL ¼ ½aij ðPi Pj þ Q i Q j Þ þ bij ðQ i Pj  P i Q j Þ ð4Þ
i¼1 j¼1
trial load voltage exponents [41].
where
2.2. DG modeling r ij rij
aij ¼ cosðdi  dj Þ; bij ¼ sinðdi  dj Þ
V iV j V iV j
Given the fact that most of the DG units are normally designed
to operate at unity power factor under the standard IEEE 1547 [42]. V i \di is complex voltage at bus i; rij + jxij = Zij is the ijth element of
Consequently, inadequacy of reactive power support for voltage impedance matrix [Zbus]; Pi and Pj are respectively active power
regulation may exist in distribution systems, given a high DG pen- injections at buses i and j; and Qi and Qj are respectively the reactive
etration. It is likely that shortage of reactive power support may be power injections at buses i and j.
an immediate concern at the distribution system level in the fu- The active and reactive power injected at bus i where a DG unit
ture. Conventional devices such as switchable capacitors, voltage is installed can be expressed as [20]:
regulators and tap changers are actually employed for automatic
Pi ¼ P DGi  PDi ð5Þ

Q i ¼ Q DGi  Q Di ð6Þ
p.u.
where PDGi and QDGi are respectively the active and reactive power
1.0 Maximum
injections from DG at bus i; PDi and QDi are respectively the active
load
and reactive power of a load at bus i.
0.8 Substituting Eqs. (5) and (6) into Eq. (4), we obtain the total ac-
Normal tive power loss with DG unit (PLDG) as follows:
work hour
load X
N X
N

0.6 PLDG ¼ aij ððPDGi  PDi ÞPj þ ðQ DGi  Q Di ÞQ j Þ
Medium i¼1 j¼1
load 
0.4 þbij ððQ DGi  Q Di ÞPj  ðP DGi  PDi ÞQ j Þ ð7Þ
Mimimum
load Finally, the active power loss index (ILP) can be defined as Eq.
0 80 2880 5820 8760 (7) divided by Eq. (4) as follows [38]:
Hour
PLDG
ILP ¼ ð8Þ
Fig. 1. Load duration curve. PL
 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 65

3.2. Reactive power loss index 3.4. Technical constraints

The total reactive power loss (QL) in a system with N buses can The maximum DG penetration, which is calculated as the total
be expressed as [46]: capacity of DG units, is limited to less than or equal to a sum of the
total system demand and the total system loss.
X
N X
N
QL ¼ ½cij ðPi Pj þ Q i Q j Þ þ nij ðQ i Pj  Pi Q j Þ ð9Þ X
N X
N X
N X
N
i¼1 j¼1 PDGi 6 PDi þ PL ; Q DGi 6 Q Di þ Q L ð13Þ
i¼2 i¼2 i¼2 i¼2
where
where PL and QL are respectively calculated using Eqs. (4) and (9).
x x The voltage at each bus is maintained close to nominal.
cij ¼ ij cosðdi  dj Þ; nij ¼ ij sinðdi  dj Þ
V iV j V iV j
V min
i 6 V i 6 V max
i ð14Þ
Substituting Eqs. (5) and (6) into Eq. (9), we obtain the total
where and V min V max
are respectively the lower and upper bounds of
i i
reactive power loss with DG unit (QLDG) as follows:
the voltage at bus i, Vi = 1 p.u. (substation
 voltage).

X N h
N X The thermal capacity of circuit n Smax n is less than the maxi-
Q LDG ¼ cij ððPDGi  PDi ÞPj þ ðQ DGi  Q Di ÞQ j Þ mum apparent power transfer (Sn).
i¼1 j¼1
 jSn j 6 Smax
n ð15Þ
þnij ððQ DGi  Q Di ÞPj  ðPDGi  PDi ÞQ j Þ ð10Þ
Finally, the reactive power loss index (ILQ) can be defined as Eq.
3.5. Energy loss and voltage stability
(10) divided by Eq. (9) as follows [38]:

Q LDG 3.5.1. Energy loss

ILQ ¼ ð11Þ The total active power loss of a system with DG unit at each per-
QL
iod t, Ploss can be obtained from Eq. (7). Here, the total period dura-
tion of a year is 8760 h, which are calculated as a sum of all the
3.3. Multi-objective index total period durations of all the load levels throughout a year as
shown in Fig. 1. The total annual energy loss in a distribution sys-
P
The multi-objective index (IMO) is a combination of the ILP and tem can be calculated as ALossy ¼ 8760 t¼1 P loss ðtÞ. Hence, the total en-
ILQ impact indices, which are respectively related to energy loss ergy loss over a given planning horizon (Ny), ELoss can be expressed
and voltage stability by giving a weight to each impact index. This as:
IMO index can be expressed by Eq. (12) which is subject to the con- X X
Ny 8760
straint on the pre-specified apparent power of DG capacity (SDGi) ELoss ¼ Ploss ðy; tÞ  Dt ð16Þ
through a relationship between PDGi and QDGi. That means: y¼1 t¼1

IMO ¼ r1 ILP þ r2 ILQ ð12Þ where Dt is 1 h, which is the time duration of period t.

subject to 3.5.2. Voltage stability

The static voltage stability can be analyzed using the relation-
S2DGi ¼ P2DGi þ Q 2DGi ship between the receiving power (P) and the voltage (V) at a cer-
P2 tain bus in a distribution power system, as illustrated in Fig. 2. This
where i ri ¼ 1:0 ^ ri 2 ½0; 1:0. This can be performed as all im-
curve is known as a P–V curve and obtained using the CPF tech-
pact indices are normalized with values between zero and one
nique [50]. The critical point (CP) or voltage collapse point in the
[25]. When DG unit is not connected to the system (i.e., base case
curve represents the maximum loading (kmax) of the system. The
system), the IMO is highest at one.
voltage stability margin (VSM) is defined as the distance from an
The weights are intended to give the corresponding importance
operating point to the critical point. As shown in Fig. 2, the scaling
to each impact index for DG connection and depend on the re-
factor of the load demand at a certain operating point (k) varies
quired analysis (e.g., planning and operation) [25,39,47–49]. Deter-
from zero to kmax. When DG unit is properly injected in the system,
mining the appropriate weights will also rely on the experience of
the loss reduces. Accordingly, the V1 and CP1 enhance to V2 and CP2,
engineers and the concerns of distribution utilities. DG integration
respectively. Hence, the maximum loadability increases from kmax1
in distribution networks has a significant impact on the energy loss
and voltage stability. Currently, the energy loss is one of the major
concerns at the distribution system level due to its impact on the
Without DG unit With DG unit
utilities’ profit, while the voltage stability is less important than
V2
the energy loss. Hence, the weight for the energy loss should be
V1
higher than that for the voltage stability. In future, if the impor-
Voltage (p.u.)

tance of voltage stability is increased due to a rise in load demands

and system security concerns, the weights can be adjusted based
on the priority. Considering the current concerns mentioned above VSM2
and referring to previous papers [25,39,47–49], this study assumes VSM1
that the active power loss related to energy loss receives a signifi-
cant weight of 0.7, leaving the reactive power loss related to volt- CP1 CP2
age stability at a weight of 0.3.
The lowest IMO implies the best DG allocation for energy loss
reduction and voltage stability enhancement. The objective func- λmax1 λmax2
Loading parameter (λ)
tion defined by the IMO in Eq. (12) is subject to technical con-
straints described below. Fig. 2. DG impact on maximum loadability and voltage stability margin.
66 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72

to kmax2 as defined by (17) [28,31,38,39] and the voltage stability power resources and other devices at the operation stage. How-
margin subsequently improves from VSM1 to VSM2. ever, redundancy of reactive power can lead to reverse power flow,
thereby leading to high losses, voltage rise, system instability, etc.
PD ¼ kPo ; Q D ¼ kQ o ð17Þ
Consequently, it becomes necessary to study the optimal power
where Poi and Qoi correspond to the initial active and reactive power factor of DG units to which the active and reactive power injections
demands, respectively. of each DG are optimized simultaneously [20].

3.6. Benefit and cost analysis 4.1. Optimal power factor

3.6.1. Utility’s benefit In practice, the choice of the best DG capacity may follow com-
The present value benefit (B) in $ given to a utility to encourage mercial standard sizes available in the market or be limited by en-
DG connection over a planning horizon from owning and sitting its ergy resource availability. Given such a pre-specified DG capacity,
own DG units can be expressed as follows: the DG power factor can be optimally calculated by adjusting the
active and reactive power sizes at which the IMO as defined by
X
Ny
Ry þ LIy þ EIy X
N
B¼ þ ND P DGi ð18Þ Eq. (12) can reach a minimum level. Using the Lagrange multiplier
y
y¼1 ð1 þ dÞ i¼2 method, the constrained problem defined by Eq. (12) can be math-
ematically converted into an unconstrained one as follows:
where all annual values are discounted at the rate d; Ry is annual  
energy sales ($/year) in year; ND is the network deferral benefit LðPDGi ; Q DGi ; ki Þ ¼ r1 ILP þ r2 ILQ þ ki S2DGi  P 2DGi  Q 2DGi ð21Þ
($/kW); PDGi is the total DG capacity connected at bus i (kW); and
Ny is the planning horizon (years). The loss incentive LIy ($/year) where ki is the Lagrangian multiplier.
can be written as [3,4]: Substituting Eqs. (8) and (11) into Eq. (21), we obtain:
r1 r2  
LIy ¼ CLossy ðTLossy  ALossy Þ L¼ PLDG þ Q LDG þ ki S2DGi  P2DGi  Q 2DGi ð22Þ
PL QL
where CLossy is the loss value ($/MW h), ALossy is the actual annual
energy loss of the system with DG units (MW h), and TLossy is the The necessary conditions for the optimization problem, given
target level of the annual energy loss of the system without DG unit by Eq. (22), state that the derivatives with respect to control vari-
(MW h). When DG units are integrated into the grid for primary en- ables PDGi, QDGi and ki become zero.
ergy supply purposes, the environmental benefit as a result of @L r1 @PLDG r2 Q LDG
reducing the usage of fossil fuel energy resources could be obtained. ¼ þ  2ki PDGi ¼ 0 ð23Þ
@PDGi PL @PDGi Q L @PDGi
The emission incentive EIy ($/year) including the emission produced
by the electricity purchased from the grid and DG units can be for- @L r1 @PLDG r2 Q LDG
mulated as [36]: ¼ þ  2ki Q DGi ¼ 0 ð24Þ
@Q DGi PL @Q DGi Q L @Q DGi
EIy ¼ CEy ðTEy  AEy Þ
@L
where CEy is the cost of each ton of generated CO2 ($/Ton CO2); AEy ¼ P2DGi þ Q 2DGi  S2DGi ¼ 0 ð25Þ
@ki
is the actual annual emission of a system with DG units (Ton CO2);
TEy is the target level of the annual emission of the system without The derivative of Eqs. (7) and (10) with respect to PDGi and QDGi
DG unit (Ton CO2). are given as:

@PLDG XN
3.6.2. Utility’s cost ¼ 2 ½aij Pj  bij Q j  ¼ 2aii Pi þ 2Ai ð26Þ
@PDGi j¼1
The present value cost (C) in $ incurred by a distribution utility
over a planning horizon can be expressed as [3,4]:
@Q LDG XN
XNy
OMy XN ¼ 2 ½cij Pj  nij Q j  ¼ 2cii Pi þ 2C i ð27Þ
C¼ y þ C DG PDGi ð19Þ @PDGi j¼1
y¼1 ð1 þ dÞ i¼2

where OMy is the annual operation, maintenance and fuel costs @PLDG XN
¼ 2 ½aij Q j þ bij Pj  ¼ 2aii Q i þ 2Bi ð28Þ
($/year) in year y; CDG is the capital cost of DG ($/kW). @Q DGi j¼1

3.6.3. Benefit–cost ratio analysis XN

@Q LDG
The benefit–cost ratio (BCR) can be expressed as follows: ¼ 2 ½cij Q j þ nij Pj  ¼ 2cii Q i þ 2Di ð29Þ
@Q DGi j¼1
B
BCR ¼ ð20Þ
C where
where the B and C are calculated using Eqs. (18) and (19), respec- X
N X
N
tively. The decision for the optimal location, size and number of Ai ¼ ðaij Pj  bij Q j Þ; Bi ¼ ðaij Q j þ bij P j Þ
DG units is obtained when the BCR as given by Eq. (20) is highest. j¼1 j¼1
j–i j–i
4. Proposed methodology
X
N X
N
The reactive power of DG units can be utilized for loss reduc- Ci ¼ ðcij Pj  nij Q j Þ; Di ¼ ðcij Q j þ nij P j Þ
tion, voltage profile and stability enhancement, and network j¼1 j¼1
investment deferral. It could be highlighted that lack of attention
j–i j–i
to reactive power support at the DG planning stage would poten-
tially result in an increase in investment costs used to add reactive Substituting Eqs. (26) and (27) into Eq. (23), we obtain:
 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 67

2r1 2r 2 Step 1: Set the apparent power of DG unit (SDGi) and the maxi-
½aii Pi þ Ai  þ ½c Pi þ C i   2ki PDGi ¼ 0 ð30Þ
PL Q L ii mum number of buses to connect DG units.
Step 2: Run load flow for the system without DG unit at the
Substituting Eq. (5) into Eq. (30), we obtain:
average load level over the planning horizon (LF) using Eq. (1).
PDi Y i  rP1LAi  rQ2 CL i Step 3: Find the optimal power factor of DG unit for each bus
PDGi ¼ ð31Þ using Eq. (3). Place this DG unit at each bus and find the IMO
Y i  ki
for each case using Eq. (12).
where Step 4: Locate the optimal bus for DG at which the IMO is min-
imum with the corresponding optimal size and power factor at
r1 aii r2 cii that bus.
Yi ¼ þ
PL QL Step 5: Run multiyear multi-period load flow with the DG size
obtained in Step 4 over the planning horizon. Calculate the
Similarly, substituting Eqs. (28) and (29) into Eq. (24), we
energy loss and its corresponding BCR using Eqs. (16) and
obtain:
(20), respectively.
2r1 2r 2 Step 6: Repeat Steps 3–5 until ‘‘the maximum number of buses is
½aii Q i þ Bi  þ ½c Q þ Di   2ki Q DGi ¼ 0 ð32Þ reached’’. These buses are defined as ‘‘a set of candidate buses’’.
PL Q L ii i
Continue to connect DG units to ‘‘these candidate buses’’ by
Substituting Eq. (6) into Eq. (32), we obtain Eq. (33), where Yi is repeating Steps 3–5.
given in Eq. (31). Step 7: Stop if any of the violations of the constraints (Sec-
tion 3.4) occurs or the last iteration BCR is smaller than the pre-
Q Di Y i  rP1LBi  rQ2 DL i
Q DGi ¼ ð33Þ vious iteration one. Obtain the results of the previous iteration.
Y i  ki
Substituting Eqs. (31) and (33) into Eq. (25), we obtain:
5. Case study
1
Y i  ki ¼ 
SDGi
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 5.1. Test systems
r 1 Ai r 2 C i 2 r1 B i r2 D i 2
 PDi Y i   þ Q Di Y i   The proposed methodology was applied to an 11 kV 69-bus ra-
PL QL PL QL
dial distribution system with four feeders that are fed by a 6 MVA
ð34Þ
33/11 kV transformer, as depicted in Fig. 3 [51]. Its complete data
Substituting Eq. (34) into Eqs. (31) and (33), we obtain: can be found in [52]. The total active and reactive power of the
  system at the average load level defined by Eq. (1) is 3.35 MW
PDi Y i  rP1LAi  rQ2 CL i SDGi and 2.30 MVAr, respectively. The proposed methodology was
PDGi ¼  rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2 ð35Þ simulated in MATLAB environment.
PDi Y i  rP1LAi  rQ2 CL i þ Q Di Y i  rP1LBi  rQ2 DL i

 
Q Di Y i  rP1LBi  rQ2 DL i SDGi
Q DGi ¼  rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2 ð36Þ
PDi Y i  rP1LAi  rQ2 CL i þ Q Di Y i  rP1LBi  rQ2 DL i

It is observed from Eq. (35) that PDGi can be positive or negative,

depending on the characteristic of system loads. However, the load
power factor of a distribution system without reactive power com-
pensation is normally in the range from 0.7 to 0.95 lagging (induc-
tive load). PDGi is assumed to be positive in this study, i.e., DG unit
delivers active power. QDGi can be positive or negative, as given by
Eq. (36). QDGi can be positive with inductive loads or negative with
capacitive loads (i.e., DG unit injects or absorbs reactive power).
Given a SDGi value is pre-defined, the optimal PDGi and QDGi val-
ues are respectively calculated using Eqs. (35) and (36) to mini-
mize the IMO as defined by Eq. (12), after running only one load
flow for the base case system. Accordingly, the optimal power fac-
tor (pfDGi) value is specified using Eq. (3). Any power factors rather
than the optimal pfDGi value will lead to a higher IMO.

4.2. Computational procedure

DG units are considered to be placed at an average load level

(LF) defined by Eq. (1) over a given planning horizon, which has
the most positive impact on the IMO. This also reduces the compu-
tational burden and the search space. The energy loss given by Eq.
(16) is calculated by a multiyear multi-period power flow analysis
over the planning horizon. The computational procedure is ex- Fig. 3. Single line diagram of the 69-bus test distribution system without DG
plained for each step as follows: [51,52].
68 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72

5.2. Assumptions and constraints

Operating voltages are limited in the range of 0.94–1.06 p.u. and

feeder thermal limits are 5.1 MVA (270 A) [3]. It is assumed that
the time-varying voltage dependent industrial load as defined in
Eq. (2) is considered in this simulation. The loading at each bus fol-
lows the industrial load duration curve across a year shown in
Fig. 1 over a planning horizon of 15 years with a yearly demand
growth of 3%. As given by Eq. (1), the load factor or average load
level over the planning horizon (LF) is 0.75. All buses are candidate
for DG investment and more than one DG units can be installed at
the same bus. The substation transformers are close to their ther-
mal rating and would need replacing in the near future, while
the conductors exhibit considerable extra headroom for further de-
mand. For the reasons of simplicity, DG units are connected at the
start of the planning horizon and operating for the whole time a
year (8760 h) at rated capacity throughout the planning horizon.
That means the average utilization factor of DG units is 100%. Nat-
ural gas engine-based DG technology is used. Its size (SDG) is pre-
specified at 0.8 MVA. The input data given in Table 1, are employed
for benefit and cost analyses.

5.3. Numerical results

The total load of the system is 4.07 MVA. Given a pre-defined

DG size of 0.8 MVA each and the constraint of DG penetration as
defined by Eq. (13), the maximum number of DG units is limited
to be five with a total size of 4 MVA. To compare the benefits
Fig. 4. Single line diagram of the 69-bus test distribution system with DG units.
brought to the utility, five scenarios (1, 2, 3, 4 and 5 DG units) have
been analyzed.
are related to the active power loss index, reactive power loss in-
5.3.1. Location, size and power factor with respect to indices dex and a multi-objective index. As shown in Fig. 5, the indices re-
Fig. 4 presents the 69-bus system with DG units. The optimal duce when the number of DG units is increased from one to five.
locations are identified at buses 62, 35, 25, 4 and 39 where five However, when the number of DG units is further increased, the
DG units (i.e., DGs 1, 2, 3, 4 and 5, respectively) are optimally total penetration of DG is higher than the total demand as previ-
placed. Table 2 shows a summary of the results of the location, ously mentioned along with an increase in the values of indices.
power factor and size of DG units for the five scenarios as men- Substantial reductions in the indices are observed in three scenar-
tioned earlier over the planning horizon of 15 years. As each DG ios (i.e., 3, 4 and 5 DG units) when compared to one and two DG
unit is pre-defined at 0.8 MVA, its power factor is adjusted such units. For each scenario, the ILP is lower than the ILQ. This indicates
that the IMO index obtained for each scenario is lowest. The opti- that the system with DG units can benefit more from minimizing
mal power factor for each location is quite different, in the range the active power loss than to the reactive power loss.
of 0.82–0.89 (lagging). The total size is increased from 0.8 to
4 MVA with respect to the number of DG units increased from 5.3.2. DG impact on energy loss and voltage stability
one to five. It has been found from the simulation that three sce- Fig. 6 presents the total energy loss of the system for different
narios (i.e., 3, 4 and 5 DG units) satisfy the technical constraints. scenarios without and with DG units over the planning horizon.
When less than three DG units are considered, the violation of For each scenario, the total energy loss for each year is estimated
the voltage constraint (i.e., the operating voltages are under as a sum of all the energy losses at the respective load levels of that
0.94 p.u.) occurs at several buses in the system. year. As shown in Fig. 6, the system energy loss with no DG units
Fig. 5 shows a comparison of the ILP, ILQ and IMO indices with increases over the planning horizon due to the annual demand
different numbers of DG units over the planning horizon, which growth of 3%. A significant reduction in the energy loss over the
planning horizon is observed for the scenarios with DG units when
compared to the scenario without DG units. The lowest energy loss
Table 1
Economic input data. is achieved for the scenario with five DG units. As the amount of
the power generation from three DG units is still not sufficient,
Gas engine-based DG capacity [3] 0.8 MVA
the system energy loss for the scenario with three DG units reduce
Investment cost [3] $976/kW
Operation and maintenance and fuel costs [3] $46/MW h insignificantly when compared to that with four or five DG units.
Electricity sales [3] $76/MW h Fig. 7 shows the impact of DG allocation on the voltage stability
Loss incentive [3] $78/MW h of the system with and without DG units over the planning horizon
Network upgrade deferral benefit for deferral of $407/kW of DG of 15 years. For each year, the simulation has been implemented at
transformer upgrades [3]
Emission factor of grid [36] 0.910 Ton CO2/
the maximum demand, where the voltage stability margin (VSM) is
MW h worst when compared to the other loading levels. In each year, as
Emission factor of 1 MVA gas engine [36] 0.773 Ton CO2/ defined in Fig. 2, the VSM of the system with 3–5 DG units signif-
MW h icantly enhances when compared to that of the system without
Emission cost [36] $10/Ton CO2
DG units. For example, when three DG units generate an amount
Discount rate [3] 9%
of 2.4 MVA at buses 62, 35 and 25 in the first year found in Table 2,
 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 69

Table 2
Location, size and power factor of DG units.

Scenarios DG location DG size (MVA) DG power factor (lag.) Total DG size (MVA) Permissible constraints?
1 DG 62 0.8 0.87 0.8 No
2 DGs 62 0.8 0.87 1.6 No
35 0.8 0.89
3 DGs 62 0.8 0.87 2.4 Yes
35 0.8 0.89
25 0.8 0.89
4 DGs 62 0.8 0.87 3.2 Yes
35 0.8 0.89
25 0.8 0.89
4 0.8 0.85
5 DGs 62 0.8 0.87 4.0 Yes
35 0.8 0.89
25 0.8 0.89
4 0.8 0.85
39 0.8 0.82

1.0 2–15 as shown in Fig. 7. Furthermore, as shown in Fig. 7, a signif-

ILP ILQ IMO
icant increase in the voltage stability margin is found for the sce-
0.8 narios with four or five DG units when compared to three DG
units. This is due to the fact that the ILP, which is related to the
Indices

0.6
reactive power loss of the system, significantly reduces for the sce-
0.4 nario with four or five DG units when compared to three DG units,
as shown in Fig. 5. In addition, it is observed from Fig. 7 that the
0.2
VSM values with and without DG units reduce with respect to a
0.0 yearly demand growth of 3%. Hence, the lowest VSM values are
1 2 3 4 5 found in the year 15.
Number of DGs Table 3 shows a summary of the results of energy losses with-
out and with DG units for each scenario over the planning horizon
Fig. 5. Indices (ILP, ILQ and IMO) for the system with various numbers of DG units of 15 years. The energy savings due to loss reduction is beneficial. A
over planning horizon. maximum energy savings is achieved for the scenario with five DG
units when compared to three and four DG units. Table 3 also pre-
sents a summary of the results of voltage stability with and with-
1500
No DG out DG units over the planning horizon. The average voltage
3 DGs stability margin of the system (AVSM) is calculated as a sum of
Energy loss (MWh)

1200
4 DGs the VSM values of all years divided by the total planning horizon.
5 DGs
900 An increase in the average voltage stability margin (DAVSM) is ob-
served after 3–5 DG units are installed in the system. It is observed
600 from Table 3 that the VSM for each scenario increases with respect
to an increase in the number of DG units installed in the system as
300
well as a reduction in the overall energy loss of the system.
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
5.3.3. Benefit and cost analysis
Year
Table 4 presents the results of the total benefit and cost for
Fig. 6. Losses of the system with and without DG units over planning horizon. three scenarios (i.e., 3, 4 and 5 DG units) without and with addi-
tional benefits over the planning horizon of 15 years. The addi-
6.0 tional benefit includes the loss incentive (LI), emission incentive
Voltage stability margin

No DG 3 DGs
(EI) and network upgrade deferral (ND). The total benefit (B) is a
4 DGs 5 DGs
4.5 sum of all the additional benefits and the energy sales (R). The total
cost (C) is a sum of the operation, maintenance and fuel cost (OM)
3.0 and the DG capacity cost (CDG RPDGi). Table 5 shows a comparison
of the results for three different scenarios without and with addi-
tional benefits over the planning horizon. The results include the
1.5
benefit–cost ratio (BCR), net present value (NPV = B  C), payback
period, and internal rate of return (IRR).
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For exclusion of the additional benefits, it is observed from Ta-
Year ble 5 that the BCR is the same for all the scenarios at 1.288. The best
solution is five DG units as the NPV is highest at k$4162. This solu-
Fig. 7. Voltage stability margin curves for all scenarios over planning horizon. tion generates an IRR of 16.48% and a payback period of 5.6 years.
For inclusion of the additional benefits, it is seen from Table 4 that
the energy sales (R) accounts for around 89–90% of the B, leaving
the VSM increases to 4.5918 from the base case value of 2.8681 the total additional benefit (LI, EI and ND) at roughly 10–11%. It
(without DG units). A similar trend has been found for years is obvious that the R has a significant impact on the BCR when
70 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72

Table 3
Energy loss and voltage stability losses over planning horizon.

Scenarios Energy loss (GW h/15 years) Energy savings (GW h/15 years) Voltage stability
AVSM DAVSM
Base case 14.91 2.1667
3 DGs 5.97 8.94 3.5745 1.4078
4 DGs 4.46 10.45 3.5758 1.4091
5 DGs 4.35 10.56 3.6855 1.5188

Table 4
Analysis of the present value benefit and cost for different scenarios over planning horizon.

Number of DGs Without additional benefits With additional benefits

3 DGs 4 DGs 5 DGs 3 DGs 4 DGs 5 DGs
R (k$) 11,421 15,090 18,624 11,421 15,090 18,624
LI (k$) – – – 336 392 387
EI (k$) – – – 244 316 379
ND RPDGi (k$) – – – 860 1137 1403
Total benefit, B (k$) 11,421 15,090 18,624 12,862 16,934 20,793
R/B (%) 88.80 89.11 89.57
(LI + EI + ND RPDGi)/B (%) 11.20 10.89 10.43

OM (k$) 6804 8990 11,095 6804 8990 11,095

CDG RPDGi (k$) 2065 2728 3367 2065 2728 3367
Total cost, C (k$) 8869 11,718 14,462 8869 11,718 14,462

Table 5
Comparison of different scenarios over planning horizon.

Number of DGs Without additional benefits With additional benefits

3 DGs 4 DGs 5 DGs 3 DGs 4 DGs 5 DGs
BCR = B/C 1.288 1.288 1.288 1.450 1.445 1.438
NPV = B  C (k$) 2552 3372 4162 3993 5216 6331
Payback period (years) 5.60 5.60 5.60 2.80 2.82 2.86
Internal rate of return, IRR (%) 16.48 16.48 16.48 32.65 32.38 31.93

compared to the total additional benefit. However, the additional

benefits, particularly the LI play a critical role in decision-making 200
No DGs 3 DGs
about the total number of DG units or the amount of DG capacity
Percentage ratio (%)

4 DGs 5 DGs
installed. This factor has an impact on the BCR. As shown in Table 5, 150
Transformer upgrade required
the BCR slightly drops from 1.450 to 1.438 when the number of DG
units is increased from three to five, respectively. The best solution 100
is three DG units with the highest BCR of 1.450. This solution gen- Trans. rating
erates an NPV of k$3993, an IRR of 32.65% and a payback period of
50
2.80 years. In general, in the absence of the additional benefits, the
optimal number of DG units is five, while in the presence of the Trans. upgrade deferred
additional benefits, this figure is three. Inclusion of the additional 0
benefits in the study can lead to faster investment recovery with 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Year

Fig. 9. Percentage ratio of the total demand plus loss to the thermal transformer
7.5 limit over planning horizon.
3 DGs
6.0 4 DGs
NPV (million $)

a higher BCR, a higher IRR and a shorter payback period when com-
5 DGs
4.5 pared to exclusion of the additional benefits. In addition, it is
shown from Table 5 that the NPV value will increase further with
3.0 increasing DG units installed. However, the system cannot accom-
modate more than five DG units due to the violation of the maxi-
1.5 mum DG penetration constraint as defined by Eq. (13).
Fig. 8 shows an increase in the NPV with the corresponding
0.0
10 20 30 40 50 60 70 80 90 100 average DG utilization factors over the planning horizon of
15 years for three scenarios (i.e., 3, 4 and 5 units) with the addi-
Average utilization of DG (%)
tional benefits. It is observed from the figure that given a certain
Fig. 8. NPV at various average DG utilization factors for different scenerios over average DG utilization factor in the range of 0–100%, the NPV is
planning horizon. highest for the scenario with five DG units, while this figure is
 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 71

lowest for the scenario with three DG units. In addition, for each [10] Borges CLT, Falcão DM. Optimal distributed generation allocation for
reliability, losses, and voltage improvement. Int J Elect Power Energy Syst
scenario, the NPV is maximum when the utilization factor is
2006;28(6):413–20.
100% as estimated in Table 5. However, in practice, this factor [11] Abdullah MA, Agalgaonkar AP, Muttaqi KM. Assessment of energy supply and
may be less than 100% due to interruption for maintenance and continuity of service in distribution network with renewable distributed
others. Consequently, the respective NPV will be reduced as shown generation. Appl Energy 2014;113:1015–26.
[12] Zhang P, Li W, Li S, Wang Y, Xiao W. Reliability assessment of photovoltaic
in Fig. 8. power systems: review of current status and future perspectives. Appl Energy
Fig. 9 shows the percentage ratio of the total demand plus total 2013;104:822–33.
loss to the thermal limit (6 MVA) of the transformer over the plan- [13] Naderi S, Pouresmaeil E, Gao WD. The frequency-independent control method
for distributed generation systems. Appl Energy 2012;96:272–80.
ning horizon with a demand growth of 3%. For the scenarios with [14] Mashhour E, Moghaddas-Tafreshi SM. Bidding strategy of virtual power plant
DG units, the curves are plotted at the average DG utilization factor for participating in energy and spinning reserve markets – Part I: problem
of 100%. Obviously, without DG connection, an investment would formulation. IEEE Trans Power Syst 2011;26(2):949–56.
[15] Rueda-Medina AC, Padilha-Feltrin A. Distributed generators as providers of
be needed to add a new transformer before year 4. However, the reactive power support – a market approach. IEEE Trans Power Syst
selected three-DG scenario can defer in upgrading this current 2013;28(1):490–502.
transformer (6 MVA) to 11 years. A higher deferral is achieved for [16] Yuen C, Oudalov A, Timbus A. The provision of frequency control reserves from
multiple microgrids. IEEE Trans Ind Electron 2011;58(1):173–83.
the scenarios with four or five DG units. [17] Katiraei F, Aguero JR. Solar PV integration challenges. IEEE Power Energy Mag
2011;9(3):62–71.
[18] Eftekharnejad S, Vittal V, Heydt GT, Keel B, Loehr J. Impact of increased
6. Conclusions penetration of photovoltaic generation on power systems. IEEE Trans Power
Syst 2013;28(2):893–901.
[19] Wang C, Nehrir MH. Analytical approaches for optimal placement of
This paper has developed an investment planning framework distributed generation sources in power systems. IEEE Trans Power Syst
for integrating multiple Distributed Generation (DG) units in 2004;19(4):2068–76.
industrial distribution systems where the DG units are assumed [20] Hung DQ, Mithulananthan N, Bansal RC. Analytical expressions for DG
allocation in primary distribution networks. IEEE Trans Energy Convers
to be owned and operated by utilities. In this framework, analytical 2010;25(3):814–20.
expressions are proposed to efficiently identify the optimal power [21] Hung DQ, Mithulananthan N, Bansal RC. Analytical strategies for renewable
factor of DG units for minimizing energy losses and enhancing distributed generation integration considering energy loss minimization. Appl
Energy 2013;105:75–85.
voltage stability. The decision for the optimal location, size and [22] Ochoa LF, Harrison GP. Minimizing energy losses: optimal accommodation and
number of DG units is achieved through a benefit–cost analysis. smart operation of renewable distributed generation. IEEE Trans Power Syst
The total benefit includes energy sales and three additional bene- 2011;26(1):198–205.
[23] Atwa YM, El-Saadany EF. Probabilistic approach for optimal allocation of wind-
fits including loss reduction, network upgrade deferral and emis-
based distributed generation in distribution systems. IET Renew Power Gener
sion reduction. The total cost is a sum of capital, operation and 2011;5(1):79–88.
maintenance costs. The results obtained on a 69-bus test distribu- [24] Atwa YM, El-Saadany EF, Salama MMA, Seethapathy R. Optimal renewable
resources mix for distribution system energy loss minimization. IEEE Trans
tion system indicated that the additional benefits, particularly the
Power Syst 2010;25(1):360–70.
loss incentive have a significant impact on decision-making about [25] Singh D, Verma KS. Multiobjective optimization for DG planning with load
the total number of DG units or the amount of DG capacity in- models. IEEE Trans Power Syst 2009;24(1):427–36.
stalled. The additional benefits together accounted for 10–11% of [26] Kansal S, Kumar V, Tyagi B. Optimal placement of different type of DG sources
in distribution networks. Int J Elect Power Energy Syst 2013;53:752–60.
the total benefit when compared to the energy sales of 89–90%. [27] Abu-Mouti FS, El-Hawary ME. Optimal distributed generation allocation and
Inclusion of these benefits in the study can lead to faster invest- sizing in distribution systems via artificial bee colony algorithm. IEEE Trans
ment recovery with a high benefit–cost ratio, a high internal rate Power Del 2011;26(4):2090–101.
[28] Ettehadi M, Ghasemi H, Vaez-Zadeh S. Voltage stability-based DG placement
of return and a short payback period. in distribution networks. IEEE Trans Power Del 2013;28(1):171–8.
When DG units are owned by DG developers, the additional [29] Aman MM, Jasmon GB, Mokhlis H, Bakar AHA. Optimal placement and sizing of
benefits should be shared between the distribution utility and a DG based on a new power stability index and line losses. Int J Elect Power
Energy Syst 2012;43(1):1296–304.
DG developer to encourage DG connection. In this situation, the [30] Esmaili M. Placement of minimum distributed generation units observing
proposed methodology could be used as guidance for the utility power losses and voltage stability with network constraints. IET Gener Transm
on how to plan and operate DG units to obtain the additional Distrib 2013;7(8):813–21.
[31] Al Abri RS RS, El-Saadany EF, Atwa YM. Optimal placement and sizing method
benefits.
to improve the voltage stability margin in a distribution system using
distributed generation. IEEE Trans Power Syst 2013;28(1):326–34.
[32] Injeti SK, Prema Kumar N. A novel approach to identify optimal access point
References and capacity of multiple DGs in a small, medium and large scale radial
distribution systems. Int J Elect Power Energy Syst 2013;45(1):142–51.
[1] El-Khattam W, Bhattacharya K, Hegazy Y, Salama MMA. Optimal investment [33] Aman MM, Jasmon GB, Bakar AHA, Mokhlis H. A new approach for optimum
planning for distributed generation in a competitive electricity market. IEEE DG placement and sizing based on voltage stability maximization and
Trans Power Syst 2004;19(3):1674–84. minimization of power losses. Energy Convers Manage 2013;70:202–10.
[2] Viral R, Khatod DK. Optimal planning of distributed generation systems in [34] Kayal P, Chanda CK. Placement of wind and solar based DGs in distribution
distribution system: a review. Renew Sustain Energy Rev 2012;16(7):5146–65. system for power loss minimization and voltage stability improvement. Int J
[3] Siano P, Ochoa LF, Harrison GP, Piccolo A. Assessing the strategic benefits of Elect Power Energy Syst 2013;53:795–809.
distributed generation ownership for DNOs. IET Gener Transm Distrib [35] Hien NC, Mithulananthan N, Bansal RC. Location and sizing of distribution
2009;3(3):225–36. generation units for loadability enhancement in primary feeder. IEEE Syst J
[4] Piccolo A, Siano P. Evaluating the impact of network investment deferral on 2013;7(4):797–806.
distributed generation expansion. IEEE Trans Power Syst 2009;24(3):1559–67. [36] Soroudi A, Ehsan M, Zareipour H. A practical eco-environmental distribution
[5] Niknam T, Taheri SI, Aghaei J, Tabatabaei S, Nayeripour M. A modified honey network planning model including fuel cells and non-renewable distributed
bee mating optimization algorithm for multiobjective placement of renewable energy resources. Renew Energy 2011;36(1):179–88.
energy resources. Appl Energy 2011;88(12):4817–30. [37] Kucuksari S, Khaleghi AM, Hamidi M, Zhang Y, Szidarovszky F, Bayraksan G,
[6] Niknam T. A new HBMO algorithm for multiobjective daily Volt/Var control in et al. An integrated GIS, optimization and simulation framework for optimal
distribution systems considering distributed generators. Appl Energy PV size and location in campus area environments. Appl Energy
2011;88(3):778–88. 2014;113:1601–13.
[7] Martinez-Rojas M, Sumper A, Gomis-Bellmunt O, Sudrià-Andreu A. Reactive [38] Hung DQ, Mithulananthan N. Loss reduction and loadability enhancement
power dispatch in wind farms using particle swarm optimization technique with DG: a dual-index analytical approach. Appl Energy 2013;115:233–41.
and feasible solutions search. Appl Energy 2011;88(12):4678–86. [39] Hung DQ, Mithulananthan N, Bansal RC. Integration of PV and BES units in
[8] Hedayati H, Nabaviniaki SA, Akbarimajd A. A method for placement of DG units commercial distribution systems considering energy loss and voltage stability.
in distribution networks. IEEE Trans Power Del 2008;23(3):1620–8. Appl Energy 2014;113:1162–70.
[9] Brown RE, Freeman LAA. Analyzing the reliability impact of distributed [40] Hung DQ, Mithulananthan N. Assessing the impact of loss reduction on
generation. In: Proc IEEE Power Eng Soc Summer Meet 2001;2:1013–8. distributed generation investment decisions. In: Proc in Australasian
72 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72

universities power engineering conference (AUPEC) Hobart, TAS, Australia, 29 [47] Ochoa LF, Padilha-Feltrin A, Harrison GP. Evaluating distributed generation
September–3 October; 2013. impacts with a multiobjective index. IEEE Trans Power Del 2006;21(3):1452–8.
[41] Asper SG, Nwankpa CO, Bradish RW, Chiang H-D, Concordia C, Staron JV, et al. [48] Ochoa LF, Padilha-Feltrin A, Harrison GP. Evaluating distributed time-varying
Bibliography on load models for power flow and dynamic performance generation through a multiobjective index’’. IEEE Trans Power Del
simulation. IEEE Trans Power Syst 1995;10(1):523–38. 2008;23(2):1132–8.
[42] IEEE 1547 standard for interconnecting distributed resources with electric [49] El-Zonkoly AM. Optimal placement of multi-distributed generation units
power systems; October 2003. including different load models using particle swarm optimisation. IET Gener
[43] Yeh HG, Gayme DF, Low SH. Adaptive VAR control for distribution circuits with Transm Distrib 2011;5(7):760–71.
photovoltaic generators. IEEE Trans Power Syst 2012;27(3):1656–63. [50] Canizares CA, Alvarado FL. Point of collapse and continuation methods for
[44] Turitsyn K, Sulc P, Backhaus S, Chertkov M. Options for control of reactive large AC/DC systems. IEEE Trans Power Syst 1993;8(1):1–8.
power by distributed photovoltaic generators. Proc IEEE 2011;99(6):1063–73. [51] Das D. A fuzzy multiobjective approach for network reconfiguration of
[45] Willis HL, Scott WG. Distributed power generation: planning and distribution systems. IEEE Trans Power Del 2006;21(1):202–9.
evaluation. New York: Marcel Dekker Inc.; 2000. [52] Harrison GP, Piccolo A, Siano P, Wallace AR. Hybrid GA and OPF evaluation of
[46] Elgerd IO. Electric energy system theory: an introduction. New York: McGraw- network capacity for distributed generation connections. Electr Power Syst Res
Hill Inc.; 1971. 2008;78(3):392–8.