Energy 66 (2014) 202e215

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A new approach for optimum simultaneous multi-DG distributed

generation Units placement and sizing based on maximization of
system loadability using HPSO (hybrid particle swarm optimization)
algorithm
M.M. Aman a, *, G.B. Jasmon a, A.H.A. Bakar b, H. Mokhlis a, b
a
Department of Electrical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
b
UM Power Energy Dedicated Advanced Centre (UMPEDAC) Level 4, Wisma R&D, University of Malaya, Jalan Pantai Baharu, 59990 Kuala Lumpur, Malaysia

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

Article history: This paper presents a new approach for optimum simultaneous multi-DG (distributed generation)
Received 7 February 2013 placement and sizing based on maximization of system loadability without violating the system con-
Received in revised form straints. DG penetration level, line limit and voltage magnitudes are considered as system constraints.
13 December 2013
HPSO (hybrid particle swarm optimization) algorithm is also proposed in this paper to find the optimum
Accepted 16 December 2013
Available online 18 January 2014
solution considering maximization of system loadability and the corresponding minimum power losses.
The proposed method is tested on standard 16-bus, 33-bus and 69-bus radial distribution test systems.
This paper will also compare the proposed method with existing Ettehadi method and present the
Keywords:
Distributed generation
effectiveness of the proposed method in terms of reduction in power system losses, maximization of
Reactive power system loadability and voltage quality improvement.
System loadability Ó 2013 Elsevier Ltd. All rights reserved.
Hybrid particle swarm optimization (HPSO)

1. Introduction
Power utilities are facing major challenges as the demand of
power system is growing exponentially. The existing transmission
line infra-structure is unable to support such a huge power de-
mand. The present need is either to invest in transmission system
to increase the capacity or provide the consumer demand locally by
DG (distributed generation). Electric power generation integrated
within the distribution systems is known as “distributed” or
“dispersed” generation. The DG source can be a traditional com-
bustion generator (such as diesel reciprocating generator and nat-
ural gas-turbine) and non-traditional generator including fuel cell,
storage device and renewable energy source (such as wind turbine
and photovoltaic) [1,2].
DG has many advantages over centralized power generation
including power losses reduction, voltage profile improvement,
system stability improvement, pollutant emission reduction,
relieving transmission and distribution congestion and others [3e
7]. The optimum placement of DG is necessary to achieve the
maximum benefits with less investment cost [8]. After deregulation
of power system, non-utility companies are investing in
* Corresponding author. Tel.: þ60 3 7967 5348.
E-mail address: mohsinaman@gmail.com (M.M. Aman).
distribution system to meet the active power demand (MW) and
get the maximum profit. For example in United States the per-
centage of nonutility generators in distribution system has
increased from 40 GW to more than 150 GW in ten years from 1990
to 2000 [9].
In deregulated operation of distribution system, the DNO
(distribution network operator) is responsible in providing good
quality service to consumer and maintain the security and effi-
ciently utilize the existing infrastructure under different un-
certainties including load changes, available generation and
operating schedule of DG Units, particularly in case of intermit-
tent energy source (e.g. wind and solar). Different artificial
intelligent based techniques have been proposed in literature for
smooth operation of distribution system under various opera-
tional scenarios [10e15]. In some countries, the DNOs are allowed
to invest DG Units based on its interests and requirements.
However, in some other countries, the DNOs are not allowed to
own DG Units [10].
Researchers have solved the problem of optimum DG place-
ment problem on the basis of minimization of power losses [16e
20] and voltage stability [21e27] approach. Authors have also
considered DG placement as a multi-objective function, consid-
ering reduction in power losses, improved voltage regulations and
voltage stability in fitness function [28]. It has also been seen in

0360-5442/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.energy.2013.12.037
 M.M. Aman et al. / Energy 66 (2014) 202e215
Ref. [29] that the DG placement also improves the reliability and
voltage profile of the system. DG with reactive power control also
helps in better network voltage profile and lower power losses.
The DG penetration level versus power losses present a U-trajec-
tory, thus the non-optimum placement of DG may increase the
power losses [30].
The impact of DG on loadability of distribution network has
also been investigated in Refs. [25,31] and concluded that the
presence of DG increases the loadability of the distribution system.
System loadability is related with voltages of the system, the
weakest voltage buses result in poor loadability of the system.
Different approaches have been proposed to increase the load-
ability of the system by improving the voltages of the system and
the DG is placed on the selected weakest voltage bus. The weakest
voltage bus is determined using eigen value determination for
optimum DG location and the maximum loading margin is
considered for optimum DG size in Ref. [27]. CPF (continuation
power flow) method is used for weakest voltage bus determina-
tion as an optimum bus for DG placement [21]. Further authors in
Ref. [21] have also extended the work for multi-DG Unit placement
after first DG Unit has been placed. Authors have considered
maximum loading up to the voltage constraint in fitness function
for optimum DG placement using genetic algorithm in Ref. [25].
Successive modal analysis and CPF are utilized for optimum multi-
DG Units placement in Ref. [22]. However techniques based on
successive selection of DG position, when one DG Unit has been
placed, cannot lead to the global optimum solution of maximum
loadability of the system.
In this paper, a new approach for optimum simultaneous
multi-DG unit placement is presented based on maximization of
system loadability. System loadability (l) is defined as the ca-
pacity of the system with which the maximum load could be
connected without violating the system and operating con-
straints. In this paper, bus voltage and line current limit are
considered as operating and system constraints respectively. In
practice, it is highly needed to utilize the existing infrastructure
without high investment to increase the capacity of existing
system. In literature, it has been seen that the system capacity or
loadability is usually limited by two factors, thermal limits and
voltage limits. Thermal limit or thermal capacity is the ampacity
or maximum current carrying capacity limit of the conductor. The
current carrying capacity is limited by the conductor’s maximum
design temperature, which is determined by the insulation class
use [32]. However the voltage limit is the allowable minimume
maximum voltage variation for safe operation of power system
and connected load [33]. The study [33] has concluded that the
maximum loadability of the distribution system is limited by the
voltage limit rather than the thermal limit. The higher loading
factor results in large current in the distribution line which re-
sults in high voltage drop and thus presents the poor voltage
regulation. From continuation power flow (CPF) theorem [34] and
results of Ref. [21], it can be concluded that increasing the
maximum loadability (lmax) improves the overall voltage profile,
as shown in Fig. 1.
From Fig. 1, it can be observed that curve A will have overall
better voltage profile than curve B at each loading. Here in Fig. 1, s is
representing the tangent (predictor vector) at different point of
loading.
This paper is organized as follows: In Section 2, the need for
reactive power compensation in presence of DG will be discussed.
In Section 3, an optimization algorithm HPSO (hybrid particle
swarm optimization) will be proposed for optimum simultaneous
multi-DG Unit placement. In DG Unit placement based on maxi-
mization of system loadability, different DG location and DG size
combinations may result into same maximum loadability. Thus in
203
τ1
Voltage Magnitude (per unit)
τ2
Point of
Collapse
B
A
Loading Factor (λ )
Fig. 1. Effect of maximum loadability of the system on voltage profile improvement.
order to reach the optimum solution, the K-matrix in HPSO algo-
rithm will sort the best result on the basis of minimum power
losses and select the best results. In Section 4, the proposed algo-
rithm for multi-DG Unit placement will be presented. In Section 5,
the proposed algorithm will be tested on 16-bus, 33-bus and 69-bus
radial distribution test systems. The present approach of optimum
multi-DG Unit placement will also be compared with Ref. [22] in
the same section.
2. Impact of distributed generation on voltage stability
Previously DGs were mainly considered as an active source of
energy [1], however at higher system loading with maximum DG
penetration, the poor voltage profile can be a big challenge for the
system operator thus the reactive power compensation approach
must be utilized to maintain the voltages in allowable limits [35].
DG presence in the system also affects the reactive power man-
agement plan. For example in case of wind generation, asynchro-
nous induction generators are used. Such generators need reactive
power from the system to which they are connected. Different
methods of reactive compensations are stated in literature [36]
including synchronous generator, shunt capacitor banks and end-
user reactive power compensation within the reactive power con-
sumption equipment. The growing trend of using non-conventional
power generation (using wind and solar energy) has also led to the
bounding that the renewable energy generation must play their
role in improving the voltage profile and providing necessary
reactive power support.
Now-a-day state of the art technology has come out to control the
active and reactive power from DG. The wind generation is now using
doubly fed induction generator and PV inverters are using special
self-commutated line inverter, capable of absorbing and supplying
reactive power at different system loading. The reactive power
capability of solar and wind power plants can be further enhanced by
the addition of SVC, STATCOMS and other reactive support equip-
ment at the plant level. Currently, inverter-based reactive capability
is more costly compared to the same capability supplied by syn-
chronous machines [37e39]. The author in Ref. [40] has analysed the
importance of reactive power in presence of DG and concluded that
the presence of DG results into better voltage profile. However at
light load in distribution system in presence of DG, the problem of
voltage rise may arise, thus the voltage regulating device must also be
presented in the system. The energy curtailment from DG is not a
good solution as this will result in revenue lost. DG has been
considered as an active source of energy in Refs. [18e20,41e43],
204 M.M. Aman et al. / Energy 66 (2014) 202e215

2. KMML (kVA margin to maximum loadability) is defined in Ref.

[45] to represent the additional load from the operating point ‘O’
to the point of voltage collapse, as shown in Fig. 2.

From Fig. 2, it can be observed that the curve A (with optimum

DG placement) will have a better voltage profile than curve B at
each loading. Further it can also be noted that using the optimum
DG placement, the operating point of the system (lV) can also be
increased from O1 to O2 (within the allowable branch current limits
(ILimit) and voltage limits VLimits).
The effect of DG placement on system performances is evaluated
by calculating the indices defined in Table 1.

3. Hybrid particle swarm optimization (HPSO)

PSO (particle swarm optimization) is introduced by Kennedy

Fig. 2. Effect of DG placement on kVA margin to maximum loadability (KMML).
however DG has been considered as an active as well as reactive
source of energy in Refs. [16,17,21,22,24,44]. Three different types of
DGs are mainly considered in literature [17].
Type 1: DG injects active power (P) only, e.g. photovoltaic
system.
Type 2: DG injects reactive power (Q) only, e.g. synchronous
compensators.
Type 3: DG injects active power (P) but absorbs reactive power,
e.g. induction generator.
Type 4: DG injects both active (P) and reactive (Q) power e.g.
synchronous generator.
[46] to solve the optimization problem, based on socialepsycho-
logical metaphor behaviour. A particle i is represented as position
vector xi (xi ¼ xi1, xi2, xi3, xi4, .. xid). In each iteration, the i-th
particle fitness value is calculated. The best fitness position asso-
ciated with the best particles (pbest) is considered as particles best
position (pposition). The overall best fitness (gposition) of the popula-
tion associated with the overall best particles gbest is recorded as
optimum result. During the iteration procedure, the velocity and
the position of ith particles are updated according to the Eqns. (2)
and (3). The rate of position vector of ith particle is represented
as velocity vector vi (vi ¼ vi1, vi2, vi3, vi4,..vid).
ðtþ1Þ ðtÞ
vid ¼ w,vid þ c1 ,r1 ,ðpbestid  xid Þ þ c2 ,r2 ,ðgbest  xid Þ
(2)
ðtþ1Þ ðtÞ ðtþ1Þ
xid ¼ xid þ vid (3)

In this paper DG will be considered as a PQ (activeereactive where t is number of iterations, d is a total number of initial par-
power) source of energy (Type 4). Following definitions are used ticles, c1 and c2 are acceleration constants, set at 2.0, r1 and r2 are
in this paper to see the potential effect of DG on distribution system. random numbers, w is inertia weight given by Eqn. (4).

wmax  wmin
1. DG penetration level is defined as the ratio of the total DG w ¼ wmax þ t (4)
power generation (SDG) over the system demand (SLoad) as given itermax
by Eqn. (1).
where,

SDG ðkVAÞ wmax and wmin are 0.9 and 0.4 respectively. HPSO parameter
DG penetration levelð%Þ ¼  100 (1)
SLoad ðkVAÞ settings (c1, c2, wmin, wmax) are defined in accordance with
standard PSO parameters [17].

Table 1
DG placement evaluation indices [22,42].

Parameters Formulae Abbreviations

Reflossesg0 ReflossesgDG
Active line-loss reduction (ALLR) TLLR% ¼ Reflossesg0
 100  Subscript (0) is representing the base case when
no DG(s) is present in the system.
Imflossesg0 ImflossesgDG
Reactive line-loss reduction (RLR) RLR% ¼ Imflossesg0
 100  Subscript (DG) is representing the case when
Pnbus Pnbus DG(s) is present in the system.
ð Vi Li Þð0Þ ð Vi Li ÞðDGÞ
Qualified Load-index Improvement (QLI) QLI% ¼ i¼1 Pnbus i¼1
 100  nbus is total number of buses.
ð i¼1
Vi Li ÞðDGÞ
lmaxð0Þ lmaxðDGÞ
System loadability improvement (SLI) SLI% ¼ lmaxð0Þ  100  Vi is voltage magnitude at bus i.

KMML KMML
kVA margin to maximum loadability improvement (KMMLI) KMMLI% ¼ ð0Þ
KMMLð0Þ
ðDGÞ
 100  Li is active load at bus i pu.
Pnbus
Voltage profile improvement (VPI) VPI% ¼ ðV
i ¼ 1 ið0Þ
 ViðDGÞ Þ2  lmax is maximum loadability of the system.
Number of bus violating voltage limit (NBVV) NBVV ¼ 0  KMML is kVA margin to maximum loadability
of the system.
for i ¼ 1:nbus
if Vi < 0.95 or Vi > 1.05
NBVV ¼ NBVV þ 1
end
end
 M.M. Aman et al. / Energy 66 (2014) 202e215 205

In the proposed method of optimum DG placement and sizing, 4.1. Fitness function calculations
HPSO is used to find the fitness function, given by Eqn. (5).
To find the maximum loadability or loading factor (l) of the
f ¼ Maxflmax g (5)
system, the active and reactive load is increased on all buses, using
where, Eqn. (7), with equal loading factor of 0.01, till the divergence is
observed in load flow analysis.
f is the fitness function and lmax is maximum loadability or
loading factor of the system. Pnew ¼ P0  Loading FactorðlÞ (7a)

In the present case of solving multi-DG Units’ placement on the

basis of maximization of system loadability, two major problems Qnew ¼ Q0  Loading FactorðlÞ (7b)
arise.
where
1. Due to multiple combinations of DG positions and DG size
(particularly in two DG Unit and three DG Unit case), it is l is a loading factor, Po and Qo are initial active and reactive
possible that the simple PSO may stuck in local minima, thus to power load, connected with ith bus. Pnew and Qnew are final
obtain the global best solution, the good results will again active and reactive power load, connected with ith bus.
iterated with new random solution (like a scout bees in ABC
algorithm). The complete block diagram for fitness function calculation is
2. One or more DG positions and DG size may result in the same shown in Fig. 3.
maximum loading factor (lmax). Thus in order to obtain the
Start
optimum DG placement, a new K-matrix variable is defined
which will contain the DG position(s), DG size(s), loadability and
Read System Data
power losses, as shown in Eqn. (6). The good results are sorted
Busdata, Linedata, Base Voltage, Base MVA,
out on the basis of minimum power losses and the first row Desired Accuracy (ep=1x10-4)
solution will be treated as the best solution. In proposed HPSO,
K-matrix variable will be formed from the updated gbest posi- Form BIBC Matrix
tions and will help in finding the optimum DG placement based
on system maximum loadability as well as minimum power
Set Limits
losses. The utilization of K-Matrix in solving multi-DG Unit Cutoff=1000, Lmb=1, Lmb(max)=10000
placement will be explained in next section.

2 3 Yes
Fitness Sorting Criteria Lmb==Lmb(max)
zfflffl}|fflffl{ zfflfflffl}|fflfflffl{
6 7
6 x11 x12 x13 x14 lmax1 Plosses1 7 No
6 7
6 x21 : : : lmax2 Plosses2 7 P=Po+Lmb x Po
6 7
6 7 Q=Qo+Lmb x Qo
K-Matrix ¼ 6 : : : : lmax3 Plosses3 7 (6)
6 7
6 : : : : lmax4 Plosses4 7
6 7 It=0
6 7
4 : : : : : : 5 Accuracy=1
xn1 : : : lmaxn Plossesn
No
Due to above two reasons, a modified form of original PSO is Accuracy>ep
proposed and referred as hybrid particle swarm optimization
Lmb=Lmb+0.01

Yes
(HPSO). The HPSO algorithm is inspired from Artificial Bee Colony
algorithm, in terms of scout bees’ behaviour, thus it is referred as It=It+1 ; Vpr=V
‘hybrid’. The detail of ABC (Artificial Bee Colony) algorithm can be
seen in Ref. [47]. Here it is noticed that HPSO differs from multi- Yes
It==Cutoff Break
objective approach, multi-objective approach is commonly used
in solving two independent quantities (e.g. cost and minimum No
power losses). HPSO is particularly useful when two quantities are
Forward Loop:
highly dependent on each other and the improvement in one Compute I using Eq. (A-1)
quantity will also result in better solution of other quantity (e.g. Backward Loop:
loadability, voltage profile or minimum power losses). The appli- Compute V using Eq. (A-2)
cation of HPSO in solving multi-DG Unit placement will be
explained in next section.
Compute Accuracy Using Eq. (A-3)

4. Proposed algorithm for optimum DG placement and sizing Stored Lmb and Vmin

In this section, a new algorithm for optimum multi-DG Units’

Print Results
placement will be proposed on the basis of maximization of system Loading Factor (λ)
loadability, given by Eqn. (5). Thukaram radial load flow used in
voltage stability and optimization tool will be used to compute the End
maximum loading of the system [48]. The generalize steps of
finding maximum loadability are given below. Fig. 3. Flow chart for calculation of maximum loadability of the system (lmax).
206 M.M. Aman et al. / Energy 66 (2014) 202e215

4.2. Problem formulation amperage limits and bus voltage constraints are also introduced in
problem formulation, given in Eqns. (11) and (12) respectively.
HPSO method is used to find the optimum fitness based on
maximization of system loadability and minimum power losses. i
IDG i
< ILimit i ¼ 1 to nbr (11)
The initial particle (xi) is a 1  2 vector (PDG1, PDG1), 1  4 vector
(PDG1, PDG1, PDG2, PDG2) or 1  6 vector (PDG1, PDG1, PDG2, PDG2, PDG3,
PDG3), representing random DG positions (PDG) and DG sizes (PDG). k  1:05
0:95  Vbus k ¼ 1 to nbus (12)
Following constraints, given in Eqns. (8)e(10), related to DG
placement are considered in problem formulation. where
X
n X
Size of DG : 0 PDG  Pload ðn ¼ No: of DG unitsÞ nbr is total number of branches, nbus is total number of buses and
k¼1 Vbus is bus voltage.
(8)
The amount of reactive power from DG is given by Eqn. (13).
Position of DG : 2  DG position  nbuses (9)  
QDG ¼ PDG  tan cos1 ðpower factorÞ ; (13)
Position of DG : PDG1 sPDG2 sPDG3 (10)
In case of fixed DG penetration, the amount of active and reac-
The placement of DG may increase the line current in some of tive power from a single DG Unit can be calculated using Eqns. (14a)
the branches and bus voltages on some of the buses. Thus the line and (14b) respectively.

Start

Initialize HPSO; Pop Size=70; It_Max=70;

Max_Cycle=3; It=1; Cycle=0

Initialize random particles, following the constrains

given in Eqns. (8) to (12)

Calculate fitness of each particles position (p)

If fitness(p) is better than

fitness(pbest) than pbest=p
Join new set of Particles
with K-matrixnew(p)
Set best of pbests as gbest

If fitness(p)==fitness(gbest) then construct the K-matrix

K-matrix=[p fitness(p) losses(p)]

Update particles(p) using Eqns. (2) and (3)

It=It+1
Initialize random particles,
following the constrains
No given in Eqns. (8)-(12)
Is It==It_max?

Yes
Sort K-matrix:
If K[fitness(p)]==gbest(fitness) then construct K-matrix(new)
K-matrix(new)=[p fitness(p) losses(p)]

Sort K-matrix(new):
Sort K-matrix(new) on the basis of losses(p)
such that losses(p1)<losses(p2)<losses(p3)….losses(pn)

Cycle=Cycle+1

Is Cycle==Max_Cycle?
No
Yes
Select the first best particles of K-matrix(new) as optimum result

End

Fig. 4. Flow Chart of Proposed Algorithm for optimum simultaneous multi DG unit Placement based on maximization of system loadability using HPSO.
 M.M. Aman et al. / Energy 66 (2014) 202e215 207

Table 2a
Maximum loadability calculations under different DG penetration and operating power factor for 16-bus radial distribution test system.

PFY No. of DGs ¼ 1 No. of DGs ¼ 2 No. of DGs ¼ 3

DG penetration (%) DG penetration (%) DG penetration (%)

80 60 40 20 80 60 40 20 80 60 40 20

1 9.63 9.14 8.62 8.1 9.41 8.96 8.5 8.03 9.32 8.88 8.45 8.01
0.95 9.69 9.18 8.66 8.12 9.52 9.04 8.56 8.06 9.44 8.98 8.51 8.04
0.9 9.64 9.15 8.64 8.11 9.51 9.03 8.55 8.06 9.43 8.98 8.51 8.04
0.85 9.58 9.11 8.61 8.1 9.47 9 8.53 8.05 9.4 8.95 8.5 8.03
0.8 9.51 9.05 8.58 8.08 9.42 8.97 8.51 8.04 9.36 8.92 8.48 8.02
0.75 9.43 9 8.54 8.07 9.36 8.93 8.48 8.03 9.3 8.88 8.45 8.01

Highlighted result shows the maximum loadability of the system at different level of DG penetration.
PF ¼ power factor.

Table 2b
Maximum loadability calculations under different DG penetration and operating power factor for 33-bus radial distribution test system.

PFY No. of DGs ¼ 1 No. of DGs ¼ 2 No. of DGs ¼ 3

DG penetration (%) DG penetration (%) DG penetration (%)

80 60 40 20 80 60 40 20 80 60 40 20

1 4.3 4.18 4.04 3.84 4.69 4.43 4.15 3.84 4.66 4.43 4.16 3.84
0.95 4.43 4.28 4.11 3.88 4.99 4.65 4.28 3.89 4.9 4.6 4.28 3.91
0.9 4.45 4.3 4.12 3.89 5.05 4.69 4.31 3.89 4.95 4.64 4.3 3.92
0.85 4.45 4.3 4.12 3.89 5.07 4.71 4.31 3.89 4.97 4.65 4.31 3.92
0.8 4.45 4.3 4.12 3.89 5.07 4.7 4.31 3.89 4.97 4.65 4.31 3.92
0.75 4.43 4.29 4.11 3.88 5.05 4.69 4.3 3.88 4.95 4.64 4.3 3.91

Highlighted result shows the maximum loadability of the system at different level of DG penetration.
PF ¼ power factor.

Table 2c
Maximum loadability calculations under different DG penetration and operating power factor for 69-bus radial distribution test system.

PFY No. of DGs ¼ 1 No. of DGs ¼ 2 No. of DGs ¼ 3

DG penetration (%) DG penetration (%) DG penetration (%)

80 60 40 20 80 60 40 20 80 60 40 20

1 4.63 4.3 3.96 3.6 4.65 4.32 3.97 3.6 4.65 4.32 3.96 3.6
0.95 4.88 4.48 4.08 3.66 4.91 4.5 4.08 3.66 4.91 4.5 3.84 3.66
0.9 4.92 4.51 4.1 3.67 4.96 4.53 4.1 3.67 4.96 4.53 4.1 3.67
0.85 4.93 4.52 4.1 3.67 4.97 4.54 4.1 3.67 4.97 4.53 4.1 3.67
0.8 4.91 4.51 4.1 3.67 4.96 4.53 4.09 3.67 4.95 4.52 4.1 3.66
0.75 4.89 4.49 4.08 3.66 4.93 4.51 4.08 3.66 4.93 4.5 4.08 3.66

Highlighted result shows the maximum loadability of the system at different level of DG penetration.
PF ¼ power factor.

systems under different power factors and different level of DG

PDG ¼ SDG  PF (14a) penetration, using HPSO method. The DG operating power factor is
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi varied from 1.0 to 0.75 (in step of 0.05) and the DG penetration level
QDG ¼ S2DG  PDG 2 (14b) has been changed from 20% up to 80% (in step of 20%). The obtained
results of maximum loadability of the system are summarized in
Table 2.
where
Table 3 shows the system details of 16-bus, 33-bus and 69-bus
SDG ¼ SLoad  DG_penetration_level  1=No: of DG Units radial distribution system.

(14c)
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u nbus Table 3
uX X
nbus
16-Bus, 33-bus and 69-bus radial distribution test system’s details.
SLoad ¼ t Pi2 þ Qi2 (14d)
i¼2 i¼2 Test Total Active Total reactive Total apparent System power
system load in load in load in factor
The complete flow chart of proposed algorithm of multi-DG Unit kW (P) kVAR (Q) kVAS ¼ Sqrt(P2 þ Q2) (PF ¼ P/S)
placement is shown in Fig. 4. 16-Bus 28 700 5900 29 300.17 0.9795
system
33-Bus 3715 2300 4369.35 0.8502
4.3. Selection of DG operating power factor system
69-Bus 3801.9 2694.1 4659.67 0.8159
In this section, the maximization of system loadability is system

computed for 16-bus, 33-bus and 69-bus radial distribution test where Sqrt ¼ square root; PF ¼ power factor.
208 M.M. Aman et al. / Energy 66 (2014) 202e215

Fig. 5. Single line diagram of modified 16-bus radial distribution test systems.

By comparing, Tables 2 and 3, it can be seen that the most op-
timum (maximum loadability) results are achieved when the DG
operating power factor is nearer to the system power factor. Thus
0.95 operating power-factor of DG is considered in case of 16-bus
test system, and 0.85 power-factor of DG is considered in case of
33-bus and 69-bus test systems. From Table 2, it can also be noticed
that higher DG penetration will result in higher system loadability,
thus the amount of DG penetration level can be decided on eco-
nomic grounds.
5. Application of proposed DG placement algorithm on radial
distribution system
Case 2 (with 40% DG penetration): In case 2, the DG pene-
tration is considered 40% of total load (SLoad). The DG constraints
given in Eqns. (9) and (10) and line amperage constraint given in
Eqn. (12) are introduced in problem formulation. Case 2 is
considered in order to compare the proposed method with [22]
method where author has considered 40% DG penetration level
and have applied the proposed method on 33-bus radial distri-
bution test system.
The applications of proposed multi-DG Unit placement on
different radial distribution system are as follows.
5.1. 16-Bus radial distribution test system

In this section, the proposed algorithm of optimum multi-DG
Unit placement will be applied on 3-phase, 12.66 kV standard 16-
bus [49], 33-bus [50] and 69-bus [51] radial distribution test sys-
tems. The system details including bus and line data are given in
Appendix A. In the proposed, DG will be considered as a ePQ load
and the DG will provide active and reactive power. The DG oper-
ating power factor is set at 0.95 for 16-bus system and 0.85 for 33-
bus and 69-bus radial distribution system. Following two case
studies are presented:
Case 1 (with maximum DG penetration): In case 1, DG pene-
tration is considered as maximum as possible, without violating DG
constraints, line amperage constraint and bus voltage constraint,
given in Eqns. (8)e(12). In case 1, it is highly possible, the bus
voltages and line current may cross the maximum limit, thus line
amperage constraint and bus voltage constraint are introduced in
problem formulation.
16-Bus system is a standard 12.66 kV network, having 3 feeders
[49]. The system details are given in appendix Table A-1. In order to
solve the 3 feeder system, the system has been transformed to a
single feeder, 15-bus radial distribution test system, as shown in
Fig. 5. When the proposed algorithm of DG Unit placement is
applied on modified 16-bus radial distribution test system, results
presented in Table 4 are obtained. The base load of modified 16-bus
system is 29 300.17 kVA.
From Table 4, following points can be concluded:
1. Using the proposed method of optimum DG placement, system
maximum loadability, kVA margin to maximum loadability
(KMML) and voltage quality has been significantly improved.
However the power losses have been increased with maximum
DG penetration and reduced with limited 40% DG penetration.
Thus the amount of DG penetration must be decided on

Table 4
Application of proposed algorithm for optimum single and multi-DG Units placement on 16-bus test system.

No. of DG Units DG size in kVA / (DG position) Active power Reactive power Maximum KMML QLI VPI NBVV lV
losses (kW) losses (kVAR) system
loadability

Base case
Without DG e 511.40 590.33 7.55 192 209.12 28.1496 e 0 1.05
Proposed method (case 1)
Single DG Unit 21859.62 / (8) 315.02 407.66 9.13 238 210.39 28.8557 0.0895 0 1.33
Two DG Units 7921.98 / (7) 492.59 585.49 9.4 246 121.43 29.0211 0.1131 0 1.34
22058.09 / (8)
Three DG Units 11506.39 / (11) 536.56 635.06 9.64 253 153.4 29.0695 0.1025 0 1.71
14430.99 / (9)
9654.92 / (8)
Proposed method (case 2)
Single DG Unit 11720.1 / (08) 176.54 227.58 8.4 216 821.26 28.5396 0.0438 0 1.30
Two DG Units 11720.1  (1/2) / (11) 200.63 248.69 8.55 221 216.29 28.5723 0.0445 0 1.45
11720.1  (1/2) / (10)
Three DG Units 11720.1  (1/3) / (11) 178.36 225.17 8.50 219 751.28 28.5623 0.0418 0 1.45
11720.1  (1/3) / (08)
11720.1  (1/3) / (10)
 M.M. Aman et al. / Energy 66 (2014) 202e215 209

70

60

50

Percentage Improvement
40

30

20

10

0
PLR RLR KMMLI SLI QLII PLR RLR KMMLI SLI QLII
-10
Proposed Method (With maximum DG Proposed Method (With 40% DG penetration)
penetration)
-20

Single DG Unit Two DG Units Three DG Unit

Fig. 6. Comparison of DG Evaluation indices in presence of DG Unit(s) for modified 16-bus radial distribution system.

economic as well as on technical grounds. The DG placement
evaluation indices defined in Table 1 are used to develop the
effectiveness of proposed method in quantitative form and the
obtained results are shown in Fig. 6.
2. To show the advantage of the proposed method of DG place-
ment, the loading margin factor (l) is also calculated from 0 up
to the voltage limit (lV), satisfying the line and bus voltage
constraints also. The higher value of lV will allow the operator to
add more load with the same existing infrastructure. It was
found that the base case can carry only 30 765.18 kVA (lV ¼ 1.05)
without violating the voltage and line constraints. However the
proposed method with 40% DG penetration can carry up to
38 969.23 kVA (lV ¼ 1.33), 39 262.23 kVA (lV ¼ 1.34) and
50 103.29 kVA (lV ¼ 1.71) in case of single, two and three DG
Units respectively. These results have been summarized in Fig. 7.
Similar results are achieved in case 2.

From above results and discussion of 16-bus test system, it can

also be observed that the performance of one DG Unit, two DG
Unit and three DG Unit placement are approximately the same.
Thus the optimum number of DG Units in case 16-bus system
must be 1, in order to reduce the installation and maintenance cost
of DG.

5.2. 33-Bus radial distribution test system

When the proposed algorithm of DG Unit placement is applied

Fig. 7. System maximum loading curve for 16-bus radial distribution system (case 1).
on 33-bus radial distribution test system (shown in Fig. 8), results

Fig. 8. Single line diagram of 33-bus test system.

210 M.M. Aman et al. / Energy 66 (2014) 202e215

Table 5
Application of proposed algorithm for optimum single and multi-DG Units placement on 33-bus test system (case 1).

No. of DG Units DG size in kVA / (DG position) Active power Reactive power Maximum KMML QLI VPI NBVV
losses (kW) losses (kW) system
loadability

Base case
Without DG e 210.99 143.01 3.41 10 530.40 3.5228 e 21
Proposed method
Single DG Unit 3623.9 / (8) 131.85 113.03 4.31 14 462.55 3.7213 0.1757 0
Two DG Units 1313.9 / (16) 87.65 73.32 5.00 17 477.40 3.7609 0.0700 0
2212.3 / (22)
Three DG Units 444.0 / (29) 84.16 68.02 5.04 17 652.18 3.7713 0.0643 0
1364.1 / (15)
1973.0 / (31)

presented in Table 5 are obtained. The base load of 33-bus system is losses have been reduced to 131.90 kW with optimum DG
4369.35 kVA. The 33-bus system details are given in appendix placement at bus-8 and DG size of 3625.2 kVA. In last cycle
Table A-2. (cycle 3), the best result with maximum loadability (lmax ¼ 4.31)
From Table 5, following points can be concluded: and corresponding minimum power losses (131.85 kW) is

1. Using the proposed method of optimum DG placement, power

losses have been reduced, while system maximum loadability,
kVA margin to maximum loadability (KMML), and voltage 1.08
quality has been significantly improved. The DG placement
1.06
evaluation indices defined in Table 1 are used to develop the 1.05
effectiveness of proposed method in quantitative form and the 1.04
obtained results are shown in Fig. 9.
1.02
2. The overall voltage profile curve for 33-bus radial distribution
Voltage Magnitude(pu)

system in presence of DG Unit(s) is shown in Fig. 10. 1

3. To show the advantage of the proposed method of DG
0.98
placement, the loading margin factor (l) is also calculated
from 0 up to the voltage limit (lV), satisfying line constraints 0.96
0.95
also. It was found that the base case can carry only
0.94
2403.14 kVA (lV ¼ 0.55) without violating the voltage and line
constraints. However the proposed method can carry up to 0.92
5898.62 kVA (lV ¼ 1.35), 8520.23 kVA (lV ¼ 1.95) and 0.9
Base case (Without DG)
Single DG (8)
8738.70 kVA (lV ¼ 2.00) in case of single, two and three DG Two DGs (16 & 32)
Units respectively. These results have been summarized in 0.88 Three DGs (15,29 & 31)
Fig. 11. 5 10 15 20 25 30 33
Bus No.
4. Fig. 12 shows iteration versus maximum loadability of the sys-
tem convergence plot of proposed HPSO algorithm for single DG Fig. 10. Voltage profile curve of 33-bus radial system (case 1).
Unit placement. From Fig. 9, it can be observed that the pro-
posed method reaches to the maximum loadability of the sys-
tem in less number of iterations with corresponding minimum
power losses. In cycle 1, proposed algorithm results in optimum
DG placement at bus-8 with DG size 3623.1 kVA having mini-
mum power loss 132.08 kW. However, in cycle 2, the power

80
70
Percentage Improvement

60
50
40
30
20
10
0
PLR RLR KMMLI SLI QLII
Single DG Unit Two DG Units Three DG Units

Fig. 9. Comparison of DG Evaluation indices in presence of DG Unit(s) for 33-bus radial

distribution system (case 1). Fig. 11. System maximum loading curve for 33-bus radial distribution system (case 1).
 M.M. Aman et al. / Energy 66 (2014) 202e215 211

4.32 80

Percentage Improvement
70
Maximum Loadability of the System

4.31 60
Cycle 1 (8, 3623.1 kVA) 50
4.3 λmax=4.31
Plosses=132.08 kW 40
4.29 30
( λmax)

Cycle 2 (8, 3625.2 kVA) 20

4.28 λmax=4.31
Plosses=131.90 kW 10
0
4.27 Cycle 3 (8,3623.9 kVA) Proposed Ettheadi Proposed Ettheadi Proposed Ettheadi
λmax=4.31 Method Method Method Method Method Method
4.26 Plosses=131.85 kW
Single DG Unit Case Two DG Units Case Three DG Units Case

4.25 PLR QLR KMMLI SLI QLII

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70
No. of Iterations
Fig. 13. Comparison of proposed method and Ettehadi method for the same level of DG
Cycle 1 Cycle 2 Cycle 3 penetration (case 2).

Fig. 12. Convergence plot of proposed HPSO algorithm.

On the basis of above discussion it can be concluded that the DG
placement based on maximization of system loadability works
obtained with optimum DG placement at bus-8 and DG size of
better than the Ettehadi method either for single or multi-DG Unit
3623.9 kVA.
placement. From above results and discussion of 33-bus test sys-
tem, it can also be observed that the performance of two DG Unit
The proposed approach of optimum DG placement is also
placement and three DG Unit placements are approximately the
compared with Ettehadi method with 40% DG penetration
same in both cases, either in case of maximum DG penetration or
(1747.7 kVA). In Ettehadi method [22], the author has considered
40% DG penetration. Thus the optimum number of DG Units in case
the voltage stability approach (based on eigen value determination)
33-bus system must be 2, in order to reduce the installation and
for optimum DG placement. A fixed DG with 40% DG penetration
maintenance cost.
(1747.7 kVA) is placed on selected weakest voltage buses. The
comparative results of proposed method with Ettehadi method are
summarized in Table 6. 5.4. 69-Bus radial distribution test system
From Table 6, following points can be concluded:
When the proposed algorithm of DG Unit placement is applied
1. It can be seen that in terms of active and reactive power losses on 69-bus radial distribution test system (shown in Fig. 14), results
reduction, system loadability improvement, voltage profile presented in Table 7 are obtained. The base load of 69-bus system is
improvement and quality load index improvement the perfor- 4659.67 kVA. The 69-bus system details are given in appendix
mance of proposed method is found better than the Ettehadi Table A-3.
method. The DG placement evaluation indices defined in Table 1 From Table 7, following points can be concluded:
are used to compare the proposed method with Ettehadi
method in quantitative terms and the comparative results are 1. Using the proposed method of optimum DG placement on 69-
presented in Fig. 13. bus radial distribution test system, power system losses have
2. From Table 6, it can also be observed that the number of been reduced, system maximum loadability, quality load index,
buses violating voltage limits (NBVV) has been reduced from kVA margin to maximum loadability and overall voltage profile
9-buses (in case of Ettehadi method) to 3 buses (proposed has been significantly improved, as compared to the base case.
method). The DG placement evaluation indices defined in Table 1 are also
3. The loading margin factor up to the voltage limit (lV) is also calculated and results are presented in Fig. 15.
found better than the Ettehadi method in all three scenarios of 2. The overall voltage profile of 69-bus test system in presence of
DG placement. DG is shown in Fig. 16.

Table 6
Comparison of proposed algorithm with Ettehadi method [22] for optimum single and multi-DG Units placement with 40% DG penetration level (case 2).

No. of DG Units DG size in kVA / (DG position) Active power Reactive power System loadability KMML QLI VPI lV NBVV
losses (kW) losses (kVAR) margin

Ettehadi method
Single DG Unit 1747.7 / (33) 148.46 116.29 3.73 11 928.33 3.6141 0.1022 0.8 9
Two DG Units 1747.7  (1/2) / (33) 103.38 75.25 4.14 13 719.76 3.6290 0.0451 1.2 0
1747.7  (1/2) / (18)
Three DG Unit 1747.7  (1/3) / (33) 99.00 70.75 4.07 13 413.91 3.6257 0.0318 1.15 0
1747.7  (1/3) / (18)
1747.7  (1/3) / (32)
Proposed method
Single DG Unit 1747.7 / (14) 112.80 81.92 4.12 13 632.37 3.6562 0.2168 1.0 3
Two DG Units 1747.7  (1/2) / (17) 52.72 38.15 4.31 14 462.55 3.6529 0.0587 1.35 0
1747.7  (1/2) / (31)
Three DG Unit 1747.7  (1/3) / (14) 59.59 41.38 4.31 14 462.55 3.6559 0.0973 1.25 0
1747.7  (1/3) / (18)
1747.7  (1/3) / (32)
212 M.M. Aman et al. / Energy 66 (2014) 202e215

36 37 38 39 40 41 42 43 44 45 46

66 67
51 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Substation
Main

68 69

47 48 49 50

53 54 55 56 57 58 59 60 61 62 63 64 65

28 29 30 31 32 33 34 35

Fig. 14. Single line diagram 69-bus test system.

Table 7
Application of proposed algorithm for optimum single and multi-DG Units placement on 69-bus test system.

No. of DG DG size in kVA / (DG position) Active power Reactive power Maximum KMML QLI VPI lV NBVV
losses losses system
loadability

Base case
Without DG e 224.95 102.15 3.22 10 344.47 3.6189 e 0.6 9
Proposed method with maximum DG penetration (case 1)
Single DG Unit 3684.7 / (61) 87.13 38.34 4.91 18 219.32 3.8713 0.1459 1.75 0
Two DG Units 3685.1 / (61) 86.68 37.17 4.91 18 219.32 3.8719 0.1459 1.75 0
547.6 / (48)
Three DG Units 3652.5 / (61) 87.00 38.18 4.91 18 219.32 3.8715 0.1459 1.75 0
32.2 / (63)
152.9 / (46)
Proposed method with 40% DG penetration (case 2)
Single DG Unit 1863.9 / (64) 41.94 24.09 4.1 14 444.98 3.7582 0.0981 1.45 0
Two DG Units 1863.9  (1/2) / (64) 29.98 18.01 4.1 14 444.98 3.7573 0.0968 1.45 0
1863.9  (1/2) / (61)
Three DG Units 1863.9  (1/3) / (62) 28.78 17.40 4.1 14 444.98 3.7569 0.0967 1.45 0
1863.9  (1/3) / (64)
1863.9  (1/3) / (61)

100
90
Percentage Improvement

80
70
60
50
40
30
20
10
0
PLR RLR KMMLI SLI QLII PLR RLR KMMLI SLI QLII
Proposed Method (With Maximum DG Proposed Method (With 40% DG penetration)
penetration)

Single DG Unit Two DG Units Three DG Unit

Fig. 15. Comparison of DG Evaluation indices in presence of DG unit(s) with maximum DG penetration (Case 1) and 40% DG penetration (Case 2) for 69-bus radial distribution
system.
 M.M. Aman et al. / Energy 66 (2014) 202e215 213

consideration of line and voltage limit constraints which allows the

1.06
planning engineers to well handle the problem of DG placement.
1.05
From 16-bus, 33-bus and 69-bus test system results, it was also
1.04
found that the optimum number of DG Unit select is also subjective
1.02 (w.r.t. the distribution system), a detailed study must be carried out
Voltage Magnitude(pu)

at planning stage, otherwise higher number of DG Unit placement

1 will result in loss of investment only.

0.98
0.96
0.95
0.94
Base case (Without DG)
Single DG (61)
0.92
Two DGs (61 & 48)
0.9
Three DGs (61,63 & 46)
10 20 30
Bus No.
Fig. 16. Voltage profile curve of 69-bus radial system (case 1).
3. From Table 7, it can also be seen that the loading margin factor
considering voltage and line limit (lV) for base case (when no
DG is present) is only 2795.80 kVA (lV ¼ 0.6). However the
proposed method can carry up to 8154.42 kVA (lV ¼ 1.75) and
6756.52 kVA (lV ¼ 1.45) in case of maximum DG penetration
and 40% DG penetration respectively, in all scenario of DG
placement (single DG Unit, two DG Units and three DG Units).
These results are also presented in Fig. 17.
4. From above results it can also be observed that in both cases
either maximum DG penetration or 40% DG penetration, the
performance of one, two and multi-DG Unit placement is
found approximately the same in terms of power loss reduction,
maximum loadability improvement and voltage quality
improvement. Thus in present case of 69-bus test system, op-
timum number of DG placement must be ‘one’ in order to get
the maximum benefit with minimum installation and operation
cost of DG.
6. Conclusion
This paper has presented a new approach for optimum
simultaneous multi-distributed generation (DG) Units’ placement
and sizing on the basis of maximization of system loadability.
Hybrid particle swarm optimization (HPSO) algorithm is also
proposed to solve the single objective and multi-constraints
40 50 60 69 problem. HPSO will help in getting the best result considering
loadability as well as minimum power losses, when two or more
DG combinations will result in the same maximum loadability.
The application of proposed algorithm on 16-bus, 33-bus and 69-
bus test system shows that with 40% DG penetration only, power
losses can be reduced up to 60e75% and maximum loadability of
the system can increase up to 15e40% and overall voltage quality
significantly improves. The proposed method was also compared
with Ettehadi method and it was found that the proposed method
works better than the existing method in terms of maximization
of system loadability, reduction in power system losses, improving
the KVA margin to maximum loadability (KMML) and improve-
ment in voltage quality.
Acknowledgements
This work was supported by the Bright Spark Programme and
the Institute of Research Management and Monitoring Fund-IPPP
(Grant Code: PV144/2012A) of University of Malaya and HIR/
MOHE Research Grant (Grant Code: D000004-16001) of Ministry of
Higher Education, Malaysia.

The proposed algorithm of optimum multi-DG Unit placement

Appendix
has been well tested on 16-bus, 33-bus and 69-bus radial distri-
bution test system and improved results are obtained in all cases.
D. Thukaram RLF method [52]
The other advantage of proposed method is simplicity and practical
Thukaram load flow is used in the present paper to carry out the
load flow analysis. Thukaram load flow method is based on for-
wardebackward sweep. In backward sweep, the load current is
computed for nth loads in the network, using the following
relation.

Ii ¼ conjðSi =Vi Þ i ¼ 1; 2; 3; .n (A-1)

where S is the connected load, V is the voltage at ith bus in nth bus
system.
However, in forward sweep, the receiving end voltage (Vr) is
calculated using the difference of sending end voltage (Vs) and line
drop, as given by Eqn. (A-2).

Vr ¼ Vs  IðR þ jXÞ (A-2)

where I is branch current, R is line resistance and X is line reactance.

Fig. 17. System maximum loading curve for 69-bus radial distribution system.
The current in each branch is formulated as a function of equivalent
current injections and the matrix is known as BIBC (bus-injection to
branch-current) [53]. Apply the same approach to nth buses in the
radial system. The convergence criteria proposed is given by the
following relation.
214 M.M. Aman et al. / Energy 66 (2014) 202e215

Table A-3
h iD 69-Bus test system data [51].
Max Vik  Vik1 tolerance i ¼ 1; 2; 3; .n (A-3)
From To P (kW) Q (kW) R (ohms) X (ohms) Imax

where, k is no. of iteration and n is total number of buses. 1 2 0 0 0.0005 0.0012 400
2 3 0 0 0.0005 0.0012 400
3 4 0 0 0.0015 0.0036 400
4 5 0 0 0.0251 0.0294 400
5 6 2.6 2.2 0.366 0.1864 400
Table A-1 6 7 40.4 30 0.381 0.1941 400
16-Bus test system data [49]. 7 8 75 54 0.0922 0.047 400
8 9 30 22 0.0493 0.0251 400
From To P (kW) Q (kVAR) R (ohms) X (ohms) Imax
9 10 28 19 0.819 0.2707 400
1 2 0 0 1.0E-10 1.0E-10 2500 10 11 145 104 0.1872 0.0619 200
2 3 2000 1600 0.1202 0.1603 1400 11 12 145 104 0.7114 0.2351 200
3 4 3000 400 0.1282 0.1763 500 12 13 8 5 1.03 0.34 200
3 5 2000 400 0.1442 0.2885 500 13 14 8 5.5 1.044 0.345 200
5 6 1500 1200 0.0641 0.0641 500 14 15 0 0 1.058 0.3496 200
2 7 4000 2700 0.1763 0.1763 1400 15 16 45.5 30 0.1966 0.065 200
7 8 5000 1800 0.1282 0.1763 1000 16 17 60 35 0.3744 0.1238 200
7 9 1000 900 0.1763 0.1763 500 17 18 60 35 0.0047 0.0016 200
8 10 600 500 0.1763 0.1763 500 18 19 0 0 0.3276 0.1083 200
8 11 4500 1700 0.1282 0.1763 500 19 20 1 0.6 0.2106 0.069 200
2 12 1000 900 0.1763 0.1763 1400 20 21 114 81 0.3416 0.1129 200
12 13 1000 1100 0.1442 0.1923 500 21 22 5 3.5 0.014 0.0046 200
12 14 1000 900 0.1282 0.1763 500 22 23 0 0 0.1591 0.0526 200
14 15 2100 800 0.0641 0.0641 500 23 24 28 20 0.3463 0.1145 200
24 25 0 0 0.7488 0.2475 200
Imax is assumed maximum line current limit in Amperes. SB ¼ 100 MVA;
25 26 14 10 0.3089 0.1021 200
VB ¼ 12.66 kV.
26 27 14 10 0.1732 0.0572 200
3 28 26 18.6 0.0044 0.0108 200
28 29 26 18.6 0.064 0.1565 200
29 30 0 0 0.3978 0.1315 200
30 31 0 0 0.0702 0.0232 200
31 32 0 0 0.351 0.116 200
Table A-2 32 33 14 10 0.839 0.2816 200
33-Bus test system data [50]. 33 34 19.5 14 1.708 0.5646 200
34 35 6 4 1.474 0.4873 200
From To P (kW) Q (kVAR) R (ohms) X (ohms) Imax
3 36 26 18.55 0.0044 0.0108 200
1 2 100 60 0.0922 0.0470 400 36 37 26 18.55 0.064 0.1565 200
2 3 90 40 0.4930 0.2510 400 37 38 0 0 0.1053 0.123 200
3 4 120 80 0.3661 0.1864 400 38 39 24 17 0.0304 0.0355 200
4 5 60 30 0.3811 0.1941 400 39 40 24 17 0.0018 0.0021 200
5 6 60 20 0.8190 0.7070 400 40 41 1.2 1 0.7283 0.8509 200
6 7 200 100 0.1872 0.6188 300 41 42 0 0 0.31 0.3623 200
7 8 200 100 1.7117 1.2357 300 42 43 6 4.3 0.041 0.0478 200
8 9 60 20 1.0299 0.7400 200 43 44 0 0 0.0092 0.0116 200
9 10 60 20 1.0440 0.7400 200 44 45 39.22 26.3 0.1089 0.1373 200
10 11 45 30 0.1967 0.0651 200 45 46 39.22 26.3 0.0009 0.0012 200
11 12 60 35 0.3744 0.1237 200 4 47 0 0 0.0034 0.0084 300
12 13 60 35 1.4680 1.1549 200 47 48 79 56.4 0.0851 0.2083 300
13 14 120 80 0.5416 0.7129 200 48 49 384.7 274.5 0.2898 0.7091 300
14 15 60 10 0.5909 0.5260 200 49 50 384.7 274.5 0.0822 0.2011 300
15 16 60 20 0.7462 0.5449 200 8 51 40.5 28.3 0.0928 0.0473 200
16 17 60 20 1.2889 1.7210 200 51 52 3.6 2.7 0.331 0.1114 200
17 18 90 40 0.7320 0.5739 200 9 53 4.35 3.5 0.174 0.0886 300
2 19 90 40 0.1640 0.1564 200 53 54 26.4 19 0.203 0.1034 300
19 20 90 40 1.5042 1.3555 200 54 55 24 17.2 0.2842 0.1447 300
20 21 90 40 0.4095 0.4784 200 55 56 0 0 0.2813 0.1433 300
21 22 90 40 0.7089 0.9373 200 56 57 0 0 1.59 0.5337 300
3 23 90 50 0.4512 0.3084 200 57 58 0 0 0.7837 0.263 300
23 24 420 200 0.8980 0.7091 200 58 59 100 72 0.3042 0.1006 300
24 25 420 200 0.8959 0.7010 200 59 60 0 0 0.3861 0.1172 300
6 26 60 25 0.2031 0.1034 300 60 61 1244 888 0.5075 0.2585 300
26 27 60 25 0.2842 0.1447 300 61 62 32 23 0.0974 0.0496 300
27 28 60 20 1.0589 0.9338 300 62 63 0 0 0.145 0.0738 300
28 29 120 70 0.8043 0.7006 200 63 64 227 162 0.7105 0.3619 300
29 30 200 600 0.5074 0.2585 200 64 65 59 42 1.041 0.5302 300
30 31 150 70 0.9745 0.9629 200 11 66 18 13 0.2012 0.0611 200
31 32 210 100 0.3105 0.3619 200 66 67 18 13 0.0047 0.0014 200
32 33 60 40 0.3411 0.5302 200 12 68 28 20 0.7394 0.2444 200
68 69 28 20 0.0047 0.0016 200
Imax is maximum line current limit in Amperes, given in Ref. [54]. SB ¼ 100 MVA;
VB ¼ 12.66 kV. Imax is maximum line current limit in Amperes, SB ¼ 100 MVA, VB ¼ 12.66 kV.
 M.M. Aman et al. / Energy 66 (2014) 202e215 215

References [28] Moradi MH, Abedini M. A combination of genetic algorithm and particle
swarm optimization for optimal DG location and sizing in distribution sys-
tems. Int J Electr Power Energy Syst 2012;34(1):66e74.
[1] Ackermann T, Andersson G, Söder L. Distributed generation: a definition.
[29] Borges CLT, Falcão DM. Optimal distributed generation allocation for reli-
Electr Power Syst Res 2001;57(3):195e204.
ability, losses, and voltage improvement. Int J Electr Power Energy Syst
[2] Chicco G, Mancarella P. Distributed multi-generation: a comprehensive view.
2006;28(6):413e20.
Renew Sustain Energy Rev 2009;13(3):535e51.
[30] Quezada VM, Abbad JR, Roman TGS. Assessment of energy distribution losses
[3] Chiradeja P, Ramakumar R. An approach to quantify the technical benefits of
for increasing penetration of distributed generation. Power Syst IEEE Trans
distributed generation. IEEE Trans Energy Convers 2004;19(4):764e73.
2006;21(2):533e40.
[4] Bayod-Rújula AA. Future development of the electricity systems with
[31] Hemdan NGA, Kurrat M. Distributed generation location and capacity effect
distributed generation. Energy 2009;34(3):377e83.
on voltage stability of distribution networks. In: Conference distributed
[5] Zangiabadi M, Feuillet R, Lesani H, Hadj-Said N, Kvaløy JT. Assessing the
generation location and capacity effect on voltage stability of distribution
performance and benefits of customer distributed generation developers
networks. p. 1e5.
under uncertainties. Energy 2011;36(3):1703e12.
[32] Wan H, McCalley JD, Vittal V. Increasing thermal rating by risk analysis. IEEE
[6] Soroudi A, Ehsan M. A distribution network expansion planning model
Trans Power Syst 1999;14(3):815e28.
considering distributed generation options and techno-economical issues.
[33] Prada RB, Souza LJ. Voltage stability and thermal limit: constraints on the
Energy 2010;35(8):3364e74.
maximum loading of electrical energy distribution feeders. IEE Proc Gener
[7] Bracco S, Dentici G, Siri S. Economic and environmental optimization model
Transm Distrib 1998;145(5):573e7.
for the design and the operation of a combined heat and power distributed
[34] Milano F. Continuation power flow analysis. Power system modelling and
generation system in an urban area. Energy 2013;55(0):1014e24.
scripting; 2010. pp. 103e30.
[8] Celli G, Ghiani E, Mocci S, Pilo F. A multiobjective evolutionary algorithm for
[35] Mahmud MA, Hossain MJ, Pota HR, Nasiruzzaman ABM. Voltage control of
the sizing and siting of distributed generation. IEEE Trans Power Syst
distribution networks with distributed generation using reactive power
2005;20(2):750e7.
compensation. In: Conference voltage control of distribution networks with
[9] Puttgen HB, Macgregor PR, Lambert FC. Distributed generation: semantic hype
distributed generation using reactive power compensation. p. 985e90.
or the dawn of a new era? IEEE Power Energy Mag 2003;1(1):22e9.
[36] Turitsyn K, Sulc P, Backhaus S, Chertkov M. Options for control of reactive
[10] Soroudi A, Ehsan M, Caire R, Hadjsaid N. Possibilistic evaluation of distributed
power by distributed photovoltaic generators. Proc IEEE 2011;99(6):1063e73.
generations impacts on distribution networks. IEEE Trans Power Syst
[37] Ellis A, Nelson R, Von Engeln E, Walling R, MacDowell J, Casey L, et al.
2011;26(4):2293e301.
Reactive power performance requirements for wind and solar plants. In:
[11] Soroudi A, Ehsan M, Caire R, Hadjsaid N. Hybrid immune-genetic algorithm
Conference reactive power performance requirements for wind and solar
method for benefit maximisation of distribution network operators and
plants. IEEE. p. 1e8.
distributed generation owners in a deregulated environment. IET Gener
[38] Xu L, Cartwright P. Direct active and reactive power control of DFIG for wind
Transm Distrib 2011;5(9):961e72.
energy generation. IEEE Trans Energy Convers 2006;21(3):750e8.
[12] Soroudi A, Caire R, Hadjsaid N, Ehsan M. Probabilistic dynamic multi-objective
[39] Tang Y, Xu L. A flexible active and reactive power control strategy for a var-
model for renewable and non-renewable distributed generation planning. IET
iable speed constant frequency generating system. IEEE Trans Power Electron
Gener Transm Distrib 2011;5(11):1173e82.
1995;10(4):472e8.
[13] Mohammadi-Ivatloo B, Zareipour H, Amjady N, Ehsan M. Application of
[40] Liew SN, Strbac G. Maximising penetration of wind generation in existing
information-gap decision theory to risk-constrained self-scheduling of Gen-
distribution networks. IEE Proc Gener Transm Distrib 2002;149(3):256e62.
Cos. Power Syst IEEE Trans 2013;28(2):1093e102.
[41] Ghosh S, Ghoshal SP, Ghosh S. Optimal sizing and placement of distributed
[14] Soroudi A, Ehsan M. IGDT based robust decision making tool for DNOs in load
generation in a network system. Int J Electr Power Energy Syst 2010;32(8):
procurement under severe uncertainty. IEEE Trans Smart Grid 2013;4(2):886e95.
849e56.
[15] Soroudi A. Robust optimization based self scheduling of hydro-thermal Genco
[42] Abou El-Ela AA, Allam SM, MM S. Maximal optimal benefits of distributed
in smart grids. Energy 2013;61(0):262e71.
generation using genetic algorithms. Electr Power Syst Res 2010;80:869e77.
[16] Duong Quoc H, Mithulananthan N. Multiple distributed generator placement
[43] Doagou-Mojarrad H, Gharehpetian GB, Rastegar H, Olamaei J. Optimal place-
in primary distribution networks for loss reduction. IEEE Trans Ind Electron
ment and sizing of DG (distributed generation) units in distribution networks
2013;60(4):1700e8.
by novel hybrid evolutionary algorithm. Energy 2013;54(0):129e38.
[17] Nguyen Cong H, Mithulananthan N, Bansal RC. Location and sizing of
[44] Moravej Z, Akhlaghi A. A novel approach based on cuckoo search for DG
distributed generation units for loadability enhancement in primary feeder.
allocation in distribution network. Int J Electr Power Energy Syst 2013;44(1):
IEEE Syst J 2013;7(4):797e806.
672e9.
[18] Caisheng W, Nehrir MH. Analytical approaches for optimal placement of
[45] Venkatesh B, Ranjan R, Gooi H. Optimal reconfiguration of radial distribu-
distributed generation sources in power systems. Power Syst IEEE Trans
tion systems to maximize loadability. IEEE Trans Power Syst 2004;19(1):
2004;19(4):2068e76.
260e6.
[19] Acharya N, Mahat P, Mithulananthan N. An analytical approach for DG allo-
[46] Kennedy J, Eberhart R. Particle swarm optimization. Conference Particle
cation in primary distribution network. Int J Electr Power Energy Syst
Swarm Optim;4. p. 1942e8.
2006;28(10):669e78.
[47] Karaboga D. An idea based on honey bee swarm for numerical optimization.
[20] Gözel T, Hocaoglu MH. An analytical method for the sizing and siting of distrib-
Techn Rep TR06. Erciyes: Erciyes Univ Press; 2005.
uted generators in radial systems. Electr Power Syst Res 2009;79(6):912e8.
[48] Gözel T, Eminoglu U, Hocaoglu MH. A tool for voltage stability and optimi-
[21] Hedayati H, Nabaviniaki S, Akbarimajd A. A method for placement of DG units
zation (VS&OP) in radial distribution systems using matlab graphical user
in distribution networks. IEEE Trans Power Deliv 2008;23(3):1620e8.
interface (GUI). Simul Model Pract Theory 2008;16(5):505e18.
[22] Ettehadi M, Ghasemi H, Vaez-Zadeh S. Voltage stability-based DG placement
[49] Zhu JZ. Optimal reconfiguration of electrical distribution network using the
in distribution networks. IEEE Trans Power Deliv 2013;28(1):171e8.
refined genetic algorithm. Electr Power Syst Res 2002;62(1):37e42.
[23] Aman MM, Jasmon GB, Mokhlis H, Bakar AHA. Optimal placement and sizing
[50] Baran ME, Wu FF. Network reconfiguration in distribution systems for loss
of a DG based on a new power stability index and line losses. Int J Electr Power
reduction and load balancing. IEEE Trans Power Deliv 1989;4(2):1401e7.
Energy Syst 2012;43(1):1296e304.
[51] Savier J, Das D. Impact of network reconfiguration on loss allocation of radial
[24] Al Abri RS, El-Saadany EF, Atwa YM. Optimal placement and sizing method to
distribution systems. IEEE Trans Power Deliv 2007;22(4):2473e80.
improve the voltage stability margin in a distribution system using distrib-
[52] Thukaram D, Wijekoon Banda H, Jerome J. A robust three phase power flow
uted generation. Power Syst IEEE Trans 2013;28(1):326e34.
algorithm for radial distribution systems. Electr Power Syst Res 1999;50(3):
[25] Alonso M, Amaris H. Voltage stability in distribution networks with DG. In:
227e36.
IEEE PowerTech conference 2009. pp. 1e6. Bucharest.
[53] Jen-Hao T. A direct approach for distribution system load flow solutions. IEEE
[26] Aman MM, Jasmon GB, Bakar AHA, Mokhlis H. A new approach for optimum
Trans Power Deliv 2003;18(3):882e7.
DG placement and sizing based on voltage stability maximization and mini-
[54] Aman MM, Jasmon GB, Bakar AHA, Mokhlis H. Optimum network reconfigu-
mization of power losses. Energy Convers Manag 2013;70(0):202e10.
ration based on maximization of system loadability using continuation power
[27] Mithulananthan N, Oo T. Distributed generator placement to maximize the
flow theorem. Int J Electr Power Energy Syst 2014;54(0):123e33.
loadability of a distribution system. Int J Electr Eng Educ 2006;43(2):107e18.