Vaiju J.

Electrical Systems 113-1 (2017): 115-130

Kalkhambkar 1*, JES
Bhanu Rawat 2, Regular paper Journal of
Rajesh Kumar 1, Electrical
Rohit Bhakar 1 Optimal Allocation of Renewable Systems
Energy Sources for Energy Loss
Minimization
Optimal allocation of renewable distributed generation (RDG), i.e., solar and the wind
in a distribution system becomes challenging due to intermittent generation and
uncertainty of loads. This paper proposes an optimal allocation methodology for single
and hybrid RDGs for energy loss minimization. The deterministic generation-load model
integrated with optimal power flow provides optimal solutions for single and hybrid
RDG. Considering the complexity of the proposed nonlinear, constrained
optimization problem, it is solved by a robust and high performance meta-heuristic,
Symbiotic Organisms Search (SOS) algorithm. Results obtained from SOS algorithm
offer optimal solutions than Genetic Algorithm (GA), Particle Swarm Optimization
(PSO) and Firefly Algorithm (FFA). Economic analysis is carried out to quantify the
economic benefits of energy loss minimization over the life span of RDGs.

Keywords: Optimal allocation, renewable distributed generation, loss minimization, symbiotic

organisms search.
Article history: Received 17 May 2016, Accepted 19 January 2017

1. Introduction

Electric power generation has entered into a new era of renewable energy generation.
Solar and the wind are the most promising renewable energy technologies, with the
potential to meet future energy demand in a sustainable way [1]. Hence, planning of
renewable sources in distribution networks is the essential task. Optimally placed and sized
solar DG (SDG) or wind DG (WDG) in distribution system reduces energy losses and
improves voltage profile [2, 3]. Hybrid DG (HDG) that is a combined employment of
energy sources enhance the utilization of energy sources and provide consistent energy
generation [4]. Optimal allocation of such HDG offers significant loss reduction [5].
Optimal allocation of RDG in a distribution system to maximize their potential is a
challenging issue due to their intermittency [6]. Also, RDG connection at non-optimal
places or inappropriate sizing increases power losses. Hence, it is essential to identify
optimal RDG allocation in a distribution network [7].
Various methodologies aim to optimally allocate RDG for minimizing losses in
distribution systems. Analytical methods result in increased computational efforts with sub-
optimal solutions [8]. Weighing factors method offers optimal solutions; however the
quality of the solution is affected by the choice of weighing factors [9]. Probabilistic
methods are formulated with large ‘generation-load’ states and these states formation
involves clustering methods. Clustering obtained by the iterative process may converge to
local minima producing sub-optimal solutions [10]. Intermittent renewable generation is
also modeled probabilistically by convolution, but a large number of states involved in
convolution and ‘generation-load state’ model affects the quality of solutions. This
underscores the necessity of planning problem formulations for the power system to
provide optimal solutions [11]. Thus, RDG allocation problems having large search space

*
Corresponding author: Vaiju Kalkhambkar, Centre for Energy and Environment Malaviya National Institute of
Technology Jaipur India-302017, E-mail: [email protected]
1
Centre for Energy and Environment , Malaviya National Institute of Technology Jaipur India - 302017
2
Department of Electrical Engineering, Malaviya National Institute of Technology Jaipur India - 302017

Copyright © JES 2017 on-line : journal/esrgroups.org/jes

 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

needs an allocation methodology to explore the search space for obtaining optimal
solutions.
The optimization algorithms used for optimal placement and sizing of DG mainly
includes genetic algorithm (GA) [12], particle swarm optimization (PSO) [13, 14],
Simulated Annealing (SA) [15], artificial bee colony (ABC) [16] and firefly algorithm
[17]. The hybrid combination of algorithms such as ant colony optimization (ACO) and
artificial bee colony (ABC) [18], (PSO) and gravitational search algorithm (GSA) [19] are
also used for optimal allocation of DGs.
This paper develops an optimal solution methodology for allocation of RDG to
minimize energy losses. A deterministic generation-load model is proposed from
probabilistic generation model and IEEE-RTS load model. The proposed deterministic
generation-load model, integrated with optimal power flow, provides optimal solutions.
Then a deterministic generation-load model is proposed for HDG allocation using the
expected generation of SDG and WDG. This nonlinear, constrained optimization
problem is solved with a robust and high-performance symbiotic organisms search (SOS)
algorithm. Comparative results from standard algorithms like GA, PSO, and FFA highlight
the efficiency of SOS algorithm to offer better results for loss minimization. In the end,
an economic model is included to get an insight of benefit of energy loss minimization by
optimal allocation of RDG over the life period of renewable DG.
The paper is organized as follows. Section 2, presents the system modeling. The detail
problem formulation is presented in section 3. The optimization algorithms, i.e., SOS is
briefly discussed in section 4. Economic analysis to quantify the economic benefits of
energy loss minimization is presented in Section 5. A case study and results are discussed
in section 6.

2. System Modeling

2.1 Historical Data Processing

Historical data is used to obtain predicted RDG generation using probability

distribution functions (pdfs). To obtain pdfs with accuracy, the whole year (Y) is
segmented into three seasons (i.e. summer, monsoon and winter) considering average
monthly temperatures. A day in the season is assumed to represent the whole season
and each day is further divided into hourly time segments. Considering each month ‘M’
to have ‘D’ days, each time segment has Y ∗ M ∗ D solar irradiance and wind speed
data points (i.e., ‘Y’ years * ‘M’ months per season * ‘D’ days per month).

2.2 Solar Power

The solar power modeling is done to obtain expected generation from it. Randomness
of solar irradiance is expressed by beta probability distribution function (i.e., beta pdf Fβ
) [5]. The Fβ indicates the probability or fraction of time for which the solar irradiance is
at a given irradiance ‘s’ . The beta pdf Fβ (s) is as given below
 Γ(α + β ) (α −1)
 s (1 − s )( β −1) 0 ≤ s ≤ 1, α , β ≥ 0
Fβ ( s ) =  Γ(α )Γ( β ) (1)

 0 otherwise
where, Γ is gamma function; s is random variable of solar irradiance in kW/m 2 ; α
and β are the shape parameters of beta distribution function. The cumulative distribution

116
 J. Electrical Systems 13-1 (2017): 115-130

function (cdf) can be used to estimate the time for which solar irradiances ‘ s ’ is within a
certain irradiance interval (e.g. s1 and s2).
Γ(α + β ) t (α −1)
f β ( s) = ∫
Γ(α )Γ( β ) 0
s (1 − s)( β −1) dt (2)

Probability of solar irradiance being between s1 and s2 can be obtained as given below.

f β ( s1 < s < s 2 ) = f β ( s 2 ) − f β ( s1 ) (3)

The power output Po (s) of PV cell at any state y with cell current Iy , cell voltage Vy , N
number of cell and fill factor ff is given as below.

Po ( s ) = N ∗ f f ∗ V y ∗ I y (4)

The total expected output power PSG at any hour can be given as,

∞
PSG = ∫ Po ( s ) * f β ( s ).ds (5)
0
2.3 Wind Power

Wind generation is modeled with Weibull pdf due to its simplicity. Weibull pdf Fw (v)
for the wind turbine is as given below [20].

k  v 
k −1
  v k 
Fw (v ) =   exp −    (6 )
c  c    c  
Here, k, c and v are the shape index, scale index and wind speed respectively.
Rayleigh distribution is a simplified case of the Weibull distribution where ‘ k ‘ is
assumed as 2 and c is approximated as 1.128 vm , where vm is the mean wind speed.
Under the Rayleigh based approach, pdf and cdf of wind velocity is given by equation (7)
and (8) respectively.
 2v    v 2 
Fw (v) =  2  exp  −    (7)
c    c  

 π  v 
−  
 4  vm  
f w (v ) = 1 − e (8)

Probability of wind speed being between v1 and v2 can be obtained as,

f w (v)(v1 < v < v2 ) = f w (v2 ) − f w (v1 ) (9)

The power delivered by a wind turbine Po (w) is represented by its power curve that
gives a relation between the wind speed and power as below.

117
 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

 0 0 ≤ vav ≤ vci

 P ∗  vav − vci  v ≤ v ≤ v
 r   ci av r
Po ( w) =   vr − vci  (10)

 Pr vr ≤ vav ≤ vco
 0 vco ≤ vav

Where, vci , vco , vr, vav and Pr represent the cut-in speed, cut-off speed, rated speed,
average speed and output power of wind turbine respectively. The total expected wind
power PWG at any time interval is given as,

∞
PWG = ∫ Po ( w) * f w (v ).dv (11)
0

2.4 Load Modeling

Load considered in this system is hourly peak load, expressed as a percentage of daily
peak load. The load profile follows IEEE-RTS system [21]. Hourly loads for three
different seasons i.e. summer, monsoon and winter are considered

3. Problem Formulation

3.1. Optimal Sizing and Location of Renewable DG

Once the number of RDGs to provide expected power (i.e., forecasted output power) at
optimal location are obtained by optimization, rated optimal size of RDG can be obtained
as given below.

3.1.1 Solar DG

The expected (PSDGE ) and rated (PSDG ) optimal size of SDG at optimal location can
be given by equations (12) and (13) respectively

PSDGE ,i = c s ,i PSG ∀i ∈ D (12)

PSDG ,i = cs ,i PSDGR ∀i ∈ D (13)

Where Cs,i, PSG, PSDGR and D give the integer variables representing number of solar
panels at i th bus, expected solar PV generation, rating of solar PV module and candidate
bus respectively

3.1.2 Wind DG

The expected (PWDGE ) and rated (PWDG ) optimal size of WDG at optimal location can
be given by equation (14) and (15) respectively.

118
 J. Electrical Systems 13-1 (2017): 115-130

PWDGE ,i = c w ,i PWG ∀i ∈ D (14)

PWDG ,i = cw ,i PWDGR ∀i ∈ D (15)
where, cw,i, PWG, and PWDGR give the integer variables representing number of wind
turbines at i th bus, expected wind power and rating of wind turbine respectively.

3.1.3 Hybrid DG

The expected (PHDGE ) and rated (PSDGH , PWDGH ) optimal size of HDG at optimal
location are given by equation (16) and (17-18) respectively.

PHDGE ,i = c sh ,i PSG + c wh ,i PWG ∀i ∈ D (16)

PSDGH ,i = csh ,i PSDGR ∀i ∈ D (17)
PWDGH ,i = c wh ,i PWDGR ∀i ∈ D (18)
where, csh,i and cwh,i , give the integer variables representing the number of solar panels
and wind generators at ith bus for hybrid combination.

3.2 Distribution System Power Flow

The expected hourly generation and load are used for load flow calculation. The
power losses Ploss can be calculated with backward/forward (BF) sweep method [22].
The system is assumed balanced and represented on per phase basis. Load current ILi of
node i can be expressed as,
Pi − Q i
I Li = (1 9 )
Vi∗

Where, Pi and Qi are active and reactive load at node i, respectively. Current through
branch i i.e., Ii will be load current of node i plus the branch currents connected to this
line,
Pi − Qi
Ii = + ∑I j (20)
Vi ∗ j∈βi

Where βi is the set of branches connected to node i. Thus for calculating branch currents,
all branches connected to the node must be determined. Sum of Ii 2 r losses in each
branch gives the total real power loss i.e., Ploss of the system. Considering renewable
DG power at node i, active power Pi gets modified to (Pi-PRDGE,i ).

3.3 Objective Function

The objective is to minimize annual energy losses.

 120 24 120 24 120 24


F = min  ∑∑Ploss ,summer ,t + ∑∑Ploss ,monsoon,t + ∑∑Ploss , winter ,t  (21)
 D =1 t =1 D =1 t =1 D =1 t =1 

119
 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

Here each season has 120 days (i.e., 30 days * 4 months). Annual energy losses are
obtained by summation of losses of three seasons.

3.4 Constraints

3.4.1 Active and reactive power balance

Assuming RDG sources are operating at unity power factor and supplying only active
power, the active and reactive powers balance are given below.
n n nb
PG1,t + ∑PRDGEi ,t − ∑Pi ,t − ∑Ploss ,b ,t = 0 (22)
i =1 i =1 b =1

n nb
QG1,t − ∑ Qi ,t − ∑ Qloss ,b ,t = 0 (23)
i =1 b =1

Where, PG1 and QG1 are injected active and reactive power at substation. PRDGE i is
RDG’s forecasted power. Pi and Qi are total active and reactive loads at ith node
respectively. PLoss,b and QLoss,b are active and reactive power losses at branch b,
respectively. nb is the total number of branches.

3.4.2 Feeder current

With the placement of renewable DG, feeder current is restricted to maximum feeder
capacity.
0 ≤ I k ,ij ≤ I ij ∀i, j , k (24)
max

3.4.3 Maximum penetration of renewable DG

RDG penetration is considered as k% of system’s total peak load (PLmax ) for single
RDG as well as HDG. Maximum penetration limits for SDG, WDG and HDG are given
by equations (25), (26) and (27) respectively. The summation of power injected by all
RDGs should be equal to the allowed maximum penetration of RDGs.
n

∑c
i =1
s ,i × PSG = k × PLmax ∀i ∈ D (25)
n

∑c
i =1
w ,i × PWG = k × PLmax ∀i ∈ D (26)
n n

∑c
i =1
s ,i × PSG + ∑cw ,i × PWG = k × PLmax
i =1
∀i ∈ D (27)
RDG penetration affects the design and operation of distribution system and may
increase cost of distribution system and consumer payments As per IEA study, 25% to
40 % penetration of renewable energy sources put a little additional cost on the system in
the long run, and hence is an acceptable penetration limit [23].

4. Optimization algorithm

120
 J. Electrical Systems 13-1 (2017): 115-130

RDG allocation is a complex, mixed integer, non-linear, constrained optimization

problem. Heuristics approaches are efficient in finding global optima with higher success
rates for better solutions [24]. Considering the exponentially large search space and
complexity of this problem, a robust and high performance algorithm called SOS is
proposed in this paper. In SOS, new solutions are generated by imitating the biological
interaction between two organisms of the ecosystem i.e., mutualism, commensalism, and
parasitism. SOS algorithm is briefly explained here [25].

4.1.1 Mutualism Phase

The mutualism phase of SOS mimics a mutualistic relationships between two organisms
e.g. bees and flowers. If Xi is an organism matched to the member of the ecosystem,
then organism Xj is randomly selected from the ecosystem that interacts with Xi . New
candidate solutions for Xi and Xj are calculated based on the mutualistic symbiosis.
The benefit factors (BF1 , BF2 ) represent the level of benefit to each organism. A vector
called ‘Mutual Vector’ represents the relationship between organism Xi and Xj . The
mutualistic efforts for their survival is given by (X best- M utual Vector ∗ BF1 ). Xbest
represents the highest degree of adaptation.
Algorithm 1: Symbiotic Organisms Search (SOS)
Ecosystem Initialization i Number of organisms (eco size), initial ecosystem,
termination criteria, num iter = 0 num fit eval= 0, max iter, max fit eval.
Go to the next iteration
Identify the best solution Xbest
Mutualism Phase
i. Select one organism randomly, Xj, where Xj ≠ Xi
ii. Determine mutual relationship vector (Mutual V ector) and benefit factor (BF)
iii. Mutual Vector = (Xi + Xj) =2
BF1 = random number either 1 or 2; BF2 = random number either 1 or 2
iv. Modify organism Xi and Xj based on their mutual relationship
v. Xi new = Xi + rand (0; 1) _ (Xbest -Mutual V ector _ BF1)
Xj new = Xj + rand (0; 1) _ (Xbest -Mutual V ector _ BF2)
Commensalism Phase
i. Select one organism randomly, Xj, where Xj ≠Xi
ii. Modify organism Xi with the assist of organism Xj
iii. Xi new = Xi + rand (-1; 1) _ (Xbest - Xj)
iv. Select Fitter organisms as solutions for the next iteration
Parasitism Phase
i. Select one organism randomly, Xj, where Xj ≠ Xi
ii. Create a Parasite (Parasite_ Vector) from Organism Xi
iii. Select Fitter organisms as solutions
Go to step 2 if the current Xi is not the last member of the ecosystem otherwise proceed
to next step Stop if one of the termination criteria is reached otherwise return to step 2
and start the next iteration

4.1.2 Commensalism Phase

This is observed in remora fish and shark. The remora attaches to the shark and eats
food leftovers. Remora receives benefit, while shark is unaffected by remora fish. Here an
organism Xj , is randomly selected to interact with Xi Organism Xi attempts to benefit
from the interaction. The organism Xj neither benefits nor suffers. New candidate
solution of Xi is calculated according to the commensal symbiosis between them.

4.1.3 Parasitism Phase

121
 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

This is observed in mosquito and human body. Here a parasite vector is created by
duplicating organism Xi. The randomly selected organism Xj serves as a host to the
parasite vector. Parasite vector tries to replace Xj in the ecosystem. Parasite Vector with
better fitness kills the organism Xj and fixes its position in the ecosystem. If the fitness
value of Xj is better, it will have immunity from the parasite and Parasite Vector will be
discarded.

Fig. 1: Optimal allocation of RDG using SOS

Flow chart for optimal allocation using SOS is shown in Fig.1. Initialization of the
algorithm requires system data, bus data, load data, expected generation of sources, no. of
organism (eco size) and termination criteria. Fitness of each organism in the ecosystem
is checked by load flow, by passing them through all phases of symbiotic relations. The
process is repeated for all elements of ecosystem (i.e. eco size) till termination criteria is
attained. Similar procedure is applied for the Genetic Algorithm (GA) [26], Particle
Swarm Optimization (PSO) [27] and Firefly Algorithm (FFA) [28].

5. Economic analysis

Economic analysis is carried out to show the economic benefits of energy loss
minimization. The costs includes investment cost and operation & maintenance cost. The
benefits are from energy generation and loss minimization due to optimal allocation. The
total investment cost CI and operation & maintenance (O&M) cost COM is given as below,
C I = PI S G C IS +P IW G C IW (28)

122
 J. Electrical Systems 13-1 (2017): 115-130

j
NY
 1 + Rinf 
COM = ∑( PISG COMS + PIWG COMW )   (29)
j =1  1 + Rint 

Where,CIS investment cost of SDG in Rs./MW ; CIW investment cost of WDG in

Rs./MW; PISG total installed capacity of SDG MW ; PIWG total installed capacity of
WDG MW; COMS O&M cost of SDG Rs./MW; COMW O&M cost of WDG Rs./MW;
NY total number of years; Rinf inflation rate; Rint interest rate or discount rate; Revenue
can be obtained by the generation of renewable energy. The total benefit by the production
of renewable generation can be obtained as below,
j
NY
 1 + Rinf 
BEG = ∑( PSG CSE + PWG CWE )   (30)
j =1  1 + Rint 
Where, CS E cost of solar energy Rs./kW h; CWE cost of wind energy Rs./kW h; With
the optimal allocation of RDG significant loss minimization is obtained. The cost of this
saved energy due to loss minimization can be given as below.
j
 1 + Rinf 
NY

BEL = ∑( PLS CSEL + PLW CWEL )   (31)

j =1  1 + Rint 

Where, PLS annual energy loss minimization by SDG M W h; PLW annual energy loss
minimization by WDG M W h; CS E L cost of energy losses for SDG Rs./kW h; CWEL
cost of energy losses for WDG Rs./kW h;
The various indices used for economic analysis of RDGs are mainly net present
value (NPV), aggregate benefit cost ratio (ABCR) and discounted payback period
(DPBP). NPV is the net value of all benefits (i.e. cash inflows) and costs (i.e. cash
outflows) of the project, discounted back to the beginning of the investment.

NPV = cash inflows − cash outflows

The project with higher NPV is most preferable.

ABCR is the ratio of the accumulated present value of all the benefits to the
accumulated present value of all costs, including the initial investment.

Present value of benefits

ABCR = (33)
Present value of costs
PBP indicates the minimum period over which the investment for the project is
recovered considering the time value of money. At payback period,

Present value of costs = present value of benefits (34)

The project with lower pay back period is most preferable

123
 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

6. Results and discussions

6.1 System under Study

The proposed methodology is applied to a 34 bus test system shown in Fig.2 [29]. The
system load is modelled as specified in IEEE-RTS system. Hourly solar and wind data of 5
years is taken for Satara (Longitude:74.05 E Latitude:17.75 N) Maharashtra state, India.

Fig. 2: 34 bus system

To get hourly predicted RDG generation more accurately, the solar irradiance data and
the wind speed data is analyzed on seasonal basis. The whole year is segmented into three
seasons and hence each season has 4 months. Thus in a 5 years span, each season has a
total of 20 months (i.e. 5 years * 4 months each season) and 600 days (i.e. 20 months *
30 days per month). Considering a typical hour of each day (say 9 am), we have 600
solar irradiance and wind speed data for this typical hour (i.e. we get 600 time slots of 9
am and hence 600 corresponding data).The solar PV module used is 325 W [30] and wind
turbine is 100 kW [31]. These specifications of SPV module and wind turbine are shown
in Appendix A. The cdfs are generated selecting different states per hour for β cdf and
Rayleigh cdf respectively. The states are selected based on the maximum solar irradiance
and maximum wind speed. The values alues of various parameters, probability and output for
different states of a typical hour (i.e. 8 am) is shown in Appendix B. Solar and wind power
is available for 11 and 24 Hrs respectively.
respect Total system peak load is 5.0 MVA. RDG
penetration is considered as 2 MW i.e., 40% of system’s total peak load (PLmax ) at unity
power factor.

6.2 Optimal location and sizing

The possible candidate bases are randomly selected and the size of RDG can be
increased in a discrete step of 100 kW
W. Maximum 5 RDGs (i.e. 500 kW) can be placed
optimally on any of the candidate buses.
uses. This maximum number of RDG at any bus can
be limited by the maximum current capacity of the feeder. Here the maximum current
capacity is 50 A. The selected ten candidate buses are {12,15,18,22,25,27,28,29,30,32}.

124
 J. Electrical Systems 13-1 (2017): 115-130

In practice, candidate buses are selected depending upon physical availability of space for
RDG and potential for RDG generation. Thereafter, optimal locations on selected buses
can be found by optimization. Optimal allocation of RDG is discussed in the following
sub-sections.

6.2.1 Solar DG

Table 1. shows the optimal number of SDG with it’s rating, optimal locations,
annual energy losses and percentage loss minimization. System’s annual energy losses
without any RDG (i.e. base case) are 1078.20 MWh. SDG’s minimum size is considered
as 100 kW. A SDG of 100 kW requires 308 rated solar modules. Thus 20 SDGs have to
be optimally allocated to obtain 2 MW penetration. About 10-11 % loss minimization is
obtained by optimal SDG placement. GA and FFA places RDG on nine candidate buses,
SOS on 8 and PSO only on 5 candidate buses.
.
TABLE I: Optimal allocation of solar DG

SOS PSO GA FFA

Location
(Bus no) DG Size DG Size DG Size DG Size
nos. (kW) nos. (kW) nos. (kW) nos. (kW)
12 2 200 5 500 3 300 2 200
15 – – – – – – – –
18 1 100 – – 2 200 2 200
22 3 300 2 200 2 200 3 300
25 4 400 3 300 2 200 3 300
27 3 300 5 500 3 300 2 200
28 – – – – 2 200 1 100
29 1 100 – – 2 200 2 200
30 1 100 – – 1 100 2 200
32 5 500 5 500 3 300 3 300
Losses
(Mwh) 958.97 959.40 963.40 963.78

Fig. 3: Convergence plot for solar DG allocation

SOS provides improved loss minimization as compared to other optimization methods.

SOS places SDG of rating (kW) 200, 100, 300, 400, 300, 100, 100 and 500 on buses
{12,18,22,25,27,29,30,32} respectively. Annual energy losses obtained by SOS are
958.97 MWh and loss minimization is 11.06 %. PSO locates SDG of rating (kW) 500,
200, 300,500 and 500 on buses {12, 22, 25, 27, 32}. GA and FFA offer similar
placement and sizing with each other as shown in Table I. Convergence plot for SDG is
shown in Fig. 3. For all algorithms, the population size and maximum iteration count was
fixed to 30 and 250 respectively. SOS offers best fitness value (i.e. loss in kW) with slow
125
 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

convergence as compared to all other algorithms. FFA offers faster convergence but
weaker fitness value, as compared to other algorithms. GA and PSO offer convergence
and fitness value intermediate to FFA and SOS.

6.2.2 Wind DG

The minimum rating of wind turbine is considered as 100 kW, hence 20 wind turbines
need to be optimally allocated. SOS offers optimal result by placing WDG on eight
buses {12,18,22,25,27,28,29,30}. FFA gives least loss minimization of 28.68%, by
allocating WDG on all buses. Loss minimization by GA is 29.18 % and by PSO is 29.46
%. WDG allocation by GA and PSO is as shown in Table II. The convergence plot of the
algorithms for wind DG allocation are as shown in Fig. 4. SOS offers best fitness value
though converges slowly. FFA has faster convergence but weaker fitness. GA and PSO
have convergence and fitness value intermediate to FFA and SOS.

TABLE II: Optimal allocation of Wind DG

SOS PSO GA FFA
Location
(Bus no) DG Size DG Size DG Size DG Size
nos. (kW) nos. (kW) nos. (kW) nos. (kW)
12 2 200 4 400 3 300 3 300
15 – – 1 100 – – 1 100
18 1 100 1 100 1 100 1 100
22 2 200 2 200 3 300 3 300
25 1 100 3 300 1 100 2 200
27 5 500 3 300 4 400 3 300
28 2 200 1 100 1 100 1 100
29 2 200 – – 2 200 2 200
30 – – – – 2 200 1 100
32 5 500 5 500 3 300 3 300
Losses
(MWh) 759.3 760.6 763.55 768.97

Fig. 4: Convergence plot for wind DG allocation

6.2.3 Hybrid DG

The combined maximum penetration for HDG is considered as 2 MW. Thus SDG or
WDG can contribute maximum 50 % penetration. Maximum 10 SDG and 10 WDG can be
optimally allocated on the candidate buses. Optimal allocation results of Table III indicates
that a maximum of 22.43% loss minimization is obtained by SOS by placing 40% SDGs
and 60% WDGs. Thus it allocates total 800 kW of SDG on buses {12,25,30,32} and
1200 kW of WDG on buses {12,25,27,29,32}. PSO, GA and FFA optimally place 35 %
of SDG and 65 % of WDG. GA and FFA place WDG on all candidate buses.
Convergence plot of algorithms for HDG optimization are shown as Fig. 5. Fitness
value obtained by SOS is better, though its convergence is slow as compared to others.

126
 J. Electrical Systems 13-1 (2017): 115-130

GA gives faster convergence. FFA and PSO has convergence value in between the GA
and SOS.
Above results shows that the optimal solutions of loss minimization by SOS provides
significant economical benefits over the life span of the RDG. Also optimization results
of SOS are found to be more consistent. Hence SOS is more suitable for these type of
optimization problems though it has a slow convergence. Optimal results were obtained by
SOS after varying the candidate buses also (e.g. 15 buses, 20 buses and 25 buses).

TABLE III: Optimal allocation of hybrid RDG

Location SOS PSO GA FFA

(Bus no) Solar Wind Solar Wind Solar Wind Solar Wind
(kW) (kW) (kW) (kW) (kW) (kW) (kW) (kW)
12 300 200 100 200 100 300 100 100
15 – – – – – 100 100 100
18 – – 200 200 – 100 – 200
22 – – – 200 100 100 – 100
25 200 100 – 200 100 200 200 100
27 – 500 200 – 100 100 – 200
28 – – 100 100 – 100 100 100
29 – 200 – – 100 100 – 100
30 200 – 100 100 100 100 200 100
32 100 200 – 300 100 100 – 200
% share of
40 60 35 65 35 65 35 65
source
Losses
836.35 837.33 842.62 850.81
(MWh)

Fig. 5: Convergence plot for hybrid DG allocation

Thus SOS provide better solution for RDG allocation as compared to FFA, GA and
PSO irrespective of the number of buses in the investigated system.
6.3 Economic Study

Economic benefits of SDG, WDG and HDG is analyzed using NPV, ABCR and DPBP.
The investment cost for SDG and WDG in Rs./MWh is 58733000 and 61916000
respectively and the O&M cost in Rs./MWh is 1300000 and 1063000 respectively. The
inflation rate and discount rate is considered as Rs. 6.10 and Rs. 10.81. The cost of energy
for SDG and WDG in Rs./kWh is 6.86 and 6.58 respectively [32]. The cost of energy for
HDG is taken as average of costs of SDG and WDG.

127
 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

TABLE IV: Economic Analysis

Loss reduction NPV ABCR DPBP

DG
Solar RS. benefit
12204507 Rs. 156530715 2.0017 7.3
Wind RS. 31310571 Rs. 732363343 5.7081 2.7
Hybrid RS. 24250792 Rs. 502532492 4.2232 3.9

The cost of energy losses are 6.86 Rs./kWh for SDG, 6.58 Rs./kWh for WDG and
6.72 Rs./kWh for HDG. Table 4. shows the NPV, ACBR and DPBP for SDG, WDG and
HDG. The useful life of the renewable projects are considered as 20 years. Higher NPV
Rs. 732363343, higher ABCR 5.7081 and lowest DPBP 2.7 years is obtained for WDG.
Thus WDG is more economical than SDG and HDG. After observing NPV, ACBR and
DPBP for renewable DG, we can conclude that energy saving by loss minimization over
the life period of renewable DG is worth to cover capital and running cost as shown in
Table IV.

7. Conclusion

The proposed methodology provides optimal solutions for allocation of single DG as

well as hybrid DG. Variability of renewable sources and load is modeled to obtain
deterministic generation-load models. Optimal allocation of WDG achieves better loss
minimization. SOS provides a better solution than other algorithms, i.e., FFA, GA, and
PSO. Thus, SOS is suitable for this non-linear constrained RDG allocation problem.
Significant economic benefits are obtained by loss minimization with optimal allocation
over the life span of RDG. This optimal allocation technique can be applied to the
planning of RDGs.

References
[1] O. H. Mohammed, Y. Amirat, G. Feld, and M. Benbouzid, “Hybrid generation systems planning
expansion forecast state of the art review: Optimal design vs technical and economical constraints,”
Journal of Electrical Systems, vol. 12, no. 1, pp. 20–32, 2016.
[2] S. C. Reddy, P. V. N. Prasad, and A. J. Laxmi, “Hybrid approach for voltage sag mitigation in
distribution system by optimal location and size of distributed generation,” J. Electrical Systems, vol. 9,
no. 2, pp.150–166, 2013.
[3] K. Muttaqi, Le, M. Negnevitsky, and G. Ledwich, “An algebraic approach for determination of
DG parameters to support voltage profiles in radial distribution networks,” IEEE Trans on Smart Grid,
vol.5.3, pp. 1351–1360, 2014.
[4] M. Kamel, A. M. Azmy, A. A. El-Ela, and A. I. Shobair, “Optimal management and operation of a
hybrid energy system based on wind energy units,” Journal of Electrical Systems, vol. 9, no. 2, pp.
191–202, 2013.
[5] Y. Atwa, E. El-Saadany, M. Salama, and R. Seethapathy, “Optimal renewable resources mix for
distribution system energy loss minimization,” IEEE Transactions on Power Systems, vol. 25, no. 1, pp.
360–370, 2010.
[6] Sarkar and Ajjarapu, “MW resource assessment model for a hybrid energy conversion system with
wind and solar resources,” IEEE Trans on Sust. Ener., vol. 2, no. 4, pp. 383–391, 2011.
[7] W.-S. Tan, M. Y. Hassan, M. S. Majid, and H. Abdul Rahman, “Optimal distributed renewable
generation
planning: A review of different approaches,” Renewable and Sustainable Energy Reviews, vol. 18,
pp.626–645, 2013.
[8] D. Q. Hung, N. Mithulananthan, and Lee, “Optimal placement of dispatchable and Non dispatchable
RDG units in distribution networks for minimizing energy loss,” Int. Jour. Electrical Power & Energy
Sys., vol. 55, pp.179-186, 2014.
[9] M. Esmaili, E. C. Firozjaee, and Shayanfar, “Optimal placement of distributed generations
considering voltage stability and power losses with observing voltage- related constraints,” Applied
Energy, vol. 113, pp. 1252–
1260, 2014.

128
 J. Electrical Systems 13-1 (2017): 115-130

[10] Wu, Advances in K-means Clustering:A Data Mining Thinking. Spring., 2012. Seifi and M. S.
Sepasian, Electric power system planning: issues, algorithms and solutions. Springer Science & Business Media, 2011.
[11] H. Seifi and M. S. Sepasian, Electric power system planning: issues, algorithms and solutions.
Springer Science & Business Media, 2011.
[12] A. E.-S. A. Nafeh, “Optimal economical sizing of a PV-wind hybrid energy system using
genetic algorithm,” International Journal of Green Energy, vol. 8, no. 1, pp. 25–43, 2011.
[13] R. Syahputra, I. Soesanti, and M. Ashari, “Performance enhancement of distribution network with DG
integration using modified PSO algorithm.” Journal of Electrical Systems, vol. 12, no. 1, 2016.
[14] K. R. Guerriche and T. Bouktir, “Optimal allocation and sizing of distributed generation with
particle swarm optimization algorithm for loss reduction,” Revue des Sciences et de la Technologie,
vol. 6, no. 1, pp. 59–69, 2015.
[15] I. S. Kumar and P. K. Navuri, “An efficient method for optimal placement and sizing of multiple
distributed generators in a radial distribution systems,” Distr. Gener. & Alter. Energy Journal, vol. 27,
no. 3, pp. 52–71, 2012.
[16] H. Sebaa, K. R. Guerriche, and T. Bouktir, “Optimal sizing and placement of renewable energy
source in large scale power system using ABC technique in presence of UPFC,” Renewable and
Sustainable Energy Conference (IRSEC), 2014 International. IEEE, 2014, pp. 294–299.
[17] K. Nadhir, D. Chabane, and B. Tarek, “Firefly algorithm based energy loss minimization approach for
optimal sizing & placement of distributed generation,” Modeling, Simulation and Applied
Optimization (ICMSAO), 2013 5th International Conference on. IEEE, 2013, pp. 1–5.
[18] M. Kefayat, A. L. Ara, and S. N. Niaki, “A hybrid of ant colony optimization and artificial
bee colony algorithm for probabilistic optimal placement and sizing of distributed energy resources,”
Energy Conversion and Management, vol. 92, pp. 149–161, 2015.
[19] W. S. Tan, M. Y. Hassan, H. A. Rahman, M. P. Abdullah, and F. Hussin, “Multi-distributed
generation planning using hybrid particle swarm optimisation-gravitational search algorithm
including voltage rise issue,” IET Gen., Trans. & Distr., vol. 7, no. 9, pp. 929–942, 2013.
[20] M. Sathyajith, Wind energy: fundamentals, resource analysis and economics. Springer Science &
Business Media, 2006.
[21] S. PM, “IEEE reliability test system,” IEEE Transactions on Power Apparatus and Systems, no. 6,
pp.2047–2054, 1979.
[22] M. Hosseini, H. A. Shayanfar, and M. Fotuhi-Firuzabad, “Modeling of D-STATCOM in
distribution systems load flow,” Journal of Zhejiang University SCIENCE A, vol. 8, no. 10, pp.
1532–1542, 2007.
[23] IEA, The Power Transformation: Wind, Sun and the Economics of Flexible Power Systems.
International Energy Agency, 2014.
[24] F. Ugranlı and E. Karatepe, “Optimal wind turbine sizing to minimize energy loss,” International
Journal of Electrical Power & Energy Systems, vol. 53, pp. 656–663, 2013.
[25] M.-Y. Cheng and D. Prayogo, “Symbiotic organisms search: A new metaheuristic optimization
algorithm,” Comp.& Stru., vol. 139, pp. 98–112, 2014.
[26] [26] S. Sivanandam and S. Deepa, Genetic Algorithm Optimization Problems. Springer, 2008.
[27] X. Yu, X.-Y. Xiong, and Y.-w. Wu, “A PSO-based approach to optimal capacitor placement with
harmonic distortion consideration,” Electric Power Systems Research, vol. 71, no. 1, pp. 27–33,
2004.
[28] X.-S. Yang, “Firefly algorithms for multimodal optimization,” in Stochastic algorithms: foundations
and applications. Springer, 2009, pp. 169–178.
[29] M. Chis, M. Salama, and S. Jayaram, “Capacitor placement in distribution systems using heuristic
search strategies,” IEE Proceedings-Generation, Transmission and Distribution, vol. 144, no. 3, pp.
225–230, 1997.
[30] www.kyocerasolar.com/assets/001/5643.pdf.
[31] www.windenergysolutions.nl/wes100.
[32] http://www.cercind.gov.in

129
 Vaiju Kalkhambkar et al: Optimal Allocation of RE Sources for Energy Loss Minimization

APPENDIX A: Specifications of Solar Panel and Wind Turbine

TABLE A1: KD325GX-LFB solar panel TABLE A2: Wind turbine

STC NOCT Temp.Coeff. Rated Power 100 kW

VM P P (V) 40.3 36.2 -0.47 Cut-in wind speed 3 m/s
IM P P (A) 8.07 6.47 0.025 Rated wind speed 13 m/s
Voc (V) 49.7 45.5 -0.36 Cut-off wind speed 25 m/s
Isc (A) 8.69 7.04 0.060 Survival wind speed 60 m/s
.
APPENDIX B: Various parameters, probability and output of states

TABLE B1: Values of various parameters

Solar PV generation Wind generation

Parameter values Parameter values
µ 0. 378 k 2
ρ 0.077 Vm 8. 6380
α 33. 217 c 9. 7437
ᵦ 54. 584

TABLE B2: Probability and output power of TABLE B3 : Probability and output
states for solar DG power of states for wind DG
Po
υ1 υ2 υ ƒw (υ) P(ω)
S1 S2 S f β ( s ) Po (s) P (s) (ω)
0 1 0.5 0.0159 0 0.00
0 0.1 0.05 0.0000 12.863 0.00 1 2 1.5 0.0463 0 0.00
0.15 0.00 2 3 2.5 0.0724 0 0.00
0.1 0.2 0.0000 38.133 3 4 3.5 0.0920 5 0.46
0.2 0.3 0.25 0.06111 62.712 3.83 4 5 4.5 0.1041 15 1.56
0.35 52.44 5 6 5.5 0.1084 25 2.71
0.3 0.4 0.60617 86.507
6 7 6.5 0.1057 35 3.70
0.4 0.5 0.45 0.32201 109.421 35.23 7 8 7.5 0.0975 45 4.39
0.55 1.40 8 9 8.5 0.0855 55 4.70
0.5 0.6 0.01063 131.360
9 10 9.5 0.0716 65 4.65
0.6 0.7 0.65 0.00001 152.227 0.00 10 11 10.5 0.0575 75 4.31
0.7 0.8 0.75 0.00000 171.929 0.00 11 12 11.5 0.0442 85 3.76
12 13 12.5 0.0327 95 3.11
0.8 0.9 0.85 0.00000 190.369 0.00
13 14 13.5 0.0233 100 2.33
0.9 1 0.95 0.00000 207.453 0.00 14 15 14.5 0.0160 100 1.60
15 16 15.5 0.0106 100 1.06
16 17 16.5 0.0067 100 0.67
17 18 17.5 0.0041 100 0.41
18 19 18.5 0.0025 100 0.25
19 20 19.5 0.0014 100 0.14

130
Copyright of Journal of Electrical Systems is the property of Journal of Electrical Systems
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.