Optimal Configuration of Wind Farms in Radial Distribution System Using Particle Swarm Optimization Technique

Bulletin of Electrical Engineering and Informatics
Vol. 7, No. 2, June 2018, pp. 286~293
ISSN: 2302-9285, DOI: 10.11591/eei.v7i2.1224  286
Journal homepage: http://journal.portalgaruda.org/index.php/EEI/index
Optimal Configuration of Wind Farms in Radial Distribution
System Using Particle Swarm Optimization Technique
Mahesh Kumar1
, Bhagwan Das2
, Perumal Nallagownden3
, Irraivan Elamvazuthi4
, Sadia Ali Khan5
1,3,4
Department of Electrical and Electronics Engineering, Universiti Teknologi PETRONAS Seri Iskandar, Perak Darul
Ridzuan, Malaysia
2
Department of Electronic Engineering, Quaid-e-Awam University of Engineering and Technology, Sindh Pakistan
5
Department of Computer Science, University of Karachi, Sindh, Pakistan
Article Info ABSTRACT
Article history:
Received Feb 01, 2018
Revised Apr 11, 2018
Accepted Apr 29, 2018
Recently, a wide range of wind farm based distributed generations (DGs) are
being integrated into distribution systems to fulfill energy demands and to
reduce the burden on transmission corridors. The non-optimal configuration
of DGs could severely affect the distribution system operations and control.
Hence, the aim of this paper is to analyze the wind data in order to build a
mathematical model for power output and pinpoint the optimal location. The
overall objective is minimization of power loss reduction in distribution
system. The five years of wind data was taken from 24o
44’ 29” North, 67o
35’ 9” East coordinates in Pakistan. The optimal location for these wind
farms were pinpointed via particle swarm optimization (PSO) algorithm
using standard IEEE 33 radial distribution system. The result reveals that the
proposed method helps in improving renewable energy near to load centers,
reduce power losses and improve voltage profile of the system. Moreover,
the validity and performance of the proposed model were also compared with
other optimization algorithms.
Keywords:
Distribution system
Optimal location and sizing
Particle Swarm Optimization
(PSO)
Wind assessment
Copyright © 2018 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
Mahesh Kumar,
Department of Electrical and Electronics Engineering,
Universiti Teknologi PETRONAS,
Seri Iskandar, Perak Darul Ridzuan, Malaysia.
Email: rathii.mahesh@gmail.com
1. INTRODUCTION
The demand of electricity is increased recently due to urbanization and industrial revolution. In
early ages, electric demand was mainly fulfilled by fossil fuel based generation. Now a days, fossil fuel based
power generation leaving a serious threat to environment. It releases large amount of greenhouse gas
emissions, which further increases the global temperature and eventually results natural disasters (like.
floods, earth-quick etc.). In order to cater this problem, the renewable energy based power generation in
distribution system is of great interest. Different types of renewable energies are available, which are
inexhaustible and freely available in the nature. The electrical power generation system is large and
centralized; it is delivering bulk power to the national grids and even international grids. Currently it is
operating optimally, adding more capacity to sustain power will increase the cost and deteriorates the
environmental standards. On the other hand, the small power generation called distributed generation (DG)
from renewable energies in distribution system will settled the problems of centralized electrical power
generation. There are number of benefits with DG integration in distribution system, firstly it is near to the
load center hence reduces the electric current from transmission system. Electric power locally to load ends
will improves the power quality and reliability. The DG has numerous advantages in terms of technical,
economic and environmental [1]-[3] such as:

BEEI ISSN: 2302-9285 
Optimal Configuration of Wind Farms in Radial Distribution System Using Particle… (Mahesh Kumar)
287
a. It improves the voltage profile.
b. Reduce power loss.
c. Enhance the voltage stability and increases the system's capacity.
d. Improve reliability and power quality.
e. Reduce transmission line congestion hence reduce investment deferral.
f. Reduce fossil fuel dependency hence reduce GHG emissions.
Power output from wind turbine is mainly depends on wind speed and complete knowledge of wind
characteristics [4]. The average monthly and yearly wind speeds gives the knowledge about suitability of
wind location. The wind speed within the range of 4 m/s to 6 m/s classifies as moderate [5] and according to
European wind energy union, wind speed with range of 6.5 m/s is considered as approximately good, 7.5 m/s
is considered as good and 8.5 m/s is very good at a given height [6]. China, Germany, and USA are top three
countries which are utilizing 45.1 percent, 10.2 percent and 9.4 percent respectively [7].
Nowadays, the wind turbines more and more are integrated in distribution system for nearest power
delivery and to fulfill energy needs. Hence to get maximum remuneration from renewable DGs and to
increase the suitable environment, it should be properly configured. Many authors havr researched on same
area such as voltage profile improvement with wind based DG in distribution. The author [8-9] uses
convolution technique with probabilistic model of wind based DG and results are discussed with expected
value of voltage on loads. The short-term forecasting model with solar PV DG by is proposed by [10]. The
effects of high wind based DG penetrating in distribution system on voltage profile is researched by [11]. The
author concludes that High wind based DG will changes the system voltage, however with proper dispersion
level, this drawback can be overcome. Wind, solar and load with probabilistic model in distribution system is
analyzed by [12]. The author uses Monte Carlo based probabilistic load flow simulation to reduce power
losses and increase the voltage profile of the system. Probabilistic model for investment deferral in
distribution system is analyzed by [13]. The author uses combined heat and power (CHP), solar and wind
based DG along with probabilistic load generation. Furthermore author also concludes that DG(s) can defer
the investment with increasing load growth. Optimal size of different types of DG using analytical method
for optimal placement and sizing is described by [14]. Linear programming method is used for optimal DG
placement and to gather maximum amount of energy, considering financial and technical constraints [15-16].
Power loss reduction and customer reliability improvement using DG in distribution system has been
proposed by [17]. The probabilistic generation-load model with its all operating condition and probabilities
has been accommodated in deterministic model. Author used the non-linear programming method to solve
the optimal DG placement and sizing problem with only wind [18] and all DG types [19].
This study offers an advancement on wind energy potential. Hourly, monthly and yearly mean and
standard deviations were utilized to form probability distributions. Furthermore, these probabilities were
explored to identify power generation output from wind turbines and finally these wind farms were
connected to radial distribution systems with optimal locations so that maximum power losses could be
curtailed. The PSO optimization technique was utilized to pinpoint optimal locations in the distribution
system.
This paper is organized as Section 2 presents the problem formulation for power loss reduction with
system constraints. Section 3 presents wind speed model. Section 4 presents the simulation results of wind
speed data and its results. Section 5 presents the optimization technique and simulation results and discussion
for optimal location of wind farm with different case studies. Section 6 presents the conclusion.
2. PROBLEM FORMULATION)
In this paper main objective function is set as minimization of real power loss of the radial
distribution system with following constraints. The objective function can be mathematically described as in
Equation 1.
min f = min(Tloss) (1)
Where, Tloss is the active power losses of the radial bus system.
Equality constraints:
The equality and inequality constraints are given in Equation (2-5).
The bus real and reactive power is limited to:
PDG,M = PLoss + ∑ PD,M (2)
QDG,M = QLoss + ∑ QD,M (3)

 ISSN: 2302-9285
BEEI, Vol. 7, No. 2, June 2018 : 286 – 293
288
Where PDG,M and PDG,M are the real and reactive power injection by DG at bus M. ΣPD,M and ΣPDG,M are the
total real and reactive power at bus M. According to these constraints the real and reactive power injection at
bus M must be equal to actual real power demand at that node and power losses of that node.
Inequality constraints:
Position of DG:
2 ≤ DG position ≤ nbuses (4)
Voltage at load bus
VM
min
≤ VM ≤ VM
max
(5)
VM
min
and VM
max
are minimum and maximum voltage at bus M. The power loss, voltage and other parameters
are found using backward forward load flow analysis, detail given in [20]
3. WIND SPEED MODEL
3.1. Wind Probability Distribution
The wind speed data are modeled by probability distributions. The purpose of these distribution are
to examine the frequency of wind speed data or in other terms that a particular wind speed falls how much in
during a time period (in a month or in a year). In literature many probability distributions are cited, like
normal distribution, lognormal distribution, weibull and Rayleigh distributions. Among all distributions, the
Weibull probability distribution functions is highly used due to its accurate and adequate specification for
wind data analyzing.
𝑓(𝑣) = (
𝑘
𝑐
) (
𝑣
𝑐
)
𝑘−1
𝑒𝑥𝑝 [− (
𝑣
𝑐
)
𝑘
] (6)
where f(v) is the probability of wind speed, k is shape parameter, c is scale parameter.
3.2. Wind Power Generation Model
The power output from wind turbine depends primarily on wind speed and it varies in non-linear
relation with wind speed. Figure.1. shows the graph between wind speed versus its probability distribution
and output power curve of wind turbine. From Figure 1, it can be seen that the output power start at wind
turbine minimum cut in speed. As wind speed increases till rated speed output power increases nonlinearly
[22], then after remains constant till wind speed reaches at wind turbine cut-out stage. Interestingly beyond
cut-out stage even speed increases, the wind turbine needs to shut down for retaining the wind turbine
mechanical strength and stability.
Figure 1. Wind speed versus Probability distribution and wind speed versus electrical output power
Mathematically output power from wind turbine can be expressed as:

BEEI ISSN: 2302-9285 
289
𝑃𝑤𝑖𝑛𝑑,𝑜𝑢𝑡𝑝𝑢𝑡 =
{
0 0 ≤ 𝑣𝑎𝑤 ≤ 𝑣𝑐𝑖
𝑃𝑟𝑎𝑡𝑒𝑑 ∗
𝑣𝑎𝑤−𝑣𝑐𝑖
𝑣𝑟−𝑣𝑐𝑖
𝑣𝑐𝑖 ≤ 𝑣𝑎𝑤 ≤ 𝑣𝑟
𝑃𝑟𝑎𝑡𝑒𝑑 𝑣𝑟 ≤ 𝑣𝑎𝑤 ≤ 𝑣𝑐𝑜
0 𝑣𝑐𝑜 ≤ 𝑣𝑎𝑤
(7)
where Prated is the rated output power from wind turbine, vaw is the average wind speed , vci is the cu-in
speed, vr is the rated speed and vco is the cut-out speed. where P(Gw) is the power generation of particular
state, vw1 and vw2 are the wind speed states.
4. SIMULATION RESULTS FOR WIND SPEED DATA
In this paper, five years of wind speed data was analyzed using MATLAB 2015a software. Each
minute by minute data of five years were further divided into yearly, monthly, and hourly basis. The Weibull
and Rayleigh probability density function (PDF) was designed for each month and year. The hourly,
monthly, and yearly data was utilized to form the mean and standard deviation. Finally, the power output
curve of wind turbine from the manufacturer was utilized to identify power output from wind turbines at
selected sites.
4.1. Yearly and Monthly Wind Speed
Figure 2 shows the annual wind speed of every minute from the year, 2003 to 2007. It can be seen
that in 2006 and 2007, the wind speed remained high as compared to 2003, 2004 and 2005. The overall mean
wind speed of each year remained at more than 5 meters per second. Figure 3 shows the monthly mean wind
speed of each year. It can be observed that in each year, from April to September, wind speed remained in the
range of 4-9 meters per second, whereas the highest wind speed was in the month of August, 2004
Figure 2. No. of wind speed samples of year 2004
to 2007
Figure 3. Mean of monthly wind speed of year
2004 to 2007
4.2. Average Wind Speed and Statistical Parameter
The average wind speed was taken from the sum of wind speeds of each year to the total number of
wind samples of that year. Figure 4 records the average wind speed of each year. It also shows that in 2004
and 2006, the average wind speed remained more than 5.6 meters per second which was higher than all the
other years.
Figure 4. Mean of yearly wind speed of year 2004 to 2007
0
5
10
Wind
Speed
(m/S)
Year 2003 Year 2004 Year 2005
Year 2006 Year 2007

 ISSN: 2302-9285
BEEI, Vol. 7, No. 2, June 2018 : 286 – 293
290
Table 1 shows the statistical parameters of wind speed, along with the mean or average wind speed
of each year, which is derived from Equation 6.
Table 1. The Statistical Wind Parameters
Parameters 2003 2004 2005 2006 2007
Avg 5.1461 5.6200 5.4062 5.634 5.2996
std 3.3562 3.2332 2.9831 3.226 3.1859
K 1.5906 1.8228 1.9073 1.8324 1.7378
C 5.7366 6.3234 6.0931 6.3414 5.9481
Vmp 3.0774 4.0874 4.1273 4.1226 3.6333
Vmax 7.7951 8.0380 7.6002 8.0427 7.7261
4.3. Electrical Power Output from Wind Turbines
The electrical output power from wind turbines depends on the wind turbine model and wind speed
data. Equation (7) is used to calculate the yearly output power and energy from Enercon-65 wind turbine.
The height of the wind turbine is set at 80 meters using the power law formula. The other parameters of this
wind turbine are vci = 4 m/s, vr = 13 m/s, vco = 25 m/s and Prated = 1.5 ∗ e6
watts. Figure 5 shows the wind
turbine power curve, the total energy output of this turbine is 562.476 kiloWatts.
Figure 5. Wind turbine power curve
5. SIMULATION RESULTS AND DISCUSSION
Optimal placement of wind farm was carried out using PSO optimization technique. The PSO
algorithm is a stochastic optimization technique introduced by [23]. This algorithm is basically inspired by
social behavior of living things such as birds flocking and fish schooling. Suppose v and S in search space are
the velocity and position of particle i . So ith
particle position can be written as (𝑆𝑖 = 𝑆𝑖,1, 𝑆𝑖,2 … … 𝑆𝑖,𝑑),
where d is the dimensional space of particle. The previous best ith
particle position will be (𝑃𝑏𝑒𝑠𝑡𝑖 =
𝑃𝑏𝑒𝑠𝑡1, 𝑃𝑏𝑒𝑠𝑡2 … … 𝑃𝑏𝑒𝑠𝑡𝑖,𝑑). The best particle among the all, will be gbest and the velocity of particle i can
be represented by (𝑉𝑖 = 𝑉𝑖,1, 𝑉𝑖,2 … … 𝑉𝑖,𝑑) . Each particle's velocity closes to pbest and gbest is calculated by
Equation (8). Each particle's current position is updated by Equation (9). Figure 6, shows the complete flow
chart of proposed method.
   
1
1 2
() ()
k k k k
id id id id d id
v v c rand pbest v c rand gbest v


          (8)
1 1
, 1,2, .. , 1,2 ,
k k k
id id id
S S v i n d m
 
      (9)
0 5 10 15 20 25
0
5
10
15
x 10
5
velocity (m/s)
turbine
power
(W)
Turbine Power Curve

BEEI ISSN: 2302-9285 
291
The IEEE 33 radial distribution system was conducted using the proposed method as shown in
Figure 7. The total real and reactive power demand on this distribution system is 3.715 MW and 2.300
MVAr, respectively. The bus and line parameters are taken from [24-25]
Figure 6. Flow chart for optimal DG configuration using single objective PSO optimization method
Two case studies were performed for the proposed work. For case I, different numbers of wind
turbines were connected on only one bus of the system, the best bus would be selected by PSO, and for case
II, the same number of wind farms were connected on two buses of the system. The following are details of
the case studies.
Base case: The system’s losses before installing wind farms were noted as 211 kW and 143 KVAr,
respectively and the minimum voltage profile was observed at bus no. 18 and the maximum voltage at slack
bus no. 1 of the distribution system.
Figure 7. IEEE 33 radial distribution system

 ISSN: 2302-9285
BEEI, Vol. 7, No. 2, June 2018 : 286 – 293
292
Case I: The main purpose of this research was to reduce power losses and improve the voltage
profile of the system. It was observed from the study that less numbers of wind farms would not increase
voltage profile as the minimum standard voltage was 0.95 volts (± 10%) and more numbers of wind farms
would reduce power losses. Hence this study used four, five, and six wind turbines as shown in Table 2. It
was observed that best results were obtained when using five wind farms at bus no. 6
Case II: The same number of wind farms as in case I were separated into two and added on two
buses of the system as shown in Table 2. Consequently, the best results were performed by four wind farms
on bus 11 and bus 29, respectively. So it can be concluded that the same number of wind farms at two buses
provided more reduction in power losses and also voltage profile improvement as compared to four wind
farms on one bus.
Table 2. Power loss before and after wind farm placement
No. of wind
farm
Wind farm
location
Ploss (kW) Qloss (KVar) Voltage (p.u)
Min Max
Base case -- 211 143 0.9037@ Bus 18 1.0@ Slack bus
Case I
4 Bus 6 112.6 82.3 0.9375@ Bus 18 1.0@ Slack bus
5 Bus 6 111.7 82.2 0.9454@ Bus 18 1.0@ Slack bus
6 Bus 7 124.5 103.6 0.9568@ Bus 18 1.0@ Slack bus
Case II
4 (2+2) Bus 11, Bus29 91.9 63.7 97.18@ Bus 18 1.0@ Slack bus
5 (2+3) Bus 10, Bus 27 100.7 71.0 96.43@ Bus 18 1.0@ Slack bus
6 (3+3) Bus 8, Bus 24 104.8 74.9 0.9480@ Bus 18 1.0@ Slack bus
Table 3. Power Loss Before and After DG Placement
Test
system
Methods
Base case
DG size
After DG
Loss reduction
Power loss Power loss
Ploss
(kW)
Qloss
(KVar)
Type-1 MW
Type-2 MVar
Ploss (kW)
Qloss
(KVar)
Ploss
(%)
Qloss
(%)
(DG)
PSO
211 143
2.6000 @bus 6 111.0 86.4 47.39 39.58
GSA (24) 2.6005@bus 6 111.0 --- --- ---
AM (23) 2.600 @bus 6 111.0 --- --- ---
The validity of the proposed method for optimal placement and sizing of the DG in radial
distribution system with PSO techniques were also compared with others, like analytical method and grid
search algorithm. The results are presented in Table 3. Hence it can be concluded that the proposed method
was correct and provided the similar results.
6. CONCLUSION
This paper presents the assessment of wind energy output from wind turbines as distributed
generations. The wind data was gathered from an area in Pakistan. Data was further explored to build
different wind farm capacities. The output power of these wind farms was added into the distribution system
in order to reduce power losses and improve voltage profile of the system. The optimal location for the wind
farm was pinpointed using PSO optimization technique. The validity of the proposed model was also
compared with other optimization techniques. On the whole, it can be concluded that the proposed method
provided better results.
ACKNOWLEDGEMENTS
The authors are grateful to Universiti Teknologi PETRONAS, Malaysia for technical and financial
help for this research.
REFERENCES
[1] R. Viral and D. Khatod, "Optimal planning of distributed generation systems in distribution system: A review,"
Renewable and Sustainable Energy Reviews, vol. 16, pp. 5146-5165, 2012.

BEEI ISSN: 2302-9285 
293
[2] P. Paliwal, N. Patidar, and R. Nema, "Planning of grid integrated distributed generators: A review of technology,
objectives and techniques," Renewable and Sustainable Energy Reviews, vol. 40, pp. 557-570, 2014.
[3] 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.
[4] R. Katyal, S. Gomathinayagam, S. Ahmed, and S. Akhtar, "Wind variability study for a complex wind farm site in
India," Indian Journal of Science and Technology, vol. 3, pp. 1043-1047, 2010.
[5] D. L. Elliott, Dominican Republic Wind Energy Resource Atlas Development: National Renewable Energy
Laboratory, 1999.
[6] A. Garrad, "Wind energy in Europe: A plan of action," Summary report of wind energy in Europe. Time for action.
Brussels: European Wind Energy Association, 1991.
[7] Z. Liao, "The evolution of wind energy policies in China (1995–2014): An analysis based on policy instruments,"
Renewable and Sustainable Energy Reviews, vol. 56, pp. 464-472, 4// 2016.
[8] P. Chiradeja and R. Ramakumar, "Voltage profile improvement with distributed wind turbine generation-a case
study," in Power Engineering Society General Meeting, 2003, IEEE, 2003.
[9] N. Roy, H. Pota, M. Mahmud, and M. Hossain, "Impacts of Distributed Wind Generation on Distribution
Networks," in Proc. 48th Annual Conference of the Australian Solar Energy Society (AuSES), Solar, 2010.
[10] Y.-S. Jeong, S.-H. Lee, K.-H. Han, D. Ryu, and Y. Jung, "Design of short-term forecasting model of distributed
generation power for solar power generation," Indian Journal of Science and Technology, vol. 8, pp. 261-270,
2015.
[11] N. Roy, M. Mahmud, and H. R. Pota, "Impact of high wind penetration on the voltage profile of distribution
systems," in North American Power Symposium (NAPS), 2011, 2011, pp. 1-6.
[12] K. Zou, A. P. Agalgaonkar, K. Muttaqi, S. Perera, and N. Browne, "Support of distribution system using distributed
wind and PV systems," in Power Engineering Conference, 2009. AUPEC 2009. Australasian Universities, 2009,
pp. 1-6.
[13] V. Mendez, J. Rivier, J. De La Fuente, T. Gomez, J. Arceluz, J. Marin, et al., "Impact of distributed generation on
distribution investment deferral," International Journal of Electrical Power & Energy Systems, vol. 28, pp. 244-252,
2006.
[14] D. Q. Hung, N. Mithulananthan, and R. Bansal, "Analytical strategies for renewable distributed generation
integration considering energy loss minimization," Applied Energy, vol. 105, pp. 75-85, 2013.
[15] A. Keane and M. O'Malley, "Optimal utilization of distribution networks for energy harvesting," Power Systems,
IEEE Transactions on, vol. 22, pp. 467-475, 2007.
[16] A. Keane and M. O'Malley, "Optimal allocation of embedded generation on distribution networks," Power
Systems, IEEE Transactions on, vol. 20, pp. 1640-1646, 2005.
[17] P. Farhadi, H. Shayeghi, T. Sojoudi, and M. Karimi, "Customer reliability improvement and power loss reduction
in distribution systems using distributed generations," Indian Journal of Science and Technology, vol. 5, pp. 2313-
2317, 2012.
[18] Y. Atwa and E. El-Saadany, "Probabilistic approach for optimal allocation of wind-based distributed generation in
distribution systems," IET Renewable Power Generation, vol. 5, pp. 79-88, 2011.
[19] Y. Atwa, E. El-Saadany, M. Salama, and R. Seethapathy, "Optimal renewable resources mix for distribution system
energy loss minimization," Power Systems, IEEE Transactions on, vol. 25, pp. 360-370, 2010.
[20] M. Haque, "Efficient load flow method for distribution systems with radial or mesh configuration," IEE
Proceedings-Generation, Transmission and Distribution, vol. 143, pp. 33-38, 1996.
[21] S. F. Khahro, K. Tabbassum, A. M. Soomro, X. Liao, M. B. Alvi, L. Dong, et al., "Techno-economical evaluation
of wind energy potential and analysis of power generation from wind at Gharo, Sindh Pakistan," Renewable and
Sustainable Energy Reviews, vol. 35, pp. 460-474, 2014.
[22] T. Yamamoto and Y. Ishida, Linear and nonlinear rotordynamics: A modern treatment with applications vol. 11:
John Wiley & Sons, 2001.
[23] Kennedy J. Particle swarm optimization. Encyclopedia of Machine Learning: Springer; 2010, pp. 760-6.
[24] S. Kansal, V. Kumar, and B. Tyagi, "Optimal placement of different type of DG sources in distribution networks,"
International Journal of Electrical Power & Energy Systems, vol. 53, pp. 752-760, 2013.
[25] S. Kansal, V. Kumar, and B. Tyagi, "Composite active and reactive power compensation of distribution networks,"
in Industrial and Information Systems (ICIIS), 2012 7th IEEE International Conference on, 2012, pp. 1-6.

Optimal Configuration of Wind Farms in Radial Distribution System Using Particle Swarm Optimization Technique

Recommended

More Related Content

What's hot (20)

More from journalBEEI (20)

Recently uploaded (20)

Optimal Configuration of Wind Farms in Radial Distribution System Using Particle Swarm Optimization Technique