Professional Documents
Culture Documents
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 novel integration technique for optimal network reconfiguration and distributed
Received 1 October 2013 generation (DG) placement in distribution system with an objective of power loss minimization and volt-
Received in revised form 12 March 2014 age stability enhancement. Fireworks Algorithm (FWA) is used to simultaneously reconfigure and allo-
Accepted 1 June 2014
cate optimal DG units in a distribution network. FWA is a new swarm intelligence based optimization
algorithm which is conceptualized using the fireworks explosion process of searching for a best location
of sparks. The radial nature of the network is secured by generating proper parent node-child node path
Keywords:
of the network during power flow. Voltage Stability Index (VSI) is used to pre-identify the optimal can-
Distributed generation
Fireworks Algorithm
didate locations for DG installation. Six different scenarios are considered during DG placement and
Network reconfiguration reconfiguration of network to assess the performance of the proposed technique. Simulations are carried
Power loss minimization out on well-known IEEE 33- and 69-bus test systems at three different load levels. The simulated results
Voltage stability enhancement demonstrate well the performance and effectiveness of the proposed method.
Ó 2014 Elsevier Ltd. All rights reserved.
Introduction the first to solve the network reconfiguration problem for loss min-
imization using a branch and bound-type optimization technique.
In recent times, electric distribution systems are becoming large Later on, several optimization algorithms have been developed for
and complex leading to higher system losses and poor voltage reg- loss minimization and/or voltage profile improvement. Authors in
ulation. Studies indicate that almost 10–13% of the total power [4] used a discrete artificial bee colony algorithm to find the max-
generated is lost as I2R losses at the distribution level [1], which imum loadability point by optimizing the distribution network,
in turn, causes increase in the cost of energy and poor voltage pro- and also used continuation power flow along with graph theory
file along the distribution feeder. Therefore, it becomes important to compute power flow. In [5], authors solved the reconfiguration
to improve the reliability of the power transmission in distribution problem assuming a series fault at a bus using Bacterial Foraging
networks. The most common methods used for voltage stability Optimization Algorithm (BFOA). Authors in [6] proposed modified
enhancement and power loss reduction in power distribution sys- honey bee mating optimization algorithm to investigate the net-
tems are network reconfiguration and DG placement [2]. The con- work reconfiguration problem with the consideration effect of
figuration of the network, placement and size of DG units should the renewable energy sources. In the very latest researches, a
be optimal to maximize the benefits and reduce their impact on new non-revisiting genetic algorithm had been proposed for solv-
the power system. Thus the optimal integration between these ing the reconfiguration problem [7].
two problems becomes a significant and complex problem. DG units are small generating plants connected directly to the
Network reconfiguration is the process of changing the topology distribution network or on the customer site of the meter. The
by altering the open/closed status of switches, so as to find a radial number of DG units installed in the distribution system has been
operating structure that minimizes the losses and improves the increasing significantly, and their technical, economical, and envi-
voltage stability while satisfying the operating constraints. Many ronmental impacts on power system are being analyzed. The most
researchers solved the network reconfiguration problems using critical factors that influence the technical and economic impacts
different methods for the last two decades [2]. Authors in [3] were are type, size, and location of DG units in power system. Recently,
several studies have been carried out on impact of DG units [8–12].
⇑ Corresponding author. Mobile: +91 9962680869. Authors in [8,9] presented an analytical and improved analytical
E-mail addresses: mohamedimran.a@vit.ac.in, imraneee20@gmail.com expression for finding the optimal location and size of DG units
(A. Mohamed Imran).
http://dx.doi.org/10.1016/j.ijepes.2014.06.011
0142-0615/Ó 2014 Elsevier Ltd. All rights reserved.
462 A. Mohamed Imran et al. / Electrical Power and Energy Systems 63 (2014) 461–472
Nomenclature
PLk real power load at bus k QLk reactive power load at bus k
Pk real power flowing out of bus k Qk reactive power flowing out of bus k
Pk+1,eff total effective real power load fed through the bus k + 1 Qk+1,eff total effective reactive power load fed through the bus
Ik equivalent current injected at node k k+1
Jk branch current in the line section between buses k and Vk voltage magnitude at bus k
k+1 DV max maximum voltage drop limit between buses 1 (substa-
Rk resistance of the line section between buses k and k + 1 tion) and k
PLoss ðk; k þ 1Þ real power loss in the line connecting buses k and Xk reactance of the line section between buses k and k + 1
k+1 PT; Loss total power loss of the system (base case)
PRT; Loss total power loss of the system after reconfiguration PDG total power loss of the system with DGs
T; Loss
Pmin
TDG minimum total real power generation limit Pmax maximum total real power generation limit
TDG
PDG,k real power supplied by DG at node k
PSS power supplied by the substation
PTDG total real power supplied by DG in the system
nd number of candidate buses for DG installation
Sk apparent power flowing in the line section between
Sk;max maximum power flow capacity limit of line section be-
buses k and k + 1
tween buses k and k + 1
V worst Worst voltage magnitude of the system.
nb total number of branches
n total number of buses
for reducing power loss along with methodologies for identifying to generate proper parent node child node path as in [17]. This
the optimal location. Authors in [10] proposed Particle Swarm path generation will ensure the radial nature of the system, also
Optimization (PSO) based technique to solve the optimal place- prevents the creation of unconnected branches or nodes, and for-
ment of different types of DGs for power loss minimization. In mation of mesh loops. Hence at each phase of distribution system
[11], authors proposed a new method which employs a modified reconfiguration, the power flow is carried out only after the gener-
teaching–learning based optimization algorithm to determine the ation of proper parent node child node path. For the sample system
optimal site and size of DG units. In [12], authors introduced a shown in Fig. 1, the generation of parent node child node path is
hybrid population-based algorithm with the combination of PSO illustrated in Table 1.
and gravitational search algorithm to determine the optimal place
and size of multi-DG units in distribution systems. More recently, Power flow equations
BFOA based approach had been proposed to solve the DG place-
ment problem at different load levels [13]. Distribution system power flow is calculated by the following
All the above researches [2–13] focussed only on the optimiza- set of recursive equations derived from the single line diagram
tion of either distribution network or the DG placement. However, shown in Fig. 2.
the objective was to minimize the power loss and/or to improve From Fig. 2, the equivalent current injected at node k is calcu-
the voltage stability, network reconfiguration usually did not take lated as
the DG units into consideration and vice versa. To benefit the
whole distribution network effectively, it is necessary to integrate
both the network reconfiguration and DG placement problems
[14,15]. In [16], authors have dealt network reconfiguration and
DG placement simultaneously using Harmony Search Algorithm 3 8 10
(HSA) based only on minimization of power losses.
The novelty of this paper is that it proposes recently developed
fireworks optimization algorithm for solving the distribution sys- 1 2 4 9
tem network reconfiguration together with DG placement for the
problem of power loss minimization and voltage stability enhance-
ment. The proposed technique makes use of VSI to pre-identify the
candidate bus locations for DG installation. Also the technique
monitors the radial nature of the network at all phases of reconfig- 6
uration by generating proper parent node-child node path during
power flow. The proposed technique has been validated on two
standard IEEE 33- and 69-bus test systems, and the results simu-
lated are compared with the results of other classical techniques 11
available in the literature to assess the performance and effective-
ness of the proposed method.
Problem formulation 5
! !
Table 1 PRT;Loss þ PDGT;Loss
Illustration of parent node child node path generation. Minimize F ¼ min: þ DV D ð8Þ
PT;Loss
Parent node 1 2 2 2 5 5 4 4 9 9
Child node 2 3 4 5 6 7 8 9 10 11 Subjected to the following operating constraints.
Power conservation limits:
X
n X
nb X
nd
PLk þ jQ Lk
Ik ¼ ð1Þ Pss ¼ P Lk þ PLoss ðk; k þ 1Þ P DG;k ð9Þ
Vk k¼2 k¼1 k¼1
Branch current in the line section connecting buses k and k + 1 Voltage deviation limits:
is calculated by using Kirchhoff’s Current law as
jV 1 V k j 6 DV max ð10Þ
J k ¼ Ikþ1 þ Ikþ2 ð2Þ
Distribution line capacity limits:
The above equation is generalised in matrix form by using Bus
current Injection to Branch Current matrix (BIBC) [18]. Now the jSk j 6 jSk;max j ð11Þ
branch current at each line can be calculated in a matrix form as
follows: Distributed generation capacity limits:
The power loss in the line section connecting buses k and k + 1 Furthermore, the radial nature of distribution network must be
is computed as maintained, and all loads must be served. If any one of the above
! constraints is violated, the resultant solution will be rejected.
P2k þ Q 2k
PLoss ðk; k þ 1Þ ¼ Rk ð5Þ
jV k j2
Voltage Stability Index for DG installation
The total power loss of the system is determined by the summa-
tion of losses in all the line sections, which is given as Voltage Stability Index (VSI) that identifies the most sensitive
bus is used to determine the candidate bus locations for DG instal-
X
nb
PT;Loss ¼ PLoss ðk; k þ 1Þ ð6Þ lation in the system [19]. The estimation of these candidate buses
k¼1 initially helps in reducing the search space significantly for the
optimization technique.
Consider a line section consisting an impedance of Rk + jXk and a
Voltage deviation index
load of Pk+1,eff + jQk+1,eff connected between k and k + 1 buses as
shown in Fig. 3.
One of the main advantages of the optimal network reconfigu-
From Fig. 3, the following equations can be written:
ration and DG installation is the fall in voltage deviation. This index
rebukes the configuration and size of DGs which gives the higher V k V kþ1
voltage deviation. The voltage deviation index (DVD) can be Jk ¼ ð14Þ
Rk þ jX k
defined as
V1 Vk Pkþ1;eff jQ kþ1;eff ¼ ðV kþ1 Þ J k ð15Þ
DV D ¼ max 8 k ¼ 1; 2; . . . ; n ð7Þ
V1
By solving the above two equations, the VSI at bus k + 1 is
During network reconfiguration and DG installation, if the state derived and given as
of the system has voltage limit violations, the proposed technique
will try to minimize DVD closer to zero and thereby improves volt- VSIðk þ 1Þ ¼ jV k j4 4:0ðPkþ1;eff X k Q kþ1;eff Rk Þ2
age stability and network performance.
4:0ðPkþ1;eff Rk þ Q kþ1;eff X k ÞjV k j2 ð16Þ
Objective function of the problem Using the above equation, VSIs are computed from the power
flows for all the buses of the given system and the values are
The proposed objective function (F) of the problem is formu- arranged in ascending order. VSIs decide the sequence in which
lated to minimize the total power loss and voltage deviation index, the buses are to be considered for DG installation. The optimal size
which is given by of DGs at the candidate buses are determined by using FWA.
Vk V k+1 V k+2
Jk P k + jQ k J k+1
th
k line P k+1 + jQ k+1
k k+1 k+2
P Lk + jQ Lk P Lk+1 + jQ Lk+1 P Lk+2 + jQ Lk+2
Vk V k+1 At each iteration of FWA, among all the current sparks and fire-
works, the best location is always kept for the next explosion gen-
R k + jX k Jk eration. After that, n 1 fireworks are selected with some
k th line probabilities proportional to their distances to other locations.
The selection probability of a location is defined by the following
k k+1 equations:
X X
P k+1,eff +jQ k+1,eff Rðxi Þ ¼ dðxi ; xj Þ ¼ jjxi xj jj ð23Þ
j2K j2K
Fig. 3. Simple distribution line.
Rðxi Þ
pðxi Þ ¼ P ð24Þ
j2K Rðxj Þ
Overview of fireworks optimization algorithm (FWA)
To keep the diversity, for a few specific sparks, an explosion The structure of above solution vector x for radial distribution
coefficient based on Gaussian distribution is applied to xkj and is network optimization is expressed by ‘‘Arc No.(i)’’ and ‘‘SW.No.(i)’’
given as: for each switch i. ‘‘Arc No.(i)’’ identifies the arc (branch) number
that contains ith open switch, and ‘‘SW.No.(i)’’ identifies the switch
xkj ¼ xkj þ Gaussianð1; 1Þ ð21Þ that is normally open on Arc No.(i). For large distribution networks,
In both the ways, if the obtained new location falls out of the it is not efficient to represent every arc in the string, as its length
search space, it is mapped to the search space according to the will be very long. Therefore, to memorize the radial configuration,
Eq. (22), it is enough to number only the open switch positions [22]. For
solving network reconfiguration problem with simultaneous DG
allocation, the first part of the solution vector is taken as the num-
xkj ¼ xkmin þ jxkj j % ðxkmax xkmin Þ ð22Þ
ber of open switches in the distribution network, and the second
A. Mohamed Imran et al. / Electrical Power and Energy Systems 63 (2014) 461–472 465
23 24 25 26 27 28 29 30 31 32 33
LP5 LP4
LP2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
LP1
LP3
19 20 21 22
Table 2
Performance analysis of proposed method on 33-bus system at different load levels.
The bold value defines the significance of proposed method in terms of power loss minimization and Voltage profile improvement.
466 A. Mohamed Imran et al. / Electrical Power and Energy Systems 63 (2014) 461–472
Table 3
Comparison of performances for 33-bus system.
part is the number of candidate buses chosen for DG installation. DG3 are sizes of DG units (kW) to be installed at candidate buses
Thus the solution vector for simultaneous reconfiguration and respectively.
DG installation is formed as follows: Step 3: Fireworks loop q = q + 1
Set off n fireworks at the selected n locations. i.e. for the
2 Open Switches DGSizes
3 selected n locations, objective function F(xi) is calculated by the
zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{ zfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{ power flows.
6 1 1 1 1 1 7
6 SW 1 SW 2 SW 3 SW 4 SW 5 DG11 DG12 DG13 7 Step 4: For each firework F(xi) 8 i ¼ 1; 2; . . . ; n
6 7
6 SW 2
SW 2 SW 3 SW 4 SW 25
2 2 2 2
DG1 DG2 DG3 7 2 27
Calculate the number of sparks that the firework yields: Si,
6 1
6 7 ð26Þ according to Eq. (19).
x¼6 7
6 7 Obtain the location of Si sparks of the firework xi using displace-
6 7
6 7
6 7 ment as in Eqs. (20) and (22);
4 SW n1 SW n2 SW n3 SW n4 SW n5 DGn1 DGn2 DGn3 5 Step 5: Generation of specific sparks loop k = k + 1.
Randomly select a firework xj and generate a specific spark for
the firework using Gaussian explosion method as in Eqs. (21) and
where SW1, SW2, SW3, SW4 and SW5 are open switches in the loops (22).
LP1 to LP5 formed corresponding to the tie switches; DG1, DG2 and Step 6: If k 6 ň, go to step 5.
33 34 35
32
31 53 54 55 56 57 58 59 60 61 62 63 64 65
47 48 49 50
30
29 LP4 66 67
28 LP5
LP2
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
36 LP1 LP3
37
38 52 51 68 69
39
40 41 42 43 44 45 46
Step 7: Evaluate the quality of all the above locations and select number of DGs installed for the given test systems is limited to
the best location that gives minimum F and keep it for the next three since it is studied that the rate of improvement of percentage
explosion generation. loss reduction decreases when the candidate locations increases
Step 8: Randomly select n 1 locations from the two types of more than three at all load levels [16,23]. However, the proposed
sparks and the current fireworks according to the probability given technique can be implemented for any number of DGs. The FWA
in Eq. (24). parameters initialized in the above section is taken common for
Step 9: if q 6 N, go to step 3. Otherwise end and display the both the test systems. In the simulation of the test systems, six dif-
results of best location. ferent scenarios at three different load levels: 0.5 (light), 1.0 (nom-
These are the prime steps of FWA employed to optimize the dis- inal), and 1.6 (heavy) are considered to analyze the superiority and
tribution network with simultaneous DG allocation for power loss performance of the proposed method.
minimization and voltage stability enhancement.
Scenario I: The system is without reconfiguration and DG units
(base case).
Results and discussions Scenario II: The system is optimally reconfigured by the avail-
able sectionalizing and tie switches.
To demonstrate and examine the applicability of the proposed Scenario III: Optimal size of DG units installed at the candidate
technique in solving the network reconfiguration and installation buses of the system.
of DG units simultaneously using FWA, it is applied to two Scenario IV: Optimal size of DG units is installed after reconfig-
12.66-kV test systems consisting of 33 and 69 buses. For both uration of the network.
the test systems, the substation voltage is considered as 1 p.u., Scenario V: The system is optimally reconfigured in presence of
and all the tie and sectionalizing switches are considered as candi- DG units installed (i.e. reconfiguring the distribution network
date switches for solving reconfiguration problem. The maximum with the distributed energy resources).
Table 4
Performance analysis of proposed method on 69-bus system at different load levels.
The bold value defines the significance of proposed method in terms of power loss minimization and Voltage profile improvement.
468 A. Mohamed Imran et al. / Electrical Power and Energy Systems 63 (2014) 461–472
1
0.995
0.955
0 10 20 30 40 50 60 70
Bus No
1
0.99
0.98
Voltage Magnitude (p.u)
0.97
0.96
0.95
0.94 Scenario I
Scenario II
0.93 Scenario III
Scenario IV
0.92
Scenario V
0.91 Scenario VI
0.9
0 10 20 30 40 50 60 70
Bus No
1.02
0.98
Voltage Magnitude (p.u)
0.96
0.94
Scenario I
0.92
Scenario II
0.84
0 10 20 30 40 50 60 70
Bus No
Scenario VI: System with simultaneous optimal network recon- normally closed sectionalize switches are 1–32 and the open tie
figuration and DG allocation. switches are 33–37. The single line diagram of 33-bus system with
the loops (LP1 to LP5) formed corresponding to each tie switch is
An analytical software tool has been developed in MATLAB shown in Fig. 4. The total real and reactive power loads of the sys-
environment for generating parent node child node path, to carry tem are 3.72 MW and 2.3 MVAr, respectively.
out power flow, calculate voltages and power losses. VSI for all nodes of the system is computed from power flow to
identify the candidate bus location of DGs for scenarios III, IV and
33-Bus test system VI. After computing VSI at all buses, they are sorted in ascending
order and ranked. Then the top three locations which are more
The line, load and tie line data of this test system are taken from sensitive are selected to install DG units in the system. The candi-
[24]. It consists of five tie lines and 32 sectionalize switches. The date locations selected for scenario III, IV and VI are given in
A. Mohamed Imran et al. / Electrical Power and Energy Systems 63 (2014) 461–472 469
Table 2. The performance of the proposed method at three different narios. At light, nominal and heavy loading conditions, the mini-
load levels is presented in Table 2. mum voltage magnitude (p.u.) is improved from 0.9583, 0.9131
From Table 2, it is observed that at nominal load, base case and 0.8529 to 0.9862, 0.9713 and 0.9554 in scenario VI. It is
power loss (kW) in the system is 202.67, which is reduced to observed that at all the three loading conditions, the power loss
139.98, 88.68, 83.91, 68.28 and 67.11 using scenarios II, III, IV, V reduced and minimum voltage improved using scenario VI is the
and VI, respectively. The percentage power loss reduction for sce- highest, which proves the superiority of the proposed technique.
nario II to VI is 30.93, 56.24, 58.59, 66.31 and 66.89, respectively. However, at all the three loading conditions,
Similarly, at light and heavy load conditions, the percentage power the improvement in percentage power loss reduction and min-
loss reduction for scenario II to VI is 29.04, 54.58, 56.82, 64.81 and imum voltage magnitude is almost the same in all the scenarios.
65.53; 33.73, 58.57, 60.90, 68.49 and 69.93 respectively. It is also Scenario V reduces the power loss almost equal to scenario VI
observed from Table 2 that, at all load levels, the minimum voltage but the size of DGs installed is higher than scenario VI. This reveals
magnitude of the system is improved impressively in all the sce- that the simultaneous network reconfiguration and allocation of
12
Power Loss (kW)
10
8
6
4
2
0
Scenario VI
Scenario V 70
60
Scenario IV 50
Scenario III 40
30
Scenario II 20
10
Scenario I 0
Scenario Line No
Fig. 9. Comparison of power loss at each line of 69 bus system at light load conditions.
50
Power Loss (kW)
40
30
20
10
0
Scenario VI
Scenario V
60 70
Scenario IV 50
Scenario III 30 40
Scenario II 10 20
Scenario I 0
Scenario Line No
Fig. 10. Comparison of power loss at each line of 69 bus system at full load conditions.
Power Loss (kW)
150
100
50
0
Scenario VI
Scenario V 70
Scenario IV 50 60
Scenario III 30 40
Scenario II 10 20
Scenario I 0 Line No
Scenario
Fig. 11. Comparison of power loss at each line of 69 bus system at heavy load conditions.
470 A. Mohamed Imran et al. / Electrical Power and Energy Systems 63 (2014) 461–472
Table 5
Comparison of performances for 69-bus system.
DG units (scenario VI) is better than other scenarios considered in each tie switch is shown in Fig. 5. The total real and reactive power
terms of quality of solutions. loads of the system are 3.80 MW and 2.69 MVAr, respectively.
In order to illustrate the performance of the proposed method, Similar to 33 bus test system, this test system is also simulated
the performance of FWA is compared with the results of HSA for six scenarios at three load levels and the results are presented
[16], GA [16] and Refined Genetic Algorithm (RGA) [16] available in Table 4. The total power loss (kW) at light, nominal and heavy
in the literature and is presented in Table 3. From the table, it is load conditions is reduced from 51.60, 224.96 and 652.42 to
perceived that at all scenarios, the performance of the FWA is bet- 9.58, 39.25 and 102.97 in scenario VI respectively. Table 2 and
ter than HSA, GA and RGA in terms of power loss minimization and Table 4 show the capability and effectiveness of the proposed
voltage stability enhancement. method in finding the optimal solutions for all the scenarios con-
sidered. From Table 4, it is also observed that scenario VI is more
69-Bus test system effective in reducing total power loss and improving minimum
voltage compared to other scenarios.
This is a large scale radial distribution system with 68 section- The voltage profiles of all six scenarios at light, nominal and
alizing switches (1–68) and five tie switches (69–73). The configu- heavy load conditions are compared and shown in Figs. 6–8,
ration, line, load and tie line data are taken from [25].The single respectively. From these figures, it is seen that the improvement
line diagram with the loops (LP1 to LP5) formed corresponding to in voltage profile of the system in scenario VI is better at all load
82.55
82.08
80.49
90
80.32
76.73
75.76
77.2
73.38
80
Percentage Power Loss Reduced (%)
65.39
61.43
61.04
60.66
70
56.17
55.85
55.42
54.08
60
50
40
30
20
10
0
Scenario II Scenario III Scenario IV Scenario VI
Scenarios FWA HSA [16] GA [16] RGA [16]
Fig. 12. Comparison of percentage power loss reduced by FWA with other classical techniques.
A. Mohamed Imran et al. / Electrical Power and Energy Systems 63 (2014) 461–472 471
0.07
0.0599
0.0589
0.0572
0.0572
0.06
0.0505
Voltage Deviation Index ( V D)
0.05
0.0389
0.0381
0.0323
0.0322
0.0313
0.04
0.0273
0.0264
0.028
0.0258
0.026
0.03
0.0204
0.02
0.01
0
Scenario II Scenario III Scenario IV Scenario VI
Fig. 13. Comparison of voltage deviation index reduced by FWA with other classical techniques.
[15] Georgilakis Pavlos S, Hatziargyriou Nikos D. Optimal distributed generation [21] Zheng Yu-Jun, Song Qin, Chen Sheng-Yong. Multiobjective fireworks
placement in power distributed networks: models, methods, and future optimization for variable-rate fertilization in oil crop production. Appl Soft
research. IEEE Trans Power Syst 2013;28(3):3420–8. Comput 2013;13:4253–63.
[16] Srinivasa Rao R, Ravindra K, Satisk K, Narasimham SVL. Power loss [22] Zhu JZ. Optimal reconfiguration of electrical distribution network using the
minimization in distribution system using network reconfiguration in the refined genetic algorithm. Elect Power Syst Res 2002;62:37–42.
presence of distributed generation. IEEE Trans Power Syst 2013;28(1):317–25. [23] Mistry Khyati D, Roy Ranjit. Enhancement of loading capacity of distribution
[17] Thukaram D, Wijekoon Banda HM, Jerome Jovitha. A robust three phase power system through distributed generator placement considering techno-
flow algorithm for radial distribution systems. Elect Power Syst Res economic benefits with load growth. Int J Electr Power Energy Syst.
1999;50:227–36. 2014;54:505–515.
[18] Teng Jen-Hao. A direct approach for distribution system load flow solutions. [24] Baran ME, Wu FF. Network reconfiguration in distribution systems for loss
IEEE Trans Power Deliver 2003;18(3):882–7. reduction and load balancing. IEEE Trans Power Deliver 1989;4(2):1401–9.
[19] Chakravorty M, Das D. Voltage stability analysis of radial distribution [25] Savier JS, Das D. Impact of network reconfiguration on loss allocation of radial
networks. Int J Electr Power Energy Syst 2001;23:129–35. distribution systems. IEEE Trans Power Deliver 2007;22(4):2473–80.
[20] Tan Ying, Zhu Yuanchun. Fireworks algorithm for optimization. Adv Swarm
Intell 2010;6145:355–64.