Soft Computing Techniques for Optimal Capacitor Placement - Complex System Modelling and Control Through Intelligent Soft Computations - page 610

Table 5 Performance of IEEE-69 bus radial distribution with capacitors (in kvar) placed using
GA and PSO
Algorithm P l (kW) Power loss cost ($/year) Total cost ($/year) Bene ts ($/year)

GA 879.07 147,683.76 542,800.46 10,243.76

PSO 830.01 139,441.70 532,556.7 7,874.83

system, leads to higher power and energy losses. Thus, total cost obtained with GA
is US$542,800.46, while it is US$532,556.7 with PSO. Thus, higher economic
bene ts are obtained with PSO, in comparison to GA.

Thus, the results show better performance of the network with PSO, than that
with GA. Both of these algorithms are stochastic in nature, where the performance
of the algorithm may vary with the conditions. The observation of these algorithms
can be summarized by running the program many times. In power system problem,
evaluating the final design solution is dominating part of the overall computational

burden. As a result, to

-
ciency, less number of design candidate are evaluated. The overall computational
ef

find the correct optimal solution high computational effi-

ciency depends upon the population size, and number of iterations or genera-
tions. The two components of PSO velocity update ( 20 ), provide the global and
local search capabilities to the algorithm, which enhances the searching capabilities
of the PSO over GA. Further, in OCP discrete nature of the variable (i.e., capacitor
sizes) allows the movement in discrete steps in the search space. Thus, the GA fails
to provide to better solution, which may be due to entrapment of the GA solutions
in the local minima, whereas PSO is able to provide better result.

6.2.2 Case-II

Since, the discussions made above show that better results are obtained with PSO.
Here, a comparison is made between the PSO and NM-PSO algorithm, using
another planning method P-2, as the objective function. For NM-PSO algorithm,
the parameters are selected experimentally, as = 0.5, and N is

the number of capacitors to be placed. The experiments are performed up to 100
iterations, for both the algorithms. The capacitors are placed at candidate buses
identi

ʱ =1, ʲ =2, ʳ = 0.5, ʴ

ed using BSi. i . The sizing of capacitors at these buses is performed using PSO
and NM-PSO separately using the P-2 (Kumar and Singh 2014).

Table 6 Capacitors placed (kvar) at the candidate IEEE 69 bus radial distribution bus system
Algorithm

Bus number Total

(kvar)

11 12 21 48 49 50 59 61 64

1 300 600 900 300 300 1,500 600 2,400 900 7,800

2 300 600 600 300 1,500 300 600 2,400 1,200 7,800