Optimization of Simultaneous Energy Storage Sizing & Network Reconfiguration in an Active 11kV Radial Distribution Network

Abstract

There is an increasing pressure for the UK to move towards a low carbon emission. The Electric power system has a major contribution by shifting to the decarbonization of power sector for a low emission target. It is important to know that the electric generation is not the only sector in the power system that affect the CO2 emissions, there is also an indirect emission as results of losses. This Thesis presents a novel approach to coordinate the simultaneous operation of network reconfiguration with the sizing and the allocation of energy storage systems in distribution networks for losses reduction aim. The major challenge of this work is to solve this hard-stochastic optimization problem with an algorithm that has the capability to find the optimum solution in a reasonable computational time to help utilities to use it for online applications. This thesis proposes a developed optimization technique for network reconfiguration to enhance the search space and improve the computational time and the convergence issue of the particle swarm optimization. The thesis also presents a novel comparison between a previously adopted engineering approach used by Western Power Distribution Company and the new proposed modified algorithm in term of losses reduction, and computational time. The similarities, the differences, the advantages and the shortcomings for both approaches were highlighted. Moreover, two different utilizations for Monte Carlo Approach were investigated in this thesis. The first is aimed to decrease the search space of the proposed modified algorithm by proposing Multi Stages Modified Particle Swarm approach for distribution network reconfiguration problem solution. The second application for Monte Carlo Method is for sizing the battery storage units for more losses’ reduction. Results show that merging the network reconfiguration and the sizing and the allocation of battery storage systems in distribution networks allow more losses reduction more than using each strategy in isolation. Furthermore, it was concluded that the new developed algorithm technique could be applied using the real distribution network giving the optimum losses reduction in a reasonable computational time, which in turn could be used for online implementation.