Improving reliability on distribution systems by network reconfiguration and optimal device placement.

Abstract

A distribution system without reliable networks impacts production; hinders economy and affects day to day activities of its customers who demand uninterrupted supply of high quality. All power utilities try to minimize costs but simultaneously strive to provide reliable supply and achieve customer satisfaction. This research has focused on predicting and thereafter improving the South African distribution network reliability. Predictive reliability modelling ensures that utilities are better informed to make decisions which will improve supply to customers. An algorithm based on Binary Particle Swarm Optimization (BPSO) was implemented to optimize distribution network configuration as well as supplemental device placement on the system. The effects on reliability, network performance and system efficiency were considered. The methodology was applied to three distribution networks in KwaZulu-Natal, each with diverse topology, environmental exposure and causes of failure. The radial operation of distribution networks as well as the practical equipment limitations was considered when determining the optimal configuration. The failure rates and repair duration calculated unique to each network was used to model the performance of each component type. Historical performance data of the networks was used as a comparison to the key performance indicators obtained from DigSILENT PowerFactory simulations to ensure accuracy and evaluate any improvement on the system. The results of a case study display improvements in System Average Interruption Duration Index (SAIDI) of up to 20% and improvements in System Average Interruption Frequency Index (SAIFI) of up to 24% after reconfiguration. The reconfiguration also reduced the system losses in some cases. Network reconfiguration provides improved reliable supply without the need for capital investment and expenditure by the utility. The BPSO algorithm is further used to optimally place and locate switches and reclosers on the networks to achieve maximum improvement in reliability for minimal cost. The results show that the discounted future benefit of adding additional protection devices to a network is approximately R27 million over a twenty-five-year period. The maximum SAIDI improvement from adding reclosers to a network was 21%, proving that additional device placement is a cost-effective means to improve system reliability.