Security of supply improvement in high voltage distribution systems

In this thesis, algorithms are proposed to improve electricity distribution network supply restoration. The practical implementation of such algorithms relies on the presence of fully automated switches located at a certain number of network substations. The algorithm has the capability to restore a maximum number of customers if an outage occurs on any section of the test network. Since a very high cost is usually involved in the implementation of the fully automated system, a second algorithm was introduced with the aim of reducing the number and location of required switches. Discrete Particle Swarm Optimisation (DPSO) was employed to identify optimal placement of a limited number of remotely operable protective devices as well as the optimal sequence of reconfiguration and restoration of the supply. The reliability of the network was determined by calculating the number of possible post-outage restored customers, considering both upstream and downstream restorations. In the selection of optimal switching sequences, network voltage and current constraints were also considered to ensure that the identified restoration was viable. Further, the proposed algorithm considered the failure rate on each section of the network in arriving at the proposed optimal locations of switches. The developed DPSO-based algorithm and Brute Force is described and applied to real 11kV urban, semi-urban and rural distribution networks each with a different number of feeders and substations. These proposed algorithms have the capability to search for an unlimited number and locations of switches pairs or clusters for all networks (urban, semi-urban and rural) with optimal locations and number of Remotely Operable Switch Pairs (ROS). It was demonstrated that less than half of fully automated switches are needed to restore more than 95% of customers for each case study network. A significant reduction in investment cost of protective devices could be achieved by applying the proposed algorithm and at the same time improving and optimising the reliability of 11kV distribution networks