(N-1) contingency planning in radial distribution networks using genetic algorithms

(N-1) contingency planning has been object of study in the area of distribution networks of several decades. Energy distribution companies have to reconnect areas affected by an outage within a very short time, and observe operational constraints, to avoid the possibility of severe financial penalties by regulatory bodies. Distribution networks are often operated with a radial topology, but, ideally, should have more than one route to deliver energy to any node of the network. Switches in the network are opened to create the radial topology used in normal operation, and, in the case of an outage, alternate routes are activated by opening or closing switches located at specific points of the network. Given an outage situation (in our case represented by the disconnection of a single branch), the choice of which switches should change their state is a combinatorial optimisation problem, with a search space of 2^k, where k is the number of switches. Because of the exponential complexity, exact methods are prohibitively time-consuming. This work presents a genetic algorithm that provides a rapid answer to network managers in terms of a switching strategy to reconnect the affected area. The method takes into account the radial topology of the power flow and the operational limits of voltage and cable load. Computational tests were conducted on a real network with 96 buses and 16 switches, located within the operational area of Energy Australia. This paper describes the genetic algorithm in detail, presents thorough computational tests, and a complete contingency plan for the test network.