More choices in water system design through hybrid optimisation

In recent years there has been an increased interest in using genetic algorithm techniques for the rehabilitation of water distribution networks. In particular, multi-objective (MOGA) techniques have found application in this domain and the results have been promising. MOGA techniques can be used to obtain Pareto optimal solutions which represent a set of compromises between two or more conflicting objectives. A desirable Pareto-curve has well-spaced points along the optimum trade-off between the two or more objectives. Pursuing this optimal trade-off though can be computationally expensive and often the optimal curve is not known and cannot be readily computed. Therefore a technique which can provide near optimal solutions which are well spread on the Pareto surface will ensure that this technique will be readily taken up by the water industry. This work investigates the possibility of combining traditional MOGA techniques and methods of local search to both improve solutions and reduce running times. Encouraging results from experiments were gained using this hybrid MOGA, applied to the New York City Tunnels water supply network rehabilitation problem.