Abstract
Application of the multi-objective particle swarm optimisation (MOPSO)algorithm to design of water distribution systems is described. An earlierMOPSO algorithm is augmented with (a) local search, (b) a modified strategy forassigning the leader, and (c) a modified mutation scheme. For one of thebenchmark problems described in the literature, the effect of each of the abovefeatures on the algorithm performance is demonstrated. The augmented MOPSOalgorithm (called MOPSO+) is applied to five benchmark problems, and in eachcase, it finds non-dominated solutions not reported earlier. In addition, forthe purpose of comparing Pareto fronts (sets of non-dominated solutions)obtained by different algorithms, a new criterion is suggested, and itsusefulness is pointed out with an example. Finally, some suggestions regardingfuture research directions are made.