Optimization of urban water pipe network design using fast-messy genetic algorithms (fmGA)

To have an efficient water distribution network, optimal design alternatives need to be identified and analysed using combined hydraulic modelling and optimization. This paper reports on the application of the fast-messy genetic algorithm (fmGA) coupled with hydraulic modelling tool EPANET to assess and select an optimum design and operational alternative for a water distribution pipe network. The sole objective of the optimization modelling was to minimize network costs subject to hydraulic and design constraints. The fmGA was first tested using the benchmark case study of the Hanoi network and then applied to the real network of Maun, Botswana which is considered as the case study. We have compared our results of the fmGA model application with other optimization techniques applied to the Hanoi network. The findings of the test revealed that the fmGA is superior to other popular metaheuristic optimization methods in terms of processing speed with comparable accuracy and pressure constraints fulfilled for all nodes. It also provides the best and least solution. For the water distribution network of Maun, the best pipeline configuration and route were determined, in which Configuration B is found to be the best and least cost solution to improve the water supply situation in Maun.