Optimal design of looped water supply network by GA-NLP hybrid approach

Abstract

Any Municipal distribution system consists of pipe network, valves and pressure generating facilities etc. Around 70% of total cost of any water distribution system is towards cost of pipe network only. Therefore, the cost of pipe network is to be reduced by selecting the appropriate lengths of candidate diameter pipes between the nodes in a pipe network.In the present study, a municipal looped water distribution network is designed for least cost by Genetic Algorithm – Non Linear Programming (GA-NLP) hybrid approach. In the present study, optimal lengths of sub pipes of various discrete diameters (sizes) available in the market making the link between the nodes are obtained by minimising the total cost of pipe network. In order to obtain the optimal solution of a looped network, non - linear hydraulic analysis and optimization of network is to be performed. In this study, non - linear hydraulic analysis is made a part of optimization by adding loop head loss and node continuity equations as constraints to the objective function. The model developed is solved by GA-NLP hybrid approach in MATLAB. Near optimal solution obtained by GA is taken as initial solution of NLP for obtaining optimal solution. By adopting GA-NLP hybrid approach, convergence is faster when compared to GA avoiding the complexity in selection of suitable trial values of GA parameters.Many researchers applied different techniques like linear programming, integer linear programming, linear programming gradient method and branch & bound algorithm etc. and optimized the looped pipe network by minimizing the cost. The results obtained by proposed model are compared with that of other investigators and found to be superior. It is also concluded that the proposed model may be applied to any larger municipal looped pipe network through demonstration by applying proposed model to a case study of Rajiv Gandhi Nagar of Guntur Municipal Corporation.