Optimal design and operation of pumping stations using NLP-GA

Abstract

This paper addresses the optimal design and operation of an irrigation pumping station system using hybrid non-linear programming and a genetic algorithm (NLP-GA), and evaluates the algorithm in a practical problem. Results of the NLP-GA are compared with existing optimisation approaches to solve the same problem. The analytical approaches considered are the Lagrange multiplier method, a genetic algorithm and the honey-bee mating optimisation algorithm. The Lagrange multiplier method, genetic algorithm, honey-bee mating optimisation and the NLP-GA hybrid are used to simultaneously optimise the minimum annualised investment cost of the pumping station and its annual operating cost. The solution includes selection of pump type, capacity, number of units and scheduling of pump operation. The hybrid algorithm takes advantage of the high speed of NLP as well as the intelligent searching of evolutionary algorithms to overcome the shortcomings of individual NLP and genetic algorithm methods such as trapping of local optima, reporting only local or near-global optimal solutions and the low convergence rate of evolutionary algorithms in this type of problem. The results highlight the advantages in design, effective operation and ease of the NLP-GA method for solving complex problems of the type considered here. Although the NLP-GA converges rapidly, the results are promising and compare well with those of the Lagrange multiplier method, the genetic algorithm and honey-bee mating optimisation.