Optimal design of drainage channels using probabilistic search

Abstract

Optimal design of artificial drainage channels is essential for the planning and management of irrigation projects. In this paper a special class of drainage channel with horizontal bottom and parabolic sides (HBPS) is designed using a recent meta-heuristic optimisation tool, namely probabilistic global search Lausanne (PGSL). The solutions of different models are found to be competitive with those obtained by classical optimisation procedures and particle swarm optimisation (PSO). The trend and sensitivity analyses show that more attention is needed in maintaining the design bed slope while implementing channels in real field conditions. Uncertainties associated with channel design parameters are incorporated in the design by applying first-order second-moment (FOSM) formulations. Two probabilistic channel design models are considered, one with overtopping probability constraint and another with a channel capacity constraint, and the solutions are obtained in the PGSL environment. To improve practical usefulness, both the models are further modified by adding geometric constraints on top width and flow depth and solutions are obtained. The presented PGSL approach is found to be capable of handling varying degrees of complexity and is proven to be suitable even for channels with more complex profiles than that of simple trapezoidal channels.