Evaluation of Optimal River Training Work Using GA Based Linked Simulation-Optimization Approach

Use of structural measures for controlling a river to minimize its devastating effect and to utilize it for the benefit of mankind is a common practice all over the world. Because of high investment, such measures require prior investigation through model study. As lab based physical model study is very expensive and time consuming, mathematical modeling is generally used for investigating different alternatives of river training works. In this study, a new approach is proposed for deciding appropriate river training measure in a particular reach of a river or channel. In this methodology, an optimization model is linked with the hydrodynamic model for obtaining cost effective combination of groynes which will maintain a user defined flow speed in a pre-decided portion of a river reach. The optimization model is developed using binary coded Genetic Algorithm (GA) and the flow simulation model uses the Beam and Warming scheme for solving the two dimensional (2D) hydrodynamic equations of unsteady flow. The performance of the model is tested by applying the methodology in a rectangular channel for attaining different target speed values at a pre-defined portion of the channel and logical results have been obtained for all the tested scenarios.