A Multi-objective Simulation–Optimization Approach for Design of Cutoff Walls and Apron of Diversion Dams

Abstract

Various seepage problems such as uplift pressure and high exit gradient lead to less efficient hydraulic structures since aims of the hydraulic design are substantial to be achieved. Cutoff walls and aprons are effective implements to control the problematic consequences of seepage flow. This paper presents a novel methodology for optimal design of cutoff walls and apron beneath Yusufkand diversion dam in Iran based on Non-dominated Sorting Genetic Algorithms-ΙΙ (NSGA-ΙΙ) multi-objective optimization model, Multi-Layer Perceptron (MLP) neural network model and Young Conflict Resolution Theory (YCRT). First, input–output data of a simulation model of seepage flow through the foundation of an existing diversion dam using Geostudio software are utilized to train and validate the MLP neural network, which can model the hydraulic performance of cutoff walls and the upstream apron. Then, the validated MLP neural network models are linked to NSGA-ΙΙ multi-objective optimization model and a trade-off curve is acquired among the hydraulic and cost criteria. Finally, YCRT is used to find the compromise optimal solution on the Pareto-front. Results show that by increasing length of the apron, when cutoff walls are at their minimum depth, the reduction of 18, 14 and 17% have been resulted in vertical exit gradient, uplift force and seepage flow discharge, respectively. Conversely, the same increment in length of the apron caused 2, 0 and 2% reduction in the mentioned parameters when other two cutoff walls are at their maximum depth. Moreover, the construction cost pertained to 1 meter of cutoff wall is nearly twice as much as for 1 meter of apron.