Performance Comparison of Artificial Fish Swarm, Shuffled Complex Evolution and Simulated Annealing Algorithms in the Optimization of Earth Dams Cross Section (Case Study: Barzok Dam)

In this study performance of evolutionary optimization methods for designing of cross section of heterogeneous earth dams is investigated. The methods applied were Artifical Fish Swarm (AFSA), Shuffled Complex Evolution (SCE) and Simulated Annealing (SA) algorithms. The model consisted of a nonlinear optimization function by applying different constraints such as slope stability constraints and geometrical dimensions. Design variables in optimization process were the geometrical parameters in cross section of earthen dam and stability safety factors constraints were determined as explicit functions according to design variables by using analyses results such as seepage and slopes stability analysis for a set of sample designs by using linear regression models. Efficiency of the optimization methods in identifying the global optimum point was compared according to mean performance and mean time required for calculations. After optimization of dimensions in Barzok dam by using SCE, AFSA and SA methods, dam volume was reduced 38, 37 and 30 percent respectively as compared to the primary design volume. Results showed that SCE method is more efficient than the SA and AFSA methods in achieving the optimal dimensions in cross section of earth dam.