Optimal mix for high performance concrete by evolution strategies combined with neural networks

Abstract

There is a growing interest in the application of non-traditional methods such as simulated annealing (SA) and genetic algorithm (GA) and evolution strategies (ES) for optimization of structural systems. In this paper, evolution strategies (ES) is used to find the optimal mix design for high performance concrete (HPC) comprised of cement, sand, coarse aggregate, water, silica fume and super pasticizer. It is required to get the optimal mix for strength for 120 MPa and slump of 120 mm. In order to get the equation for strength and slump, the sequential learning neural network (SLNN) proposed by Zhang and Morris is used. The cost function to be minimized is the cost of HPC/unit weight of HPC subjected to strength and slump constraints. It is concluded that the method proposed is highly suitable for getting the optimal mix for high performance concrete in practice.