Modeling the mechanical properties of rubberized concretes by neural network and genetic programming

Using the discarded tire rubber as the concrete aggregate may be a solution to disposal of this waste material. However, presence of the rubber aggregates in the concrete mixture decrease the mechanical properties of such concretes depending mainly on the type and the content of the rubber used. In this paper, neural network (NN) and genetic programming (GEP) based explicit models are proposed for the prediction of mechanical properties of rubberized concretes. Data used in both training and testing of NN and GEP models were obtained from an experimental study containing a total of 70 rubberized concretes. The models were constructed using eight design variables and one response as the inputs and output, respectively. Compressive strength, splitting tensile strength, and static elastic modulus of the concretes were employed as the outputs of the models developed in this study. It is found that both NN and GEP provided high prediction capability with certain accuracy. The proposed formulations also showed perfect agreement with the experimental study, thus leading to beneficial and practical estimation of the mechanical properties of the rubberized concretes.