Using the Response Surface Method and Artificial Neural Network to Estimate the Compressive Strength of Environmentally Friendly Concretes Containing Fine Copper Slag Aggregates

Replacing natural sand with other waste materials will not only reduce the environmental damage caused by the destruction of mountains and rivers (to produce natural sand), but will also solve the problems of disposing of waste materials in nature. Hence, effort has been made in this study to propose a model to estimate the compressive strength of green concretes containing fine copper slag aggregates to reduce the sample making/testing cost/time using the artificial neural network (ANN) and response surface method (RSM). To this end, use was made of a set of 318 available experimental data including such input parameters as the ratios of water to powder material, powder material to total aggregates, fine aggregates to total aggregates, copper slag, and curing age. According to the results, a single-layer ten-neuron ANN and a fifth-order polynomial RSM predicted the compressive strength of this type of concrete quite well. A better RSM correlation coefficient (R-2 = 0.97) compared to that of the ANN model (R-2 = 0.92) indicated a proper correlation between the predicted and experimental results.