Artificial neural network methodology: application to predict plasticity of clay soil treated with sand

Clay soils consist of fine particles and are generally characterized by low strength. The clays and the water interactions result to clays high-plasticity that can easily deform or crack. It should be noted that the sand addition to expansive soils can help to improve their particle size, reduce their plasticity, and consequently increase their strength. The present study aims mainly to develop prediction model of the plasticity index (PI) using the artificial neural network (ANN) method for soil treated with sand at various contents. For the purpose of predicting the plasticity index values of sand-treated soils, the experimental results were conducted on 38 soil samples, collected and thoroughly analyzed. The experimental investigation carried out on the clayey soil from the region of Medea, Algeria and treated with dune sand. It was decided to consider three inputs, namely the plastic limit, liquid limit, and sand percentage, while there was only one output, i.e. the plasticity index. The proposed model performance has been validated by the experimental test results and show that the relative error is very low (MRE = 1.06%). The proposed model could be used for predicting the plasticity index of treated clayey with sand.