Prediction of bearing capacity for thin-wall spread foundations using ICA-ANN predictive model

Abstract

Thin - wall spread foundations are used in places where the soil has relatively low strength. In this paper , a literature review was conducted to investigate the beneficial effect of providing thin - walls to spread foundations on bearing capacity. Overall, the literature suggest that , in terms of bearing capacity, thin - wall foundation works better compare d to surface (simple) foundations. Nevertheless, due to the fact that famous bearing capacity equations are proposed for conventional footings rather than thin - wall footings, developing a predictive model in this regard is advantageous. Therefore, in this study, apart from the relatively extensive literature review, an effort was made to develop a predictive model of bearing capacity for the aforementioned footings using an artificial neural network (ANN) technique enhanced with imperialist competitive algorithm (ICA). For this reason, a relatively large dataset comprising 149 recorded cases of thin - wall footing load tests were compiled from literature. The dataset consisted of footing width, wall to width ratio of footings, internal friction angle and unit weight of soil as well as bearing capacity of footings. Apart from the latter, other parameters were used as input parameters of the predictive model. The correlation coefficient and mean square error equal to 0.9 5 and 0.01 for testing data respectively indicate the relative reliability of the ICA - based ANN predictive model of bearing capacity for thin - wall spread foundations.