Application of gene expression programming for modeling bearing capacity of aggregate pier reinforced clay

Utilizing the aggregate piers is one of the methods to improve and increase the bearing capacity of soft soils. The ultimate bearing capacity of these piers is affected by parameters such as the physical properties of the piers, structural conditions, the embedment depth and replacement ratio of piers, which complicates the estimation of bearing capacity. In this study, the Gene Expression Programming method was used for the prediction of the ultimate bearing capacity of clay soils reinforced with aggregate piers. For this purpose, two different models were developed, of which the first model (GEP2) utilized two input variables, the undrained shear strength of clay (Su) and replacement ratio (ar), while the second model (GEP4) used four input variables including the undrained shear strength of clay (Su), replacement ratio (ar), slenderness ratio (Sr), and embedment depth of piers (df). The coefficient of determination of the GEP2 model, and the GEP4 model is 0.921 and 0.942, respectively. Besides, comparing the GEP4 model of this research with the developed models of previous studies confirms the superior performance of the GEP4 model, considering both the accuracy and number of input parameters. The results of sensitivity analysis showed that the undrained shear strength of clay (Su), replacement ratio (ar), slenderness ratio (Sr), and embedment depth of piers (df) have the highest impact on the prediction of bearing capacity, respectively. Furthermore, the parametric analysis demonstrated that increasing the Su, ar, Sr, and df would improve the bearing capacity of the aggregate piers reinforced clay.