A new approach for prediction of collapse settlement of sandy gravel soils

Abstract

Collapse settlement of the granular soils has become one of the main concerns of geotechnical engineers since their increased usage for constructing rockfill dams. Behavioral characterization of soil collapse settlement is a challenging task because this phenomenon is influenced by several parameters. In this study, a multi-gene genetic programming (MGGP) approach is proposed for the prediction of the collapse settlement of sandy gravel soils. The main goal is to formulate the collapse settlement and coefficient of stress release (CSR) in terms of sand content, normal stress, shear stress level, and relative density. The proposed models are compared with linear regression and ANN models. The results indicate that the MGGP models have superb accuracy and efficiency while they are simple and practical. The contribution of each parameter is evaluated through a sensitivity analysis. A parametric study is performed, and the obtained results are discussed.