Prediction of Swell Pressure of Expansive Soils using Multi-Expression Programing

Abstract

This project presents the development of a new empirical prediction model to evaluate swell pressure of expansive soils (Ps-ES). An extensive database comprising 168 Ps records was established after a comprehensive literature search. The performance of developed model was tested using mean absolute error (MAE), root squared error (RSE), root mean square error (RMSE), correlation coefficient (R), regression coefficient (R2). The results in the increasing order of the contribution of each input parameter is in the order of OMC (28.27) > PI (27.59) > CF (14.59) > MDD (12.59) > SP (10.40) > silt (6.55). The MEP model outperformed the other AI models found in literature for the prediction of swell pressure in terms of closeness of training, validation and testing data set with the ideal fit slope. The findings of this study can help researchers and designers to evaluate the swell characteristics of the expansive soils in pre-planning and pre-design phases of a construction project and for validation of the laboratory and field test results.