Influence of Dilatancy Behavior on the Numerical Modeling and Prediction of Slope Stability of Stabilized Expansive Soil Slope

Soil dilatancy behavior is crucial in the numerical modeling of slopes for safety factor (SF) computation. The effect of dilatancy on the SF of stabilized expansive soils has been rarely reported. The present study thus focused on the effect of dilatancy on the numerical modeling (NM) of coconut shell ash (CSA)-treated expansive soil slopes using the strength reduction finite element analysis in PLAXIS for SF computation. The NM input parameters were mainly obtained from laboratory experiments performed for the stabilization of the expansive soil using CSA. The degree of non-associativity, which was incorporated during the NM for SF computation, was considered as a crucial variable to check the dilatancy effect on the SF. Predictive modeling of the SF was subsequently done using multiple linear regression (MLR), artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS) and support vector regression (SVR). The results obtained show that the dilatancy has a pronounced effect for overconsolidated clays. Furthermore, consideration of the degree of non-associativity as a variable quantifying the dilatancy behavior resulted in better predictions of SF. Through parametric studies, optimal parameters for the ANN, ANFIS and SVR that affected the SF prediction were established. Overall, the ANN and ANFIS had superior predictive capability when compared with the MLR and SVR.