Prediction of Pile Response in Lateral Spreading Soil Using Multigene Genetic Programming

The present study investigates the performance of single fixed-head vertical and batter piles (10° and 20° batter angles) subjected to lateral spreading displacements adopting an integrated approach of multigene genetic programming (MGGP) with design of experiments. Three-dimensional fully coupled numerical analyses are carried out using the pressure dependent multiyield (PDMY02) material model considering variation in permeability of liquefiable soil. The parameters influencing the response of pile foundation (maximum bending moment and pile head displacement) are initially screened using the fractional factorial design approach. Central composite design (CCD) is then adopted to generate the design matrices for conducting detailed numerical analyses. Using the output, prediction models are developed for responses of vertical piles and batter piles (both positive and negative) in laterally spreading soil using MGGP. Further, parametric analyses are conducted to study the influence of various parameters on pile response in laterally spreading soil conditions. Finally, the performances of vertical and batter piles are compared by estimating the reliability index using the Monte Carlo technique.