Optimal design of soil–cement columns to stabilise soft soil under seismic loading using ANN and GWO

This study mainly sought to evaluate the performance of soil–cement (SC) columns in the stabilisation of soft soils under seismic loads and to develop a comprehensive model for the optimal design of SC columns through experimental and numerical analyses. The work was divided into two phases: (i) SC column performance evaluation and (ii) optimal SC column design model development. The first phase experimentally evaluated the performance of SC columns in stabilising soft soils under two conditions, including a rigid overburden and embankment. A numerical model was developed using the finite-element method in ABAQUS and validated. The numerical model showed an error below 10% under static and dynamic conditions. The second phase developed a model through grey wolf optimisation for the optimal design of SC columns to stabilise soft soils under seismic loads. To evaluate the responses of SC column–stabilised soil, an artificial neural network model was developed to measure the settlement. It was found that the average soil settlement decreased by 62.7% under a rigid overburden in five seismic scenarios. The proposed model showed acceptable performance in determining optimal SC column parameters in the seismic strengthening of soft soils and could be exploited as a comprehensive framework.