Optimization Design of Stiffened Deep Cement Mixing Piles and Geogrid for Ground Improvement to Support High-Speed Railway Embankments in Central Vietnam

High-speed railways (HSR) have been developed remarkably in recent decades as a major public transportation mode in many countries. The stability of embankments under HSR operation is a critical and technical challenge. Thus, this study aims to investigate the effectiveness of stiffened deep cement mixing piles (SDCM) combined with geogrid to minimize embankment settlement under HSR conditions. PLAXIS 2D axisymmetric models were created and automatically controlled using the Python wrapper for remote scripting to investigate a wide range of input variables and several combination scenarios. Selected input parameters include the thickness of the soft soil layer, SDCM pile dimensions (diameter and length), embankment height, and train load characteristics (axle loads, dynamic factors, train speeds, and dead loads). As a result, approximately 324 points of data on the settlement of the HSR embankment were extracted from FEM analysis conveniently. Besides, a Gene-Expression Programming (GEP) algorithm was applied to generate a reliable model for estimating the settlement of the HSR embankment. A parametric study indicates that the HSR embankment settlement increases considerably whether the soft soil layer is thick, the embankment is high, or the train is heavy and fast. In contrast, enlarging the diameter and the length of SDCM piles could significantly reduce overall settlement. The study offers practical insights for reducing design efforts using advanced FEM analysis for HSR embankments. These findings are a valuable resource for engineers involved in upcoming HSR projects in Vietnam.