Optimisation design of geocell-reinforced high-speed railway embankments

• Investigating the settlement of high embankments under HSR train loads. • Applying an advanced simulation model by Plaxis API with Python remote scripting. • Developing a reliable settlement estimation using the GEP algorithm. • Optimising the design of the GRPS system with geocell reinforcement. The construction of high-speed railway (HSR) embankments on soft ground areas is a geotechnical challenge due to excessive settlement and long-term deformation. Thus, this study evaluates the effectiveness of deep cement mixing (DCM) piles reinforced by a geocell system (GPRS) as a ground improvement method. An advanced automated simulation framework using Python remote scripting was developed to examine the effects of train load, embankment height, soft soil thickness, and the number of geocell layers. Approximately 336 data points on the settlement of the HSR embankment was extracted from Plaxis analysis. The highest equivalent train load was 100 kN/m 2 and the settlement of the embankment fluctuated from 22 to 65 mm. A Gene Expression Programming (GEP) algorithm was then applied to develop a reliable regression among the key variables with high accuracy (R = 0.99) and low errors. The parametric study indicated that the geocell reinforcement could significantly improve the lateral resistances, mobilise the membrane effect within the fill material, and reduce vertical stress and cumulative settlement. The research findings provide critical insights for improving the design process, optimising GRPS reinforcement using geocell, and enhancing predictive capabilities for HSR embankment settlement under various ground conditions.