Probabilistic performance analysis of TDM column-supported embankments considering soil spatial variability

Abstract

This study presents a numerical analysis method that considers the spatial variation of both surrounding soil and soil-cement materials in evaluating the performance of embankments supported by T-shaped deep cement mixing columns (TDM). The approach applies spatial variability simulation combined with probabilistic analysis using the Adaptive Kriging Monte Carlo Simulation (AK-MCS) method. The study found that considering the spatial variability of soil and soil-cement materials significantly affects the probability assessment of embankment system failure. Although deterministic analysis results indicated that both embankments supported by deep cement mixing (DCM) and TDM columns meet the specified deformation safety criteria, probabilistic analysis clearly showed that both systems have high failure probabilities (exceeding specified criteria), especially regarding maximum settlement. DCM columns demonstrated better efficiency in controlling failure probability in terms of maximum settlement and lateral movement, while TDM columns were superior in reducing the failure probability related to differential settlement between columns and surrounding soil. This study highlights the importance of probabilistic failure analysis that considers spatial variability for efficient and safe design and performance evaluation of embankments on soft clay.