Geomechanics and Engineering

Volume 42, Number 6, 2025, pages 409-423

DOI: 10.12989/gae.2025.42.6.409

Role of geogrid-encased deep soil mixing columns in enhancing foundation performance on the reinforced collapsible sandy Sabkha soil

Mohamed Elsawy , Abderrahim Lakhouit , Turki S. Alahmari , Hossam AbdelMeguid , Mahmoud Shaban

Abstract

The current research aims to reinforce collapsible Sabkha soil by encased deep soil mixing columns (EDSMCs). Full-scale three-dimensional numerical models are created to analyze the performance of footings on both untreated and treated soil. Various parameters such as columns configuration, lime content, collapse index and geogrid stiffness are considered . The results demonstrated that the conventional DSMCs significantly increase the bearing capacity of the collapsible soil under immersion conditions, up to three times that of non-treated soil. However, the bearing capacity of the footing on the reinforced soil still requires further enhancement. Utilizing geogrid encasement for DSMCs improves effectively the foundation bearing capacity, and minimizes the foundation settlement and the columns lateral bulging compared to conventional DSMCs. The minimum settlement and lateral bulging, and the greater loads carried by EDSMCs are achieved when utilizing higher geogrid stiffness . In addition to the numerical analyses, multiple machine learning models including Logistic Regression (LR), Nonlinear Regression (NLR), Support Vector Machine (SVM), Gaussian Process Regression (GPR), Random Forest (RF), Decision Tree (DT), and Extreme Gradient Boosting (XGBoost) are developed. These models exhibited strong performance in predicting the properties of treated Sabkha soil, with Coefficient of Determination (R2-Score) exceeding 0.95. The machine learning analyses support the findings of the numerical analyses, emphasizing the significant role of geogrid encasement in enhancing footing performance on the reinforced soil.

Key Words

collapsed settlement; deep soil mixing columns; geosynthetics; load transfer; machine learning

Address

PDF Viewer

Preview is limited to the first 3 pages. Sign in to access the full PDF.

Loading…