Numerical Analysis of the Behavior of Geopolymer GCLs in Radioactive Waste Landfill Liners under a Machine Learning Approach

Numerical Analysis of the Behavior of Geopolymer GCLs in Radioactive Waste Landfill Liners under a Machine Learning Approach

Amirmasoud Naderi1 Ramin Nemati Amooghin2 Maryam Hematian Dehkordi3

1) M.Sc. in Civil Engineering – Geotechnical Engineering, Faculty of Engineering, Shahrekord University, Shahrekord, Iran
2) M.Sc. Student in Civil Engineering – Geotechnical Engineering, Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran
3) Ph.D. in Urban Planning, Faculty of Art and Architecture, Kerman Branch, Islamic Azad University, Kerman, Iran

IOI Code : AACA-ABBZZZ
Publication : 10th.International Conference on Researches in Science & Engineering & 7th.International Congress on Civil, Architecture and Urbanism in Asia(10icrsie.com)
Abstract :
    This research aims to analyze the coupled thermo-hydro-mechanical-chemical (THMC) behavior of geopolymer clay liners (GCLs) under the influence of radioactive radiation and long-term periods. A combined approach of numerical modeling based on the finite element method in PLAXIS 2D software and the Random Forest machine learning algorithm is employed. A two-dimensional model with dimensions of 100×20 meters, utilizing a 15-node triangular mesh and coupled flow-deformation analysis, simulated four main scenarios (low/high radiation levels and 2/10-year periods). Output data, including settlement, pore pressure, and permeability, were extracted. The Random Forest algorithm was trained on this data and, with low RMSE values (0.018–0.022 meters for settlement), predicted nonlinear patterns. Results indicated that increased radiation raises settlement by 45–75% and pore pressure by 25–44%, while permeability decreases by 30–40%, confirming the chemical stability of geopolymers. Feature importance analysis highlighted the dominant role of elasticity modulus (0.52 for settlement) and radiation (0.50 for permeability). This approach enhances prediction accuracy by 30%, reduces computational time by 50%, and provides a basis for optimizing cover designs and mitigating environmental risks, which can contribute to the development of IAEA standards.

Keywords : geopolymer clay liner (gcl) radioactive landfill numerical analysis machine learning and nuclear waste