Advancing Bentonite Material Modelling in the KI-Stoff Project
| Project leader | Project manager | Project duration |
|---|---|---|
| Prof. Dr. Thomas Nagel (Soil Mechanics sub-project) | M.Sc. Aqeel Afzal Chaudhry (Soil mechanics sub-project) | 01.02.2025 - 31.01.2028 |
About the project
Background
Starting in February 2025, BGE TECHNOLOGY GmbH (BGE TEC) will coordinate the KI-Stoff project—a collaborative initiative involving research teams from BGE TEC, ³Ô¹ÏÍø, and TU Braunschweig. At the Chair of Soil Mechanics at ³Ô¹ÏÍø, our involvement focuses on refining material modelling approaches for bentonite, which is crucial for the safety assessment of radioactive waste repositories.
Bentonite is widely used as a sealing and buffering material; however, its complex behaviour under varying environmental conditions presents challenges for current material models. Rather than creating new constitutive laws, the project adopts modern inverse modelling methods to automatically select and calibrate the most suitable existing material model. This process integrates full-field measurements gathered from both laboratory and in-situ experiments, considers full hydraulic-mechanical coupling, and incorporates advanced artificial intelligence techniques to enhance the selection and calibration process.
By rigorously quantifying uncertainty and applying state-of-the-art inverse modelling combined with machine learning, our goal is to narrow the gap between experimental observations and numerical simulations. The automatic model selection framework is designed to reliably identify the best-fitting model from a predefined collection, ensuring that our predictions are robust and aligned with experimental data. This targeted and scientifically grounded approach aims to enhance the reliability of post-closure safety assessments in radioactive waste disposal.
At ³Ô¹ÏÍø, the Chair of Soil Mechanics is dedicated to advancing our understanding of bentonite behaviour. Our contribution to the KI-Stoff project reflects a commitment to employing modern, data-driven techniques—including machine learning—to improve material modelling and support safer radioactive waste management.
Beteiligte Projektpartner
BGE TECHNOLOGY GMBH
Institut für Rechnergestützte Modellierung im Bauingenieurwesen - Technische Universität Braunschweig
