// Closure of Underground Cavities, Long-term Safety Studies
KI-Stoff
AI-supported Material Modelling of Bentonite behaviour
The research project “KI-Stoff” focuses on developing a methodology for the automated selection and adaptation of material models for bentonite using machine learning (ML). Bentonite materials are an integral part of the geotechnical barriers in most repository concepts worldwide and exhibit complex behavioural patterns such as non-linearity, heterogeneity, and coupled interactions. The process of developing material laws is therefore extremely demanding and time-consuming. In addition, the properties of bentonite are subject to natural variations depending on its occurrence, meaning that it has not yet been possible to develop a universally valid constitutive law. The use of ML opens up new possibilities that can contribute to accelerated model selection, parameter identification, and further development of material models.
A major challenge is to select and calibrate a material model suitable for new measurement data of different bentonite types. The aim of this project is therefore to develop an ML methodology that automates this process at various points and thus provides the best possible support for modellers and engineers. At the same time, model comparisons can be systematised on this basis.
The ultimate aim of the research project is to reduce the gap between experiments and simulations for bentonite.
Contact
Research & Development
info@bge-technology.de
Short Infos
Runtime: 2025 - 2028
Client:
Federal Ministry for the Environment, Nature Conservation, Nuclear Safety and Consumer Protection (BMUV)
Partner:
Technischen Universität Bergakademie Freiberg (TUBAF), Deutschland
Abteilung Datengetriebene Modellierung mechanischer Systeme (IRMB-DDM-TUBS), Deutschland
Publications
// Conference paper
Automated Generation of Site‑Specific Discrete Fracture Networks: Mixture‑Model Orientation Clustering and Fracture‑Count Calibration
Author(s): U. Kelka, C. Müller, A. Monnamitheen, P. Herold
In: European Geosciences Union (EGU) (ed.) EGU General Assembly 2026. Vienna, Austria, 3. – 8. Mai 2026.
Language: Englisch
Publication number: 10.5194/egusphere-egu26-17621
04.05.2026
Kelka, U.; Müller, C.; Monnamitheen, A.; Herold, P. (2026): Automated Generation of Site‑Specific Discrete Fracture Networks: Mixture‑Model Orientation Clustering and Fracture‑Count Calibration. In: European Geosciences Union (EGU) (ed.) EGU General Assembly 2026. Vienna, Austria, 3. – 8. Mai 2026.. KI-Stoff: AI-supported Material Modelling of Bentonite behaviour. 10.5194/egusphere-egu26-17621. BGE TECHNOLOGY GmbH (BGE TEC). Vienna
// Conference paper
Automatised model calibration for benonite buffer material
Author(s): T. Gerasimov, U. Kelka, W. Schmidt, et al.
In: Helmholtz-Zentrum für Umweltforschung GmbH (UFZ) (ed.) 5. Tage der Standortauswahl 2026. Leipzig, 23. – 24. April 2026. Bundesgesellschaft für Endlagerung (BGE); Technische Universität Dresden. Leipzig. pp. 132 - 134
Language: Englisch
23.04.2026
Gerasimov, T.; Kelka, U.; Schmidt, W. A.; et al. (2026): Automatised model calibration for benonite buffer material. In: Helmholtz-Zentrum für Umweltforschung GmbH (UFZ) (ed.) 5. Tage der Standortauswahl 2026. Leipzig, 23. – 24. April 2026. Bundesgesellschaft für Endlagerung (BGE); Technische Universität Dresden. Leipzig. pp. 132 - 134. KI-Stoff: AI-supported Material Modelling of Bentonite behaviour. BGE TECHNOLOGY GmbH (BGE TEC), Bedretto Underground Laboratory for Geosciences and Geoenergies (Bedretto URL), Technischen Universität Bergakademie Freiberg (TUBAF), Helmholtz-Zentrum Dresden-Rossendorf (HZDR). Leipzig
// Newsletter
BGE TECHNOLOGY NEWS 2025 Q4
Author(s): J. Froböse, et al.
Fourth newsletter of BGE TECHNOLOGY GmbH in 2025
- 25 Years of BGE TEC: Scientific Exchange and International Dialogue
- KI-Stoff Update: AI-assisted Calibration of the VIBE Model
- Launch of the KerWerk Research Project
- The Fourth Six Years of BGE TECHNOLOGY GmbH (2019–2025)
Language: Englisch
25.03.2026
// Newsletter
BGE TECHNOLOGY NEWS 2024 Q4
Author(s): T. von Berlepsch, et al.
Fourth newsletter of BGE TECHNOLOGY GmbH in 2024
- BGE TECHNOLOGY GmbH supports BGE in Area Screening in the Site Selection Procedure
- Accident Analyses for the Operating Phase of the KONRAD Mine – Effect of Fire on Structural Stability?
- Evaluation of the Impact of Metal Corrosion on the Integrity of Waste Containers in Crystalline Rock (R&D Project BEnKo)
- AI-supported Material Modelling of Bentonite behaviour (R&D Project KI-Stoff)
- Crushed Salt as Engineered Backfill Material (R&D Project MEASURES)
Language: Englisch
20.12.2024