HERMES#DITOCO2023: Integrating Process Modelling into Digital Twin Concepts

Sergey Churakov, Eric Laloy, Enrique Garcia, Anne-Catherine Dieudonné, Nikolaos Prasianakis, Attila Baksay, Javier Samper, Jan Brezina, Lars Bilke, Réka Szőke, Olaf Kolditz

Digital twins are the basic concept for creating digital representations of real systems, such as deep geological repositories for radioactive waste disposal (Kolditz et al. 2023). This involves continuous data collection and integration, model prediction, and the use of virtual reality (VR) methods to represent and interact with complex systems. Process simulation plays an important role in digital twin concepts, giving digital twins predictive power in addition to representing the actual state of the system. This is a fundamental characteristic of digital twins and a strong competitive advantage for modern and innovative planning tools.

DITOCO2030 is a strategic study in the EURAD-2 programme that aims to develop a common understanding of the specific requirements of different disciplines and stakeholders within the general digital twin framework (Szoke et al. 2025). The HERMES work package aims to develop high-fidelity numerical models for the simulation of strongly coupled thermo-hydro-mechanical-chemical (THMC) processes in the near-field of repositories, for the optimisation of repository design, and for the interpretation of mock-up experiments, using a combination of physics-based models and accelerated computing supported by machine learning and artificial intelligence (Churakov et al. 2024). HERMES is divided into following main tasks: simulation of THMC processes,  development of suitable surrogate models, and finally application of process modelling for safety and performance assessment for the safe deep geological disposal of radioactive waste.

Integration of HERMES components into DITOCO2030 should take place on several levels and steps. First examples are (1) The automation of computational steps: repository systems are so complex that the individual repository parts and components (e.g. emplacement drifts and sealing structures) have to be modelled separately before an overall model of the entire repository system can be developed (Samper et al. 2024); (2) Flexible  simulation methods: Various simulation methods for THMC processes will be made available for plausibility and functionality testing. In addition to the usual code comparison (e.g. in DECOVALEX), process- and data-based (machine learning) methods (Prasianakis et al. 2025) will be integrated into HERMES for the first time. Technically, this will take place within the HERMES Model Hub. These will be developed and tested in (1) and (2) and then implemented in a generalised form in DITOCO2030.

References:

Churakov et al. (2024): Environ. Earth Sci. 83 (17), art. 521, 10.1007/s12665-024-11832-7

Kolditz et al. (2023): Digitalisation for nuclear waste management: predisposal and disposal. Environ. Earth Sci. 82 (1), art. 42, 10.1007/s12665-022-10675-4

Prasianakis et al. (2025): Geochemistry and machine learning: methods and benchmarking. Environ. Earth Sci. 84 (5), art. 121, 10.1007/s12665-024-12066-3

Samper  et al. (2024): Multiphase flow and reactive transport benchmark for radioactive waste disposal. Environmental Earth Sciences, 83 (22), art. 619, 10.1007/s12665-024-11887-6

Szoke et al. (2025): DITOCO - Digital Twins (DT) to support Optimisation (including communication of safety), Construction and Operation of radioactive waste management facilities abstract submitted to SafeND 2025

Co-funded by European Union under Grant Agreement n° 101166718 and Research Council of Norway under the International Calls International Collaborative Project - Project number 355507