Enhancing Comprehension through Interactive Visualization of Geological Simulation Results under Uncertainty

Digital models and simulations play a critical role in site selection for a deep geological repository for radioactive waste. One of the key aspects of this process is evaluating the integrity of the containment-providing rock zone, which is vital for ensuring the long-term safety of repository sites [1].  Computational models, which assess the integrity of geological barriers, generate results that are often intricate and difficult to interpret, requiring expert knowledge. The complexity of these model outcomes increases when input parameter uncertainties are considered. As the site-selection process progresses, a growing need exists to make these complex results more accessible to a broader audience. This accessibility is crucial for better decision-making and fostering public trust and acceptance, particularly when considering pre-selected and rejected sites.
The results from deterministic finite element simulations provide detailed data in both space and time for various quantities of interest, such as pore water pressure, stress, temperature, and other integrity criteria. When uncertainty is introduced into the model through stochastic or parametric variations in the input parameters, the model‘s dimensionality increases significantly, complicating the analysis and the understanding of the model outputs.
To make these results more comprehensible, we propose using a surrogate model created through adaptive state-space sampling [2, 3]. This surrogate efficiently captures the full state space of the system, establishing a functional relationship between input parameter values in realistic ranges and the resulting outputs from the finite element simulations. Importantly, no data reduction occurs during this process, allowing for an accurate and complete mapping of the system’s behavior under varying conditions.
Based on this data, a real-time visualization dashboard allows users to interactively explore the effects of input parameter changes within the complete physical and time domains of the model. This tool can significantly enhance the understanding and interpretation of complex geological models, making them more accessible to both experts and non-experts and ultimately supporting better-informed decision-making processes in geological repository site selection.

 

[1] J. Maßmann, J. et al. (2022). Methode und Berechnungen zur Integritätsanalyse der geologischen Barriere für ein generisches Endlagersystem im Tongestein. Projekt ANSICHT-II. Ergebnisbericht. Bundesanstalt für Geowissenschaften und Rohstoffe (BGR).

[2] Bittens, M. (2024). OpenGeoSysUncertaintyQuantification.jl: a Julia library implementing an uncertainty quantification toolbox for OpenGeoSys. Journal of Open Source Software, 9(98), 6725.

[3] Bittens, M. und Gates, R.L (2023): DistributedSparseGrids.jl: A Julia library implementing an Adaptive Sparse Grid collocation method, Journal of Open Source Software 8.83.