Our society uses the geological subsurface both as a source of various natural resources and for the storage and disposal of different types of human waste, such as nuclear waste. However, a complete picture of the subsurface is often unavailable, as boreholes or geophysical surveys are sparse and reveal only some aspects. Therefore, complementary physics-based simulations are indispensable for reconstructing the complete picture of the subsurface, including various thermofluid-mechanically coupled processes such as groundwater flow and temperature distribution. The simulations provide a way to forecast future system behaviors, such as the potential distribution of groundwater contaminants. While these simulation models require calibration against observed and measured data, identifying which datasets are most informative for this task remains a challenge. Additionally, the high computational cost of the full-complexity physical models can limit the ability to assess a wide range of possible outcomes.
In the context of nuclear waste repository site selection, the project aims to optimize data acquisition and 'smart monitoring' strategies. The main objective of the project is to develop a systematic framework for evaluating different data acquisition strategies with regard to their value added for a specific requirement, such as the quantification of radioactive contamination within a region of the subsurface due to contaminated water. These methods will lead to the development of smart-monitoring strategies, which are critical for ensuring the long-term safety of subsurface nuclear waste repositories.
Within this project, we have three main goals:
- The development and provision of benchmark scenarios for the possible host rocks of a nuclear waste deposit: claystone, salt, and crystalline rocks. Parameter estimation, optimization of data acquisition, and the stability of forecasts shall be systematically analyzed based on these scenarios.
- The implementation of a model cascade, which includes both the thermo-fluid-dynamical processes in the subsurface and subsequent models, such as the spatial distribution of the accumulated radiation level due to contaminated groundwater. The automatization of this process allows for simulations for different scenarios as well as the following statistical analysis and evaluation, for example, in terms of radiation zone maps.
- The development of a modern and robust parameter estimation and data assimilation based on meta models and Bayesian Active learning. Beginning with methods of optimal experimental design, optimal data acquisition, and smart-monitoring strategies, these will be developed.
