Plot for first- and second-order sensitivity indices for a set of observed quantities in a THM problem. Circles along the diagonal represent S1, while non-diagonal units (with the exception of crosses) show interactions S2 between parameters.

Uncertainty quantification and sensitivity analyses in THMC models for geotechnical safety assessment

Coupled thermo-hydro-mechanical-chemical (THMC) models are used for the assessment of nuclear waste disposal, reservoir engineering, and geotechnical engineering. With the availability and continuous increase in computational resources, the tendency to employ increasingly complex mathematical models for geotechnical safety assessment is becoming more prevalent. Among others, an important aspect of this increased complexity is the increased number of model inputs. When the models in question describe complex physical phenomena in the subsurface on large spatial and temporal scales, these parameters can usually not be determined without considerable uncertainty or imprecision. Deterministic analyses with a selected parameter set only insufficiently open up the spectrum of possible system responses and provide only a limited view during the evaluation of the analyses. Thus, model-based decision making and optimization require sensitivity analyses (SA) and uncertainty quantification (UQ). Assessment of the suitability of different UQ and SA methods for coupled THMC problems on an engineering scale is required. The objective of this work is to systematically compare the information gained by different approaches to sensitivity analysis using both local and global approaches. The analyses are performed for different spatio-temporal settings to observe both near and far-field effects as well as early- and late-stage system response and link them to monitoring approaches. We analyze which parameters and parameter interactions control the results in these different domains and how this knowledge can inform an enhanced physical interpretation.