THM-TH Abstraction of HLW Nearfield

The evolution of the nearfield around a disposal tunnel for high-level radioactive waste (HLW) depends on complex coupled thermal-hydrological-mechanical processes. However, the compu-tationally expensive performance assessment (PA) calculations are preferably conducted using a simpler conceptual model that only explicitly accounts for thermal-hydrological (TH) processes. This simplification requires a THM-to-TH abstraction step. It is proposed that THM processes can approximately be represented by determining effective TH parameters. These effective parameters are estimated by the inversion of data reflecting THM processes using a TH model. The validity of this process is examined by generating synthetic THM data – calculated by a THM simulation of a two-dimensional cross section of a HLW disposal tunnel – and inverting the synthetic THM data using a TH model of the same cross section.

The THM numerical model was developed using Code_Aster (EDF 2024), which simulates the coupled non-isothermal two-phase flow and hydro-mechanical phenomena associated with repository-induced effects in the nearfield of an SF/HLW repository. Estimation of the effective parameters is performed using the iTOUGH2 simulation-optimization framework (Finsterle et al. 2017), which provides inverse modelling capabilities for TOUGH2 (Pruess et al. 2012), a code that simulates multiphase, multicomponent, non-isothermal fluid flow in porous media.

The primary goal of the modelling study was to obtain an abstracted iTOUGH2 TH model that could effectively predict the maximum pressures in the Opalinus Clay. The impact of potential sources of uncertainty in the iTOUGH2 model on predictions of interest was evaluated through Monte Carlo simulations with Latin Hypercube Sampling. It was demonstrated that the impact of overall model uncertainty, such as small amounts of uncertainty in the permeability of geological formations, affects prediction uncertainty far more than the small amount of uncertainty introduced from the model abstraction step.

Another important achievement was, that estimating effective pore expansivity and pore compressibility coefficients (a) from high-fidelity THM models, or (b) directly from site charac-terisation data (preferred) is an acceptable THM-to-TH abstraction approach. Uncertainties in the predicted maximum overpressures caused by uncertainties in the estimated, effective pore expansivity and pore compressibility coefficients are acceptable; the influence of residual errors from the model abstraction process on predictions of interest (e.g., maximum overpressures) is considerably smaller than the influence of random and systematic errors caused by the uncertainties on other parameters and model assumptions.