This study investigates the influence of primary variables selection on modeling non-isothermal two-phase flow, using numerical simulation based on the full-scale engineered barrier system (EBS) experiment conducted at the Horonobe Underground Research Laboratory (URL) as part of the DECOVALEX-2023 project. A thermalhydraulic coupled model was validated against analytical solution and experimental data before being applied to simulate the heterogeneous porous media within the EBS. Two different primary variable schemes were compared for discretizing the governing equations, revealing substantial differences in results. Notably, using capillary pressure as a primary variable instead of saturation resulted in closer alignment with analytical solutions and real-world observations. While the modeling work at the Horonobe URL generally exhibited trends consistent with experimental data, discrepancies were attributed to the operational conditions of the heater and the influence of the Excavation Damaged Zone (EDZ) near the borehole.
