Modeling of buoyancy driven flow-mixing experiments in Rocom test facility using 3D capabilities of Cathare code

In the present study, the 3-D features of the thermal-hydraulic system code CATHARE2 were thoroughly assessed against the OECD/PKL-2 ROCOM tests T1.1, T2.1 and T2.2 which focus on the evaluation of the thermal mixing in the reactor pressure vessel (RPV) under asymmetric buoyant cooling loop conditions. Furthermore, the calculation of the recent ROCOM test T2.3 (OECD/PKL-3), which aims at detailed investigation of the thermal-hydraulic behavior inside the RPV during the emergency core cooling injection from accumulator to the cold leg, was performed in the framework of OECD/NEA/CSNI international benchmark activity. A successful application of CFD code to support a set-up of best estimate 3-D SYS-TH nodalization of RPV was demonstrated. This was achieved by using ANSYS CFX code to evaluate the pressure losses throughout the vessel with further application of additional loss coefficients for the sieve drum and core support plate in CATHARE reference model in order to match the pressure drops predicted by the CFD model. The effect of nodalization refinement (axial, radial and azimuthal) of the ROCOM pressure vessel as well as the influence of the singular pressure losses in the sieve drum and core support plate on the predicted results was studied. It was observed that the number of azimuthal nodes is the most influential parameter on the mixing scalar. It also important to notice that the calculations conducted with refined grids show better general agreement with the experimental data. This is due to the fact that an increase of the node numbers causes a decrease of the numerical diffusion, which plays a vital role in the simulations. The performed qualitative analysis has shown the ability of CATHARE 3-D models to capture the main features of the mixing phenomena in reactor pressure vessel using appropriate modelling, however from the quantitative point of view, the effectiveness of the thermal mixing is generally overpredicted. The obtained results are qualitatively comparable to the ones obtained with CFD codes, like ANSYS CFX, CODE_SATURNE, STAR-CCM+, however require considerably less computational resources. It also important to notice that applicability of the findings of the present studies to NPP scale has to be further investigated both experimentally and analytically, due to scaling distortions which are inevitably present in any scaled test facility.