Assessment of the depressurisation and mixing effects of two spray systems in the PANDA facility, part II – Simulations using GOTHIC8.3(QA)

Abstract

The spray safety system is one of the most effective solutions to limit the pressure and temperature rise in the containment of a nuclear power plant during a design basis accident. However, its actuation under severe accident conditions may have either a positive effect on the hydrogen risk by mixing the containment atmosphere and avoiding high H2 local concentrations or a negative impact by condensing the steam and de-inerting the atmosphere. The experiments performed during the OECD/NEA HYMERES-2 Project aimed to extend the experimental database dedicated to assessing the capabilities of the codes used to optimise the operation of the safety systems. Specifically, the H2P5 series consisted of two experiments that compared the activation of a central full cone spraying nozzle (H2P5_2) with nine smaller spraying nozzles, eight equally distributed at 0.5 m from the metallic walls of PANDA (H2P5_1). The ninth nozzle was positioned in the vessel vertical axis for the latter configuration. These experiments created a demanding scenario for the computational codes, which need to consider three different fields: three components of the gas phase (air, helium, and steam), the dispersed liquid (droplets), and the continuous liquid accumulated on the walls and at the bottom of the PANDA vessel during the test. Furthermore, the proximity of the nozzles to the walls in H2P5_1 requires considering the strong interaction of the fluid with these metallic structures.In this work, the GOTHIC 3D models were able to qualitatively represent all the relevant phase change phenomena affecting the cooling effect of the spray and provided a reasonable qualitative representation of the depressurization behaviour observed during the experiments. However, the code overestimated the steam condensation on droplets. To achieve quantitative agreement between the experiments and computer models it was necessary to enhance the spatial distribution of the droplets (to compensate for anomalies due to a conical injection in a relatively coarse cartesian mesh) and to diminish the steam condensation on the droplets by decreasing the spray mass flow rate. Regarding the helium mixing, the simulations revealed a potential explanation for the faster mixing observed in the single nozzle (H2P5_2) experiment. Furthermore, the droplet diameter showed to be the parameter inducing the most significant variations in the mixing.