The understanding of the reflood process of a severely damaged reactor core represents a challenge in the prediction of safety margin of existing and future pressurized water reactors. In case of loss of coolant accident with no availability of safety injection systems, the severe accident scenario assumes that the reactor core heats up due to the residual power. This leads to fragmentation of fuel rods and melting of reactor core materials that can result in the formation of a “debris bed”. In a debris bed the particles might reach few millimeters (characteristic lengthscale: 1 to 5 mm). All accident mitigation possibilities should be used to avoid the progression of
the core degradation and the radioactive materials into environment. In particular, if the water source is renewed, the injection of water into the damaged reactor core could, in certain conditions, mitigate the progression of a severe accident. On the other hand, the injection of water (reflood) enhances the hydrogen formation and the risk damage of containment. Moreover, it could contribute to the pressure increase in the primary circuit. After the TMI-2 accident in 1979 that resulted in partial melting of the core and the debris formation, the coolability of a heat-generating debris bed was studied experimentally and theoretically. The first experiments were made at Brookhaven National Laboratory and University of Los Angeles in the 80’s. These experimental programs proposed yet an interesting database for top and bottom reflood analysis. Currently, the French “Institut de Radioprotection et de Sûreté Nucléaire” (IRSN) sets up two experimental facilities, PEARL and PRELUDE. The series of PRELUDE experiments achieved in 2010 constitutes a significant complement to the database of high temperature bottom reflood experimental data. They provide relevant data to understand the progression of the quench front and the intensity of heat transfers. On the basis of those experimental results, the analyses of the thermal hydraulic features at the quench front have been analysed. The intensity of the maximum heat flux is estimated, as well as the front velocity. From that, a reflooding model applicable to porous media was improved. The model is implemented into a computer code for severe accident analysis ICARE-CATHARE. The proposed two-phase flow model is validated with PRELUDE experimental results. A comparative assessment with respect to previous experimental data is also briefly provided.