Detailed uncertainty and sensitivity study on OECD-PRISME pool fire experiment PRS-SI-D1

Fire simulations as well as analytical validation procedures have gained more and more significance, particularly in the context of deterministic as well as probabilistic fire safety analyses for operating nuclear power plants (NPP). Meanwhile, fire simulation codes have been adapted as reliable analytical tools for risk-oriented fire safety assessments. The use of fire simulation results could be demonstrated, on the one hand, for improving and upgrading fire protection by the licensees and, on the other hand, as a tool for reproducible, traceable and clearly understandable estimations within the frame of assessing the available and/or foreseen fire protection means by the regulators and their experts. To assess quality and predictability of fire simulation codes these codes are validated against different experiments. In earlier times conservative assumptions related to different phenomena and boundary conditions were often used. Due to the complexity of fire scenarios leading to possible contradictions of conservative assumptions this procedure is not straight forward so that nowadays so-called best-estimate simulations are performed. Using this kind of simulations additional knowledge about uncertainty bands for different results is needed. In this contribution recent COCOSYS results of a detailed uncertainty and sensitivity study using the GRS tool SUSA will be presented. The study is based on the simulation of the PRS-SI-D1 fire experiment inside the DIVA facility at IRSN (Institut de Radioprotection et de S urete) in Cadarache (France). This experiment is part of the OECD PRISME (Propagation d'un incendie pour des scenarious multi-locaux elementaires) project. (orig.)