Computation of Quantitative Fire Risks Using Two-Model Monte Carlo Simulation and CFD Fire Model

Quantitative fire risk analysis requires integration of uncertainties and variability of the input data to computation of the fire consequences. In industrial environments with complicated geometries and fire loads, the consequence assessment must usually be carried out by fire simulation based on Computational Fluid Dynamics (CFD). Yet, integration of the probabilistic description of the input data and the computationally demanding CFD is a challenging task. We present a new Two-Model Monte Carlo (TMMC) technique enabling to perform Monte Carlo simulation in conjunction with computationally expensive tools like CFD. In TMMC, two models of different accuracy are used to simulate the same problem so that during the simulation, the input-distribution random space is covered by the less accurate but faster model and in a post-processing phase, a correction is made based on a smaller number of simulations using the more accurate model. The technique is validated using artificial cases and applied to assessment of fire-induced failure probabilities and time available for fire fighting in a nuclear power plant (NPP) electronics room using the FDS code. Finally, implementation of the TMMC results to NPP PSA systems is addressed using a classification approach of the NPP rooms according to their fire and risk characteristics.