2020

Tyrväinen, Tero; Kling, Terhi; Korhonen, Timo

This report studies probabilistic methods for the analysis of fire risks. In this study, an old nuclear power plant cable room fire scenario has been successfully implemented in a new tool, simulation-based event tree of FinPSA. The main components of the model are Monte Carlo fire simulations and a stochastic operation time model for firefighting. The fire simulations were performed separately using deterministic Fire Dynamics Simulator (FDS), and the results of the simulations were imported to FinPSA. The stochastic operation time model is implemented in FinPSA scripts in eight parts corresponding to different operational phases, including fire detection, guard centre actions, control room actions and fire brigade actions.<br/><br/>The results calculated using FinPSA are approximately same as the results calculated by the old tool, Excel-based Probabilistic Fire Simulator (PFS), but FinPSA offers better model structure, better readability and better maintainability than PFS. Developing complex computation rules is much easier in the script files of FinPSA than it is in the Excel format. FinPSA is therefore considered a useful and practical tool for probabilistic fire risk modelling.<br/><br/>The study may be continued e.g. by more realistic modelling of firefighting, updating the case study for the present needs of PRA-modelling and the latest changes in the control room guidance, analysing failures of multiple cables, modelling fire spreading and analysing uncertainties. The long-term goal is to develop a general-purpose approach for probabilistic risk analysis of fires. It however requires analysis of a wider range of fire scenarios, because other cases may introduce challenges not present in this particular case study.