A five-year research project involving BRANZ and the University of Canterbury has an objective to develop a building design fire tool to simulate building fires. Fire simulation results will be presented in a probabilistic form and will allow the variability and uncertainty associated with the predictions of the fire environment to be quantified. The research is funded by the Foundation for Research Science and Technology, Building Research and the Department of Building and Housing.
A better description of the uncertainty applying to fire engineering outputs for life safety will help ensure that fire risks in buildings are better managed and ensure there is more robustness in fire safety solutions. Reflecting on current practice, the treatment of uncertainty in the existing methods of analysis is usually poor and sensitivity analysis is often forgotten. It is hoped that this research will lead to more transparency and confidence in the level of safety provided in fire engineered designs.
The proposed approach is to use the BRANZFIRE fire simulation software within a larger probabilistic-risk model involving Monte Carlo simulation techniques to generate time-dependent probability distributions for the output variables that describe the survivability of the fire environment. BRANZFIRE is a popular fire zone model used by New Zealand fire engineers in support of alternative fire-engineering solutions. It provides time-dependent predictions of smoke layer height, temperatures and concentrations of combustion products based on a user-specified design fire (e.g. rate of heat release).
The research will include the development of an extensive database of building contents and prediction of item to item fire spread. The reliability and efficacy of fire safety systems (active and passive) will also be included. Fire scenarios will be constructed by randomly selecting an initial burning object and its location (appropriate to the type of occupancy e.g. residential, office, retail) and then modelling the fire spread to adjacent fuel packages along with calculation of critical measures of survivability. The tool would require hundreds or thousands of simulations each representing a possible outcome of the fire event and would generate outputs in the form of probability distributions that could be used to demonstrate compliance with a ‘probabilistic statement of performance’. A possible example of a probabilistic statement of performance for life safety might be - ‘The design must allow for a 0.9 probability of having 20 minutes or more before a smoke layer height falls below 2 m’.
The researchers will include staff from both BRANZ and University of Canterbury along with the involvement of ME and PhD fire engineering students.