A COMPARISON OF A PRIORI MODELLING PREDICTIONS WITH EXPERIMENTAL RESULTS TO VALIDATE A DESIGN FIRE GENERATOR SUBMODEL

摘要

A new fire safety engineering model called B-RISK has been developed in New Zealand asrnpart of a collaborative research project. B-RISK uses iterative Monte-Carlo algorithms to samplernvalues from input parameter distributions, and has a submodel that generates a heat release ratern(HRR) curve as input for each iteration, called a design fire generator (DFG). The DFG submodelrnproduces HRR curves by modelling fire spread between combustible items in a compartmentrnscenario. In this paper part of an experimental validation programme, in which compartment fire datarnis compared with a priori DFG modelling predictions, is described. The validation experiments werernbased on three generic materials typically found in New Zealand residential occupancies: non fireretardantrnflexible polyurethane foam used in residential furniture; ABS plastic sheet used as therncasings for electronic goods; and MDF used in household cabinetry. Firstly, Cone Calorimeter (ISOrn5660) experiments were carried out to generate material data for use as input to the DFG modelling.rnThis was then followed by a series of experiments where mock-ups of single items were tested in arnFurniture Calorimeter to generate free-burning HRR data, also for use in the modelling. Havingrngenerated the necessary input data, a multi-item layout in the ISO 9705 compartment was finalisedrnand DFG modelling conducted. Once this a priori modelling had been completed, actual experimentsrnmatching the multi-item layout of the DFG modelling were conducted where HRR data were gatheredrnand compared with the DFG results. The DFG submodel predicted earlier ignition, and hence a morernrapid fire development, than actually occurred in the experimental simulations.