We consider in this study the development of a prototype inverse fire model (IFM) aimed at predicting the size of a building fire using smoke layer information gained from a temperature sensor. The proposed methodology consists in: performing a search for the unknown heat release rate (HRR) by performing tens or hundreds of different zone model simulations; comparing model predictions to observation data and thereby formulating an error function; using an optimization technique to minimize the error that gives in turn the most probable variations of HRR. The prototype IFM algorithm uses: MATLAB as the programming language; BRI2002 (developed by the Building Research Institute in Tsukuba, Japan) as the zone model; and a genetic algorithm for optimization. The IFM algorithm is applied to a series of test configurations corresponding to: steady or unsteady fire conditions; single- or multi-compartments; single or multiple vents; different ventilation capacities ranging from well-ventilated to under-ventilated fire conditions; and up to 19 unknown parameters. In all tests, the IFM algorithm is found to be remarkably robust and to successfully minimize the error function.
展开▼
