Fire modeling — Numerical modeling tools for fire development

Although fire has been mastered by man for almost 790,000 years, our understanding of the corresponding phenomena developed mainly in the second half of the 20th century. The equation of combustion and transport phenomena is therefore very recent.

The first analytical models, developed by pioneers such as Howard Emmons, Philip Thomas and Kunio Kawagoe between the 1950s and 1970s, led to the establishment of today's numerical models, in parallel with the development of computer science.

The purpose of these models is to represent a fire situation, to enable predictive calculations of the course of disasters. They are therefore the engineer's main tool for quantifying the effects of fire in a structure, for a given fire scenario. This tool is designed to provide information on the evolution of a structure's environment as a function of the scenario, in particular variations in temperature (local or global) and fire power (heat output). The aim is to ensure the safety of people in the vicinity of smoke displacement, and to design the structural resistance of structures in the event of fire.

Today, with the development of fire safety engineering in transport and building regulations as a means of proving the level of fire safety performance, engineers have a range of modeling tools at their disposal to quantify the development and consequences of fires.

In a fire safety engineering study, these models are used after a risk analysis stage, followed by the selection of design scenarios. The models are then used as quantification tools to certify a structure's level of fire safety performance, either by comparison with predefined thresholds (absolute approach) or by comparison with designs considered acceptable by default (relative approach). In such applications, the digital tool may overestimate certain parameters, as it serves as a sizing tool. Care must be taken to ensure that it never underestimates the calculated safety level.

Other uses for these models are possible, such as the reconstitution of real claims, for legal purposes or to draw lessons from them and advance regulations. In such applications, a representation as close as possible to reality is essential.