Fire growth prediction heat transfer

In the context of fire CFD, it is sometimes appropriate to distinguish between flame and fire spread simulations. Flame spread simulation usually means the ability to predict the fire growth starting from a small initial fire or ignition point, where all the three subprocesses are important but the second subprocess dominates the heat transfer. Fire spread, in turn, means the ignition of solid surfaces in the presence of a relatively large initial fire dominating the heat transfer by radiation. In practice,... 

FIRE SIMULATOR predicts the effects of fire growth in a 1-room, 2-vent compartment with sprinkler and detector. It predicts temperature and smoke properties (Oj/CO/COj concentrations and optical densities), heat transfer through room walls and ceilings, sprinkler/heat and smoke detector activation time, heating history of sprinkler/heat detector links, smoke detector response, sprinkler activation, ceiling jet temperature and velocity history (at specified radius from the flre i, sprinkler suppression rate of fire, time to flashover, post-flashover burning rates and duration, doors and windows which open and close, forced ventilation, post-flashover ventilation-limited combustion, lower flammability limit, smoke emissivity, and generation rates of CO/CO, pro iri i post-flashover. 

Convective heating in fire conditions is principally under natural convection conditions where for turbulent flow, a heat transfer coefficient of about 10 W/m2 K is typical. Therefore, under typical turbulent average flame temperatures of 800 °C, we expect convective heat fluxes of about 8 kW/m2. Consequently, under turbulent conditions, radiative heat transfer becomes more important to fire growth. This is one reason why fire growth is not easy to predict. 

Typically, a fire growth model is evaluated by comparing its calculations (predictions) of large-scale behavior to experimental HRR measurements, thermocouple temperatures, or pyrolysis front position. The overall predictive capabilities of fire growth models depend on the pyrolysis model, treatment of gas-phase fluid mechanics, turbulence, combustion chemistry, and convective/radiative heat transfer. Unless simulations are truly blind, some model calibration (adjusting various input parameters to improve agreement between model calculations and experimental data) is usually inherent in published results, so model calculations may not truly be predictions. 

---