Turbulent wall fires

Abstract

Turbulent natural convection fires at the base of a vertical wall were considered. The burning surface was simulated by 51–305 mm high wicks soaked with methanol, ethanol or 1-propanol. Measurements were made of burning rates in the pyrolysis zone; radiative and convective heat fluxes to the wall (and radiative heat flux to the ambiance) in the wall plume above the pyrolysis zone; and profiles of mean velocity, temperature and concentrations in all regions of the flow. The measurements are compared with solutions of the laminar boundary layer equations, solutions of the turbulent boundary layer equations employing a mixing length model, and integral models approximating the more complete theories. The theoreis provided a unified correlation of laminar and turbulent burning rate measurements from both the present and earlier studies. The correlation was relatively insensitive to the radiant heat flux from the flame to the burning surface, which varies in the range 0–86 percent of the total surface heat flux; this behavior is convenient, but has not been fully explained and deserves further attention. Laminar wall flames are 2–3 times longer than turbulent wall flames, when length is normalized by the length of the pyrolysis zone; theory provides good length predictions in both cases. The soot content of the flames was low for the present tests and radiation comprised only 10–20% of the heat flux to the surface; the flux of radiation to the ambiance was larger than the wall component. Predicted wall heat fluxes were within 20% of the measurements. Predictions in the noncombusting portion of the plume employed earlier results for weakly buoyant plumes using average physical properties at each wall position; profiles of mean velocity and temperature approximated local similarity, based on local plume energy flux and height along the wall; however, effects of combustion near the flame tip, variable property effects, and an additional Rayleigh number dependence were observed.