Citation

Tsai KC, Drysdale D. Fire Mater. 2002; 26(6): 279-287.

Copyright

(Copyright © 2002, John Wiley and Sons)

DOI

unavailable

PMID

unavailable

Abstract

Previous work has demonstrated that flame height is one of the two most important parameters determining the rate of vertical flame spread on a wall. Flame spread models rely on empirical flame height correlations of the form X-f = K(Q) over dot(/n), but there have been no carefully controlled experiments designed to establish the validity of such correlations and there are no data for values of (Q) over dot(/) less than c.25 kW/m. Two new sets of data are presented here, one based on experiments with 6 mm thick PMMA slabs (heights 25-250 mm, width 50-150 mm) as the fire source, measuring (Q) over dot (/) and X-f simultaneously. This set relates to the early stages of a wall fire when (Q) over dot(/) less than 30 kW/m. The other set of data was obtained with a vertical gas-fired panel which consisted of an array of 14 independent burners arranged to allow the aspect ratio of the burning surface to be varied. The data confirm that the flame height correlates with (Q) over dot(/), but reveal (inter alia) that there are two regions, for values of (Q) over dot(/) greater than and less than c. 20 kW/m. An existing upward flame spread model was modified to allow these two correlations to be incorporated to predict the development of a small fire on a vertical surface for these two regions, which would provide more realistic modelling information.