Computational models have been utilized to investigate the penetration capa
bility of sprinkler sprays directly above a fire source with respect to wat
er flow rate, spray drop size, and spray momentum. The spray models are gen
erated by assigning a representative drop size, mass flow rate, discharge s
peed, and discharge angle for each of 275 trajectories in such a way that t
hey produce computed results which match the measured water flux distributi
on and spray momentum in the absence of a fire. The spray/fire plume intera
ction models are created by combining the spray models using a Lagrangian p
article tracking scheme with free-burn fire plume models. Actual delivered
densities and penetration ratios are computed through the interaction simul
ations at six how rates, three fire sizes, and two ceiling heights. Drop si
zes and spray momentum at two flow rates are increased by 25 and 50% from t
he original values without changing the other spray characteristics in orde
r to investigate the effects of each parameter on penetration capability in
dependently. The study indicates that there is an optimal flow rate for a g
iven sprinkler that gives the highest penetration ratio within a practical
flow range. It is also shown that increasing drop size is a much more effec
tive way for obtaining a higher penetration ratio compared to increasing sp
ray momentum. (C) 1999 Elsevier Science Ltd. All rights reserved.