International Journal on Engineering Performance-Based Fire Codes, Volume 4, Number 3, p.95-103, 2002 95 A SIMPLE ENGINEERING MODEL FOR SPRINKLER SPRAY INTERACTION WITH FIRE PRODUCTS V. Novozhilov School of Mechanical and Production Engineering Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798 (Received 21 February 2002; Accepted 2 September 2002) ABSTRACT Analytical model is developed to predict smoke layer temperatures after activation of sprinkler or water mist system. The governing equations are solved to determine water droplet trajectory, temperature history and evaporation rate. As a result, a rate of heat absorption by water spray and temperature history of the hot smoke layer can be predicted. A good agreement is demonstrated between the analytical model and the results of full CFD simulations. 1. INTRODUCTION The problem of predicting smoke layer temperature is typically arising in many fire safety applications. A part of this general problem is a prediction of smoke layer behaviour upon activation of fire suppression systems. Typical examples of such systems are water sprinklers and water mist nozzles. Heat absorption by water spray generally leads to the drop of the overall temperature inside the compartment. Quantitative prediction of this effect is quite important for estimation of attainable conditions inside buildings during fire events. Primary differences between water sprinkler and water mist systems are in water discharge rates and droplet size distribution. Water spray for conventional sprinkler is relatively “coarse”, with significant amount of droplets larger than 1 mm. In contrast, mist spray is defined by NFPA as the spray containing 99% of its volume in the droplets smaller than 1 mm. Recent interest in water mist application is mostly driven by phasing out of halons, and the need to find a replacement which can act as volumetric suppression agent. Due to very fine dispersion, mist is capable of flooding the room, although it does not have as good flooding properties as gaseous agents. Due to high surface area of the droplet phase, water mist evaporates very effectively, and therefore water discharge quantities may be kept much smaller, compared to conventional sprinklers. Water mist spray does not generally penetrate to the surface of burning material (due to low momentum of individual droplets), and suppress flame in gaseous phase. In contrast, sprinkler sprays are most effective for unshielded fuel surfaces, as droplets easily penetrate to burning material, cool the surface, and suppress pyrolysis process. Computational Fluid Dynamics (CFD) modelling of water sprinkler sprays is well-established [1-3]. However, this technique is relatively expensive. Typically, tens of thousands of trajectories are tracked to represent water spray statistically. For real fire engineering applications, this may not be necessary in all situations. Basic understanding of water spray behaviour can be achieved using simple estimations of droplet dynamics, similar to considerations applied in the pioneering work of Chow and Tang [4]. The model described in the present paper provides a more accurate estimation of droplet interaction with the hot layer of combustion products. In contrast to the paper [4], heat transfer problem for the moving droplet is solved in addition to the dynamic problem. The major advancement of the model is the analytical solution for the droplet motion, obtained for the case of real drag coefficient correlations, applicable for spherical particles moving in a turbulent gas stream. The equations and the method to obtain analytical solutions are discussed first. From the solutions, the droplet trajectories and heat absorption rates for individual droplets can be predicted. Such predictions are illustrated for various droplet sizes and different temperatures of the hot layer. Finally, the model is utilized to make predictions of temperature histories in compartments after activation of water sprinklers. The results from the present analytical model are compared to the CFD simulations of the similar problem.