Experimental measurements of water mist systems and implications for modelling in CFD

Abstract: The use of water mist for fire extinguishment has increased rapidly in recent years. The main reason is the abandonment of halon-based extinguishing systems in favour of environmentally friendlier systems. Furthermore the use of water mist systems has spread from mainly marine applications to also include the protection of buildings. The main problem in this regard is to verify the effectiveness of the system. At present time this can only by done by full-scale tests. This is however expensive and in some cases also unrealistic and expensive when it comes to water mist systems for buildings. The aim of this thesis is to provide experimental data that can serve as a basis for simulations of the interaction of water mist and a fire and to demonstrate that CFD can predict the performance of a water mist system. The physics of water mist systems has been studied by theoretical considerations as well as experimental work. Measurements of droplet velocities, diameters and volumetric water distribution were carried out on the spray from a high-pressure system of 100 bars. Experiments have been conducted on a hollow cone nozzle without fire and with fire, as well as a full cone nozzle without fire. Relevant measurement results were obtained with Phase Doppler Anemometry and Particle Image velocimetry as well as Laser Tomography and High Speed camera. Suggestions were made for improvement of the water density apparatus. The measurements have been the basis for simulations of water mist with CFD. Initial simulations involving the complex zone around the nozzle resulted in droplets with radial velocities and insufficient transfer of momentum to the air. A new approach has been used for the simulations with the LES model in FDS 4.07. In this approach the simulations of the water mist spray is not done in the zone close to the nozzle. Instead the boundary conditions are set further downstream, based on the conducted measurements. This approach resulted in droplets and air moving downwards at relatively high velocities as expected. However, the momentum transfer is limited, and the simulations did not give sufficient mixing. Suggestion are made of how sufficient mixing can be obtained with the new approach, with regards to implementation of spray boundary conditions and treatment of the turbulence model interacting with the movement of the droplets.