Investigation Of Downward Displacement Of Fire Smoke Under The Effect Of A Spray Of Water Droplets Discharged From A Nozzle

Fire smoke, causing most of the fire victims, is the most serious threat for people evacuating during fire disasters. Although fixed water spray/sprinkler systems are being widely used for the fire protection of buildings, the water droplets discharged from the nozzles of these systems are able to disturb a stratification of fire smoke and cause a serious downward smoke displacement (’smoke logging’) to threaten the people evacuating in fire environments. Due to lack of sufficient knowledge to understand this phenomenon, it is a potential risk for fire safety, so that installing the fixed water spray/sprinkler system in tunnels to improve the tunnel fire safety is still a controversial topic in the world. So far, hardly any European countries take this measure as a regular safety facility in tunnels.The precondition for good prediction of the downward smoke displacement induced by water droplets is essentially to understand the interaction between water droplets and hot smoke, including mass and heat transfer between each other. This is a very complex topic related to the two-phase flow. Normally, the cooling effect and downward drag force from water droplets are recognised as two mechanisms for the smoke logging. Based on some simplifications, even ignoring the heat transfer between water droplets and hot smoke, the classical theory determines the smoke logging from the balance between the drag force exerted on the smoke by the droplets and the buoyancy force on the smoke. This force balance is revisited in the present work, including the net downward buoyancy in the cooled region in the smoke layer, which has been ignored in the existing models. It is illustrated that the classical theory is not valid in general. An improved model is developed. It can be incorporated into existing fire zone models to quantify the downward smoke layer displacement, the main goal of this thesis.This research topic is also investigated by means of experiments and CFD (computational fluid dynamics) simulations.Thus, this thesis consists of four parts. The first part concerns the experimental study of the downward displacement of fire-induced smoke by water droplets. The downward smoke layer displacement as a result of the interaction with the water droplets is quantified for a wide range of settings, including variable fire conditions, smoke layer temperature, smoke layer thickness and water spray operation pressure. Useful conclusions are drawn from the phenomenological observations. It is illustrated that smoke with low temperature under high spray operation pressure is likely to have a large downward smoke displacement. The experiments also include temperature measurements, allowing highlighting the importance of entrainment processes in the downward smoke displacement phenomenon. The experimental outcomes have been used as the basis for the development of the model in the second part of this thesis. At the same time, they serve as experimental data set for the CFI) simulations (third part of the thesis).In the second part of this thesis, a stand-alone analytical model is described to quantify the downward smoke displacement caused by a water spray from a sprinkler/spray nozzle, for use in two-zone model simulations. The underlying assumptions are identified and the global balance is described between downward drag force, potentially downward buoyancy due to a cooling effect within the water spray envelope in the smoke layer, and the upward buoyant force in the ambient air below the smoke layer. From this balance, the downward smoke displacement is quantified. It is explained that the classical Bullen criterion for smoke layer stability is in general not valid. There is always downward smoke displacement, although potentially small, depending on the circumstances. The tracking of individual water droplets leads to the evolution of the spray envelope radius and provides the total downward drag force on the smoke. Smoke flow (’entrainment’) into and out of the water spray envelope is incorporated, as well as a model for the heat exchange between the smoke and the water droplets. The model input quantities are:water flow rate, orifice diameter, droplet diameter, spray angle, initial smoke layer thickness and temperature, and ambient air temperature. The output quantities are:droplet velocity, droplet trajectory, droplet diameter, drag force, smoke buoyancy, smoke velocity and smoke temperature inside spray region, entrainment mass flow rate, heat exchange rate between droplets and smoke. The downward smoke displacement could be determined by the position where smoke velocity equals to zero. The quality of the model predictions for downward smoke displacement is illustrated for a range of experimental conditions. Results are shown to be sensitive to the mean droplet diameter and spray angle, though, and combinations of values of these parameters are suggested for the sprinkler considered. In addition, the quality of the model predictions of heat transfer rate between droplets and hot smoke is also illustrated by good agreement with another set of experiments. Additionally, an extensive model test study is presented, varying the water spray angle at the nozzle, the water droplet diameter, the droplet initial velocity, the water flow rate of the nozzle, the smoke layer temperature and the smoke layer thickness. Based on these testing results, the motions of a single water droplet in different conditions, the interaction between droplets and ambient gas phase, the smoke downward displacement varying with different conditions are all investigated very carefully. It is highlighted that the drag force and the shape of the spray region are in close relation to the droplet diameter. Some conclusions drawn in the experimental part, e.g. smoke with low temperature under high spray operation pressure is likely to have a large downward smoke displacement, are also confirmed by the test results. It is also highlighted that the shift in downward smoke displacement due to the differences in initial smoke layer temperature is relatively independent of smoke thickness, except for very thin or low temperature smoke layer. Furthermore, a practically important phenomenon is identified, namely an abrupt strong downward smoke layer displacement when the initial smoke layer thickness is below to a certain value. The explanation is that, due to entrainment of cool air into the water spray envelope, the upward buoyancy flow diminishes. This phenomenon is more serious under the high water operation pressure (large water flow rate and small mean diameter of the droplets), small spray angle and low smoke temperature. Therefore, this is particularly dangerous during the smoke layer build-up phase and it might be advantageous to opt for sprinklers with moderate water flow rates and late response in this respect.The third part of this thesis focuses on CFD simulations of downward smoke layer displacement under different spray conditions. In order to verify whether the downward smoke displacement under spray conditions can be predicted accurately by means of CFD simulation, it is important to make the comparison of simulation results to the experimental observations as described in the second part. The simulations are carried out by FDS (Fire Dynamics Simulator) which is a CFD model for fire simulation using the LES (large eddy simulation) approach for turbulence. The results, focusing on the temperature distribution and the downward smoke displacement, reveal that FDS can reconstruct the experimental observations successfully under the typical experimental conditions. Based on the success of the simulations, the entrainment process highlighted in the analysis of the experiments is confirmed, as seen in detail from the results of the simulations. In the final part of this thesis, the smoke downward displacement induced by droplets in a tunnel fire is briefly investigated with FDS simulations. Two types of sprinklers, as well as two kinds of tunnel ventilation conditions, natural ventilation and longitudinal ventilation with2m/s, are considered in the simulations. Ihe fire is assumed to be a t2growing fire. The simulation results indicate that the severity of smoke downward displacement depends on the fire scale, sprinkler and the ventilation conditions. In the tunnel with the natural ventilation condition, single sprinkler can induce the fire smoke to the floor directly at the initial stage of the fire, because the smoke layer is thin and its temperature is relative low. As it is illustrated in the model tests that a large amount of cool air entrained into the downward smoke, the visibility of the space occupied by the downward smoke decreases not very seriously at this stage. After that, the visibility would go worse when more smoke is deduced down continually. However, this trend could be stopped and turn to a better visibility, as the increase of the smoke temperature due to the increase of HRR with time. In the tunnel with longitudinal ventilation condition, the longitudinal ventilation could worsen the stratification of the smoke layer in the downstream of the fire, but could keep the back-layering in upstream of the fire more stable. It is highlighted that in both ventilation conditions, the serious smoke logging could happen when increasing the number and the mass flow rate of the sprinklers. Additionally, the entrainment effect of the fire would accelerate the downward smoke spreading in the bottom space of the tunnel, and make the visibility worse. All these findings can be explained well by the theory as developed in this thesis. The model developed in the second part even has good predictions to reflect the smoke logging in the condition of a single sprinkler interacting with the tunnel fire smoke under the natural ventilation. The description in the model that the temperature inside the spray region is relatively lower than the surrounding hot smoke is also confirmed by the simulation results. Based on the above findings, some conclusions for the fixed water spray/sprinkler systems in practical tunnels to avoid serious smoke logging are drawn:the severity of the smoke logging is related closely to spray/sprinkler condition, so that the sprinkler type and water operation pressure should be considered carefully in the stage of design, which could be assisted by the model in this thesis and the FDS model; It is better only to actuate the sprinkler/spray nozzle near the fire; In the longitudinal ventilation tunnel, the spray/sprinkler should be controlled very carefully if there are still evacuators under the sprinklers in the downstream of the fire; Otherwise, it is better to provide the largest longitudinal ventilation to avoid back-layering in upstream, or keep the back-layering more stable under the spray/sprinkler conditions.To summarise, this thesis helps to better understand the interaction between water droplets and fire smoke, and reveals how the downward smoke displacement is affected by the water droplets discharged from a sprinkler or spray nozzle. The developed model could be a useful tool for practice to quantify the downward smoke layer displacement. Furthermore, some research results could be referred to as guideline for designing a fixed water fixed water spray/sprinkler system in tunnels to avoid or reduce the downward smoke displacement.