The “cabin” concept has been used widely for fire safety in big public transport terminals not fully protected by sprinkler nor smoke exhaust system in the Far East for the past 20 years. There, only the areas with a higher fire risk, such as retail shops, are protected. The big terminals are used to be crowded, and so there are concerns on whether the cabin design can provide adequate fire protection in such big halls, and whether the design will give problems in firefighting. Consequently, additional fire safety provisions such as long-throw sprinkler had been installed in areas with higher fire risk. Smoke exhaust system have to be added to the large hall. In the new edition of fire safety provisions required for railway systems, the open type of cabin is specified. No in-depth study supported by systematic experiments on how the fire can be limited to 2 MW reported in the literature. On the other hand, preliminary experimental studies on cabin with sprinkler and smoke exhaust carried out indicated that there are many concerns.In this paper, five tests on cabin fire carried out in a big burning hall at the State Key Laboratory of Fire Science, University of Science and Technology of China (USTC) will be taken out to demonstrate the problems associated with such open type cabins. Problems in big halls without fully covered by smoke exhaust system had been demonstrated clearly in an airport terminal fire in 1998. It took a long time for firefighters to locate the fire source in that small fire at the airport terminal. The fire safety must be upgraded by appropriate hardware engineering systems and software safety management to reduce the chance of having big fire, and facilitate firefighting.

Experiments are carried out in a simplified drop system, which can provide a 2.3s 10−2g reduced gravity period, to investigate the transition of solid material combustion behavior from normal- to reduced-gravity. A CCD camera records the diffusion flames produced by a solid fuel, Methenamine (C6H12N4), which easily sublimes and decomposes to flammable CH4 and H2. The temperature distribution within the flame is analyzed by a monochromatic line of sight CCD measurement calibrated by a thermocouple measurement. It is revealed that during the transition period from normal- to reduced-gravity, the change of the flame shape, from tall and thin (pear like) to spherical, is faster than that of the luminance which represents the radiation intensity, indicating a relatively shorter adjustment time scale of dynamical gas flow buoyancy than that of the heat transfer when subjected to such a sudden change of gravity condition. It is also found that owing to the change of the flow affecting the oxygen and heat supply, the luminance of the flame as well as the burning rate, which was deduced based on projected area of the solid fuel according to d2-law, first decreases followed by an increment near the end of the drop where the gravity level slightly increases. The flame temperature gradually decreases, as the flame suddenly goes into reduced-gravity. Interpretation based on analysis of heat transfer balance and pyrolysis rate as well as the experimental results both indicate that the transitional effect is less pronounced for a smaller fuel particle than for a larger one, where buoyancy is stronger under normal-gravity whereas the flame has a larger standoff distance with increasing importance of the radiation losses from the fuel particle surface under reduced-gravity. The present work is relevant in assessing fire hazards of solid material during space travel as gravity levels change.

Carbon monoxide (CO) stratification and its relationship with thermal stratification are quite important for dealing with issues related to building fire safety. This experimental study compares the CO stratification and thermal stratification in channel fires. The results show that the relationship between CO stratification and thermal stratification depends on heat loss intensity from smoke flow to walls. In the conditions with considerable amount of heat loss, the vertical gradients of CO volume concentration are smaller than those of temperature rise beneath the ceiling. However, in the conditions with negligible heat loss, the vertical profile of dimensionless CO volume concentration becomes similar with dimensionless temperature rise. The longitudinal ventilation, as a prevailing smoke control or ventilation method, has a strong effect on the relationship between CO stratification and thermal stratification. A larger longitudinal air flow velocity leads to smaller heat loss intensity from smoke flow to walls and thus a higher similarity between CO stratification and thermal stratification.

The present paper is engaged in an experimental study over a rise-time of smoke layer interface and deduced the relation between dimensionless rise-time of smoke layer interface and dimensionless height of the shaft. Based on the theoretical results, the risetime of smoke layer interface is measured by using small scale shaft which are located at the PolyU/USTC Atrium. Our study shows that the rise-time of the smoke layer interface be inversely proportional to the 1/3 power of heat release rate of the fire source, and proportional to the 1.05 and the 1.50 power of the height in open and closed shafts, respectively. Furthermore, it is also proved that the smoke layer interface movement in open shaft be faster than that in a free space and closed shaft due to the stack effect. Under the same fire source and shaft dimensions, the velocity of smoke spreading in closed shaft is the least one.

Front velocity of transient ceiling jet is a major concern for building fire safety. This work concentrates on the front velocity of transient ceiling jet in corridor fires. Experiments with both steady fire sources and a developing fire source are performed to evaluate the applicability of existing theoretical correlations for a relatively long corridor. To account for the conditions with significant heat loss from smoke flow, some modifications for these correlations are proposed by this work. Results show that the classical Hinkley’s correlation without consideration of the heat loss can significantly over-predict the front velocity of transient ceiling jet at the remote locations with relatively low temperatures (e.g., less than 80 °C). Jones’ model, Benjamin’s model and CFAST’s model also over-predict the front velocity at these low temperatures, but with much smaller deviations in relation to Hinkley’s correlation. The modified Hinkley’s correlation, with consideration of the heat loss from smoke flow, produces more acceptable predictions for both steady and growing fires.

In a channel fire, the transports of the two primary hazards, heat and CO species, are dominated by different mechanisms. Experiments were conducted in a horizontal channel with dimensions of 66.0 m long x 1.5 m wide x 1.3 m high to investigate the spatial distributions of CO volume concentration and those of temperature rise. The effects of longitudinal ventilation were also considered. Both in the longitudinal direction and in the vertical direction, profiles of CO volume concentration differ with those of temperature rise. In the downstream direction, the temperature rise of smoke flow decays rapidly with increase in the distance from the fire origin, while CO volume concentration remains constant in the longitudinal direction. A larger longitudinal ventilation velocity leads to a slower decay in temperature rise along the longitudinal direction. However, longitudinal ventilation has a small influence on the longitudinal profiles of normalized CO volume concentration. As compared with temperature rise, the CO volume concentration decays more slowly with the decrease in height under the conditions without longitudinal ventilation. However, a relatively large longitudinal ventilation velocity leads to a high similarity between the vertical profile of CO volume concentration and that of temperature rise. (C) 2010 Elsevier Ltd. All rights reserved.

The temperatures of buoyant fire plumes were measured vertically and compared with the Zukoski model, the Heskestad model, and the McCaffrey model. Predictions by the Heskestad and McCaffrey models were closer to the measured value with average deviation of 18.4% and 18.6%, respectively, while those by the Zukoski model were much lower than the measured value with average deviation of 36.7%. The power law decay index of z was found to increase slightly with fire size, rather than being constant in the models. The coefficient for the variation of ΔT with Q2/3 z−5/3 it was found to be 20.9 experimentally, near to that of the Heskestad and McCaffrey models.

To assess the impact of smoke on the ceiling in subway stations, the maximum smoke temperature under the ceiling was studied theoretically and experimentally with two sets of small-scale experiments conducted. The results show that the maximum smoke temperature under the ceiling complies with the Alpert equation in which fire keeps distant from the walls in subway stations whereas fire adjacent to the end wall leads to the maximum smoke temperature under the ceiling decaying exponentially against the increased distance between the fire and the wall. In addition to the Alpert equation, a correlation determining the maximum smoke temperature is developed by taking the end wall effect into account. Consequently, a simplified calculation method involving the Alpert equation and the correlation is established. The method is applicable to practical fire engineering designs for subway stations. (C) 2011 Elsevier Ltd. All rights reserved.

The performance of mixture fraction models FDS4 and FDS5 is investigated under different global equivalence ratios (GER). Predictions of heat release rate (HRR), upper-layer temperature, and CO yield are compared with measurements considering their sensitivities to the lower limit of fuel, mixing time scale, and turbulence model constants. When using FDS4, the inclusion of an extinction model can result in significant variations in both total and volumetric HRR prediction. When using FDS5, the mixing model constant has significant effects on volumetric HRR prediction. At low GER (< 0.23). the prediction of upper-layer temperature shows dependency on both the lower fuel limit and the mixing model constant, but the predicted temperature is always lower than measured temperature, with deviations in excess of 30%. At higher GER (0.53 < GER < 0.81), the upper-layer temperature prediction shows significant dependency on the mixing model constant but can be over-predicted, with deviations up to 24%. The variations of CO yield prediction with lower fuel limit or with the mixing model constant show an opposite trend to that of upper-layer temperature. Furthermore, the prediction of CO yield shows a much greater dependency on the Smagorinsky constant and on the turbulent Schmidt number than do those of HRR and upper-layer temperature. (C) 2010 Elsevier Ltd. All rights reserved.

Experiments were carried out in a reduced-scale horizontal channel to investigate the fire-induced buoyant flow stratification behavior, with the effect of the velocity shear between the hot buoyant flow and the cool air flow considered. This shear intensity was controlled and varied by changing the exhaust rate at the ceiling with one of the end of the channel opened. The flow pattern was visualized by the aid of a laser sheet. The horizontal traveling velocity, vertical temperature profile and stratification interface height of the buoyant flow were measured. The stratification pattern was found to fall into three regimes. Buoyancy force and inertia force, as the two factors that dominate the buoyant flow stratification, were correlated through the Froude number and the Richardson number. At Region I (Ri > 0.9 or Fr < 1.2), the buoyant flow stratification was stable, where a distinct interface existed between the upper smoke layer and the lower air layer. At Region II (0.3 < Ri <0.9 or 1.2 < Fr < 2.4), the buoyant flow stratification was stable but with interfacial instability. At Range III (Ri < 0.3 or Fr > 2.4), the buoyant flow stratification becomes unstable, with a strong mixing between the buoyant flow and the air flow and then a thickened smoke layer. (C) 2010 Elsevier Ltd. All rights reserved.

Longitudinal decay profiles of CO concentration and smoke temperature in a tunnel fire smoke flow are theoretical analyzed and compared, with their difference investigated, under different longitudinal ventilation velocities. Experimental data on longitudinal CO distribution achieved from a set of full scale road tunnel fire tests are presented to compare with the theoretical equation. CFD simulations are also carried out by Fire Dynamics Simulator (FDS). It is found that the longitudinal profile of CO concentration along the tunnel yields a function of Cx/C0=1/(1+bx), and its difference with that of the smoke temperature increases along the tunnel by a function of Cx/C0-ΔTx/ΔT0≈λ(1-e-Kx). The smoke temperature decays much faster than the CO concentration along the tunnel. Their longitudinal profile difference decreases as the longitudinal ventilation velocity increases, and increases along with the distance away from the fire asymptotically to a quasi-steady value. The value of b decreases as the longitudinal ventilation velocity increases, which indicates that the CO concentration decays relatively slower along the tunnel under a higher longitudinal ventilation velocity. And its value is shown to be less affected by the longitudinal ventilation velocity for a relative larger fire. The increase in the longitudinal ventilation velocity leads to the enhancement of the air mass entrainment, thus results in the decrease of the longitudinal decay profile difference between the CO concentration and the smoke temperature. The value of λ is found to decrease with the increase of the longitudinal ventilation velocity, following a reciprocal function of λ∼1/(ϕ+αu). Its value at zero longitudinal ventilation velocity is higher for a larger fire, but decreases faster with the increase of the longitudinal ventilation velocity than a smaller fire. The full scale experimental data and the CFD simulation results both agree well with the theoretical analysis and equations.

The adverse effects of a sprinkler on smoke exhaust systems used to be debated in designing fire safety provisions. However, very few experimental studies on the interaction of a sprinkler with the smoke layer have been reported; therefore, this study investigates and reports on the interaction of a smoke layer with a sprinkler water spray by a specially designed rig with 25 tests. Smoke spreading from a fire chamber was collected in an adjacent hall. The stability of the smoke layer was then studied by the discharged water spray. The ratio of spray drag to smoke buoyancy is proposed to determine smoke stability.

A mathematical model has been developed to investigate the effect of a sprinkler spray on adjacent horizontal smoke venting and in particular the water droplet drag component. The pressure difference across a roof vent and the volumetric flow of smoke vented are determined by considering the interaction between the drag force of the sprinkler spray and the buoyancy of the smoke layer in the spray region. Smoke venting may become progressively more inefficient as the sprinkler operating pressure increases due to the cooling and drag effect of the sprinkler spray. Full scale experiments were carried out to validate the model. Results show that the mathematical model can predict the observed trend of a decrease in vented volumetric flow with an increase in sprinkler operating pressure, which eventually leads to ineffective smoke venting. Experiments with different smoke venting areas show that vent area has little influence on smoke flow once sprinkler pressure causes a loss in smoke flow efficiency or vent function.

Square pool fires with length of 5, 7.5, 10, 15, 20, 25 and 30cm and rectangular pool fires with dimensions of 10cm×20cm and 10cm×40cm were burned in a wind tunnel, under a longitudinal air flow ranged from 0 to 3m/s with incremental change of about 0.5m/s. Methanol and gasoline were burned and compared, with results indicated that their burning rates showed different response to the longitudinal air flow. With the increase of the longitudinal air flow speed, the burning rates of methanol pool fires, except the 5cm square one, first decreased and then increased, but those of the 5cm methanol square one and the gasoline pool fires increased monotonously. The burning rate of smaller square pool fires increased more significantly than that of the larger ones, as well as the enlargement of their flame attachment length along the ground. The burning rate of a rectangular pool fire with longer rim parallel to the longitudinal flow increased faster, but the flame attachment length seemed to increase more gradually, with the increase of the longitudinal air flow speed than that perpendicular to.

Experiments were conducted to investigate the smoke movement in a full-scale six-storey stairwell induced by a fire in an adjacent compartment. The smoke flow pattern in the stairwell was observed with vertical temperature and velocity field distribution measured. Corresponding Computational Fluid Dynamics (CFD) simulations were carried out by Fire Dynamics Simulator (FDS) and compared with the measured data. It was observed that the smoke flow moved upward in circular patterns and formed vortices under the stairwell tread of each storey. The smoke temperature was found to generally decrease exponentially with height in the stairwell, except the top and the bottom levels, where the temperature is relatively higher. The upward smoke flow velocity first increased and then decreased with height. The predicted smoke flow temperature and velocity by FDS were shown to agree well with the measured values. Temperature distribution is identified to be a key factor affecting the smoke movement in the stairwell.

In fire safety engineering, rapid fire detection and suppression in large spaces are very important issue. Development of new techniques is always based on the good understanding of the basic principle of smoke movement, especially the temperature of spill plume. Aiming to this objective, this paper specifies the multi-sectional characteristics on spill plume free developing out of a wide-door cabin in a large space atrium by means of full-scale experiment, theoretical analysis and CFD modeling. First the physical development process of the spill plume is analyzed from the horizontal curved regime at the door, the near-field two-dimensional linear plume regime, through to the far-field axisymmetric plume regime. It demonstrates the coupling correlation existing between the linear virtual origin, critical transition height from linear to axisymmetric regime, the axisymmetric virtual origin and other parameters, such as the spill plume depth at the door, the mass and heat flow rate outside the cabin door, the cabin door width and height etc. The basic mathematic equations regarding each regime's centerline temperature have been obtained. The virtual origin position of linear plume regime and axisymmetric plume regime are given, the critical transition height of spill plume from the two-dimensional linear plume to axisymmetric plume regime is also proposed, and finally the double-part prediction model (MEDP) for the spill plume centerline temperature is put forward. By validation with the full-scale experiment and the large eddy CFD modeling, it shows that the proposed spill plume model is capable to well describe the centerline temperature varying characteristics of the spill plume free developing out of the cabin.

Discharge rate of a horizontal adjacent smoke vent under sprinkler spray is experimentally investigated. Temperature of smoke layer and velocity of smoke venting were measured, under different sprinkler operating pressures and smoke venting areas. CO concentration at the smoke vent center and velocity of vent flow with fresh air outside were recorded in tests under different smoke venting conditions. Experimental results have shown that efficiency of smoke venting is controlled by a combination of smoke buoyancy and drag force of sprinkler spray. Only when buoyancy is greater than drag force the smoke could be extracted by venting. Velocity of smoke venting has shown to decrease as the operating pressure increases. Smoke venting logging, which represents the failure of smoke venting, was experimentally found from certain operating pressure called initial logging pressure. The CO concentration was found to increase after sprinkler was operated as the smoke is constrained in the spray region with horizontal momentum decreased. Negative pressure difference is caused at the vent when there is smoke venting logging, which might practically bring the exterior fresh air into the fire building. Additionally, experiments results have shown that the venting area has little influence on smoke flow under smoke venting logging.

A mathematical model was developed for predicting the downward descending behavior of the buoyant smoke layer under sprinkler spray. The behavior of the smoke layer was determined by considering the interaction between the drag force of the sprinkler spray and the buoyancy force of the hot smoke layer itself in the spray region. The smoke layer may be pulled down with its thickness increased at the center of the spray region due to the cooling and drag effects of the sprinkler spray, thus to form a downward "smoke logging" plume. In the mathematical model developed in this paper, the critical condition under which the smoke layer lost its stability, as a serious concern, was predicted. Additionally, the length of the downward plume, which was rarely investigated before, was also further calculated. Full-scale experiments were carried out to validate the model. Results showed that the predictions, including the critical condition and the length of the plume, by the mathematical model agreed well with that observed and measured in the experiments. The length of the downward plume was shown to increase with the sprinkler operating pressure by an approximately linear correlation.