Elsevier

Fire Safety Journal

Volume 40, Issue 3, April 2005, Pages 213-244
Fire Safety Journal

The critical ventilation velocity in tunnel fires—a computer simulation

https://doi.org/10.1016/j.firesaf.2004.11.001Get rights and content

Abstract

In ventilated tunnel fires, smoke and hot combustion products may form a layer near the ceiling and flow in the direction opposite to the ventilation stream. The existence of this reverse stratified flow has an important bearing on fire fighting and evacuation of underground mine roadways, tunnels and building corridors. In the present study, conducted by the National Institute for Occupational Safety and Health, a CFD program (fire dynamics simulator) based on large eddy simulations (LES) is used to model floor-level fires in a ventilated tunnel. Specifically, the critical ventilation velocity that is just sufficient to prevent the formation of a reverse stratified layer is simulated for two tunnels of different size. The computer code is verified by checking the computed velocity profile against experimental measurements. The CFD results show the leveling-off of the critical ventilation velocity as the heat release rate surpasses a certain value. At this critical ventilation, the ceiling temperature above the fire reaches a maximum for both tunnels. The velocity leveling-off can be explained from this observation. An extended correlation of Newman (Combust. Flame 57 (1984) 33) is applied to the temperature profiles obtained by CFD. At the critical ventilation, temperature stratification exists downstream from the fire. The computed critical ventilation velocity shows fair agreement with available experimental data taken from both horizontal and inclined fire tunnels. The CFD simulations indicate that the Froude modeling is an approximation for tunnel fires. The Froude-scaling law does not apply to two geometrically similar fire tunnels. The CFD results are compared with two simple theories of critical ventilation by Kennedy et al. (ASHRAE Trans. Res. 102(2) (1996) 40) and Kunsch (Fire safety J. 37 (2002) 67).

Introduction

When a fire is started on the floor of a straight tunnel without a ventilation (cross) flow, a hot plume rises above the fire and entrains the surrounding cold air into the plume. The plume, upon reaching the ceiling, forms two gas streams flowing in opposite directions along the ceiling. When a cross ventilation current exists, the symmetry in the rising plume and in the ceiling gas streams is broken. The ventilation current bends the plume and the length of the ceiling layer flowing against the ventilation current is reduced. This situation is depicted in Fig. 1, where the ceiling layer is represented by the gas temperature contours. In Fig. 1, the velocity vectors are also shown to indicate the general flow pattern in the tunnel.

In the event of a tunnel fire or smoke emergency, a main concern is to maintain an evacuation path that is free of smoke and hot gases. The existence of reverse stratified layer (also called back-layer) of hot combustion products has an important bearing on fire fighting and evacuation of underground mine roadways, tunnels and building corridors. Consider a scenario involving a stopped vehicle on fire in a tunnel, disrupting traffic and requiring worker and passenger evacuation. Another scenario involves a conveyer-belt fire in an underground mine entry, producing smoke and toxic combustion products. The consideration for emergency planning must focus on determining the ventilation required to maintain a single evacuation path from the fire source clear of smoke and hot gases.

Experimental data show that the length of a back-layer upstream of the fire source is a function of uin for fixed heat release rate from the fire [1], [2], [3]. As the value of uin increases, the length of the back-layer decreases. The “critical ventilation velocity” ucr is defined as the value of the ventilation velocity uin that is just able to prevent the formation of a back-layer. In the present CFD study, the ventilation velocity for a given Q is ucr when the front end of the back-layer is at x=0 (front end of the fire source).

The risk from accidental fires as well as the subsequent smoke movement depends largely on this applied ventilation current uin. It is of practical importance to understand the physical parameters and flow conditions under which the reverse stratified flow can be prevented.

Previous authors employed simple empirical models to determine the critical ventilation velocity needed to prevent upstream movement of smoke from a fire in a tunnel [1], [4], [5], [6], [7]. In general, these models considered the buoyancy head and the dynamic head in the system, and deduced appropriate quantities for correlation. Hwang et al. [8] and Charters et al. [9] employed phenomenological models that provided more detail than the simple empirical approaches. A recent review of tunnel fires by Grant et al. [10] pointed out that existing experimental data still show an inadequate fundamental understanding of the interaction between buoyancy-driven combustion products and forced ventilation, the validity of extrapolation of small-scale results to large scales, the influence of slope on smoke movement, and the effect of tunnel geometry. Wu and Bakar [11] carried out experimental tests of tunnel fires using tunnels of different cross sections. They used the hydraulic diameter as the characteristic length in the dimensionless groups for correlation. The correlation was able to encompass their own data and large tunnel data of other workers. The correlation shows that at large values of Q, ucr levels off. Kunsch [12] derived an expression that shows the decreasing effect of Q on uin as Q increases. His equation shows that the aspect ratio of the tunnel cross-section is also a parameter.

The present study addresses the problem of critical ventilation velocity in longitudinally ventilated tunnels. Specifically, the leveling-off of ucr for large values of Q is analyzed. Computer simulations are made to see whether the leveling-off of ucr can be observed. If leveling-off of ucr can be simulated, a possible cause is searched and studied. The simulation results are compared with available experimental data and simple theories.

Section snippets

Experimental data

Fig. 2 shows a plot of existing experimental data on the critical ventilation velocity as a function of the heat release rate in fire tunnels. The tunnel size is expressed by the hydraulic tunnel height, H¯, defined as 4×(cross-sectional area)/(perimeter). The values of H¯ range from 0.18 to 7.72 m, a factor of 43. Note that the scaling factor H¯ is not included in the correlation of Fig. 2. The factor H¯ will be included in the correlation in the next section. If we ignore the local variations

Turbulence model

In our previous study of the reverse stratified flow generated by a floor-level fire, the standard k-ε turbulence model was employed for simulation [15]. The advantages of this model are its simplicity and cost effectiveness. For fire applications, one of fundamental limitations of this model is the averaging procedure at the root of the model equations. Since the k-ε model was developed as a time-averaged approximation to the conservation equations of fluid dynamics, the results of fire

Critical ventilation velocity, ucr*

For a fixed heat generation rate Q, the program was run with a selected ventilation velocity (volumetric flow rate) and the formation of the back-flow was checked. Runs were repeated until a range of ventilation velocities encompassed the conditions exhibiting back-flow and no back-flow along the ceiling relative to the upwind edge of the fire zone. Fig. 9, Fig. 10 show the plots of ventilation velocity uin versus the heat generation rate Q for the two tunnels used in the simulation. As shown

Discussions and conclusions

In the present investigation, a CFD code FDS2 was employed to predict the critical ventilation velocity in fire tunnels. Tunnels of different sizes and fire-source geometries were selected for simulations. The following observations are made from the present study.

  • (1)

    When the critical ventilation velocity uin is plotted against the fire heat generation rate Q as shown in Fig. 2, ucr is roughly proportional to the 1/5 power of Q. This plot encompasses all available data of tunnel sizes. It is noted

References (31)

  • J.G. Quintiere

    Scaling applications in fire research

    Fire Safety J

    (1989)
  • P.H. Thomas

    Movement of smoke in horizontal corridors against an air flow

    Inst Fire Engrs Q

    (1970)
  • Massachusetts Highway Department. Boston, MA: Memorial tunnel fire ventilation test program, Test Report,...
  • Heselden AJM. Studies of fire and smoke behavior relevant to tunnels. In: Proceedings of the second international...
  • Guelzim A, Souil JM, Vantelon JP, Sou DK, Gabay D, Dallest D. Modelling of a reverse layer of fire-induced smoke in a...
  • Cited by (179)

    View all citing articles on Scopus
    View full text