FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW

ISFV14-14 th International Symposium on Flow Visualization June 21-24, 2010, EXCO Daegu, Korea FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW Gallo M.*, Kunsch J. P. and Rösgen T. Institute of Fluid Dynamics, ETH Zürich, Zürich 8092, Switzerland *Corresponding author, e-mail: gallo@ifd.mavt.ethz.ch KEYWORDS: Main subject: hot jet in confined cross-flow Fluid: air Visualization method: thermocouples Other keywords: tunnel fire simulation ABSTRACT: In order to optimize the emergency ventilation strategies put into operation during a tunnel fire, a better knowledge of the temperature and flow fields generated by a tunnel fire is required. Tests have been conducted in a model tunnel at ETH for different ventilation regimes indicating strong changes in the temperature field. 1 Introduction During tunnel fires, smoke control by adequate ventilation is fundamental for safe escape and evacuation of tunnel users. Owing to the confined geometry of a tunnel, the temperature can easily exceed 1000 C and the dispersion of smoke gases contributes to asphyxia risk. One important strategy to control the smoke dispersion in case of a fire is the longitudinal ventilation. The velocity of the longitudinal air flow is critical when the smoke front is arrested, i.e. when the smoke is prevented from propagating upstream against the fresh air current. Many experimental investigations related to the critical ventilation velocity have been performed, e.g. by Vauquelin [1], Wu and Bakar [2], Lee and Ryou [3], Atkinson and Wu [4] and Gallo et al [5]. From a theoretical point of view, the analytical models of Danziger and Kennedy [6] or Kunsch [7] may be pointed out. Numerous numerical studies are devoted to different aspects of emergency ventilation strategies and the corresponding flow regimes, see e.g. Brandeis and Bergmann [8] and Fletcher et al. [9]. Ballesteros-Tajadura et al. [10] analyzed the influence of tunnel slope on smoke behavior while Hu et al. [11] focused on the maximum temperature under the ceiling. The simplified model of Migoya et al. [12] for the simulation of fires in road tunnels with longitudinal ventilation may complete this survey. At ETH Zurich, Rusch et al. [13] acquired systematically temperature and velocity data, in order to have well defined boundary conditions for CFD simulations and experimental data for their validation. The present paper is focused on the experimental analysis of the critical ventilation regime preventing the propagation of smoke against a longitudinal air flow. In this context, the backlayering zone containing the smoke gases between the seat of the fire and the smoke front is an important concept. We anticipate here Fig.10 which highlights the corresponding geometry and which suggests the flow physics dominated by stratification. The experimental facility at ETHZ (Fig.1), which has been designed for the physical modeling of tunnel fires, allows specifically the investigation of the backlayering phenomena. Three-dimensional flow structures, which are generated by the interaction between the hot jet and the cross flow, call for detailed visualization techniques. The most relevant 1

GALLO M., KUNSCH J. P. AND RÖSGEN T. quantities shown here are the temperature signals, acquired with type-k thermocouples. An accurate analysis, however, involves not only the mean temperature, but also the root mean squares (RMS) of the temperature signals. 2 Experimental apparatus The tunnel, sketched in Fig. 1, is mounted on a frame made of aluminum profiles. The square cross section of the tunnel has a height of 0.8m. The air intake consists of a nozzle, with a contraction in only one direction and an area ratio of 1.75. Upstream of the nozzle three fabric screens and a screen made of stainless steel create enough pressure loss to damp incoming disturbances from the laboratory. The ceiling and the side walls consist of insulating material, because the hot air propagates under the ceiling due to buoyancy. The floor is made of glass plates except at the location of the hot air source where the floor is made of insulating material able to sustain the high temperatures. The hot air source is realized by an electric heater which allows heating up the supplied air from the ambient temperature to about 500 C. In order to smooth the velocity profiles downstream of the heater, two screen layers made of fine stainless steel meshes are inserted in the heater exit plane. A diffuser with an opening angle of 4 degrees is mounted downstream of the screens, followed by a pipe with an inner diameter of 0.2m. Both the diffuser and the pipe are insulated with two layers of 30mm thick stone wool mats. The extraction fan drives the cross-flow and is connected to the tunnel by a 0.5m long nozzle. A 40mm thick honeycomb with a pitch of 5mm is mounted between the nozzle and the fan in order to prevent the formation of an upstream vortex generated by the fan. Three PID controllers are employed to assure the repeatability and stable operation of the experiments. The first PID controller uses the analog signal of a velocity anemometer, located immediately after the tunnel inlet nozzle, to control the motor of the extraction fan. The second PID unit controls the blower which supplies the electric heater with air. It receives the input signal from a volumetric flow rate sensor and keeps the cold air volume supply constant. The third PID controller reads its signal from a type-k thermocouple which is placed downstream of the electric heater and maintains the jet temperature constant during the operation. y z x Fig. 1 Experimental set-up 2

FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW 3 Temperature measurements In this section the measurement technique and the data processing procedure, applied for the extraction of the RMS from the acquired temperature signals, are described. Subsequently the experimental results will be discussed. 3.1 Experimental procedure The dimensionless numbers representing the thermo-fluid-dynamical behavior of the flow in the tunnel are the heat release rate Q & * and the Richardson number Ri. The heat release rate is given by the following relations: Q& Q& * = 2 (1) ρ C T D gd hj cf p p cf ( T T ) Q & = m& C (2) hj a and the Richardson number is defined by the following ratio: ρgh Ri = (3) ρ 2 au cf where ρ is the density, T the temperature, D the hot jet diameter, C p the specific heat of the air, g the acceleration of gravity, ρ the density rise above ambient, H the tunnel height, m& the mass flow rate and U the velocity. The subscript a refers to the ambient condition, cf to the cross-flow and hj to the hot jet. The choice of Q & *, which characterizes the fire plume from the energetic point of view, is very important to simulate realistic fire conditions. Indeed, as discussed by Heskestad [14], its value must be small enough ( 1) to allow physical modeling of the buoyancy-driven pool fire regime. The cross-flow velocity, appearing in the Ri number which characterizes stratification, is an important parameter to be chosen carefully in the experiments. When the cross-flow velocity is too small, backlayering occurs, i.e. the hot-air plume can propagate upstream in a direction opposite to the fresh air current. Backlayering may obstruct partially the cross section with the velocity anemometer used for the control of the extraction fan. This partial blockage would lead to an acceleration of the cold air flow and the velocity anemometer in the control loop would trigger a reduction of the cross flow velocity. As a consequence, enhanced backlayering would lead to global instability of the tunnel. However, when the cross-flow velocity is too large, entrainment of cold air into the hot jet would reduce the stratification below a relevant level. The coordinate system adopted to localize the various investigated cross-sections downstream of the hot jet is shown in Fig. 1. The locations and the test conditions are also summarized in Table 1. The temperature measurements were conducted using NiCr-Ni type-k thermocouples whose signals are acquired and amplified with a custom-built multichannel amplifier and are then logged with a A/D converter on a data acquisition board linked to a computer. In the generic tunnel cross section the temperature distributions are measured by means of rakes which can contain up to 15 probes but, in order to avoid an excessive blockage, only eight or seven probes, respectively are traversed at the same 3

GALLO M., KUNSCH J. P. AND RÖSGEN T. time. The probes are equally spaced at 5cm in the span-wise direction and the step size in vertical direction is chosen to be 2.5cm. The first and the last port of the rake are both placed at a distance of 5cm from the two side walls. The measurement position of the thermocouples closest to the ceiling is 1.25cm. At each vertical station the temperatures are recorded with a sampling rate of f sr =50Hz over 3 minutes, for the investigated sections positioned downstream of the hot jet and, over 1 minute, for the sections located upstream of the hot jet. In the different investigated tunnel cross sections the distributions of the mean and RMS of the temperature signals are evaluated interpolating onto a square grid (31x31), the temperature signals being measured on the 15x31 acquisition points. The mean temperature distributions, acquired downstream and upstream of the hot jet, are normalized by means T = m& T + m& T m& + m& and by the ambient temperature T a, of the bulk temperature ( ) ( ) b cf cf hj hj respectively. The maps of the mean and RMS of the temperature signals, acquired downstream of the heat source, have been used to perform a 3D-reconstruction of the temperature field which should provide, for this tunnel region, a better insight into the fluid thermal behavior along the tunnel axis. In all the figures shown in the next subsection, the center of the circular exit section of the hot jet is located at x/d=0, y/d=0 and z/d=0.5 and the cross-flow direction corresponds to the x-axis. For the computation of the RMS and main frequency, the time resolved temperature signals are first low-pass filtered using the M 4 filter function as discussed by Monaghan [15] (Fig. 2a). In a second step, the high-pass filtered signal is evaluated by the difference of the original signal and the low-pass filtered signal (Fig. 2b). From the high-pass filtered signal the RMS is computed. Ri * Q & Downstream of the hot jet cf hj Investigated sections From To Pitch 4.75 0.95 x/d=0.625 x/d=6.875 1.25D 10.3 0.95 x/d=0.625 x/d=6.875 1.25D Upstream of the hot jet Ri * Q & Investigated sections From To Pitch 10.3 0.95 x/d=-0.5 x/d=-1.75 1.25D 12.1 0.95 x/d=-0.5 x/d=-1.75 1.25D 14.5 0.95 x/d=-0.5 x/d=-1.75 1.25D Tab. 1 Test conditions and investigated planes The choice of the low-pass filter width is of crucial importance because it should be able to keep the largest coherent structures and, at the same time, remove the large scale fluctuations introduced by the control loops. Therefore the integral time scale for each temperature signal is computed by integrating the auto-correlation function R of the mean-free temperature signal from zero up to the first zero crossing (Fig. 3). 4

FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW T (ºC) T (ºC) a) t (s) b) t (s) Fig. 2 Measured temperature signal at (x/d, y/d, z/d)=(0.625, 0.5, 0.5): a) original and low-pass filtered temperature, b) high-pass filtered temperature, After computing the integral time scale for each temperature signal, an average value is computed for each x/d location. The found mean values vary between 5 and 15 seconds and the median value finally chosen is 10 seconds. This value is used as low-pass filter width and the cut-off frequency is given by f co =0.1Hz. R t (s) Fig. 3 Auto-correlation of the mean-free original signal 4 Results 4.1 Temperatures measured downstream of the hot jet Fig. 4 reports, for the tunnel volume explored during the investigations, both the iso-t/t b surfaces relative to the value of T/T b =1.37 and the slice z/d=0.2 for the two tested Richardson numbers. It is well known from the literature that the unbounded jet in cross-flow is characterized, downstream of the exit section of the jet, by a counter-rotating vortex pair (CVP) with its axis along the cross-flow direction [16]. For both tested Richardson numbers, it is possible to observe for T/T b =1.37 two separated iso-t/t b surfaces which are associated with the CVP cited above. 5

GALLO M., KUNSCH J. P. AND RÖSGEN T. T/T b Ri=4.75 a) Cross-flow b) Fig. 4 Dimensionless temperature distribution at z/d=0.2 and iso-t/t b surface for T/T b =1.37: a) Ri=4.75; b). For the lower Richardson number (Fig. 4a), i.e. for the higher cross-flow velocity, it can be noted that the cross flow exhibits a better capability to wrap the CVP because the shear stresses on the boundaries between the two counter-rotating vortices and the cross-flow are more intense. The high kinetic energy contained in the cross-flow promotes a better cooling of the two hot counter-rotating vortices. In addition, the buoyancy effects of the hot jet become clearly visible. This is highlighted by the fact that, for 0.625<x/D<3.125, in proximity of the ceiling the measured temperatures are lower than the ones measured, in the same region, for (Fig. 4b). When the cross-flow velocity is reduced, the buoyancy effects become stronger. Indeed, from the T/T b distribution relative to z/d=0.2 and (Fig. 4b), it is possible to see in proximity of the ceiling (0<x/D<1.875) both a significant increase of the temperature and an upstream movement of the iso-t/t b curves. This can be explained by a lower cross-flow rate and hence a higher temperature of the two hot counter-rotating vortices reaching the ceiling and splitting in both directions, upstream and downstream. The upstream movement of the flow at the ceiling, known as backlayering, is absent in the case of Ri=4.75 (Fig. 4a). In fact at the inlet of the test region close to the ceiling, the measured temperatures appear to be very low. After the section x/d=3.125, Fig. 4 shows, for both tested Richardson numbers, a stratification of the temperature, but the one with a stronger temperature gradient is that for (Fig. 4b). From Fig. 5, reporting the iso-t/t b surfaces for the value of T/T b =1.02 for the two tested Richardson numbers, it is possible to see in case of a higher cross-flow velocity (Fig. 5a) that the iso-t/t b surface shows a peak between x/d=2 and x/d=4, while the iso-t/t b surface related to the lower cross-flow velocity (Fig. 4b) is rather flat. This different behavior can be explained by the fact that a high cross flow velocity produces a strong depression immediately after the hot jet that together with the rotation direction of the CVP, should promote secondary flow fields responsible for the upwards movement of the cooler air close to the floor of the tunnel. 6

FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW Ri=4.75 a) Cross-flow b) Fig. 5 Iso-T/T b surface for T/T b =1.02: a) Ri=4.75; b) In Fig. 6 the iso-rms surfaces are reported. For both Richardson numbers the iso-rms surfaces are relative to the value RMS=2.09ºC. Comparing the iso-t/t b surfaces, shown in Fig. 4, with the iso-rms surfaces, reported in Fig. 6, it is possible to see that the regions of the tunnel volume presenting high RMS values are those positioned immediately downstream of the hot jet and crossed by the high temperature counter rotating vortices. Ri=4.75 a) Cross-flow b) Fig. 6 Iso-RMS surfaces relative to RMS=2.09 C: a) Ri=4.75; b) 4.2 Temperatures measured upstream of the hot jet In order to better understand the backlayering phenomenon, temperature measurements were performed at two different x/d locations upstream of the hot jet (Table 1). In Fig. 7 the normalized temperature distributions T/T a are placed in two rows, each one acquired in one investigated section. The Richardson number increases by moving from the left to the right. Two preliminary features are apparent in Fig. 7. First, the temperature distributions do not exhibit significant variations in the spanwise direction z, except in proximity of the side walls, where a temperature decrease can be attributed to heat transfer effects. Second, for the front of the backlayering region does not reach the section x/d=-1.75, where the measured temperatures are effectively equal to the ambient temperature. At this point it is necessary to introduce two definitions which will be used to better describe and characterize the backlayering phenomenon. In the present contribution the backlayering thickness d bl is defined as the distance from the ceiling to the location where the measured temperature profile presents a kink, which is identified by calculating the maximum of the first derivative of the vertical temperature profile. An inspection of Fig.10 confirms the plausibility of this definition. The backlayering thickness is normalized by means of the hydraulic diameter of the tunnel D. The mean temperature measured in the rectangular region identified by the backlayering thickness and by the tunnel width is defined as backlayering temperature T bl. This quantity (T bl ) si normalized by means of 7

GALLO M., KUNSCH J. P. AND RÖSGEN T. T a. As can be seen from the maps reported in Fig. 7 and, more quantitatively, from the trends reported in the diagrams of Fig. 8, both the temperature T bl /T a and the thickness d bl /D of the backlayering increase as the Richardson number increases for both the investigated planes. T/T a x/d=-0.5 x/d=-0.5 Ri=12.1 x/d=-0.5 Ri=14.5 No backlayering T/T a x/d=-1.75 x/d=-1.75 Ri=12.1 x/d=-1.75 Ri=14.5 Fig. 7 Normalized temperature distributions T/T a acquired upstream of the hot source T bl /T a d bl /D a) b) Ri Ri Fig. 8 Dependence of the normalized backlayering temperature T bl /T a and thickness d bl /D on the Richardson number 8

FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW RMS x/d=-0.5 x/d=-0.5 Ri=12.1 x/d=-0.5 Ri=14.5 RMS No backlayering x/d=-1.75 x/d=-1.75 Ri=12.1 x/d=-1.75 Ri=14.5 Fig. 9 RMS temperature distributions acquired upstream of the hot source. In Fig. 9, the normalized temperature distributions of Fig. 7 are replaced by RMS temperature distributions, using the same locations x/d and the same parameter values for the Richardson number. An interesting observation is that, for both x/d locations, the maximum RMS values can be found in the backlayering region. In the section x/d=-1.75 however, the RMS values of the temperature are larger in the shear flow region between the backlayering and the cross flow; these values tend to decrease towards the ceiling. On the contrary at x/d=-0.5, the RMS values measured within the backlayering zone are quite uniform, except for. For this ventilation regime, the RMS distribution is quite similar to the ones at x/d=-1.75 for Ri= 12.1 and Ri=14.5. These considerations lead to the conclusion that the high RMS values observed in the shear flow region between the backlayering zone and the cross flow decrease with increasing distance from the backlayering front. As a consequence, the distribution of the RMS values becomes more uniform. In Fig. 10 the side view of the tunnel is shown for three different flow regimes, characterized by the Richardson numbers 14.5, 12.1 and 10.3. Temperature profiles obtained by spatial averaging in the span-wise direction (z) are presented in two cross sections located at x/d=-0.50 and x/d=-1.75. The contours of the backlayering region (see dashed lines in Fig. 10) can be qualitatively drawn by using the definition of the backlayering thickness introduced at the beginning of the present sub section. Following the three sketches in Fig. 10 from top to bottom, it can be observed qualitatively, how the backlayering zone becomes smaller as the cross-flow velocity increases i.e. as the Richardson number decreases. Fig. 11 shows that the RMS temperature distributions are analogous to Fig. 10 for the same experimental conditions. 9

GALLO M., KUNSCH J. P. AND RÖSGEN T. Fig. 10 Overview of the shape of the backlayering zones (delimited by the dashed lines), depending on the stratification parameter i.e. Richardson number; Temperature profiles obtained by spatial averaging in the spanwise direction (z) relative to the cross sections located at x/d=-0.50 and x/d=-1.75. 10

FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW Fig. 11 Overview of the shape of the backlayering zones (delimited by the dashed lines), depending on the stratification parameter i.e. Richardson number; RMS temperature profiles obtained by spatial averaging in the span-wise direction (z) relative to the cross sections located at x/d=-0.50 and x/d=-1.75. 5 Conclusions Measurements of the temperature field in different cross sections of a model tunnel, used for physical modeling of a tunnel fire, have been carried out by using a rake of thermocouples. The measurements performed downstream of the hot source were used to perform a 3D reconstruction of the temperature field highlighting significant differences between the two tested ventilation regimes. These differences concern both the temperature distributions showing a strong stratification, expressed in terms of Richardson numbers, and the behavior of the iso-rms surfaces characterizing the counter rotating vortex pair CVP, which is generated by the interaction between the hot jet and the cross flow. The measurements performed upstream of the hot source show that the thickness and the temperature of the backlayering zone increase with increasing Richardson number. In addition, a homogenization of 11

GALLO M., KUNSCH J. P. AND RÖSGEN T. the temperature distribution in the backlayering zone can be observed with increasing distance from the front of the backlayering zone. References 1. Vauquelin O. Parametrical study of the back flow occurrence in case of a buoyant release into a rectangular channel. Experimental Thermal and Fluid Science, Vol. 29, No. 5, pp 725-731, 2005. 2. Wu Y and Bakar M.Z.A. Control of smoke flow in tunnel fires using longitudinal ventilation system a study of critical velocity. Fire Safety Journal, Vol. 25, No. 4, pp 363-390, 2000. 3. Lee S R and Ryou H S. An experimental study of the effect of the aspect ratio on the critical velocity in longitudinal ventilation tunnel fires. Journal of Fire Sciences, Vol. 23, pp 119-138, 2004. 4. Atkinson G T and Wu Y. Smoke control in sloping tunnels. Fire Safety Journal, Vol. 27, pp 335-341, 1996. 5. Gallo M, Kunsch J P and Rösgen T. Temperature sensing array for parallel measurements in tunnel fire models. In Proceedings of the 10 th International Conference Fluid Controls, Measurements and Visualization, Moscow, 2009. 6. Danziger N H and Kennedy W D. Longitudinal ventilation analysis for the Gleenwood Canyon Tunnels. 4 th International Symposium on Aerodynamics and Ventilation of Vehicles Tunnels. BHr Group, pp 169-186, 1982 7. Kunsch J P. Simple model for control of fire gases in a ventilated tunnel. Fire Safety Journal, Vol. 37, pp 67-81, 2002. 8. Brandeis J and Bergmann D J. A numerical study of tunnel fires. Combustion Science and Technology, Vol. 35, pp 133-135, 1983. 9. Fletcher D F, Kent J H, Apte J H and Green A R. Numerical simulations of smoke movement from a pool fire in a ventilation tunnel. Fire Safety Journal, Vol. 23, pp 305-325, 1994. 10. Ballesteros-Tajadura R, Santolaria-Morros C and Blanco-Marigorta E. Influence of the slope in the ventilation semi-transversal system of an urban tunnel. Tunneling and Underground Space Technology, Vol. 21, pp 21-28, 2006. 11. Hu L H, Huo R, Li Y Z, Peng W, Chow W K and Yang R X. On the maximum smoke temperature under the ceiling in the tunnel fires. Tunneling and Underground Space Technology, Vol. 21, pp 650-655, 2006. 12. Migoya E, Crespo A, Garcia J and Hernandez J. A simplified model of fires in road tunnels. Comparison with three-dimensional models and full-scale measurements. Tunneling and Underground Space Technology, Vol. 24, pp 37-52, 2009. 13. Rusch D, Blum L, Moser A and Roesgen T. Turbulence model validation for fire simulation by CFD and experimental investigation of a hot jet in cross-flow. Fire Safety Journal, Vol. 43, pp 429-441, 2008. 14. Heskestad G. Fire plumes, flame height and air entrainment. The SFPE Handbook of Fire Protection Engineering, National Fire Protection Association, NY, 2002. 15. Monaghan J. Extrapolating B-splines for interpolation. Journal of computational physics, Vol. 60, No. 2, pp 253-262, 1985. 16. Fric T F and Roshko A. Vortical structure in the wake of a transverse jet. Journal of Fluid Mechanics, Vol. 279, pp 1-47, 1994. 12

FLUID TEMPERATURE MEASUREMENTS FOR A HOT JET IN CONFINED CROSS-FLOW Copyright Statement The authors confirm that they, and/or their company or institution, hold copyright on all of the original material included in their paper. They also confirm they have obtained permission, from the copyright holder of any third party material included in their paper, to publish it as part of their paper. The authors grant full permission for the publication and distribution of their paper as part of the ISFV14 proceedings or as individual off-prints from the proceedings. 13