The Effect of Cross-sectional Area and Air Velocity on the Conditions in a Tunnel during a Fire

The Effect of Cross-sectional Area and Air Velocity on the Conditions in a Tunnel during a Fire Anders Lönnermark and Haukur Ingason SP Technical Research Institute of Sweden Fire Technology SP Report 27:5

The Effect of Cross-sectional Area and Air Velocity on the Conditions in a Tunnel during a Fire Anders Lönnermark and Haukur Ingason

2 Abstract Tests in a model-scale tunnel (scale 1:2) were performed to study the effect of the height and width of a tunnel on the mass loss rate, heat release rate, and gas temperatures. The tunnel was 1 m long. The widths used were.3 m,.45 m, and.6 m and the height was varied between.25 m and.4 m. Two different types of fuels were used: pools of heptane and wood cribs. The experimental results show that the tunnel dimensions have an effect on the Mass Loss Rate (MLR) of the fuel, the Heat Release Rate (HRR) and the gas temperatures. The dependency of the MLR and the HRR on the tunnel dimensions are different from each other. This is especially obvious for pool fires. The results also indicate that the influence of tunnel dimensions is not only a radiation effect, as often is assumed. It is probably a combination of radiation from surfaces and hot gases, influence of air flow patterns, the shape and position of the flame and combustion zone, and temperature distribution. The tests show that the gas temperature depends on the tunnel dimension. Generally the temperature decreases with increasing dimensions. However, the extent of the effect varied at different positions in the tunnel and the effect in some positions was relatively small. Furthermore, the effect on the temperature in the vicinity of the fire was opposite to the effect further downstream. The analysis shows that as several factors and processes are interacting, it is important to know the starting conditions (type and geometry of the fuel, geometry and dimension of the tunnel, ventilation, etc.) to be able to predict the effect of a change in a specific parameter, e.g. HRR or MLR. Key words: tunnel, fire, gas temperature, gas velocity, heat release rate, cross-sectional area SP Tekniska SP Technical Research Forskningsinstitut Institute of Sweden SP Rapport 27:5 SP Report 27:5 ISBN 91-85533-69-6 ISSN 284-5172 Borås 27 Postal address: Box 857, SE-51 15 BORÅS, Sweden Telephone: +46 1 516 5 Telefax: +46 33 13 55 2 E-mail: info@sp.se http://www.sp.se

3 Contents Abstract 2 Contents 3 Preface 4 Summary 5 Nomenclature 6 1 Introduction 9 2 Theoretical aspects 11 2.1 Scale modelling 11 2.2 The Heat Release Rate (HRR) 12 2.3 Gas temperature 12 2.4 Wood porosity 13 3 Experimental set-up 15 3.1 Tunnel 15 3.2 Fuel 17 3.2.1 Wood cribs 17 3.2.2 Heptane pools 18 3.3 Measurements 19 3.3.1 Gas temperature measurements 19 3.3.2 Gas velocity 2 3.3.3 Gas analysis 2 3.3.4 Heat flux 2 4 Test procedure 21 5 Results 24 5.1 HRR from free burn tests 24 5.1.1 Heptane pool 24 5.1.2 Wood cribs 24 5.2 Effect of geometry on MLR and HRR 26 5.3 Effect of ventilation on MLR and HRR 3 5.4 Effect of geometry on gas temperature 32 5.5 Temperature distribution along the tunnel 37 5.6 Effect of blockage on HRR 38 6 Discussion 4 6.1 Effect of geometry on MLR and HRR 4 6.2 Effect of ventilation on MLR and HRR 41 6.3 Effect of geometry on gas temperature 41 7 Conclusions 43 8 References 45 Appendix A Measurement positions 48 Appendix B Time-resolved test results 51

4 Preface This project was financed by the Swedish Research Council (FORMAS). The technicians Michael Magnusson, Joel Blom, Lars Gustafsson and Sven-Gunnar Gustafsson at SP Fire Technology are acknowledged for their valuable assistance during performance of the tests. They were also responsible for the construction of the test rig. Jonatan Hugosson is acknowledged for his assistance in producing time-resolved graphs.

5 Summary This report focuses on the effect of the tunnel dimensions, i.e. the height and the width of the tunnel together with the effects of ventilation rate on Mass Loss Rate (MLR) of the fuel, the Heat Release Rate (HRR) and gas temperatures. Therefore, tests in a model-scale tunnel (scale 1:2) were performed. Two different types of fuels were used: pools of heptane and wood cribs. The model tunnel was 1 m long with varying width and height dimensions. The main conclusions from the tests are: The dependency of the MLR of the fuel and the HRR on the tunnel dimensions are different from each other, especially for pool fires. The effect of the height and the width of the tunnel on the MLR and HRR depends on the starting conditions. For a pool fire in a tunnel with relatively small height or width, an increase in tunnel dimensions leads to an increase in fire size while for larger height and width, an increase in tunnel dimensions instead leads to a decrease in fire size. The situation for wood crib fires is more complex and relations opposite to those found for heptane fires can be found for some conditions. It was found that an increasing ventilation rate increases the maximum HRR. The increase in the maximum HRR for well ventilated cribs is in the range of 1.3 to 1.7 times the value measured outside the tunnel under ambient conditions and for wind velocities equal to or higher than.45 m/s. The increase in the maximum HRR of low ventilated cribs was 1.8 (.67 m/s) and 2. (1.12 m/s), respectively, for the higher velocities. When compared to ambient conditions inside the tunnel based on MLR, this increase is lower. For the case with a velocity of.67 m/s, the fire growth rate increased by a factor of 5 1 times the free burn case, depending on the dimensions of the tunnel crosssection. Above a certain air velocity, the ratio for HRR does not seem to vary significantly with the velocity. The gas temperature is affected by the dimension of the tunnel. In general there is a decrease in temperature with increasing dimension, but not near the fire where an increase in gas temperature can be seen. The effects on tests with the two different fuels are similar, but not identical. Near the seat of the fire the burning characteristics for the two fuels are different. In this region the three-dimensional effects in the wood crib case and the effect of the flame position for the heptane case become significant. For the tunnels with a small cross-section, the temperature is relatively uniform, while an increased stratification and difference between average temperature and ceiling temperature can be observed when the width or height of the tunnel is increased. The model of the average temperature correlates relatively well with the measured temperature at half the tunnel height and the average value calculated from temperatures measured in a vertical thermocouple tree, at least a distance away from the fire. Close to the fire the difference can be large.

6 Nomenclature A Area [m 2 ] b The side of the square cross section of the sticks in the wood cribs [m] B Area of the bottom of a wood crib [m 2 ] c p D H E Heat capacity [kj/kg/k] Hydraulic diameter [m] Amount of energy developed per consumed kilogram of oxygen, 13.1 [MJ/kg] Fr Froude number g Acceleration of gravity [m/s 2 ] h Heat transfer coefficient [kw/m 2 /K] H Height [m] Δh c k p L m m& Heat of combustion [MJ/kg] Correlation coefficient for the bi-directional probe Length [m] Mass [kg] Air mass flow rate [kg/s] M O 2 Molecular weight for oxygen [g/mol] M air Molecular weight for air (actually the molar weight for the gas flow in the duct, n l n L N l N L P Δp Q Q & S l S L t T u u [g/mol] Number of sticks in a layer with short sticks in a wood crib Number of sticks in a layer with long sticks in a wood crib Number of layers with short sticks in a wood crib Number of layers with long sticks in a wood crib Perimeter of the tunnel [m] Porosity [m] Pressure difference [Pa] Energy [kj] Heat Release Rate (HRR) from the fire [kw] Distance between the long sticks in a wood crib [m] Distance between the short sticks in a wood crib [m] Time [s] Gas temperature [K] Gas velocity [m/s] Average velocity (for the cross-section) [m/s]

7 u max V & W x Velocity in the centre of the cross-section [m/s] Volume flow rate [m 3 /s] Width [m] Distance [m] X O 2 Mole fraction for O 2 in the ambient air, measured on dry gases [mol/mol] X CO 2 Mole fraction for CO 2 in the ambient air, measured on dry gases [mol/mol] X H 2 O Mole fraction for H 2 O in the ambient air [mol/mol] X O 2 Mole fraction for O 2 in the flue gases, measured on dry gases [mol/mol] X CO 2 Mole fraction for CO 2 in the flue gases, measured on dry gases [mol/mol] x Distance from centre of fire source [m] Greek α ζ Ratio between the number of moles of combustion products including nitrogen and the number of moles of reactants including nitrogen (expansion factor) A theoretically determined mass flow correction factor ρ Density at ambient temperature [kg/m 3 ] Subscript Corresponds to ambient conditions avg Average fb free burn F Full scale H Hydraulic diameter M Model scale s Exposed surface v refer to the cross sectional area of the vertical crib shafts Abbreviations C HF HGV HRR Centre Heat flux Heavy goods vehicle Heat release rate

8 L MLR PT R TC Left Mass loss rate Plate thermometer Right Thermocouple

9 1 Introduction In the light of the recent catastrophic fires in tunnels and results from large-scale fire tests in tunnels during the past few years, much effort has been put into finding representative design fires for tunnel safety. Such work usually includes the definition of different types of fire scenarios in different types of tunnels, depending on tunnel length, tunnel location, traffic intensity, type of transport through the tunnel, etc. However, a central part of this work is to determine what fire development and maximum heat release rate (HRR) can be assigned to each defined fire scenario. Such information can be based on conclusions from real fires or data from fire tests in tunnels. For some cases, HRR data is only available from free burn tests, i.e. not from tests inside a tunnel. In such cases it is important to be able to convert the results so that they are relevant for the corresponding case inside a tunnel. Even if a test similar to the scenario of interest has been performed inside a tunnel, it is not always straight forward to use the results if various parameters are changed. The main objective of the study presented here is to find out experimentally using model-scale tests, the impact of the geometrical cross section and ventilation on the maximum HRR, maximum ceiling gas temperatures and fire growth rates of liquid and solid fuels such as wood cribs. Conditions inside the model tunnel are compared to situation outside the tunnel with and without longitudinal ventilation. The reason for this is that many of the studies carried out previously do lack a systematic data focusing on these parameters. For example, using free-burn results can lead to an underestimation of the HRR, since both the fire growth and fire spread rate and the maximum HRR can increase in a tunnel, relative to free-burn. This is thought to be mainly due to the increased feedback from the flames and the hot smoke layer. This is a well known phenomenon from enclosure fires [1], but has also been seen in tunnel fire tests [2]. Takeda and Akita studied this phenomenon and showed that it is also related to the ventilation factor [3]. Carvel et al. have studied both the effect of the tunnel on the HRR [4] and the effect of the ventilation on the HRR from a fire in a tunnel [5-7]. Ingason performed pool fire tests in a model scale tunnel, using heptane, methanol, and xylene as fuels [8, 9]. For heptane, the maximum increase of the burning rate due to the tunnel was by a factor of 3.3 (.13 kg/s/m 2 (u = 1 m/s) compared to.4 kg/s/m 2 (free- burn)). Saito et al. [1] showed that the mass loss rate (MLR) for liquid fires increased in a tunnel compared to free-burn conditions. The tests were performed with pool fires with methanol (.1 m,.15 m,.2 m, and.25 m in diameter) and heptane (.15 m in diameter). For the two smallest pools the effect of the tunnel (with an air velocity.8 m/s) on the MLR of methanol was only a few percent, while for the.25 m diameter pool the MLR in the tunnel was increased by a factor 2.7 compared to free-burn conditions. For heptane, the tunnel (with an air velocity of.43 m/s) increased the MLR by approximately a factor 4. For both fuels the MLR was significantly decreased with increasing air velocity. This illustrates the importance of the heat feed back from the flames, hot gases, and tunnel structure on the MLR. Ingason [11] carried out fire tests in a 1:23 model scale tunnel. Fire loads corresponding to a HGV trailer were simulated with aid of wood cribs of two different sizes. Longitudinal ventilation was tested under different fire conditions. The effects of different ventilation rates on the fire growth rate, fire spread, flame length, gas temperatures and backlayering were investigated. An essential aim of Ingason s study was to investigate the influence of the ventilation rate on the maximum HRR and the fire growth rate for solid fuels (wood crib in this case). An important aspect was to study it for fuels that are comparable from a porosity point of view. It was found that an increasing ventilation rate increases the maximum HRR per unit fuel area. The increase in the maximum HRR per unit fuel area is in the range of 1.4 to 1.55 times the value measured outside the tunnel under ambient conditions. This is much lower than earlier studies by Carvel et al. [5, 12] on influences of ventilation on maximum

1 HRR for solid fuels. One possible explanation was thought to be the way the fuel was compared. The influence of the longitudinal ventilation rate on the fire growth rate was found to be much higher than on the maximum HRR. The increase at a velocity corresponding to 3 m/s (.62 m/s) in large scale is in the order of 2 times the fire growth rate under ambient conditions and for 5 m/s (1.4 m/s) it is in the order of 3. The results obtained in Ingason [11] study confirm the results of Carvel et al., i.e., increasing longitudinal ventilation rate increases the fire growth rate although the numbers were not the same. Carvel et al. performed an analysis of the HRR enhancement of a tunnel compared to corresponding fire situation in the open air [4]. Results from a number of different experimental test series published in the literature were studied. These test series included a wide range of cross-sectional areas: from model-scale (.9 m 2 ) to real-scale (8 m 2 ). Bayes theorem concerning conditional probability was used to study which of the parameters, i.e., tunnel height, tunnel width, blockage ratio of the fire load, and mean hydraulic diameter, had a significant influence on the HRR enhancement. The work includes an extensive summary of experiments involving liquid pools, wood cribs, and cars and how the HRR from tests in tunnels is compared to similar free burn tests. The authors drew the conclusion that the width of the tunnel (or the distance from the fire load to the walls) has a significant influence on the HRR from a fire in a tunnel. This result can be explained by the radiation to the pool (from walls and from the ceiling layer), the temperature inside the tunnel, and also the flow pattern near the fire. The analysis also gave the result that the height of the tunnel (or distance from the fire load to the ceiling) did not significantly affect the HRR enhancement. However, in most of the test series the distance between the fire load and the ceiling was not varied and therefore support for this conclusion is limited. This is one of the reasons for performing the test series presented in this paper. It is important to understand that several parameters affect the development of the HRR curve, e.g. the type of fuel used to represent the scenario, the air velocity inside the tunnel, and the tunnel geometry. This report focuses on the effect of the tunnel dimensions, i.e. the height and the width of the tunnel, on the maximum HRR, maximum ceiling gas temperature and the fire growth rate using wood cribs and pools of heptane in a model tunnel.

11 2 Theoretical aspects 2.1 Scale modelling When using scale modelling it is important that the similarity between the full-scale situation and the scale model is well-defined. A complete similarity involves for example both gas flow conditions and the effect of material properties. The gas flow conditions can be described by a number of non-dimensional numbers, e.g. the Froude number, the Reynolds number, and the Richardson number. For a perfect scaling all of these numbers should be the same in the model-scale model as in the full-scale case. This is, however, in most cases not possible and it is often enough to focus on the Froude number: 2 u Fr = (1) gl where u is the velocity, g is the acceleration of gravity, and L is the length. This so called Froude scaling has been used in the present study, i.e. the Froude number alone has been used to scale the conditions from the large scale to the model scale and vice versa. Information about scaling theories can be obtained from for example references [1, 13-15]. Table 2.1 A list of scaling correlations for the model tunnel. Type of unit Scaling model Equation number Heat Release Rate (HRR) 5/ 2 (2) L F (kw) & & Q F = QM LM Volume flow (m3/s) 5 / 2 = (3) LF V F V M LM Velocity (m/s) 1/ 2 (4) LF u = F um LM Time (s) 1/ 2 (5) L F t = F tm LM Energy (kj) 3 LF Δh (6) c, M QF = QM L M Δhc, F Mass (kg) 3 (7) L F m = F mm LM Temperature (K) T = (8) F T M a) Index M corresponds to the model scale and index F to the full scale (L M =1 and L F =2 in the present case). The model-scale tunnel used in the study presented here was built in scale 1:2, which means that the size of the tunnel is scaled geometrically according to this ratio. The parameters considered in the study and how they are scaled between the reality and the model are presented in Table 2.1. This includes the HRR, the time, flow rates, the energy content, and mass. The influence of the thermal inertia of the involved material is neglected. Since the Reynolds number is not kept the same in the different scales, the turbulence intensity in not considered in the study. A parameter important in studies of fires and fire spread is the

12 radiation but this parameter is not considered here. Previous studies have, however, proved that model-scale studies can give interesting results and give important information on fire behaviour when different parameters are varied [11, 16-18]. 2.2 The Heat Release Rate (HRR) The HRR was calculated with aid of oxygen calorimetry [19, 2] according to the equation: O2 O2 ( M O M ) ( ) 2 air 1 X H 2O 1 ( X O ( 1 X ) 2 CO2 ( X ( 1 X ) ( 1 X ) E m& Q& = (9) α 1 + X X where E is the amount of energy developed per consumed kilogram of oxygen (MJ/kg), M is the molar mass, X is the mole fraction (the superscript refers to ambient conditions), and α is expansion factor (the ratio between the number of moles of combustion products including nitrogen and the number of moles of reactants including nitrogen). The mass flow rate, m& a, was calculated from the following equation: O2 CO2 CO2 where T m& = ζ A u ρ (1) T u ζ = (11) u max is the ratio between the average velocity and the maximum (central) velocity in the exhaust tube, where the measurements for the HRR calculations were performed. This was controlled to be.868. In equation (1), A is the cross-sectional area of the tube, ρ the density of air at ambient temperature, T the ambient temperature, and T the actual temperature. The velocity inside the tunnel was obtained according to the following equation: u 1 k p 2ΔpT ρ T = (12) where k p is a correlation constant related to the bi-directional probe [21] and Δp is the differential pressure. 2.3 Gas temperature The temperature distribution along the tunnel is of interest to this study. Also how the shape of this distribution is altered when the test conditions are varied. The experimental data will be compared to a theoretical calculation of the average cross-sectional temperature along the tunnel. This model is described by Ingason [1]. It is based on the convective HRR from the fire, i.e. a fixed fraction of the total HRR is lost through flame radiation. The radiation losses are linearized and expressed in the same way as the convective losses. The total losses included as a function of the distance from the fire and are described by the total heat transfer coefficient, h, (h =.2 kw/(m 2 K) has been used in the calculation presented in this paper).the average temperature can be calculated as

13 T hpx 2 Q ρuat c avg = T + e 3 ρuac & (13) where T is the temperature, Q & is the HRR, ρ is the density, u is the velocity, A t is the crosssectional area of the tunnel, c is the heat capacity, h is the heat transfer coefficient, and x is the distance from the seat of the fire. The index corresponds to conditions of the incoming air. Equation (13) will be used to compare experimental data. 2.4 Wood porosity It has been shown that the burning rate of a wood crib can be related to its porosity [13, 22, 23]. Therefore, it is important to be able to define the porosity of a specific wood crib. The original works mentioned above, all relate to wood cribs with a square horizontal crosssection, i.e. all sticks in the crib were of the same length. In the present work the wood cribs were to simulate a vehicle, e.g. an HGV, and therefore the length was not the same as the width of the wood cribs. This means that the wood cribs constituted of both layers with long sticks and layers with short sticks (see Figure 2.1 and Figure 3.5) s L s l L b b l S l Figure 2.1 Drawing of a wood crib and definition of different lengths. Based on the original works mentioned above on porosity of wood cribs, Ingason developed the formulas to be valid also for wood cribs with sticks of different lengths in the two horizontal directions [11]. The porosity of a wood crib with the length L, the width l, and with a square cross-section of the sticks with the side b (see Figure 2.1) can be defined as: P A v 2 1 2 = s 1 H b (14) As where A v is the total cross-sectional area of the vertical crib shafts A v = ( L n b)( l n b) l L (15)

14 and A s is the total exposed surface area of the crib (based on the assumption that bottom and the top layers are with long sticks) A = 4 b s 2 ( n N l + n N L) + 2b ( n N + n N n n N ) B l l L L l l L L l L L (16) In the equation n l, n L, N l, and N L are the number of sticks in a layer with short sticks, the number of sticks in a layer with long sticks, the number of layers with short sticks, and the number of layers with long sticks, respectively. The parameter B is included to represent the area of the bottom of the wood crib that is not exposed. In the calculation of the porosity according to equation (14) the parameter s H is included. This parameter corresponds to the hydraulic diameter of the rectangular space defined by the parameters s l and s L : i.e. l nlb s = l nl 1 (17) L nlb s = L nl 1 (18) s H sl sl = 2 s + s l L (19)

15 3 Experimental set-up 3.1 Tunnel To study the influence of different parameters on fire development, gas temperatures and combustion conditions in a tunnel, free-burn tests and fire tests inside a model-scale tunnel were performed. The tunnel was 1 m long (see Figure 3.1 and Figure 3.2). The width and height of the tunnel were varied during the test series. The widths used were:.3 m,.45 m, and.6 m, and the heights were:.25 m and.4 m. The different cross sections used are presented in Figure 3.3. The scale of the tunnel was assumed to be 1:2 and therefore the widths correspond to 6 m, 9 m and 12 m in real scale, while the heights correspond to 5 m and 8 m, respectively. The ceiling, floor, and one of the walls were made of 15 mm thick PROMATECT -H boards. One of the walls was comprised of 15 windows of 5 mm thick fire proof glass set in steel frames (555 mm 41 mm visual access plus frame). Figure 3.1 The photo shows the fan, the mixing box, and the tunnel with its hood and exhaust. 125 125 125 125 125 125 125 125 Air flow x3 Fire x3 1.5H 3 Window 1 load cell 1 88 thermocouple pile thermocouple velocity thermocouple pile 35mm 3mm 3mm 78mm 125mm 172mm 22mm Plate thermometer wall thermocouple heat flux gage target gasanalysis FBG Figure 3.2 Side view of the model-scale tunnel with positions of measurements and gas sampling. Dimensions in mm.

16 45 A B C 6 3 3 45 125 4 4 4 25 25 25 D E F 6 Figure 3.3 The different cross-sectional areas (A to F) used during the test series. Dimensions in mm. A longitudinal ventilation were established by a fan in the upstream end of the tunnel (see Figure 3.1). The fan itself was.95 m long with an inner diameter of.35 m and.8 HP motor yielding a maximum capacity of 2 m 3 /h (at 14 rpm and 7.5 mmh 2 O). In the tunnel a maximum velocity of approximately 1.3 m/s could be reached with this fan. The rotation speed, and thereby the capacity, could be controlled by an electrical device coupled to the motor. The swirls created by the axial fan, was hampered by filling a mixing box consisting of 1 mm thick Promatect H boards with straw fibres and wooden wool. This mixing box created a smooth flow profile in the tunnel. The mixing box was.85 m long,.45 m wide and.25 m high (see Figure 3.4). Figure 3.4 a) b) The mixing box filled with straw, seen a) from the fan and b) from the tunnel. The fuel was positioned (and centred) at the centre of the tunnel, i.e. 5 m from the inlet and outlet of the tunnel. This position is denoted and different positions in the tunnel are related to this zero position with negative values in the upstream direction and positive values in the downstream direction. For convenience, centimetres have been used to name the different positions. The fuel placed on four.5 m high piles with pieces of PROMATECT -H standing on a.34 m.55 m.1 m PROMATECT -H board connected by metals rods to a digital balance beneath the tunnel floor. The top of the board was.2 m above the floor. This means that the bottom of the wood crib was.7 m above the tunnel floor. In the heptane tests, the fuel pan was standing directly on the PROMATECT -H board. The initial fuel surface was in

17 these cases.7 m above the tunnel floor. This means that the distance between the ceiling and the (initial) top of the fuel was 18 cm and 33 cm, respectively, for the heptane tests, and 12.5 cm and 27.5 cm, respectively, for the tests with wood cribs. The values correspond to the two different tunnel heights used. Four small wooden targets (15 mm 15 mm 17 mm) were placed at two different heights (on the floor and.2 m above the floor) at +.815 m and +2.3 m. The targets were dried together with the fuel and the Promatect H boards (6 ºC). The average weight of the dried targets was 1.5 g (1.6 g for the targets before drying). The wood cribs, targets, and pieces of Promatect H were dried in 6 ºC at least 24 hours before each test. The targets in Test 1 were dried in 5 C. During the tests the fuel on the Promatect board was weighed on a scale (Inv 7164; Sartorius EA6EDE-I). The fire gases were after the tunnel collected in a built hood and guided into an exhaust duct (.25 m in diameter). The hood was.88 m long,.88 m wide, and.9 m high. The wall opposite to the tunnel opening was extended downwards an extra.3 m giving a total height of 1.2 m. The measurements performed in the exhaust duct were used to calculate the HRR. 3.2 Fuel 3.2.1 Wood cribs Two types of fuels were used: wood cribs and the liquid fuel heptane. The wood cribs were of three different types. Two were wood cribs with two different porosities (P1 and P2) and one had the short pieces of wood replaced by pieces of polyethene. The standard wood crib (P1) was constructed of four layers of long sticks (.5 m) with four sticks in each layer and three layers of short sticks (.15 m) with three sticks in each layer (see Figure 3.5). The sides of the square cross section of the sticks were.15 m for both the long and the short sticks. This gave a total height of.15 m. According to equation (14) the porosity of this wood crib (P1) was 2.1 mm and according to equation (16) the exposed fuel surface A is.54 m 2. s To study the effect of the porosity of the wood crib on the results (e.g. the effect of the ventilation on the HRR), tests with a wood crib, with a porosity (P2) differing from the standard wood crib, were performed. In this case the sides of the square cross section of the sticks were.1 m. The wood crib was constructed of five layers of long (.5 m) sticks and four layers of short (.15 m) sticks. For both long and short sticks there were seven sticks in each layer. This gave a total height of.9 m. According to equation (14), the porosity of this wood crib (P2) was.62 mm and according to equation (16) the exposed fuel surface A s is.8 m 2. In the third type of wood crib all the short sticks were replaced by sticks of polyethene. The dimensions were the same as for the standard wood crib, but the weights were different. The weights for the wood cribs are given in Table 4.1 together with the presentation of the test conditions.

18 a) Figure 3.5 b) c) Wood cribs: a) with porosity P1, b) with porosity P2, and c) with wood and plastics (porosity P1). In the tests with a height of.7 m between the floor and the base of the fuel, the wood cribs was were placed on four 5 cm high piles with pieces of Promatect H (two pieces.2 m.15 m.2 m and one piece.2 m.15 m.1 m). 3.2.2 Heptane pools In some tests, the liquid fuel heptane was used. In all cases an almost square pan (.155 m.16 m) was used with a.5 m thick layer of heptane. The fuel pan was standing directly on the Promatect H board (see Figure 3.6). The initial fuel surface was in these cases.7 m above the tunnel floor.

19 Figure 3.6 The heptane pool placed on a Promatect H board connected to a scale. 3.3 Measurements In this Section the different types of measurements performed are described. Details on measurement positions are given in Appendix A. 3.3.1 Gas temperature measurements The gas temperatures were measured at many different positions in the tunnel: near the ceiling, along the centreline of the tunnel and in a number of thermocouple trees registering the temperature distribution over the cross section. In most cases bare thermocouples Type K,.25 mm in diameter were used. Type K is suitable for continuous measurements within the temperature range from -25 C to 126 C. There are number of sources of errors in gas temperature measurement that can be more or less important. These are mainly due to the measurement situation, e.g. temporal and spatial resolution, radiation effects, conduction heat transfer along the wire, and chemical catalysis on the thermocouple surface, especially radical recombination reactions. These problems and different solution to quantify and decrease the errors have been presented in the literature [24-28]. In many fire situations, the radiation effect can have a significant influence on the measured result and this can lead to a large discrepancy between the measured temperature and the actual temperature of the gas. Lönnermark et al. [29, 3] have discussed this problem for tunnels and presented some estimations on the errors obtained. The wall temperature inside and outside the Promatect board was measured at half the height (i.e. 125 mm in the tunnels with the height 25 mm and at 2 mm in the tunnel with the height 4 mm) 375 cm from the seat of the fire. Plate thermometers were installed at three positions. This is further described in Section 3.3.4.

2 3.3.2 Gas velocity The gas velocity in the tunnel was measured at the centre of the tunnel at two different positions in the tunnel: -2.75 m and +3.65 m. The velocity was measured with bi-directional probes [21] (inner diameter = 14 mm, outer diameter = 16 mm, and length = 32 mm) connected to differential pressure instruments (Furness Controls Differential Pressure Transmitter Model 332) and the velocity was calculated according to equation (12). The gas velocity in the tunnel was adjusted according to measurement with hand held velocity measurement instrument (TSI VELOCICHECK 833-M-S). The same instrument was also used to register the velocity contour over the cross-section before each test. 3.3.3 Gas analysis Gas analyses were performed on gases sampled at two different heights (3 mm from the ceiling and at half the tunnel height) 3.85 m from the seat of the fire. Concentrations of oxygen (O 2 ), carbon monoxide (CO), and carbon dioxide (CO 2 ) were analysed. Gas analyses for the same species were also performed in the exhaust duct from the hood at the end of the tunnel. These gas analyses were used to calculate the HRR (based on oxygen consumption calorimetry [19, 2]) from the fires, equation (9). The gas analysis was performed approximately 1.5 m from the exit of the hood. The gas velocity and temperature in the duct were measured.5 m closer to the hood. The gas analysers were calibrated every day of testing. The pressure transducers were calibrated before the test series and checked again in the middle of the test series. The hood system was controlled by performing three calibration tests. Pans (two tests with the same pan as used in the tunnel tests,.155 m.16 m, and one test with a.1 m high pan with a diameter of.8 m) with heptane were placed on a scale beneath the hood. The bottom of the pan was positioned 8 cm from the top of the hood. 3.3.4 Heat flux The heat flux to the tunnel floor was measured with two different devices: heat flux meters (Schmidt-Boelter) [31] and plate thermometers (PT) [32]. The PT consists of a stainless steel plate, 1 mm x 1 mm and.7 mm thick, with a 1 mm thick insulation pad on the backside. A thermocouple is welded to the centre of the plate. The plate thermometers and the heat flux meters were mounted at the centre of the floor at distances -75 mm, +75 mm and +3.75 m. The heat flux using data from the PT can be obtained with aid of the procedure presented by Ingason and Wickström [33]. The heat flux based on the PT has not been analysed in this report.

21 4 Test procedure A total of 42 tests were performed. The tunnel height and tunnel width varied as well as the ventilation rate. The tests included three fuel types; heptane, wood crib and wood crib with plastic ribs. The porosity of the wood cribs was varied. The reason to vary the porosity was to observe the effects of the ventilation rate on the maximum HRR and the fire development. The fuel was in all cases placed on the Promatect H board resting on the stands connected to the scales. The height between the fuel and this board varied between the tests. In the tests with wood cribs, the ignition source was placed under the upstream end (see Figure 4.1). The ignition sources consisted of pieces of fibre board (3 mm 3 mm 24 mm) soaked with 9 ml heptane and wrapped in a piece of polyethene. The procedure was such that the dried Promatect board, fuel (and ignition source), and small targets were put into place before the air velocity in the tunnel was adjusted and registered. All measurements were started two minutes before ignition to register background conditions and to se that everything was running. No test was manually extinguished, but the fuel was let burn until only glowing embers were left or the heptane was totally consumed. During the test series the type of fuel, porosity of the fuel, air velocity, the height of the tunnel, the width of the tunnel, and the height between the floor of the tunnel and the fuel were varied. In Table 4.1 the details on the test series and how the mentioned parameters were varied are presented. Four different air velocities in the tunnel were used:.22 m/s,.45 m/s, 9.67 m/s, and 1.12 m/s. This corresponds to 1 m/s, 2 m/s, 3 m/s, and 5 m/s, respectively, according to scaling laws given in Table 2.1. A fifth ventilation condition, i.e. a case where no forced ventilation was present, was also used. In this case the fan and the mixing box on the upstream side and the exhaust duct on the downstream side were disconnected. In the case with a.6 m wide and.4 m high tunnel, the highest velocity reached was approximately.5 m/s and in these cases the velocities.22 m/s and.5 m/s were used. In some of the tests (see Table 4.1) the fuel was placed on a 13 cm high metal frame too rase the top of the fuel to a height of 2 cm above the floor (see Figure 4.2). Note that in the table it is referred to as the height above the Promatect board, i.e..18 m. Figure 4.1 In the tests with wood cribs, the ignition source was placed under the upstream end. The ignition source consisted of pieces of fibre board (3 mm 3 mm 24 mm) soaked with 9 ml heptane and wrapped in a piece of polyethene.

22 Figure 4.2 A wood crib place on a metal frame to rase the top of the fuel to 2 cm above the floor.

23 Fuel b) Table 4.1 Description of the test conditions for the tests in the test series. Test Mass c) H t (m) B t (m) H d) b (m) u (m/s) P nr a) (kg) 1 Heptane.856.25.45.5.67-2 Wood.832.25.45.5.67 P1 3 Wood.82.25.45.5.22 P1 4 Wood.82.25.45.5.45 P1 5 Wood.86.25.45.5 1.12 P1 6 Wood.96.25.45.5.67 P2 7 Wood.964.25.45.5 1.12 P2 8 Wood 1.4.25.45.5.22 P2 9 Wood.86.25.45.5.67 P1 1 Wood.98.25.45.5.67 P1 +plastics (.256) 11 Wood.882.25.45.5 P1 12 Wood.88.25.6.5.67 P1 13 Wood.816.25.6.5.22 P1 14 Wood.936.25.6.5.67 P1 +plastics (.252) 15 Heptane.852.25.6.5.67-16 Wood.782.25.3.5.67 P1 17 Wood.81.25.3.5.22 P1 18 Wood.982.25.3.5.67 P1 +plastics (.252) 19 Heptane.852.25.3.5.67-2 Wood.862.4.3.5.67 P1 21 Wood.756.4.3.5.22 P1 22 Heptane.868.4.3.5.67-23 Wood.86.4.45.5.67 P1 24 Wood.946.4.45.5.67 P1 +plastics (.252) 25 Wood.82.4.45.5.22 P1 26 Wood.822.4.45.18.67 P1 27 Heptane.872.4.45.5.67-28 Wood.91.4.45.18.67 P1 +plastics (.266) 29 Heptane.876.4.45.18.67-3 Wood.774.4.45.5 P1 31 Wood.844.4.6.5.5 P1 32 Wood.867.4.6.5.22 P1 33 Heptane.876.4.6.5.5-34 Wood.838.4.6.5.5 P1 35 Wood 1.4.4.6.5.5 P1 +plastics (.258) 42 e) Wood.88.4.6.5.5 P1 a) Tests 36 to 41 were free-burn tests under the hood and these tests are presented in Section 5.1.2. b) Heptane means the.155 m.16 m pan, while wood means wood crib. c) In the case of wood crib with plastics the first number corresponds to the total mass, while the number inside parenthesis corresponds to the mass of plastics. d) Measured from the top of the Promatect board under the fuel. An extra.2 m should be added to get the distance between the tunnel floor and the bottom of the fuel. In the case of heptane, the start level of the liquid surface is used as the bottom of the fuel. e) A piece of Promatect board (15 cm 14 cm) was positioned at each end of the wood crib as a blockage.

24 5 Results In this section results from different tests are compared and discussed. Time-resolved graphs for different parameters for all the tests are given in Appendix B. 5.1 HRR from free burn tests Free-burn tests were performed with different fuels used during the test series: the heptane pool and the three types of wood cribs. 5.1.1 Heptane pool Three free-burn tests with the pool (15 mm 16 mm) with heptane used as fire source during the test series. In Figure 5.1 the HRR curves for the three tests are presented. The HRR is calculated from the mass signal assuming a combustion enthalpy of 44.56 MJ/kg and a combustion efficiency of.92. The curves are similar to each other, but vary somewhat in the development in time. The reason for this is probably effects of air currents. These small fires are very sensitive to fluctuations in flame angle. The levels in HRR are approximately the same for the three tests. The average maximum HRR was 5.3 kw. There is no real steadysate, but there is a period of slower increase in the middle of the test. During this period the HRR was between 3 and 4 kw. 8 7 6 Test 44 Test 45 Test 46 Heptane pool HRR [kw] 5 4 3 2 1 Figure 5.1 5 1 15 2 25 Time [min] HRR curves for the three free burn tests with a pool with heptane. 5.1.2 Wood cribs Six free-burn tests were performed, two with each type of wood crib. For each type the position of the ignition source was varied, one test with the ignition source at one end of the wood crib (the position used during the main test series) and one test with the wood crib under the centre of the wood crib. As can be seen below (Figure 5.2), there was a significant difference between these cases for all types of wood cribs. The development of the case with the ignition source beneath the centre of the crib was faster and led to a higher maximum HRR. The maximum HRRs are summarized in Table 5.1. It can there be seen that the maximum HRR in the case of centre ignition is for the wood cribs approximately 5 % higher

25 than the corresponding case with ignition at the end. For the wood/plastic crib the increase is approximately 2 %. For each free-burn test the combustion enthalpy was calculated from the measurements during the time period where the HRR was higher than 2 kw in each test. The results are presented in Table 5.2. Table 5.1 Maximum HRR for the different wood cribs free burn beneath a hood. Item Maximun HRR [kw] HRRmax/HRRmax,end Wood crib, P1, ignition at one end 52.7 1 Wood crib, P1, beneath the centre 77.9 1.48 Wood crib, P2, ignition at one end 56. 1 Wood crib, P2, beneath the centre 85.6 1.53 Wood/plastic crib, P1, ignition at 84.5 1 one end Wood/plastic crib, P1, beneath the centre 1.5 1.19 Table 5.2 Combustion enthalpy for the crib material based on the free burn tests. Item Δh c [MJ/kg] Wood crib, P1, ignition at one end 17. Wood crib, P1, beneath the centre 16.1 Wood crib, P2, ignition at one end 17.8 Wood crib, P2, beneath the centre 17.6 Wood/plastic crib, P1, ignition at one end 23.4 Wood/plastic crib, P1, beneath the centre 22.8 12 Wood crib, P1 12 Wood crib, P2 1 Ignition under one end Ignition under centre 1 Ignition under one end Ignition under centre 8 8 HRR [kw] 6 HRR [kw] 6 4 4 2 2 5 1 15 2 25 Time [min] 5 1 15 2 25 Time [min] 12 1 Wood/plastic crib, P1 Ignition under one end Ignition under centre 8 HRR [kw] 6 4 2 Figure 5.2 5 1 15 2 25 Time [min] Heat release rate of the different cribs free burn beneath hood.

26 In Figure 5.3 the MLR as function of time for the different wood cribs is presented. The results from the two different ignition positions are presented in the two figures. In both cases the fastest increase is for porosity P2. The case with porosity P2 also reach a semi steady state, while the cases with porosity P1 reach either one or two clear peaks. 4 6 Mass loss rate [g/s] 3.5 3 2.5 2 1.5 1.5 Test 36 Test 38 Test 4 Mass loss rate [g/s] 5 4 3 2 1 Test 37 Test 39 Test 41 5 1 15 Time [min] Figure 5.3 5 1 15 Time [min] a) b) Mass loss rate for different wood cribs freely burning beneath hood with the ignition source placed a) under one end and b) under the centre of the wood crib, respectively. 5.2 Effect of geometry on MLR and HRR In addition to the tunnel height and width, the velocity in the tunnel was varied during the tests series. Most tests were performed with forced ventilation to be able to measure the HRR by oxygen calorimetry, but two tests were performed without forced ventilation (i.e., with natural ventilation). These tests provide a direct comparison of the effect of the tunnel on the combustion compared to free-burn conditions. To be able to study the effect of the tunnel dimensions in more detail, results from test with a velocity of.67 m/s (corresponding to a velocity of 3 m/s in real scale) measured 2.75 m upstream of the fire at the centre of the cross section, are also presented and discussed. The tests with natural ventilation were preformed with wood cribs and a tunnel width of.45 m. Two different tunnel heights were used,.25 m and.4 m. For these tests only the MLR can be compared, since the HRR measurements are not available as it was not possible to measure the HRR when fire gases are exhausted from both tunnel portals. The results are summarized in Table 2, where the maximum MLR for the wood crib tests with natural ventilation in the tunnel are compared to the free-burn test. The MLR increased inside the tunnel for both tunnel heights, but the increase was higher in the case with the lower height (a 4 % increase for the height of.25 m compared to a 3 % increase for the height of.4 m). Table 5.3 Mass loss rate of wood crib tunnel tests with natural ventilation compared to free burning test results. Type of test H (m) MLR max (g/s) MLR max /MLR max,fb Free-burn - 3.16 1 Tunnel.25 4.44 1.4 Tunnel.4 4.13 1.3 fb = free-burn.

27 MLRmax/MLRmax,fb 4. 3.5 3. 2.5 2. 1.5 1..5. H=.25 m H=.4m.2.4.6 W (m) HRRmax/HRRmax,fb 4. 3.5 3. 2.5 2. 1.5 1..5. H=.25 m H=.4m.2.4.6 W (m) 4. 4. MLRmax/MLRmax,fb 3.5 3. 2.5 2. 1.5 1..5. W=.3 m W=.45m W=.6m.1.2.3.4.5 H (m) HRRmax/HRRmax,fb 3.5 3. 2.5 2. 1.5 1..5. W=.3 m W=.45m W=.6m.2.4.6 H (m) Figure 5.4 MLR and HRR as function of tunnel width and tunnel height for tests with heptane. The MLR and HRR as functions of height and width of the tunnel are presented in Figure 5.4. For both the width and the height, the MLR decreases with increasing dimension. For the HRR the situation is more complicated. The HRR increases with increasing width for the height H =.25 m, while it decreases with increasing width for the height H =.4 m. In the same way, the HRR increases with the height for the smallest width, W =.3 m, while it decreases with increasing height for the other two widths. The results from the tests with wood cribs (see Figure 6) indicate an even more complex situation than for the pool fires with heptane. Here the general trend is that for the larger cross sections, the MLR and HRR increase with increasing dimension, while for the smallest crosssections the MLR and HRR decreases or shows a relatively small change with increasing dimensions.

28 2.5 2.5 MLRmax/MLRmax,fb 2. 1.5 1..5 H=.25 m H=.4m HRRmax/HRRmax,fb 2. 1.5 1..5 H=.25 m H=.4m..1.2.3.4.5 W (m)..1.2.3.4.5 W (m) 2.5 2.5 MLRmax/MLRmax,fb 2. 1.5 1..5 W=.3 m W=.45m W=.6m HRRmax/HRRmax,fb 2. 1.5 1..5 W=.3 m W=.45m W=.6m..1.2.3.4.5 H (m)..1.2.3.4.5 H (m) Figure 5.5 MLR and HRR as function of tunnel width and tunnel height for tests with wood cribs. A parameter often convenient to use is the hydraulic diameter, especially when establishing how various parameters depend on dimensions. The hydraulic diameter, D H, is defined as 4A D H = (2) P where A is the cross-sectional area and P is the perimeter of the tunnel. In Figure 5.6 and Figure 5.7 the dependencies of the MLR and HRR on D H is presented for heptane and wood cribs, respectively. The dependencies for wood cribs are almost mirror images of those for pool with heptane. The general trend for heptane pool fires is that the MLR decreases with increasing hydraulic diameter. The function for the HRR is more of a negative parabolic shape. For the wood cribs, the MLR increases slightly with increasing hydraulic diameter, while the function for the HRR in this case is somewhat positive parabola. Three of the graphs have a clear notch in the curve. This shows that both MLR and HRR depend individually on the height and the width of the tunnel and that the hydraulic diameter alone cannot be used to explain variations in MLR and HRR.

29 MLRmax/MLRmax,fb 4. 3.5 3. 2.5 2. 1.5 1..5..2.4.6 D H (m) HRRmax/HRRmax,fb 4. 3.5 3. 2.5 2. 1.5 1..5..2.4.6 D H (m) Figure 5.6 MLR and HRR as function of hydraulic diameter for tests with heptane. MLRmax/MLRmax,fb 2.5 2. 1.5 1..5..2.4.6 HRRmax/HRRmax,fb 2.5 2. 1.5 1..5..2.4.6 D H (m) D H (m) Figure 5.7 MLR and HRR as function of hydraulic diameter for tests with wood cribs. The cases presented show that there is a significant difference between the effect on HRR compared to the effect on MLR when placing a specific fire inside of a tunnel instead of having it free burning. In Table 2 these effect are summarized for each case, using MLR ratio /HRR ratio, defined as: MLR HRR ratio ratio MLR max MLR max,fb = (21) HRR HRR max max,fb The results in Table 5.4 show that both the height and width affect the results. The table also emphasizes that the MLR in most cases is affected more than the HRR. In a few cases the HRR is affected to a larger extent. In all these cases the height was.4 m. Table 5.4 MLR ratio /HRR ratio (defined by Eq. [5]) for heptane and wood crib fires in tunnels with different heights and widths. H (m) W (m) D H (m) MLR ratio /HRR ratio Heptane Wood crib.25.3.273 1.31 1.6.25.45.321 1. 1.26.4.3.343.92.96.25.6.353 1.15 1.31.4.45.423.88 1.14

3 5.3 Effect of ventilation on MLR and HRR The effects of the ventilation rate on fire development (fire growth) and maximum HRR is analysed in this section. The relevant HRR results are presented in Figure 5.8. There are two apparent groups of curves in each figure. For both porosities, the HRR for the highest velocities (.45 m/s,.67 m/s, and 1.12 m/s) increase rapidly to a high level. The HRR curves for the cases with lower velocity (.22 m/s) have a different shape: slower increase and a lower, not so sharp peak. The free-burn HRR curve is more similar to the group of lower velocity. The test with natural ventilation is not included since no HRR was measured. Instead the MLR can be used to study how the burning behaviour is affected by the variation in velocity. In Figure 5.9 MLR curves corresponding to different ventilation conditions are presented. One can see that the MLR curves have the same shapes as the HRR curves. However, the difference between the case with.22 m/s and the free burn case is larger in the MLR than in the HRR. The MLR for the case with natural ventilation develops more slowly but reaches a higher maximum value than MLR for the test with the velocity.22 m/s. In Figure 5.1 HRR curves for wood cribs with the porosity P1 and P2 are compared for two different velocities. The tests were performed in a tunnel with the width.45 m and a height of.25 m. Again, the HRR for the cases with a velocity of.22 m/s is significantly lower than the HRR for the higher velocity (.67 m/s). There is, however, a difference between the two porosities. The maximum HRR at.22 m/s is for P2 lower than the corresponding for P1, but at.67 m/s the maximum HRR for P2 is significantly higher than the corresponding for P1. The total area of exposed fuel surface, A s, for P1 is.54 m 2 and.8 m 2 for P2. Therefore, when both P1 and P2 are in a well ventilated flow we should expect differences in the maximum HRR. Thus, there is a larger effect of the ventilation on wood cribs with the porosity P2 then with the porosity P1. This further discussed below. 1 H=.25m; W=.45m; P1 12 H=.25m; W=.45m; P2 HRR [kw] 8 6 4 Free-burning.22 m/s.45 m/s.67 m/s 1.12 m/s HRR [kw] 1 8 6 4 Free-burning.22 m/s.67 m/s 1.12 m/s 2 2 5 1 15 Time [min] 5 1 15 Time [min] a) b) Figure 5.8 Heat release rate curves for the tests in the tunnel with the width.45 m and height.25 m. The fuel was wood cribs with the porosity a) P1 and b) P2, respectively.

31 Mass loss rate [g/s] 6 5 4 3 2 1 H=.25m; W=.45m; P1 Free burning Natural ventilation.22 m/s.45 m/s.67 m/s 1.12 m/s 5 1 15 Time [min] Figure 5.9 Comparison of mass loss rates for different ventilation conditions inside the tunnel with a width of.45 m and a height of.25 m. A wood crib with the porosity 2.1 mm (P1) was used as fuel. According to Ingason [11] for a wood crib similar in shape and size, the maximum HRR was in average for all tests 155 kw per exposed fuel surface area (kw/m 2 ). This means that we could expect a maximum of 84 kw for P1 and 124 kw for P2. These values are very close to the one measured here for P1 and P2 under well ventilated conditions inside the tunnel (see Figure 5.8). The reason why P1 and P2 do not show higher peak HRR in a free-burn and natural ventilation is mainly due to the fact that the fire did not engulf the entire wood crib during the test at one time, i.e. it spreads much slower across the wood crib compared to when the higher ventilation was used. Furthermore, when P2 is burning under ambient conditions or low velocity conditions, the influence of the porosity reduces additionally the maximum value due to lower access of oxygen to the core of the wood crib. When the forced ventilation is activated the relative increase in maximum HRR will be higher for P2 compared to P1. HRR [kw] 12 1 8 6 4 H=.25m; W=.45m P1,.22 m/s P2,.22 m/s P1,.67 m/s P2,.67 m/s 2 5 1 15 Time [min] Figure 5.1 Comparison of HRR for porosity P1 and P2 for two velocities. The tests were performed in a tunnel with the width.45 m and a height of.25 m. So far the time resolved curves have been presented. If a certain time or time period for each test is selected, the differences between the different cases can be quantified. In Figure 5.11 the ratio between the maximum HRR for the test with wood cribs (P1) inside the tunnel and the maximum HRR for the free burning test is presented as function of the air velocity. The test data from Ingason [11], which is approximately P1 crib, is plotted as well for comparison

32 since it was very similar conditions. Also the P2 data is shown in order to show the difference in increase for P1 and P2 wood crib. There is a slight increase in maximum HRR with increasing velocity for a P1 crib. The ratio between the HRRmax inside the tunnel and the free-burn maximum seems to level off at a level somewhat higher the 1.5 for velocities equal to or higher than.67 m/s. Largest spread in the results is it for the velocity.22 m/s where the ratio actually is lower than one for three of the tunnel cross section. All these three tests were performed in the tunnel with the lower height,.25 m. Note that the comparisons here are made to the free-burn case, while many reported comparisons are made to a case with natural ventilation. The agreement of P1crib with the test data from Ingason [11] is good. HRRmax/HRRmax,fb 4. 3.5 3. 2.5 2. 1.5 1..5. P1;W=.45m; H=.25m P1;W=.6m; H=.25m P1;W=.3m; H=.25m P1;W=.3m; H=.4m P1;W=.45m; H=.4m P2;W=.45m;H=.25m Ingason [11] W=.4m;H=.3m Ingason [11],W=.4m;H=.2 m.2.4.6.8 1 1.2 u (m/s) Figure 5.11 Maximum HRR as function of velocity for different tunnel cross sections, compared to the free burn case. As can be seen in Figure 5.11, the relative increase of HRR for P2 is higher than for P1 at corresponding conditions. For the low velocity (.22 m/s) the ratio was lower than one or.8 and for the higher velocities it was 1.8 (.67 m/s) and 2. (1.12 m/s), respectively. This shows that it is important to have in mind the porosity of the fuel when comparing the ambient fuel set-up with a fuel in a forced ventilation flow. 5.4 Effect of geometry on gas temperature The combustion zone, position of the flame, and the temperature distribution is affected by the dimensions of the tunnel. This means that the effect of a change of one length parameter can be different at different distances from the seat of fire. In Figure 5.12 to Figure 5.16 the maximum temperatures near the ceiling at different distances from the seat of the fire are presented. Each figure contain graphs for two different fuels (heptane and wood cribs), presented as variations in tunnel width, tunnel height, and hydraulic diameter, respectively. Note that in each graph, the point representing the largest cross section (W =.6 m and H =.4 m) corresponds to an air velocity of.5 m/s instead of.67 m/s.

33 Tmax [ o C] Tmax, cm, Heptane 7 6 5 4 3 2 H=.25 m 1 H=.4 m.2.4.6.8 WT (m) Tmax [ o C] 12 1 8 6 4 2 Tmax, cm, Wood crib H=.25 m H=.4m.2.4.6.8 W T (m) Tmax [ o C] 7 6 5 4 Tmax, cm, Heptane 3 W=.3 m 2 W=.45 m 1 W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] 12 1 8 6 4 2 Tmax, cm, Wood crib B=.3 m B=.45m B=.6m.1.2.3.4.5 H T (m) Tmax [ o C] Tmax, cm, Heptane 7 6 5 4 3 2 Tmax y = 1162.6x + 64.174 1 Linear (Tmax) R 2 =.698.2.4.6 D H (m) Tmax [ o C] Tmax, cm, Wood crib 12 1 8 6 4 Tmax 2 Linear (Tmax) y = -572.74x + 111.4 R 2 =.265.2.4.6 D H (m) Figure 5.12 Maximum temperatures near the ceiling above the centre of the fuel. The maximum temperature above the centre of the fire increases with the width of the tunnel for both heptane and wood crib fires (see Figure 5.12). There is a significant difference in the maximum temperature between the different tunnel widths. Even if the absolute temperatures are lower in the heptane case, the increase with the tunnel width is almost as distinct as in the wood crib case. One significant difference between the two fuels is that in the heptane case the temperatures are higher for the largest tunnel height, at least for the smallest tunnel width, while in the wood crib case the lowest tunnel has the highest maximum temperatures above the centre of the fire. This is probably related to the geometry of the fuel and how the flame zone is affected by the ventilation. In the case with the liquid fuel heptane, an increased ceiling height probably moves the flame volume towards the seat of the fire due to increased mixing. In the case with the wood crib as a fuel, the combustion zone is already near the seat of the fire due to the three-dimensional burning inside the wood crib. This can be seen in the higher temperatures above the seat of the fire in the wood crib case. An increased ceiling height does not seem to improve the combustion efficiency. Instead the distance to the hot surface is increased and the increased space for air flow probably cools the gases above the fire. This difference in influence of the tunnel height can also be seen in the middle graphs where the maximum temperatures are plotted against the tunnel height. From these graphs it is apparent that the temperature increases with increasing height for the heptane case while it is the other way around for the wood crib tests. The use of hydraulic diameter, D H, can be seen as an efficient way of combining the effects of both height and width. In the case with heptane where the effect of variation in height and

34 width are similar, there is a clear relationship between the change in maximum temperature and variation of the hydraulic diameter. The outlier with the relatively high temperature in the graph for heptane can most probably be explained by two different processes that in this case influence the temperature in the same direction. The point corresponds to the case with W =.6 m and H =.3 m. The first effect is due to the large width is an increase in the mixing thereby moving the combustion zone closer to the seat of the fire and increasing the maximum temperature in this region. The other effect is due to the low tunnel height giving a radiation effect increasing the burning rate. This effect can also be enhanced by the wider tunnel moving the flame volume closer to the seat of the fire. Even if the walls (and possible radiation from them) are further away from the fuel. Tmax [ o C] 12 1 8 6 4 2 Tmax, 25 cm, Heptane H=.25 m H=.4 m.2.4.6.8 W T (m) Tmax [ o C] Tmax, 25 cm, Wood crib 12 1 8 6 4 H=.25 m 2 H=.4 m.2.4.6.8 W T (m) Tmax [ o C] 12 1 8 6 Tmax, 25 cm, Heptane 4 W=.3 m 2 W=.45 m W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] 12 1 8 6 4 2 Tmax, 25 cm, Wood crib W=.3 m W=.45 m W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] Tmax, 25 cm, Heptane 12 1 8 6 4 Tmax 2 Linear (Tmax) y = -277.5x + 868.78 R 2 =.253.1.2.3.4.5.6 D H (m) Tmax [ o C] Tmax, 25 cm, Wood crib 12 1 8 6 4 Tmax y = -46.64x + 1143.5 2 Linear (Tmax) R 2 =.462.2.4.6 D H (m) Figure 5.13 Maximum temperatures near the ceiling 25 cm downstream the seat of the fire. The measurement at 25 cm downstream of the seat of the fire (see Figure 5.13) is situated downstream of the heptane pool and above the downstream edge of the wood crib. These temperature measurements represent an intermediate region. There is no clear relationship between the temperature and the width of the tunnel. Only between the width.3 m and.45 m for the height H =.4 m is there a significant change. For heptane, the trend is an increasing temperature for increasing tunnel width. The variation in maximum temperature with tunnel height is not extremely large, but still clear and the temperature decreases with increasing height. The trends are the same for the two types of fuels. There is no clear function of the hydraulic diameter for any of the two fuels.

35 Tmax [ o C] Tmax, 5 cm, Heptane 12 1 8 6 4 H=.25 m 2 H=.4 m.2.4.6.8 W T (m) Tmax [ o C] Tmax, 5 cm, Wood crib 12 1 8 6 4 H=.25 m 2 H=.4 m.2.4.6.8 W T (m) Tmax [ o C] 12 1 8 6 4 2 Tmax, 5 cm, Heptane W=.3 m W=.45 m W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] 12 1 8 6 4 2 Tmax, 5 cm, Wood crib W=.3 m W=.45 m W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] Tmax, 5 cm, Heptane 12 1 y = -3477.6x + 2116 R 2 =.729 8 6 4 Tmax 2 Linear (Tmax).1.2.3.4.5.6 Tmax [ o C] Tmax, 5 cm, Wood crib 12 1 y = -3539.8x + 1994.5 R 2 =.7353 8 6 4 Tmax 2 Linaer (Tmax).1.2.3.4.5.6 D H (m) D H (m) Figure 5.14 Maximum temperatures near the ceiling 5 cm downstream the seat of the fire. In Figure 5.14, the maximum temperatures near the ceiling 5 cm downstream of the seat of the fire are shown. The maximum temperature decreases with tunnel width for both heptane and the wood crib, especially between W =.45 and W =.6 m for the case with H =.4 m. The same trend with decreasing temperatures can be seen as a function of the tunnel height, especially for W =.6 m. Even if the not all graphs in the upper four figures show significant trends, there is a clear function between the maximum temperature and variation in hydraulic diameter. For both fuels, heptane and wood cribs, the case with W =.45 m and H =.4 m seems to obtain combustion conditions leading to relatively high temperatures in the position 5 cm downstream of the fire. The points corresponding to this case do not fall on the regression line for the function of D H, especially for wood cribs. In Figure 5.15, the maximum temperatures near the ceiling 75 cm downstream of the seat of the fire are presented. The effect of a change in the tunnel width is much more pronounced in the case with wood crib as a fire source than for the case with heptane, where the change was negligible. Almost the same situation can be seen for a change in the tunnel height, although the effect in the wood crib case is not as clear for a change in tunnel height as for a change in tunnel width. In total there is a negligible effect of a change in the hydraulic diameter for the case with heptane, while there, on the other hand, is a clear decrease in maximum temperature with increasing hydraulic diameter for the wood crib case.

36 Tmax [ o C] Tmax, 75 cm, Heptane 12 1 8 6 4 H=.25 m 2 H=.4 m.2.4.6.8 W T (m) Tmax [ o C] Tmax, 75 cm, Wood crib 12 1 8 6 4 H=.25 m 2 H=.4 m.2.4.6.8 W T (m) Tmax [ o C] 12 1 8 6 4 2 Tmax, 75 cm, Heptane W=.3 m W=.45 m W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] 12 1 8 6 Tmax, 75 cm, Wood crib 4 W=.3 m 2 W=.45 m W=.6 m.1.2.3.4.5 H T (m) Tmax, 75 cm, Heptane Tmax, 75 cm, Wood crib 12 12 1 1 Tmax [ o C] 8 6 4 Tmax y = -223.72x + 154.9 2 Linear (Tmax) R 2 =.1286.2.4.6 D H (m) Tmax [ o C] 8 6 4 Tmax y = -997.75x + 151.5 2 Linear (Tmax) R 2 =.6351.2.4.6 D H (m) Figure 5.15 Maximum temperatures near the ceiling 75 cm downstream the seat of the fire. In Figure 5.16, the maximum temperatures near the ceiling 1 cm downstream of the seat of the fire are presented. The temperature decreases with increasing tunnel width for both heptane and wood crib. The lines for heptane are almost identical for the two heights. This can also be seen in the middle graphs where there is no clear function between the temperature and the tunnel height. For the wood crib case, on the other hand the function between temperature and tunnel height is relatively clear. This difference between heptane and wood crib is also clearly seen in the lower graphs where the functions of the temperatures are decreasing for increasing hydraulic diameter for both heptane and wood cribs, but while the function in the heptane case is only slowly decreasing and is based on somewhat scattered data, the wood crib case shows a clearly decreasing function and an almost perfectly linear correlation.

37 Tmax [ o C] Tmax, 1 cm, Heptane 12 1 8 6 4 H=.25 m 2 H=.4 m.2.4.6.8 W T (m) Tmax [ o C] Tmax, 1 cm, Wood crib 12 1 8 6 4 H=.25 m 2 H=.4 m.2.4.6.8 W T (m) Tmax [ o C] 12 1 8 6 4 2 Tmax, 1 cm, Heptane W=.3 m W=.45 m W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] 12 1 8 6 Tmax, 1 cm, Wood crib 4 W=.3 m W=.45 m 2 W=.6 m.1.2.3.4.5 H T (m) Tmax [ o C] 12 1 8 6 Tmax, 1 cm, Heptane 4 Tmax y = -541.76x + 18.8 2 Linear (Tmax) R 2 =.4351.1.2.3.4.5.6 D H (m) Tmax [ o C] 12 1 8 6 4 2 Tmax Tmax, 1 cm, Wood crib y = -1157.2x + 141.1 Linear (Tmax) R 2 =.9725.1.2.3.4.5.6 Figure 5.16 Maximum temperatures near the ceiling 1 cm downstream the seat of the fire. DH (m) 5.5 Temperature distribution along the tunnel The temperature distribution along the tunnel is an important factor for many reasons, e.g. for the risk for fire spread, for occupants inside the tunnel, for the rescue service, etc. In Section 2 a way of calculating the average temperature along the tunnel was presented. Equation (14) has been used to calculate the average temperatures and in Figure 5.17 this calculated temperatures are compared to the temperature measurements 3 mm below the ceiling, the measurements 125 mm from the ceiling (half the tunnel height) and an average temperature calculated as an arithmetic average of the five thermocouples in the thermocouple tree at the centre line of the tunnel. An effective heat transfer coefficient of h =.2 kw/m 2 /K has been used. In Figure 5.17 it can be seen that the tests with heptane have their maximum temperature further downstream (5 cm 75 cm) than what is the case for the wood crib tests (25 cm downstream of the seat of the fire). The temperature measured at half the tunnel height and the average temperature of a vertical thermocouple tree are in most cases relatively similar to each other. The largest difference can be seen for the heptane test in the largest tunnel where the average over height at the cross section 75 cm from the seat of the fire is more than 1 C lower than the half height temperature. The temperatures measured with the lowest two thermocouples are in this case much lower than for the other thermocouples, i.e. there was a clear stratification. The calculations of average temperature based on Eq. (14) correlate relatively well with the temperature measured at half the tunnel height, at least when excluding the measurements at

38 positions close to maximum temperature. For the tunnel with the smallest cross section, Equation (14) gives a temperature of 141 C (not shown in the graph) above the seat of the fire, while the measured temperature is significantly lower. Heptane, W=.3 m, H=.25 m Wood crib, W=.3 m, H=.25 m 12 1 8 Tavg HRRconv Tmeas, 3mm Tmeas, 125mm Tmeas, avg 12 1 8 Tavg HRRconv Tmeas, 3mm Tmeas, 125mm Tmeas, avg T ( o C) 6 T ( o C) 6 4 4 2 2 1 2 3 4 5 x (m) 1 2 3 4 5 x (m) Heptane, W=.45 m, H=.25 m Wood crib, W=.45, H=.25 m 12 1 8 Tavg HRRconv Tmeas, 3mm Tmeas, 125mm Tmeas, avg 12 1 8 Tavg HRRconv Tmeas, 3mm Tmeas, 125mm Tmeas, avg T ( o C) 6 T ( o C) 6 4 4 2 2 1 2 3 4 5 x (m) 1 2 3 4 5 x (m) Heptane, W=.45 m, H=.4 m Wood crib, W=.45 m, H=.4 m 12 1 8 Tavg HRRconv Tmeas, 3mm Tmeas, 2mm Tmeas, avg 12 1 8 Tavg HRRconv Tmeas, 3mm Tmeas, 2mm Tmeas, avg T ( o C) 6 T ( o C) 6 4 4 2 2 1 2 3 4 5 x (m) 1 2 3 4 5 x (m) Figure 5.17 Temperatures measured near the ceiling along the tunnel. The position x= represents a position above the centre of the fire. Included are also temperatures measured at a height equal to half the tunnel height and averages of the tunnel height, respectively. The line included in the graphs is based on Equation (14). For heptane, it is apparent that the average temperature and the temperature measured in the ceiling are more similar to each other, the smaller the tunnel cross section. For a tunnel with a larger height a clear interface between upper and lower gas layer can be seen. This can be seen also in the case with wood crib, but since the maximum temperatures can be found closer to the seat of the fire, this stratification could be registered also in the middle graph (low but wide tunnel). 5.6 Effect of blockage on HRR In one of the tests (Test 42) a blockage made of Promatect H board was positioned at each end of the wood crib, see Figure 5.18. The air velocity was in this case approximately.5 m/s and the aim of the test was to study the effect of the blockages.

39 Figure 5.18 Test 42: a) Set-up with blockages at the two ends, b) Photo from the test. The HRR curves in Figure 5.19 shows a clear effect of the blockages on the HRR. Especially the development is affected. The fire was developed much slower in the case with blockages. The increase in HRR was approximately twice as fast for Test 31 (6 kw/min) and Test 34 (63 kw/min) compared to Test 42 (31 kw/min). Also the maximum HRR for Test 42 was lower than that for the two tests without blockage. Test 31 and Test 34 were performed under the same conditions and the HRR curves in Figure 5.19 show a very good repeatability. 1 8 Test 31 Test 34 Test 42 HRR [kw] 6 4 2 5 1 15 Time [min] Figure 5.19 Comparison of HRR in wood crib tests with and without blockage in front of the wood crib.

4 6 Discussion 6.1 Effect of geometry on MLR and HRR The main explanation for the effect of the tunnel and tunnel dimensions on the HRR from a pool fire is the re-radiation from surfaces and hot gases to the fuel. This factor can, however, not alone explain the results presented here. For the larger heights and widths of the tunnel, the HRR decreases with increasing tunnel dimensions. However, for the smallest tunnel dimensions in the test series the HRR increases with both the height and the width of the tunnel. The explanation for this can probably be found in the flow field and combustion zone. For the smallest dimensions, complete combustion is probably not easily achieved and therefore increased dimensions leads to a more complete combustion and higher HRR. Both for the lowest height and the smallest width of the tunnel, higher CO concentrations were measured than in corresponding cases with larger dimensions. The combustion situation is even more complicated in the case of wood cribs due to the threedimensional nature of crib burning. A change in tunnel dimension will affect the air flow around and through the wood crib. This will affect the temperature and combustion inside the wood crib. It will also affect the shape and position of the flame. This could also be seen in the evaluation of the maximum temperatures along the tunnel, see section 5.4, where the dependencies on width and height were significantly different between positions near the seat of the fire and positions further downstream. For the heptane pool fires the factor of increase from free burning conditions to tunnel conditions varied between 2.5 and 3.. For the wood crib fires (with a centre air velocity of.67 m/s) this factor varies between 1.4 and 1.7. For heptane the variation in MLR is even larger, between 2.4 and 3.5. All these results are taken from tests with an air velocity of.67 m/s. This was chosen since most data was available for this velocity. However, it is of special interest to see what influence the tunnel has on the result for the situation with natural ventilation. MLR data is available for two such tests with wood cribs. If it is assumed that the quotient between MLR and HRR is approximately the same for the same type of fuel and the same tunnel dimensions, maximum HRR can be estimated for these two tests. The factors of increase from free-burn conditions to tunnel conditions would then be between 1.12 and 1.15 for the case with natural ventilation. Furthermore, this indicates an increase in maximum HRR of 1.3 or 1.5 for wood cribs when the ventilation conditions are changed from natural ventilation to a situation of forced ventilation with a velocity of.67 m/s (3 m/s in real scale). That the ventilation in a tunnel has an important impact on the HRR has been shown and discussed previously in the literature [[6, 7, 34, 35]]. The air velocity was also varied in the presented test series and the effect of the ventilation should be kept in mind (see also Section 6.2). The values of the factors of increase from free burning conditions to tunnel conditions reported here are in the range of what has been presented in the literature for other types of fires. It is, however, of relatively small value to compare individual values from different test series where many parameters can differ between the test series. Detailed knowledge on the conditions is needed to be able to perform a complete comparison. It is of more interest to determine how variation in different parameters affects the results. Using our understanding of this, results from different experimental series could be normalised to reduce the influence of any given parameter and foster comparison between various tests.

41 6.2 Effect of ventilation on MLR and HRR The results show that there is an effect of the ventilation both on the maximum HRR and on the fire growth rate. The results on maximum HRR and on fire growth rate of P1 correspond very well to the data presented by Ingason [11]. Further, the results show that the effect of the velocity on the increase factor depends on the tunnel dimensions and the conditions inside the tunnel. The height and the width of the tunnel influence the results. The effect of initial conditions of the fuel and the tunnel flow is also illustrated by the HRR ratio between the lowest velocity and the cases with higher velocity. For the case with the lowest velocity (.22 m/s), the fire develops more slowly and reaches a lower maximum value compared to the cases with higher velocities. Almost all cases with the velocity.22 m/s (corresponding to a velocity of 1 m/s in real scale) became somewhat under-ventilated at the time of the highest MLR. This is important to keep in mind since this can also be the case for real tunnels. This can significantly affect the effect of the velocity on the increase factors. Also the type of fuel, here represented by the porosity, is important. The wood crib with the lower porosity (P2) was more affected by the velocity than was the wood crib with the higher velocity (P1). It is also interesting to note that the increase factor, both for the maximum HRR and for the fire growth rate, seems to reach a constant value or at least increase more slowly when reaching above a certain velocity. 6.3 Effect of geometry on gas temperature In general it is observed that the maximum gas temperatures are affected by the variation of the width and the height, but the result of the variation differ depending on at which position in the tunnel the temperature is measured. In most positions the temperature decreases with increasing dimension, but in several of these positions the change in temperature is relatively small. As an exception it should be noted that close to the fire the gas temperature increases with increasing dimension, especially for heptane. In general, the trends are similar for heptane and wood cribs, even if the magnitude of the effect on the temperature varies. Again, one exception is the position near the fire, where the influence of the height on the temperature above the seat of the fire is different for heptane and wood crib fires. One reason for this might be the difference in burning, combustions zone and flame extension for the two types of fuels. Heptane burns with a flame volume arising from the fuel surface. The ventilation flow makes the flame tilt and a change in height or width alters the combustion zone and thereby the position of maximum temperature. The wood cribs burn threedimensionally, e.g. inside the wood crib. This means that they are affected differently by the ventilation where the flow to some extent passes through the burning wood crib. A change in the tunnel dimensions can alter the air flow and thereby the combustion process. The geometrical effects of the tunnel on the maximum temperatures is more pronounced further downstream in the tunnel where the temperature is more two dimensional in its character. On the other hand closer to the seat of the fire, the combustion behaviour and flame volume are very much affected by the dimensions of the cross-section. For example, when the width is increased the oxygen mixes better with the pyrolysis gases from the fuel, and therefore the combustion zone moves closer to the seat of the fire. The consequence is that the maximum temperature is reached closer to the seat of the fire. The geometry of the fuel clearly plays an important role in this process.

42 The maximum ceiling temperatures does not always need to be at the centreline of the tunnel. It was observed in some tests that the maximum occurred in between the centreline and the wall. For example in Test 23 the temperature was 64 C at the centre line, but 68 C on both the left side and the right side (all measured 3 mm below the ceiling). Furthermore, the temperatures can be affected by the ventilation as well. A change in ventilation can affect the temperatures in several different ways, or for several different reasons, see section 6.2. It was shown earlier that an increase in the air velocity can increase the HRR. The magnitude of the change depends on the starting conditions [36]. An increase of the air velocity will also increase the total air flow and this can lead to a cooling of the gases inside the tunnel. Another effect an increase of the air velocity can have is that it can change the flame angle and combustion zone and thereby locally significantly alter the temperature and resulting in a large effect on individual temperature measurements. These issues are important to take into account when applying the results for different cases. Note that in each graph showing the effect of the width and the height in section 5.4, the point representing the largest cross section (W =.6 m and H =.4 m) corresponds to an air velocity of.5 m/s instead of.67 m/s. This might have had some effect even if it has not been quantified. In most cases, however, this point seems to follow the general trends. It is only at the position 5 cm where there is a significant difference between the large cross section and the other tests. The different velocity might have some effect, but some of the difference might also be due to the larger cross-section itself. Note that the theory on the temperature distribution along the length of the tunnel, shown in section 5.5, was based on a constant HRR. In the comparisons in this paper, the maximum HRR and the temperatures recorded at the time for maximum HRR have been used. In some cases the increase in HRR has been fast and the peak had been sharp leading to only a short time period of high HRR. That will affect the correlation between the theoretical and experimental results.

43 7 Conclusions This report focuses on the effect of the tunnel dimensions, i.e. the height and the width of the tunnel together with the effects of ventilation rate on Mass Loss Rate (MLR) of the fuel, the Heat Release Rate (HRR) and gas temperatures at different locations in a model tunnel. The main conclusions from the tests are: The dependency of the MLR of the fuel and the HRR on the tunnel dimensions are different from each other, especially for pool fires. The effect of the height and the width of the tunnel on the MLR and HRR depends on the starting conditions. For a pool fire in a tunnel with relatively small height or width, an increase in tunnel dimensions leads to an increase in fire size while for larger height and widths, an increase in tunnel dimensions instead leads to a decrease in fire size. The situation for wood crib fires is more complex and relations opposite to those found for heptane fires can be found for some conditions. The MLR and the HRR depend on both height and width. The hydraulic diameter alone does not fully explain the variations. The position of the ignitions source affects the fire development and the maximum HRR. It was found that an increasing ventilation rate increases the maximum HRR. The increase in the maximum HRR of the P1 crib is in the range of 1.3 to 1.7 times the value measured outside the tunnel under ambient conditions and for wind velocities equal to or higher than.45 m/s. When comparing the MLR for a case with forced ventilation to a case with natural ventilation conditions inside the tunnel this increase is lower. For all cases when the velocity was.22 m/s and the low ceiling height was used, the ratio was found to be lower than one. This was not the case when high ceiling height was used. For the case with a velocity of.67 m/s, the fire growth rate increased by a factor of 5 1 times the free burn case, depending on the dimensions of the tunnel crosssection. The increase factor of the maximum HRR (compared to the free burn case) of P2 was 1.8 (.67 m/s) and 2. (1.12 m/s), respectively, for the higher velocities. For the wood crib P1, this increase factor was lower. Above a certain air velocity, the ratio for HRR does not seem to vary significantly with the velocity. In general we can say that this ratio is highly dependent on the starting conditions of the fuel, i.e. its porosity, the velocity, the ceiling height etc. In particular, if the wood crib is vitiated under ambient conditions, we can expect significant changes in the maximum HRR when the ventilation conditions are changed. The influence of velocity on the fire growth rate is much higher than on the maximum HRR. As the fire growth rate is one of the most important design parameters for tunnel safety these results are considered as important. The gas temperature is affected by the dimensions of the tunnel. In general there is a decrease in temperature with increasing dimension, but not near the fire where an increase in gas temperature can be seen.

44 The effects on tests with the two different fuels are similar, but not identical. Near the seat of the fire the burning characteristics for the two fuels are different. In this region the three-dimensional effects in the wood crib case and the effect of the flame position for the heptane case become significant. For the tunnels with a small cross-section, the temperature is relatively uniform, while an increased stratification and difference between average temperature and ceiling temperature can be observed when the width or height of the tunnel is increased. The model of the average temperature correlates relatively well with the measured temperature at half the tunnel height and the average value calculated from temperatures measured in a vertical thermocouple tree, at least a distance away from the fire. Close to the fire the difference can be large. The cases where the conditions have been the same for different tests show that the there is good repeatability in the test results. In general it can be concluded that several factors and processes are interacting and it is important to know the initial boundary conditions to be able to predict the effect of a change in any specific dimensional parameter.

45 8 References 1. Drysdale, D., An Introduction to Fire Dynamics, John Wiley & Sons, 1992. 2. Casale, E., and Marlair, G., "Heptane Fire Tests with Forced Ventialtion", Proceedings of the International Conference on Fires in Tunnels, 36-5, Borås, Sveden, 1-11 October, 1994. 3. Takeda, H., and Akita, K., "Critical Phenomenon in Compartment Fires with Liquid Fuels", Eighteenth Symposium (International) on Combustion, 519-527, Waterloo, Canada, 17-22 August, 198. 4. Carvel, R. O., Beard, A. N., and Jowitt, P. W., "How Much do Tunnels Enhance the Heat Release Rate of Fires?" Proc. 4th Int. Conf on Safety in Road and Rail Tunnels, 457-466, Madrid, Spain, 2-6 April, 21. 5. Carvel, R. O., Beard, A. N., Jowitt, P. W., and Drysdale, D. D., "Variation of Heat Release Rate with Forced Longitudinal Ventilation for Vehicle Fires in Tunnels", Fire Safety Journal, 36, 6, 569-596, 21. 6. Carvel, R. O., Beard, A. N., and Jowitt, P. W., "A Bayesian Estimation of the Effect of Forced Ventilation on a Pool Fire in a Tunnel", Civil. Eng. and Env. Syst., 18, 279-32, 21. 7. Carvel, R. O., Beard, A. N., and Jowitt, P. W., "The Influence of Longitudinal Ventilation and Tunnel Size on HGV Fires in Tunnel", 1th International Fire Science & Engineering Conference (Interflam 24), 815-82, Edinburgh, Scotland, 5-7 July, 24. 8. Ingason, H., "Fire Experiments in a Model Tunnel using Pool Fires - Experimental Data", SP Swedish National Testing and Research Institute, SP AR 1995:52, Borås, Sweden, 1995. 9. Ingason, H., "Effects of Ventilation on Heat Release Rate of Pool Fires in a Model Tunnel", SP Swedish National Testing and Research Institute, SP REPORT 1995:55, Borås, Sweden, 1995. 1. Saito, N., Yamada, T., Sekizawa, A., Yanai, E., Watanabe, Y., and Miyazaki, S., "Experimental Study on Fire Behavior in a Wind Tunnel with a Reduced Scale Model", Second International Conference on Safety in Road and Rail Tunnels, 33-31, Granada, Spain, 3-6 April, 1995. 11. Ingason, H., "Model Scale Tunnel Fire Tests - Longitudinal ventilation", SP Swedish National Testing and Research Institute, SP REPORT 25:49, Borås, Sweden, 25. 12. Carvel, R. O., Beard, A. N., and Jowitt, P. W., "The Influence of Longitudinal Ventilation Systems on Fires in Tunnels", Tunnelling and Underground Space Technology, 16, 3-21, 21. 13. Heskestad, G., "Modeling of Enclosure Fires", Proceedings of the Fourteenth Symposium (International) on Combustion, 121-13, The Pennsylvania State University, USA, August, 1972. 14. Quintiere, J. G., "Scaling Applications in Fire Research", Fire Safety Journal, 15, 3-29, 1989. 15. Heskestad, G., "Physical Modeling of Fire", Journal of Fire & Flammability, Vol. 6, p. 253-273, 1975. 16. Ingason, H., Wickström, U., and van Hees, P., "The Gothenburg Discoteque Fire Investigation", 9th International Fire Science & Engineering Conference (Interflam 21), 965-976, Edinburg, Scotland, 17-19 September, 21. 17. Ingason, H., "Model Scale Railcar Fire Tests", SP Swedish National Testing and Research Institute, SP REPORT 25:48, Borås, 25. 18. Lönnermark, A., and Ingason, H., "Fire Spread in Large Industrial Premises and Warehouses", SP Swedish National Testing and Research Institute, SP Report 25:21, Borås, Sweden, 25.

46 19. Huggett, C., "Estimation of Rate of Heat Release by Means of Oxygen Consumption Measurements", Fire and Materials, 4, 2, 61-65, 198. 2. Parker, W. J., "Calculations of the Heat Release Rate by Oxygen Consumption for Varios Applications", National Bureau of Standards, NBSIR 81-2427, Gaithersburg, USA, 1982. 21. McCaffrey, B. J., and Heskestad, G., "Brief Communications: A Robust Bidirectional Low-Velocity Probe for Flame and Fire Application", Combustion and Flame, 26, 125-127, 1976. 22. Gross, D., "Experiments on the Burning of Cross Piles of Wood", Journal of Research of the National Bureau of Standards-C. Engineering and Instrumentation, Vol. 66C, 2, 99-15, 1962. 23. Croce, P. A., and Xin, Y., "Scale modeling of quasi-steady wood crib fires in enclosures", Fire Safety Journal, Vol. 4, 245-266, 25. 24. Heitor, M. V., and Moreira, A. L. N., "Thermocouples and Sample Probes for Combustion Studies", Progress in Energy and Combustion Science, 19, 259-278, 1993. 25. Luo, M., "Effects of Radiation on Temperature Measurement in a Fire Environment", Journal of Fire Sciences, 15, November/December, 443-461, 1997. 26. Blevins, L. G., and Pitts, W. M., "Modeling of bare and aspirated thermocouples in compartment fires", Fire Safety Journal, 33, 239-259, 1999. 27. Brohez, S., Delvosalle, C., and Marlair, G., "A two-thermocouples probe for radiation corrections of measured temperatures in compartment fires", Fire Safety Journal, 39, 399-411, 24. 28. Pitts, W. M., Braun, E., Peacock, R. D., Mitler, H. E., Johnsson, E. L., Reneke, P. A., and Blevins, L. G., "Temperature Uncertainties for Bare-bead and Aspirated Thermocouple Measurements in Fire Environments", Annual Conference on Fire Research, 15-16, Gaithersburg, Mariland, USA, 2-5 November, 1998. 29. Lönnermark, A., and Ingason, H., "Gas Temperatures in Heavy Goods Vehicle Fires in Tunnels", Fire Safety Journal, 4, 56-527, 25. 3. Lönnermark, A., Hedekvist, P. O., and Ingason, H., "Gas Temperature Measurements Using Fibre Bragg Grating during Fire Experiments in a Tunnel", Fire Safety Journal, 27. 31. Persson, B., and Wetterlund, I., "Tentative Guidelines for Calibration and Use of Heat Flux Meters", SP Swedish National Testing and Research Institute, SP Report 1997:33, 1997. 32. Ingason, H., and Wickström, U., "Measuring incident radiant heat flux using the plate thermometer ", Fire Safety Journal, Vol. 42, 2, 161-166, 27. 33. Ingason, H., and Wickström, U., "The international FORUM of fire research directors: A position paper on future actions for improving road tunnel fire safety", Fire Safety Journal, 41, 111-114, 26. 34. Huijben, J. W., "The Influence of Longitudinal Ventilation on Fire Size and Development", Fourth International Conference on Tunnel Fires, 115-124, Basel Switzerland, 2-4 December, 22. 35. Lönnermark, A., "On the Characteristics of Fires in Tunnels", Doctoral Thesis, Department of Fire Safety Engineering, Lund University, Lund, Sweden, 25. 36. Lönnermark, A., and Ingason, H., "The Influence of Tunnel Dimensions on Fire Size", Proceedings of the 11th International Fire Science & Engineering Conference (Interflam 27), 1327-1338, London, UK, 3-5 September, 27.

47

48 Appendix A Type of instrument Distance a) (cm) Measurement positions Height from Ceiling (mm) Placement Sideways b) Comments TC.25 mm -375 3 mm C TC.25 mm -35 3 mm C TC.25 mm -312.5 3 mm C TC.25 mm -275 3 mm C TC.25 mm -275 125 mm C TC.25 mm -25 3 mm C TC1.8 mm -25 3 mm C TC.25 mm -225 3 mm C TC.25 mm -225 78 mm C TC.25 mm -225 125 mm C TC.25 mm -225 172 mm C TC.25 mm -225 22 mm C TC.25 mm -225 3 mm C H T =.4 m TC.25 mm -225 35 mm C H T =.4 m TC.25 mm -187.5 3 mm C TC.25 mm -15 3 mm C TC.25 mm -125 3 mm C TC1.8 mm -125 3 mm C TC.25 mm -1 3 mm C TC.25 mm -75 3 mm C TC.25 mm -75 78 mm C TC.25 mm -75 125 mm C TC.25 mm -75 172 mm C TC.25 mm -75 22 mm C TC.25 mm -75 3 mm C H T =.4 m TC.25 mm -75 35 mm C H T =.4 m TC.25 mm -75 3 mm L TC.25 mm -75 78 mm L TC.25 mm -75 125 mm L TC.25 mm -75 172 mm L TC.25 mm -75 22 mm L TC.25 mm -75 3 mm R TC.25 mm -75 78 mm R TC.25 mm -75 125 mm R TC.25 mm -75 172 mm R TC.25 mm -75 22 mm R TC1.8 mm -75 3 mm C TC1.8 mm -75 78 mm C TC1.8 mm -75 125 mm C TC1.8 mm -75 172 mm C TC1.8 mm -75 22 mm C TC1.8 mm -75 3 mm C H T =.4 m TC1.8 mm -75 35 mm C H T =.4 m TC.25 mm -5 3 mm C TC1.8 mm -5 3 mm C TC.25 mm -25 3 mm C TC.25 mm 3 mm C TC.25 mm 25 3 mm C TC.25 mm 5 3 mm C TC1.8 mm 5 3 mm C TCS 5 3 mm C TC.25 mm 75 3 mm C

49 Type of instrument Distance (mm) Height from Ceiling (mm) Placement sideways Comments TC.25 mm 75 78 mm C TC.25 mm 75 125 mm C TC.25 mm 75 172 mm C TC.25 mm 75 22 mm C TC.25 mm 75 3 mm C TC.25 mm 75 35 mm C TC.25 mm 75 3 mm L TC.25 mm 75 78 mm L TC.25 mm 75 125 mm L TC.25 mm 75 172 mm L TC.25 mm 75 22 mm L TC.25 mm 75 3 mm R TC.25 mm 75 78 mm R TC.25 mm 75 125 mm R TC.25 mm 75 172 mm R TC.25 mm 75 22 mm R TC1.8 mm 75 3 mm C TC1.8 mm 75 78 mm C TC1.8 mm 75 125 mm C TC1.8 mm 75 172 mm C TC1.8 mm 75 22 mm C TC1.8 mm 75 3 mm C H T =.4 m TC1.8 mm 75 35 mm C H T =.4 m TC.25 mm 1 3 mm C TC.25 mm 125 3 mm C TC1.8 mm 125 3 mm C TC.25 mm 15 3 mm C TC.25 mm 187.5 3 mm C TC.25 mm 225 3 mm C TC.25 mm 225 78 mm C TC.25 mm 225 125 mm C TC.25 mm 225 172 mm C TC.25 mm 225 22 mm C TC.25 mm 225 3 mm C H T =.4 m TC.25 mm 225 35 mm C H T =.4 m TC.25 mm 25 3 mm C TC1.8 mm 25 3 mm C TC.25 mm 275 3 mm C TC.25 mm 312.5 3 mm C TC.25 mm 35 3 mm C TC.25 mm 365 125 mm C TC.25 mm 375 3 mm C TC.25 mm 375 78 mm C TC.25 mm 375 125 mm C TC.25 mm 375 172 mm C TC.25 mm 375 22 mm C TC.25 mm 375 3 mm C H T =.4 m TC.25 mm 375 35 mm C H T =.4 m TC1.8 mm 375 3 mm C

5 Type of instrument Distance (mm) Height from Ceiling (mm) Placement sideways Comments TC1.8 mm 375 78 mm C TC1.8 mm 375 125 mm C TC1.8 mm 375 172 mm C TC1.8 mm 375 22 mm C TC1.8 mm 375 3 mm C H T =.4 m TC1.8 mm 375 35 mm C H T =.4 m Wall TC inside 375 125 mm Wall TC outside 375 125 mm BD-probe -275 125 mm C BD-probe 365 125 mm C PT -75 floor R 75 mm PT 75 floor R 75 mm PT 375 floor R 75 mm HF -75 floor L 75 mm HF 75 floor L 75 mm HF 375 floor L 75 mm O 2 385 3 mm C CO 385 3 mm C CO 2 385 3 mm C O 2 385 125 mm C CO 385 125 mm C CO 2 385 125 mm C TC.25 mm duct BD-probe duct O 2 duct CO duct CO 2 duct Weight a) The position is at the centre of the fire; negative coordinates are upstream of the fire while positive coordinates are downstream of the fire. The word "duct" means that the measurements were performed in the exhaust duct from the hood at the end of the tunnel. These measurements were used for the calculation of the HRR. b) C = centre, R = right, and L = left; as seen from the fan (see Figure 3.1 and Figure 3.2) TC = thermocouple TCS = thermocouple, type S (otherwise type K) BD = Bi-directional PT = Plate thermometer HF = Heat flux H T = height of tunnel

51 Appendix B Test1 Time-resolved test results

Test2 52

Test 3 53

Test 4 54

Test5 55

Test 6 56

Test 7 57

Test 8 58

Test 9 59

Test 1 6

Test 11 61

Test 12 62

Test 13 63

Test 14 64

Test 15 65

Test 16 66

Test 17 67

Test 18 68

Test 19 69

Test 2 7

Test 21 71

Test 22 72

Test 23 73

Test 24 74

Test 25 75

Test 26 76

Test 27 77

Test 28 78

Test 29 79

Test 3 8

Test 31 81

Test 32 82

Test 33 83

Test 34 84

Test 35 85

Test 36 86

Test 37 87

Test 38 88

Test 39 89

Test 4 9

Test 41 91

Test 42 92

SP Technical Research Institute of Sweden Our work is concentrated on innovation and the development of value-adding technology. Using Sweden's most extensive and advanced resources for technical evaluation, measurement technology, research and development, we make an important contribution to the competitiveness and sustainable development of industry. Research is carried out in close conjunction with universities and institutes of technology, to the benefit of a customer base of about 9 organisations, ranging from start-up companies developing new technologies or new ideas to international groups. SP Technical Research Institute of Sweden Fire Technology Box 857, SE-51 15 BORÅS, SWEDEN SP Report 27:5 Telephone: +46 1 516 5, Telefax: +46 33 13 55 2 ISBN 91-85533-69-6 E-mail: info@sp.se, Internet: www.sp.se ISSN 284-5172 www.sp.se