Multiple Fires in a Mine Drift with Longitudinal Ventilation

Transcription

1 Multiple Fires in a Mine Drit with Longitudinal Ventilation Analysis o ire behaviour o perormed ire experiments in a model scale mine drit Rickard Hansen

2 Table o Contents Abstract... 3 Nomenclature Introduction General ire behavior in a mine drit single ire source Perormed model scale ire experiments Results and discussion o ire experiments Fire gas temperatures Flame behaviour Spread mechanisms Fire growth rate and ire spread rate The heat release rates Conclusions Reerences... 29

3 Abstract The ire behaviour involving multiple ires in a mine drit with longitudinal ventilation was analysed. The conditions and ire phenomena occurring were described. The analysis was based upon experimental data rom model-scale ire experiments. A ire involving several uel items may lead to lames tilted horizontally and illing up the entire cross section, leading to earlier ignition, higher ire growth rates, higher ire spread rate and severe ire behaviour. Longer lame lengths will also result due to decreased air entrainment. A correlation or the continuous lame length was proposed. The results o the analysis will help identiying and preventing potentially dangerous ire situations with several large combustible items distributed along a mine drit. Multiple Fires in a Mine Drit with Longitudinal Ventilation 3

4 Nomenclature A is the cross-sectional area o the mine drit [m 2 ] c is the speciic heat [kj/kg K] p D is the diameter o the ire [m] E is the energy [kj] H is the mine drit height [m] H is the vertical distance between the ire source centre and the mine drit ceiling [m] H c is the heat o combustion [kj/g] L is the length scale L is the lame length [m] * L is the dimensionless lame length m is the mass [kg] m is the mass low rate o air [kg/s] a Q is the heat release rate [kw] * Q is the dimensionless heat release rate S is the ree distance between combustible items [m] S avg is the average ree distance between a number o combustible items [m] t is the time [s] T is the ambient temperature [K] a u is the longitudinal velocity [m/s] V is the dimensionless ventilation velocity * V is the dimensionless longitudinal velocity * w is the characteristic plume velocity φ is the ventilation/uel controlled criterion ϕ is the lame angle ormed by the straight line between the centre o the ire and the position o the maximum temperature beneath the mine drit ceiling ρ is the density o the ambient air [kg/m 3 ] a Multiple Fires in a Mine Drit with Longitudinal Ventilation 4

5 1. Introduction In a mine drit with combustible items distributed along the drit, one o the greatest risks would be a continuous ire spread rom one larger uel item to numerous adjacent items where the ire spread is anned by the longitudinal ventilation. This will result in severe ire behaviour and a major risk to the miners and the ire and rescue personnel. This paper ocuses on the phenomena occurring during the continuous ire spread rom item to item in a mine drit with longitudinal ventilation. What conditions and phenomena can be expected during such a ire? Knowing what phenomena that may occur, will allow or better ire design and ire saety at potential sites. When investigating the ire phenomena, experimental data rom earlier model-scale ire experiments were applied and analysed. The experimental data was obtained rom the modelscale experiments presented by Hansen and Ingason [1] as these tests were ound to it very well to the aim o the work presented here. The purpose o this paper is to describe the ire behaviour and its impact on the surroundings during a ire with multiple uel items on ire in a mine drit with longitudinal ventilation. Better knowledge about ire behaviour is needed or underground hard rock mines [2] and as the number o studies in the ield are generally limited to open space cases or cases with only natural ventilation the need or this knowledge is urther underlined. A number o studies have been conducted on the ire behaviour o merging lames rom multiple ires in open space [3-5]. The individual lames will merge into a single lame with longer lame length when placed suiciently close together. This is due to the reduction o air entrainment due to the adjacent ires which will cause a pressure gradient resulting in lames leaning towards each other and inally merging. The lame length will increase substantially with a decreasing distance when the lames start to emerge, but ater the lames are ully merged the distance between the ires will have little eect on the lame length. The studies have been conducted on ires with natural ventilation, using ire sources such as gas burners, pool ires and wood crib ires. As opposed to ires in the open, multiple ires in a mine drit or a tunnel will be aected by the surrounding suraces. Wan et al [6] perormed a number o ire experiments in a model scale tunnel with two propane burners acting as separate ires. During the experiments the heat release rate and distance between the burners were varied. It was ound that the spacing between the gas burners had a more signiicant eect on lame length in open space than in a tunnel. A criterion o beginning merging and ully merging lames were proposed. Models or predicting the ceiling gas temperature proiles and lame lengths were presented. Ji et al [7] presented a study where experiments with two pool ires in a model scale tunnel were conducted, studying the interaction between the ires as the heat release rates and distance between the ires were varied. The mass loss rate o the ires was ound to initially increase and then decrease as the distance was increased. The maximum mass loss rate was ound to occur at a shorter distance between the ires in a tunnel compared with the open case. The mass loss rate o a pool ire - with an adjacent pool ire in a tunnel - was higher than the corresponding case in the open. No orced, longitudinal ventilation was used during either o the studies. It is somewhat troublesome and doubtul to draw any extensive conclusions based upon experiments conducted with pool or gas burner ires on the ire behaviour or ires in a mine drit, as these ires will oten be ires in solid materials, with varying heat release rates, varying lame spread, varying lame lengths and where the re-radiation to the uel surace will have an inluence on the ire development (as opposed to gas burner ires). Ingason [8] perormed a number o model scale ire experiments simulating a tunnel ire involving several HGV trailers, using wood cribs as ire load. The main purposes o the work were to investigate the eects o variable ventilation rate on the maximum heat release rate and ire growth rate. It was ound that an increase in the ventilation rate lead to an increase in the maximum heat release rate per unit uel are which was 1.4 to 1.55 times higher than the corresponding open ire case. The increase in the ire growth rate was 2 to 3 times higher when increasing the ventilation rate to 3 m/s and 5 m/s respectively. The inluence o the longitudinal ventilation on the heat release rate o ventilation controlled ires will have an even urther impact where the porosity actor will be a key parameter. Multiple Fires in a Mine Drit with Longitudinal Ventilation 5

6 Hansen and Ingason [1] [9] presented studies ocusing on the ignition o individual uel items and the calculation o the heat release rate o multiple objects located in an underground structure. The output o the calculations were compared to model scale ire experiments with varying number o wood cribs and pallets placed at equal distances as well as varying distances rom the ignited wood crib and the pile o wooden pallets. The experiments were perormed in a model scale mine drit/tunnel with longitudinal ventilation. In the ollowing, the general ire behaviour o a single ire source in a mine drit is outlined and discussed, model scale experiments by Hansen and Ingason [1] are described and the results o the model scale experiments are presented and analysed with respect to the ire behaviour and the impact on the surroundings. Multiple Fires in a Mine Drit with Longitudinal Ventilation 6

7 2. General ire behavior in a mine drit single ire source The uel load in an underground mine can be considerable at speciic and isolated positions. Positions with a more or less continuous, spatial distribution o combustible items could or example be a mine drit used as a parking lot or larger mining vehicles. The uel load in a mine drit will mostly be ound in the lower regions and will thereore not necessarily be enguled in hot ire gases or largely aected by ceiling impingement o lames, thus limiting the eect o the ire spread mechanisms [2]. Several dierent types o combustibles can be ound in a mine drit depending on the type o activity. Types o combustibles could or example be: vehicles, lammable liquid, wooden pallets, electrical cables, tyres, hoses, conveyor belts etc. Vehicle ires are the most common type o ire in underground hard rock mines [10-11] and may result in high heat release rates and extensive smoke spread. The maximum heat release rate o larger mining vehicles can be several tens o megawatts, putting the ire protection systems in a mine to the test and causing great problems and risks to the personnel in the mine. Depending on the type o vehicle the uel load and construction will vary, but generally combustibles such as tyres, lammable liquids, cables and hoses are ound on the vehicles. Fires in lammable liquids can be characterized by the rapid ire growth; considerable and rapid smoke production. Fires in tyres and larger amounts o hydraulic hoses are distinguished by the extensive smoke production and the long ire duration which will increase the risk to evacuating miners. The heat release rates o ires in tyres, cables, hoses and other solid materials will be dictated by the size o the involved surace area at the time. A ire in a mine drit will either be ventilation controlled or uel controlled. The heat release rate o a ventilation controlled ire will be dictated by the amount o oxygen available (i.e. there is no excess o oxygen) and the amount o uel will dictate the heat release rate o a uel controlled ire. The ollowing equation is used or determining whether the ire is ventilation controlled or uel controlled [12]: m = 3000 a Q φ (1) The ire is uel controlled i equation (1) is larger than 1. Due to the generally large dimensions o a mining drit and the presence o longitudinal ventilation, a ire in a mine drit will in most cases be a uel controlled ire. Unshielded lames in a mine drit are aected by the longitudinal ventilation low rom the mechanical ventilation, which can visually be seen in the tilting o lames. Tilting lames lead to aster lame spread, higher ire growth rate, higher maximum heat release rates and aster ignition o adjacent uel as the view actor will increase and lame impingement may occur as well. The ollowing lame tilt angle relationships o Li and Ingason [13] were developed or tunnel ires with longitudinal ventilation and validated against model scale as well as large scale ire experiments: sin ϕ = 1 V (2) 3 / 5 ( 5.26 ) sinϕ = V V > Q * (3) Multiple Fires in a Mine Drit with Longitudinal Ventilation 7

8 sin 1/ 5 D * 3 ϕ = 0.25 V / H > V * > Q (4) u V = (5) * w w * g Q = D ρ a c 2 p T a 1/ 3 (6) V * = u g H (7) = Q ρ * 1/ 2 5 / 2 (8) a c p Ta g H Q The lame length will also play an important part with respect to the spread o ire to adjacent items due to the importance o the lame radiation mechanism. The ollowing correlations o Ingason and Li [14] have been ound to match observed non-dimensional lame lengths rom perormed ull-scale ire experiments in a mine drit [15]: L * * 4.3 Q = (9) L * = * L H = Q ρ c T g (11) Q a p a 1/ 2 A H 1/ 2 (10) The mine drit geometry is accounted or in the correlations above, which underlines the inluence o the geometry on the lame lengths in a mine drit. As the longitudinal ventilation low increases, the lame lengths and lame temperatures in a mine drit will decrease due to increased air entrainment which will increase the combustion eiciency. The initial increase in the heat release rate and ire growth rate due to increased longitudinal ventilation low will eventually reach a maximum value at a certain ventilation velocity. A urther increase in the ventilation velocity will result in a decrease in the heat release rate and ire growth rate as the convective cooling overtakes the radiative heating rom the lames. In igure 1 the measured incident heat lux at loor level in a model scale mine drit/tunnel ire experiment can be seen. The experiments are described urther in the coming chapter. The heat lux was measured 1 m downstream o a burning pile o wooden pallets at three dierent ventilation velocities. As can be seen, a higher maximum heat lux Multiple Fires in a Mine Drit with Longitudinal Ventilation 8

9 was measured or the 0.6 m/s case compared with the 0.3 m/s and 0.9 m/s cases. The ventilation velocities o the model scale experiments equal 1.16 m/s, 2.32 m/s and 3.49 m/s respectively in ull scale Heat lux (W/m 2 ) u=0.3 m/s u=0.6 m/s u=0.9 m/s t (s) Figure 1. Heat luxes at dierent ventilation velocities in model scale ire experiments. The dominating ire spread mechanism will vary with the distance rom the ire, the position o the uel items with respect to the ventilation low direction as well as the spread o the ire gases, and type o material involved in the ire. For shorter distances the lame radiation will play a crucial part with respect to spread mechanisms. Thus the lame radiation mechanism will be very important during the ire spread between the various uel components o a vehicle. The convective heat transer mechanism will increase in importance with increasing distance. The ventilation low may push the ire gases away rom an adjacent uel component, thus decreasing the impact o the convective heat transer term. On the other hand, a longitudinal ventilation low will tilt lames towards adjacent uel items and increase the importance o the lame radiation transer term. The mine drit itsel will also inluence the ire behaviour. The surrounding rock close to the ire will increase the re-radiation mechanism back to the ire and increase the heat release rate. The rock urther downstream o the ire will instead have a cooling eect on the ire gases and thereore decrease the stratiication o the smoke and the risk o ignition o uel items downstream. The mine drit height will in many cases be considerable, decreasing the risk o ceiling impingement o lames. The inclination o the mine drit will also inluence the ire behaviour, where the inclination will lead to increasing lame tilt which in turn will lead to earlier ignition o uel items, aster lame spread etc. The general openness and cooling eect o a mine drit - together with the generally limited amount o combustible material - will decrease the likelihood o a lashover. Even though a lashover is less likely to occur; severe, ast spreading and high intensity ires are not unlikely. A mine drit with a continuity o large amount o combustibles could very well be the place o a large and dangerous ire. The smoke spread in a mine drit is largely determined by the longitudinal ventilation low. The longitudinal ventilation velocity will determine the occurring smoke stratiication, together with the dimensions o the mine drit, the heat release rate as well as the distance to the ire. With a low or no orced air velocity the smoke stratiication will be high in the vicinity o the ire and at high air velocities the smoke stratiication will be low downstream rom the ire. With increasing mine drit height and increasing distance to the ire, the vertical temperature gradients as well as the smoke stratiication will decrease. An increase in the heat release rate will result in an increase in the vertical temperature gradients and the smoke stratiication. Figure 2 displays the smoke stratiication along a mine drit, where the vertical temperature gradient, stratiication as well as average ire gas temperature decrease with increasing distance rom the ire. Multiple Fires in a Mine Drit with Longitudinal Ventilation 9

10 Figure 2. The smoke stratiication in a mine drit or a single ire source [2]. Multiple Fires in a Mine Drit with Longitudinal Ventilation 10

11 3. Perormed model scale ire experiments In the ollowing a brie summary o the perormed ire experiments is presented. For a more detailed description o the wooden pallets, the experimental procedures and heat release rate results o the experiments see Hansen and Ingason [16]. Hansen and Ingason [1] presented a series o model scale mine drit/tunnel ire tests, carrying out a total o 12 tests in a 1:15 model scale tunnel (the size o the tunnel was 10 m long, 0.6 m wide and 0.4 m in height). The parameters tested during the tests were: the distance between piles o wooden pallets and longitudinal ventilation rate (longitudinal ventilation velocities o 0.3 m/s, 0.6 m/s and 0.9 m/s were applied during the experiments). The aim o the study was to investigate the eect o varying distances between the wooden pallet piles on ire spread under dierent longitudinal ventilation rates. See igure 3 or a sketch o the model tunnel and igure 4 or the position o thermocouples, probes and instruments. exhaust pipe Φ 260 mm 525 mm 850 mm 2500 mm 2500 mm 2500 mm 2500 mm axial an steel net X 800 mm 1800 mm 880(L)*880(W) *880(H) 880 mm smooth pipes net connected to central system ignited wood crib target Figure 3. A schematic drawing o the model tunnel [8] X T1 T2 T3 T4 T5 T6 T7 T9 T10 T11 T12 B22 B23 T13 S24 S25 S26 G28 G W pile A mm S27 pile B T=thermocouple B=bi-directional probe S=Schmidt-Boelter gage G=gasanalysis W=Weightloss T8 T14 T15 T16 T17 40mm 120mm 200mm 280mm 360mm T18 T20 T12 T19 T21 Thermocouple K 0.25 mm bi-directional probe lux meter Thermcouple pile Gasanalysis Thermocouple pile A Thermocouple pile B Figure 4. Position o thermocouples, probes and instruments [8]. The applied ire load in the experiments consisted o piles o scaled down sotwood pallets (pine), where each pile was composed o ive individual wooden pallets. Three test ires were used as reerence tests and consisted o a single pile o pallets, whereas in the other tests the ire load consisted o our piles o wood pallets placed at dierent distances between each other. During the tests the ollowing parameters were measured or calculated: > Time o ignition o adjacent piles were clocked and documented manually. > The total heat release rate. Multiple Fires in a Mine Drit with Longitudinal Ventilation 11

12 > Fire gas temperatures; where most o the thermocouples were placed along the ceiling. Two sets o thermocouples were also positioned 4.65 m and 8.75 m rom the inlet opening, measuring the temperatures at various vertical positions rom which an average ire gas temperature could be calculated. > Gas concentrations at the end o the tunnel. > Heat luxes at loor level at several positions. > The centreline low velocity. > The centreline pressure dierence. All experiments were visually documented using a video camera, recording the ire behaviour along the long side o the tunnel. The long side o the tunnel was covered with a ire resistant 5 mm window glaze, allowing or visual observations o ignition, lame lengths, tilting lames etc. See table 1 or the results rom the model scale experiments. Experiment #2 comprised o our piles o wooden pallets but where only the initial pile took part in the ire and was thereore not included in the analysis. The ignition time o the irst pile o wooden pallets was set to 0 s. Multiple Fires in a Mine Drit with Longitudinal Ventilation 12

13 Table 1. Results rom the model scale ire experiments. Test # u [m/s] Number o piles Free distance between the piles [m] Maximum heat release rate [kw] Ignition times o second and third pile [s] Engulment times o irst, second and third pile [s] Time ater ignition when lames became horizontal [s] Heat release rate when lames became horizontal [kw] Measured continuous lame lengths [m] Heat release rate at measured continuous lame lengths [kw] ; 0.7; ; ; 131; ; 0.7; ; ; 158; ; 0.8; ; 0.8; ; 0.8; ; 0.8; ; ; 142; ; ; 140; ; ; 135; ; ; 144; ; 0.8; ; ; 166; ; 0.9; ; ; 177;

14 The use o model scale ire experiments is convenient as it allows or less costly experiments where the results can be translated into a ull scale system, using the theory o dimensionless groups as a basis. A relatively well deined similarity exists between model scale and ull scale experiments. The size o the model scale mine drit was scaled geometrically according to a 1:15 ratio. The scaling models o the heat release rate, low rate, time, energy content, mass and temperature, can be ound in table 2. The index F relates to the ull scale (i.e. 15 in our case) and the index M relates to the model scale (i.e. 1 in our case). The inluence o the thermal inertia o the involved material, the turbulence intensity and radiation was neglected. Table 2. List o scaling models [8]. Unit Scaling model Equation number Heat release rate [kw] 5 / 2 LF Q = F Q M LM Velocity [m/s] 1/ 2 LF u = F um LM Time [s] 1/ 2 LF t = F t M LM (12) (13) (14) Energy [kj] E F = E M L L F M 3 H H c, M c, F (15) Mass [kg] 3 LF m = F mm LM (16) Temperature [K] T F = T (17) M Multiple Fires in a Mine Drit with Longitudinal Ventilation 14

15 Multiple Fires in a Mine Drit with Longitudinal Ventilation Results and discussion o ire experiments A ire in a mine drit involving multiple separate ires and with longitudinal ventilation will be complex and aected by numerous actors: lame tilt, varying degree o air entrainment depending on the position o the ire versus the other ires and the ventilation direction, distance between the ires, number o ires, heat release rates o the individual ires, ire interaction, eect o surrounding suraces and dimensions o the mine drit, orientation o uel suraces versus ventilation low, low o ire gases and lames etc. In some cases the various actors will aect each other and urther complicate the ire behaviour. 4.1 Fire gas temperatures A ire gas layer temperature criterion o 600 C - which is a lashover criterion or remote ignition [17] - was applied at roo level to study the extent o the ire gas layer with lashover potential. In igure 5 the roo temperatures or the thermocouples in experiment #7 that exceeded 600 C can be seen. The ire gas layer with lashover potential will not be stationary, instead it will move in the direction o the ventilation low. Also, the extent o the layer will initially increase with each pile o pallets being ignited and then gradually decrease as the number o piles is limited. A spatial continuity in the combustible material is essential to allow the lashover conditions to increase and progress along the mine drit/tunnel. The span o distances between the piles o pallets in the model scale experiments (0.4 m m) correspond to a span between 6 m and 16.5 m in ull scale. Figure 6 displays the roo temperatures o experiment #4 where the ire gas layer temperature exceeded 600 C. Comparing with igure 5, it is clear that the extent o the zone where lashover may occur is limited - both in time and space - compared with experiment #7 due to the lack o continuity in combustible material. Even though it is generally less likely that a lashover would occur in a mine drit compared with a ire in an enclosure, the experiments clearly show the possibility given the right circumstances. A lashover in an open mine drit would not include the entire mine drit at one time, but would instead have a limited spatial extent and non-stationary characteristics. The extent o the zone where lashover may occur was 16.5 m (ull scale) downstream o the last pile o pallets in experiment # Temperature (C) t (s) T6 T7 T8 T9 T10 T11 Figure 5. The roo temperatures o thermocouples exceeding 600 C in experiment #7.

16 Temperature (C) t (s) T6 T7 Figure 6. The roo temperatures o thermocouples exceeding 600 C in experiment #4. The temperature readings at pile A in experiment #4 can be seen in igure 7. The experiment only involved one pile o pallets. As noted there is a distinct temperature gradient between the lower two thermocouples and the upper three thermocouples. A clear stratiication resulted rom the experiment during a majority o the ire except at the very end. The corresponding temperature readings o experiment #7 can be seen in igure 8. As opposed to the temperature readings rom experiment #4, the temperatures o experiment #7 display a period subsequent to the early growth phase and prior to the decay phase, where the temperature gradient is very low and resulting in very little stratiication. The lowest thermocouple (T17) in igure 7 displays a temperature that starts to level out ater approximately 90 seconds, while the same thermocouple in igure 8 displays a temperature that continues to increase more and more. At this point o time, the second pile o wooden pallets was ignited in experiment #7 while experiment #4 only involved one pile total. Thus the ire o experiment #7 continued to increase in intensity and temperature due to the continuation o combustible material. In igure 8 the period with little stratiication is initiated when temperature readings o the two lowest thermocouples (T16 and T17) start a very rapid climb ater about 150 s. Studying the video recordings rom the experiment, it was seen that the ire had spread to the lower parts o the second pile and the lames started to tilt more or less horizontally at this time, directed towards the thermocouples o pile A. Figure 9 displays the temperature readings o pile B in experiment #11, where the last pile o pallets was positioned closest to pile B (approximately 2.1 m, which equals 31.5 m in ull scale) among all experiments. Same as or experiment #7, igure 9 displays a period with smaller temperature gradient and thus less stratiication. This period starts ater approximately 280 s where the lowest thermocouple (T21) initiates a rapid increase in temperature. Studying the video recordings, the same phenomenon occurs as in experiment #7 where the pile o pallets closest to pile B is ignited shortly beore and the lames toward pile B is tilted horizontally even rom the lower parts o the pile. The model scale experiments displayed a dierent temperature distribution and stratiication along the mine drit compared with the case o only one ire source (seen in igure 2). Instead o a clear temperature gradient and a clear stratiication, the experiments showed a more or less uniorm vertical temperature distribution - with high temperatures - and little stratiication or a considerable distance along the model scale mine drit. Multiple Fires in a Mine Drit with Longitudinal Ventilation 16

17 Multiple Fires in a Mine Drit with Longitudinal Ventilation T (C) T8 T14 T15 T16 T t (s) Figure 7. The temperature readings at pile A in experiment # T (C) t (s) T8 T14 T15 T16 T17 Figure 8. The temperature readings at pile A in experiment # T (C) T12 T18 T19 T20 T t (s) Figure 9. The temperature readings at pile B in experiment #11.

18 4.2 Flame behaviour During the experiments with multiple ires, it was ound that during all experiments did the lames tilt horizontally and illed up the entire cross section (see igure 10). Applying equations (2)-(8) or the model scale experiments, the results did not predict horizontal lames. This is not surprising as Li and Ingason [13] ocus on the ceiling temperature when predicting the tilt o the lames and will not give any clues on the temperatures and lame tilts closer to the ground. Also, or very large ires the lame volume would be across the entire cross section and not only at the ceiling. Figure 10. The lame volume across the entire cross section in experiment #5. When studying the video recordings o the experiments it could be seen that cut-o ceiling impingement - o the lames occurred or the irst pile o pallets. But as the lames started to tilt horizontally or the subsequent piles, the ceiling impingement more or less ceased. Studying the video recordings rom the experiments, the lames did not ully merge with each other (other than very occasionally and only at the very top and base). Instead they tied into each other, orming a more or less continuous lame along the mine drit due to the longitudinal ventilation low and lame volume across the entire cross section. When trying to visually observe the individual lame lengths o the piles o wooden pallets it was ound to be diicult due to the soot that covered the lame tip, soot on the window glaze and the luctuating nature o the lame tips. Instead the lame length at the various thermocouples positioned along the mine drit and pile A and B was determined by using the temperature readings o the various positions, assuming a lame tip Multiple Fires in a Mine Drit with Longitudinal Ventilation 18

19 Multiple Fires in a Mine Drit with Longitudinal Ventilation 19 temperature o 500 K above ambient and that the continuous lame has reached pile A or B when the temperature at any o the thermocouple is constantly above 500 K o ambient [18]. See table 1 or the measured continuous lame lengths and corresponding heat release rates o the experiments. The measured lame lengths indicated increasing and longer lame lengths with each ignited pile o pallets urther downstream (when comparing with the lame length o the reerence test). Figure 11 displays the continuous lame lengths o experiment #4 (reerence test) and #7 as a unction o time until the maximum lame length was attained. The maximum continuous lame length o experiment #7 was more than our times larger than the maximum lame length o experiment #4 even though the maximum continuous lame length o experiment #7 only encompassed three piles o wooden pallets. The longitudinal ventilation low will increase the combustion eiciency o the irst pile o wooden pallets, resulting in increased heat release rate and ire growth rate, and decreasing lame lengths and lame temperatures or the irst pile. But urther downstream the air entrainment will decrease with each uel item, resulting in increased lame lengths and lame temperatures which in turn will lead to earlier ignition o adjacent uel items. The ire growth rate and heat release rate o uel items urther downstream will still increase despite decreased air entrainment as more uel items will be ignited at an earlier stage o the ire and a larger portion o the uel surace o each uel item will be ignited at an earlier stage due to lames tilted horizontally and illing up the entire cross section. This will increase in importance as the distance between the ires decrease and the number o ires increase. Flame length (m) 5 4,5 4 3,5 3 2,5 2 1,5 1 0, t (s) Experiment #7 Experiment #4 Figure 11. The lame length developments o experiment #4 and #7. When applying equations (9)-(11) or calculating the lame lengths, it was ound that the calculated lame lengths were considerably longer than the measured lame lengths (see igure 12 or the calculated and measured lame lengths o experiment #3). The use o the lame length correlations o Ingason and Li [14] would in this case seem to it a single ire source better than or multiple ire sources. A possible explanation could be that the air entrainment or several ire sources would be more eicient than the air entrainment o a single ire source with the same heat release rate, thus resulting in a longer lame length or a single ire source.

20 6 Flame length (m) Measured lame length Calculated lame length t (s) Figure 12. Flame lengths o experiment #3. Applying dimensional analysis on the continuous lame length along the model scale mine drit, the ollowing governing parameters were included based upon earlier studies [3], [7], [19], [20]: L ( Q, ρ, c, T, g, H S ) = (18) a p a, An earlier paper by Ingason and Li [14] had ound that the lame length was only a weak unction o the longitudinal ventilation low and was thereore not included in the dimensional analysis. Furthermore, when analyzing the model scale ire experiments it was ound that the uel diameter did not play a signiicant role when determining the lame length due to the long lame lengths encountered or multiple ires. The uel diameter was thereore also excluded rom the dimensional analysis. The normalized lame length: L S = ρ a c p Q T g a 1/ 2 S 5 / 2 H, S (19) L S ρ a c Q g = 1/ 2 3 / 2 p Ta S H (20) Setting the dimensionless heat release rate equal to: = Q ρ * (21) 1/ 2 3 / 2 a c p Ta g S H Q Perorming a regression analysis on the measured lame lengths or various total heat release rates and Multiple Fires in a Mine Drit with Longitudinal Ventilation 20

21 Multiple Fires in a Mine Drit with Longitudinal Ventilation 21 average ree distances between the piles o wooden pallets (as the distances varied in the same experiment), resulted in the ollowing correlation: L * = 1.61 S avg Q (22) The plotting o the correlation between the normalized lame length and normalized heat release rate can be seen in igure 13. A R 2 value o resulted, showing that the continuous lame length can be reasonably well described by using equation (22). The high correction coeicient value urther motivates the exclusion o the longitudinal ventilation low and uel diameter rom the dimensional analysis. Equation (22) expresses the continuous lame length o multiple ires in a mine drit with longitudinal ventilation low. Cases with several separate lames are not accounted or in equation (22) L/S ,5 1 1,5 2 2,5 3 3,5 4 4,5 5 Q* Figure 13. Correlation o normalized lame length against normalized heat release rate. 4.3 Spread mechanisms The main ire spread mechanisms in all experiments were convection, lame radiation and lame impingement, but the dierence lays in the size o the uel area being aected as a unction o the time. Figure 14 displays the moment just beore ignition o the second pile o wooden pallets o test ire #5, as can be seen lame radiation will aect the entire side and upper surace o the pile but mostly the upper part due to the tilted lame. Convection and lame impingement will mostly aect the upper part o the pile. Figure 15 displays the moment where the third pile o wooden pallets o test ire #5 is ignited. As can be seen, convection, lame radiation and lame impingement will aect the entire side o the pile as well as the upper surace. Even though the upper part o the pile is ignited irst, the entire side o the pile will ignite more or less momentarily shortly aterwards. Thus a larger uel surace area will be on ire at an early stage, leading to a aster ire growth rate and a higher maximum heat release rate. A ire scenario with uel suraces at the lower regions o the mine drit being exposed to a larger degree o convection, lame radiation and lame impingement may have a severe impact on the ire saety. This is due to the act that the larger amount o combustible materials is ound in the lower regions o the mine drit [2] and the ire protection measures in the mine drit may not have been addressed with respect to such ire behaviour.

22 Figure 14. Photograph taken just beore ignition o the second pile o wooden pallets o test ire #5. Multiple Fires in a Mine Drit with Longitudinal Ventilation 22

23 Multiple Fires in a Mine Drit with Longitudinal Ventilation 23 Figure 15. Photograph taken at the ignition o the third pile o wooden pallets o test ire # Fire growth rate and ire spread rate The heat release rate o the reerence test was subtracted rom the total heat release rate o the other experiments in order to analyse the heat release rate o the second pile o wooden pallets, assuming that the heat release rate o the reerence test was similar to the heat release rate o the irst pile in the dierent tests. The ire growth rate and heat release rate o the second pile o wooden pallets were examined and compared with the corresponding values o the irst pile, rom the ignition o the pile until the time when the third pile was ignited. The resulting heat release rate rom experiment #9 can be seen in igure 16. As can be seen the ire growth rate as well as the heat release rate increased dramatically or the second pile.

24 Heat release rate (kw) Reerence test Second pile o test # t (s) Figure 16. The initial heat release rate o the reerence test and the second pile o experiment #9. The resulting ire growth rates (a parabolic growth rate was applied as the ire growth rates had a parabolic appearance) and heat release rates can be seen in table 3. As can be seen, the ire growth rate o the second pile increased with decreasing distance between the irst and second pile. The ire growth rate and heat release rate (at t=30 s ater ignition) were considerably higher or the second pile compared with the reerence test o the irst pile. Indicating an increasing ire growth rate, more rapid ignition, higher maximum heat release rate, escalating ire severity, worsening ire behaviour with increasing number o adjacent combustible items downstream o the ire. Table 3. Fire growth rates and heat release rates o the reerence test and the second pile o the experiments with multiple piles o wooden pallets. Experiment # Parabolic ire growth rate [kw/s 2 ] Heat release rate at t=30 s [kw] (reerence) Free distance between irst and second pile [m] Using the heat release rate curve o the reerence test (#4), adding the curve at the recorded ignition times o the individual piles and summing it up into a total heat release rate o the experiment in question, igure 17 Multiple Fires in a Mine Drit with Longitudinal Ventilation 24

25 Multiple Fires in a Mine Drit with Longitudinal Ventilation 25 displays the summed up heat release rate o experiment #3 and the measured heat release rate. As can be seen, the measured heat release rate displays a more rapid ire growth rate ater the initial phase (where only the irst pile was involved in the ire) compared with the summed up heat release rate, showing that the ire growth rate o the second, third and ourth pile was higher compared with the irst pile. 600 Heat release rate (kw) Summed up heat release rate Measured heat release rate t (s) Figure 17. The summed up heat release rate and measured heat release rate o experiment #3. The video recordings rom the experiments were studied in order to obtain approximate times rom ignition o a pile and until the pile was ully involved in the ire. The pile was assumed to be ully involved in the ire when lames were seen lowing rom the bottom to the top o the pile on the downstream side. Due to soot ormation on the window glaze it was diicult to observe lames emitted rom the ourth pile o pallets and the results o the ourth pile was thereore omitted. See table 1 or the ignition times and times o engulment or the experiments. The average time duration - rom ignition to ully involved ire - was 62 s (standard deviation o 5.2) or the irst pile, 39 s (standard deviation o 4.9) or the second pile and 25 s (standard deviation o 7.5) or the third pile. Thus a decreasing time duration and an increasing ire growth rate or each additional pile downstream o the initial ire. Please observe that the irst pile was ignited at the bottom o the pile and the buoyancy clearly acilitated the lame spread along the pallet suraces. In the case o the second pile, the pile was ignited at the top and the lame spread was not assisted by the buoyancy. The time dierence between the irst and second pile should thereore have been larger i the ignition point would have been identical in both cases. The measured heat lux values at the loor level were studied in order to obtain approximate values o the ire spread along the mine drit/tunnel. A heat lux criterion o 13.1 kw/m 2 was set based upon the critical heat lux o the wooden pallets [1]. The ire was assumed to reach the position o the lux meter when the measured heat lux exceeded the critical heat lux. It was ound that the ire accelerated considerably during the course; three to our times higher spread rates were measured between the second and third heat lux meter compared with the spread rate between the irst pile o pallets and the second heat lux meter (the irst heat lux meter was positioned upstream o the irst pile). The spread rates between the third and ourth heat lux meter showed diverging results, which is due to that no pile o pallets was positioned downstream o the ourth heat lux meter. The highest spread rates - between the third and ourth heat lux meter - were recorded in the cases where the last pile o pallets was positioned closest to the ourth heat lux meter. The overall average ire spread rate o the various experiments was ound to be similar, corresponding to approximately 0.1 m/s in ull scale. See igure 18 or ire spread rates o experiment #3, #8, #9 and #11.

26 Spread rate (m/s) Start-S25 S25-S26 S26-S27 Experiment #3 Experiment #8 Experiment #9 Experiment #11 Figure 18. The ire spread rate between the dierent heat lux meters along the model scale mine drit/tunnel. Similar results were obtained when applying the ire gas layer temperature criterion o 600 C at roo level to pinpoint the position o the ire at a given point o time. The ire accelerated considerably during parts o the distance, the ire spread rate increased approximately our times during the course. In igure 19, the ire spread rates between the dierent thermocouples at the roo level are displayed (starting with the distance between the irst pile and thermocouple 6, see igure 4 or the position o the thermocouples) or experiments #4, #7, #8 and #9. As can be seen, the ire spread rate started to accelerate between thermocouple T8 (positioned in between the second and third pile) and T9 (positioned right ater the third pile). The ire behaviour changed signiicantly ater the second pile. Experiment #4 never underwent the acceleration phase as the other experiments, as experiment #4 was a reerence experiment only containing one pile o wooden pallets. A spread rate o 0.1 m/s in the model scale experiments corresponds to a spread rate o approximately 0.4 m/s in ull scale. Ater T9 and the acceleration phase the spread rates o the ires decreased and levelled out. Thermocouple T10 was positioned at or very close to the last pile o wooden pallets, the decreasing ire spread was due to that the ire lost momentum when it ran out o uel. 0,12 0,1 Spread rate (m/s) 0,08 0,06 0,04 0,02 Experiment #4 Experiment #7 Experiment #8 Experiment #9 0 Start-T6 T6-T7 T7-T8 T8-T9 T9-T10 T10-T11 Figure 19. The ire spread rate between dierent thermocouples along the model scale mine drit/tunnel. 4.5 The heat release rates Multiple Fires in a Mine Drit with Longitudinal Ventilation 26

27 Multiple Fires in a Mine Drit with Longitudinal Ventilation 27 The ire experiments were ound to be uel controlled applying equation (1) and the measured maximum heat release rates ound in table 1 - and not ventilation controlled - throughout the entire duration o the ires. This is in line with the environment o a mine drit, where the longitudinal ventilation low together with the large air masses available, making ventilation controlled ires less likely compared with or example compartment ires [2]. The dimensions o the model scale mine drit correspond to a height o 6 m and a width o 9 m in ull scale. With a longitudinal ventilation velocity o 2.32 m/s (the ull scale velocity corresponding to the model scale velocity o 0.6 m/s) the maximum heat release rate would be approximately 450 MW, which is a very large ire and not very likely in a mine drit. Ater the ignition o the second pile o wooden pallets, lames started to tilt horizontally and ill up the entire cross section. See table 1 or the time points where the lames became horizontal and the corresponding heat release rates. I applying the heat release rate at the time o when the lames started to display this behaviour as a threshold or the severe ire behaviour, an average heat release rate o 240 kw (standard deviation o 40 kw) was calculated or the experiments. The equivalent ull scale heat release rate would be approximately 209 MW. Potential sites or such a high heat release rate could be places where larger vehicles are accumulated: a parking drit with larger vehicles parked side by side or a workshop servicing a number o larger vehicles.

28 5. Conclusions The ire behaviour involving multiple uel items in a mine drit with longitudinal ventilation was analysed. Focusing on the phenomena occurring during the ire spread between multiple items in a mine drit. It was ound based upon perormed ire experiments in a model-scale mine drit that: > Even though a lashover in an open mine drit is less likely compared with an enclosure ire, a progressively advancing lashover in a mine drit with a spatial continuity o combustible items is a possibility and a risk. > A ire involving multiple uel items may lead to high ire gas temperatures and lames across the entire cross section o the mine drit and with lames tilted horizontally. Causing earlier ignition, higher ire growth rate, higher maximum heat release rate, increasing ire spread rate and more severe ire behaviour. > A ire scenario in a mine drit where uel suraces in the lower regions are exposed to a larger degree o convection, lame radiation and lame impingement may have a severe impact on the ire saety. Larger amount o combustible materials are ound in the lower regions and ire protection systems may not have been designed with respect to the severe ire behaviour. > The measured lame lengths showed increasing and longer lame lengths with each ignited uel item urther downstream, due to decreasing air entrainment. A correlation or the continuous lame length was analysed and proposed or multiple ires in a mine drit. > The analysis urther underlines the importance o preventing the ire rom spreading to an adjacent, larger uel item. Preventing a severe and accelerating ire behaviour that could have disastrous eects in an underground mine. The indings o the analysis would help identiying and preventing potentially dangerous ire situations and improve the design o ire saety measures with respect to the ire behaviour. Multiple Fires in a Mine Drit with Longitudinal Ventilation 28

29 Multiple Fires in a Mine Drit with Longitudinal Ventilation Reerences [1] Hansen R, Ingason H. Heat Release Rates o Multiple Objects at Varying Distances. Fire Saety Journal 2012;52:1-10. [2] Hansen R. Study o heat release rates o mining vehicles in underground hard rock mines. Doctoral thesis. Västerås: Mälardalen University; [3] Thomas PH, Baldwin R, Heselden JM. Buoyant Diusion Flames: Some Measurements o Air Entrainment, Heat Transer, and Flame Merging. Tenth International Symposium on Combustion 1965; [4] Baldwin R. Flame Merging in Multiple Fires. Combustion and Flame 1968;12: [5] Kamikawa D, Weng WG, Kagiya K, Fukuda Y, Mase R, Hasemi Y. Experimental study o merged lames rom multiire sources in propane and wood crib burners. Combustion and Flame 2005;142: [6] Wan H, Gao Z, Ji J, Li K, Sun J, Zhang Y. Experimental study on ceiling gas temperature and lame perormances o two buoyancy-controlled propane burners located in a tunnel. Applied Energy 2017;185: [7] Ji J, Wan H, Gao Z, Fu Y, Sun J, Zhang Y, Li K, Hostikka S. Experimental study on lame merging behaviors rom two pool ires along the longitudinal centerline o model tunnel with natural ventilation. Combustion and Flame 2016;173: [8] Ingason H. Model Scale Tunnel Fire Tests, SP report 2005:49. Borås: Swedish National Testing and Research Institute; [9] Hansen R, Ingason H. An engineering tool to calculate heat release rates o multiple objects in underground structures. Fire Saety Journal 2011;46: [10] De Rosa MI. Analysis o mine ires or all US metal/non-metal mining categories, NIOSH; [11] Hansen R. Investigation on ire causes and ire behaviour vehicle ires in underground mines in Sweden Work report SiST 2013:3. Västerås: Mälardalen University; [12] Beard A, Carvel R. The handbook o tunnel ire saety. London: Thomas Telord Ltd; [13] Li YZ, Ingason H. Maximum ceiling temperature in a tunnel ire. SP Report 2010:51, Borås: SP Swedish National Testing and Research Institute; [14] Ingason H, Li YZ. Model scale tunnel ire tests with longitudinal ventilation. Fire Saety Journal 2010;45: [15] Hansen R. Analysis o Methodologies or Calculating the Heat Release Rates o Mining Vehicle Fires in Underground Mines. Fire Saety Journal 2015;71: [16] Hansen R, Ingason H. Model scale ire experiments in a model tunnel with wooden pallets at varying distances. Research report SiST 2010:8. Västerås: Mälardalen University; [17] Quintiere JG. Fundamentals o ire phenomena. Chichester: John Wiley & Sons Ltd; [18] Heskestad G. Fire Plumes, Flame Height and Air Entrainment. In: The SFPE Handbook o Fire Protection Engineering, Ed: DiNenno PJ, NFPA, Quincy, USA; [19] Quintiere JG. Scaling applications in ire research. Fire Saety Journal 1989;15:3-29. [20] Zukoski EE. Properties o ire plumes, Combustion undamentals o ire. London: Academic Press; 1995.