- Email: [email protected]
A comprehensive study of two fire sources in a road tunnel: Considering different arrangement of obstacles
Tunnelling and Underground Space Technology 59 (2016) 91–99
Contents lists available at ScienceDirect
Tunnelling and Underground Space Technology journal homepage: www.elsevier.com/locate/tust
A comprehensive study of two fire sources in a road tunnel: Considering different arrangement of obstacles Ghassem Heidarinejad ⇑, Maryam Mapar, Hadi Pasdarshahri Faculty of Mechanical Engineering, Tarbiat Modares University, PO Box 14115-143, Tehran, Iran
a r t i c l e
i n f o
Article history: Received 26 October 2015 Received in revised form 3 June 2016 Accepted 4 June 2016
Keywords: Tunnel Fire LES Critical ventilation velocity
a b s t r a c t This paper uses Fire Dynamics Simulator (FDS) to study various arrangements of different vehicles at upstream of two fire sources. In order to make a comprehensive study, the effects of two fire sources in both lateral and longitudinal directions are investigated. The results reveal that the behavior of two fire sources, in both perpendicular directions, is directly influenced by distance between them. For small vehicles, variations of the arrangement and distance between the vehicles and fire sources do not affect the calculated Critical Ventilation Velocity (CVV). However, the presence of medium vehicles strengthens the influence of inertia force rather than buoyant force of fire plume in the tunnel. Accordingly, when there is a short distance between fires and medium obstructions, less air ventilation is needed to prevent smoke back-layering. Eventually, far distance between the vehicles and the fires results in vanishing obstruction effects. Consequently, CVV is the same as the case in which there is no vehicle in the tunnel. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Generally, fire in a road tunnel forms a complicated structure. This physical phenomenon which involves chemical reaction, turbulence, and radiation is affected by various parameters, including geometry, tunnel slope, ventilation velocity, sidewalls restriction, and pressure of passing air among other influential variables. Once a fire occurs in a road tunnel; inhaling the distributed smoke throughout the tunnel is more dangerous issue than direct contact of individuals to the fire (Hu et al., 2005). Therefore, optimized control of smoke distribution is one of the crucial studies in design of ventilation systems, and it is indispensable to comprehend the characteristics of smoke distribution in order to reach successful design. When a fire occurs in the tunnel, the longitudinal ventilation system turns on in order to prevent back-layering flow of smoke and provide a safe region for people to escape. The low ventilation velocity allows combustion products to have counterflow movement. The magnitude of Critical Ventilation Velocity (CVV) and fire Heat Release Rate (HRR) are two substantial initial conditions required for designing longitudinal ventilation system in tunnels. Thomas (1968) depicted an analytical relation between original ventilation velocity and back-layering flow based on a comparison between buoyant force of the fire plume and inertia force of ⇑ Corresponding author. E-mail address: [email protected] (G. Heidarinejad). http://dx.doi.org/10.1016/j.tust.2016.06.016 0886-7798/Ó 2016 Elsevier Ltd. All rights reserved.
ventilation velocity. This method was studied by numerous investigators. Oka and Atkinson (1995) studied smoke movement in a real tunnel which was modeled by 1:10 scale and observed that critical velocity behavior in the modeled tunnel is similar to real one. They investigated the effect of various parameters such as tunnel shape, size, and obstruction on CVV. They found that when 12% of cross section of tunnel is occupied with the obstruction, the ratio of (VCr, Unoccupied-VCr, Occupied)/VCr, Unoccupied is about 15% and when 40–45% of tunnel is occupied this ratio would increase to 32%. Wu and Bakar (2000) studied the effect of cross section geometry of a small scale tunnel on CVV. They proposed that in order to apply the effect of geometry of the cross section, it is appropriate to use hydraulic height of tunnel in CVV relation. Modic (2003) considered the effect of tunnel slope in the relation which was presented by Thomas (1968). Also, experimental studies were carried out in a 1:10 reduced-scale modeled tunnel to study the influence of slope on CVV (Yi et al., 2014). In this study the experimental results agreed well with previous studies. Hwang and Edwards (2005) investigated ventilation velocity in the fire plume and jet region, close to the ceiling, and concluded that CVV has a direct proportion to Q1/5 (Q is HRR) and this achievement was based on Quintiere (1989) analysis. Quintiere studied fluid flow in the tunnel and stated that cross section geometry of tunnel has influence on CVV. When this geometry changes, the fire dynamic would change as well. Gao et al. (2014) investigated the influence of sidewall restriction on the maximum ceiling gas temperature in a
92
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
Nomenclature Cp d D g hs L _ 000 m b p q_ q_ 000 q_ 00 q_ 00r R t
constant pressure specific heat (m2 s1) distance between two fires (m) diffusion coefficient acceleration of gravity (m s2) sensitive enthalpy (kJ m3) distance between vehicles and fire mass production rate of species by evaporating droplets/particles (kg m3 s1) pressure (kg m1 s2) heat release rate (kW) heat release rate per unit volume (kW m3) heat flux vector (kW m2) radiative flux to a solid surface (kW m2) universal gas constant (J K1 mol1) time (s)
buoyancy-driven thermal flow experimentally. They concluded that decreasing the distance of fire source to the walls does not have significant effect on the maximum temperature when fire is far from the walls, but next to the wall the maximum temperature soars drastically. Finally, they proposed a new correlation for the maximum ceiling gas temperature. Ji et al. (2012) set an experimental model to study the effect of various transverse fire locations on maximum smoke temperature under the tunnel ceiling. They also stated that due to the restriction effect of the sidewalls of tunnel the maximum smoke temperature goes up compared with unconfined space. Gao et al. (2015) implemented 48 experiments with four experimental setups to study the details of flame shape and flame length under the ceiling of a channel and concluded that the flame shape is a function of HRR and flame position. Li et al. (2010) investigated the effect of the presence of the accident vehicles obstructions on CVV and length of backlayering in which the obstacles occupied 20% of cross section of the tunnel. They presented that due to presence of the vehicles, CVV reduced about 23%. Tsai et al. (2010) evaluated the value of CVV with two heat sources in a tunnel numerically and experimentally. They depicted that CVV descends when the distance between two sources is far significantly and in this case, the ventilation velocity must be calculated only by considering the more powerful source. Since smoke flow of the more powerful source has dominant power with respect to less powerful one, the back-layering effect of small source becomes comparatively negligible. Lee and Tsai (2012) studied the effect of vehicles obstruction on CVV and length of back-layering experimentally. They applied Different arrangement of vehicles in which 3–31% of the cross section was occupied and stated that in the case that longitudinal ventilation flow is in the direct access to fire, CVV reduces. Whereas, in the case that vehicles arrangement is placed in the way which prevents direct ventilation flow to the heat source, it leads to soar HRR and consequently increase of CVV. The influence of an obstacle blockage according to its location compared with the tunnel floor on the back-layering flow behavior and the critical velocity is performed numerically by using FDS (Gannouni and Maad, 2015). This study depicted that the effect of obstacle blockage brings about a decrease of CVV compared to those obtained with an empty tunnel. Previous studies show that the two passes road tunnels which were plugged by traffic jam and car accident causes fire adventure (Ingason, 2008). Mapar et al. (2013) investigated the effect of fire longitudinal location on CVV. They observed that increase of distance between fire and entrance of the tunnel leads to decrease in CVV.
T u V Cr Wa X Ya Z
temperature (k) velocity vector (m s1) critical ventilation velocity (m s1) molecular weight of gas species (kg mol1) longitudinal direction of tunnel (m) mass fraction of species a heightened-direction of tunnel (m)
Greek symbols q0 initial air density (kg m3) sij viscous stress tensor (kg m1 s2) e dissipation rate (kg m1 s3) H total pressure divided by density (m2 s2) p pressure perturbation (kg m1 s2) x vorticity vector (s1)
Based on the earlier investigations, the state which two fire sources in the presence of multi vehicles phenomenon has not been evaluated. Nevertheless, the history of car accident in the tunnel such as Mont Blanc (Henke and Gagliardi, 2004) and Gotthard (Vuilleumier et al., 2002) suggests that usually the number of fire has been more than one and other vehicles are in the tunnel during accident. Therefore, in the present study to increase human safety level that passes through the tunnel, CVV with Different arrangement of vehicles and diverse sizes in the upstream of two fire sources is investigated. In order to simulate this phenomenon, this study uses Fire Dynamics Simulator (FDS) code, version 6. Experimental results from Lee and Tsai (2012) allow this study to validate the numerical simulation results with experimental measurements. Moreover, the effects of distance between obstruction from the fire and also the distance between two fire sources on the critical ventilation velocity have been evaluated.
2. Physics of tunnel fire There are several stages in developing a fire scenario including ignition, spread, flashover, fully developed fire, and decay. Since the heat release rate of a fully developed fire is in its maximum value, the most challenging state for firefighters to control situation is when fire is fully developed (Pasdarshahri et al., 2013). In the present study, this scenario, i.e. the fully developed fire in a tunnel is studied. A schematic of fire flow is shown in Fig. 1. When a fire occurs in a tunnel due to buoyant force, the combustion products move up and reach to the tunnel ceiling. These hot gases initiate to distribute along the longitudinal sides of the tunnel and consequently great volume of smoke and heat disperse throughout the tunnel. The presence of smoke particles provides a region which not only has got insufficient oxygen but also reduces the magnitude of individual’s sight for escaping. One of the best scenarios for escaping people in such venturesome situations is the time which longitudinal ventilation system is considered in tunnel. In other words, longitudinal ventilation causes an air flow in this direction. In this case, fresh air blows to the fire from one side of tunnel and therefore it prevents distribution of soot and combustion products to the upstream part of the fire. In addition, using longitudinal ventilation causes temperature and concentration of smoke to decline and prevents back-layering phenomenon; consequently, it enables the individuals to move away more convenient. CVV is a minimum ventilation velocity which ceases smoke backlayering phenomenon. Moreover, the back-layering length, showed in Fig. 1, becomes zero on account of this ventilation velocity.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
93
Fig. 1. A schematic of flow of soot in the tunnel in presence of longitudinal ventilation.
Finding a precise value for CVV and applying it in the ventilation system will stop smoke back-layering with minimum energy consumption. 3. Dimensionless Froude number Thomas proposed that flow form is dependent on the ratio of inertia force to buoyant force in a specific cross sectional area of the tunnel, suggesting to define this ratio as the Froude number (Thomas, 1968):
Fr ¼
V2 T gH DT
ð1Þ
where g is gravity force, V and H are specific values of velocity and length scale, respectively. T is the average temperature of hot layer which computed as:
T¼
q_
q0 C p AV
þ T0
ð2Þ
In the case which an obstruction in tunnel is created by vehicles, it is indispensable to use hydraulic diameter and wetted area in Eqs. (1) and (2) (Floyd et al., 2012a). Furthermore, Thomas stated that for vanishing smoke backflow the Froude number should be equal to 1. In such a case, the critical ventilation velocity is proposed to be calculated as follow (Thomas, 1968):
V cr ¼
_ g qH
1=3
q0 C p TA
ð3Þ
In addition, Thomas introduced the back-layering length as a function of Froude number. Thomas discussed that the backlayering length in a tunnel with longitudinal ventilation system could be expressed as (Thomas, 1958):
l ¼
L gHq_ H q0 C p TAV 3
ð4Þ
4. Numerical simulation FDS is developed for studying fire behavior and performance of fire extinguishing systems by National Institute of Standard and Technology (NIST). This software package solves Navier Stokes equations in low Mach number flow using Large Eddy Simulation (LES) or Direct Numerical Simulation (DNS) methods (Floyd et al., 2012b). Governing equations in fluid flow is introduced as (Floyd et al., 2012b): Mass conservation equation:
@q þ r qu ¼ 0 @t
_ 000 Here m a is rate of a creation which introduced as a source term in Eq. (6), Da is diffusion coefficient of species a, and Ya is mass fraction of species a. Momentum conservation equation:
@u r 1 ¼ 1 ½ðq q Þg þ r sij u x þ rH P n @t q q
H are vorticity where sij is tension tensor for Newtonian fluids, x; P; vector, pressure perturbation and total pressure divided by density respectively. Energy conservation equation:
@ DP ðqhs Þ þ r ðqhs uÞ ¼ þ q_ 00 þ e @t Dt
Pm ðz; tÞ ¼ qRT
X Ya a Wa
ð9Þ
5. Geometry The effect of obstructive vehicles arrangement in the case which a fire occurs in a tunnel is studied by Lee and Tsai (2012) experimentally. In this experimental study a small tunnel has been utilized (Fig. 2). It is observed that the utilized model has 0.6 m 0.6 m square cross section area in which these proportions are 1/15 of a practical tunnel (Lee and Tsai, 2012). In the simulation of mentioned geometry, the inlet velocity and pressure outlet boundary conditions have been used respectively in entrance and output of the tunnel. When the velocity at the entrance of tunnel is specified, FDS uses the velocity inlet boundary condition. Moreover, walls are insulated ð@T=@n ¼ 0Þ and no slip boundary condition is conducted on the walls. It is worth noticing that the value of yþ on the walls is changed in the range of 10–50. In FDS, wall function is computed through Werner and Wengle model (1993). A uniform air flow by 20 °C temperature inters from left side of the tunnel. For modeling fire the propane is used as the fuel.
ð5Þ
ð6Þ
ð8Þ
where hs is the sensible enthalpy of fluid which is a function of temperature value. Moreover, e is dissipation term in energy equation, q_ 000 is HRR in a volume by chemical reaction, and q_ 00 denotes conductive and radiative heat transfer. Finally, ideal gas equation of state is introduced as:
Species conservation equation in the general form:
@ _ 000 ðqY a Þ þ r ðqY a uÞ ¼ r ðqDa rY a Þ þ m a @t
ð7Þ
Fig. 2. A schematic view of the considered geometry.
94
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
through a vertical line which is located at one meter distance from the fire in the downstream is depicted for different grid sizes and 9.5 kW heat power in Fig. 3. It is observed that changing the number of computational grid from 92,000–161,280 cells, the differences between velocity profiles become comparatively insignificant. Conversely, the length scale for Q_ ¼ 9:5 kW is in order of 15 cm. Therefore, due to Eq.
Fig. 3. Study of grid independency for velocity changes in vertical line, x = 3.5 m.
dY s Y e ðtÞ Y s ðtÞ ¼ L=V dt
Table 1 The dimension of vehicles in the simulation.
Small vehicle Medium vehicle
Long (m)
Width (m)
Height (m)
0.24 0.61
0.09 0.13
0.07 0.22
An appropriate length scale for solving a problem including plume and boundary flow as a function of heat power could be introduced as (Floyd et al., 2012b):
D ¼
!0:4 Q_ pffiffiffi q1 T 1 C P g
D < 16 dx
ð12Þ
where L denotes length characteristic of the detector geometry, V is free stream velocity, Ys is mass fraction of smoke, and Ye is mass fraction of external free stream. Therefore, in each step, this study solves the problem through the time to obtain a steady value of smoke mass fraction.
5.1. Results validation
ð10Þ
This length scale evaluates an effective region next to fire which is influenced by any change in fire behavior. Floyd et al. (2012b) stated that, when the proportion length scale to grid size follows Eq. (11), the values of large scale, which are obtained from direct simulation of equations, are truly solved. In addition, it is important to notice that this is an evaluation of computational grid size and the procedure to reach to independent results requires to satisfy the following condition:
4<
(11) it is necessary that the grid size considered in the range 1 < dx < 4 cm. Consequently, choosing a uniform 3 3 3 centimeter grid including 92,000 cells is a proper grid to simulate the model. It is worth mentioning that in FDS simulation, a smoke detector has been considered above the fire location. Therefore, the ventilation velocity was increased as far as the sensor detects no smoke. This procedure has been done in several steps to find out the exact CVV. Moreover, the mentioned detector computes the smoke density using ensuing correlation:
ð11Þ
The effect of changing grid proportions on temperature, velocity, and species concentration profiles in different points of the computational field is investigated. The velocity distribution
Lee and Tsai (2012) consider two different sizes of obstacles, showed in Table 1, as small and medium vehicles, arranged in two and three rows (according to Fig. 4a–d). It is worth mentioning that the obstacles dimensions are chosen in 1:15 scale of actual vehicles (Lee and Tsai, 2012). Lee and Tsai (2012) defined the blockage percentage, specified in Eq. (13), as a ratio of unoccupied area to total area. Table 2
Table 2 The percentage of blockage for Different arrangement of vehicles. Arrangement type
Blockage (%)
Small vehicles in two arrays (case a) Medium vehicles in two arrays (case b) Small vehicles in three arrays (case c) Medium vehicles in three arrays (case d)
3.5 16 5 24
Fig. 4. Different vehicles arrangements: case (a) small vehicles in two arrays, case (b) medium vehicles in two arrays, case (c) small vehicles in three arrays, case (d) medium vehicles in three arrays.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
95
summarizes the blockage percentage for different arrangement of vehicles illustrated in Fig. 4a–d.
Blockage ð%Þ ¼
A ATotal Blocks A
ð13Þ
where A is the tunnel cross area, and ATotal Blocks is the occupied area via obstacles in the tunnel. It is seen that the diagrams trend are the same as experimental results (Figs. 5 and 6). Generally, the comparison between numerical and experimental results demonstrates that the CFD results overestimate critical velocity at about 15% more than the experimental measurements. 6. Results and discussions After assessing the validation of results and procedure, the magnitude of CVV for different scenarios, which have been shown in Fig. 7a and b, is surveyed. The variation of CVV versus alteration of the distance between two fires in two directions, which are introduced in Fig. 7a and b, have been represented in Fig. 8a and b, respectively. As it is shown in Fig. 8a, the increase of distance between two fires leads to decrease in the magnitude of CVV. While two fires are adjacent to each other, CVV becomes equal to 0.61 m s1. This result is the same as the state which a fire by 9 kW is burned in the tunnel. Also, when d = 60 cm CVV reaches to 0.52 m s1 which is equal to the state that a 4.5 kW fire is burning. Thus, it could be Fig. 7. A schematic of two simultaneous fires in the tunnel. (a) The distance between two fires increases in longitudinal direction. (b) The distance between two fires increases in lateral direction.
Fig. 5. A comparison between numerical results of the simulation and experimental results of Lee and Tsai (2012) in three arrays state.
Fig. 6. A comparison between numerical results of the simulation and experimental results of Lee and Tsai (2012) in two arrays state.
Fig. 8. The variation of CVV for different distances between two fires with the same HRR = 4.5 kW. (a) The distance between two fires in longitudinal direction is increasing. (b) The distance between two fires in lateral direction is increasing.
96
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
stated that whenever two fire sources locate adjacent to each other the ventilation velocity is equal to a fire by twice magnitude of HRR. Furthermore, increasing the distance between two fires tends CVV value to decrease. As the distance increases sufficiently, the fire plumes force does not have any effect on each other. Hence, the ventilation velocity must prevent back-layering flow from the first fire. This result is strictly similar to Tsai et al. (2010) investigation. In contrast, in accordance with Fig. 8b, when two fires are completely placed next to each other, it is shown that CVV value becomes equal to 0.61 m s1. An increase in the lateral distance between two fires leads to reduce CVV magnitude by 0.55 m s1 and it follows by a stationary value. Nevertheless, as two fires approach the walls, because the inertia force has to overcome both buoyant force and the shear stress of the walls simultaneously, CVV soars 20% with respect to 0.55 m s1. In this context, for better comprehension, a temperature contour of two fire plumes is demonstrated in Fig. 9. Fig. 9a remarks that when there is no space between two fires (d = 0), the two fire plumes intensify each other. In this case, critical ventilation velocity needs to be 0.61 m s1 to overcome buoyant force. In accordance with Fig. 9c, when the distance goes up adequately (d = 0.6 m), the two fire plumes separate completely and do not have any influence on each other and consequently, the entrance velocity of the tunnel dwindles to 0.52 m s1 to just overcome buoyant force of the first fire. To investigate the effect of obstruction on CVV in the state that two fires burn in the tunnel simultaneously, the Different arrangement of vehicles which demonstrated in Fig. 4, is studied. Moreover, for the purpose of a comprehensive investigation, the two fires move far away from each other in longitudinal and lateral directions respectively (Fig. 7a and b). 6.1. Variation of fire locations in longitudinal direction The variation of CVV in various distances between fires and vehicles for a state which d = 60 cm is illustrated in Fig. 10. As mentioned in the previous section, when the distance between two fires is 60 cm, the two plumes do not have any effect on each other. Parameter / for each four states in Table 3, is introduced to provide additional quantitative insights:
Fig. 10. Alternation of CVV for different distances between vehicles and fires in all arrangements of obstructions, d = 60 cm.
Table 3 The percentage of difference of CVV in proportion of occupied tunnel to unoccupied tunnel, d = 60 cm. L (cm)
5 10 15 20
/ (%) Case a
Case b
Case c
Case d
0 0 0 0
7 2 0 0
1 0 0 0
11 4 0 0
The minus sign shows that CVV is decreased rather than CVV in unoccupied tunnel.
/¼
V 0cr V cr 100 V 0cr
ð14Þ
where V 0cr is CVV in the unoccupied tunnel and Vcr is CVV in the occupied tunnel. In accordance with Fig. 10 and Table 3 it is concluded that: In cases a and c, i.e. while the small vehicles are respectively in two and three rows with 3.5% and 5% blockages, located in front of two fires, CVV is equal to unoccupied state for all L distances because this blockage percentage does not prevent the access of
Fig. 9. Contour of temperature of simultaneous two fires; (a) d = 0 m, Vcr = 0.61 m s1; (b) d = 0.3 m, Vcr = 0.59 m s1; (c) d = 0.6 m, Vcr = 0.52 m s1.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
Fig. 11. The variation of Froude number versus percentage of blockage.
ventilation flow to the fire plume. These results are similar to that of empirical study have been accomplished by Lee and Tsai (2012).
97
In case b, two rows of medium vehicles by 16% blockage, when the distance between fires and vehicles (L) is trivial, the ventilation velocity declines. As L = 5 cm this velocity is 7% less than unoccupied state. While L is great enough, the effect of obstacle on the fire behavior is vanished and Vcr is equal to unoccupied state. Lee and Tsai (2012) have reached to the same result in investigating the presence of two rows of medium vehicles. Comparing the inertia and buoyancy forces could interpret obtained results. Therefore, the Froude number variation at different blockage percentages of arrangement is shown in Fig. 11. It could be seen that increasing blockage percentage leads Froude number to augment. In other words, increase of blockage tends to empower inertia force of wind flow with respecting the buoyant force of the fire plume in the cross section. In case d, three rows of medium vehicles by 24% blockage, not only the Froude number is greatened dramatically but also a vortex structure would be generated in the fire upstream. It is shown in Fig. 12a–d that while distance between vehicles and fires is small; a part of the first fire plume energy is encompassed in generated eddy. Consequently, it could be said that CVV reduces noticeably. According to Table 3 this reduction is about 11% in the case which L = 5 cm. While L increases and fire plume gets out of the turbulence region, Vcr soars remarkably. The turbulence region is more clarified in Fig. 13.
Fig. 12. The effect of blockage of 9 vehicles in medium size which are arranged in three rows, d = 60 cm; (a) L = 5 cm, Vcr = 0.45 m s1; (b) L = 10 cm, Vcr = 0.5 m s1; (c) L = 15 cm, Vcr = 0.52 m s1; (d) L = 20 cm, Vcr = 0.52 m s1.
Fig. 13. Generated turbulent region by medium-size vehicles, L = 20 cm.
98
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
Studying the effect of case d on one fire presented that Vcr increases about 20% with respect to unoccupied state (Mapar et al., 2013). Whereas, in the present paper 24% blockage in the presence of two fires cause a decrease in Vcr with respect to unoccupied state. The effect of obstruction via Different arrangement of vehicles on CVV in the state that the distance between fires is 60 cm, is evaluated for different values of L (Fig. 14). Comparing the results for all four arrangements of vehicles in presence of two fires with d = 60 cm, the parameter / is shown in Table 4. Based on Fig. 14 and Table 4, it could be concluded that: Presence of obstacle reduces Vcr. The reason of this reduction could be found in dimensionless Froude number. According to Fig. 11 existence of obstacles in the tunnel causes augmentation in inertia force rather than buoyant force of fire. In this state increasing the distance of vehicles leads Vcr to growth and reach to unoccupied state as well. Once three rows of vehicles are located in front of fires, indeed the access of entrance air to the fire plume reduces owing to existence of middle row and this issue causes Vcr to be plunged as the distance between vehicles and the fire is trivial. 6.2. Variation of fire locations in lateral direction In this case the effect of blockage, while L is increasing, on CVV for the most critical state, i.e. two fires stick to the walls, is represented in Fig. 15. Fig. 15 shows that the effect of blockage on CVV is the same as longitudinal state. It means that when the set of fires approaches the vehicles due to inertia force augmentation in fire location, CVV declines significantly. Whereas, as L increases, CVV magnitude approaches to unoccupied state. For instance, in case d, when L is equal to 5 cm, the critical ventilation velocity tends to be
Fig. 14. Alteration of CVV for different distances between vehicles and fires in all arrangements of obstructions, d = 30 cm.
Table 4 The percentage of difference of CVV in proportion of occupied tunnel to unoccupied tunnel, d = 30 cm. L (cm)
5 10 15 20
/ (%) Case a
Case b
Case c
Case d
0 0 0 0
6 6 6 0
7 7 5 3
18 16 11 3
The minus sign shows that CVV is decreased rather than CVV in unoccupied tunnel.
Fig. 15. Alteration of CVV for different distances between vehicles and fires in all arrangements of obstructions, two fires stick to the walls.
0.55 m s1 to overcome inertia force; however, when L is 20 cm, it reaches to the unoccupied state value (Vcr = 0.66 m s1). It is worth mentioning that, owing to the effect of occupied blockage on the inertia force, the value of critical velocity in cases a and c is greater than cases b and d, respectively. Furthermore, the variation of Vcr for cases b and d which have larger blockage percentage (Case b = 16% and case d = 24%) is more significant than cases a and c, respectively.
7. Conclusions In the present investigation, the behavior of fire in a road tunnel and the effect of obstacle presence at the upstream of two fire sources are studied. Comparing the present numerical results with experimental ones indicates the numerical results are within a good agreement of the experimental results. The results indicate in both longitudinal and lateral directions, the Critical Ventilation Velocity (CVV) alters when the distance between two fires increases. Moreover, in longitudinal state, as the distance between fire sources increases: (1) the effect of two fire plumes on each other vanishes and (2) critical ventilation velocity in the occupied tunnel (Vcr) becomes equal to the state which only a fire exists in the tunnel. In lateral state, when two fires stick to the walls CVV soars remarkably. When the obstacles occupy the tunnel Vcr decreases because of increase in inertia force of air with respect to buoyant force. When the obstacles are far from the fire, the value of critical ventilation velocity approaches to the state that tunnel is devoid of obstacles. Furthermore, it is observed that existence of small vehicles does not have any effect on CVV. As the percentage of blockage increases, the critical ventilation velocity decreases. Overall, it could be deduced that while this percentage augments, the existence of a vortex structure in front of vehicles leads to encompass a part of fire plume energy and consequently, diminish the value of critical ventilation velocity.
References Floyd, J., Forney, G., Hostikka, S., Korhonen, T., McDermott, R., McGrattan, K.B, 2012a. Fire Dynamics Simulator User’s Guide. NIST Special Publication 1019. Floyd, J., Forney, G., Hostikka, S., Korhonen, T., McDermott, R., McGrattan, K.B., 2012b. Fire Dynamics Simulator Technical Reference Guide, Volume 1: Mathematical Model. NIST Special Publication 1018. Gannouni, S., Maad, R.B., 2015. Numerical study of the effect of blockage on critical velocity and backlayering length in longitudinally ventilated tunnel fires. Tunn. Undergr. Sp. Technol. 48, 147–155. Gao, Z.H., Ji, J., Fan, C.G., Sun, J.H., Zhu, J.P., 2014. Influence of sidewall restriction on maximum ceiling gas temperature of buoyancy-driven thermal flow. Energy Build. 84, 13–20.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99 Gao, Z.H., Ji, J., Wan, H., Li, K., Sun, J.H., 2015. An investigation of the detailed flame shape and flame length under the ceiling of a channel. Proc. Combust. Inst. 35, 2657–2664. Henke, A., Gagliardi, M., 2004. The 2001 Gotthard road tunnel fire. Tunn. Manage. Int. 7 (1), 33–39. Hu, L.H., Huo, R., Li, Y.Z., Wang, H.B., Chow, W.K., 2005. Full-scale burning tests on studying smoke temperature and velocity along a corridor. Tunn. Undergr. Sp. Technol. 20, 223–229. Hwang, C.C., Edwards, J.C., 2005. The critical ventilation velocity in tunnel fires – a computer simulation. Fire Saf. J. 40 (3), 213–244. Ingason, H., 2008. State of the art of tunnel fire research. In: Proceeding of the Ninth IAFSS Symposium, Germany. Ji, J., Fan, C.G., Zhong, W., Shen, X.B., Sun, J.H., 2012. Experimental investigation on influence of different transverse fire locations on maximum smoke temperature under the tunnel ceiling. Int. J. Heat Mass Transf. 55, 4817–4826. Lee, Y.P., Tsai, K.C., 2012. Effect of vehicular blockage on critical ventilation velocity and tunnel fire behavior in longitudinally ventilated tunnels. Fire Saf. J. 53, 35– 42. Li, Y.Z., Lei, B., Ingason, H., 2010. Study of critical velocity and backlayering length in longitudinally ventilated tunnel fires. Fire Saf. J. 45 (6–8), 361–370. Mapar, M., Heidarinejad, G., Zargartalebi, H., 2013. Numerical investigation of fire in tunnel and developing an analytical equation for critical velocity. In: 21th Annual International Conference on Mechanical Engineering, Faculty of Mechanical Engineering K.N. Toosi University of Technology, 7–9 May (in Persian).
99
Modic, J., 2003. Fire simulation in road tunnels. Tunn. Undergr. Sp. Technol. 18 (5), 525–530. Oka, Y., Atkinson, G.T., 1995. Control of smoke flow in tunnel fires. Fire Saf. J. 25 (4), 305–322. Pasdarshahri, H., Heidarinejad, G., Mazaheri, K., 2013. Evaluation of induced-flow in a two-room compartment fire using large eddy simulation. J. Modares Mech. Eng. 13 (4), 74–85 (in Persian). Quintiere, J.G., 1989. Scaling applications in fire research. Fire Saf. J. 15 (1), 3–29. Thomas, P.H., 1958. The Movement of Buoyant Flow Against a Stream and Venting of Underground Fire. Fire Research Station Note No. 351. Thomas, P.H., 1968. The Movement of Smoke in Horizontal Passages Against an Air Flow. Fire Research Station Note No. 723. Tsai, K.C., Chen, H.H., Lee, S.K., 2010. Critical ventilation velocity for multi-source tunnel fires. J. Wind Eng. Ind. Aerodyn. 98 (10–11), 650–660. Vuilleumier, F., Weatherill, A., Crausaz, B., 2002. Safety aspects of railway and road tunnel: example of the Lotschberg railway tunnel and Mont-Blanc road tunnel. Tunn. Undergr. Sp. Technol. 17, 153–158. Werner, H., Wengle, H., 1993. Large-eddy simulation of turbulent flow over and around a cube in a plate channel. Turb. Shear Flows 8, 155–168. Wu, Y., Bakar, M.Z.A., 2000. Control of smoke flow in tunnel fires using longitudinal ventilation systems – a study of the critical velocity. Fire Saf. J. 35 (4), 363–390. Yi, L., Xu, Q., Xu, Z., Wu, D., 2014. An experimental study on critical velocity in sloping tunnel with longitudinal ventilation under fire. Tunn. Undergr. Sp. Technol. 43, 198–203.
Contents lists available at ScienceDirect
Tunnelling and Underground Space Technology journal homepage: www.elsevier.com/locate/tust
A comprehensive study of two fire sources in a road tunnel: Considering different arrangement of obstacles Ghassem Heidarinejad ⇑, Maryam Mapar, Hadi Pasdarshahri Faculty of Mechanical Engineering, Tarbiat Modares University, PO Box 14115-143, Tehran, Iran
a r t i c l e
i n f o
Article history: Received 26 October 2015 Received in revised form 3 June 2016 Accepted 4 June 2016
Keywords: Tunnel Fire LES Critical ventilation velocity
a b s t r a c t This paper uses Fire Dynamics Simulator (FDS) to study various arrangements of different vehicles at upstream of two fire sources. In order to make a comprehensive study, the effects of two fire sources in both lateral and longitudinal directions are investigated. The results reveal that the behavior of two fire sources, in both perpendicular directions, is directly influenced by distance between them. For small vehicles, variations of the arrangement and distance between the vehicles and fire sources do not affect the calculated Critical Ventilation Velocity (CVV). However, the presence of medium vehicles strengthens the influence of inertia force rather than buoyant force of fire plume in the tunnel. Accordingly, when there is a short distance between fires and medium obstructions, less air ventilation is needed to prevent smoke back-layering. Eventually, far distance between the vehicles and the fires results in vanishing obstruction effects. Consequently, CVV is the same as the case in which there is no vehicle in the tunnel. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Generally, fire in a road tunnel forms a complicated structure. This physical phenomenon which involves chemical reaction, turbulence, and radiation is affected by various parameters, including geometry, tunnel slope, ventilation velocity, sidewalls restriction, and pressure of passing air among other influential variables. Once a fire occurs in a road tunnel; inhaling the distributed smoke throughout the tunnel is more dangerous issue than direct contact of individuals to the fire (Hu et al., 2005). Therefore, optimized control of smoke distribution is one of the crucial studies in design of ventilation systems, and it is indispensable to comprehend the characteristics of smoke distribution in order to reach successful design. When a fire occurs in the tunnel, the longitudinal ventilation system turns on in order to prevent back-layering flow of smoke and provide a safe region for people to escape. The low ventilation velocity allows combustion products to have counterflow movement. The magnitude of Critical Ventilation Velocity (CVV) and fire Heat Release Rate (HRR) are two substantial initial conditions required for designing longitudinal ventilation system in tunnels. Thomas (1968) depicted an analytical relation between original ventilation velocity and back-layering flow based on a comparison between buoyant force of the fire plume and inertia force of ⇑ Corresponding author. E-mail address: [email protected] (G. Heidarinejad). http://dx.doi.org/10.1016/j.tust.2016.06.016 0886-7798/Ó 2016 Elsevier Ltd. All rights reserved.
ventilation velocity. This method was studied by numerous investigators. Oka and Atkinson (1995) studied smoke movement in a real tunnel which was modeled by 1:10 scale and observed that critical velocity behavior in the modeled tunnel is similar to real one. They investigated the effect of various parameters such as tunnel shape, size, and obstruction on CVV. They found that when 12% of cross section of tunnel is occupied with the obstruction, the ratio of (VCr, Unoccupied-VCr, Occupied)/VCr, Unoccupied is about 15% and when 40–45% of tunnel is occupied this ratio would increase to 32%. Wu and Bakar (2000) studied the effect of cross section geometry of a small scale tunnel on CVV. They proposed that in order to apply the effect of geometry of the cross section, it is appropriate to use hydraulic height of tunnel in CVV relation. Modic (2003) considered the effect of tunnel slope in the relation which was presented by Thomas (1968). Also, experimental studies were carried out in a 1:10 reduced-scale modeled tunnel to study the influence of slope on CVV (Yi et al., 2014). In this study the experimental results agreed well with previous studies. Hwang and Edwards (2005) investigated ventilation velocity in the fire plume and jet region, close to the ceiling, and concluded that CVV has a direct proportion to Q1/5 (Q is HRR) and this achievement was based on Quintiere (1989) analysis. Quintiere studied fluid flow in the tunnel and stated that cross section geometry of tunnel has influence on CVV. When this geometry changes, the fire dynamic would change as well. Gao et al. (2014) investigated the influence of sidewall restriction on the maximum ceiling gas temperature in a
92
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
Nomenclature Cp d D g hs L _ 000 m b p q_ q_ 000 q_ 00 q_ 00r R t
constant pressure specific heat (m2 s1) distance between two fires (m) diffusion coefficient acceleration of gravity (m s2) sensitive enthalpy (kJ m3) distance between vehicles and fire mass production rate of species by evaporating droplets/particles (kg m3 s1) pressure (kg m1 s2) heat release rate (kW) heat release rate per unit volume (kW m3) heat flux vector (kW m2) radiative flux to a solid surface (kW m2) universal gas constant (J K1 mol1) time (s)
buoyancy-driven thermal flow experimentally. They concluded that decreasing the distance of fire source to the walls does not have significant effect on the maximum temperature when fire is far from the walls, but next to the wall the maximum temperature soars drastically. Finally, they proposed a new correlation for the maximum ceiling gas temperature. Ji et al. (2012) set an experimental model to study the effect of various transverse fire locations on maximum smoke temperature under the tunnel ceiling. They also stated that due to the restriction effect of the sidewalls of tunnel the maximum smoke temperature goes up compared with unconfined space. Gao et al. (2015) implemented 48 experiments with four experimental setups to study the details of flame shape and flame length under the ceiling of a channel and concluded that the flame shape is a function of HRR and flame position. Li et al. (2010) investigated the effect of the presence of the accident vehicles obstructions on CVV and length of backlayering in which the obstacles occupied 20% of cross section of the tunnel. They presented that due to presence of the vehicles, CVV reduced about 23%. Tsai et al. (2010) evaluated the value of CVV with two heat sources in a tunnel numerically and experimentally. They depicted that CVV descends when the distance between two sources is far significantly and in this case, the ventilation velocity must be calculated only by considering the more powerful source. Since smoke flow of the more powerful source has dominant power with respect to less powerful one, the back-layering effect of small source becomes comparatively negligible. Lee and Tsai (2012) studied the effect of vehicles obstruction on CVV and length of back-layering experimentally. They applied Different arrangement of vehicles in which 3–31% of the cross section was occupied and stated that in the case that longitudinal ventilation flow is in the direct access to fire, CVV reduces. Whereas, in the case that vehicles arrangement is placed in the way which prevents direct ventilation flow to the heat source, it leads to soar HRR and consequently increase of CVV. The influence of an obstacle blockage according to its location compared with the tunnel floor on the back-layering flow behavior and the critical velocity is performed numerically by using FDS (Gannouni and Maad, 2015). This study depicted that the effect of obstacle blockage brings about a decrease of CVV compared to those obtained with an empty tunnel. Previous studies show that the two passes road tunnels which were plugged by traffic jam and car accident causes fire adventure (Ingason, 2008). Mapar et al. (2013) investigated the effect of fire longitudinal location on CVV. They observed that increase of distance between fire and entrance of the tunnel leads to decrease in CVV.
T u V Cr Wa X Ya Z
temperature (k) velocity vector (m s1) critical ventilation velocity (m s1) molecular weight of gas species (kg mol1) longitudinal direction of tunnel (m) mass fraction of species a heightened-direction of tunnel (m)
Greek symbols q0 initial air density (kg m3) sij viscous stress tensor (kg m1 s2) e dissipation rate (kg m1 s3) H total pressure divided by density (m2 s2) p pressure perturbation (kg m1 s2) x vorticity vector (s1)
Based on the earlier investigations, the state which two fire sources in the presence of multi vehicles phenomenon has not been evaluated. Nevertheless, the history of car accident in the tunnel such as Mont Blanc (Henke and Gagliardi, 2004) and Gotthard (Vuilleumier et al., 2002) suggests that usually the number of fire has been more than one and other vehicles are in the tunnel during accident. Therefore, in the present study to increase human safety level that passes through the tunnel, CVV with Different arrangement of vehicles and diverse sizes in the upstream of two fire sources is investigated. In order to simulate this phenomenon, this study uses Fire Dynamics Simulator (FDS) code, version 6. Experimental results from Lee and Tsai (2012) allow this study to validate the numerical simulation results with experimental measurements. Moreover, the effects of distance between obstruction from the fire and also the distance between two fire sources on the critical ventilation velocity have been evaluated.
2. Physics of tunnel fire There are several stages in developing a fire scenario including ignition, spread, flashover, fully developed fire, and decay. Since the heat release rate of a fully developed fire is in its maximum value, the most challenging state for firefighters to control situation is when fire is fully developed (Pasdarshahri et al., 2013). In the present study, this scenario, i.e. the fully developed fire in a tunnel is studied. A schematic of fire flow is shown in Fig. 1. When a fire occurs in a tunnel due to buoyant force, the combustion products move up and reach to the tunnel ceiling. These hot gases initiate to distribute along the longitudinal sides of the tunnel and consequently great volume of smoke and heat disperse throughout the tunnel. The presence of smoke particles provides a region which not only has got insufficient oxygen but also reduces the magnitude of individual’s sight for escaping. One of the best scenarios for escaping people in such venturesome situations is the time which longitudinal ventilation system is considered in tunnel. In other words, longitudinal ventilation causes an air flow in this direction. In this case, fresh air blows to the fire from one side of tunnel and therefore it prevents distribution of soot and combustion products to the upstream part of the fire. In addition, using longitudinal ventilation causes temperature and concentration of smoke to decline and prevents back-layering phenomenon; consequently, it enables the individuals to move away more convenient. CVV is a minimum ventilation velocity which ceases smoke backlayering phenomenon. Moreover, the back-layering length, showed in Fig. 1, becomes zero on account of this ventilation velocity.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
93
Fig. 1. A schematic of flow of soot in the tunnel in presence of longitudinal ventilation.
Finding a precise value for CVV and applying it in the ventilation system will stop smoke back-layering with minimum energy consumption. 3. Dimensionless Froude number Thomas proposed that flow form is dependent on the ratio of inertia force to buoyant force in a specific cross sectional area of the tunnel, suggesting to define this ratio as the Froude number (Thomas, 1968):
Fr ¼
V2 T gH DT
ð1Þ
where g is gravity force, V and H are specific values of velocity and length scale, respectively. T is the average temperature of hot layer which computed as:
T¼
q_
q0 C p AV
þ T0
ð2Þ
In the case which an obstruction in tunnel is created by vehicles, it is indispensable to use hydraulic diameter and wetted area in Eqs. (1) and (2) (Floyd et al., 2012a). Furthermore, Thomas stated that for vanishing smoke backflow the Froude number should be equal to 1. In such a case, the critical ventilation velocity is proposed to be calculated as follow (Thomas, 1968):
V cr ¼
_ g qH
1=3
q0 C p TA
ð3Þ
In addition, Thomas introduced the back-layering length as a function of Froude number. Thomas discussed that the backlayering length in a tunnel with longitudinal ventilation system could be expressed as (Thomas, 1958):
l ¼
L gHq_ H q0 C p TAV 3
ð4Þ
4. Numerical simulation FDS is developed for studying fire behavior and performance of fire extinguishing systems by National Institute of Standard and Technology (NIST). This software package solves Navier Stokes equations in low Mach number flow using Large Eddy Simulation (LES) or Direct Numerical Simulation (DNS) methods (Floyd et al., 2012b). Governing equations in fluid flow is introduced as (Floyd et al., 2012b): Mass conservation equation:
@q þ r qu ¼ 0 @t
_ 000 Here m a is rate of a creation which introduced as a source term in Eq. (6), Da is diffusion coefficient of species a, and Ya is mass fraction of species a. Momentum conservation equation:
@u r 1 ¼ 1 ½ðq q Þg þ r sij u x þ rH P n @t q q
H are vorticity where sij is tension tensor for Newtonian fluids, x; P; vector, pressure perturbation and total pressure divided by density respectively. Energy conservation equation:
@ DP ðqhs Þ þ r ðqhs uÞ ¼ þ q_ 00 þ e @t Dt
Pm ðz; tÞ ¼ qRT
X Ya a Wa
ð9Þ
5. Geometry The effect of obstructive vehicles arrangement in the case which a fire occurs in a tunnel is studied by Lee and Tsai (2012) experimentally. In this experimental study a small tunnel has been utilized (Fig. 2). It is observed that the utilized model has 0.6 m 0.6 m square cross section area in which these proportions are 1/15 of a practical tunnel (Lee and Tsai, 2012). In the simulation of mentioned geometry, the inlet velocity and pressure outlet boundary conditions have been used respectively in entrance and output of the tunnel. When the velocity at the entrance of tunnel is specified, FDS uses the velocity inlet boundary condition. Moreover, walls are insulated ð@T=@n ¼ 0Þ and no slip boundary condition is conducted on the walls. It is worth noticing that the value of yþ on the walls is changed in the range of 10–50. In FDS, wall function is computed through Werner and Wengle model (1993). A uniform air flow by 20 °C temperature inters from left side of the tunnel. For modeling fire the propane is used as the fuel.
ð5Þ
ð6Þ
ð8Þ
where hs is the sensible enthalpy of fluid which is a function of temperature value. Moreover, e is dissipation term in energy equation, q_ 000 is HRR in a volume by chemical reaction, and q_ 00 denotes conductive and radiative heat transfer. Finally, ideal gas equation of state is introduced as:
Species conservation equation in the general form:
@ _ 000 ðqY a Þ þ r ðqY a uÞ ¼ r ðqDa rY a Þ þ m a @t
ð7Þ
Fig. 2. A schematic view of the considered geometry.
94
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
through a vertical line which is located at one meter distance from the fire in the downstream is depicted for different grid sizes and 9.5 kW heat power in Fig. 3. It is observed that changing the number of computational grid from 92,000–161,280 cells, the differences between velocity profiles become comparatively insignificant. Conversely, the length scale for Q_ ¼ 9:5 kW is in order of 15 cm. Therefore, due to Eq.
Fig. 3. Study of grid independency for velocity changes in vertical line, x = 3.5 m.
dY s Y e ðtÞ Y s ðtÞ ¼ L=V dt
Table 1 The dimension of vehicles in the simulation.
Small vehicle Medium vehicle
Long (m)
Width (m)
Height (m)
0.24 0.61
0.09 0.13
0.07 0.22
An appropriate length scale for solving a problem including plume and boundary flow as a function of heat power could be introduced as (Floyd et al., 2012b):
D ¼
!0:4 Q_ pffiffiffi q1 T 1 C P g
D < 16 dx
ð12Þ
where L denotes length characteristic of the detector geometry, V is free stream velocity, Ys is mass fraction of smoke, and Ye is mass fraction of external free stream. Therefore, in each step, this study solves the problem through the time to obtain a steady value of smoke mass fraction.
5.1. Results validation
ð10Þ
This length scale evaluates an effective region next to fire which is influenced by any change in fire behavior. Floyd et al. (2012b) stated that, when the proportion length scale to grid size follows Eq. (11), the values of large scale, which are obtained from direct simulation of equations, are truly solved. In addition, it is important to notice that this is an evaluation of computational grid size and the procedure to reach to independent results requires to satisfy the following condition:
4<
(11) it is necessary that the grid size considered in the range 1 < dx < 4 cm. Consequently, choosing a uniform 3 3 3 centimeter grid including 92,000 cells is a proper grid to simulate the model. It is worth mentioning that in FDS simulation, a smoke detector has been considered above the fire location. Therefore, the ventilation velocity was increased as far as the sensor detects no smoke. This procedure has been done in several steps to find out the exact CVV. Moreover, the mentioned detector computes the smoke density using ensuing correlation:
ð11Þ
The effect of changing grid proportions on temperature, velocity, and species concentration profiles in different points of the computational field is investigated. The velocity distribution
Lee and Tsai (2012) consider two different sizes of obstacles, showed in Table 1, as small and medium vehicles, arranged in two and three rows (according to Fig. 4a–d). It is worth mentioning that the obstacles dimensions are chosen in 1:15 scale of actual vehicles (Lee and Tsai, 2012). Lee and Tsai (2012) defined the blockage percentage, specified in Eq. (13), as a ratio of unoccupied area to total area. Table 2
Table 2 The percentage of blockage for Different arrangement of vehicles. Arrangement type
Blockage (%)
Small vehicles in two arrays (case a) Medium vehicles in two arrays (case b) Small vehicles in three arrays (case c) Medium vehicles in three arrays (case d)
3.5 16 5 24
Fig. 4. Different vehicles arrangements: case (a) small vehicles in two arrays, case (b) medium vehicles in two arrays, case (c) small vehicles in three arrays, case (d) medium vehicles in three arrays.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
95
summarizes the blockage percentage for different arrangement of vehicles illustrated in Fig. 4a–d.
Blockage ð%Þ ¼
A ATotal Blocks A
ð13Þ
where A is the tunnel cross area, and ATotal Blocks is the occupied area via obstacles in the tunnel. It is seen that the diagrams trend are the same as experimental results (Figs. 5 and 6). Generally, the comparison between numerical and experimental results demonstrates that the CFD results overestimate critical velocity at about 15% more than the experimental measurements. 6. Results and discussions After assessing the validation of results and procedure, the magnitude of CVV for different scenarios, which have been shown in Fig. 7a and b, is surveyed. The variation of CVV versus alteration of the distance between two fires in two directions, which are introduced in Fig. 7a and b, have been represented in Fig. 8a and b, respectively. As it is shown in Fig. 8a, the increase of distance between two fires leads to decrease in the magnitude of CVV. While two fires are adjacent to each other, CVV becomes equal to 0.61 m s1. This result is the same as the state which a fire by 9 kW is burned in the tunnel. Also, when d = 60 cm CVV reaches to 0.52 m s1 which is equal to the state that a 4.5 kW fire is burning. Thus, it could be Fig. 7. A schematic of two simultaneous fires in the tunnel. (a) The distance between two fires increases in longitudinal direction. (b) The distance between two fires increases in lateral direction.
Fig. 5. A comparison between numerical results of the simulation and experimental results of Lee and Tsai (2012) in three arrays state.
Fig. 6. A comparison between numerical results of the simulation and experimental results of Lee and Tsai (2012) in two arrays state.
Fig. 8. The variation of CVV for different distances between two fires with the same HRR = 4.5 kW. (a) The distance between two fires in longitudinal direction is increasing. (b) The distance between two fires in lateral direction is increasing.
96
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
stated that whenever two fire sources locate adjacent to each other the ventilation velocity is equal to a fire by twice magnitude of HRR. Furthermore, increasing the distance between two fires tends CVV value to decrease. As the distance increases sufficiently, the fire plumes force does not have any effect on each other. Hence, the ventilation velocity must prevent back-layering flow from the first fire. This result is strictly similar to Tsai et al. (2010) investigation. In contrast, in accordance with Fig. 8b, when two fires are completely placed next to each other, it is shown that CVV value becomes equal to 0.61 m s1. An increase in the lateral distance between two fires leads to reduce CVV magnitude by 0.55 m s1 and it follows by a stationary value. Nevertheless, as two fires approach the walls, because the inertia force has to overcome both buoyant force and the shear stress of the walls simultaneously, CVV soars 20% with respect to 0.55 m s1. In this context, for better comprehension, a temperature contour of two fire plumes is demonstrated in Fig. 9. Fig. 9a remarks that when there is no space between two fires (d = 0), the two fire plumes intensify each other. In this case, critical ventilation velocity needs to be 0.61 m s1 to overcome buoyant force. In accordance with Fig. 9c, when the distance goes up adequately (d = 0.6 m), the two fire plumes separate completely and do not have any influence on each other and consequently, the entrance velocity of the tunnel dwindles to 0.52 m s1 to just overcome buoyant force of the first fire. To investigate the effect of obstruction on CVV in the state that two fires burn in the tunnel simultaneously, the Different arrangement of vehicles which demonstrated in Fig. 4, is studied. Moreover, for the purpose of a comprehensive investigation, the two fires move far away from each other in longitudinal and lateral directions respectively (Fig. 7a and b). 6.1. Variation of fire locations in longitudinal direction The variation of CVV in various distances between fires and vehicles for a state which d = 60 cm is illustrated in Fig. 10. As mentioned in the previous section, when the distance between two fires is 60 cm, the two plumes do not have any effect on each other. Parameter / for each four states in Table 3, is introduced to provide additional quantitative insights:
Fig. 10. Alternation of CVV for different distances between vehicles and fires in all arrangements of obstructions, d = 60 cm.
Table 3 The percentage of difference of CVV in proportion of occupied tunnel to unoccupied tunnel, d = 60 cm. L (cm)
5 10 15 20
/ (%) Case a
Case b
Case c
Case d
0 0 0 0
7 2 0 0
1 0 0 0
11 4 0 0
The minus sign shows that CVV is decreased rather than CVV in unoccupied tunnel.
/¼
V 0cr V cr 100 V 0cr
ð14Þ
where V 0cr is CVV in the unoccupied tunnel and Vcr is CVV in the occupied tunnel. In accordance with Fig. 10 and Table 3 it is concluded that: In cases a and c, i.e. while the small vehicles are respectively in two and three rows with 3.5% and 5% blockages, located in front of two fires, CVV is equal to unoccupied state for all L distances because this blockage percentage does not prevent the access of
Fig. 9. Contour of temperature of simultaneous two fires; (a) d = 0 m, Vcr = 0.61 m s1; (b) d = 0.3 m, Vcr = 0.59 m s1; (c) d = 0.6 m, Vcr = 0.52 m s1.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
Fig. 11. The variation of Froude number versus percentage of blockage.
ventilation flow to the fire plume. These results are similar to that of empirical study have been accomplished by Lee and Tsai (2012).
97
In case b, two rows of medium vehicles by 16% blockage, when the distance between fires and vehicles (L) is trivial, the ventilation velocity declines. As L = 5 cm this velocity is 7% less than unoccupied state. While L is great enough, the effect of obstacle on the fire behavior is vanished and Vcr is equal to unoccupied state. Lee and Tsai (2012) have reached to the same result in investigating the presence of two rows of medium vehicles. Comparing the inertia and buoyancy forces could interpret obtained results. Therefore, the Froude number variation at different blockage percentages of arrangement is shown in Fig. 11. It could be seen that increasing blockage percentage leads Froude number to augment. In other words, increase of blockage tends to empower inertia force of wind flow with respecting the buoyant force of the fire plume in the cross section. In case d, three rows of medium vehicles by 24% blockage, not only the Froude number is greatened dramatically but also a vortex structure would be generated in the fire upstream. It is shown in Fig. 12a–d that while distance between vehicles and fires is small; a part of the first fire plume energy is encompassed in generated eddy. Consequently, it could be said that CVV reduces noticeably. According to Table 3 this reduction is about 11% in the case which L = 5 cm. While L increases and fire plume gets out of the turbulence region, Vcr soars remarkably. The turbulence region is more clarified in Fig. 13.
Fig. 12. The effect of blockage of 9 vehicles in medium size which are arranged in three rows, d = 60 cm; (a) L = 5 cm, Vcr = 0.45 m s1; (b) L = 10 cm, Vcr = 0.5 m s1; (c) L = 15 cm, Vcr = 0.52 m s1; (d) L = 20 cm, Vcr = 0.52 m s1.
Fig. 13. Generated turbulent region by medium-size vehicles, L = 20 cm.
98
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99
Studying the effect of case d on one fire presented that Vcr increases about 20% with respect to unoccupied state (Mapar et al., 2013). Whereas, in the present paper 24% blockage in the presence of two fires cause a decrease in Vcr with respect to unoccupied state. The effect of obstruction via Different arrangement of vehicles on CVV in the state that the distance between fires is 60 cm, is evaluated for different values of L (Fig. 14). Comparing the results for all four arrangements of vehicles in presence of two fires with d = 60 cm, the parameter / is shown in Table 4. Based on Fig. 14 and Table 4, it could be concluded that: Presence of obstacle reduces Vcr. The reason of this reduction could be found in dimensionless Froude number. According to Fig. 11 existence of obstacles in the tunnel causes augmentation in inertia force rather than buoyant force of fire. In this state increasing the distance of vehicles leads Vcr to growth and reach to unoccupied state as well. Once three rows of vehicles are located in front of fires, indeed the access of entrance air to the fire plume reduces owing to existence of middle row and this issue causes Vcr to be plunged as the distance between vehicles and the fire is trivial. 6.2. Variation of fire locations in lateral direction In this case the effect of blockage, while L is increasing, on CVV for the most critical state, i.e. two fires stick to the walls, is represented in Fig. 15. Fig. 15 shows that the effect of blockage on CVV is the same as longitudinal state. It means that when the set of fires approaches the vehicles due to inertia force augmentation in fire location, CVV declines significantly. Whereas, as L increases, CVV magnitude approaches to unoccupied state. For instance, in case d, when L is equal to 5 cm, the critical ventilation velocity tends to be
Fig. 14. Alteration of CVV for different distances between vehicles and fires in all arrangements of obstructions, d = 30 cm.
Table 4 The percentage of difference of CVV in proportion of occupied tunnel to unoccupied tunnel, d = 30 cm. L (cm)
5 10 15 20
/ (%) Case a
Case b
Case c
Case d
0 0 0 0
6 6 6 0
7 7 5 3
18 16 11 3
The minus sign shows that CVV is decreased rather than CVV in unoccupied tunnel.
Fig. 15. Alteration of CVV for different distances between vehicles and fires in all arrangements of obstructions, two fires stick to the walls.
0.55 m s1 to overcome inertia force; however, when L is 20 cm, it reaches to the unoccupied state value (Vcr = 0.66 m s1). It is worth mentioning that, owing to the effect of occupied blockage on the inertia force, the value of critical velocity in cases a and c is greater than cases b and d, respectively. Furthermore, the variation of Vcr for cases b and d which have larger blockage percentage (Case b = 16% and case d = 24%) is more significant than cases a and c, respectively.
7. Conclusions In the present investigation, the behavior of fire in a road tunnel and the effect of obstacle presence at the upstream of two fire sources are studied. Comparing the present numerical results with experimental ones indicates the numerical results are within a good agreement of the experimental results. The results indicate in both longitudinal and lateral directions, the Critical Ventilation Velocity (CVV) alters when the distance between two fires increases. Moreover, in longitudinal state, as the distance between fire sources increases: (1) the effect of two fire plumes on each other vanishes and (2) critical ventilation velocity in the occupied tunnel (Vcr) becomes equal to the state which only a fire exists in the tunnel. In lateral state, when two fires stick to the walls CVV soars remarkably. When the obstacles occupy the tunnel Vcr decreases because of increase in inertia force of air with respect to buoyant force. When the obstacles are far from the fire, the value of critical ventilation velocity approaches to the state that tunnel is devoid of obstacles. Furthermore, it is observed that existence of small vehicles does not have any effect on CVV. As the percentage of blockage increases, the critical ventilation velocity decreases. Overall, it could be deduced that while this percentage augments, the existence of a vortex structure in front of vehicles leads to encompass a part of fire plume energy and consequently, diminish the value of critical ventilation velocity.
References Floyd, J., Forney, G., Hostikka, S., Korhonen, T., McDermott, R., McGrattan, K.B, 2012a. Fire Dynamics Simulator User’s Guide. NIST Special Publication 1019. Floyd, J., Forney, G., Hostikka, S., Korhonen, T., McDermott, R., McGrattan, K.B., 2012b. Fire Dynamics Simulator Technical Reference Guide, Volume 1: Mathematical Model. NIST Special Publication 1018. Gannouni, S., Maad, R.B., 2015. Numerical study of the effect of blockage on critical velocity and backlayering length in longitudinally ventilated tunnel fires. Tunn. Undergr. Sp. Technol. 48, 147–155. Gao, Z.H., Ji, J., Fan, C.G., Sun, J.H., Zhu, J.P., 2014. Influence of sidewall restriction on maximum ceiling gas temperature of buoyancy-driven thermal flow. Energy Build. 84, 13–20.
G. Heidarinejad et al. / Tunnelling and Underground Space Technology 59 (2016) 91–99 Gao, Z.H., Ji, J., Wan, H., Li, K., Sun, J.H., 2015. An investigation of the detailed flame shape and flame length under the ceiling of a channel. Proc. Combust. Inst. 35, 2657–2664. Henke, A., Gagliardi, M., 2004. The 2001 Gotthard road tunnel fire. Tunn. Manage. Int. 7 (1), 33–39. Hu, L.H., Huo, R., Li, Y.Z., Wang, H.B., Chow, W.K., 2005. Full-scale burning tests on studying smoke temperature and velocity along a corridor. Tunn. Undergr. Sp. Technol. 20, 223–229. Hwang, C.C., Edwards, J.C., 2005. The critical ventilation velocity in tunnel fires – a computer simulation. Fire Saf. J. 40 (3), 213–244. Ingason, H., 2008. State of the art of tunnel fire research. In: Proceeding of the Ninth IAFSS Symposium, Germany. Ji, J., Fan, C.G., Zhong, W., Shen, X.B., Sun, J.H., 2012. Experimental investigation on influence of different transverse fire locations on maximum smoke temperature under the tunnel ceiling. Int. J. Heat Mass Transf. 55, 4817–4826. Lee, Y.P., Tsai, K.C., 2012. Effect of vehicular blockage on critical ventilation velocity and tunnel fire behavior in longitudinally ventilated tunnels. Fire Saf. J. 53, 35– 42. Li, Y.Z., Lei, B., Ingason, H., 2010. Study of critical velocity and backlayering length in longitudinally ventilated tunnel fires. Fire Saf. J. 45 (6–8), 361–370. Mapar, M., Heidarinejad, G., Zargartalebi, H., 2013. Numerical investigation of fire in tunnel and developing an analytical equation for critical velocity. In: 21th Annual International Conference on Mechanical Engineering, Faculty of Mechanical Engineering K.N. Toosi University of Technology, 7–9 May (in Persian).
99
Modic, J., 2003. Fire simulation in road tunnels. Tunn. Undergr. Sp. Technol. 18 (5), 525–530. Oka, Y., Atkinson, G.T., 1995. Control of smoke flow in tunnel fires. Fire Saf. J. 25 (4), 305–322. Pasdarshahri, H., Heidarinejad, G., Mazaheri, K., 2013. Evaluation of induced-flow in a two-room compartment fire using large eddy simulation. J. Modares Mech. Eng. 13 (4), 74–85 (in Persian). Quintiere, J.G., 1989. Scaling applications in fire research. Fire Saf. J. 15 (1), 3–29. Thomas, P.H., 1958. The Movement of Buoyant Flow Against a Stream and Venting of Underground Fire. Fire Research Station Note No. 351. Thomas, P.H., 1968. The Movement of Smoke in Horizontal Passages Against an Air Flow. Fire Research Station Note No. 723. Tsai, K.C., Chen, H.H., Lee, S.K., 2010. Critical ventilation velocity for multi-source tunnel fires. J. Wind Eng. Ind. Aerodyn. 98 (10–11), 650–660. Vuilleumier, F., Weatherill, A., Crausaz, B., 2002. Safety aspects of railway and road tunnel: example of the Lotschberg railway tunnel and Mont-Blanc road tunnel. Tunn. Undergr. Sp. Technol. 17, 153–158. Werner, H., Wengle, H., 1993. Large-eddy simulation of turbulent flow over and around a cube in a plate channel. Turb. Shear Flows 8, 155–168. Wu, Y., Bakar, M.Z.A., 2000. Control of smoke flow in tunnel fires using longitudinal ventilation systems – a study of the critical velocity. Fire Saf. J. 35 (4), 363–390. Yi, L., Xu, Q., Xu, Z., Wu, D., 2014. An experimental study on critical velocity in sloping tunnel with longitudinal ventilation under fire. Tunn. Undergr. Sp. Technol. 43, 198–203.