Study of Baffle Height and Wind Velocity Effect on the Characteristics of Pool Fires in a Wind Tunnel

Abstract

Wind-blown pool fire accidents in a ventilated tunnel usually pose and present serious risks to properties and personal safety. In this paper, the flame characteristics and flow field changes of pool fires under the combined influence of baffles and crosswinds were investigated using wind tunnel experiments and numerical simulations. The fire experiments reveal that the flame length increases first, then decreases with increasing crosswind speeds up to 3 m/s and then becomes almost constant. Finally, as the wind speed continues to increase, the flame length remains constant. The flame height decreases with crosswind speed up to 3 m/s and then remains almost unchanged. As for the flame tilt angle, the flame above the baffle increases first and then remains unchanged when the velocity of crosswind increases from 2 to 5 m/s. According to the simulation results, the clockwise rotating eddy will be generated downstream of the baffle. The recirculation zone on the surface of the pool transfers the fresh air to the flame. The vortex zone downstream of the pool can retain heat and send back unignited fuel vapor to be burned rather than being blown away. The results may provide beneficial suggestions for understanding flame development behind obstacles for ventilated tunnels.

1. Introduction

Wind-blown pool fires, which occur as a result of the spill and ignition of hydrocarbon fuels in ventilated environments, such as aircraft engines [1], shafts [2], subways [3,4,5,6], etc., present serious risks to people and properties. They are a crucial foundational topic in the research area of flames. Wind tunnel fire experiments have been the most widely used method for studying the phenomena associated with the development of wind-blown fires. Apte and Green [7] experimented on the burning rate of a 1 m-diameter aviation fuel pool fire in a wind tunnel and reported that when the wind speed increased from 0.57 to 2 m/s, the combustion rate decreased by nearly 25%. Carvel [8] studied the influence of the HRR (heat release rate) of a pool fire in a wind tunnel, and it was found that the HRR of large-scale pools continuously increased with increasing wind velocity, while the HRR of small-scale pools increased first and then decreased. Oka et al. [9] investigated the effect of 0.55 to 2.2 m/s cross airflow on fires by changing the length–width ratio of a rectangular pan, and an empirical flame tilt angle prediction model was suggested. Hu [10] conducted an experimental investigation on heptane and ethanol fires with square pool sizes of 0.1 to 0.25 m in wind speeds of 0 to 2.5 m/s and found that the flame length elongated as the wind speed increased.

Previous research has primarily concentrated on crosswind influences on fire development, and the effect of blockage has been ignored. In practice, congested vehicles in tunnels, clutters in aircraft engines, and obstructions in ship engine rooms can be considered as blockages in fire accidents. Besides crosswinds, these blockages in a ventilated environment have a significant influence on fire development, which will affect the airflow field and thus the flame shape [11]. Hirst [12] studied the influence of baffle height and wind speed on the stability of an aviation kerosene pool fire, and it was shown that the most stable flame was obtained with a 2.54 cm-high baffle in a wind tunnel when the wind speed was 3.96 m/s. Chen [13] experimentally investigated the effect of a 10–40 cm-high baffle and 0–3 m/s crosswind on an 8 cm pool fire outside the wind tunnel, and an empirical model for predicting flame height and inclination was established.

In general, reports on the condition of combining factors of baffles and crosswinds on pool fire behavior are rare, and researchers have mainly focused on pool fires at 0–3 m/s wind speeds. In this paper, wind tunnel fire experiments were carried out to obtain the flame morphology characteristics of pool fires under the combined influence of obstacles and crosswinds. Numerical simulations were performed to obtain an airflow field together with the HRR of an N-heptane pool fire under a 2 to 10 cm-high baffle and a 1 to 5 m/s crosswind.

2. Experiments and Numerical Models

2.1. Experimental Methods

The experimental setup, mainly composed of a wind tunnel, is illustrated in Figure 1a. The test cabin, with an internal size of 2.0 m (L) × 1.6 m (W) × 1.3 m (H), was made of 3 mm-thick steel plates. The ambient air temperature was 28 °C ± 2 °C. A 0.08 m (L) × 0.08 m (W) × 0.03 m (H) square steel pan containing 100 mL N-heptane was placed at the longitudinal center line of the tunnel. There was a baffle of 1 m in length at the leading edge of the square plate, and the baffle height varied from 2 to 10 cm in the experiments.

The crosswind speeds in the experiments ranged from 0 to 5 m/s. Before the fire experiments, a hot wire anemometer probe (±0.01 m/s accuracy) was installed 0.1 m upstream of the baffle to record the wind velocity in real time, as shown in Figure 1b. From Figure 1b, the ratio of the standard deviation of the ventilation velocity δu to the average ventilation velocity u_means characterizes the wind stability in this case.

As summarized in

Table 1. A summary of the experimental and simulation scenarios.

No.	Baffle Height (H cm)	Crosswind Velocity (V m/s)
1–6	2	0, 1, 2, 3, 4, 5
7–12	4	0, 1, 2, 3, 4, 5
13–18	6	0, 1, 2, 3, 4, 5
19–24	8	0, 1, 2, 3, 4, 5
25–30	10	0, 1, 2, 3, 4, 5

, a total of 30 scenarios were tested in this research under a series of crosswind velocities and 5 different baffle heights. All of the scenarios were repeated twice to avoid random errors.

2.2. Simulation Details

Meanwhile, wind tunnel fires were simulated using the LES (large eddy simulation) method in the FDS (Fire Dynamics Simulator) [14] to obtain the flow velocity field and the HRR of the flame. The calculation model was established in the same size as the fire test cabin. In the numerical calculation process, one of the side boundaries was set as “wind” while the matching export was set as “open”. The other four side boundaries of the wind tunnel were 3 mm-thick steel. The ambient air temperature was 23 °C. The N-heptane (C₇H₁₆) fire was set as a flame source (8 cm × 8 cm × 6 cm) at the center axis point of the wind tunnel.

Table 2. Thermal characteristic parameters of heptane and steel.

Material	Density (kg/m3)	Special Heat Capacity (kJ/kg·k)	Thermal Conductivity (W/m·k)	Absorption Coefficient (m−1)	Heat of Reaction (kJ/kg)	Boiling Temperature (°C)
Heptane	675	2.24	0.14	333	317	98.5
Steel	7850	0.46	45.8	-	-	-

indicates the thermal characteristic parameters of the N-heptane and steel provided by the FDS website for this simulation calculation.

The grid size, δx, was determined based on the fire characteristic diameter, $D^{*}$, given by ${\left[\dot{Q}/\left({\rho }_{\infty }{c}_p{T}_{\infty }\sqrt{g}\right)\right]}^{2/5}$ [14], where $\dot{Q}$ is the heat release rate, ${\rho }_{\infty }$ is the ambient air density, c_p is the heat capacity of the air at atmospheric pressure, $T_{\infty }$ is the ambient temperature, and ɡ is the gravity. Full-scale and model-scale experiments have proven that the simulation calculation can obtain reliable results when δx/$D^{*}$ ≤ 0.1. Based on the calculation result, the grid size was set as 0.01 m × 0.01 m × 0.01 m in these simulations. The duration of the fire simulation was 60 s, before which the stable burning states of the N-heptane fire were reached. The primary goal of the simulation was to analyze the flow field in wind situations, so the 0 m/s wind speed numerical simulation scenario was omitted. The simulation scenarios were the same as the 25 experimental scenarios shown in Table 1.

3. Results and Discussions

3.1. Flame Evolution

For the free-burning N-heptane pool fire, a high concentration of fuel vapor was present above the pool surface with insufficient oxygen. However, for the pool fire burning in ventilated conditions, the amount of oxygen near the fuel surface was supplemented, and the heat feedback changed, leading to changes in the evolution of wind-blown pool fires [15,16,17]. The coupling effect of flame buoyancy and wind drives flame behaviors, which tilts the flame and changes its geometry. The presence of a blockage baffle enhances the interaction between the crosswind and fire and consequently creates complex turbulent flow conditions.

Figure 2 displays the baffle height’s influence on flame shape changes in the pool fire in the experimental section at a 3 m/s crosswind. After igniting, the flame quickly spread across the pool, and the flame size increased until a stable regime was established. Then, the fire gradually decreased until it was extinguished owing to a lack of fuel. As the baffle height increased, the flame body became brighter, and the flame volume expanded. When the height of the baffle in front of the wind-blown flame increased from 2 to 10 cm, the burnout time reduced from 360 to 211 s. The entire combustion process can be separated into three stages: the initial burning stage, steady combustion stage, and decline stage. Figure 3 depicts the pool fire flame shape under a stable combustion stage at a 0–5 m/s crosswind velocity. When the wind speed increased from 1 to 2 m/s, the flame length above the baffle was significantly reduced due to the high wind speed which prevented the fuel vapor from timely combustion and reaction. Meanwhile, due to the increase in wind speed, the turbulence and velocity in the backflow area behind the baffle also increased. The flame under the baffle had a full combustion reaction. The combustion intensity increased, and the brightness of the flame increased. When the baffle height increased, the flame morphology changed from an inclined cone shape to a curved shape with a corner. When the baffle height was 2 cm, the flame morphology was similar to that without the baffle, with the flame inclined downstream. When the height of the baffle increased to more than 4 cm, it can be seen that the flame bent. The lower part shows the vertical flame close to the baffle and the upper part shows the flame inclined in the downstream direction.

3.2. Numercial Analysis

3.2.1. Flow Field

A baffle set in the wind’s path will obviously make the airflow field change and thus affect the flame shape. FDS was used to calculate the change of flow field when the wind velocity was 1 to 5 m/s, and Smokeview showed the velocity vector slice in the numerical calculation zone. One slice, Y = 0, was set to measure the flow velocity in horizontal directions. Considering that the flow velocity vector fields were similar for different wind speeds, the slice of the flow velocity vector fields of 1 m/s was set as an example in Figure 4 to depict the velocity vector field and the oxygen volume fraction. For a more intuitive representation of the flow field, the velocity vector slice in the wind tunnel was simplified as a schematic illustration, as shown in Figure 5. The shear layer was generated at the top of the baffle as the wind passed the baffle. Meanwhile, two vortex regions formed downstream of the baffle: a recirculation zone near the pool and a much larger recirculation zone downstream of the pool.

As the baffle height increased, the recirculation downstream of the pool became larger. When the baffle was placed in front of the pool, it blocked the bottom airflow. The flow field downstream of the baffle was divided into two parts. The airflow above the baffle formed a mainstream zone with higher velocity, while the bottom airflow formed low-speed recirculation zones. The range of the recirculation region is determined by a variety of flow characteristics, the most important of which are the baffle height and the cross-stream turbulence level [18]. The process of flame combustion is influenced by reverse flow in the recirculation zones [19]. The recirculation zone above the pool surface transfers the fresh air to the flame. The recirculation zone downstream of the pool can retain heat and send back unignited fuel vapor to be burned rather than being blown away as is normally the case without a baffle [20,21].

3.2.2. Heat Release Rate (HRR)

The HRR is the amount of energy released by a burning fire per unit time, influenced by fuel characteristics, ambient circumstances, and other factors [22]. It is an important parameter indicating the intensity of heat energy released by the fire, regardless of whether the fire is in still air or wind-blown conditions [23]. When wind is present, the heat feedback changes. The natural convective boundary layer caused by buoyancy is replaced by the wind-forced convective boundary layer, which may lead to more intense convective heat feedback on the fuel surface. As the HRR developing trends were similar regardless of the baffle height, the heat release rate of the pool fire with a 6 cm baffle height was set as an example in Figure 6a, while the wind speeds varied from 1 to 5 m/s. The variation trend of the average HRR of a pool fire with the wind velocity (V = 1–5 m/s) at different baffle heights in 30 s of stable combustion stage is shown in Figure 6b.

As shown in Figure 6b, the average HRR under the different baffle heights (H = 2–10 cm) presented a similar increasing trend with the crosswind varying from 1 m/s to 5 m/s. When small-scale pool fires burn in wind-blown states, the convection heat feedback percentage increases, which becomes the major heat feedback mechanism and influences the flame characteristic [19]. In the condition of the combined effect of the baffle and cross airflow, the flow field and heat convection near the oil pool become more complicated compared to the case of wind alone [24,25,26]. The recirculation zone downstream of the pool shown in Figure 5 can retain heat and send back unignited fuel vapor to be burned rather than being blown away, as is the normal case without a baffle [27]. There is a turbulent heat exchange between the recirculation zones and the surrounding flow, which constantly supplies the energy required for ignition to the fuel vapor. The increase in crosswind speed may increase the velocity of the airflow in the recirculation zone and improve the mixing efficiency of fuel and oxidant. Furthermore, more high-temperature gas and fuel vapor returning to the fire zone can increase the HRR of the flame. As a result, the increase in the HRR exhibited an increase in flame brightness and volume, as shown in Figure 3. Under the experimental scenario of a 10 cm baffle height, an explanation for the decrease in the heat release rate may be the recirculation zone being formed far from the pool fire at a relatively low wind speed. Consequently, the heated gas cannot be kept near the oil pool in time, resulting in a drop in the heat release rate.

3.3. Flame Length, Height, and Tilt Angle

Although the crosswind was almost constant, the flame continuously oscillated during the experiments because of the combined impact of the airflow turbulence and the turbulent motion in the combusting behavior. In order to analyze the flame shape, a CCD (charge-coupled device) camera was used to record the flame video for 30 s at a steady rate of 30 frames per second. Adobe Premiere was utilized to transform the video into multi-frame pictures, and then the pictures were further converted into greyscale and binary images using a threshold value calculated by the Otsu method [28]. The number of 1 in each pixel point in the binary images was counted to determine the position of the flame, and the threshold of the mean flame contour was determined by the 50% flame occurrence probability to quantify the flame geometric features.

3.3.1. Definition of Flame Length, Height, and Tilt Angle

The geometrical characteristics of a flame govern its radiation heat feedback to the surrounding environment and the surface of the oil pool and thus affect the fire spread. When set in a crosswind environment, a flame is inclined by the crosswind, which enhances the radiation intensity to the downstream area. Therefore, the flame length and flame tilt are the basic behavioral parameters for analyzing a wind-blown fire, which have been defined by previous researchers. The flame length (L_f) under a crosswind is expressed as the linear distance from the center of the fuel surface to the tip of the flame [29]. The flame tilt angle α is the angle at which the connecting line from the center of the pool to the inclined flame tip deviates from the vertical direction, as defined by Oka et al. [30].

Figure 7 depicts two typical flame features deduced from flame photos based on the experimental results. For the relatively lower baffle heights, the flame shape was similar to the wind-blown pool fire scenarios without a baffle (Figure 7a) and the flame will tilt towards the direction of the wind. As the baffle height exceeded 2 cm, the flame shape changes as per Figure 7b. The flame part below the baffle height was close to the baffle due to the existence of the recirculation zone, and the flame part over the baffle tilted downstream due to the crosswind. The flame length and tilt angle were redefined as a result of the bending of the flame in this article. The flame length was divided into bottom flame length L_f1 and upper flame length L_f2 based on the baffle height, as shown in Figure 7b. Hence, the total length of the flame can be expressed as L_f = L_f1 + L_f2. The flame height H_a is defined as the height from the pool surface to the flame tip.

As for the flame tilt angle, the former definition by Oka et al. was applicable to describe a flame without a baffle blockage or a flame with a baffle height less than 2 cm, as shown in Figure 7a. However, the flame shape changed when a baffle over 2 cm in height was set. In this case, the flame tilt angle α was used to describe the angle of inclination of the upper flame above the baffle height relative to the vertical direction, as shown in Figure 7b.

3.3.2. Variation of Flame Length, Height, and Tilt Angle

In windy conditions, variations of pool fire length are mainly determined by the coupling of the pool size, flame lift, fuel evaporation rate, and airflow environment. The interactions between wind force and flame buoyancy affect air entrainment and lead to irregular flame length variations [31].

Figure 8 shows the flame characteristics of length, height, and tilt angle under a 0–5 m/s wind velocity and a 6 cm-high baffle. It can be seen that as the wind velocity increased from 0 to 5 m/s, the flame was more inclined and the flame base area was larger. The flame height decreased when the wind velocity increased.

Figure 9a,b shows the trends of the average flame lengths and heights with the increase in wind speed. The flame length increased first and then decreased with increasing crosswinds. The flame height decreased with an increase in wind speed.

When the wind velocity increased to 1 m/s, the flame length first increased due to the wind, which increased the spread of flammable gas and maintained the combustion. When the wind velocity exceeded 2 m/s, the speed of fuel vapor diffusion and combustion could not keep up with the wind speed in the mainstream zone. Under the condition of the high baffle (4–10 cm), the flame buoyancy and gas cycle in the recirculation region were the most important influencing elements of bottom flame length L_f. Therefore, the flame length decreased when the wind speed increased. The flame height of the heptane pool fire with a baffle at various crosswind velocities is shown in Figure 9b. The flame height reduced considerably when the crosswind velocity was less than 2 m/s. The downward trend gradually slowed down as the crosswind velocity increased to 5 m/s. What distinguished this result from the pool fire in the crosswind was that the flame height was limited by baffle blockage, as seen in Figure 2. We compared the flame height with the wind–fire experiment when the baffle–fire distance was 20 cm. As for the difference in the flame height, we refer to Chen’s study [13] when there is a distance between the flame and upstream baffle. The bottom part of the flame inclines to the baffle side causing there to be no more fuel vapor to maintain the upper flame, which finally influences the flame height.

Flame tilting is the essential morphological behavior of wind-blown pool fires, which is critical for fire propagation. When a pool fire occurs in the presence of wind, the flame tilt angle is governed by the coupling effect of its vertical buoyancy and the horizontal momentum given by the wind [19]. However, the overall change in flame tilt due to wind involvement and baffle obstruction is more complex. Overall, the flame average inclination angles at varied wind speeds, with baffle heights ranging from 2 to 10 cm in the experiments, are shown in Figure 10.

The flame inclination changed with the presence of a baffle due to an imbalance in air suction at the edge of the oil pool. The upper flame was inclined by the high-speed crosswind in the mainstream area and was also influenced by the recirculation flow and flame buoyancy. For the experimental scenarios, the flame tilt α of pool fires varied from −7.1 to 68.0°. The existence of the baffle caused the flame to be slightly tilted to the baffle side when the wind speed was 0 m/s, so the angle was negative. Then, the tilt angle α increased quickly when the crosswind velocity rose to 2 m/s and almost remained constant when the wind speed continued to rise to 5 m/s. The airflow in the mainstream zone had the main influence on the upper inclination α. It can be seen from the flow field that the higher the baffle height, the greater the wind speed in the mainstream zone. As a result, the higher the height of baffle H is, the larger the inclination angle of the upper flame α is.

4. Conclusions

The impact of baffle height on the flame characteristics of a pool fire was investigated at different crosswind speeds ranging from 0 to 5 m/s using wind tunnel experiments. The flow field near the baffle was also studied to illustrate the flame characteristics based on the simulation computation. Our major findings are listed as follows:

When a baffle was placed at the leading edge of the pool fire, it blocked the bottom airflow. Two vortex regions formed downstream of the baffle: a recirculation zone on the surface of the pool and a much larger recirculation zone downstream of the pool.

The recirculation zone on the surface of the pool transferred fresh air back to the vicinity of the flame. The recirculation zone downstream of the pool could retain heat and send back unignited gas to be burned rather than being blown away. In addition, increasing the crosswind speed and baffle height may improve the mixing of fuel and the oxidant. For these reasons, the HRR increased with the increase in baffle height and wind velocity.

As the flame shape changed when the baffle height was 4–10 cm, a new definition was introduced to illustrate the flame length and flame tilt angle. Firstly, the flame length increased, then decreased with increasing crosswind speeds up to 3 m/s. Finally, as the wind speed continued to increase, the flame length remained constant. The flame height decreased as the wind velocity increased (V ≤ 3 m/s), and when the wind velocity continued to increase to 5 m/s, the flame height was basically unchanged. The flame tilt angle α increased quickly as the wind velocity rose to 2 m/s and then remained almost constant as the wind speed rose further to 5 m/s.