Experimental Study of Spatiotemporal Variations in Lab-Scale Fire Whirls

Abstract

This study focused on the unsteady behavior of fire whirls. A laboratory-scale fire whirl was generated, and temporal variations in flame height were measured from images taken by a high-speed camera and subjected to frequency analysis. The flame height fluctuations of the fire whirl also showed intermittent behavior, such as the puffing of a pool flame. However, the period and amplitude were irregular compared to the pool flame. In addition, the fire whirl exhibited a greater amplitude spectrum at higher frequencies than the pool flame. To investigate the velocity distribution in the horizontal plane, particle image velocimetry (PIV) was employed. The results demonstrated that the mean velocity increased from the outer radial direction toward the inner radial direction, peaked, and decreased. Conversely, the coefficient of velocity variation decreased from the outer to the inner radial direction, exhibited a minimum, and then increased. Finally, the flame was photographed from horizontal and vertical directions under two conditions with different flow velocities from the fan to generate the fire whirl. Image analysis was employed to investigate the relationship between the center position of the flame and the flame height. The results demonstrated that under conditions where the flow velocity from the fan was low, the fire whirl was intermittent and moved following the circular path drawn by the swirling flow, exhibiting unstable behavior. Furthermore, the flame height was lower when the center of the flame was further from the liquid fuel pool.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1. Introduction

Fire whirls are a common phenomenon in large fires, causing significant damage. Numerous researchers have conducted theoretical, numerical, and experimental studies on fire whirls, particularly focusing on the correlation between flame shape and intrinsic parameters. For instance, Lei and colleagues have employed experiments and physical analysis to investigate the effects of heat generation rate and circulation on flame height and width in fire whirls [1–4]. Hartl and Smits conducted experiments using particle image velocimetry (PIV) and performed dimensional analysis to demonstrate that the dimensionless flame height solely depends on the dimensionless circulation [5, 6]. They also generated fire whirls from the burner or pool flames and compared the two [7, 8]. Experimental, numerical, and theoretical analyses by Hayashi et al. revealed a correlation between flame height and vortex parameters in fire whirls [9, 10]. Chow et al. conducted experimental studies on the internal fire whirl in a 15 m-high shaft under different ventilation conditions, deriving a correlation between flame height and fuel consumption [11]. Dobashi et al. investigated the relationship between flame height and radiant heat in fire whirls through small- and medium-scale experiments [12]. In addition to studies on flame height, Hariharan et al. investigated bright and blue flames in fire whirls and the transitions between these flames, as well as particulate matter generated from fire whirls [13, 14]. Other studies have focused on vortex-buoyancy interactions [15, 16] and the interaction between the central flame and the surrounding flames in a fire whirl [17].

Most of the above studies evaluated time-averaged fire whirl characteristics, such as flame height. However, it is important to note that fire whirls fluctuate in flame height and width, with the center of rotation shifting. This unstable behavior of fire whirls is believed to affect fire damage. The movement of the fire whirl has been studied in depth, with research conducted on fire whirls moving on a linear fire [18–22]. In addition, experimental studies have been conducted on the horizontal undulation of fire whirls [23]. Watanabe and Okamoto have also conducted numerical analyses to study the gas temperature distribution and its variation in a lab-scale fire whirl [24]. However, few studies have focused on temporal flame height and shape variations. This study aimed to characterize the unsteady behavior of flame height in fire whirls. Specifically, high-speed images of a laboratory-scale fire whirl were first taken. Subsequently, the images were analyzed to compare the oscillation frequencies of the flame height in a pool flame and a fire whirl generated from a pool flame. The velocity field in the horizontal plane was measured with PIV, and the temporal variation of velocity was also investigated. Furthermore, the correlation between the center of the fire whirl and the flame height was investigated by simultaneously taking images from two directions and analyzing the images.

2. Experimental Setup

Figure 1 shows a schematic of the experimental apparatus used in this experiment, and Figure 2 shows the arrangement of the measurement apparatus. Four partitions surrounded the 500 mm square table, and fans blew the air through two 50 mm gaps between the partitions. The table was 150 mm high, and the partition was 1000 mm high. The liquid fuel pool (50, 60, 75, and 90 mm in diameter) was placed in the center of the table, and the fuel was anhydrous alcohol. The height of the 50 mm diameter fuel pan was 15 mm, and the height of the other fuel pans was 20 mm. During PIV, tracer particles (water-soluble glycol-based smoke material, particle size: 1 μm) generated by a smoke generator (KFS-70 pressurized seeding device, Kato Koken Co. Ltd.) were injected at a flow rate of 10 L/min from the position shown in Figure 1. Only the flow around the flame was visualized as the particles vaporized inside the flame. The PIV laser (PIV Laser G1000, maximum power: 1 W, wavelength: 532 nm, Kato Koken Co. Ltd.) was directed horizontally at a height of 60 mm from the table. A mirror was placed at a height of 900 mm from the table, inclined at 45° from the horizontal plane, and a high-speed camera (HAS D71, Detect Co. Ltd.) was used to capture the images of the flow field in the horizontal plane through the mirror. A band-pass filter (wavelength: 532 nm, FWHM: ±5 nm) was attached to the lens. Since the image was captured through a mirror, the direction of rotation of the flow in the captured image was opposite to the actual direction. The flame height was measured from images of the flame taken from the side with a high-speed camera (Flux k-240, Kato Koken Co. Ltd.). The two cameras were synchronized to simultaneously capture images of the flame from two directions. The commercial software Flownizer 2D (Detect Co. Ltd.) was used for PIV analysis. The flow velocity of the fan was measured with a hot-wire anemometer (Anemomaster Lite Model 6006-D0, Kanomax Japan Inc.). If the flow velocity of the fan was too low, the flame did not swirl; however, even if the flow velocity was too high, the flame was swept away and tilted by the flow from the fan, and the fire whirl did not occur. In this study, measurements were mainly made at a flow velocity of 2.5 m/s from the fan, which produces a stable fire whirl, and experiments were also conducted at 1.5 m/s, which produces an intermittent fire whirl, for comparison.

[figure(s) omitted; refer to PDF]

2.1. Oscillation Frequency of Flame Height

High-speed images were taken of a pool flame and a fire whirl generated from a pool flame, and the oscillations in flame height were studied by image analysis. The table used for measuring pool flames was surrounded by four partitions, with no gaps between them. In the fire whirl experiment, the flow velocity from the blower was set at 2.5 m/s. An 850 × 680 mm area in the vertical plane containing the central axis of the liquid fuel pool was filmed at 1280 × 1024 pixels at a frame rate of 500 fps. Table 1 lists the diameters of the liquid fuel pool, the amount of anhydrous ethanol, and the combustion time (time from ignition to flame extinction). The amount of anhydrous ethanol was set at the amount that would result in a 5 mm depth of ethanol in the pool. The combustion time was the average of the results of three measurements. Figure 3 shows the fuel consumption rates calculated from the amount of anhydrous ethanol and the combustion time. The fuel consumption rate increased linearly with increasing pool diameter. In the case of fire whirls, the fuel consumption rate was greater than that of pool flames, and the slope of the increase in fuel consumption rate with increasing pool diameter increased.

Table 1

Pool diameter, fuel quantity, and combustion time.

Pool diameter (mm)	Amount of anhydrous alcohol (mL)	Combustion time of pool flame (sec)	Combustion time of fire whirl (sec)
50	10	330	220
60	14	324	189
75	22	296	167
90	32	300	167

[figure(s) omitted; refer to PDF]

2.2. Pool Flames

Irregular and violent oscillations in flame height were observed immediately after ignition and just before the flame was extinguished. In the context of a pool flame, the transfer of heat from the flame to the fuel pool is attributed to radiation and convective heat transfer mechanisms [25–28]. The latter is said to be more pronounced when the container size is reduced. Following the heating of the fuel, its evaporation occurs, leading to its mixture with air and subsequent combustion. Immediately after ignition, the fuel is not sufficiently heated, and just before extinction, there is little remaining fuel in the container. Therefore, it is thought that sufficient fuel vapor is not supplied stably, and the flame height oscillates violently. In contrast, the flame had a relatively stable oscillatory motion during the rest of the time. The flame height was defined as the vertical distance from the top of the pool to the flame tip and was measured using an image analysis software (HAS-XViewer Ver. 1.6.0.3, Detect Co. Ltd.). Figure 4(a) shows the time variation of the flame height of the pool flame at a pool diameter of 50 mm, and Figures 4(b) and 4(c) show flame images corresponding to a peak and a trough of the flame height oscillation. The flame heights shown in Figures 4(b) and 4(c) are approximately 130 and 70 mm, respectively. The bright area in the background is an image reflected in the glass. Figure 5 shows the frequency analysis results by applying a fast Fourier transform (FFT) to Figure 4(a). The recording started 90 s after ignition, and the data used for the FFT were 1.022 s (512 frames).

[figure(s) omitted; refer to PDF]

In pool flames, a flame oscillation phenomenon called puffing occurs when the pool diameter exceeds 30 mm. Puffing is an intermittent phenomenon that repeats the process of the flame nodding upward in a continuously burning area, forming a mushroom-shaped flame due to the entrapment of ambient air, and then breaking away [29, 30]. In this experiment, the puffing phenomenon was also confirmed, and the flame height increased until the flame grew upwards and formed a mushroom-shaped flame. After this, the flame height decreased rapidly as the flame tip separated. The discontinuity in the flame height, from its peak to its bottom, as illustrated in Figure 4(a), can be attributed to the puffing phenomenon. Figures 4(a) and 5 show that flame height oscillations occur with a frequency of about 10 Hz. Temporal flame height fluctuations were also measured for 60, 75, and 90 mm pool diameters. Although flame height oscillations that appeared to be puffing were observed, the period and amplitude of the oscillations were not as constant as for the 50 mm pool diameter. The following discussion examines the possible causes of unstable flame behavior in larger diameter pools. In this experiment, the pool was placed on a table and four partitions were constructed around the table. The mobility of these partitions resulted in slight gaps between the partitions and between the partitions and the table. When the smoke was made to flow around the partitions and tables to visualize the flow, it was observed that air was flowing in through these gaps. Although we have not made quantitative measurements, it is believed that as the diameter of the pool increases, the fuel consumption rate also increases, resulting in a greater influx of air into the unit through these gaps. This phenomenon is believed to be a contributing factor to the irregular oscillation of the pool flame, especially in larger diameter pools. The following discussion focuses on the case of a 50 mm diameter pool, where the effect of drafts is minimal.

2.3. Pool Fire Whirls

A swirling flame was generated by airflow at 2.5 m/s from fans installed at two locations. Filming started 90 s after ignition for the 50 and 60 mm pool diameters and 60 s after ignition for the 75 and 90 mm pool diameters. The data used for the FFT were 1.022 s (512 frames). Figure 6(a) shows the time evolution of the flame height in the fire whirl for the 50 mm pool diameter, and Figures 6(b) and 6(c) show flame images corresponding to a peak and a trough of the flame height oscillation. The flame heights shown in Figures 6(b) and 6(c) are approximately 330 and 140 mm, respectively. Flame height oscillations were also observed in the fire whirl, but unlike the pool flame, the amplitude of the flame height fluctuations was large and irregular, and the period also appeared irregular. Deviations in the flame position from the center of the swirling flow may result in tilt or waver, which is believed to contribute to irregular fluctuations in flame height. Furthermore, the swirling flow augments the heat feedback to the surface of the fuel and the radial velocity near the base of the flame, thereby increasing the combustion rate [31]. In addition, fuel vapor may be driven away by the flow, which can lead to a reduction in fuel supply from the liquid surface. The occurrence of these effects, which promote or hinder combustion, in succession is believed to be the reason for the irregular fluctuations. Figure 7 compares the flame height frequency distribution of the pool flame and the pool fire whirl. The fire whirl had a higher amplitude spectrum for all pool diameters than the pool flame at relatively higher frequencies.

[figure(s) omitted; refer to PDF]

2.4. Fluctuation of Flow Velocity

A high-speed camera was used to capture a 555 × 740 mm horizontal surface at a resolution of 480 × 640 pixels and a frame rate of 500 fps to determine the velocity distribution on the horizontal surface by PIV for a fire whirl generated as a 50 mm pool diameter and 2.5 m/s flow velocity from a fan. The start time of filming was 60 s after ignition, and the duration of filming was 2.0 s (1000 frames). The distance between the velocity measurement points by PIV was 10 mm, and 625 measurement points were placed, 25 vertically and 25 horizontally. The inspection area size was set to 20 pixels × 20 pixels, and the exploration area was set to 10 pixels each for the top, bottom, left, and right. The validity of the PIV was confirmed by comparison with hot-wire velocimetry measurements. Figure 8 shows the average velocity vector diagram for 2.0 s. The direction of rotation is opposite to the actual one because the image was taken through a mirror. Since the tracer particles were liquid, they evaporated as they approached the flame. Therefore, only the area outside the flame where the tracer particles were visible was observed. A swirling flow could be observed around the flame, indicating that the velocity increased near the flame.

[figure(s) omitted; refer to PDF]

Figure 9 shows the frequency analysis results of the velocity variation at a measurement point of 55 mm from the center of the pool by FFT. The data used for the FFT were for 512 frames. Where tracer particles could not be detected, and velocity data were missing, the data were supplemented by linear interpolation from data at earlier and later times. No characteristic peaks in the frequency distribution indicated that the velocity fluctuations included a wide range of frequency fluctuations.

[figure(s) omitted; refer to PDF]

Figure 10 shows the radial distribution of the time-averaged velocity and the coefficient of variation of the velocity. It should be noted that the velocity referred to here is the two-dimensional velocity in the horizontal plane, as shown in Figure 8. The average velocity increased from the outer radial direction toward the center, reached a peak, and then decreased. The average flow velocity exhibited a decline in the vicinity of the center, a phenomenon attributed to the presence of an updraft in the proximity of the flame, thereby reducing the horizontal velocity component. On the other hand, the coefficient of variation showed the opposite trend to the average velocity, decreasing once from the outer radius toward the center and then increasing, with the velocity fluctuation increasing near the flame.

[figure(s) omitted; refer to PDF]

2.5. Relationship Between the Center of the Fire Whirl and Flame Height

The flame was photographed simultaneously from two directions, horizontally and vertically, to investigate the relationship between the change in the center position of the fire whirl and the flame height. The center position of the flame was determined as follows. If the outline of the flame could be found in the image taken from above the flame, the center of the two points showing the maximum width of the flame was taken as the flame center (Figure 11(a)). If the outline of the flame was not visible, the center of the flame was defined as the center of the line connecting the furthest point from the center of the pool where the flame could be seen and the center of the pool (Figure 11(b)). It should be noted that the image utilized to determine the flame center encompasses the entire flame and does not depict the flame center at a particular elevation. The pool diameter was 50 mm, and the volume of anhydrous ethanol was 20 mL. Two fan flow velocities were selected as follows: 2.5 m/s for a relatively stable fire whirl and 1.5 m/s for an intermittent fire whirl. Table 2 shows the recording conditions of the cameras installed in two directions. Data from 2.4 s (1200 frames) after 60 s of ignition for a flow velocity of 2.5 m/s and 9.8 s (4900 frames) after 66 s of ignition for a flow velocity of 1.5 m/s were used for analysis.

[figure(s) omitted; refer to PDF]

Table 2

Recording conditions.

		2.5 m/s	1.5 m/s
Vertical plane	Photographic area (mm × mm)	512 × 384	615 × 460
	Number of pixels	640 × 480
	Frame rate (fps)	500
Horizontal plane	Photographic area (mm × mm)	744 × 558	630 × 472
	Number of pixels	640 × 480
	Frame rate (fps)	500

Figure 12 shows the variation of the flame’s center position. The center position was measured at 0.2 s intervals (100 frames), with the center of the liquid pool as the origin. At a flow velocity of 2.5 m/s, the maximum movement of the flame center from the origin was within 30 mm in the X and Y directions, respectively, and was stable with slight fluctuation. On the other hand, at the flow velocity of 1.5 m/s, the displacement of the flame center from the origin was about 50 mm in the X and Y directions, which was larger than that at the flow velocity of 2.5 m/s. It has been established that an increase in the rotational speed of the swirling flow increases fuel consumption rate [5]. In circumstances where the flow velocity is elevated, the supply of oxidants to the flame is also increased, thereby accelerating the combustion rate and generating a powerful updraft. This updraft, in turn, accelerates the flow towards the inner radius at the base of the flame, resulting in a decrease in the range of movement of the center of the flame.

[figure(s) omitted; refer to PDF]

Figure 13 shows the swirling flow and flame motion photographed from above at a flow velocity of 1.5 m/s. Time progresses from the left figure to the right figure. The red frame shows the center of the swirling airflow, and the yellow circle shows the pool’s position. Below the pool, there is a smoke inlet. The bright area outside the yellow circle is the flame. At a flow velocity of 1.5 m/s, the center of the swirling flow moved in a circular orbit. Since the surrounding swirling flow forms the fire whirl, it can be seen that the center of the flame also moved significantly with the movement of the swirling flow.

[figure(s) omitted; refer to PDF]

In the case of a flow velocity of 2.5 m/s, a large vortex is formed around the fuel pool. It is hypothesized that this vortex is drawn towards the center by the inward flow induced by the flame, contributing to the generation of a fire whirl. Conversely, at a flow velocity of 1.5 m/s, a vortex is formed near the corner of the device, and it is hypothesized that this is drawn toward the flame by the inward flow, moving in a circular orbit around the fuel pool, as illustrated in Figure 13. This affects the movement of the flame.

Figure 14 shows the relationship between the center position of the fire whirl and the flame height. Approximate straight lines are also superimposed. Since the flame height varied in space and time, it was not easy to see a strong correlation between the center position of the flame and the flame height. However, when Pearson’s correlation coefficient was calculated, it was found that for a flow velocity of 2.5 m/s, the value was −0.45, and for a flow velocity of 1.5 m/s, the value was −0.66. These results indicated that the flame height tends to decrease as the center of the flame moves away from the origin.

[figure(s) omitted; refer to PDF]

3. Conclusions

This study investigated the temporal variation of the flame height and velocity distribution using image analysis and PIV to understand the unsteady characteristics of the fire whirls. The results are summarized as follows:

1. Intermittent flame height fluctuations were observed in the fire whirl, similar to the puffing of a pool flame, but the flame height oscillation period and amplitude were not constant. The flame height oscillation period of the fire whirl was generally shorter than that of the pool flame, but the oscillation period was sometimes longer than that of the pool flame. This irregular fluctuation in flame height is due to the flame tilting and wavering due to the interaction between the flame and the swirling flow. In addition, it is thought that the irregular fluctuation in flame height is also affected by the swirling flow promoting fuel supply at the base of the flame and, conversely, blowing away the fuel vapor. A comparison of the frequency analysis results of the flame height fluctuations showed that the fire whirl contained more fluctuations in the high-frequency range than the pool flame.

2. In the velocity distribution on the horizontal plane, the time-averaged velocity once increased from the outer radial direction toward the center, reached a peak, and then decreased. This decrease is because the horizontal velocity component decreases due to the updrafts that occur near the flame. On the other hand, the coefficient of variation of the velocity decreased from the outer radial direction toward the center and then increased, indicating a large velocity fluctuation in the vicinity of the flame.

3. The stability of the fire whirl and the flame height varied with the air velocity of the fan used to generate the whirl. When the airflow velocity was high, combustion accelerated, a strong updraft was generated, and the flow toward the center at the base of the flame was accelerated, so the position of the flame stabilized. When the air velocity was low, the center of the swirl flow moved in a circular path, and the flame followed it, making the fire whirl unstable. In addition, the flame height tended to decrease as the distance from the pool to the center of the flame increased.

To predict the extent of damage caused by the fire whirl’s unsteady behavior, it is essential to continue investigating the following aspects: changes in flame shape, fluctuations in velocity, and temperature inside and outside the flame. In the future, to elucidate the time history and correlation of parameters, we will conduct simultaneous velocity field measurements from two directions, and temperature field measurements in addition to performing numerical simulations.

Funding

This work was supported by JSPS KAKENHI (Grant no. JP23K03718).

Acknowledgments

This work was supported by JSPS KAKENHI (Grant no. JP23K03718).

References

[1] J. Lei, N. Liu, L. Zhang, "Experimental Research on Combustion Dynamics of Medium-Scale Fire Whirl," Proceedings of the Combustion Institute, vol. 33 no. 2, pp. 2407-2415, DOI: 10.1016/j.proci.2010.06.009, 2011.

[2] J. Lei, N. Liu, L. Zhang, K. Satoh, "Temperature, Velocity and Air Entrainment of Fire Whirl Plume: A Comprehensive Experimental Investigation," Combustion and Flame, vol. 162 no. 3, pp. 745-758, DOI: 10.1016/j.combustflame.2014.08.017, 2015.

[3] J. Lei, P. Huang, N. Liu, L. Zhang, "On the Flame Width of Turbulent Fire Whirls," Combustion and Flame, vol. 244,DOI: 10.1016/j.combustflame.2022.112285, 2022.

[4] J. Lei, Z. Liu, P. Huang, L. Zhang, N. Liu, "Turbulent Cylindrical Fire Whirl," Combustion and Flame, vol. 255,DOI: 10.1016/j.combustflame.2023.112879, 2023.

[5] K. H. Chuah, G. Kushida, "The Prediction of Flame Heights and Flame Shapes of Small Fire Whirls," Proceedings of the Combustion Institute, vol. 31 no. 2, pp. 2599-2606, DOI: 10.1016/j.proci.2006.07.109, 2007.

[6] K. H. Chuah, K. Kuwana, K. Saito, "Modeling a Fire Whirl Generated over a 5-cm-Diameter Methanol Pool Fire," Combustion and Flame, vol. 156 no. 9, pp. 1828-1833, DOI: 10.1016/j.combustflame.2009.06.010, 2009.

[7] K. A. Hartl, A. J. Smits, "Scaling of a Small Scale Burner Fire Whirl," Combustion and Flame, vol. 163, pp. 202-208, DOI: 10.1016/j.combustflame.2015.09.027, 2016.

[8] K. A. Hartl, A. J. Smits, "Comparing Burner Fire Whirls to Pool Fire Whirls," Combustion and Flame, vol. 258,DOI: 10.1016/j.combustflame.2023.113045, 2023.

[9] Y. Hayashi, K. Kuwana, R. Dobashi, "Influence of Vortex Structure on Fire Whirl Behavior," Fire Safety Science, vol. 10, pp. 671-679, DOI: 10.3801/iaffs.fss.10-671, 2011.

[10] Y. Hayashi, K. Kuwana, T. Mogi, R. Dobashi, "Influence of Vortex Parameters on the Flame Height of a Weak Fire Whirl via Heat Feedback Mechanism," Journal of Chemical Engineering of Japan, vol. 46 no. 10, pp. 689-694, DOI: 10.1252/jcej.13we078, 2013.

[11] W. K. Chow, Z. He, Y. Gao, "Internal Fire Whirls in a Vertical Shaft," Journal of Fire Sciences, vol. 29 no. 1, pp. 71-92, DOI: 10.1177/0734904110378974, 2011.

[12] R. Dobashi, T. Okura, R. Nagaoka, Y. Hayashi, T. Mogi, "Experimental Study on Flame Height and Radiant Heat of Fire Whirls," Fire Technology, vol. 52 no. 4, pp. 1069-1080, DOI: 10.1007/s10694-015-0549-z, 2016.

[13] S. B. Hariharan, Y. Hu, M. J. Gollner, E. S. Oran, "Effects of Circulation and Buoyancy on the Transition from a Fire Whirl to a Blue Whirl," Physical Review Fluids, vol. 5 no. 10,DOI: 10.1103/physrevfluids.5.103201, 2020.

[14] S. B. Hariharan, H. F. Farahani, A. S. Rangwala, J. L. Dowling, E. S. Oran, M. J. Gollner, "Comparison of Particulate-Matter Emissions From Liquid-Fueled Pool Fires and Fire Whirls," Combustion and Flame, vol. 227, pp. 483-496, DOI: 10.1016/j.combustflame.2020.12.033, 2021.

[15] F. Battaglia, R. G. Rehm, H. R. Baum, "The Fluid Mechanics of Fire Whirls: An Inviscid Model," Physics of Fluids, vol. 12 no. 11, pp. 2859-2867, DOI: 10.1063/1.1308510, 2000.

[16] J. M. McDonough, A. Loh, "Simulation of Vorticity-Buoyancy Interactions in Fire-Whirl-Like Phenomena," Proceedings of ASME Summer Heat Transfer Conference, pp. 195-201, DOI: 10.1115/ht2003-47548, .

[17] R. Zhou, Z.-N. Wu, "Fire Whirls Due to Surrounding Flame Sources and the Influence of the Rotation Speed on the Flame Height," Journal of Fluid Mechanics, vol. 583, pp. 313-345, DOI: 10.1017/s0022112007006337, 2007.

[18] K. Kuwana, K. Sekimoto, T. Minami, T. Tashiro, K. Saito, "Scale-Model Experiments of Moving Fire Whirl over a Line Fire," Proceedings of the Combustion Institute, vol. 34 no. 2, pp. 2625-2631, DOI: 10.1016/j.proci.2012.06.092, 2013.

[19] K. Kuwana, K. Saito, F. A. Williams, "Fire-whirl Movement Along Line Fires–A Separated-Flow Rotating-Cylinder Analogy," Combustion Science and Technology, vol. 192 no. 7, pp. 1160-1172, DOI: 10.1080/00102202.2020.1712376, 2020.

[20] K. Zhou, N. Liu, P. Yin, X. Yuan, J. Jiang, "Fire Whirl Due to Interaction Between Line Fire and Cross Wind," Fire Safety Science, vol. 11, pp. 1420-1429, DOI: 10.3801/iafss.fss.11-1420, 2014.

[21] K. Zhou, N. Liu, X. Yuan, "Effect of Wind on Fire Whirl over a Line Fire," Fire Technology, vol. 52 no. 3, pp. 865-875, DOI: 10.1007/s10694-015-0507-9, 2016.

[22] K. Zhou, Y. Wang, A. Simeoni, R. Dong, "Fire Whirls Induced by a Line Fire on a Windward Slope: A Laboratory-Scale Study," International Journal of Wildland Fire, vol. 33 no. 1,DOI: 10.1071/wf23048, 2023.

[23] P. Wang, N. Liu, X. Liu, X. Yuan, "Experimental Study on Flame Wander of Fire Whirl," Fire Technology, vol. 54 no. 5, pp. 1369-1381, DOI: 10.1007/s10694-018-0734-y, 2018.

[24] M. Watanabe, K. Okamoto, "Gas Temperature Distribution and Fluctuation in a Lab-Scale Fire Whirl," Journal of Flow Control, Measurement & Visualization, vol. 11 no. 02, pp. 15-29, DOI: 10.4236/jfcmv.2023.112002, 2023.

[25] F. A. Williams, Combustion Theory, 1985.

[26] A. Nakakuki, "Heat Transfer in Small Scale Pool Fires," Combustion and Flame, vol. 96 no. 3, pp. 311-324, DOI: 10.1016/0010-2180(94)90018-3, 1994.

[27] A. Nakakuki, "Heat Transfer Mechanisms in Liquid Pool Fires," Fire Safety Journal, vol. 23 no. 4, pp. 339-363, DOI: 10.1016/0379-7112(94)90003-5, 1994.

[28] J. A. Fay, "Model of Large Pool Fires," Journal of Hazardous Materials, vol. 136 no. 2, pp. 219-232, DOI: 10.1016/j.jhazmat.2005.11.095, 2006.

[29] W. M. G. Malalasekera, H. K. Versteeg, K. Gilchrist, "A Review of Research and an Experimental Study on the Pulsation of Buoyant Diffusion Flames and Pool Fires," Fire and Materials, vol. 20 no. 6, pp. 261-271, DOI: 10.1002/(sici)1099-1018(199611)20:6%3C261::aid-fam578%3E3.3.co;2-d, 1996.

[30] D. Moreno-Boza, W. Coenen, J. Carpio, A. L. Sánchez, F. A. Williams, "On the Critical Conditions for Pool-Fire Puffing," Combustion and Flame, vol. 192, pp. 426-438, DOI: 10.1016/j.combustflame.2018.02.011, 2018.

[31] K. Zhou, N. Liu, J. S. Lozano, Y. Shan, B. Yao, K. Satoh, "Effect of Flow Circulation on Combustion Dynamics of Fire Whirl," Proceedings of the Combustion Institute, vol. 34 no. 2, pp. 2617-2624, DOI: 10.1016/j.proci.2012.06.053, 2013.