Available online at www.sciencedirect.com Proceedings of the Combustion Institute Proceedings of the Combustion Institute 34 (2013) 2607–2615 www.elsevier.com/locate/proci Experimental research on flame revolution and precession of fire whirls Jiao Lei a, Naian Liu a,⇑, Jesse S. Lozano b, Linhe Zhang a, Zhihua Deng a, Kohyu Satoh a a State Key Laboratory of Fire Science, University of Science and Technology of China, Hefei, Anhui 230026, PR China b Mechanical Engineering Department, University of California, Riverside, USA Available online 13 July 2012 Abstract This paper presents the first experimental effort to explore the large scale 3-D flame instabilities of fire whirls, including the inclined flame revolution during the transition from a general pool fire to fire whirl, and the swirling flame precession in a quasi-steady fire whirl. The experimental medium-scale fire whirls were produced by a fixed-frame facility. Experimental observations indicate that flame revolution is an important flame instability during the formation of fire whirl, showing that the entire flame is inclined and revolves around the geometrical axis of symmetry with increasing angular velocity until the critical point, without the self-rotation of the flame. It is found that the inlet velocity fluctuates synchronously with the flame revolution. As soon as the fire whirl forms, the erect swirling flame starts to precess around the geometrical axis of symmetry. Analysis indicates that during flame precession the periodic fluctuations of inlet velocity disappear and a local annular external recirculation zone (ERZ) is produced outside the flame (vortex core), while the flow is upward inside. It is found that the inlet velocities are nearly constant within the continuous flame in order to maintain a stable generating eddy. A good linear correlation exists between the average inlet velocities and average ambient circulations for all fuel pan sizes. The precession frequency is relatively stable during one test. The frequencies of flame revolution and precession are both proportional to the average inlet velocity, and the corresponding Strouhal numbers are constants of 0.42 and 0.80, respectively. The flame revolves and precesses in the same direction as the self-rotation of the fire whirl flame in all tests. The flame revolution is related to the periodical fluctuations of inlet flow, while the flame precession is considered to be linked to the occurrence of ERZ in fire whirls. Ó 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Fire whirl; Flame revolution; Flame precession; Fixed-frame facility; External recirculation zone 1. Introduction erty and life [1,2]. As indicated, the formation of a fire whirl depends on three essential conditions – a Fire whirl is a special swirling diffusion flame generating eddy, a fluid sink within the eddy, and that may occur in both urban and wildland areas some friction or drag offered to the movement of with possibility of causing great damage to prop- air in the ground boundary of the eddy [3].Ina fire whirl, the hot gas generated by the fire itself ⇑ Corresponding author. Fax: +86 551 3601669. serves as the fluid sink which entrains the ambient E-mail address: [email protected] (N. Liu). air with angular momentum from the generating 1540-7489/$ - see front matter Ó 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.proci.2012.06.126 2608 J. Lei et al. / Proceedings of the Combustion Institute 34 (2013) 2607–2615 eddy to the flame (vortex core). As summarized by frequencies of flame revolution and flame preces- Byram and Martin [3], the different types of facil- sion were extracted from data of periodic temper- ities which can be used to produce single steady ature fluctuations at specific points, and the flow fire whirls in laboratory are based on the concepts characteristics were obtained by comprehensive of a generating eddy and a fluid sink. The facilities velocity measurements. The relationships between can be generally classified into two types, depend- the instability frequencies and inlet velocity were ing on whether the generating eddy is imposed revealed. The differences of precession in fire mechanically by a spinning screen (Emmons type whirls and ordinary swirling jet flames were exam- [4,5]) or induced by the entrained air flowing ined extensively. through well-arranged spiral paths (fixed-frame type [6–9], fire-wall type [10]). Some quantitative experimental researches have been performed that 2. Experimental focused on burning rates, flame plume tempera- ture, velocity distributions, and radiation heat flux The fixed-frame fire whirl facility, located in a of fire whirls [2,4,7,8,11]. large test hall, was a square enclosure made of As usually observed in fire whirl experiments, a tempered glass, with a base dimension of distinct difference between fire whirls and general 2m 2 m and a height of 15 m (Fig. 1). The pool fires is the presence of an obvious flame wan- channel was open at the top. Each channel wall der around the geometrical centerline in experi- had a uniform 20 cm wide vertical gap at its cor- mental fire whirls, which can induce large ner to ensure that the entrained air induced by fluctuations of data, leading to great difficulties the burning flame could enter the channel, thus for the subsequent data analysis [4,8]. Emmons imparting the rotational flow necessary for fire and Ying [4] reported that the flame wander whirl formation. The base table (2 m  2 m) was involved amplitude of several flame diameters made of pine wood with a round hole (60 cm in with a lower frequency when compared to the diameter and 10 cm in depth) in the center. The large turbulent eddies. They attributed this wan- steel circular fuel pans (with diameters d of 10– der to some inherent flame instability since the 55 cm with increments of 5 cm, and a depth of facility was symmetrical and careful precautions 10 cm) were positioned in the center of the facility were taken for turbulence suppression. and placed such that their rims were flush with the Flame instability was also found by numerical top of the base table. The initial height of the simulations of fire whirls. Snegirev et al. [12] liquid fuel (n-heptane, 97%) surface was 7 cm. A reported the periodic flame wander, but without water layer was used in the larger fuel pans to further discussion. Recently Su et al. [13], using maintain the initial fuel surface height constant Fire Dynamics Simulation (FDS) for simulating and also ensure safety. The lip height has proven fire whirls in a ship engine room, found that the to have little effect on the burning rate of fire calculated angular velocity of flame wander fit whirls [8]. The fuel mass versus time was recorded well with a cube polynomial function of the inlet using an electronic balance with a precision of velocity along the four walls and increased rapidly 0.1 g. when the inlet velocity was over a critical value The temperatures were measured using Type K (6.58 m/s). However, no experimental data was (chromel–alumel) thermocouples (bead diameter: presented to compare to the numerical results. It 1 mm). They were arranged in the radial direction is doubtful that using FDS without code modifica- in intervals of 1.5 cm from r = À30 to 30 cm tion would accurately represent the complex inter- action of combustion and strong rotation of a fire Table 1 whirl. The experimental data in the present work. In appearance, the flame wander in a single a d V in ci fri V in cs fps C Rfs Rps Q quasi-steady fire whirl is characterized by its self (cm) (m/s) (Hz) (m/s) (Hz) (m2/s) (cm) (cm) (kW) rotation and the simultaneous rotation of the entire flame body around a certain geometrical 10 0.31 0.07 0.28 0.14 1.49 4.0 1.5 22.3 15 0.52 0.11 0.52 0.25 2.00 5.6 2.4 62.4 axis (generally the central axis of the pool). There- 20 0.58 0.12 0.67 0.28 2.67 7.6 2.7 102.6 fore, the flame wander should be more exactly 25 0.61 0.13 0.78 0.32 3.21 8.9 2.8 156.1 termed flame precession. Before the formation of 30 0.71 0.16 0.87 0.41 3.78 10.7 3.4 240.8 the steady fire whirl, the tilted flame would revolve 35 0.77 0.17 1.08 0.46 4.60 12.2 3.7 343.4 around the geometrical central axis, but without 40 0.92 0.20 1.19 0.48 5.08 13.6 3.6 441.5 self rotation. We call this phenomenon of flame 45 0.89 0.20 1.35 0.51 5.83 15.0 4.0 566.4 instability as flame revolution, so as to distinguish 50 1.03 0.21 1.48 0.55 6.31 15.7 4.6 669.0 it from flame precession. 55 1.13 0.24 1.64 0.62 6.81 16.9 5.1 789.4 This paper presents the first effort to a The heat release rates Q were calculated by assum- quantitatively explore the characteristics of flame ing unity of combustion efficiency in the quasi-steady instability in fire whirls. We utilized a four- state, and the effective heat of combustion for heptane is wall fixed-frame facility for experiments. The 44.6 kJ/g. J. Lei et al. / Proceedings of the Combustion Institute 34 (2013) 2607–2615 2609 (cm) -30 -20 -10 0 10 20 30 2.0m 400 4.0m Pitot tubes for tangential Pitot tubes for axial 20cm HWA velocity measurement velocity measurement 350 3.15m 300 3.0m 3.0m 2.65m 250 2.6m Liquid 2.15m Fuel pool 200 1.9m 2.0m d 2.0m 1.65m 150 1.2m 1.15m Tempered glasses wall Tempered glasses 100 0.91m T T T Entrained air -20 0 20 0.65m Corner gap 50 0.5m HWA for inlet velocity measurement 0.17m 0.15m d Raidial thermocouples Liquid Fuel, =10,15,20,...,55cm (Spacing of 1.5cm,T ~T ) -20 20 Wood base Electronic balance Fig.