- Email: [email protected]
air flame propagation in a closed combustion tube
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Effects of ignition location on premixed hydrogen/ air flame propagation in a closed combustion tube Huahua Xiao, Qiangling Duan, Lin Jiang, Jinhua Sun* State Key Laboratory of Fire Science, University of Science and Technology of China, Hefei 230027, PR China
article info
abstract
Article history:
The dynamics of premixed hydrogen/air flame ignited at different locations in a finite-size
Received 17 December 2013
closed tube is experimentally studied. The flame behaves differently in the experiments
Received in revised form
with different ignition positions. The ignition location exhibits an important impact on the
16 March 2014
flame behavior. When the flame is ignited at one of the tube ends, the heat losses to the
Accepted 22 March 2014
end wall reduce the effective thermal expansion and moderate the flame propagation and
Available online 21 April 2014
acceleration. When the ignition source is at a short distance off one of the ends, the tulip flame dynamics closely agrees with that in the theory. And both the tulip and distorted
Keywords:
tulip flames are more pronounced than those in the case with the ignition source placed at
Hydrogen/air mixture
one of the ends. Besides, the flameepressure wave coupling is quite strong and a second
Ignition location
distorted tulip flame is generated. When the ignition source is in the tube center, the flame
Flame dynamics
propagates in a much gentler way and the tulip flame can not be formed. The flame os-
Tulip flame
cillations are weaker since the flameepressure wave interaction is weaker.
Distorted tulip flame
Copyright ª 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
Flame oscillation
Introduction Flame dynamics in tubes is one of the important subjects of combustion research [1e11]. The premixed flame propagation in a hydrogen/air mixture is of particular interest in the application of safety and operation of internal combustion engines [12e16]. The flame in a tube is more complex than a freely propagating flame. The thermal expansion of the combustion products plays a key role in the early flame acceleration [2,4]. The flame behavior can be influenced by different parameters such as boundary layer, acoustic waves and heat losses at the walls. And the flame is unstable due to the intrinsic hydrodynamic instability that results from thermal expansion. Various observations on premixed flame in closed and half-open tubes were reported. One of the curious
phenomena is the tulip flame which is characterized by a shape concaved from the unburned mixture to the burnt gas [4,17e19]. The visual results of premixed flame propagation in closed tubes were first reported by Ellis [20]. In his experiments, a curved flame convex toward the unburned gas suddenly flattens and turns concave toward the burnt gas subsequent to a rapid deceleration. Generally, the flame maintains this tulip shape through the rest of the propagation. The tulip flame can be also formed in half-open tubes [4]. Many studies have been focused on the tulip flame phenomenon because of its complexity [2,4,7,10,11,18,21,22]. A variety of explanations of the tulip flame formation as well as several analytical models have been proposed. The possible explanations include: viscosity effect [11,21], interaction of flame with pressure wave [23], hydrodynamic instability [5,18], effect of vortex motion in the burnt region [10,22], and
* Corresponding author. Tel.: þ86 (0) 55163607572; fax: þ86 (0) 55163601669. E-mail addresses: [email protected], [email protected] (J. Sun). http://dx.doi.org/10.1016/j.ijhydene.2014.03.164 0360-3199/Copyright ª 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
8558
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
Taylor instability [4]. The experimental and analytical work by Clanet and Searby [4] suggests that both the pressure wave and boundary layer are unimportant for the tulip formation. Though the initiation of tulip flame could be caused by the hydrodynamic instability, the tulip flame formation can not be explained by the linear analysis of this instability [22,24]. Matalon and Metzener [10] thought that the actual tulip formation results from the vortical flow in the burnt gas and proposed a mathematical theory to support this mechanism. Clanet and Searby [4] pointed out that the tulip phenomenon is very similar to the inversion in Markstein’s experiments [25] that is caused by the interaction between a curved flame and a shock wave. Based on this similarity, they concluded that the tulip flame formation could a manifestation of the Taylor instability. However, this mechanism has not been fully supported by an analytical theory. The numerical study by Lee and Tsai [26] indicates that the heat losses at the sidewalls of the tube have an impact on the tulip flame development. Dunn-Rankin and Sawyer [17] found that the tulip flame is sensitive to the geometry of combustion tube. Clanet and Searby [4] suggested an empirical model for the tulip flame propagation. Following this work, Bychkov et al. [2] developed a theoretical model for the early flame acceleration and tulip flame evolution, and found that the tulip flame formation does not depend on the Reynolds number. Recent investigations [14,19,27] demonstrated that a distorted tulip flame can be produced after the full formation of a classical tulip flame during premixed hydrogen/air flame propagation in a closed tube with hydrogen concentration in the range of 26e64% by volume. A distorted tulip flame is the flame shape with two secondary cusps (distortions) formed on the two primary tulip lips, e.g. the flame at t ¼ 6.867 ms in Fig. 3(b) in Ref. [14]. The onset of a distorted tulip flame coincides with the sudden deceleration of both the flame front and the pressure growth. The distorted tulip flame develops into a triple tulip front as the secondary cusps (distortions) nearly arrive at the center of the primary tulip lips [14,27]. Before the disappearance of the first distorted tulip flame, a second one can be created with a cascade of distortions superimposed on the primary lips [27]. The pressure wave (acoustic wave) triggered by the first drastic flame deceleration drives the flame front to decelerate periodically and assumes an important role in the distorted tulip flame formation [14,27]. The flame displacement speed oscillates in phase with the pressure growth rate under the effect of the pressure wave. The distorted tulip flame formation could result from the Taylor instability driven by the abrupt deceleration of the flame front [27]. Although the characteristics and mechanism of the distorted tulip phenomenon has been studied, the conditions required for the generation of a distorted tulip flame are not well understood, e.g. the effects of ignition location, tube length and heat losses at the walls. The objective of the present work is to investigate the effects of ignition location on the premixed flame dynamics in a hydrogen/air mixture in a closed combustion vessel. Experiments are performed to examine the transient behavior and characteristics of premixed hydrogen/air flame ignited at different ignition locations. Then, the experimental results are compared with theoretical analysis to provide further knowledge of the influence of the ignition position.
Experimental approaches The experimental setup was detailed in the earlier work [19]. It is mainly composed of a constant volume combustion tube, a gas mixing system, a high-voltage ignition system, a highspeed schlieren cinematographic system, and a synchronization controller. The experimental combustion vessel, schematically shown in Fig. 1, is a horizontal closed duct with square crosssection 82 mm 82 mm. The length of the duct is 530 mm. The two side panels of the tube are constructed of quartz glass to provide optical access. The rest of the walls are made of TP304 stainless steel. The temporal variations of flame shape and location in the process are recorded using the high-speed schlieren photography system. The operating speed of the high-speed video camera is 15,000 frame/s. The duct is filled with a rich hydrogen/air mixture at an equivalence ratio of 1.58 (hydrogen concentration 40% by volume). The flame in the mixture is diffusively stable and can assume both tulip and distorted tulip shapes [19]. A short time delay of about 30 s is incorporated into the gas filling process to allow a quiescent mixture before ignition. The previous studies [8,14,17] indicate that this time delay is sufficient for the mixture to become quiescent. The initial temperature and pressure inside the chamber are T0 z 298 K and p0 z 101325 Pa, respectively. The combustible mixture is ignited by using a single spark gap in each test. Three different ignition sites are employed in the experiments to examine the impact of ignition position on the flame dynamics, as shown in Fig. 1. The first ignition location (ignition location 1) is at the center of the left end of the duct. The second one (ignition location 2) is at a short distance off the left end wall (5.5 cm from the left end on the duct axis). The third one (ignition location 3) is in the center of the duct (26.5 cm from the left end on the duct axis). The high-speed video camera and igniter are initiated simultaneously by the synchronization controller.
Results and discussion Development of the flame ignited at different locations Flame ignited at the center of the left end of the tube The development of the flame ignited at the center of the left end of the tube (ignition location 1) is shown in Fig. 2 through a sequence of high-speed schlieren images. Fig. 2(a) presents the early stage of the flame propagation while Fig. 2(b)
Fig. 1 e Sketch of the combustion vessel. The height of the duct is 8.2 cm.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
8559
Fig. 2 e Sequences of high-speed schlieren images of premixed hydrogen/air flame initiated at the center of the left end of the tube (ignition location 1): (a) early stage, (b) later stage.
displays the later stage. The photographs illustrate the shape variations of the traveling flame. A flame front convex toward the fresh mixture is observed before the tulip flame formation. The flame inversion occurs at about t ¼ 6.4 ms, and a tulip flame is subsequently produced, as shown in Fig. 2(a). After a well-pronounced tulip flame is established, a distorted tulip is initiated at around t ¼ 8.4 ms, as shown in Fig. 2(b). Two distortions (secondary cusps) are generated from the tips of the primary tulip lips later on, and move toward the primary tulip cusp. The distorted tulip flame collapses as the secondary cusps meet the primary one at approximately t ¼ 10.8 ms. After the disappearance of the distorted tulip flame, additional weak distortions are created on the tulip lips. Nevertheless, no remarkable secondary distorted tulip flame can be observed in the experiment.
Flame ignited at a short distance off the left end wall (on the tube axis 5.5 cm from the left end) Fig. 3 gives a series of schlieren photographs of the premixed hydrogen/air flame initiated at the second ignition location which is on the duct axis at a short distance (5.5 cm) off the left end wall. The schlieren images indicate that the flame expands spherically after ignition. Thereafter, the flame is elongated in the horizontal direction, as shown in Fig. 3(a). The flame is flattened at about t ¼ 5.2 ms, and a tulip flame is subsequently generated, as shown in Fig. 3(b). The tulip flame is more pronounced with a much deeper cusp than that produced in the case with ignition location 1, e.g. the flame
shapes at t ¼ 6.0 and 6.4 ms. The distorted tulip flame is initiated after t ¼ 6.4 ms. The distortions travel along the primary tulip lips toward the primary cusp. As the distortions approach the center of the primary tulip lips a salient distorted tulip flame is formed, e.g. the flame shape at t ¼ 7.2 ms. Like the tulip flame, the distorted tulip flame in this case is also more noticeable than that in the case of ignition location 1. The secondary distortions are produced from the tips of the primary tulip lips before the vanishing of the primary distorted tulip flame, as shown in Fig. 3(b) (e.g. the flame at t ¼ 8.0 ms). Nevertheless, the second distorted tulip flame is much less pronounced than the first one.
Flame ignited in the center of tube (on the tube axis 26.5 cm from the left end) The schlieren photographs of the developing hydrogen/air flame initiated in the center of duct are shown in Fig. 4. Note that the flame develops approximately in the same manner in the two horizontal directions, and only the flame propagating toward the right end of the duct is given here. The flame evolution is simpler than that in the cases of ignition locations 1 and 2. Following the ignition, a spherical smooth flame is generated. The flame is horizontally elongated when approaching the tube sidewalls. After the flame front reaching the sidewalls, the curvature radius increases. However, tulip flame is not formed during the burning process. The flame front is considerably wrinkled by the flame instabilities at the later stage.
8560
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
Fig. 3 e Sequences of high-speed schlieren images of premixed hydrogen/air flame ignited at a short distance of 5.5 cm from the left end (ignition location 2): (a) early stage, (b) later stage.
Dynamics of the flame front with different ignition locations Flame front dynamics with ignition located at the center of the tube left end Fig. 5 shows the flame tip position and propagation speed (displacement speed) as a function of time in the experiment with the ignition attached to the center of the left end of the tube (ignition location 1). The position is defined as the distance from the ignition point. The flame front along the centerline of the duct is treated as the flame tip before the flame inversion, and the flame front near the top wall is taken as the flame tip for the tulip and distorted tulip flames. The flame tip position is extracted from the high-speed schlieren images. The uncertainty in the measurement of the flame position is less than 3%. Remarkable periodic oscillations are clearly seen in the traces of both the location and propagation speed of the flame tip in the case with
ignition position 1. In general, the flame accelerates exponentially in the early stage due to the quick increase of the flame surface area [2,4,14]. The acceleration stops as the flame touches the sidewall of the duct. The contact of the flame with the duct sidewalls results in a rapid reduction in the flame surface area, and consequently a quick reduction in the expansion of the burnt matter. As demonstrated in Refs. [14,27], the periodic oscillating behavior results from the interaction between the flame front and pressure wave. The pressure wave is triggered by the first contact of the flame skirt with the tube sidewalls and travels back and forth inside the tube. When the pressure maximum passes the flame front in the direction from unburned mixture to burnt gas, the flame undergoes a sudden deceleration. A detailed description of the coupling between flame front and pressure wave in a closed combustion vessel could be found in previous works [14,27].
Fig. 4 e Schlieren photographs of premixed hydrogen/air flame ignited in the center of duct (ignition location 3, on the axis 26.5 cm from the left end).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
8561
smaller acceleration/deceleration and weaker flameepressure interaction. The smaller deceleration leads to less pronounced tulip and distorted tulip flames since the both phenomena are driven by flame deceleration. It means that the heat losses at the left end wall can significantly affect the flame behavior and characteristics.
Flame front dynamics with ignition location on the tube axis 26.5 cm from the left end (center of the tube)
Fig. 5 e History of flame tip position and propagation speed in the experiment with ignition location attached to the center of the left end of the tube (ignition location 1).
Flame front dynamics with ignition located on the tube axis 5.5 cm from the left end The time history of flame tip position and propagation speed in the case with the ignition located on the tube axis 5.5 cm from the left end (ignition location 2) is given in Fig. 6. It can be seen that both the flame position and speed oscillates significantly, similar to those of the case with ignition location 1 (see Fig. 5). The reason for the periodic flame oscillations is also due to the coupling between the flame front and pressure wave, as remarked in Section 3.2.1. With respect to the case with ignition location 2, the numerical simulation with the same ignition location in Ref. [14] shows that the flame front reaches the left end wall after the flame has touched the duct sidewalls, which implies the heat losses at left end wall can not influence the early flame propagation. The flame propagates much faster approximately after t ¼ 2.0 ms when the ignition source is placed at a short distance (5.5 cm) off the left end wall (ignition location 2) than that in the case of ignition location 1, as shown in Fig. 6. It is well known that the thermal expansion of the combustion products plays a very important role in the flame propagation and acceleration [2,4]. When the flame is ignited at the left end of the duct, the cooling effect of the left end wall reduces the effective thermal expansion and consequently leads to smaller flame propagation speed,
Fig. 6 e Position and propagation speed of flame tip as a function of time in the case with ignition located on the tube axis 5.5 cm from the left end (ignition location 2).
Fig. 7 presents the flame tip position and propagation speed of the flame tip versus time in the experiment with ignition site placed on the tube axis 26.5 cm from the left end (center of the tube, ignition location 3). The oscillating behavior is much weaker in the case of ignition position 3, as shown in Fig. 7. The flame propagation speed is also much smaller in the later stages, as expected. This is due to the fact that the inability of the gases to expand freely and flow toward the right end of the duct considerably moderates the flame acceleration since the flame front is much closer to the right closed end. The flame deceleration in the experiment with ignition location 3 is smaller, and thus the interaction of flame front with pressure wave is weaker, leading to smaller flame oscillations.
Comparisons of the experimental results with theoretical prediction Four stages can be distinguished in the tulip flame propagation according to Clanet and Searby [4], i.e. spherical/hemispherical flame, finger-shaped flame, flame with its skirt touching the tube sidewalls, and tulip flame. Based on the flame mechanism proposed by Clanet and Searby, Bychkov et al. [2] suggested a theoretical model for the tulip flame evolution in long half-open tubes. In the study by Bychkov et al. [2], the flame is initiated at the center of the closed end and moves toward the open one under adiabatic boundary conditions. It has been put into evidence that the early flame acceleration and tulip evolution in a closed tube are pretty similar to those in a half-open tube with adiabatic walls when the flame is ignited at a short distance of 5.5 cm from the left end on the axis (the same as ignition location 2) [14]. The theory by Bychkov is compatible with the tulip flame propagation in a closed tube at the same dimensions as that in the
Fig. 7 e Position and propagation speed of flame tip versus time in the case with ignition location on the tube axis 26.5 cm from the left end (center of the tube, ignition location 3).
8562
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
present work. Following the analytical model, the flame develops from a spherical kernel to a finger-shaped front when the flame skirt has traveled about halfway to the sidewall. The time of the transition from spherical to finger flame tsph is given in the theory as: (1) tsph ¼ 1=ð2sÞ$H=SL0 ; pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where s ¼ E$ðE 1Þ, H is half of the tube width, E ¼ 6.5 is the expansion ratio of combustion products, and SL0 ¼ 2.42 m/s [28] is the laminar burning velocity. The time when the flame reaches the sidewall of the duct twall is calculated as: twall ¼ 1=ð2sÞ$H=SL0 $ln½ðE þ sÞ=ðE sÞ;
(2)
The onset time of tulip flame (flame inversion time) is estimated as: ttulip ¼ twall þ tinv ;
(3)
where tinv¼H$l/(2s$SL0) denotes the interval between twall and ttulip, and l is a model coefficient comparable to unity which is set as 1.0 in this work as in Ref. [2]. The location of flame tip in the early acceleration stage can be predicted as follows: Ztip ¼ E=ð2sÞ$sinhð2sSL0 t=HÞ$H;
(4)
The comparison between the characteristic times predicted by the theory and the experimental results is given in Table 1. The flame expands freely for a short period after ignition at the first stage, unaffected by the tube sidewalls. It is thus no doubt that the spherical flame time with ignition location 2 is equal to that with ignition location 3, as shown in Table 1. The flame is elongated under the confinement of the tube sidewalls at the second stage. The flame accelerates fast due to the exponential increase of the flame surface area caused by the elongation. The flame acceleration stage (second stage) terminates as the flame arrives at the duct sidewalls. The flame surface area is reduced quickly at the third stage, resulting in a rapid flame deceleration. The inversion of the flame front takes place subsequent to the flame deceleration. In the experiment, the time twall is determined as the time when the flame acceleration stops. The termination time of the flame acceleration twall depends on only the initial laminar flame speed and expansion ratio. Therefore, the time twall with ignition location 2 is equal to that with ignition location 3, as shown in Table 1. As mentioned above, the effective thermal expansion is weakened by the effect of heat losses at the left end wall of the duct when the flame is ignited at the center of the left end. Therefore, the time twall with ignition location 1 is larger than in the theory and the cases with ignition locations 2 and 3. Similarly, the tulip initiation time is also larger in the experiment with ignition location 1. Note that although the flame front development at the first
stage in the case of ignition location 1 is not captured in the present study, it is can be expected following Eq. (1) that the time tsph with a reduced expansion coefficient should be larger than the values in the cases of ignition locations 2 and 3. Fig. 8 shows the position of the flame tip in the first and second stages in a comparison between the theory and experiments with different ignition locations. It is shown that the flame tip position in the experiment with ignition location 2 is in good agreement with the theoretical calculation. The flame tip position in the case with ignition location 1 is smaller than that in the theory and the discrepancy becomes larger with time. As remarked above, this is caused by the reduction of the effective thermal expansion due to the heat transfer through the left end wall. The flame tip position in the case with ignition location 3 is approximately equal to that in the theory at the first stage. Starting from the second flame stage, the flame tip location is much smaller, even smaller than for ignition 1 in the later phase of the second stage due to the inability of the unburned mixture to move freely toward the right closed end.
Conclusions An experimental investigation of premixed hydrogen/air flame dynamics in a finite-size closed combustion tube has been presented. High-speed schlieren photometry was employed to obtain an understanding of the flame behavior and characteristics in the experiments with different ignition positions. Comparison between the experimental results and theoretical predictions was carried out to provide an insight into the effects of the ignition location on the flame propagation. The following conclusions have been obtained: (1) It was found that the ignition location has a significant influence on the flame dynamics, including the flame shape changes and arrival times. When the flame is ignited at one of the ends (left end) of the duct, the flame can develop into tulip and distorted tulip flames. The heat losses to the left end wall reduce the effective thermal expansion of the combustion products and
Table 1 e Characteristic times in the theory and the experiments with different ignition positions. Theoretical Ignition location 1 Ignition location 2 Ignition location 3
tsph(ms)
twall(ms)
ttulip(ms)
1.2 e 1.133 1.133
4.28 4.333 4.067 4.067
5.7 6.4 5.2 e
Fig. 8 e Comparison of flame tip position at the early stages between the theory and experiments with different ignition locations.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
make the flame acceleration noticeably weaker. And both the contact of the flame with the sidewalls and the initiation of the tulip flame occur later than those in the theory. (2) When the ignition location is at a short distance (5.5 cm) off the one of the tube ends (left end), the flame acceleration and characteristics agree well with the theoretical predictions since the heat transfer through left end wall can not affect the early flame development. In addition, both the tulip and distorted tulip flames are more pronounced than those in the experiment with the flame ignited at the left end, and a second distorted tulip flame is produced. (3) When the flame is initiated in the center of duct, the flame propagation is largely restrained due to the inability of the mixture to expand freely toward the end wall of the duct. The flame acceleration is much weaker, and no tulip flame or distorted tulip flame can be formed. (4) The coupling of the flame with the pressure wave is very strong when the ignition source is at a short distance (5.5 cm) off the left end wall, and leads to very drastic periodic flame oscillations. The interaction of the flame with pressure wave in the cases with ignition source located at the left end wall and at the center of the duct are weaker, causing gentler flame oscillations, especially for the case with ignition location at the duct center.
Acknowledgments The authors thank the financial supports provided by National Natural Science Foundation of China (No. 51376174), and the Chinese Postdoctoral International Exchange Program (2013).
references
[1] Akkerman VB, Bychkov VV, Bastiaans RJM, de Goey LPH, van Oijen JA, Eriksson LE. Flow-flame interaction in a closed chamber. Phys Fluids 2008;20:055107. [2] Bychkov V, Akkerman V, Fru G, Petchenko A, Eriksson LE. Flame acceleration in the early stages of burning in tubes. Combust Flame 2007;150:263e76. [3] Ciccarelli G, Dorofeev S. Flame acceleration and transition to detonation in ducts. Prog Energy Combust Sci 2008;34:499e550. [4] Clanet C, Searby G. On the ‘tulip flame’ phenomenon. Combust Flame 1996;105:225e38. [5] Dold JW, Joulin G. An evolution equation modeling inversion of tulip flames. Combust Flame 1995;100:450e6. [6] Dorofeev SB. Flame acceleration and explosion safety applications. Proc Combust Inst 2011;33:2161e75.
8563
[7] Dunn-Rankin D, Barr PK, Sawyer RF. Numerical and experimental study of “tulip” flame formation in a closed vessel. Proc Combust Inst 1986;21:1291e301. [8] Fairweather M, Hargrave GK, Ibrahim SS, Walker DG. Studies of premixed flame propagation in explosion tubes. Combust Flame 1999;116:504e18. [9] Gonzalez M. Acoustic instability of a premixed flame propagating in a tube. Combust Flame 1996;107:245e59. [10] Matalon M, Metzener P. The propagation of premixed flames in closed tubes. J Fluid Mech 1997;336:331e50. [11] Starke R, Roth P. An experimental investigation of flame behavior during cylindrical vessel explosions. Combust Flame 1986;66:249e59. [12] Bychkov VV, Liberman MA. Dynamics and stability of premixed flames. Phys Rep 2000;325:116e237. [13] Verhelst S, Wallner T. Hydrogen-fueled internal combustion engines. Prog Energy Combust Sci 2009;39:490e527. [14] Xiao HH, Makarov D, Sun JH, Molkov V. Experimental and numerical investigation of premixed flame propagation with distorted tulip shape in a closed duct. Combust Flame 2012;159:1523e38. [15] Molkov V, Verbecke F, Makarov D. LES of hydrogen-air deflagrations in a 78.5-m tunnel. Combust Sci Technol 2008;180:796e808. [16] Dorofeev SB. Hydrogen flames in tubes: critical run-up distances. Int J Hydrogen Energy 2009;34:5832e7. [17] Dunn-Rankin D, Sawyer RF. Tulip flames: changes in shape of premixed flames propagating in closed tubes. Exp Fluids 1998;24:130e40. [18] Gonzalez M, Borghi R, Saouab A. Interaction of a flame front with its self-generated flow in an enclosure e the fulip flame phenomenon. Combust Flame 1992;88:201e20. [19] Xiao HH, Wang QS, He XC, Sun JH, Shen XB. Experimental study on the behaviors and shape changes of premixed hydrogen-air flames propagating in horizontal duct. Int J Hydrogen Energy 2011;36:6325e36. [20] Ellis OCdeC. Flame movement in gaseous explosive mixtures. Fuel Sci 1928;7:502e8. [21] Marra FS, Continillo G. Numerical study of premixed laminar flame propagation in a closed tube with a full NaviereStokes approach. Proc Combust Inst 1996;26:907e13. [22] Metzener P, Matalon M. Premixed flames in closed cylindrical tubes. Combust Theor Model 2001;5:463e83. [23] Guenoche H. Flame propagation in tubes and in closed vessels. In: Markstein GH, editor. Nonsteady flame propagation. New York: Pergamon Press; 1964. p. 107. [24] Matalon M, Mcgreevy JL. The initial development of a tulip flame. Proc Combust Inst; 1994:1407e13. [25] Markstein GH. A shock-tube study of flame front-pressure wave interaction. Proc Combust Inst 1956;6:387e98. [26] Lee ST, Tsai CH. Numerical investigation of steady laminar flame propagation in a circular tube. Combust Flame 1994;99:484e90. [27] Xiao HH, Wang QS, Shen XB, Guo S, Sun JH. An experimental study of distorted tulip flame formation in a closed duct. Combust Flame 2013;160:1725e8. [28] Lamoureux N, Djebaili-Chaumeix N, Paillard CE. Laminar flame velocity determination for H2eaireHeeCO2 mixtures using the spherical bomb method. Exp Therm Fluid Sci 2003;27:385e93.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Effects of ignition location on premixed hydrogen/ air flame propagation in a closed combustion tube Huahua Xiao, Qiangling Duan, Lin Jiang, Jinhua Sun* State Key Laboratory of Fire Science, University of Science and Technology of China, Hefei 230027, PR China
article info
abstract
Article history:
The dynamics of premixed hydrogen/air flame ignited at different locations in a finite-size
Received 17 December 2013
closed tube is experimentally studied. The flame behaves differently in the experiments
Received in revised form
with different ignition positions. The ignition location exhibits an important impact on the
16 March 2014
flame behavior. When the flame is ignited at one of the tube ends, the heat losses to the
Accepted 22 March 2014
end wall reduce the effective thermal expansion and moderate the flame propagation and
Available online 21 April 2014
acceleration. When the ignition source is at a short distance off one of the ends, the tulip flame dynamics closely agrees with that in the theory. And both the tulip and distorted
Keywords:
tulip flames are more pronounced than those in the case with the ignition source placed at
Hydrogen/air mixture
one of the ends. Besides, the flameepressure wave coupling is quite strong and a second
Ignition location
distorted tulip flame is generated. When the ignition source is in the tube center, the flame
Flame dynamics
propagates in a much gentler way and the tulip flame can not be formed. The flame os-
Tulip flame
cillations are weaker since the flameepressure wave interaction is weaker.
Distorted tulip flame
Copyright ª 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
Flame oscillation
Introduction Flame dynamics in tubes is one of the important subjects of combustion research [1e11]. The premixed flame propagation in a hydrogen/air mixture is of particular interest in the application of safety and operation of internal combustion engines [12e16]. The flame in a tube is more complex than a freely propagating flame. The thermal expansion of the combustion products plays a key role in the early flame acceleration [2,4]. The flame behavior can be influenced by different parameters such as boundary layer, acoustic waves and heat losses at the walls. And the flame is unstable due to the intrinsic hydrodynamic instability that results from thermal expansion. Various observations on premixed flame in closed and half-open tubes were reported. One of the curious
phenomena is the tulip flame which is characterized by a shape concaved from the unburned mixture to the burnt gas [4,17e19]. The visual results of premixed flame propagation in closed tubes were first reported by Ellis [20]. In his experiments, a curved flame convex toward the unburned gas suddenly flattens and turns concave toward the burnt gas subsequent to a rapid deceleration. Generally, the flame maintains this tulip shape through the rest of the propagation. The tulip flame can be also formed in half-open tubes [4]. Many studies have been focused on the tulip flame phenomenon because of its complexity [2,4,7,10,11,18,21,22]. A variety of explanations of the tulip flame formation as well as several analytical models have been proposed. The possible explanations include: viscosity effect [11,21], interaction of flame with pressure wave [23], hydrodynamic instability [5,18], effect of vortex motion in the burnt region [10,22], and
* Corresponding author. Tel.: þ86 (0) 55163607572; fax: þ86 (0) 55163601669. E-mail addresses: [email protected], [email protected] (J. Sun). http://dx.doi.org/10.1016/j.ijhydene.2014.03.164 0360-3199/Copyright ª 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
8558
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
Taylor instability [4]. The experimental and analytical work by Clanet and Searby [4] suggests that both the pressure wave and boundary layer are unimportant for the tulip formation. Though the initiation of tulip flame could be caused by the hydrodynamic instability, the tulip flame formation can not be explained by the linear analysis of this instability [22,24]. Matalon and Metzener [10] thought that the actual tulip formation results from the vortical flow in the burnt gas and proposed a mathematical theory to support this mechanism. Clanet and Searby [4] pointed out that the tulip phenomenon is very similar to the inversion in Markstein’s experiments [25] that is caused by the interaction between a curved flame and a shock wave. Based on this similarity, they concluded that the tulip flame formation could a manifestation of the Taylor instability. However, this mechanism has not been fully supported by an analytical theory. The numerical study by Lee and Tsai [26] indicates that the heat losses at the sidewalls of the tube have an impact on the tulip flame development. Dunn-Rankin and Sawyer [17] found that the tulip flame is sensitive to the geometry of combustion tube. Clanet and Searby [4] suggested an empirical model for the tulip flame propagation. Following this work, Bychkov et al. [2] developed a theoretical model for the early flame acceleration and tulip flame evolution, and found that the tulip flame formation does not depend on the Reynolds number. Recent investigations [14,19,27] demonstrated that a distorted tulip flame can be produced after the full formation of a classical tulip flame during premixed hydrogen/air flame propagation in a closed tube with hydrogen concentration in the range of 26e64% by volume. A distorted tulip flame is the flame shape with two secondary cusps (distortions) formed on the two primary tulip lips, e.g. the flame at t ¼ 6.867 ms in Fig. 3(b) in Ref. [14]. The onset of a distorted tulip flame coincides with the sudden deceleration of both the flame front and the pressure growth. The distorted tulip flame develops into a triple tulip front as the secondary cusps (distortions) nearly arrive at the center of the primary tulip lips [14,27]. Before the disappearance of the first distorted tulip flame, a second one can be created with a cascade of distortions superimposed on the primary lips [27]. The pressure wave (acoustic wave) triggered by the first drastic flame deceleration drives the flame front to decelerate periodically and assumes an important role in the distorted tulip flame formation [14,27]. The flame displacement speed oscillates in phase with the pressure growth rate under the effect of the pressure wave. The distorted tulip flame formation could result from the Taylor instability driven by the abrupt deceleration of the flame front [27]. Although the characteristics and mechanism of the distorted tulip phenomenon has been studied, the conditions required for the generation of a distorted tulip flame are not well understood, e.g. the effects of ignition location, tube length and heat losses at the walls. The objective of the present work is to investigate the effects of ignition location on the premixed flame dynamics in a hydrogen/air mixture in a closed combustion vessel. Experiments are performed to examine the transient behavior and characteristics of premixed hydrogen/air flame ignited at different ignition locations. Then, the experimental results are compared with theoretical analysis to provide further knowledge of the influence of the ignition position.
Experimental approaches The experimental setup was detailed in the earlier work [19]. It is mainly composed of a constant volume combustion tube, a gas mixing system, a high-voltage ignition system, a highspeed schlieren cinematographic system, and a synchronization controller. The experimental combustion vessel, schematically shown in Fig. 1, is a horizontal closed duct with square crosssection 82 mm 82 mm. The length of the duct is 530 mm. The two side panels of the tube are constructed of quartz glass to provide optical access. The rest of the walls are made of TP304 stainless steel. The temporal variations of flame shape and location in the process are recorded using the high-speed schlieren photography system. The operating speed of the high-speed video camera is 15,000 frame/s. The duct is filled with a rich hydrogen/air mixture at an equivalence ratio of 1.58 (hydrogen concentration 40% by volume). The flame in the mixture is diffusively stable and can assume both tulip and distorted tulip shapes [19]. A short time delay of about 30 s is incorporated into the gas filling process to allow a quiescent mixture before ignition. The previous studies [8,14,17] indicate that this time delay is sufficient for the mixture to become quiescent. The initial temperature and pressure inside the chamber are T0 z 298 K and p0 z 101325 Pa, respectively. The combustible mixture is ignited by using a single spark gap in each test. Three different ignition sites are employed in the experiments to examine the impact of ignition position on the flame dynamics, as shown in Fig. 1. The first ignition location (ignition location 1) is at the center of the left end of the duct. The second one (ignition location 2) is at a short distance off the left end wall (5.5 cm from the left end on the duct axis). The third one (ignition location 3) is in the center of the duct (26.5 cm from the left end on the duct axis). The high-speed video camera and igniter are initiated simultaneously by the synchronization controller.
Results and discussion Development of the flame ignited at different locations Flame ignited at the center of the left end of the tube The development of the flame ignited at the center of the left end of the tube (ignition location 1) is shown in Fig. 2 through a sequence of high-speed schlieren images. Fig. 2(a) presents the early stage of the flame propagation while Fig. 2(b)
Fig. 1 e Sketch of the combustion vessel. The height of the duct is 8.2 cm.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
8559
Fig. 2 e Sequences of high-speed schlieren images of premixed hydrogen/air flame initiated at the center of the left end of the tube (ignition location 1): (a) early stage, (b) later stage.
displays the later stage. The photographs illustrate the shape variations of the traveling flame. A flame front convex toward the fresh mixture is observed before the tulip flame formation. The flame inversion occurs at about t ¼ 6.4 ms, and a tulip flame is subsequently produced, as shown in Fig. 2(a). After a well-pronounced tulip flame is established, a distorted tulip is initiated at around t ¼ 8.4 ms, as shown in Fig. 2(b). Two distortions (secondary cusps) are generated from the tips of the primary tulip lips later on, and move toward the primary tulip cusp. The distorted tulip flame collapses as the secondary cusps meet the primary one at approximately t ¼ 10.8 ms. After the disappearance of the distorted tulip flame, additional weak distortions are created on the tulip lips. Nevertheless, no remarkable secondary distorted tulip flame can be observed in the experiment.
Flame ignited at a short distance off the left end wall (on the tube axis 5.5 cm from the left end) Fig. 3 gives a series of schlieren photographs of the premixed hydrogen/air flame initiated at the second ignition location which is on the duct axis at a short distance (5.5 cm) off the left end wall. The schlieren images indicate that the flame expands spherically after ignition. Thereafter, the flame is elongated in the horizontal direction, as shown in Fig. 3(a). The flame is flattened at about t ¼ 5.2 ms, and a tulip flame is subsequently generated, as shown in Fig. 3(b). The tulip flame is more pronounced with a much deeper cusp than that produced in the case with ignition location 1, e.g. the flame
shapes at t ¼ 6.0 and 6.4 ms. The distorted tulip flame is initiated after t ¼ 6.4 ms. The distortions travel along the primary tulip lips toward the primary cusp. As the distortions approach the center of the primary tulip lips a salient distorted tulip flame is formed, e.g. the flame shape at t ¼ 7.2 ms. Like the tulip flame, the distorted tulip flame in this case is also more noticeable than that in the case of ignition location 1. The secondary distortions are produced from the tips of the primary tulip lips before the vanishing of the primary distorted tulip flame, as shown in Fig. 3(b) (e.g. the flame at t ¼ 8.0 ms). Nevertheless, the second distorted tulip flame is much less pronounced than the first one.
Flame ignited in the center of tube (on the tube axis 26.5 cm from the left end) The schlieren photographs of the developing hydrogen/air flame initiated in the center of duct are shown in Fig. 4. Note that the flame develops approximately in the same manner in the two horizontal directions, and only the flame propagating toward the right end of the duct is given here. The flame evolution is simpler than that in the cases of ignition locations 1 and 2. Following the ignition, a spherical smooth flame is generated. The flame is horizontally elongated when approaching the tube sidewalls. After the flame front reaching the sidewalls, the curvature radius increases. However, tulip flame is not formed during the burning process. The flame front is considerably wrinkled by the flame instabilities at the later stage.
8560
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
Fig. 3 e Sequences of high-speed schlieren images of premixed hydrogen/air flame ignited at a short distance of 5.5 cm from the left end (ignition location 2): (a) early stage, (b) later stage.
Dynamics of the flame front with different ignition locations Flame front dynamics with ignition located at the center of the tube left end Fig. 5 shows the flame tip position and propagation speed (displacement speed) as a function of time in the experiment with the ignition attached to the center of the left end of the tube (ignition location 1). The position is defined as the distance from the ignition point. The flame front along the centerline of the duct is treated as the flame tip before the flame inversion, and the flame front near the top wall is taken as the flame tip for the tulip and distorted tulip flames. The flame tip position is extracted from the high-speed schlieren images. The uncertainty in the measurement of the flame position is less than 3%. Remarkable periodic oscillations are clearly seen in the traces of both the location and propagation speed of the flame tip in the case with
ignition position 1. In general, the flame accelerates exponentially in the early stage due to the quick increase of the flame surface area [2,4,14]. The acceleration stops as the flame touches the sidewall of the duct. The contact of the flame with the duct sidewalls results in a rapid reduction in the flame surface area, and consequently a quick reduction in the expansion of the burnt matter. As demonstrated in Refs. [14,27], the periodic oscillating behavior results from the interaction between the flame front and pressure wave. The pressure wave is triggered by the first contact of the flame skirt with the tube sidewalls and travels back and forth inside the tube. When the pressure maximum passes the flame front in the direction from unburned mixture to burnt gas, the flame undergoes a sudden deceleration. A detailed description of the coupling between flame front and pressure wave in a closed combustion vessel could be found in previous works [14,27].
Fig. 4 e Schlieren photographs of premixed hydrogen/air flame ignited in the center of duct (ignition location 3, on the axis 26.5 cm from the left end).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
8561
smaller acceleration/deceleration and weaker flameepressure interaction. The smaller deceleration leads to less pronounced tulip and distorted tulip flames since the both phenomena are driven by flame deceleration. It means that the heat losses at the left end wall can significantly affect the flame behavior and characteristics.
Flame front dynamics with ignition location on the tube axis 26.5 cm from the left end (center of the tube)
Fig. 5 e History of flame tip position and propagation speed in the experiment with ignition location attached to the center of the left end of the tube (ignition location 1).
Flame front dynamics with ignition located on the tube axis 5.5 cm from the left end The time history of flame tip position and propagation speed in the case with the ignition located on the tube axis 5.5 cm from the left end (ignition location 2) is given in Fig. 6. It can be seen that both the flame position and speed oscillates significantly, similar to those of the case with ignition location 1 (see Fig. 5). The reason for the periodic flame oscillations is also due to the coupling between the flame front and pressure wave, as remarked in Section 3.2.1. With respect to the case with ignition location 2, the numerical simulation with the same ignition location in Ref. [14] shows that the flame front reaches the left end wall after the flame has touched the duct sidewalls, which implies the heat losses at left end wall can not influence the early flame propagation. The flame propagates much faster approximately after t ¼ 2.0 ms when the ignition source is placed at a short distance (5.5 cm) off the left end wall (ignition location 2) than that in the case of ignition location 1, as shown in Fig. 6. It is well known that the thermal expansion of the combustion products plays a very important role in the flame propagation and acceleration [2,4]. When the flame is ignited at the left end of the duct, the cooling effect of the left end wall reduces the effective thermal expansion and consequently leads to smaller flame propagation speed,
Fig. 6 e Position and propagation speed of flame tip as a function of time in the case with ignition located on the tube axis 5.5 cm from the left end (ignition location 2).
Fig. 7 presents the flame tip position and propagation speed of the flame tip versus time in the experiment with ignition site placed on the tube axis 26.5 cm from the left end (center of the tube, ignition location 3). The oscillating behavior is much weaker in the case of ignition position 3, as shown in Fig. 7. The flame propagation speed is also much smaller in the later stages, as expected. This is due to the fact that the inability of the gases to expand freely and flow toward the right end of the duct considerably moderates the flame acceleration since the flame front is much closer to the right closed end. The flame deceleration in the experiment with ignition location 3 is smaller, and thus the interaction of flame front with pressure wave is weaker, leading to smaller flame oscillations.
Comparisons of the experimental results with theoretical prediction Four stages can be distinguished in the tulip flame propagation according to Clanet and Searby [4], i.e. spherical/hemispherical flame, finger-shaped flame, flame with its skirt touching the tube sidewalls, and tulip flame. Based on the flame mechanism proposed by Clanet and Searby, Bychkov et al. [2] suggested a theoretical model for the tulip flame evolution in long half-open tubes. In the study by Bychkov et al. [2], the flame is initiated at the center of the closed end and moves toward the open one under adiabatic boundary conditions. It has been put into evidence that the early flame acceleration and tulip evolution in a closed tube are pretty similar to those in a half-open tube with adiabatic walls when the flame is ignited at a short distance of 5.5 cm from the left end on the axis (the same as ignition location 2) [14]. The theory by Bychkov is compatible with the tulip flame propagation in a closed tube at the same dimensions as that in the
Fig. 7 e Position and propagation speed of flame tip versus time in the case with ignition location on the tube axis 26.5 cm from the left end (center of the tube, ignition location 3).
8562
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
present work. Following the analytical model, the flame develops from a spherical kernel to a finger-shaped front when the flame skirt has traveled about halfway to the sidewall. The time of the transition from spherical to finger flame tsph is given in the theory as: (1) tsph ¼ 1=ð2sÞ$H=SL0 ; pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where s ¼ E$ðE 1Þ, H is half of the tube width, E ¼ 6.5 is the expansion ratio of combustion products, and SL0 ¼ 2.42 m/s [28] is the laminar burning velocity. The time when the flame reaches the sidewall of the duct twall is calculated as: twall ¼ 1=ð2sÞ$H=SL0 $ln½ðE þ sÞ=ðE sÞ;
(2)
The onset time of tulip flame (flame inversion time) is estimated as: ttulip ¼ twall þ tinv ;
(3)
where tinv¼H$l/(2s$SL0) denotes the interval between twall and ttulip, and l is a model coefficient comparable to unity which is set as 1.0 in this work as in Ref. [2]. The location of flame tip in the early acceleration stage can be predicted as follows: Ztip ¼ E=ð2sÞ$sinhð2sSL0 t=HÞ$H;
(4)
The comparison between the characteristic times predicted by the theory and the experimental results is given in Table 1. The flame expands freely for a short period after ignition at the first stage, unaffected by the tube sidewalls. It is thus no doubt that the spherical flame time with ignition location 2 is equal to that with ignition location 3, as shown in Table 1. The flame is elongated under the confinement of the tube sidewalls at the second stage. The flame accelerates fast due to the exponential increase of the flame surface area caused by the elongation. The flame acceleration stage (second stage) terminates as the flame arrives at the duct sidewalls. The flame surface area is reduced quickly at the third stage, resulting in a rapid flame deceleration. The inversion of the flame front takes place subsequent to the flame deceleration. In the experiment, the time twall is determined as the time when the flame acceleration stops. The termination time of the flame acceleration twall depends on only the initial laminar flame speed and expansion ratio. Therefore, the time twall with ignition location 2 is equal to that with ignition location 3, as shown in Table 1. As mentioned above, the effective thermal expansion is weakened by the effect of heat losses at the left end wall of the duct when the flame is ignited at the center of the left end. Therefore, the time twall with ignition location 1 is larger than in the theory and the cases with ignition locations 2 and 3. Similarly, the tulip initiation time is also larger in the experiment with ignition location 1. Note that although the flame front development at the first
stage in the case of ignition location 1 is not captured in the present study, it is can be expected following Eq. (1) that the time tsph with a reduced expansion coefficient should be larger than the values in the cases of ignition locations 2 and 3. Fig. 8 shows the position of the flame tip in the first and second stages in a comparison between the theory and experiments with different ignition locations. It is shown that the flame tip position in the experiment with ignition location 2 is in good agreement with the theoretical calculation. The flame tip position in the case with ignition location 1 is smaller than that in the theory and the discrepancy becomes larger with time. As remarked above, this is caused by the reduction of the effective thermal expansion due to the heat transfer through the left end wall. The flame tip position in the case with ignition location 3 is approximately equal to that in the theory at the first stage. Starting from the second flame stage, the flame tip location is much smaller, even smaller than for ignition 1 in the later phase of the second stage due to the inability of the unburned mixture to move freely toward the right closed end.
Conclusions An experimental investigation of premixed hydrogen/air flame dynamics in a finite-size closed combustion tube has been presented. High-speed schlieren photometry was employed to obtain an understanding of the flame behavior and characteristics in the experiments with different ignition positions. Comparison between the experimental results and theoretical predictions was carried out to provide an insight into the effects of the ignition location on the flame propagation. The following conclusions have been obtained: (1) It was found that the ignition location has a significant influence on the flame dynamics, including the flame shape changes and arrival times. When the flame is ignited at one of the ends (left end) of the duct, the flame can develop into tulip and distorted tulip flames. The heat losses to the left end wall reduce the effective thermal expansion of the combustion products and
Table 1 e Characteristic times in the theory and the experiments with different ignition positions. Theoretical Ignition location 1 Ignition location 2 Ignition location 3
tsph(ms)
twall(ms)
ttulip(ms)
1.2 e 1.133 1.133
4.28 4.333 4.067 4.067
5.7 6.4 5.2 e
Fig. 8 e Comparison of flame tip position at the early stages between the theory and experiments with different ignition locations.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 8 5 5 7 e8 5 6 3
make the flame acceleration noticeably weaker. And both the contact of the flame with the sidewalls and the initiation of the tulip flame occur later than those in the theory. (2) When the ignition location is at a short distance (5.5 cm) off the one of the tube ends (left end), the flame acceleration and characteristics agree well with the theoretical predictions since the heat transfer through left end wall can not affect the early flame development. In addition, both the tulip and distorted tulip flames are more pronounced than those in the experiment with the flame ignited at the left end, and a second distorted tulip flame is produced. (3) When the flame is initiated in the center of duct, the flame propagation is largely restrained due to the inability of the mixture to expand freely toward the end wall of the duct. The flame acceleration is much weaker, and no tulip flame or distorted tulip flame can be formed. (4) The coupling of the flame with the pressure wave is very strong when the ignition source is at a short distance (5.5 cm) off the left end wall, and leads to very drastic periodic flame oscillations. The interaction of the flame with pressure wave in the cases with ignition source located at the left end wall and at the center of the duct are weaker, causing gentler flame oscillations, especially for the case with ignition location at the duct center.
Acknowledgments The authors thank the financial supports provided by National Natural Science Foundation of China (No. 51376174), and the Chinese Postdoctoral International Exchange Program (2013).
references
[1] Akkerman VB, Bychkov VV, Bastiaans RJM, de Goey LPH, van Oijen JA, Eriksson LE. Flow-flame interaction in a closed chamber. Phys Fluids 2008;20:055107. [2] Bychkov V, Akkerman V, Fru G, Petchenko A, Eriksson LE. Flame acceleration in the early stages of burning in tubes. Combust Flame 2007;150:263e76. [3] Ciccarelli G, Dorofeev S. Flame acceleration and transition to detonation in ducts. Prog Energy Combust Sci 2008;34:499e550. [4] Clanet C, Searby G. On the ‘tulip flame’ phenomenon. Combust Flame 1996;105:225e38. [5] Dold JW, Joulin G. An evolution equation modeling inversion of tulip flames. Combust Flame 1995;100:450e6. [6] Dorofeev SB. Flame acceleration and explosion safety applications. Proc Combust Inst 2011;33:2161e75.
8563
[7] Dunn-Rankin D, Barr PK, Sawyer RF. Numerical and experimental study of “tulip” flame formation in a closed vessel. Proc Combust Inst 1986;21:1291e301. [8] Fairweather M, Hargrave GK, Ibrahim SS, Walker DG. Studies of premixed flame propagation in explosion tubes. Combust Flame 1999;116:504e18. [9] Gonzalez M. Acoustic instability of a premixed flame propagating in a tube. Combust Flame 1996;107:245e59. [10] Matalon M, Metzener P. The propagation of premixed flames in closed tubes. J Fluid Mech 1997;336:331e50. [11] Starke R, Roth P. An experimental investigation of flame behavior during cylindrical vessel explosions. Combust Flame 1986;66:249e59. [12] Bychkov VV, Liberman MA. Dynamics and stability of premixed flames. Phys Rep 2000;325:116e237. [13] Verhelst S, Wallner T. Hydrogen-fueled internal combustion engines. Prog Energy Combust Sci 2009;39:490e527. [14] Xiao HH, Makarov D, Sun JH, Molkov V. Experimental and numerical investigation of premixed flame propagation with distorted tulip shape in a closed duct. Combust Flame 2012;159:1523e38. [15] Molkov V, Verbecke F, Makarov D. LES of hydrogen-air deflagrations in a 78.5-m tunnel. Combust Sci Technol 2008;180:796e808. [16] Dorofeev SB. Hydrogen flames in tubes: critical run-up distances. Int J Hydrogen Energy 2009;34:5832e7. [17] Dunn-Rankin D, Sawyer RF. Tulip flames: changes in shape of premixed flames propagating in closed tubes. Exp Fluids 1998;24:130e40. [18] Gonzalez M, Borghi R, Saouab A. Interaction of a flame front with its self-generated flow in an enclosure e the fulip flame phenomenon. Combust Flame 1992;88:201e20. [19] Xiao HH, Wang QS, He XC, Sun JH, Shen XB. Experimental study on the behaviors and shape changes of premixed hydrogen-air flames propagating in horizontal duct. Int J Hydrogen Energy 2011;36:6325e36. [20] Ellis OCdeC. Flame movement in gaseous explosive mixtures. Fuel Sci 1928;7:502e8. [21] Marra FS, Continillo G. Numerical study of premixed laminar flame propagation in a closed tube with a full NaviereStokes approach. Proc Combust Inst 1996;26:907e13. [22] Metzener P, Matalon M. Premixed flames in closed cylindrical tubes. Combust Theor Model 2001;5:463e83. [23] Guenoche H. Flame propagation in tubes and in closed vessels. In: Markstein GH, editor. Nonsteady flame propagation. New York: Pergamon Press; 1964. p. 107. [24] Matalon M, Mcgreevy JL. The initial development of a tulip flame. Proc Combust Inst; 1994:1407e13. [25] Markstein GH. A shock-tube study of flame front-pressure wave interaction. Proc Combust Inst 1956;6:387e98. [26] Lee ST, Tsai CH. Numerical investigation of steady laminar flame propagation in a circular tube. Combust Flame 1994;99:484e90. [27] Xiao HH, Wang QS, Shen XB, Guo S, Sun JH. An experimental study of distorted tulip flame formation in a closed duct. Combust Flame 2013;160:1725e8. [28] Lamoureux N, Djebaili-Chaumeix N, Paillard CE. Laminar flame velocity determination for H2eaireHeeCO2 mixtures using the spherical bomb method. Exp Therm Fluid Sci 2003;27:385e93.