- Email: [email protected]
Visualization of spontaneous ignition under controlled burst pressure
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 x x x ( 2 0 1 6 ) 1 e6
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Visualization of spontaneous ignition under controlled burst pressure K. Yamashita a, T. Saburi b,*, Y. Wada b, M. Asahara a, T. Mogi c, A.K. Hayashi a a
Graduate School of Science and Engineering, Aoyama Gakuin University, 5-10-1 Fuchinobe, Chuo-ku, Sagamiharashi, Kanagawa 252-0206, Japan b Research Institute of Science for Safety and Sustainability, National Institute of Advanced Industrial Science and Technology, 16-1 Onogawa, Tsukuba-shi, Ibaraki 305-8569, Japan c Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
article info
abstract
Article history:
A high-pressure hydrogen jet released into the air has the possibility of igniting in a tube
Received 1 April 2016
without any ignition source. The mechanism of this phenomenon, called spontaneous
Received in revised form
ignition, is considered to be that hydrogen diffuses into the hot air caused by the shock
28 June 2016
wave from diaphragm rupture and the hydrogen-oxidizer mixed region is formed enough
Accepted 28 June 2016
to start chemical reaction. Recently, flow visualization studies on the spontaneous ignition
Available online xxx
process have been conducted to understand its detailed mechanism, but such ignition has not yet been well clarified. In this study, the spontaneous ignition phenomenon was
Keywords:
observed in a rectangular tube. The results confirm the presence of a flame at the wall of
High-pressure hydrogen
the tube when the shock wave pressure reaches 1.2e1.5 MPa in more than 9 MPa burst
Spontaneous ignition
pressure and that ignition occurs near the wall, followed by multiple ignitions as the shock
Shock wave
wave propagates, with the ignitions eventually combining to form a flame.
High-speed camera
© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction When emergencies occur in hydrogen stations and chemical plants, high-pressure hydrogen is released from a rupture disc through a pipe into the air. During this process, the highpressure hydrogen may spontaneously ignite, yielding a flame that propagates without any ignition source. This phenomenon, called spontaneous ignition, is not caused by an ignition source but by a diffusion of hydrogen into heated by ski shock oxidizer, as experimentally confirmed by Wolan et al. [1]. However, the detailed mechanism of spontaneous ignition in a tube has not yet been well clarified. It is necessary
to define safety criteria to prevent a flame from developing when high-pressure hydrogen is accidentally released in a hydrogen station or chemical plant. The spontaneous ignition process should be clarified in detail based on this background. A variety of experimental and numerical studies have been conducted on this subject. An experimental shock tube system equipped with a diaphragm to release hydrogen using a rupture disc is generally used to study the phenomenon of hydrogen spontaneous ignition. The possibility of a hydrogen jet igniting because of a sudden release depends on the burst pressure of the diaphragm and the tube geometry [2e5]. The minimum burst pressure sufficient to cause spontaneous
* Corresponding author. Fax: þ81 29 861 8760. E-mail address: [email protected] (T. Saburi). http://dx.doi.org/10.1016/j.ijhydene.2016.06.240 0360-3199/© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
2
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 x x x ( 2 0 1 6 ) 1 e6
ignition decreases with increasing tube length to a certain critical length, then increases with further tube extension [5]. The tube geometry downstream of the diaphragm affects the ignition behaviour, whereas that upstream of the diaphragm does not [2]; however, the configuration of the tube and connectors may affect the downstream gas dynamics. The behaviour of the diaphragm rupture can be varied by changing various rupture and diaphragm characteristics, such as the rupture speed and the position, shape, and material of the diaphragm, allowing the possibility of ignition to be altered [6e8]. The flame structure and flammable range outside the tube have been investigated both experimentally [3,5,9], and numerically [4,7]. The ignition process in the tube has also been numerically simulated [7,8,10e12], and experimentally investigated [13] in recent years. However, the quantitative discussion remains insufficient because some experimental studies are not reproducible as a result of the inaccuracy of the rupture process and rupture pressure. In particular, only a few studies have focused on the visualization of the phenomenon of spontaneous ignition in a tube. In this study, a shock tube experiment was conducted to clarify the processes of ignition and flame formation for a high-pressure hydrogen flow in a tube. A circular shock tube apparatus was equipped with a plunger system to control the burst pressure. A circular-to-rectangular conversion flange was used to avoid discontinuous changes in cross section, and a rectangular tube with glass windows was used for optical observation. Two high-speed cameras were used to simultaneously take shadowgraphs and direct photographs.
Experimental setup The experimental apparatus used in this study is based on a shock tube system that consists of a high-pressure hydrogen section and an ambient pressure air section separated by a
diaphragm (Fig. 1). The diaphragm, which is made of polyethylene terephthalate (PET) film (Melinex, Teijin DuPont), is used to keep the hydrogen section at a high pressure. The thickness of the PET film was adjusted according to the burst pressure such that the film could be expected to fail and rupture at a predetermined pressure, but thickness-based rupture control is generally insufficiently accurate. The plunger system used in this apparatus is composed of a barrel with a solenoid coil, a tungsten needle positioned at the coil end, and a step-up circuit and is mounted diagonally on the high-pressure side of the diaphragm flange. When the step-up circuit is switched on, a large pulsed current flows through the coil and forms a magnetic field along the barrel. The magnetic force pulls and accelerates the tungsten needle. The needle moves along the barrel, then hits and ruptures the diaphragm. The plunger system is able to precisely rupture the diaphragm by base pressure control, and the pressure is measured by a strain gauge pressure transducer (PHA-L-20MP, Kyowa Electronic Instruments). The rectangular test section, which is equipped with silica glass windows and sensors, is connected to the tube end of the ambient pressure section. The high-pressure hydrogen propagation, ignition, and flammable behaviours were observed by high-speed cameras through the windows, and the pressure and luminescence histories were recorded at 30 mm intervals using pressure sensors (PCB M111A24, Piezotronics, Inc.) and light detectors (PDA25K, Thorlabs), respectively. When the step-up circuit is switched on, a trigger signal is sent to start recording with a data logger and high-speed cameras. Two high-speed cameras record the shadowgraph system to allow the visualization of the density gradient and direct flame images. The illumination source of the shadowgraph system is a xenon lamp (LS75, KATOKOKEN). The light from the lamp is reflected and collimated off a Schlieren mirror, then passes through the windows of the measurement section. The image then passes through a Fresnel lens and is focused on the first shadowgraph
Fig. 1 e Experimental schematic of entire system. Left and right broken line areas indicate circular driver (high pressure) and rectangular test (ambient pressure) sections, respectively. Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
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 x x x ( 2 0 1 6 ) 1 e6
3
camera positioned normal to the glass window. The flame dynamics are directly captured using the second camera oriented slightly oblique to the window. The cross section of the rectangular tube originally at ambient pressure measures 10 mm 10 mm, and its length ranges from 200 to 800 mm. The circular-to-rectangular conversion flange is connected between the circular high-pressure section and rectangular ambient pressure section, and its cross-sectional shape gradually changes from circular-to-rectangular over a length of 20 mm to avoid a sudden change in the cross-sectional shape of the tube.
propagate, but the degree of decay is sufficiently small in the tube used in this study because of its small length (1 m) [5]. As the burst pressure was increased, the experimental shock wave velocity in the present study increased, and its gradient decreased. The results of the present experimental shock wave velocities agree with the theoretical curve and the results of past studies [3,5]. The present experimental data remain at 80e90% of the theoretical curve, and the results of past studies show similar values. Possible causes of the deviation of the experimental results from the theoretical curve are considered to be the boundary layer effect, the finite time of the rupture process, and the three-dimensional effect.
Results and disscusion
Sensor measurement to determine occurrence of spontaneous ignition
An experimental study was conducted to investigate the spontaneous ignition of high-pressure hydrogen injection into the tube. First, the shock wave (pressure wave) velocities obtained at various burst pressures in the present study were compared with those from past studies [3,5]. The pressure profiles at different burst pressures were then considered. Finally, spontaneous ignition and flame propagation in a tube were observed using high-speed cameras.
Comparison of shock wave velocities from past and present studies with theoretical values Fig. 2 compares the experimental shock wave velocities calculated from its arrival time at each sensor with the theoretically obtained velocities and those from previous studies. The theoretical values were calculated as sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi g 1 g þ 1 p2 þ ; Cs ¼ a1 2g 2g p1 where Cs [m/s] is the theoretical shock speed; a [m/s] is the speed of sound; g is the specific heat ratio; p is the pressure; and the subscripts 1 and 2 indicate conditions in front of and behind the shock, respectively. Shock waves decay as they
The pressure and luminescence histories recorded by the pressure gauges P1eP5 and light sensors L1eL5, respectively, were compared to determine whether spontaneous ignition occurred in three cases with different burst pressures. The locations of P1eP5 with respect to the diagram correspond to those of L1eL5 in all cases. The time histories for the cases in which no ignition occurred (burst pressure of 5.1 MPa; Case 1), ignition occurred in the measurement section (burst pressure of 9.5 MPa; Case 2), and ignition occurred before the measurement section (burst pressure of 10 MPa; Case 3) are shown in Figs. 3e5, respectively. It should be noted that L1 detected no useful data because it failed in the experiments. Additionally, the time t ¼ 0 represents the trigger time, not the time at which the diaphragm ruptured. Generally, pressure suddenly increases upon the arrival of a shock wave, and luminescence is detected from the flame passing the sensor position. In Case 1, the arrival of the shock wave at each pressure sensor caused the measured pressure to increase to approximately 1 MPa, and no luminescence was detected, as shown in Fig. 3. In Case 2, the pressure first increased to approximately 1.5 MPa at P3, P4, and P5, and luminescence from the flame was detected by L3, L4, and L5, as shown in Fig. 4. In contrast,
Pressure wave velocity [m/s]
2000 1800 1600 1400 1200 1000 800 600
Theory Mogi et al. [3] Kitabayashi et al. [5] Present experiment
400 200 0 0
2
4
6
8
10
12
14
Burst pressure [MPa] Fig. 2 e Effect of burst pressure on pressure wave velocity. Comparison of theoretical and experimental pressure wave velocity results for a tube.
Fig. 3 e Pressure histories at the tube wall for Case 1. In Case 1, no ignition occurred. The pressure histories were measured by pressure sensors P1eP5 at a burst pressure of 5.1 MPa.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
4
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 x x x ( 2 0 1 6 ) 1 e6
In the experimental results of hydrogen ignition obtained by Dryer et al. [2], by Golub et al. [4], and in the present study, ignition was observed at storage pressures of 2, 4e9, and 4e11 MPa, respectively. One reason for this difference may have been that in the experiments by Dryer et al. [2], there was a small step between the tube and the connector, causing ignition to occur. The results obtained by Golub et al. [4] and in the present study yielded similar ignition storage pressures. Hence, the experiment by Golub et al. [4] and the present experiment have a similar accuracy for the experimental tube and measurement system.
Fig. 4 e Pressure and luminescence histories at the tube wall for Case 2. In Case 2, ignition occurred in the measurement section. The pressure and luminescence histories were measured by pressure sensors P1eP5 and light sensors L3eL5, respectively, at a burst pressure of 9.5 MPa.
Observation of ignition phenomenon in tube The shadowgraph and direct images taken in the case in which ignition occurred near the measurement section in the tube are shown in Figs. 6 and 7, respectively. The following points about these images should be noted prior to the discussion. ➢ These images were taken under experimental conditions (burst pressure of 9.0 MPa; Case 4) differing from those discussed in the previous section. ➢ The time t ¼ 0 indicates the arrival time of the shock wave in the measurement section.
Fig. 5 e Pressure and luminescence histories at the tube wall for Case 3. In Case 3, ignition occurred before the measurement section. The pressure and luminescence histories were measured by pressure sensors P1eP5 and light sensors L2eL5, respectively, at a burst pressure of 10.0 MPa.
the pressure increased to only approximately 1.2 MPa at P2, and no flame luminescence was detected at L2. It is believed that there is the flame genesis position just before the L3 position where the flame is detected. Finally, in Case 3, all pressure sensors detected pressures above 1.5 MPa after the first pressure rise, and all functioning light sensors detected flame luminescence, as shown in Fig. 5. Therefore, based on these three cases, ignition can be expected to occur when the wall pressure reaches 1.2e1.5 MPa. In addition, each light sensor detected luminescence from the flame for approximately 100 ms, which indicates that the length of the flame region is approximately 10 cm if the flame is assumed to propagate at approximately the same rate as the shock wave. It should be noted that the burst pressure in the case of ignition was higher than that obtained in previous related studies. This higher burst pressure arises from differences in the tube configuration, such as the cross-sectional area, because ignition occurs at a lower burst pressure in tubes with smaller cross-sectional areas [2e4].
Fig. 6 e The x-t diagram from shadowgraph at a burst pressure of 9.0 MPa. Propagation behaviors of shockwave front and frame front, frame tail, and H2 flow are shown every 10 ms.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
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 x x x ( 2 0 1 6 ) 1 e6
5
flame shown in Fig. 7. A high-density gradient was observed near the wall in the mixed region following the shock wave. This high-density gradient region may have been generated by fluid dynamic instability, which was confirmed in numerical simulations [5,7,10,11]. Fig. 7 shows a flame kernel arising from the ignition, appearing near the wall at 60 ms. Another flame kernel emerged from the other side of the wall at 70 ms and gradually expanded as it propagated. Additional flame kernels emerged throughout the tube at approximately 130e150 ms. Finally, these flames combined to form a large flame region spanning the tube width. The position of the initially appearing flame kernel corresponds to the mixed region near the wall in the shadowgraph image. This indicates that the ignition first occurred in the mixed region near the wall; the same trend was confirmed in a previous experimental study [13] and numerical 3D study [14]. Additionally, ignition occurred in many places, and the resulting flames merged together to form the flame region. As measured by the direct images, the length of the entire flame region reached approximately 10 cm, which is almost the same as that obtained from the light sensor measurement. Thus, the flame length evaluated by the light sensors and that measured by the photographic images are consistent with each other.
Conclusion Fig. 7 e The x-t diagram from direct photograph at a burst pressure of 9.0 MPa. Generation and propagation behaviors of frame kernels are shown every 10 ms. ➢ The lines in the shadowgraph images illustrate the shock wave, flame front, flame tail, and developed turbulent hydrogen flow. The area between shock wave and flame front is a high temperature air, and the area between flame tail and developed turbulent hydrogen flow is an unburned hydrogen/air mixture area. ➢ As mentioned in Section Experimental setup, the camera for the shadowgraph system was installed normal to the glass window, and the camera for the direct photography was oriented slightly oblique to the window. Therefore, lights along the right-hand rounded edge in the two direct photography images at 160 and 170 ms in Fig. 7 are the reflected light at observing window frame. The shadowgraph in Fig. 6 shows the shock wave entering the measurement section first, followed by the high-density fluctuation region corresponding to a mixed region and finally the hydrogen jet. The mixed region is divided into two parts: one following the shock wave and another accompanying the shock wave at approximately 70 ms. In the part of mixture region accompanying the shock wave, the flame weakened as it propagated and seemed to disappear, as shown in Fig. 7. Conversely, the part of mixture region following the shock wave corresponds to the propagating
The following conclusions were reached in the present experimental investigation of shock waves and flames in a shock tube using pressure and light sensor measurements and observations by high-speed camera photography. ➢ Luminescence from the flame was confirmed when the shock wave pressure at the wall reached 1.2e1.5 MPa. ➢ Ignition was found to occur first in the mixed region near the wall, followed by multiple subsequent ignitions as the shock wave propagated and the eventual formation of the flame region. This is consistent with the results obtained in a previous study [14]. ➢ It requires that the mixed region maintains its duration time to meet the local ignition delay time to cause ignition. Based on these findings, to prevent spontaneous ignition during high-pressure hydrogen discharge, it is important to ensure that the shock wave pressure does not reach the pressure required for ignition. The possibility of ignition can be reduced by controlling the temperature near the outlet. Furthermore, the spontaneous ignition limit is generally discussed by using the spontaneous ignition curve, which has specified the tube length and the burst pressure as parameters in past studies [5].
references
ski PS, Wo jcicki S. Investigation into the mechanism [1] Wolan of the diffusion ignition of a combustible gas flowing into an oxidizing atmosphere. P Combust Inst 1973;14:1217e23.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
6
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 x x x ( 2 0 1 6 ) 1 e6
[2] Dryer FL, Chaos M, Zhao Z, Stein JN, Alpert JY, Homer CJ. Spontaneous ignition of pressurized release of hydrogen and natural gas into air. Combust Sci Technol 2007;179(4e6):663e94. [3] Mogi T, Wada Y, Ogata Y, Hayashi AK. Self-ignition and flame propagation of high-pressure hydrogen jet during sudden discharge from a pipe. Int J Hydrogen Energy 2009;34(14):5810e6. [4] Golub VV, Baklanov DI, Bazhenova TV, Golovastov SV, Ivanov MF, Laskin IN, et al. Experimental and numerical investigation of hydrogen gas auto-ignition. Int J Hydrogen Energy 2009;34(14):5946e53. [5] Kitabayashi N, Wada Y, Mogi T, Saburi T, Hayashi AK. Experimental study on high pressure hydrogen jets coming out of tubes of 0.1e4.2 m in length. Int J Hydrogen Energy 2013;38(19):8100e7. [6] Golovastov S, Bocharnikov V. The influence of diaphragm rupture rate on spontaneous self ignition of pressurized hydrogen: experimental investigation. Int J Hydrogen Energy 2012;37(14):10956e62. [7] Wen JX, Xu BP, Tam VHY. Numerical study on spontaneous ignition of pressurized hydrogen release through a length of tube. Combust Flame 2009;156(11):2173e89. [8] Lee HJ, Park JH, Kim SD, Kim S, Jeung IS. Numerical study on the spontaneous-ignition features of high-pressure
[9]
[10]
[11]
[12]
[13]
[14]
hydrogen released through a tube with burst conditions. P Combust Inst 2015;35(2):2173e80. Kim S, Lee HJ, Park JH, Jeung IS. Effects of a wall on the selfignition patterns and flame propagation of high-pressure hydrogen release through a tube. P Combust Inst 2013;34(2):2049e56. Yamada E, Watanabe S, Hayashi AK, Tsuboi N. Numerical analysis on auto-ignition of a high pressure hydrogen jet spouting from a tube. P Combust Inst 2009;32(2):2363e9. Asahara M, Yokoyama A, Hayashi AK. Numerical simulation of auto-ignition induced by high-pressure hydrogen release with detailed reaction model: fluid dynamic effect by diaphragm shape and boundary layer. Int J Hydrogen Energy 2014;39(35):20378e87. Bragin MV, Makarov DV, Molkov VV. Pressure limit of hydrogen spontaneous ignition in a T-shaped channel. Int J Hydrogen Energy 2013;38(19):8039e52. Kim YR, Lee HJ, Kim S, Jeung IS. A flow visualization study on self-ignition of high pressure hydrogen gas released into a tube. P Combust Inst 2013;34(2):2057e64. Bragin MV, Molkov VV. Physics of spontaneous ignition of high-pressure hydrogen release and transition to jet fire. Int J Hydrogen Energy 2011;36(3):2589e96.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Visualization of spontaneous ignition under controlled burst pressure K. Yamashita a, T. Saburi b,*, Y. Wada b, M. Asahara a, T. Mogi c, A.K. Hayashi a a
Graduate School of Science and Engineering, Aoyama Gakuin University, 5-10-1 Fuchinobe, Chuo-ku, Sagamiharashi, Kanagawa 252-0206, Japan b Research Institute of Science for Safety and Sustainability, National Institute of Advanced Industrial Science and Technology, 16-1 Onogawa, Tsukuba-shi, Ibaraki 305-8569, Japan c Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
article info
abstract
Article history:
A high-pressure hydrogen jet released into the air has the possibility of igniting in a tube
Received 1 April 2016
without any ignition source. The mechanism of this phenomenon, called spontaneous
Received in revised form
ignition, is considered to be that hydrogen diffuses into the hot air caused by the shock
28 June 2016
wave from diaphragm rupture and the hydrogen-oxidizer mixed region is formed enough
Accepted 28 June 2016
to start chemical reaction. Recently, flow visualization studies on the spontaneous ignition
Available online xxx
process have been conducted to understand its detailed mechanism, but such ignition has not yet been well clarified. In this study, the spontaneous ignition phenomenon was
Keywords:
observed in a rectangular tube. The results confirm the presence of a flame at the wall of
High-pressure hydrogen
the tube when the shock wave pressure reaches 1.2e1.5 MPa in more than 9 MPa burst
Spontaneous ignition
pressure and that ignition occurs near the wall, followed by multiple ignitions as the shock
Shock wave
wave propagates, with the ignitions eventually combining to form a flame.
High-speed camera
© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction When emergencies occur in hydrogen stations and chemical plants, high-pressure hydrogen is released from a rupture disc through a pipe into the air. During this process, the highpressure hydrogen may spontaneously ignite, yielding a flame that propagates without any ignition source. This phenomenon, called spontaneous ignition, is not caused by an ignition source but by a diffusion of hydrogen into heated by ski shock oxidizer, as experimentally confirmed by Wolan et al. [1]. However, the detailed mechanism of spontaneous ignition in a tube has not yet been well clarified. It is necessary
to define safety criteria to prevent a flame from developing when high-pressure hydrogen is accidentally released in a hydrogen station or chemical plant. The spontaneous ignition process should be clarified in detail based on this background. A variety of experimental and numerical studies have been conducted on this subject. An experimental shock tube system equipped with a diaphragm to release hydrogen using a rupture disc is generally used to study the phenomenon of hydrogen spontaneous ignition. The possibility of a hydrogen jet igniting because of a sudden release depends on the burst pressure of the diaphragm and the tube geometry [2e5]. The minimum burst pressure sufficient to cause spontaneous
* Corresponding author. Fax: þ81 29 861 8760. E-mail address: [email protected] (T. Saburi). http://dx.doi.org/10.1016/j.ijhydene.2016.06.240 0360-3199/© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
2
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 x x x ( 2 0 1 6 ) 1 e6
ignition decreases with increasing tube length to a certain critical length, then increases with further tube extension [5]. The tube geometry downstream of the diaphragm affects the ignition behaviour, whereas that upstream of the diaphragm does not [2]; however, the configuration of the tube and connectors may affect the downstream gas dynamics. The behaviour of the diaphragm rupture can be varied by changing various rupture and diaphragm characteristics, such as the rupture speed and the position, shape, and material of the diaphragm, allowing the possibility of ignition to be altered [6e8]. The flame structure and flammable range outside the tube have been investigated both experimentally [3,5,9], and numerically [4,7]. The ignition process in the tube has also been numerically simulated [7,8,10e12], and experimentally investigated [13] in recent years. However, the quantitative discussion remains insufficient because some experimental studies are not reproducible as a result of the inaccuracy of the rupture process and rupture pressure. In particular, only a few studies have focused on the visualization of the phenomenon of spontaneous ignition in a tube. In this study, a shock tube experiment was conducted to clarify the processes of ignition and flame formation for a high-pressure hydrogen flow in a tube. A circular shock tube apparatus was equipped with a plunger system to control the burst pressure. A circular-to-rectangular conversion flange was used to avoid discontinuous changes in cross section, and a rectangular tube with glass windows was used for optical observation. Two high-speed cameras were used to simultaneously take shadowgraphs and direct photographs.
Experimental setup The experimental apparatus used in this study is based on a shock tube system that consists of a high-pressure hydrogen section and an ambient pressure air section separated by a
diaphragm (Fig. 1). The diaphragm, which is made of polyethylene terephthalate (PET) film (Melinex, Teijin DuPont), is used to keep the hydrogen section at a high pressure. The thickness of the PET film was adjusted according to the burst pressure such that the film could be expected to fail and rupture at a predetermined pressure, but thickness-based rupture control is generally insufficiently accurate. The plunger system used in this apparatus is composed of a barrel with a solenoid coil, a tungsten needle positioned at the coil end, and a step-up circuit and is mounted diagonally on the high-pressure side of the diaphragm flange. When the step-up circuit is switched on, a large pulsed current flows through the coil and forms a magnetic field along the barrel. The magnetic force pulls and accelerates the tungsten needle. The needle moves along the barrel, then hits and ruptures the diaphragm. The plunger system is able to precisely rupture the diaphragm by base pressure control, and the pressure is measured by a strain gauge pressure transducer (PHA-L-20MP, Kyowa Electronic Instruments). The rectangular test section, which is equipped with silica glass windows and sensors, is connected to the tube end of the ambient pressure section. The high-pressure hydrogen propagation, ignition, and flammable behaviours were observed by high-speed cameras through the windows, and the pressure and luminescence histories were recorded at 30 mm intervals using pressure sensors (PCB M111A24, Piezotronics, Inc.) and light detectors (PDA25K, Thorlabs), respectively. When the step-up circuit is switched on, a trigger signal is sent to start recording with a data logger and high-speed cameras. Two high-speed cameras record the shadowgraph system to allow the visualization of the density gradient and direct flame images. The illumination source of the shadowgraph system is a xenon lamp (LS75, KATOKOKEN). The light from the lamp is reflected and collimated off a Schlieren mirror, then passes through the windows of the measurement section. The image then passes through a Fresnel lens and is focused on the first shadowgraph
Fig. 1 e Experimental schematic of entire system. Left and right broken line areas indicate circular driver (high pressure) and rectangular test (ambient pressure) sections, respectively. Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
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 x x x ( 2 0 1 6 ) 1 e6
3
camera positioned normal to the glass window. The flame dynamics are directly captured using the second camera oriented slightly oblique to the window. The cross section of the rectangular tube originally at ambient pressure measures 10 mm 10 mm, and its length ranges from 200 to 800 mm. The circular-to-rectangular conversion flange is connected between the circular high-pressure section and rectangular ambient pressure section, and its cross-sectional shape gradually changes from circular-to-rectangular over a length of 20 mm to avoid a sudden change in the cross-sectional shape of the tube.
propagate, but the degree of decay is sufficiently small in the tube used in this study because of its small length (1 m) [5]. As the burst pressure was increased, the experimental shock wave velocity in the present study increased, and its gradient decreased. The results of the present experimental shock wave velocities agree with the theoretical curve and the results of past studies [3,5]. The present experimental data remain at 80e90% of the theoretical curve, and the results of past studies show similar values. Possible causes of the deviation of the experimental results from the theoretical curve are considered to be the boundary layer effect, the finite time of the rupture process, and the three-dimensional effect.
Results and disscusion
Sensor measurement to determine occurrence of spontaneous ignition
An experimental study was conducted to investigate the spontaneous ignition of high-pressure hydrogen injection into the tube. First, the shock wave (pressure wave) velocities obtained at various burst pressures in the present study were compared with those from past studies [3,5]. The pressure profiles at different burst pressures were then considered. Finally, spontaneous ignition and flame propagation in a tube were observed using high-speed cameras.
Comparison of shock wave velocities from past and present studies with theoretical values Fig. 2 compares the experimental shock wave velocities calculated from its arrival time at each sensor with the theoretically obtained velocities and those from previous studies. The theoretical values were calculated as sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi g 1 g þ 1 p2 þ ; Cs ¼ a1 2g 2g p1 where Cs [m/s] is the theoretical shock speed; a [m/s] is the speed of sound; g is the specific heat ratio; p is the pressure; and the subscripts 1 and 2 indicate conditions in front of and behind the shock, respectively. Shock waves decay as they
The pressure and luminescence histories recorded by the pressure gauges P1eP5 and light sensors L1eL5, respectively, were compared to determine whether spontaneous ignition occurred in three cases with different burst pressures. The locations of P1eP5 with respect to the diagram correspond to those of L1eL5 in all cases. The time histories for the cases in which no ignition occurred (burst pressure of 5.1 MPa; Case 1), ignition occurred in the measurement section (burst pressure of 9.5 MPa; Case 2), and ignition occurred before the measurement section (burst pressure of 10 MPa; Case 3) are shown in Figs. 3e5, respectively. It should be noted that L1 detected no useful data because it failed in the experiments. Additionally, the time t ¼ 0 represents the trigger time, not the time at which the diaphragm ruptured. Generally, pressure suddenly increases upon the arrival of a shock wave, and luminescence is detected from the flame passing the sensor position. In Case 1, the arrival of the shock wave at each pressure sensor caused the measured pressure to increase to approximately 1 MPa, and no luminescence was detected, as shown in Fig. 3. In Case 2, the pressure first increased to approximately 1.5 MPa at P3, P4, and P5, and luminescence from the flame was detected by L3, L4, and L5, as shown in Fig. 4. In contrast,
Pressure wave velocity [m/s]
2000 1800 1600 1400 1200 1000 800 600
Theory Mogi et al. [3] Kitabayashi et al. [5] Present experiment
400 200 0 0
2
4
6
8
10
12
14
Burst pressure [MPa] Fig. 2 e Effect of burst pressure on pressure wave velocity. Comparison of theoretical and experimental pressure wave velocity results for a tube.
Fig. 3 e Pressure histories at the tube wall for Case 1. In Case 1, no ignition occurred. The pressure histories were measured by pressure sensors P1eP5 at a burst pressure of 5.1 MPa.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
4
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 x x x ( 2 0 1 6 ) 1 e6
In the experimental results of hydrogen ignition obtained by Dryer et al. [2], by Golub et al. [4], and in the present study, ignition was observed at storage pressures of 2, 4e9, and 4e11 MPa, respectively. One reason for this difference may have been that in the experiments by Dryer et al. [2], there was a small step between the tube and the connector, causing ignition to occur. The results obtained by Golub et al. [4] and in the present study yielded similar ignition storage pressures. Hence, the experiment by Golub et al. [4] and the present experiment have a similar accuracy for the experimental tube and measurement system.
Fig. 4 e Pressure and luminescence histories at the tube wall for Case 2. In Case 2, ignition occurred in the measurement section. The pressure and luminescence histories were measured by pressure sensors P1eP5 and light sensors L3eL5, respectively, at a burst pressure of 9.5 MPa.
Observation of ignition phenomenon in tube The shadowgraph and direct images taken in the case in which ignition occurred near the measurement section in the tube are shown in Figs. 6 and 7, respectively. The following points about these images should be noted prior to the discussion. ➢ These images were taken under experimental conditions (burst pressure of 9.0 MPa; Case 4) differing from those discussed in the previous section. ➢ The time t ¼ 0 indicates the arrival time of the shock wave in the measurement section.
Fig. 5 e Pressure and luminescence histories at the tube wall for Case 3. In Case 3, ignition occurred before the measurement section. The pressure and luminescence histories were measured by pressure sensors P1eP5 and light sensors L2eL5, respectively, at a burst pressure of 10.0 MPa.
the pressure increased to only approximately 1.2 MPa at P2, and no flame luminescence was detected at L2. It is believed that there is the flame genesis position just before the L3 position where the flame is detected. Finally, in Case 3, all pressure sensors detected pressures above 1.5 MPa after the first pressure rise, and all functioning light sensors detected flame luminescence, as shown in Fig. 5. Therefore, based on these three cases, ignition can be expected to occur when the wall pressure reaches 1.2e1.5 MPa. In addition, each light sensor detected luminescence from the flame for approximately 100 ms, which indicates that the length of the flame region is approximately 10 cm if the flame is assumed to propagate at approximately the same rate as the shock wave. It should be noted that the burst pressure in the case of ignition was higher than that obtained in previous related studies. This higher burst pressure arises from differences in the tube configuration, such as the cross-sectional area, because ignition occurs at a lower burst pressure in tubes with smaller cross-sectional areas [2e4].
Fig. 6 e The x-t diagram from shadowgraph at a burst pressure of 9.0 MPa. Propagation behaviors of shockwave front and frame front, frame tail, and H2 flow are shown every 10 ms.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
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 x x x ( 2 0 1 6 ) 1 e6
5
flame shown in Fig. 7. A high-density gradient was observed near the wall in the mixed region following the shock wave. This high-density gradient region may have been generated by fluid dynamic instability, which was confirmed in numerical simulations [5,7,10,11]. Fig. 7 shows a flame kernel arising from the ignition, appearing near the wall at 60 ms. Another flame kernel emerged from the other side of the wall at 70 ms and gradually expanded as it propagated. Additional flame kernels emerged throughout the tube at approximately 130e150 ms. Finally, these flames combined to form a large flame region spanning the tube width. The position of the initially appearing flame kernel corresponds to the mixed region near the wall in the shadowgraph image. This indicates that the ignition first occurred in the mixed region near the wall; the same trend was confirmed in a previous experimental study [13] and numerical 3D study [14]. Additionally, ignition occurred in many places, and the resulting flames merged together to form the flame region. As measured by the direct images, the length of the entire flame region reached approximately 10 cm, which is almost the same as that obtained from the light sensor measurement. Thus, the flame length evaluated by the light sensors and that measured by the photographic images are consistent with each other.
Conclusion Fig. 7 e The x-t diagram from direct photograph at a burst pressure of 9.0 MPa. Generation and propagation behaviors of frame kernels are shown every 10 ms. ➢ The lines in the shadowgraph images illustrate the shock wave, flame front, flame tail, and developed turbulent hydrogen flow. The area between shock wave and flame front is a high temperature air, and the area between flame tail and developed turbulent hydrogen flow is an unburned hydrogen/air mixture area. ➢ As mentioned in Section Experimental setup, the camera for the shadowgraph system was installed normal to the glass window, and the camera for the direct photography was oriented slightly oblique to the window. Therefore, lights along the right-hand rounded edge in the two direct photography images at 160 and 170 ms in Fig. 7 are the reflected light at observing window frame. The shadowgraph in Fig. 6 shows the shock wave entering the measurement section first, followed by the high-density fluctuation region corresponding to a mixed region and finally the hydrogen jet. The mixed region is divided into two parts: one following the shock wave and another accompanying the shock wave at approximately 70 ms. In the part of mixture region accompanying the shock wave, the flame weakened as it propagated and seemed to disappear, as shown in Fig. 7. Conversely, the part of mixture region following the shock wave corresponds to the propagating
The following conclusions were reached in the present experimental investigation of shock waves and flames in a shock tube using pressure and light sensor measurements and observations by high-speed camera photography. ➢ Luminescence from the flame was confirmed when the shock wave pressure at the wall reached 1.2e1.5 MPa. ➢ Ignition was found to occur first in the mixed region near the wall, followed by multiple subsequent ignitions as the shock wave propagated and the eventual formation of the flame region. This is consistent with the results obtained in a previous study [14]. ➢ It requires that the mixed region maintains its duration time to meet the local ignition delay time to cause ignition. Based on these findings, to prevent spontaneous ignition during high-pressure hydrogen discharge, it is important to ensure that the shock wave pressure does not reach the pressure required for ignition. The possibility of ignition can be reduced by controlling the temperature near the outlet. Furthermore, the spontaneous ignition limit is generally discussed by using the spontaneous ignition curve, which has specified the tube length and the burst pressure as parameters in past studies [5].
references
ski PS, Wo jcicki S. Investigation into the mechanism [1] Wolan of the diffusion ignition of a combustible gas flowing into an oxidizing atmosphere. P Combust Inst 1973;14:1217e23.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240
6
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 x x x ( 2 0 1 6 ) 1 e6
[2] Dryer FL, Chaos M, Zhao Z, Stein JN, Alpert JY, Homer CJ. Spontaneous ignition of pressurized release of hydrogen and natural gas into air. Combust Sci Technol 2007;179(4e6):663e94. [3] Mogi T, Wada Y, Ogata Y, Hayashi AK. Self-ignition and flame propagation of high-pressure hydrogen jet during sudden discharge from a pipe. Int J Hydrogen Energy 2009;34(14):5810e6. [4] Golub VV, Baklanov DI, Bazhenova TV, Golovastov SV, Ivanov MF, Laskin IN, et al. Experimental and numerical investigation of hydrogen gas auto-ignition. Int J Hydrogen Energy 2009;34(14):5946e53. [5] Kitabayashi N, Wada Y, Mogi T, Saburi T, Hayashi AK. Experimental study on high pressure hydrogen jets coming out of tubes of 0.1e4.2 m in length. Int J Hydrogen Energy 2013;38(19):8100e7. [6] Golovastov S, Bocharnikov V. The influence of diaphragm rupture rate on spontaneous self ignition of pressurized hydrogen: experimental investigation. Int J Hydrogen Energy 2012;37(14):10956e62. [7] Wen JX, Xu BP, Tam VHY. Numerical study on spontaneous ignition of pressurized hydrogen release through a length of tube. Combust Flame 2009;156(11):2173e89. [8] Lee HJ, Park JH, Kim SD, Kim S, Jeung IS. Numerical study on the spontaneous-ignition features of high-pressure
[9]
[10]
[11]
[12]
[13]
[14]
hydrogen released through a tube with burst conditions. P Combust Inst 2015;35(2):2173e80. Kim S, Lee HJ, Park JH, Jeung IS. Effects of a wall on the selfignition patterns and flame propagation of high-pressure hydrogen release through a tube. P Combust Inst 2013;34(2):2049e56. Yamada E, Watanabe S, Hayashi AK, Tsuboi N. Numerical analysis on auto-ignition of a high pressure hydrogen jet spouting from a tube. P Combust Inst 2009;32(2):2363e9. Asahara M, Yokoyama A, Hayashi AK. Numerical simulation of auto-ignition induced by high-pressure hydrogen release with detailed reaction model: fluid dynamic effect by diaphragm shape and boundary layer. Int J Hydrogen Energy 2014;39(35):20378e87. Bragin MV, Makarov DV, Molkov VV. Pressure limit of hydrogen spontaneous ignition in a T-shaped channel. Int J Hydrogen Energy 2013;38(19):8039e52. Kim YR, Lee HJ, Kim S, Jeung IS. A flow visualization study on self-ignition of high pressure hydrogen gas released into a tube. P Combust Inst 2013;34(2):2057e64. Bragin MV, Molkov VV. Physics of spontaneous ignition of high-pressure hydrogen release and transition to jet fire. Int J Hydrogen Energy 2011;36(3):2589e96.
Please cite this article in press as: Yamashita K, et al., Visualization of spontaneous ignition under controlled burst pressure, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.06.240