Design of safe hydrogen refueling stations against gas-leakage, explosion and accidental automobile collision

Transcription

1 Design of safe hydrogen refueling stations against gas-leakage, explosion and accidental automobile collision Yoshihide SUWA, Hideo MIYAHARA, Keiji KUBO, Kenji YONEZAWA, Yoshiyuki ONO and Kenji MIKODA Obayashi Corporation, Technical Research Institute, Shimokiyoto, Kiyose-shi, Tokyo , Japan suwa.yoshihide@obayashi.co.jp ABSTRACT: An establishment of safe application ware is now the most important problem to generalize the use of hydrogen energy. This paper describes the result of research project performed as one of the NEDO projects during 2003 to Accident potentials of hydrogen refueling stations were analyzed based on the case research, and we found that the explosive accident brings significant disasters such as fatal accidents. In the next stage, measures against the explosion accidents were studied, and the way to design more effective protective walls was investigated. As for the damage of walls or the building structures by the blast, explosion tests were performed and the response and damage of reinforced concrete construction was observed. Finally, protection against the collisions by recklessly driven or miss-operated vehicles was studied. These results are applicable to realize safe design of hydrogen refueling stations. KEYWORDS: hydrogen gas, safety, accident potential, explosion, accidental automobile collision. 1.Introduction Recently, the use of hydrogen energy is drawing attention with the improvement in fuel cell technologies. Hydrogen energy rarely influences environment, and it doesn't depend on a petroleum. So, hydrogen energy is expected as the energy in next generations. Now the fuel cell batteries started to be applied in the home-use generators, automobiles, personal computers and so on. However, hydrogen is still a very inflammable and explosive material. Especially, in the FC automobiles and the hydrogen refueling stations, high-pressure compressed hydrogen gas such as higher than 40MPa is used. This means that a safe technology not only for the hydrogen gas itself but also for the compressed gases in addition is necessary. An establishment of safe application ware is now the most important problem to generalize the use of hydrogen energy. Authors performed a research project on the safety of hydrogen refueling stations as one of the project of NEDO (New Energy and Industrial Technology Development Organization in Japan) held during 2003 to 2004 [1]. This paper describes the results of this project and the result of our studies about safe technologies continued until now.

2 2. Analysis of the accident potentials based on the case-research The 195 accidents, which were occurred after 1949 concerned with hydrogen (valid data were in 172 accidents), were studied to investigate accident potentials [2]. Result of the analysis is shown in Table1 and Fig.1. As the result, 68% of the accidents were caused by the artificial factors, such as imperfect inspections, operational errors, judging errors or incorrect usage, and were much more than the accidents caused by the equipments. The 55% of occurrence parts of the accidents were storages or manufacturing equipments, and the 38% were in the piping systems. In many examples that report the occurrence parts as the equipments, exact occurrence parts were the attachments such as flanges or valves. We conclude that the majority of the accidents were related to the piping systems. Accidents resulted into the fire were about 40% and the accidents resulted into the explosion (including burst accidents) were also about 40%. Only the 13% of accidents stopped in the leak. Fatal accidents have occurred only in the case of the explosion accidents. Fortunately, no one has died in the fire or the leak accidents in Japan. To realize safe hydrogen stations, we have to reduce artificial accident potentials by considering a fail-safe design. When the leak accident has occurred, the most important thing is never to make it reach to the fire or the explosion accidents. Division of the accident's type (Larege) (Middle) Human related Machine related Automobile accident operation, judgement and use acknowledgement verification Mistakes of design or structure Degradation (Small) Table 1 Patterns of accidents in hydrogen supply stations screwing up welding Impulities piling up Prohibited operation Control mistake valve operation Mistake of used parts purge verification Error at Carrying Taking over mistake sub-standard material making design Erosion and corrosion Flange particulars from background to occurrence Frequient Human suffering Tendency Examples (At causal error) (At accident) (Time lag) Disaster (Death) (Ratio) of times (subtotal) (total) Past inspection? Steady/unsteady Instant/ Outbreak Fire No Small Increase 21 Pipe At repair? Startup Middle Leak No Small Decrease 3 All Past inspection? Steady Middle Explosion Yes Small Decrease 5 Pipe Degradation? Steady Outbreak Fire No Small - 3 Equipment Unspecified? Simultaneously Instant Explosion Yes Middle Parallel 24 accessory of Equipment Valve Pipe and cleaning Startup and shutdown and repair? Simultaneously Instant/ Middle Fire Yes Large Increase 10? Simultaneously Instant Explosion Yes Small Parallel 10? Steady Short Fire No Small Parallel 6 Equipment Shutdown? Next work Short Explosion Yes Middle Increase 7 Equipment Shutdown? Next work Instant Explosion Yes Large Decrease 14 Receptacle Loading? - and moving Driving and unloading Middle Fire No Small Decrease 11? Simultaneously Instant Fire No Small - 3 Pipe At making? Steady Middle Fire Yes Large Decrease 4 Flange At making? Steady Middle Explosion Yes Small Decrease 6 Pipe At planning? Steady Middle Explosion No Middle Increase 6 Pipe Degradation? Steady/unsteady Outbreak Fire Yes Small Increase 12 Receptacle Pipe Degradation? Steady/unsteady Outbreak Fire Yes Middle Decrease 12 Viveration of machine Collision accident Frequent area Accessory of Equipment Degradation? Steady/unsteady Outbreak Fire No Small - 3 Receptacle At driving? Simultaneously Instant Leak - - Parallel 5 Overspeed Receptacle At driving? Simultaneously Instant Leak - - Parallel 4 In factory Hose At moving? Simultaneously Instant Fire No (19%) 50 (29%) 35 (20%) 16 (9%) 27 (16%) 11 (6%) Natural disaster Thunderbolt Ventstack Thunderbolt? Gas discharging Instant Fire No - - (3) Others - -? (1%)

3 screwing up welding inspection Impurities piling up Prohibited operation Control mistake valve operation Mistake of used parts purge verification Error at Carrying Taking over mistake Sub-standard material making design Erosion and Corrosion Crack and degradation Vibration of machine Manufacture Storage Pipe Valve,Flange Receptacle Others screwing up welding inspection Impurities piling up Prohibited operation Control mistake valve operation Mistake of used parts purge verification Error at Carrying Taking over mistake Sub-standard material making design Erosion and Corrosion Crack and degradation Vibration of machine Explosion and rupture Fire Leak Others Explosion Rupture Fire Leak Fig.1 Relationship between patterns, locations and disaster phenomena of accidents 3. Measures to the explosion accidents 3.1 Reduction of the blast using protective walls When the explosion accident has occurred, the measures to save human life and to minimize the damage to the surroundings are very important. The effects of protective walls were studied using the computer simulation technique [3]. Simulation was performed using an impact analysis code AUTODYN-3D, and the detonation of pre-mixed hydrogen gas with 30% concentration was assumed (explosion energy becomes largest because of the stoichiometry ratio). Fig.2 shows the time history of overpressure at each point. Simulated result showed a good agreement with the experiment in the time history of overpressure at each point. Fig.3 compares overpressure and impulse in the free field and that in the case with a protective wall at the distance of 5m from the explosion center. We can observe the blast behind the wall reduced in both overpressure and impulse, and we can recognize the effect by the protective wall. In the previous paper, the effects by a lot of kinds of walls were studied, and we found the higher walls reduced the blast more effectively and the effect didn't depend on their sectional shapes [4], [5]. In this work, more cases were studied in such as different wall distances, different height of explosion sources. As the result, higher walls again showed larger advantage in reducing the blast. On the other hand, we also found that the impulse acts on the walls became larger in the higher walls. Therefore, higher walls receive larger load rather than the increase in their exposed area to the blast. We concluded that higher walls were more effective to reduce the blast, but they need higher strength when we design these walls.

4 P2 P3 Hydrogen wall tent P4 P1 P2 P3 P1 P4 5m 3m 5m 10m Fig.2 Time history of overpressure at each point Fig.3 Comparison of overpressure and impulse in the free field and that in the case with a protective wall 3.2 Damage of reinforced-concrete-construction by the blast Before our research, a lot of experiments and analyses already had been performed concerned with the explosion of hydrogen gas, and many parts of the phenomenon of the explosion itself had been clear. However, as for the damage of the building structures by the blast, we didn't have enough data to achieve the safe design. In the case of very large explosion accident, we might have to consider not only about the direct damage by the blast, but also about secondary disasters caused by the fragments or the collapse of walls. Authors performed the experiment on the cooperation with SRI International, and we collected fundamental data [6], [7]. Fig.4 and Fig.5 show the tested system and some of high-speed-camera images, which captured the moment of the explosion. In this experiment, a pre-mixed hydrogen gas with 30% in concentration filled in a plastic tent with 37m 3 in volume was detonated using a small amount of C-4 explosive, and the response and the damage of reinforced-concrete walls stand near the tent were observed.

5 Fig.4 Tested systems used for pre-tests (left) and main tests (right) Fig.5 some of high-speed-camera images, which captured the moment of the explosion The 22 kinds of walls with different height, thickness and reinforcements were tested, and a wide range of response was observed. Some walls showed elastic or plastic behaviors and others were collapsed (Table2 and Fig.6). Probably, the experiment, which provided such wide range of response from a single experimental system, was the first one in the world. Authors obtained much valuable data from this experiment. Fig.7 shows the time histories of blast overpressure and that of the displacement of the wall (this figure shows displacements at the top and at the middle of the wall). Overpressure showed a steep peak with the arrival of shock front, and its duration time was very short. On the other hand, displacement started to rise delaying Table2 Test parameters and the results Fig.7 Time histories of blast overpressure and that of wall displacement

6 1m-height-walls 1.5m-height-walls Collapsed 1m-height-walls Collapsed 1.5m-height-walls 2m-height-walls Fig.6 Cracks and Damages of RC walls observed after the explosion tests Collapsed 2m-height-walls

7 after the peak of overpressure. Displacement at the top was smaller than that at the middle in the early stage, but it exceeded that at the middle after 0.01ms. This means the mode of wall's deformation changed during the response. Damage of structures has often been considered to occur with the arrival of blast, and many of the experimental results have been classified using the peak overpressure. However, our result indicated these are not correct ways. We considered that following process has occurred based on our experiments. At the first stage, a blast impact of the explosion makes the stress concentration at the bottom of the wall. At the next stage, a local displacement, caused by this stress concentration, shifts to the deformation of whole wall, and we thought this makes significant damage on the wall. Displacement data explains the change of deformation mode. Seto et.al. [8] explained our result of experiment using a sophisticated theoretical approach, and they proved the existence of such change of deformation mode. We considered that the dominant deformation modes which have made damages appeared depending on the eigen frequency of the wall. Cracks and breakages of walls after the experiment showed the traces of n=2 or n=3 mode deformations. In the previous paragraph, we indicated the necessity of higher strength in the design of tall walls. We consider that a rigid and reasonable wall design becomes possible, observing the damage mechanism obtained our experiment. We think it important to design an eigen frequency of the wall not to meet the duration time of the blast. 4. Defense from the vehicle crash accidents 4.1 Defense from the reckless driving large-sized freight vehicles Authors thought that a study on the accident potentials such as the vehicle crash accidents was also necessary, because hydrogen stations would be used to refuel FC vehicles in the future. An ability of guard fences to defend the station from reckless vehicles was studied using the computer simulation technique [9]. Fig.8 shows the result of computer simulations assuming the crash accident of large-sized freight vehicle into the guard fence. This simulation was performed using the 3D crash analysis code LS-DYNA. FEM Vehicle models were provided from NCAC (National Crash Analysis Center). We simulated the crash of a vehicle into the guard fence, which had stiffness from 60 to 130 kj. Assumed vehicle had 15.8ton in weight and 40km/h in speed, and we assumed the collision angle as 15 degrees. As the result, vehicles easily invaded in the area in all cases. On the other hand, in case if we assumed guard fence built on a step, vehicles couldn't invade the area across the border. Steps with height of 150mm showed enough effect. Desirable step height was more than 250mm. 4.2 The crash defense by the mis-starting vehicles In the area of refueling stations, sometimes the crash accident occurs caused by the mis-starting vehicles. Measures to the mis-starting vehicles were also studied. In most of hydrogen stations, high-pressure piping systems are installed under the pit in order to prevent crash accidents. However, stations should have refueling dispensers, which contains high-pressure pipes at the ground. Usually, refueling dispensers are guarded with bollards (hard posts). We studied how much stiffness was necessary on the bollard against the vehicle collisions. Fig.9 shows an example of the simulation results. As a result of the simulation assuming the head-on collisions and the offset collisions, necessary stiffness for the bollard was 130kJ against the

8 Guard fence with a stiffness of 60kJ Guard fence with a stiffness of 120kJ Guard fence with a stiffness of 120kJ + 250mm step Fig.8 Simulated crash accident of large-sized freight vehicle into the guard fence Fig.9 An example of simulations on the effect of bollards against the crash of mis-started compact vehicles collision of compact vehicles with 30km/h in speed. Steps were also effective if the collision angle was very shallow as less than 30 degrees. The well considered design on the vehicle's routes and the layout of islands (area with different level by the step surrounding dispensers) will enable high crash defend-ability stations. 5. Conclusions Results of our works on the safety use of hydrogen energy were described in this paper. These results are applicable to the safe design of a lot of kinds of hydrogen facilities. The official report of this research project, which contains detailed experimental data, is now on the public domain and it can be downloaded from the Internet [1]. We hope our result will be used widely to realize the safety on the use of hydrogen energy. ACKNOWLEDGEMENTS: This research was performed in the project of New Energy and Industrial Technology Development Organization in Japan (NEDO) titled "Research on the safety of hydrogen infrastructure and building frame structure against hydrogen explosion and earthquake" held during 2003 to We had a lot of advices from Professor Tomonori Ohno in the National Defense Academy of Japan, and these advices were always helpful for this project. We performed the explosion tests based on the complete cooperation of SRI International. Dr. James Colton, Dr. Mark Groethe, Dr. Paul Gefken, Dr. Erik Merilo in the Poulter laboratory in SRII and Dr.

9 Seiki Chiba in SRII Japan made a maximum effort to succeed in the experiment, and they also provided us a lot of fruitful advices. Professor Kenji Seto in the Hokkai-Gakuen University explained our experimental result using a very sophisticated theoretical approach. Dr. Dhafer Marzougui in National Crash Analysis Center (NCAC) kindly permitted us to use their FEM vehicle models for our crash simulations. We would like to thank all of the people who supported and gave us helpful cooperation to succeed in our research project. REFERENCES: [1] NEDO: Research on the safety of hydrogen infrastructure and building frame structure against hydrogen explosion and earthquake, NEDO and NEDO , (2005), [2] Kenji, Mikoda and Yoshihide Suwa: Countermeasures to accidents due to hydrogen-gas explosion (part4) -Analysis of hydrogen-gas accidents and systemization of factors contributing to them-, Obayashi Technical Research Institute Annual Reports, 69(2005). [3] Yoshiyuki Ono, Yoshihide Suwa and Kenji Yonezawa: Effects of the hydrogen explosion on the structure, part1 computational simulation of the pressure characteristics by the hydrogen explosion, Hydrogen Energy Systems (submitted). [4] Yoshihide Suwa, Kenji Yonezawa and Yoshiyuki Ono: Countermeasures to accidents due to hydrogen-gas explosion (part1) -Effect of protective walls against explosion accidents in hydrogen refueling stations-, Obayashi Technical Research Institute Annual Reports, 68(2004). [5] Hideo Miyahara, Keiji Kubo, Mitsuru Saito, Yoshihide Suwa, Kenji Yonezawa, Kazuhiro Naganuma and Katsuyoshi Imoto: Research on the safety of hydrogen infrastructure and building frame structure against hydrogen explosion and earthquake, 15th. World Hydrogen Energy Conf. WHEC 2004, ( Yokohama). [6] Kenji Yonezawa, Yoshihide Suwa, Yoshiyuki Ono, Kazuhiro Naganuma, Katsuyoshi Imoto and Keiji Kubo: Experimental study on nonlinear response of RC wall subjected to hydrogen explosive load, Obayashi Technical Research Institute Annual Reports, 69(2005). [7] Kenji Yonezawa, Yoshihide Suwa, Yoshiyuki Ono and Kazuhiro Naganuma: An experimental study on nonlinear response of RC wall subjected to hydrogen explosive load, J. of Struct. Constr. Eng., Architectural Institute of Japan, 601 (2006), pp [8] Kanji Seto and Taijiro Nonaka: Closed-form solution of elastic panel under impulsive loading (will be published). [9] Yoshihide Suwa: The protecting structures against collisions by recklessly driven or miss-operated vehicles, Proc. of Ann. Conf. of Architectural Institute of Japan (2005).