COMPUTER SIMULATION OF EXPLOSION OF A HYDROGEN STORAGE TANK


Mohsen Sanai
SRI International
Poulter Laboratory
333 Ravenswood Avenue
Menlo Park, California 94025

Abstract

SRI International, a not-for-profit research organization chartered in the United States, has been tasked to evaluate the safety of hydrogen in various operations and processes of the World Energy Network (WE-NET) system. Based on the available experimental data and advanced computer simulation of explosion accidents, SRI has developed a novel computer algorithm called accident analysis computer algorithm (AACA:AACA is a protected intellectual property of SRI International.) that employs the pressure-impulse (PI) methodology to quickly calculate the response of humans and structures to postulated explosion accidents. The results are displayed graphically in the form of color contours of equal damage levels. Because the probability of occurrence is included as part of the PI isodamage curves, the AACA can also function as a probabilistic risk assessment tool. Also, an economic analysis module can be incorporated into the AACA to allow the safety engineers to find the most cost-effective solution to hazards mitigation. The load-damage correlations used in this algorithm are being expanded to include other WE-NET accident scenarios.

1. Introduction

SRI has been tasked with the evaluation of hydrogen safety in various operations and processes of the World Energy Network (WE-NET) system including hydrogen production, post-production storage, marine transportation, pre-use storage, land transportation, and industrial utilization. Our general approach to determining the consequences of postulated accidents involves four steps: (1) identifying and characterizing the hazard sources including explosion, fire, and fragments, (2) determining the load environment experienced by objects that surround the accident site, (3) determining the resulting damage and whether it leads to system failure, and (4) assembling the results into a user-friendly PC-based algorithm that would allow the safety engineers to quickly investigate postulated accident scenarios and evaluate possible mitigation schemes.
Examples of computer simulations that cover steps 1 through 3 of several postulated accidents are discussed in References 1 and 2. Detailed modeling of the explosion accidents have included the detonation of the energetic source and subsequent expansion of the detonation products, formation and propagation of the blast wave, loading of the surrounding objects and structures, and modeling of the structures to estimate the damage caused by the blast loads. Examples described in Reference 1 are the hazard evaluation of an operation involving energetic materials, modeling an accidental explosion of energetic materials stored in a bunker near a residential area, explosion of a vapor cloud floating above the ground, and modeling an accidental airplane crash above a concrete instrumentation bunker. In Reference 2, we have illustrated the use of advanced computer simulation techniques to analyze the explosion of a pressurized hydrogen tank and the response of structures and people exposed to the resulting blast environment.
The remainder of this paper addresses the fourth step mentioned above, namely, development of a PC-based computer algorithm for quick evaluation of postulated accident scenarios.

2. ACCIDENT ANALYSIS COMPUTER ALGORITHM (AACA)

Although significant advances have been made in computer simulation techniques, performing hundreds of dedicated calculations needed to evaluate each and every postulated accident scenario is not practical because these calculations are usually (a) computationally intensive and time consuming, (b) technically complex requiring a high degree of expertise, and (c) accessible only to a select few due to licensing, maintenance costs, and export restrictions. The SRI AACA is based on the premise that the results from the computer simulations and experimental data can be assembled into pressure-impulse (PI) diagrams in which the response of structures (or humans) are correlated with the blast environments resulting from postulated accidents. This methodology, developed and used at SRI since 1976 (Reference 3), is forming the basis of a continually expanding library of PI load and damage curves. These curves serve as reference look-up charts and tables in the AACA, thus allowing the user to perform almost instantaneous calculations of several postulated accident scenarios.
The output of the AACA is illustrated in Figures 1 through 6. Figures 1 shows the entry page into the AACA and allows the user to select any one of the available accident analysis programs. For example, Figure 2 shows the leading page of a program devised to assess the effect of a terrorist bomb onboard an aircraft. The type of aircraft, location, and weight of the bomb are selected by answering the screen prompts. The resulting damage, expressed as a fraction of the strain necessary to rupture the panels, is displayed graphically (Figure 3) based on the (generic) PI diagram shown in Figure 4.
For explosion of a pressurized hydrogen tank, the graphical display is in the form of concentric circles that indicate the human response to the blast environment generated by the explosion. Figures 5 and 6 show typical displays of the eardrum damage and lung collapse, respectively. As before, the input required for performing these calculations is provided by answering the prompts that appear on the screen. For the hydrogen tank explosion, the input required is the tank size, initial pressure, location of interest, and the type of response to be calculated.
The PI library for human responses includes the probability of occurrence, so the PI methodology used in the AACA qualifies as a probabilistic risk assessment method, which has gained acceptance as part of normal plant operation. Moreover, an economic analysis module can be readily incorporated into the AACA to allow the safety engineers to find the most cost-effective solution to hazards mitigation.

3. FUTURE WORK

The AACA is being extended to accidents that involve fire. For this case, the pressure is replaced by a flux modulus that represents the peak thermal flux absorbed by the object of interest, and the impulse is replaced by a fluence modulus that represents the thermal exposure. The AACA would then be able to calculate significant thermal responses, such as the probability of skin burn or ignition of wood, based on the input conditions that characterizes the fire size, absorptivity, and density and thermal conductivity of the material.

4. Acknowledgment

The hydrogen tank explosion analysis discussed in this paper was conducted under the program "International Clean Energy Network Using Hydrogen (WE-NET) Subtask 3," consigned to the Institute of Applied Energy from the New Energy and Industrial Technology Development Organization, which is carried under the New Sunshine Project administered by the Agency of Industrial Science and Technology, the Ministry of Trade and Industry, Japan.

5. References

1. M. Sanai, J. D. Colton, and G. G. Greenfield, "Consequence Analysis Based on Numerical Simulation of Postulated Accidents," Proceedings of the Conference on Process Plant Safety Symposium, Houston, Texas, 18-19 February 1992.
2. M. Sanai, "Use of Advanced Computer Simulation Techniques and Pressure-Impulse Methodology for Investigation and Mitigation of Postulated Explosion Accidents," Proceedings of the International Hydrogen and Clean Energy Symposium, Tokyo, Japan, February 1995.
3. G. R. Abrahamson and H. E. Lindberg, "Peak Load-Impulse Characterization of Critical Pulse Loads in Structural Dynamics," Nuclear Engineering and Design, 37, 35-46 (1976).