1. Canadian Hydrogen Safety Program Comparative Risk Estimation of Hydrogen and CNG Refuelling Options NHA Annual Hydrogen Conference 2007 San Antonio, March 20, 2007 Andrei Tchouvelev / Bob Hay / Pierre Benard

2. 2 Outline  Acknowledgements  Project Rationale and Scope  Qualitative Risk Comparison  Failure Scenarios  Release and Dispersion Modeling  Ignition Probabilities  Thermal Effects Analysis  Risk Metrics and Quantitative Risk Estimation  Conclusions

3. 3 Acknowledgements Financial Support Natural Resources Canada AVT, TISEC and HRI and Collaborating industrial partners Special Thanks Stefan Unnasch from TIAX Marsh FM Global

4. 4 Project Rationale  Facilitate acceptance of the products, services and systems of the Canadian Hydrogen Industry by the Canadian Hydrogen Stakeholder Community: Industrial – to facilitate trade Insurers – to ensure fair insurance rates Regulators – to ensure effective and efficient approval procedures General Public – to ensure diverse interests are accommodated

5. 5 Project Technical Rationale  Deficiencies of existing RA tools: Both phenomena and algorithms are tailored to hydrocarbons – “big refinery” approach “Black box” approach – out of user’s reach  Need to develop framework for “ultimate” risk estimation tool: Phenomena and algorithms are understood and applicable to hydrogen Open for examination, technical input and use by the hydrogen community

6. 6 Scope Rationale  Comparative risk estimation project of CH 2 refuelling options: With CNG in order to compare with similar compressed gas system Examine only stand-out scenarios Concentrate on most obvious consequence – jet fire radiative heat Detailed use of CFD and fault / event trees software Estimate risk from individual components at selected coordinates / locations as examples

7. 7 Scope Quantitative risk comparison of hydrogen and natural gas refuelling options Project Scope Sourcing Options HydrogenNatural Gas Delivery Compressed Gas Pipeline On-Site Generation Reforming fuel comparison scenario Electrolysis

8. 8 Qualitative RA – TIAX FMEA Study FMEA for Hydrogen Fueling Options, CEC-600-2005-001 Methodology and Risk Binning Matrix:

9. 9 Qualitative RA – TIAX FMEA Study FMEA for Hydrogen Fueling Options, CEC-600-2005-001 Design Baseline Considerations:

10. 10 QRA Project Failure Scenarios  Tube Trailer: S1: Small horizontal leak during unloading S2: Catastrophic horizontal leak during unloading  Electrolyser: S3: Catastrophic internal leak at hydrogen rinser S4: Venting of catastrophic internal leak S5: H2 Leak outdoors between compressor and storage  Reformer: S6 (and S9): NG supply line leak outdoors S7: NG line leak between NG compressor and reformer S8: Catastrophic internal leak at PSA unit  CNG Station: S10: CNG leak outdoors between compressor and storage  Gas Storage: S11: Horizontal jet release (H2 and CH4) at equal pressure and orifice S12: Venting of H2 and CH4 at equal flow rate

11. 11 Tube Trailer Failure Scenarios Small leak via 1 mm orifice, 2640 psig: LFL horizontal extent 4.26 m Catastrophic leak via ½” OD orifice, 2640 psig: LFL horizontal extent 40.5 m

12. 12 Horizontal Jets Comparison Storage: hydrogen and CNG release and dispersion from a shut-off valve fitting through a ½” OD at 4125 psi Leak orifice ½” OD, 8.48 mm ID Leak direction/location: horizontal leak 1 m above the ground Domain size: symmetric, 100m by 8 m by 25 m. Water volume of cylinders: 2.811 m 3 Initial stagnation pressure: 284.4 bars Choked leak duration: H 2 : 80 seconds, CH 4 : 240 seconds Simulation time: 0-90 seconds for CH 4, 0-60 seconds for H 2

13. 13 Steady state CFD results: H 2 vs. CH 4  Steady CFD results by PHOENICS 3.6.1 Horizontal leak 1m above the ground, OD ½” (ID 8.48 mm) orifice High pressure 284 bars  H 2 : 43 m, CH 4 : 68 m

14.  Steady CFD results by FLUENT Horizontal leak 1m above the ground, OD ½” (ID 8.48 mm) orifice High pressure 284 bars  H 2 : 32 m, CH 4 : 46 m Steady state CFD results: H 2 vs. CH 4 14

15. 15 Transient CFD results: H 2 vs. CH 4

16. 16 Ignition Probabilities Approach Developed by DNV with input from AVT  Starting point – historical data from DNV database RELEASE RATE CATEGORY RELEASE RATE (kg/s) GAS LEAK CRUDECLASS ICLASS IICLASS III Small< 10.010 0.0060.0040.002 Large1 – 500.0700.0300.0180.0120.007 Massive> 500.3000.0800.0490.0310.018 Historical Ignition Probability Data for Hydrocarbons (Cox, Lees & Ang)

17. 17 Ignition Probabilities Considerations  All considered H2 leaks are less than 1 kg/s  Historically reported ratio of immediate to delayed ignition probability is 2 to 1  What is realistic probability for H2 – 1% seems low  Key considerations in comparison with methane: For a given mass leak, H2 would produce appr. 8 times bigger flammable cloud than methane (their LFL’s are close) Delayed ignition probability is proportional to the flammable cloud size. Hence, 1 kg/s leak for methane is “equivalent” in volume to 0.125 kg/s for H2 Though the flammable range of H2 (4 to 75% vol.) is 7.3 times greater than that of methane (5 to 15% vol.), for both gases the size of a cloud above 15 % vol. is about 16% of the total size of cloud above LFL

18. 18 Ignition Probabilities Considerations Minimum ignition energy vs H2 concentration in air As presented by M.Swain on May 24, 2004 Hydrogen ConcentrationMinimum Ignition Energy Required (mJ) 29% (stoicheometric)0.02 10%0.15 9%0.21 8%0.33 7%0.56 6%1.0 5%3.0 4%10.0

19. 19 Ignition Probabilities Approach Adopted ignition probability As proposed by DNV RELEASE RATE CATEGORY HYDROGEN RELEASE RATE (kg/s) HYDROGEN TOTAL IGNITION PROBABILITY HYDROGEN IMMEDIATE IGNITION PROBABILITY HYDROGEN DELAYED IGNITION PROBABILITY Small Leak < 0.1250.0120.0080.004 Large Leak 0.125 – 6.250.080.0530.027 Massive Leak > 6.250.350.230.12 Flammable H2 gas mixture within closed systems Not Applicable 1-1

20. 20 Jet Fire Thermal Radiative Flux Model References Y. R. Sivathanu and J. P. Gore, 1993. W. Houf and R. Schefer (SNL), 2004-6.  Has been recently validated by SNL for free H 2 jet flames T. Mogi et al (AIST), 2005.  Used for verification purposes TNO “Yellow Book”, Part 2, p.6.48, 1997.  Used for vertical flares  Still needs to be validated for hydrogen

21. 21 Tube Trailer Thermal Effects Scenario 1: Small leak via 1 mm orifice, 2640 psig LFL horizontal extent 4.26 m

22. 22 Tube Trailer Thermal Effects Scenario 2: Catastrophic leak via ½” OD orifice, 2640 psig LFL horizontal extent 40.5 m

23. 23 Storage Horizontal Jet Thermal Effects Scenario 11: Catastrophic leak via 8.48 mm orifice at 282 bars from CNG storage system

24. 24 Storage Horizontal Jet Thermal Effects Scenario 11: Catastrophic leak via 8.48 mm orifice at 282 bars from H 2 storage system

25. 25 Storage Venting Thermal Effects Thermal Flux from CNG and H2 Storage Venting

26. 26 Risk Estimation Risk Metrics  Location Specific Individual Risk (LSIR) where the summation is extended for all scenarios and: Fs is the frequency of the scenario S P F is the probability of death in the scenario for an individual at the location The frequency of the scenario is taken as: where F O is the end outcome frequency calculated from the post-incident event trees with the formula: where F i, is the failure frequency of the initiating event for the scenario calculated using a fault tree analysis P b is the probability of an individual segment of the event tree such as probability of immediate ignition or probability of delayed ignition

27. 27 Risk Estimation Risk Metrics  Potential Loss of Life (PLL) where n present is the number of persons present and exposed to the event where A = -36.38; B = 2.56; I = heat radiation load any value greater than 1.6 Kw/m 2 ; t = 20 or 60 seconds (except for electrolyser and reformer cases where leak durations were very short)  Specific locations with coordinates x=1, R=1 and x=5, R=1 were selected for comparison for all scenarios  Probit equations (TNO “Purple Book”) were used to estimate consequence of each fire scenario:

28. 28 Tube Trailer Risk Estimation

29. 29 Gas Storage Risk Estimation

30. 30 Summary  Qualitative Assessment: Good tool for rough / preliminary risk ranking Maybe be misleading regarding real consequences / risks  Comparison Among Hydrogen Options: Sourcing hydrogen on-site and off-site present almost the same risk From the individual risk, the electrolysis process presents the lowest risk  CNG and Hydrogen Storage Comparison: Hydrogen storage facility presents a marginally lower (within 20%) risk compared to an identical CNG storage In terms of storage venting, a CNG storage facility may require either a larger clearance than an identical hydrogen storage facility or a higher vent stack to achieve the same level of thermal radiation from a vertical flare  Risk Comparison : In summary, an electrolysis refuelling option that includes compressed hydrogen storage presents the lowest risk among the refuelling options that were considered including a CNG station of equal refuelling capacity to provide equivalent travel mileage

31. 31 Next Steps Integration of RA tools:  Integrate radiative heat estimation software with risk estimation software  Integrate frequency / probability software Develop new and refine existing algorithms:  Continue refinement of existing algorithms, e.g.: Wall and transient effects Vertical flares – validate for hydrogen  Develop / validate algorithms fully applicable to hydrogen, e.g. flash fires / VCE thermal effects, DDT Expand risk estimation scope:  Compare individual components and complete refuelling options in terms of distances to maximum acceptable risk levels

32. 32 Contact Information Andrei Tchouvelev: atchouvelev@tchouvelev.org Bob Hay: bobhay@tisec.com Pierre Bénard: Pierre.Benard@UQTR.CA  CTFCA website http://ctfca.nrcan.gc.cahttp://ctfca.nrcan.gc.ca (English) http://acpct.rncan.gc.cahttp://acpct.rncan.gc.ca (French)  www.tchouvelev.org www.tchouvelev.org  www.hydrogensociety.net www.hydrogensociety.net