1. Risk informed separation distances for hydrogen refuelling stations
Frederic Barth
Air Liquide Hydrogen Energy

2. Background and general motivation
Approach developed for ISO/DIS Gaseous Hydrogen – Fuelling stations within TC197/WG 11 Fueling stations by TG1 Separation distances
To substantiate lay-out requirements for HRS sub-systems
Applied to gaseous hydrogen systems
Hydrogen supply system (e.g. tube trailer)
Hydrogen compression skid
Hydrogen buffer storage
Hydrogen dispensers
Hydrogen is being developed for generalized use as an energy carrier:
Higher operating pressures than previously considered
Installation and use in public settings
Variety of applications (e.g. RV fuelling stations, back-up power, materials handling…)
Inherently safe designs and built-in safety measures
 Need of a robust rationale and approach for addressing these new applications consistently

3. Separation distances in codes & standards Rationale
Purpose : a generic means for mitigating the effect of a foreseeable incident and preventing a minor incident escalating into a larger incident (EIGA IGC 75/05)
Apply separation as appropriate, along with other means, to achieve freedom from unacceptable risk
Separation is not always necessary, nor most appropriate means
Where applied, appropriate separation can be defined by application of a risk criterion
Protection against catastrophic events is essentially achieved by other means than separation, such as prevention, specific means of mitigation, or emergency response, which are also addressed. 3

4. Separation distances in codes & standards Form of specification
Continue to express requirements by means of a good table that is suitable for the covered application
Most practical
Tabled distances have been checked
Same distance for similar systems supports standardization
Relying on a formulas raises the risk that design parameters will be chosen to minimise safety distance requirements although this choice does not reduce the actual risk level to exposures
Practical value added of specifying distance by means of formulas is not clear
Different applications may require different tables
e.g. Fuelling stations, bulk hydrogen storage systems, hydrogen installations in non industrial environment

5. Table based separation distances specification – Basic steps
Table Lines : Exposures or sources of hazard ; Columns: system category
Select system characteristics that fundamentally determine actual risk impact
Define system categories associated to a graduation of risk impact
Taking into account different types of equipment actually used
Limit the number of categories to justified need
Use a risk model to determine the separation distances for each category, by application of a criterion on estimated residual risk,
Based on max values for the category
Higher risk  Greater separation
Populate the distance table and evaluate the result.

6. Selection of system characteristics that fundamentally determine actual risk impact
Separation distances should not be determined only by Pressure and Internal Diameter. Need to integrate fundamental factors determining actual risk impact, such as inventory, system complexity, and exposure criticality
Over sensitivity to a detail design parameter such as internal diameter needs to be avoided

7. Selection of system characteristics that fundamentally determine actual risk impact
Storage system size
Small
Large
Complexity level as reflected by number of components
Very simple (for Small systems only)
Simple
Complex
For Small systems only : pressure
Regular
High

8. Categorization of compressed hydrogen storage systems
Boundaries defined according to equipment types in use

9. Resulting categorization for gaseous hydrogen storage systems
8 categories

10. Cumulated frequency of feared effects from leaks greater than X g/s
Risk model for determination of a separation distance requirement from a system
Separation distance
To be applied 1 10 3 30
Separation
distance (m)
Reference
leak size
occ./yr
Leaks 10 10 - 1
Cumulated frequency of feared effects from leaks greater than X g/s 10 - 2
Feared Effect 10
Frequency 10 10 - 3 - 3 10 - 4 10 - 4
Target 10 10 - 5 - 5 10 - 6 10 - 6
Leak
Leak
rate
rate
0,01
0,01
0,1
0,1 1 1 10 10
100
100
(g/s)
(g/s)

11. Key parameters of risk model
Cumulative leak frequency vs leak size
See next slides
Probability of having the feared event (injury) when a leak occurs
Pignition x Geometric factor = 0,04 x 0,125 = 0,005
Consequence model providing distance up to which leaks can produce the feared event, in function of leak size and type of feared effect (e.g thermal effects or 4% H2 concentration)
Sandia National Laboratories jet release and fire models
Target value for the feared event frequency,
Non-critical exposure: 10-5 /yr
Critical exposure: /yr
Risk model does not provide an accurate evaluation of risk, but allows to take into consideration the main risk factors consistently
 Separation distances are risk informed

12. Determination of system leak frequency distribution in function of component leak frequency distribution
Consider main contributors to leaks
Joints, Valves, Hoses, Compressors
Estimate cumulated leak frequency in function leak size (% of flow section) for each type of component, from available statistical data
Estimate cumulated leak frequency in function of leak size for the whole system, by summation of contributing component leak frequency data

13. Component leak frequency – Source of input to risk model
Risk model requires leak frequency input for following leak size ranges : [0.01% ; 0,1%], [0.1% ; 1%][1% ; 10%][10% ; 100%]
Use of published leak frequencies compiled by SNL (J. LaChance) Extract for valves, where information on leak size is provided (34% of records):
Data input to risk model: Leak size range [0.01% ; 0,1%] [0.1% ; 1%] [1% ; 10%] [10% ; 100%] Log. average freq. of extrapolated “Small leaks” “Large leaks” “Ruptures”

14. Risk model leak frequency input for valves (1)
Frequency and size of “small leaks”

15. Risk model leak frequency input for valves (2)
Frequency and size of “small leaks”

16. Risk model leak frequency input for valves (3)
Note : adequacy of using log-average of “Small leak”, “Large leak”, and “Rupture” frequencies as risk model input for [0.1% ; 1%], [1% ; 10%],[10% ; 100%] ranges was verified for all types of components

17. Risk model component leak frequency functions

18. Consequence Model
Interpolation of flame length and flammable cloud length formulas developed by SNL (Bill Houf) :

19. Risk informed leak diameters & separation distances for storage/transfer systems

20. Separation distance requirements for compressed for gaseous hydrogen storage/transfer systems

21. Thank you frederic.barth@airliquide.com