1. Funded by FCH JU (Grant agreement No. 256823) 1 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 1

2. Funded by FCH JU (Grant agreement No. 256823) 2 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 2 1.Inherently safer design of FC systems (coupling of technological and safety requirements) 2.Separation distances for unignited leak and jet fire: which is longer? 3.Strategy for safer use of forklift in warehouse (calculation of PRD diameter by the similarity law for unignited releases to exclude cloud formation) 4.Effect of restrictor in a pipeline (the dimensional flame length correlation) 5.Momentum-dominated and buoyancy-controlled leaks: decrease of separation distances for hydrogen transport pipelines 6.Plane jets and reduction of separation by innovative PRD (Pressure Relief Devices) 7.Increase of fire resistance of onboard storage

3. Funded by FCH JU (Grant agreement No. 256823) 3 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 3  Let us consider 50 kW FC system (multi-family dwellings) and 150 kW FC for vehicle. Assuming electrical efficiency of FC is 45%, heat of reaction 132.5 kJ/g, the mass flow rate for functioning of FC can be calculated as only (50 kW)/0.45/(132.5 kJ/g)= 0.84 g/s and 2.52 g/s respectively.  For example, the mass flow rate 0.84 g/s can be provided at P =5 bar and D =1.8 mm restrictor ( or P =2 bar, D =2.9 mm )  Unfortunately, current FC systems are often designed that piping diameter is D =5-15 mm (25 mm!?) and pressure is P =5-15 bar.  Maximum mass flow rate for D =15 mm, P =15 bar is 170 g/s. Significant excess of required mass flow rate (below 3 g/s ). FC design to be changed.  Expected result: essential decrease of separation distance and improvement of safety of FC and hydrogen systems. For example, flame length is proportional to mass flow rate as ( m ) 1/3. Matching technology and safety would decrease flame length by (170/0.84) 1/3 = 6 times!.

4. Funded by FCH JU (Grant agreement No. 256823) 4 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 4  Interrelationships between prescriptive and performance-based approaches to regulation are well established for fire safety engineering are now under discussion for hydrogen safety engineering.  Usefulness of prescriptive codes for separation distances, e.g. International Fire Code (edition 2006) is questionable. For example, IFC provides separation distance from non-reacting leaks in Tables without any reference to the original source of information or reasoning.  In particular, it is impossible to determine separation distance without real parameters of a particular system:  Storage pressure  Leak diameter  Without science-informed approach such kind of RCS should be avoided for safe introduction of HFC systems and infrastructure.

5. Funded by FCH JU (Grant agreement No. 256823) 5 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 5  An unignited jet deterministic separation distance is the distance where hydrogen concentration decays to 4% by volume (e.g. to exclude penetration of flammable hydrogen-air mixture from intake through ventilation systems to buildings). The similarity law gives the distance to 4% in a momentum-dominated jet (conservative estimate compared to a buoyancy-controlled jet)  The unignited jet separation distance is clearly a function of a leak diameter D and hydrogen density in the leak exit  N (directly related to storage pressure).  The dimensionless jet flame length, L F /D, is a function of storage pressure too (see next slide: flame tip is located where the unignited jet from the same source would decay to 8%-16% of hydrogen by volume).  Which separation from a leak is longer: for unignited jet or jet fire?

6. Funded by FCH JU (Grant agreement No. 256823) 6 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 6 Flame tip location: from 8% to 16% in unignited jet (best fit – 11%). Conclusion (see chapter F): flame is longer than the distance to axial concentration 29.5% in unignited jet (stoichiometric hydrogen-air mixture), where location of the flame tip was mistakenly identified previously, by 2.2 times (16%) to 4.7 times (8%)! 11% 8% 16%

7. Funded by FCH JU (Grant agreement No. 256823) 7 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 7 x =2. L F – “death” limit (300 o C, 20 s) x =3. L F – pain limit (115 o C, 5 min) x =3.5. L F – “no harm” limit (70 o C)

8. Funded by FCH JU (Grant agreement No. 256823) 8 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 8  The ratio of distances to 8% of hydrogen by volume (conservative estimate of the flame length) to 4% (lower flammability limit = LFL ) is x 8% / x 4% = L F / x 4% =0.48 (by the similarity law for decay of hydrogen in momentum-dominated expanded and under-expanded jets).  Three separation distances for jet fire are: death limit x/L F =2 ; pain limit x/L F =3 ; no harm limit x/L F =3.5.  Thus, the ratio of three separation distances from jet fire to the separation distance to LFL in unignited jet from the same leak source is:  Death limit (309C, 20 s): x T=309C(8%) /x 4% =0.48* 2 = 0.96 ;  Pain limit (115C, 5 min): x T=115C(8%) /x 4% =0.48* 3 = 1.44 ;  No harm distance (70C): x T=70C(8%) /x 4% =0.48* 3.5 = 1.68 ;  “Unexpected” conclusion: all three separation distance for jet fire are equal or longer then separation distance to LFL (non-reacting release).

9. Funded by FCH JU (Grant agreement No. 256823) 9 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 9  Warehouse is closed space and probably the worst-case scenario is deflagration of flammable hydrogen-air cloud created during unscheduled hydrogen leak from storage tank (through pipeline to FC or through PRD).  One possible safety strategy is to exclude formation of flammable hydrogen-air mixture under the ceiling.  If this safety strategy is realised then only a small volume of the warehouse (within flammable part of the jet) will be occupied by flammable mixture and deflagrate (assessment of this deflagration overpressure effects should be done separately – not part of this example).  Let us calculate PRD diameter that in case of unscheduled release, e.g. by PRD fault, would not create flammable cloud under the ceiling: hydrogen concentration should decay to at least 4% by volume (LFL) before reaching the ceiling.

10. Funded by FCH JU (Grant agreement No. 256823) 10 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 10  Example : upward (worst case in sense of flammable cloud formation) release from the forklift onboard storage at 35 MPa. The distance from PRD to warehouse ceiling is 10 m.  To realize the suggested strategy the concentration on the jet axis at distance 10 m should be below 4% of hydrogen by volume (corresponding mass fraction is C ax =0.00288 ). The similarity law to be applied  The under-expanded jet theory gives  N =14.6 kg/m 3 for storage pressure 35 MPa (simplified calculation would give the estimate: 350/2*0.084= 14.7 ). Thus, the PRD diameter can be calculated straight forward 1.5 mm

11. Funded by FCH JU (Grant agreement No. 256823) 11 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 11  A restrictor is often used on the exit from a high pressure storage to a pipeline leading to a fuel cell (FC). By definition the restrictor orifice diameter is smaller than internal diameter of the pipeline.  There are examples of systems when diameter of pipeline from high pressure storage to dispenser at a hydrogen refuelling station (RS) is 25 mm (?!).  Let us consider an imaginary scenario that 1 mm restrictor placed in a pipeline from storage to dispenser at RS. Mass flow rate through 1 mm orifice at storage pressure 1000 bar is about 2.1 kg/min (sufficient to fill in the car in couple of minutes) = 0.035 kg/s.  How the flame length would change if a designer would choose either 3 mm or 25 mm internal diameter of the pipeline?  The dimensional correlation for flame length is convenient tool to be applied for this case (see next slide).

12. Funded by FCH JU (Grant agreement No. 256823) 12 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 12 For D =0.025 m ( m =0.035 kg/s, d =1 mm) L F =6.6 m (“no harm” – 6.6x3.5= 23 m ). For D =0.003 m ( m =0.035 kg/s, d =1 mm) L F =3.2 m (“no harm” – 3.2x3.5= 11.2 m ). Conservative separation is 50% longer ( 34.5 m and 16.8 m ). Conclusion: separation depends on pipe diameter even if restrictor diameter is the same!

13. Funded by FCH JU (Grant agreement No. 256823) 13 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 13  Let us consider unignited release for the same two scenarios.  How the separation distance would change if the unignited release is through the full bore of broken pipeline with internal diameter 25 mm and 3 mm (for both cases the same restrictor orifice diameter of 1 mm )?  The similarity law is applied for the assessment of separation distance  Let us assume: mass flow rate is defined by the restrictor only (no losses), and flow is choked, i.e. velocities are the same (no difference in T during expansion). Thus:  Thus, we can conclude from the similarity law that the separation distance for unignited release in assumptions made does not depend on pipe diameter (only on restrictor size). This is not the case for fire!

14. Funded by FCH JU (Grant agreement No. 256823) 14 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 14  We have just considered high pressure under-expanded jet fires.  Let us now estimate how flame length would change if our flow parameters are in the momentum-controlled regime for expanded jets when the dimensionless flame length is the constant ( L F /D =230 ).  Conclusion: for momentum- controlled expanded jet fires the flame length, and thus separation distances, grow proportional to diameter.

15. Funded by FCH JU (Grant agreement No. 256823) 15 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 15 Example of unignited separation distance decrease by buoyancy  Since 1938 the chemical industries in Hϋls, Ruhr area (Germany): pipeline 215 km, maximum pressure 25 bar, inner diameter D =16.8-27.3 cm (after expansion of jet D eff =98 cm ).  Full bore rupture mass flow rate 90 kg/s (by the way such pipeline could serve a big city with population of 10 million people: 15 cars/s (6 kg/fill), 3000 cars/3 min (time of fill), 1.4M cars/day, 10M cars/week).  Logarithm of the Froude number is Log ( Fr =U 2 /gD eff )=5.2.  If the similarity law is applied (conservative assumption of momentum- controlled jet) then the horizontal distance to 4% is (  N =1.267) 465 m.  If the Shevyakov’s graph (next slide) is applied in the assumption that jet is momentum-controlled where hydrogen concentration is 4% then separation distance is 418 m ( Log ( x/D )=2.63) that is close to the similarity law result. However…

16. Funded by FCH JU (Grant agreement No. 256823) 16 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 16  However, the buoyancy will direct the horizontal jet (gives the maximum separation distance) up and thus will decrease the separation distance.  Transition from momentum to buoyant jet will happen for Log Fr =5.2 at Log ( x/D )=2.04 when concentration is about 12% (not 4% for which Log ( x/D ) is 2.63 and the jet is assumed to be fully momentum). momentum buoyant downward jet Log(x/D) =2.63 Log(x/D) =2.04 Separation distance: 465 m reduces to 107 m ( more than 4 times!)

17. Funded by FCH JU (Grant agreement No. 256823) 17 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 17  Theory predicts faster concentration decay in round jet compared to infinite plane jet of the same size  Is this theoretical conclusion applicable to plane jets with finite aspect ratio (ratio of slot length to width)? The first difference between infinite and finite ratio plane jets is so-called switch-of-axis phenomenon.

18. Funded by FCH JU (Grant agreement No. 256823) 18 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 18 Flame length is proportional to nozzle diameter.

19. Funded by FCH JU (Grant agreement No. 256823) 19 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 19 Storage pressure 40 MPa, and constant nozzle area of 0.8 mm 2.

20. Funded by FCH JU (Grant agreement No. 256823) 20 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 20 Flame length reduction: 7.5 –> 1.8 m (more than 4 times)

21. Funded by FCH JU (Grant agreement No. 256823) 21 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 21 Flame length reduction: 6.1 –> 1.8 m (more than 3 times)

22. Funded by FCH JU (Grant agreement No. 256823) 22 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 22 Back view Side view Innovative PRD flame Current PRD flame Innovative PRD flame

23. Funded by FCH JU (Grant agreement No. 256823) 23 © HyFacts Project 2012/13 CONFIDENTIAL – NOT FOR PUBLIC USE 23 An example of strategy and engineering solutions for storage tank and PRD is described in a patent application (University of Ulster). Preventing fireball PRD Higher fire resistance (orders of magnitude!)