Advances in Hydrogen Safety Engineering Vladimir Molkov

Home , Hydrogen safety

13th International Hydrail Conference: HYDRAIL 2018 Rome, 6-8 May 2018 Advances in hydrogen safety engineering Vladimir Molkov

Hydrogen Safety Engineering and Research Centre (HySAFER) ulster.ac.uk Outline

. About HySAFER Centre. . Fundamentals of hydrogen safety engineering. . Tool to simulate hydrogen tank fuelling. . Blast wave of storage tank rupture in a tunnel fire. . Quantitative risk assessment (QRA): requirements for acceptable risk of onboard storage (London). . Breakthrough solution: leak-no-burst safety technology in a fire composite tanks. HySAFER Centre Hydrogen Safety Engineering and Research

. International multi-disciplinary team of dozen researchers and academics (UK, Egypt, Iran, Iraq, Ireland, Italy, Norway, Russia, Ukraine) since 2008. . External funding in hydrogen safety since 2004 totals €8M. . More than 100 man-years of accumulated knowledge in hydrogen safety. . HySAFER is one of the key providers of high quality hydrogen safety research and education/training globally. . HySAFER champions hydrogen safety research and education in the UK. . HySAFER is a co-founder of the International Association for Hydrogen Safety, and actively participates in related international activities, including but not limited to: o UN GTR#13 IWG Sub-Group Safety, o ISO TC197 Hydrogen Technologies (BSI mirror committee), o CEN/CLC/TC6 Hydrogen in Energy Systems, o IEA Hydrogen Implementation Agreement Task 37 Hydrogen Safety, o Working Groups of Cross-Cutting Research Activities and Transport of HER, o European Hydrogen Safety Panel (EHSP), o Regulations, Codes and Standards Strategy Coordination Group of FCH JU, o etc. Fundamentals of hydrogen safety engineering

(free download book at www.bookboon.com) Hydrogen in rail systems Problem parameters

. Storage system pressure 350 bar (roof mounted). . Up to 100 kg onboard storage unit. . Daily usage by a train up to 150-170 kg. . Need in 1500-2500 kg storage per day. Bulk logistics issue. . Some of safety related issues: o Leaks and their dispersion. o Leak ignition, jet fires, their thermal and pressure effects. o Explosions: deflagrations, detonations, tank rupture in fire. o Refuelling. o Confined space hazards and risks, e.g. tunnels. o Hazards and quantitative risk assessment (QRA), etc. Hydrogen safety engineering Definition and RCS

. Hydrogen safety engineering (HSE) is defined as an application of scientific and engineering principles to the protection of life, property and environment from adverse effects of incidents/accidents involving hydrogen. . Design for hydrogen safety should be treated as an engineering responsibility rather than as a matter for detailed regulatory control; designers should develop a greater understanding of hydrogen safety. . There is an overestimation of expectations from the role of RCS in safety design of hydrogen systems and infrastructure (at least three years old, often only general statements, “naturally” fragmented, etc.) Hydrogen safety engineering What do we already know: examples

. The similarity law for concentration decay in unignited under- expanded hydrogen jets (hazard distance for unignited release). . Passive and forced ventilation of hydrogen. . Criteria for spontaneous ignition of hydrogen releases into air. . Thermal and pressure effects from jet fires. . Three hazard distances for jet fires (fatality, injury, no-harm). . Revealed in 2010 the pressure peaking phenomenon. . Mitigation of hydrogen-air explosions, including limitation of inventory (warehouse) and venting of non-homogeneous deflagrations. . Criteria for prevention of deflagration-to-detonation transition. . Fire resistance rating of onboard storage for vehicles. . Parameters of blast wave and fireball size after onboard tank rupture in the open (stand-alone and under-vehicle tanks). . How to improve reproducibility of GTR#13 fire test protocol (see next slide), etc. GTR#13 fire test protocol reproducibility Fire resistance rating saturation phenomenon 20 CH4-air fire, HRR=290.8 kW/m2, Type IV tank

18 CH4-air fire, HRR=617.5 kW/m2, Type IV tank 16 C3H8 fire, HRR=652 kW/m2, Type III tank 14 C3H8 fire, HRR=1632.2 kW/m2, Type IV tank Pool fire, HRR=4270.8 kW/m2, Type IV tank 12 Pool fire, HRR=1563 kW/m2, Type IV tank 10 8

Fire resistance rating, min 6 4 0 1000 2000 3000 4000 HRR/A, kW/m2

Saturation at FRR=6 min for specific Heat Release Rate HRR/A>1000 kW/m2. Knowledge gap: FRR=3 min for use of hydrogen jet fire as a fire source. Hydrogen safety engineering Knowledge gaps and technical bottlenecks

. Dispersion and ventilation of leaks in realistic conditions of use (congestion, etc.). . Probability of ignition in realistic conditions. . Fire resistance of equipment to hydrogen impinging jets. . Consequences of large-scale tank rupture in a fire. . Innovative prevention and mitigation technologies for hydrogen releases, jet fires, chemical (combustion) and physical (burst) explosions for various applications. . Validated models of refuelling and corresponding efficient equipment, e.g. “adiabatic hoses”, etc. . Hazards and associated risk of hydrogen trains in tunnels, etc. Hydrogen tank fuelling model

with M Dadashzadeh, S Kashkarov and D Makarov Developing fuelling model Benefits and applicability range

Benefits: . Fast . Inexpensive . Tank independent Applicability range is beyond the J2601 protocol range: 0 0 . Temperature: -40 C Tgas 85 C . Pressure: 0.5 MPa P 1.25×NWP ≤gas ≤ . SOC: not to exceed 100% ≤ ≤ . Filling time: 3-5 min

Note: state of charge (SOC) = , ( , 15 ) 0 𝜌𝜌 𝑃𝑃 𝑇𝑇 ⁄𝜌𝜌 𝑁𝑁𝑁𝑁𝑁𝑁 𝐶𝐶 Fuelling SAE J2601 protocol

Fuelling protocol for light duty gaseous hydrogen surface vehicles SAE J2601: Two approaches: . Look-up table: utilising a fixed pressure ramp . Formula based: utilising a dynamic pressure ramp Currently J2601 protocol is designed for: . Delivery temperature categories: -400C, -300C, -200C . Pressure classes: 35 MPa and 70 MPa . Compressed hydrogen storage: 49.7 L to 248.6 L Future development (J2601): . Warmer fuel delivery temperatures (-100C or ambient) . Smaller compressed hydrogen storage sizes The model Current version (to be extended to pipes) Formulation Filling Equation Reference model . Form of energy conservation equation Molkov et al., 2009 Gas . Real gas EOS (Abel-Noble) Johnson, 2005 . Unsteady heat transfer equation Patankar, 1980 . Nu correlations for convective for inside heat transfer Woodfield, 2008 Tank . Constant heat transfer coefficient on external wall . Original approach based on the entrainment theory Ricou & Spalding, 1961 . Tank properties: volume; internal surface, diameter and length; external diameter; load-bearing wall and liner thicknesses and their material thermal properties (thermal conductivity, specific heat capacity, density); external heat transfer coefficient; nozzle diameter; initial temperature Input . Hydrogen properties: co-volume constant; specific heat capacity; thermal conductivity; specific gas constant; dynamic viscosity; initial pressure and temperature; pressure ramp . Other inputs: ambient temperature; air viscosity; fuelling time . Gas temperature inside the tank Output . Temperature profile within the tank wall and liner . Gas density or SOC Validation Type IV tank, 29 L

Test (Miguel et al., 2016): . Initial pressure 2 MPa; target pressure 77 MPa . 3 mm orifice Tank properties (Acosta et al., 2014): . Volume 29 L (external length 827 mm; external diameter 279 mm; internal diameter 230 mm) . CFRP: thermal conductivity 0.74 W m-1 K-1; specific heat capacity 1120 J kg-1 K-1; density 1494 kg m-3 . HDPE liner: thermal conductivity 0.385 W m-1 K-1; specific heat capacity 1580 J kg-1 K-1; density 945 kg m-3 Validation results Type IV tank, 29 L

120 Measured gas temperature [Miguel et al., 2016 experiment] 105 Simulated gas temperature [Ulster model]

90 C) 0 75

45 Temperature ( Temperature

0 0 20 40 60 80 100 120 140 160 180 200 220 240 Time (s)

Maximum experimental temperature difference in the tank is 30C (Cebolla et al., 2014). The model gives maximum deviation 50C. Validation Type III tank, 74 L

Test (Zheng et al., 2013): . Initial pressure 5.5 MPa; target pressure 70 MPa . 5 mm orifice Tank material properties (Zheng et al., 2013): . Volume 74 L (external length 1030 mm; external diameter 427 mm; internal diameter 354 mm) . CFRP: thermal conductivity 0.612 W m-1 K-1; specific heat capacity 840 J kg-1 K-1; density 1570 kg m-3 . Aluminium liner: thermal conductivity 238 W m-1 K-1; specific heat capacity 902 J kg-1 K-1; density 2700 kg m-3 Validation result Type III tank, 74 L

120 Measured gas temperature [Zheng et al., 2013 experiment] 105 Simulated gas temperature [Ulster model]

90 C) 0 75

45 Temperature ( Temperature

0 0 80 160 240 320 400 480 560 640 Time (s)

Maximum experimental temperature difference in the tank is 50C (Zheng et al., 2013). The model gives the same maximum deviation 50C. Model application: Type IV tank, 50 L 0 Ambient temperature Tamb = 50 C No pipe with heat losses in simulations (assuming double wall vacuumed hose, i.e. “adiabatic hose”). 90 87.5 Inside gas temperature Inside gas pressure 85

80 T 85 0C 66.1

70 44.8

65 Pressure (MPa) Temperature (C)Temperature

60 23.4

55 End of End of fuelling 50 2.0 0 20 40 60 80 100 120 140 160 180 Time (s) End of fuelling is at SOC 100%.

J2601: fuelling at Tamb=50 C with Tdel=-40 C from 2 to 77.8 MPa takes with ramp 3.2 MPa/min (77.8-2)/3.2=24 min?! Fuelling with “adiabatic” hose gives only 2 min 45 s! Model application: Type IV tank, 50 L 0 0 Effect of pressure ramp: Tamb 50 C; Tdel -10 C

85 T 85 0C 80

70 87.50 65 Temperature (C)Temperature 44.75 60

55 2.00 Pressure (MPa)

0 90 180 End of fuelling 50 0 20 40 60 80 100 120 140 160 180 Time (s)

End of fuelling is at SOC 100% Concluding remarks (fuelling)

. The model for simulating hydrogen tank fuelling is validated against fuelling tests with Type III and Type IV tanks. . The model predictions are “instantaneous” (could be used for fuelling control systems). . Deviation from test in temperature prediction in a tank is within ± 5 (within experimentally observed temperature scatter in a tank).0 𝐶𝐶 . The model could be used to design efficient fuelling protocols. . The “adiabatic hose” fuelling efficiency and different thermal isolation methods should be tested experimentally. Blast wave of hydrogen storage tank rupture in a tunnel fire

with V Shentsov and D Makarov Blast wave study outline

Problem formulation . CFD model details . Tunnel cross-section . Calculation domain and grid . Initial and boundary conditions Physical tank explosion (rupture without combustion) . Blast wave overpressure and propagation velocity . Blast wave dynamics Tank rupture in a fire (with combustion) . Max overpressure: unignited vs ignited scenario . Hazard distances Concluding remarks Problem formulation CFD model details

. The Reynolds averaged Navier-Stokes (RANS) CFD model. . The realizable k-ε turbulence model. . The finite rate chemistry (FRC) combustion model. . The 37-step chemical reaction mechanism of hydrogen combustion in air with 13 species. . The in-situ adaptive tabulation (ISAT) algorithm accelerating chemistry calculations by 2-3 orders of magnitude (Pope) together with stiff chemistry solver. . The discrete ordinates (DO) radiation model. . Real gas Redlich-Kwong EoS. Problem formulation Tunnel cross-section (9 m width)

Reference: Maidl, 2014 Problem formulation Calculation domain and grid

Polyhedral grid outside the tunnel

Hexahedral block- structured grid inside the tunnel

Total CVs: 396,272 External: 79,712 Tunnel: 30,760 Hexahedral block- Tank: 14,800 structured grid inside the tunnel Problem formulation Initial and boundary conditions

Initially quiescent atmosphere (u=v=w=0 m/s). High-pressure hydrogen storage tank (floor level): . Hemisphere 140 L, H2=5.5 kg, p=700 bar Rest of the domain (atmosphere):

. Mass fraction: 23% O2 and 77% N2 . T=300 K and p=101325 Pa Opaque walls (non-slip boundary conditions for velocity). Non-reflecting calculation domain boundaries have ambient properties. Unignited tank rupture in the tunnel Blast wave overpressure and velocity 1000 Average velocity of the leading shock is 400 m/s 900 50 m 800 45 m a 700 40 m k P

, 35 m e 600 r u

s 30 m s 500 e r 25 m p

r 400 e v 300 20 m O 200 15 m 100 10 m 0 5 m 0 20 40 60 80 100 120 140 160 180 200 Time, ms Unignited tank rupture in the tunnel Blast wave propagation dynamics

Overpressure, kPa Ignited rupture in the tunnel Blast wave and fireball dynamics Max overpressure in the tunnel Unignited vs ignited tank rupture scenario Unignited and ignited tank rupture Hazard distances (5 m step after first 5 m) 1010006 Lung damage 50% probability threshold 3 m eardrum rupture ) 3 m

a 4 m

P 5 "Fatality" 5 m k 11000 (

P 50 m

- "Injury" 50 m ,

e 4 r

u 1010

s Temporary shift s

e threshold - "No harm" r

r p 3 Skull fracture threshold e 101 v O Without combustion, 0.25 m Lethality from body With combustion, 0.25 m translation threshold 100.12 10-1 100 101 102 103 104 105 106 Impulse, I (Pa . s) Concluding remarks (blast in a tunnel)

. Overpressure is higher at the bottom (in the beginning) and then stabilises at constant value throughout the whole height of the tunnel: . 25 kPa at 42 m the case without combustion . 37 kPa at 27 m for the case with combustion . Combustion increased the blast wave overpressure by about 50% when stabilised. . Even for unignited case the overpressure and impulse up to 10 m exceeds the “fatality” limit and within the rest of the tunnel length it is above the “injury” limit. . Research is ongoing to find out the maximum onboard inventory to exclude injuries and fatalities. Quantitative risk assessment (QRA): requirements for acceptable risk of onboard storage (London roads)

with M Dadashzadeh, S Kashkarov, D Makarov QRA methodology Benefits

. Considers previously “missed” hazards of blast wave and fireball following the tank rupture in a fire. . Uses innovative validated engineering tools to calculate hazard distances. . Merges contemporary probabilistic and deterministic methods. . Assesses two risks: o Fatality per vehicle per year, and o Cost per an accident with vehicle fire QRA methodology Statistics, scenario, assumptions Fire o About 15,500 accidental car fires annually in GB only (Fire Fire due to the car or Fire due to HP or Fire while filling statistics, Great Britain, 2010- accident fittings, connections hydrogen/tow away 2011)

TPRD failure o TPRDs failure rate is not zero: localised fire, blockage of TPRD during accident, etc. (Saw et al., 2016) Failure of emergency Failure of TPRD 50 CNG tank rupture (1976-2010): 35% due to TPRD and operation o failure (Lowell, 2013)

Catastrophic tank rupture & hazards o Carbon-fibre reinforced polymer (CFRP) Type 3 and 4 tanks fire resistance rating before catastrophic rupture is 6-12 min (Weyandt, 2005; 2006)

Blast New accounted hazards: fireball and a blast wave (Dadashzadeh et al., and Fireball o wave ICHS 2017, IJHE 2018)

Open literature data for: o Onboard storage - type 4 tank, 62.4 L, 70 MPa, localised fire (Toyota, 2016; Yamashita et al, 2015) o Statistical data on car accidents, vehicle fires and population density in the UK (Statistica, 2016; GLA, 2015) o Hazards distances for fatality due to blast wave and fireball (Molkov et al., 2015; Kashkarov et al., 2016; Hord, 1978) o Failure probability of safety barriers (Saw et al., 2016; Landucci et al., 2015) o Cost of fatality due to the catastrophic rupture (HSE, 2015) -5 o Risk acceptance criteria - 10 fatality/year (LaChance & Houf, 2009; LaChance et al., 2011; Haugom et al., 2003) QRA methodology published (IJHE)

First QRA of onboard storage for hydrogen-powered cars. QRA methodology Risk assessment frameworks: fatality and cost

Consequence analysis Blast wave and fireball Harm criteria and hazard distances Blast wave Fireball Area used Fatality per Hazard Hazard Effect Hazard Hazard in QRA, accident distance, diameter, area, m2 area, m2 m2 ( ) m m 𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇 𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂𝒂 Death 1.68 9 35 962 962 4.56 Serious injury* 13.4 555 - - - - Slight injury* 76 17,573 - - - -

. Hazard distances are from: Molkov, Kashkarov, IJHE, 2015:40, 12581-12603. . Serious Injury and Slight injury hazard distances are for information only. . Fireball diameter calculation is based on: Experimental data (Weyandt, 2005 and 2006) o 𝐷𝐷𝑓𝑓 . Thermal effects on humans are lethal within fireball and negligible (thermal doze) outside. . London area: mean population density = 0.008 . Ref: GLA, 2015. . Considering 1.5 persons are in the vehicle. Ref: Assael & Kakosimos2 , 2010. 𝑁𝑁0 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝⁄𝑚𝑚 . = + 1.5 = 0.008 962 + 1.5 = 9.19

𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑝𝑝𝑝𝑝𝑝𝑝 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑁𝑁0 � 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 � 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓⁄𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 Frequency analysis Frequencies of three initiating fire events

Fire frequencies due to three different reasons Value I. Fires due to a car accident 1 Number of car accidents (www.statista.com, 2016) 2.67 10 2 Number of cars (www.statista.com, 2016) 3.11 105 � 3 Frequency of car accidents (accident/year): “1”/”2” 8.57 10 7 � Probability of accident leading to fire −3 4 4.54 � 10 (Rodionov, Wikening, Moretto, IJHE, 2011) −3 5 Frequency of initiating fire (fire accident/vehicle/year): “3”x”4” 3.89 � 10 −5 II. Fires due to leaks of HP fittings, valves and piping connections� Frequency of initiating fire (fire/vehicle/year) 6 1.00 10 (Saw et al., in: Hazards 26, Edinburgh, 2016) −3 III. Fires due to hydrogen filling/tow away � Frequency of initiating fire, (fire/vehicle/year) 7 1.00 10 (Saw et al., in: Hazards 26, Edinburgh, 2016) −6 � Frequency analysis Failure probabilities

. PRD failure probability (Reliability Analysis Center, 1991) -3 o Engulfing fire 6.04⋅10 -1 o Localised fire 5.03⋅10 (2 orders larger failure probability!) o TPRD failure probability is equal to that of “ordinary” PRD (FCH JU FireComp project, Saw et al., Hazards 26, 2016) . Emergency actions failure probability ( ) (Landucci et al., 2015) 𝑬𝑬𝑬𝑬 o Probit function: = . . o = + 𝒀𝒀( 𝟗𝟗 𝟐𝟐𝟐𝟐 − 𝟏𝟏 𝟖𝟖𝟖𝟖 � 𝐥𝐥𝐥𝐥 𝐅𝐅𝐅𝐅𝐅𝐅 𝟏𝟏 𝐘𝐘−𝟓𝟓 . Example𝑬𝑬𝑬𝑬 for𝟐𝟐 the𝟏𝟏 fire𝐞𝐞𝐞𝐞 𝐟𝐟 resistance𝟐𝟐 � rating (FRR) of 8 min: . = 8 = 5.04030 = 6.57 10 −1 𝐹𝐹𝐹𝐹𝐹𝐹 ⇒ 𝑌𝑌 ⇒ 𝐸𝐸𝐸𝐸 � Frequency analysis Tank rupture frequency Catastrophic tank rupture frequency scenario assumptions: . Localised fire . Fire escalation probability for FRR=8 min = ( ) 3 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 ∑𝑖𝑖=1 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑖𝑖 � 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 � 𝐸𝐸𝐸𝐸 Effect of fire resistance rating (FRR) Fatality rate =

3.50E-03 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅Acceptable𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 Level𝑅𝑅 of𝑅𝑅 Risk𝑅𝑅𝑅𝑅𝑅𝑅 [LaChance𝑅𝑅𝑅𝑅 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 & Houf, 2009;� 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 LaChance et𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 al., 2011; Haugom et al., 2003] Risk (fatality/vehicle/year) for localised fire, current study 3.00E-03

2.50E-03 6.00E-05

2.00E-03 4.50E-05

3.00E-05 1.50E-03 1.50E-05

1.00E-03 0.00E+00 Risk (fatality/vehicle/year) Risk 33 38 43 48 53 58 63 67 77 87 5.00E-04

0.00E+00 8 13 18 23 28 30 33 38 43 48 53 58 63 67 77 87 Fire Resistance Rating (min)

For mean population density 0.008 acceptable FRR level is 47 min!

𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝟐𝟐𝒑𝒑𝒑𝒑 𝒎𝒎 Effect of fire resistance rating (FRR) Accident cost

= 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑜𝑜𝑜𝑜 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 � 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 Effect Value (£/person) � 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒Fatality𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 1,336,800 Serious injury* 207,200 Slight injury* 300 * - for information only, not used in this QRA

Source: Health and Safety Executive, Cost Benefit Analysis, 2015, http://www.hse.gov.uk/risk/theory/alarpcheck.htm Effect of fire resistance rating (FRR) Accident cost

£4,500,000

£4,000,000

£80,000 £3,500,000 £64,000 £3,000,000 £48,000 £2,500,000 £32,000 £2,000,000 £16,000 Cost (£/accident) Cost £1,500,000 £0 £1,000,000 33 38 43 48 53 58 63 67 77 87

£500,000

£0 8 13 18 23 28 30 33 38 43 48 53 58 63 67 77 87 Fire Resistance Rating (min)

. For FRR=8 min cost of fire accident is £4.00M! . Drastic reduction of cost of fire accident by increasing tank’s FRR Concluding remarks

. The QRA methodology is developed and applied to onboard hydrogen storage. For the first time the risk is shown to be a function of tank fire resistance rating (FRR). . The previous “missed” hazards of blast wave and fireball during the tank rupture in a fire are considered. . Current Type 4 tanks with FRR=6-12 min have unacceptable risk of human life loss 3.14⋅10-3, and the cost associated with a fire accident of £4.00M. . Increasing FRR of the tank reduces the risk value significantly to acceptable level of 10-5 and negligible cost. Leak-no-burst safety technology for composite pressure vessels (explosion-free in a fire tanks)

with S Kashkarov and D Makarov

International (PCT) Application No PCT/EP2018/053384 "Composite Vessel for Hydrogen Storage" Ulster University Supporting projects

Conceived in: EPSRC SUPERGEN Hydrogen and Fuel Cells Hub, EP/J016454/1, 2012-2017, £218,677. Tested (intumescent paint part 1 of 3) in: EPSRC SUPERGEN Hydrogen and Fuel Cells Challenge: Safety Strategies for Onboard Hydrogen Storage Systems, EP/K021109/1, 2013-2017, £1.2M Current manufacturing and testing projects (hybrid tank part 2 of 3): . Invest NI PoC “Composite tank prototype for onboard compressed hydrogen storage based on novel Ulster’s leak-no-burst safety technology”, £106k . Interreg Atlantic Area ERDF HYLANTIC “Atlantic network for renewable generation and supply of hydrogen to promote high energy efficiency EAPA_204/2016”, €2.5M (€250k). Proposals under review: . Horizon 2020 “PNR for safety of hydrogen driven vehicles and transport through tunnels and similar confined spaces” (HyTunnel-CS), €2.5M . H2020-MSCA-ITN-2018. 2019-2022. SAFIT: Safety of Alternative Fuels Infrastructure Training Network, 10 partners, €4.18M (UU coordinator – €909.5k). . EPSRC Future Composites Manufacturing Research Hub: 2 year fellowship. Design, manufacturing, testing

Samples manufacture

Materials Thermal selection properties Prototype design (UU)

Prototype Prototype testing manufacture Composite material measurements Thermal conductivity

Hoop and helical fibre plies Cross-plies conductivity (lower)

Epoxy matrix

Along-plies conductivity (HIGHER) Composite material measurements Why samples should be flat

Cross-plies Cross-plies conduction conduction

Fibres aligned horizontally in a In flat sample machined from initially curved- flat samples will give right wound sample fibres at a certain angle will through-plies conductivity contribute as non-horizontal component General prototyping targets

. Cheaper (than current proof of concept design)

. Equal or smaller outside volume for the same storage capacity (or equal to ordinary tanks)

. Lighter Global partnership Ulster communicates with 43 composite pressure vessel and material suppliers from 15 countries. Global partnership Ulster communicates with 10 testing laboratories from 6 countries. Concluding remarks (LNB tanks)

. The LNB safety technology tanks are being manufactured and tested in different countries: . No long flames from TPRD . No blast wave (solving tunnel safety issue) . No fireball (largest hazard distance . No pressure peaking phenomenon in confined space . The Ulster IP is planned for licensing from 2019. . The requirement for onboard storage hydrogen tanks to be “explosion-free in a fire” will be suggested for inclusion to UN Global Technical Regulation for Hydrogen and Fuel Cell Vehicles #13 to protect life, property, and the built environment.