Presentation of the French national project DRIVE:Experimental Data for the Evaluation of hydrogen Risks

onboard vehicles, the Validation of numerical toolsand the Edition of guidelines

Lionel PERRETTEINERIS / France

Henri PAILLERECEA (LTMF) / France

Guillaume JONCQUETPSA, Centre Technique de Carrières-sous-Poissy / France

ABSTRACT

The everyday use of hydrogen in the transport sectorrequires high safety standards. Safety requirementsmust be addressed as a key issue for fuel cell cardevelopment. Therefore, it becomes crucial to haveexperimental data on hand in order to provide realisticand reliable risk assessment and to be able to reallyknow the extent of safety margins taken. In such acontext, the National Institute of Industrial Environmentand Risks (INERIS) along with the French AtomicEnergy Commission (CEA), the French automotivemanufacturer PSA PEUGEOT CITROËN and theresearch Institute on unstable phenomena (IRPHE)recently started a research programme entitled DRIVE.This programme aims at providing experimental andnumerical results for the safe design of hydrogenvehicles. Fields of investigation cover the whole range ofphenomena that can be encountered in hydrogenaccidents, from leakage to dispersion, ignition and finallycombustion. The work programme and early results arepresented in this article.

INTRODUCTION

The everyday use of hydrogen in the transport sectorrequires high safety standards. Safety requirementsmust be addressed as a key issue for fuel cell cardevelopment. It is therefore a necessity to fullyunderstand and investigate explosion and fire hazardsassociated with the use of hydrogen and to come upwith practical and realistic means to control thesehazards. Although the experience and safety record of

hydrogen as an industrial gas is significant, it turns outnot to be applicable to the future decentralised use ofmoderate quantities of hydrogen. Data specific to thatuse are becoming crucial in order to provide realistic riskassessment and to be able to specify expected safetyperformances.

PROGRAMME OBJECTIVES

This project which gathers together industrial expertiseof the car manufacturer PSA PEUGEOT CITROËN aswell as technical and scientific skills of CEA, IRPHE andthe co-ordinator INERIS aims at providing experimentaland numerical results to better assess hazards whenhandling hydrogen onboard vehicles. Results willcontribute to strengthen risk assessment studies interms of hazard quantification as well as to proposeinnovative solutions for the handling of hydrogen in carscompatible with public use.

The work programme is split into 4 items: the automotivesystem, hydrogen dispersion, hydrogen ignition andcombustion and finally results dissemination. Thesecond and the third items focus on the hazardousphenomena one would consecutively encounter whendealing with hydrogen accidents. The accidental chain isillustrated by the figure below.

Accidental

leakage •

H2 build up andexpansion of

explosiveatmosphere onboard

nr outside vehicle

•

•

Hydrogen jet

Explosion

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4Author manuscript, published in "Society of automative engineer world Congress 2007, Detroit : United States (2007)"

The list below indicates some of the safety challengesthat face car manufacturers:

• Safe pressurised storage design (thermal insulation,use and location of pressure relief devices (PRD),enclosed or open compartment for storage...) andlocation of the pressurised storage in order toprevent burst due to thermal and mechanicalaggressions as well as to control consequences ofaccidental leakage or PRD releases.

• Adequate control of standing flames fed by minorundetected leaks,

• Appropriate equipment design (choice of materialand technology) and location (segregation,,enclosure,...) in order to control ignition sourceswhether mechanical, electrical or electrostatic innormal operations,

• Safe handling of hydrogen purged outside vehicle,• Effective control of explosive atmospheres in

confined or semi-confined spaces on-board vehicles(engine compartment) or outside the vehicle(underbody, garages...). That also includes ensuringthat hydrogen will never penetrates into thepassenger compartment.

Aside from the items above, one of the great challengesconsists in setting an appropriate limit to segregatetolerable accidental releases versus unacceptable ones.Such a limit would contribute to set the expectedperformance of the leak detection and leak interruptionsystems whether based on atmospheric detection or ondirect measurement such as flow or pressure drop.Ideally, a vehicle should be able to run without leakdetection systems having in mind that reliability of suchdetection system is so far moderate and thatmaintenance is rather constraining. Chronic leakage isalso an issue since the no-leak objective does not seemto be realistic. Therefore, upper tolerable limit shouldalso be specified for chronic leakage along with designmeasures to ensure that hydrogen can neveraccumulate to form an explosive mixture.

Finally, CFD tools currently and intensively used by carmanufacturers to design vehicles (aerodynamic studiesfor instance) should also be able to provide reliableresults when computing normal or accidental releasescenarios for hydrogen vehicles.

Safe design practices should finally be shared amongmanufacturers to promote the high safety standardexpected for the public use of hydrogen.

DETAILED PRESENTATION OF THE WORKPROGRAMME AND EARLY RESULTS

WORK PROGRAMME

This project started early 2006 in the framework of thefirst call for proposals of the National Research Agency(ANR). It will last for three years. Its work programme ismade of 4 main modules:

• vehicle safety,• hydrogen release and dispersion,• hydrogen ignition and combustion (explosion or jet

flame),• and finally results compilation and dissemination.

The vehicle safety module includes a thorough riskassessment in order to identify hazardous situations(causes, consequences). Outputs from this module willshape the content of the experimental and modellingwork in terms of hazardous situations to investigate withthe objective of obtaining better quantification andtherefore better risk assessment and risk control. Thevehicle safety module that concentrates on hydrogenequipment and potential ignition sources onboardvehicles goes down to component level.

Within the hydrogen release and dispersion module, wewill then experimentally look at hydrogen componentleakage in order to quantify them. Both chronic andaccidental hydrogen leakage, as well as hydrogenemissions (purges, pressure relief systems) will beconsidered. These values will then serve as referencesto build the test matrix for hydrogen releases (free andimpinged) on board the vehicle as well as in confinedspaces.

These release experiments will be carried out withhelium as well as with hydrogen. The size and locationof explosive atmospheres will be monitored for variousleaking rate and location.

Within the third module, we will concentrate on hydrogenignition with vehicle components as well on hydrogencombustion (flame or explosion). Ignition potentiality ofcar components (fans, electronic cards, electric engines,braking systems...) will be investigated through practicaltesting. In parallel, explosion experiments will beundertaken in order to assess the destructive potential ofexplosive atmospheres (in accordance with hydrogendispersion findings) in more or less obstructedenvironments. In a later stage, combustion tests will takeplace on board vehicles. Hydrogen flames will also beconsidered at two different scales; on one hand, effect oflarge hydrogen flame resulting for instance from a PRDrelease and on the other hand, standing flames fed byminor leaks. For the latter point we will look at the easewith which ignition can occur as well as flamesustainability depending on upstream pressure andleaking orifice size. Excessive fire hazards due to theuse of hydrogen can possibly be more critical thanexplosive atmospheres themselves that concentratemost of today's concerns. In the frame of the lastmodule, experimental learning will be turned intopractical graphs to facilitate risk quantification.Recommendations on best practices for vehicle designwill also be drafted. Topics such as maximum allowableexplosive volume, control of ignition sources andlocation of hydrogen components... will be addressed infinal public reports. Results will be discussed duringdedicated seminars.

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

This project will not investigate the safety of high-pressure hydrogen storage in terms of potentiality forcatastrophic failure when exposed to mechanicalaggression or external fire. Issues such as fuse locationon tanks or thermal insulation will not be looked at in theframe of this project. However, potential leaks as well asPRD releases consequences (explosive atmosphereunder vehicle, jet fire) will be experimentally considered.Leakage during refuelling will not be considered either.

More detailed content and progress of some of the workthemes is given below.

WORK TOPICS

Objectives, methodology, tools, basic backgroundinformation and early results will be presented for eachof the topics below:

1. vehicle risk assessment,2. leak quantification,3. hydrogen dispersion on board and outside the

vehicle,4. hydrogen ignition by vehicle components,5. hydrogen explosion,6. & hydrogen flames.

VEHICLE RISK ASSESSMENT

OBJECTIVE

Risk assessment helps to evaluate and demonstratethrough a systematic and consistent approach that ahydrogen vehicle whether of a prototypic or a seriesproduction type is as safe as a conventional one. Withinthis project this early work will help us to selecthazardous situations to be studied and therefore toshape the experimental programme.

STEPS

Usual risk assessment steps are the following:

• identify unwanted events and associated causesand consequences,

• e va I u ate th e i r criti city,• and take appropriate actions to bring risk down to an

acceptable level.

TOOLS

Four tools are mainly used to contribute to the riskassessment studies: functional analysis preliminary riskanalysis, fault trees and then Failure Mode and EffectAnalysis (FMEA).

First of all, a functional analysis was undertaken. It helpsto identify system functions and associated internal andexternal constrains. Then a preliminary risk assessmentwas carried out. This simple methodology helps to

highlight the most unwanted events and to initiatediscussions on severity and severity levels. For eachvehicle mode (running, stand by, maintenance...) thefollowing unwanted events are investigated:

• shift in internal system functions and systemconstrains,

• impact from the system on its environment,• and finally, impact from the environment on the

studied system.

All identified unwanted events were then rankedaccording to severity criteria. The severity scale consistsof four levels. The highest ones 4 and 3 are those furtherconsidered in the study. Level 4 is associated with deathor injury on people and level 3 means that the vehiclecan no longer be used because of a break down. Level 3and 4 unwanted events were then further investigatedusing the fault tree methodology. This graphicalmethodology helps to visualise steps and failurecombinations that lead to the top unwanted event.

Finally, a failure mode and effect analysis (FMEA) wascarried out. This systematic and inductive methodologyhelps:

• to assess consequences of component failures (lossor alteration of associated function) on the vehicleand its user,

• to identify component failure modes that can impactsafety,

• and finally to propose corrective safety actions in thecourse of the design process.

A typical FMEA table similar to the one used in theproject is pictured below:

oQ.

0)

<ô a:

CD >

E

Hi

|

P

.E ô t

E =

E 5SSICO CD

B e l.2 .2 ;>,[ D O E .E Q..SE x

ai

<D Ë S «•Q o) 2 2

2 cIS °O) (J

is <

0) >

T3 D-

IICO CD

Ia

Table 1: FMEA table used in the risk assessmentprocess

EARLY RESULTS

Identified unwanted events when using hydrogenonboard vehicle include:

• fire onboard vehicles,

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

• explosive atmospheres and ignition whereveronboard or outside the vehicle,

• the burst of the pressurised tank,• high pressure jets.

All the selected unwanted events are linked either withthe use of hydrogen or with the handling of pressurisedgas. Unwanted events associated with high voltage orother non specific hazards are not considered in thisprogramme.

LEAK QUANTIFICATION

OBJECTIVE

The objective of this task consists in experimentallymeasuring chronic and accidental leak rates (externalaggression, human error...) associated with keycomponents of the hydrogen feed line and the hydrogenconversion system. Accidental leaking rates to beevaluated include likely situation such as bad fitting(insufficient clamping), erroneous fitting, missing ordamaged sealant. Ageing will be looked at throughrepeated dismantling of hydrogen connections. Leakcaused by organic failures of components (qualityissues, internal ageing) quantification will be forwardedto component manufacturers. Design related hydrogenemissions (PEMFC purge, PRV...) will also be taken intoconsideration. Investigated pressures will cover the fullrange from the 700 bar working pressure for the high-pressure line down to few bars for the hydrogenconversion system. Both hydrogen and helium gas willbe used in order to assess the appropriateness to usehelium for leak quantification.

These realistic and practical figures gained from thesetests will serve as input parameters for further hydrogendispersion and combustion experiments. These valueswill also be of a great help for risk assessment studies.

STEPS

The study includes four main steps. The first oneconsists in extracting and referencing leak scenariosfrom the risk assessment study. The next one willconsist in selecting components and test conditions(pressure, damage free / damaged equipment).Selection criteria include for instance likelihood andseverity of leak, lack of data from manufacturers or gascomponent not initially designed for hydrogen use. Then,tests will be carried out with the experimental set upshown further down. Finally results will be recorded andanalysed.

BACKGROUND INFORMATION

(Cfcp1 [2]) with an average value around 10-20 g/min. Itis interesting to notice that these values vary in a largedomain. Besides, except for the BMW tests, there is notany obvious link between investigated cases andspecific component leakage.

Regarding component leakage the Cfcp documentindicates that the fill up point and the pressure reliefpoint are considered as the only way for the hydrogentank to leak out. As far as other components areconcerned, the pressure regulator, piping and fittingswere considered as components able to leak outwhatever the operating pressure (high-pressure andlow-pressure). The value of 20 CFM (12.77g/min),referenced in this document, is based on the assumptionthat that the onboard computer is capable of shuttingdown hydrogen flow upon receiving a signal detecting aflow larger than 20 CFM when the engine is off. TheJapanese regulation for hydrogen vehicle2 proposes aleak threshold of 11,8 g/min (or 131 L/min) that shouldnot be reached in case of leakage after vehicle collision.This leakage is assumed to have the same calorificvalue as today's admissible value for gasoline. However,associated explosive atmosphere and consequenceshave not been taken into consideration. The samedocument indicates that helium can be used instead ofhydrogen for leak tests as long as a 1.33 multiplicationfactor of the leak volumetric flow is used.

Regarding purge management, the Japanese regulationindicates also that no purged gas in excess of 4% ofhydrogen concentration shall be discharged or leaked tothe atmosphere.

As far as the GRPE Informal Group on Hydrogen/FuelCell Vehicles work is concerned [3], no maximumallowable leakage value or leakage value to be detectedis proposed.

The work carried out in the framework of the DRIVEproject will propose some realistic data specific tohydrogen components and hydrogen fittings. These datawill enrich somehow empirical data proposed so far.

TOOLS

Components will be connected to pressure sourcesaccording to the figure below. Components will beplaced inside a closed 2 m3 insulated sphere insidewhere both pressure and temperature are monitored.Component leak rate measurement is directly linked topressure variation inside the sphere.

Literature shows that various vehicles related leak rateshave been investigated in the frame of dispersionexperiments. As shown in table 2, these investigatedrates vary from 0.5 g/min (BMW [1]) up to 100 g/min

1 California Fuel Cell Partnership2 Attachment 100 - Technical standard for fuel systemsof motor vehicles fueled by compressed hydrogen gas

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

Temperature Openinq 2

Opening 1

Hydrogencompressor(700 bar)

Pressuretransducer

Pressure transducerinside the sphere

Purae line

Temperature

Figure 1: Experimental set up for leak quantification

EARLY RESULTS

At least, the following components will be tested:hydrogen connections, pressure regulator, phaseseparator, hydrogen pump, solenoid valve &components made of polymeric or plastic type materials.

HYDROGEN DISPERSION ON BOARD ANDOUTSIDE VEHICLE (CONFINED SPACES)

OBJECTIVES

The objectives of this study are to investigate,experimentally and numerically, different types of leaksand the dispersion of hydrogen either inside the vehicleor outside the vehicle in the most critical situation, that iswhen the vehicle is parked in an enclosed space suchas a private garage. The mechanisms for the build-up ofa potentially explosive mixture of hydrogen and air willbe studied in detail, and effects of confinement andventilation will be assessed.

STEPS

The experimental programme will consist of leak anddispersion tests in a full-scale garage as well as on-board a vehicle. Both helium and hydrogen will be used.The test matrix will be elaborated from the leakquantification tests, the preliminary risk assessmentstudy, and some pretest calculations using CFD(Computational Fluid Dynamics) codes will be performed

to optimise the instrumentation set-up (location, numberof sensors). After the tests have been performed, theCFD tools will be validated on a series of experiments,including some involving passive or active ventilation.Finally, the CFD tools will be used to perform parametricstudies and to identify worse case scenarios and howthey can be mitigated (for example by design of passiveor active ventilation).

BACKGROUND INFORMATION

The table below summarises past hydrogen leakhypotheses and dispersion experiments in confinedspaces.

Datasource

Swaingarage -1998 [4] &

[5]

Swainhallway -1999 [6}

Californiafuel cell

Partnership - July2004 [2]

JARI [7]

BMWtests [1]

MotorVehicle

FireResearchInstituteUSA [8]

Experimentalconfiguration and

objectives

Assess how shouldexisting garages bemodified to make

them suitable for H2vehicles

CFD validation

The objective was tonumerically

evaluateconsequences

(explosiveatmosphere volume)of hydrogen leakage

in confinedenvironment such as

garages.Two undercarriage

configurationsconsidered

Examine thediffusion behaviourof hydrogen in an

enclosed space withvarious releasespeed Hydrogen

was releasedcontinuously from

the center of the boxbottom surface

BMW wanted toexperimentally

demonstrate thatmaximum leakage in

a confined room(garage) would notinduce hazardous

explosiveatmosphere withminimum natural

ventilation.

Assessconsequences of

hydrogen released ifignited in the vehicleunderbody as wellas in the engine

compartment

Leak rateproposed or

investigated ing/min

10.2 g/min

4.8 g/min

Residentialgarages : 5, 15,20 CFM leak(respectively12.77, 38.30 and51.07 g/mn) &Maintenance andservice facilities :continuous 24and 36 CFM leak(respectively61.29 and 91.93g/mn)

0,84 g/min or 10L/min

1 g/min0.5 g/min

48 g/min & 24g/min through a1.6 mm orifice

nozzle

Remarks

Low exit velocity (0.1 m/s)

Low exit velocity (0.1 m/s)

The assumption is that theon-board computer is

capable of shutting downhydrogen flow upon

receiving a signal detectinga flow larger than 20 cubicfeet per minute (CFM) flow,'leak', when the vehicle isdormant. This leak rate

corresponds to a fuel cellpower output of about 50

kW

It was found that withchange in release speed

(0.1, 0.2 and 3.4 m/s(corresponding exit

diameters are 46, 33 and 8mm respectively) there

was a moderate variationin the horizontal hydrogendiffusion along the bottom

surface and in the lift of the4%+ combustible mixture

1 g/min is considered to bethe maximum escape

volume of hydrogen in theevent of a fault.

0,5 g/s concerns lesssevere cases. Hydrogen is

stored under cryogenicform. These values are

based on hydrogen boil off.

No indication given on whythese data have been

chosen.

Table 2: Hydrogen leak cases in confined spacesextracted from the literature

Compared with existing experiments and data, DRIVEwill investigate wider leak rates and release conditionsspectrum with more elaborate measurement techniques,in terms of gas concentration and velocity andturbulence measurements. The new tests will be the

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

most detailed and finely instrumented tests for the studyof hydrogen leaks. The objective is clearly to producedata both for risk assessment purpose, and to validateCFD tools, which are now well recognised to bepotentially the most accurate and reliable computationaltool for assessing consequences of accident scenarios.

TOOLS

The development of a full-scale garage and a quasi non-intrusive (due to their small size) sensor network(catharometric measurement type) is currently inprogress. Mass flow controller of various ranges will beused in order to control leak flow. Hydrogen or heliumconcentration will be measured with the help of thecatharometric sensors grid. Non intrusive velocimetrymeasurements will be undertaken through particle imagevelocimetry (PIV) and Laser Doppler Velocimetry (LDV).Transparent walls will be mounted on the garage toallow measurements. Ventilation will be managed withthe help of either an extractor or a fan. The dimensionsof the garage are those of a typical European garage fora single vehicle, and it is fitted with an overhead pivotingdoor. The garage will be fitted with vents which canprovide either passive or active ventilation or extraction.

Computational Fluid Dynamics tools will be used for pre-test calculations, as well as to analyse the tests once theCFD codes have been completely validated. One of theCFD outputs that will be of most use for safetyassessment is the time-evolution of the volume ofexplosive atmosphere (for example the volume of gasmixtures above 4% of H2). The effect of leak location,direction, impingement on nearby obstacles as well aspresence or absence of ventilation on the evolution ofthis cloud will be investigated in detail.

Length (m)

Width (m)

High(m)

Volume (m3)

Externaldimensions

5.84

3.04

2.5

44.88

Internaldimensions

5.76

2.69

2.42

41.26

Figure 2: Garage skeleton (top) and garage dimensions(bottom)

EARLY RESULTS

The first tests are scheduled in the second semester of2006, and will first address simple release scenarioswith no obstacles and then release scenarios with avehicle mock-up. Extensive work is under way to buildand test the experimental set up, as well as to test andcalibrate the sensors, both under air/helium andair/hydrogen flows.

HYDROGEN IGNITION BY VEHICLECOMPONENTS

OBJECTIVE

The objective of this work is to assess the ignitionpotential of some of the electrical and mechanicalcomponents and to identify best practices for the designand the positioning of vehicle components.

STEPS

Though not applicable to vehicles, the same strategy asthe one of the ATEX 94/9CE [9] & ATEX 99/92CE [10]is followed. Basically, it consists in demonstrating thatcomponents have no ignition potential under normalcircumstances whenever an explosive atmosphere canform accidentally.

The first step therefore consists in listing mechanical andelectrical components in the vicinity of hydrogensystems. Then, some of them will be selected toundertake ignition tests in dedicated chambers. Basedon results, recommendations will be made regarding thedesign and the location of the components.

BACKGROUND INFORMATION

In the course of this project, we have been looking atcross-experiences from natural gas vehicles (NGV) aswell as from industrial vehicles that could accidentallydrive into an explosive atmosphere (fork lift trucks).Regarding NGV vehicles, no recommendation at all isfound regarding protection or segregation of electricalequipment [11]. As far as forklift trucks are concerned(category 3 vehicle for type 2 zone), the EN 1755 [12]standard gives a few hints which include:

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

• minimum distance to prevent mechanical frictions orlimit speed at which friction can occur (brakes,clutch,...),

• bounding of all mechanical parts to ensure sameelectric potential,

• enclosing of electrical equipment to limit combustiblegas ingress whenever the forklift drives through anexplosive cloud,

• installing gas detection onboard forklift truck. A 10%LFL concentration triggers an alarm. A 25% LFLconcentration stops the forklift,

• and also, limiting maximum surface temperatureswhen in use.

Now, looking at GRPE [3] latest draft provisions, onecould notice quite a few recommendations on electricalequipment. That includes:

• preventing hydrogen ingress into power supplyconnections where hydrogen leaks are possible(paragraph 14.11.4.),

• providing sufficient insulation of electricalcomponents to ensure that no current passesthrough hydrogen containing parts (paragraph 6.5.1)

• & ensuring that electrical connections andcomponents in the gas tight housing are constructedsuch that no sparks are generated (paragraph14.10.3.)

TOOLS

Tests will be carried out inside an explosion chamberdedicated to test electrical or mechanical equipment.The picture below features a typical test chamber.

Figure 3: Front view of a typical test chamber to testignition potential of equipment

EARLY RESULTS

So far, the following components have been selected forpractical testing:

• the hydrogen compartment fan,• the electric motor that drives the vehicle,• the electrical converter,• and the braking system.

Moreover, some non-conductive material componentshave been selected in order to check whether or notthey can lead to an electrostatic ignition hazard.

Regarding early recommendations for the design ofvehicle components in order to control ignition sourcesin normal use, the following can be proposed so far:

Component

Brakingsystem

Fans

Sparkless drivemotors

References

EN1755[12]

EN50021

[13]

EN50021

[13]

State of the art design to control of ignition sources

Blocks should be made of a combination of non-metallic materials and cast iron or an equivalentmaterial. Light alloys are not allowed. Non-metallic materials should not contain more than40% metal in weight and should be made ofpowder or fine wires (sizes given in standard).Blocks should be glued or pressed on brakeshoes. Holes or assembly should not be foundon friction surfaces,Temperature should be monitored,And wear should be checked.IP10 rating (IP:Protection Index for water anddust penetration),on the output side and IP 20rating on the input side,Distances between moving and fixed partsgreater than 1/100 of the fan diameter (at leastbetween 1 and 5 mm),3,5 J shock test without hazardous deformation(or 2 J if fans are physically protected againstshocks).IP 54 rating,Dielectric strength test to ensure sufficientelectric isolation,Expectations on electrical couplings,Minimum distance between rotor and stator tobe implemented.

Table 3: State of the art design to control ignitionsources for some key components

HYDROGEN COMBUSTION

OBJECTIVE

This work aims at defining maximum tolerable explosivevolumes for various leakage environments and providingstraightforward pressure/effect prediction rules for riskassessment. Test results will also be useful to evaluatethe prediction potential of CFD tools.

STEPS

This experimental programme will first investigatepressure effects caused by the ignition of moderate andcalibrated explosive volumes for various levels ofconfinement from free environment up to congestedenvironment similar to those we can find onboardvehicles. Pressure prediction rules will be derived fromthese early tests. Then explosions will be carried outonboard vehicle following leakage flows and locationsconsistent with those found in the earlier workprogramme. These tests will also help us to evaluate thedestructive potential of onboard vehicle explosions.

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

BACKGROUND INFORMATION

Although not wanted, leaks will inevitably occur onboardvehicles. It is therefore a necessity to work on amaximum allowable explosive volume with tolerableassociated effects. Once this limit is set, it will be asignificant input to size safety barriers for leak detectionand interruption. So far, the only "tolerable explosivevolume" we found come from the industry. TheEuropean guide document for the implementation ofATEX 99/92CE directive [10] indicates, independently ofthe combustible gas considered that:

• an unconfined explosive atmosphere should alwaysbe considered hazardous when its volume is above10 litres,

• and an explosive atmosphere in a closed roomshould always be considered hazardous when itsvolume is greater than 1/10 000 of the room volume.

Other work [15] investigated the limit of acceptableexplosive volumes in natural gas turbine casings. Therecommended safety criterion is that the natural gasflammable cloud volume should be less than 0.1% of thenet enclosure volume. In addition to the above conditionon the allowable size of a flammable gas cloud relativeto the net enclosure volume, the flammable gas cloudvolume should also be less than 1 m3 in all cases,irrespective of the enclosure volume. Safety systemshave to be designed (detection, ventilation) in order toinsure that these limits are never exceeded. One shouldmention that people would not be present inside theenclosure volume but are expected to stand nearby.Proposed acceptable explosive volumes are notsusceptible to cause casing rupture and blast injuries.

Examples above illustrated confined and unconfinedexplosions as well as explosion with and without directexposure of people. These various cases will be takeninto consideration.

Regarding existing data on explosion experiments formoderate and calibrated explosion volumes in more orless confined environment, the review has not yetstarted. However, similar experiments will be carried outby FZK3 in the framework of the HySafe network.

For explosion onboard vehicles, we can mentionexperiments carried out by the Motor Vehicle FireResearch Institute USA [8] where they investigatedhydrogen leakage in the engine compartment. Theyinvestigated various release rate and ignition delay.Test showed that the maximum engine compartmentoverpressure was 220 mbar. This maximumoverpressure that caused the hood to bend wasobtained for a 64 sec release of 51 g of hydrogen.Shorter released produced overpressure of 10 mbar to152 mbar.

TOOLS

Forschungszentrum Karlsruhe

These tests are planed for 2007. The experimental setup has not been designed yet. Tests will most probablybe combined with those undertaken by FZK in theframework of the HySafe network.

HYDROGEN FLAME

OBJECTIVE

This work aims on one hand at providing data onhydrogen flame length and on hydrogen flame radiationfor some severe scenarios (significant leak under thehood, hydrogen escaping through the thermal fuse).Besides these tests, less severe cases will beinvestigated in order to study conditions under which ahydrogen flame cannot be sustained. Ease of ignitionshall also be investigated for high and moderatepressures with various leak orifices. These latter casesaddress minor / non detectable leakage that do notcause any explosion hazard but could however cause asignificant fire hazard whenever the vehicle is parked ordriven.

STEPS

The work programme includes both an experimental anda numerical part. The first step will consist in selecting aset of realistic scenarios/situations to be investigated.The experimental set up for flame length and flameradiation has not yet been designed. Regarding flamelength, both free and impinged jets will be looked at.Seeding will probably be used to visualise flames. As faras micro leaks are concerned, we will look at leakorifices most probably below 1 mm for various values ofpressure.

BACKGROUND INFORMATION

It has been established experimentally that hydrogendischarges through circular orifices larger than a criticaldiameter sustains stable lifted flames irrespective of thereservoir pressure driving the release. At smallerdiameter however, stable burning will only be achievedat operating pressures higher than a particular,diameter-dependant threshold [16].

Regarding vehicle related literature contributions, wecan mention two practical contributions to the topic. Firstof all, we can mention the experimental work from theMotor Vehicle Fire Research Institute USA [8] where jetflames in the engine compartment have beeninvestigated. Some more data was previously releasedby the Japan Automobile Research Institute [7]. JARIcompared for a given leakage rate (11.8 g/min forhydrogen and 28.6 g/min for natural gas/methane)critical diameter under which a gas flame was notsustained. They found out that the methane flamebecame no more viable with a rather large diameter of2.00 mm or less. On the other hand, the hydrogen flamewas sustained at the leak orifice even with a nozzlediameter of 1.0 mm. Authors concluded that thefavourable stability of hydrogen flames is advantageous

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

as a means to avoid explosion. Needless to say that the1 mm orifice considered is rather large. Much smallerleak orifice could be expected without being noticed.They could eventually cause a fire hazard. It is thereforenecessary to investigate leak orifices smaller than 1 mm.The DRIVE project will investigate these situations.

Moreover, this research institute also investigated theflame behaviour of hydrogen released from 350 barhydrogen tanks. They found out that the hydrogen flamereleased upwards reached its maximum height of 7-8meters in 5 to 10 seconds. This flame length is verysignificant and could cause severe hazards for carsoccupants as well as for rescue services. It is thereforeof a prime importance to provide advice on how thisrelease should be dealt with (upward, downward underthe vehicle...). The DRIVE project will also address thisissue within this work programme.

RESULTS AND DISSEMINATION

The DRIVE project aims at investigating experimentallyand numerically hydrogen vehicle related accidentalscenarios in order to strengthen risk assessment as wellas to promote safe hydrogen handling solution onboardvehicle. Three documents will be produced:

• the first one will address hazard quantificationresults and hazard quantification tools for hydrogenvehicles. Hazardous phenomena will be explained.Issues such as flame length or explosiveatmosphere volume versus pressure and diameterwill be detailed,

• the second one will address best practices whenusing CFD tools to predict hydrogen behaviour whenreleased accidentally or under normal situation,

• and the final one will address best practices whenhandling hydrogen onboard vehicles. Its content willinclude both generic risk assessment as well as bestpractices for the design of hydrogen vehicles.

These results will be presented in the course of openseminars.

CONCLUSIONS

Field of investigation covers the all range of phenomenathat can be encountered in hydrogen accidents fromleakage, to dispersion, ignition and finally combustion.This programme aims at providing experimental andnumerical results in order to provide realistic riskassessment and to be able to specify precisely expectedhydrogen vehicles safety performances. DRIVE outputswill be turned into practical graphs to facilitate riskquantification. Recommendations on best practicevehicle design will also be drafted. Actually, topics suchas maximum allowable explosive volume, control ofignition sources and location of hydrogen componentswill be addressed.

This project has just started. However preliminaryreviews of existing experimental data and guidelineshave been performed, to better define the expectedresults of the project. These reviews and early thoughtsand findings represent important contributions to theelaboration of safety guidelines for hydrogen vehicles.We can without any doubt consider today that DRIVEoutputs will be of a great help for designing safehydrogen vehicles.

ACKNOWLEDGMENTS

Authors would like to thank the French NationalResearch Agency ANR, and its hydrogen programmePAN-H, which financially supports this project.

REFERENCES

[1] Dr Fùrst, S. & al, Safety of Hydrogen-fueled motorvehicles with IC engines; Proceeding from the 1st

International Conference on Hydrogen Safey, Pisa,September 2005

[2] CaFCP Technical report, 2004, 'Support facilities forhydrogen fuelled vehicles-Conceptual design and costanalysis study', Prepared for California Fuel CellPartnership by Parsons and Brinckerhoff in associationwith TIAX and University of Miami.

[3] UN GRPE Informal Group on Hydrogen/Fuel CellVehicles, Proposal for a new draft regulation. Uniformprovisions concerning the approval of:

specific components of motor vehicles usingcompressed gaseous hydrogen,

vehicles with regard to the installation of specificcomponents for the use of compressed gaseoushydrogen.

[4] Swain M.R., Schriber J.A. and Swain M.N., (1998a)"Addendum to Hydrogen Vehicle Safety Report:Residential Garage Safety Assessment. Phase-1 : Risksin Indoor vehicle storage", Prepared for DirectedTechnologies Inc., pp 118. & Swain M.R., Grilliot E.S.

[5] Swain M.N., (1998b) "Addendum to HydrogenVehicle Safety Report: Residential Garage SafetyAssessment. Phase-2: Risks in Indoor vehicle storage",Prepared for Directed Technologies Inc., pp 35.

[6] Swain M.R., Grilliot E.S. and Swain M.N., (1999)"Experimental verification of a hydrogen risk assessmentmethod", Chemical Health & Safety, pp 28-32.

[7] T. Hayashi and S. Watanabe, (2004) "HydrogenSafety for Fuel Cell Vehicles", Proceeding of WHEC-15,Yokohama.

[8] Presentation from Bob Zalosh at IEA Task 19meeting on March 16-17th, 2006

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4

[9] ATEX 99/92/EC - Equipment and protective systemsintended for use in potentially explosive atmospheres

[10] ATEX 99/92/EC - Minimum requirements forimproving the safety and health protection of workerspotentially at risk from explosive atmospheres

[11] Regulation No. 110 - Uniform provisions concerningthe approval of:1. Specific components of motor vehicles usingcompressed natural gas (CNG) in their propulsionsystem;2. Vehicles with regard to the installation of specificcomponents of an approved type for the use ofcompressed natural gas (CNG) in their propulsionsystem

[12] EN 1755 : Safety of industrial trucks - Operation inpotentially explosive atmospheres - Use in flammablegas, vapour, mist and dust

[13] EN 50021 : Electrical apparatus for potentiallyexplosive atmospheres - Type of protection "n".

[14] Guide de bonne pratique à caractère noncontraignant pour la mise en œuvre de la Directive1999/92/CE du Parlement européen et du Conseil

concernant les prescriptions minimales visant àaméliorer la protection en matière de sécurité et desanté des travailleurs susceptibles d'être exposés aurisque d'atmosphères explosives, CommissionEuropéenne, DG Emploi et affaires sociales, Santé,sécurité et hygiène au travail, Version finale avril 2003

[15] Health & Safety Laboratory, Outstanding safetyquestions concerning the use of gas turbines for powergeneration - Executive report - CM/04/02

[16] C.B. Devaud & al, Stability of underexpandedsupersonic jet flames burning H2-CO mixtures, Shockwaves (2002) 12: 241-249

CONTACT

DRIVE project coordinator: Lionel PERRETTE

Tel: +33 3 44 55 63 39

Lionel.perretteneris.fr

iner

is-0

0976

178,

ver

sion

1 -

9 Ap

r 201

4