Evaluation of Safety Distances Related to Unconfined Hydrogen Explosions

Sergey DorofeevFM Global

1st ICHS, Pisa, Italy, September 8-10, 2005

Motivation

• H2 releases in confined and semi-confined geometries (tunnels, parking, garages, etc.) represent a significant safety problem Possibility of hydrogen accumulation, Promoting role of confinement for FA and pressure build-

up

• Unconfined H2 explosions can also be a significant safety problem Releases in obstructed areas (refuelling stations,

hydrogen production units, etc.) Relatively fast dilution of H2-air mixtures at open air and

inefficient FA without confinement On the other hand: large quantities of H2

Confined versus unconfined

Motivation

• Potential consequences of unconfined hydrogen explosions important for safety distances Blast effects Thermal effects Effects of explosion-generated fragments

• Blast effects are usually of the prime interest for safety distances

• May be especially important for hydrogen because of their potential severity

• Unconfined hydrogen explosions and their blast effects are the focus of the present study

Consequences

Motivation

• A detailed analysis of blast effects should include Hydrogen release and distribution Flame propagation and blast generation in complex 3D

geometry Blast wave propagation and its effect on the surrounding

objects

• This would generally require an application of 3D CFD simulations Limited variety of the cases / applications

• A simple approximate analytical tool should be useful Screening tool to select the cases where detailed

analysis may be necessary

Analysis strategy

Objective

• Develop a simple approximate method for evaluation of blast effects and safety distances for unconfined hydrogen explosions Model for evaluation of hydrogen flame speeds in

obstructed areas Model for properties of “worst case” hydrogen distribution Model for blast parameters Set of blast damage criteria

Methodology

• Pressure effect of a gas explosion essentially depends on the maximum flame speed

• It is important to have a reliable estimate for the flame speed

• Flame speed increases due to: Increase of the flame area in an obstacle field Increase of the turbulent burning velocity during flame

propagation

Flame speeds

R

fTf A

ASV

Methodology

• Flame folding due to obstacles

• Plus Bradley correlation for turbulent burning velocity:

Flame speeds

3/12

2

)(3

41)1(

T

Lf

L

x

R

x

ySbaV

ba

x

x

yR

R

Methodology

• Experimental data

Flame speeds

10

100

1000

0.01 0.1 1 10 100

Distance, m

Exp

eri

men

tal f

lam

e s

peed

, m/s

x=45, y=4 mm H2

x=33, y=4 mm H2

x=31, y=4 mm H2

x=18, y=1 mm H2

x=12, y=1 mm H2

x=10, y=1 mm H2

x=9, y=0.65 mm H2

x=7, y=0.65 mm H2

x=6, y=0.65 mm H2

x=39, y=5 cm C2H4

x=22, y=5 cm C2H4

no obstacles H2

no obstacles C3H6

Methodology

• Correlation

Flame speeds

0

50

100

150

200

250

300

0 50 100 150 200 250 300

Model flame speed, m/s

Exp

eri

men

tal f

lam

e s

peed

, m/s

x=45, y=4 mm H2 x=33, y=4 mm H2

x=31, y=4 mm H2 x=18, y=1 mm H2

x=12, y=1 mm H2 x=10, y=1 mm H2

x=9, y=0.65 mm H2 x=7, y=0.65 mm H2

x=6, y=0.65 mm H2 x=39, y=5 cm C2H4

x=22, y=5 cm C2H4 no obstacles H2

no obstacles C3H6

Methodology

• There is clearly a variety of release scenarios, which can affect the resulting hydrogen distribution

• Continuous release Slow: jet or plume with size of flammable volume break

size Fast: jet with size of flammable volume >> break size

• Instantaneous release – most dangerous Pressure vessel rupture LH2 release or vessel rupture

• Other scenarios

Hydrogen distribution

Methodology

• Instead of considering specific scenarios here, a simple general model for instantaneous releases is analysed

• This model assumes that the released hydrogen forms a cloud with a non-uniform concentration

• The form of the cloud is assumed to be semi-spherical, for simplicity

• Hydrogen concentration reachesmaximum in the centre and decreases linearly with radius

• Stoichiometric H2/air – unrealistic and overconservative!

Model for gas distribution

r

Cmax

Methodology

• Variable: maximum concentration in the centre, Cmax

• ‘Worst case’: maximum of < >=<(-1)SL>, averaged between UFL and LFL

• Properties of ‘worst case’: Cmax = 88% vol.

< > = 0.1max

<E> = 60% of total chemical energy

‘Worst case’ distribution

LFL

Cmax

UFL

Methodology

• Calculations of blast parameters are based on our method published in 1996

• Dimensionless overpressure and impulse are functions of flame speed, Vf

Blast parameters

),min( *2

*1

* PPP ),min( *2

*1

* III 3*2*3/4**

1 )/(0033.0)/(062.0)/(34.0 RRRP 968.0**

1 )/(0353.0 RI

))/(14.0/83.0(1 2**

20

2*2 RR

c

VP f

))/(0025.0)/(04.0/06.0(1

4.011 3*2**

00

*2 RRR

c

V

c

VI ff

Methodology

• An assessment of damage potential is made here using pressure-impulse (P, I) damage criteria

Damage potential

kPPII aa ))((

Damage description Pa, Pa Ia, Pa∙s k, Pa2∙s

Total destruction of buildings 70100 770 866100

Threshold for partial destruction; 50 to 75% of walls destroyed

34500 520 541000

Threshold for serious structural damage; some load bearing members fall

14600 300 119200

Border of minor structural damage 3600 100 8950

Results

• High congestion, x = 0.2 m; y = 0.1 m: a technological unit with multiple tubes / pipes.

• Medium congestion, x = 1 m; y = 0.5 m: a technological unit surrounded by other units / boxes.

• Low congestion, x = 4 m; y = 2 m: a large technological unit surrounded by other large units (e. g., refueling station)

Characteristic obstacle geometry

Results

• Obstacle geometry affects significantly flame speeds

• To reach 300 m/s: 1 kg, 40 kg, and 1000 kg high, medium, and low congestion

Flame speeds

0

100

200

300

400

500

600

0.1 1 10 100 1000

m, kg

Fla

me

spee

d, m

/s

Low congestion

Medium congestion

High congestion

Results

• Example for medium congestion

Radii for selected levels of damages

0

50

100

150

200

250

300

1 10 100

m, kg

R,

m

Full destruction (buildings)

50-75% destruction

Significant damage

Minimum damage

Results

• Scenarios

• Consequences Pressure Thermal Fragments

• Acceptance criteria Population Regulations Costs

Safety distances – contributing factors

Results

• Defined, as an example, by minimum building damage criterion for unconfined H2 explosions

Safety distances - example

1

10

100

1000

0.1 1 10 100 1000

m, kg

R, m

Low congestion

Medium congestion

High congestion

TNT equal energy

Results

• The same method applied to: hydrogen, ethylene, propane, methane – medium congestion

Safety distances – fuel comparison

1

10

100

1000

1 10 100 1000 10000

m, kg

R, m

CH4

C3H8

C2H4

H2

Results

• The same as a function of total combustion energy of released gas

Safety distances – fuel comparison

1

10

100

1000

100 1000 10000 100000 1000000

E, MJ

R, m

CH4

C3H8

C2H4H2

Conclusions

• A simple approximate analytical method for evaluation of blast effects and safety distances for unconfined H2 explosions has been presented

• Potential blast effects of unconfined H2 explosions strongly depends on the level of congestion

• Certain threshold values of the mass of hydrogen released may be defined as potentially damaging

• This minimum mass varies by several orders of magnitude depending on the level of congestion

• In terms of potential blast effects, hydrogen may represent a significantly high threat as compared to ethylene, propane, and methane