An Intercomparison Exercise on the Capabilities of CFD Models

to Predict Deflagration of a Large-Scale H2-Air Mixture

in Open Atmosphere J. García, E. Gallego, E. Migoya, A. Crespo (UPM)

A. Kotchourko, J. Yañez (FZK), A. Beccantini (CEA),

O.R. Hansen (GexCon), D. Baraldi (JRC), S. Høiset (N-H),

M.M. Voort (TNO), V. Molkov (UU)

• Standard Benchmark Exercise Problems SBEPs

• Objectives: – establishing a framework for validation of

codes and models for simulation of problems relevant to hydrogen safety,

– identifying the main priority areas for the further development of the codes/models.

SBEPs in HySafe

• The experiment was performed by the Fraunhofer Institut Chemische Technologie (Fh-ICT), Germany in 1983.

• 20 m diameter polyethylene hemispheric balloon (total volume 2094 m3).• Homogeneous stoichoimetric hydrogen-air mixture.

Experiment description

• Initial conditions:– Pressure: 98.9 kPa

– Temperature: 283 K.

• Pressure dynamics was recorded using 11 transducers, installed on the ground level in a radial direction at different distances from the centre.

• The deflagration front propagation was filmed using high-speed cameras.

Experiment description

Experiment description

Variation of flame front contours with time.

Experiment results

The flame front radius vs. time

Experiment results

Organisations and codes participating

Participant Organisations CodesCEA, Commissariat à l’Energie Atomique, France CAST3M

FZK, Forschungszentrum Karlsruhe, Germany COM3D

GexCon, GexCon AS, Norway FLACSv8.1

JRC, Joint Research Centre, European Commission Reacflow

NH, Norsk Hydro ASA, Norway FLACSv8.0

TNO, The Netherlands AutoReaGas v3.0

UU, University of Ulster, UK FLUENTv6.1.18

Participant & Code

Turbulence model

Chemical model

CEA CAST3M

- CREBCOM combustion model

GexCon FLACS v8.1

k- standard Beta flame modelReaction rate based on one step model with burning velocity from flame-library

FZKCOM3D

k- standard CREBCOM combustion model. Adjustable parameter Cf, governing the rate of chemical

interaction and therefore a visible flame speed.

JRC Reacflow

k- standard Modified Eddy Dissipation combustion model

NH FLACS v8

k- standard Beta flame model

TNO AutoReaGas

v3.0

k- standard Combustion rate depends on the mean composition of the mixing region. Flame speed correlates via empirical relations with the calculated turbulence parameters

UU FLUENT

v6.1.18

LES (RNG) Gradient method

Models

ModelsParticipant & Code

Resolution method & discretisation scheme

GridComputer &CPU time

CEACAST3M

Operator splitting technique. First order

1D spherical domainCell size 0.1 m

Not available

GexCon FLACS v8.1

SIMPLESecond order

3D-Cartesian Cell size: 0.5 m

1 CPU PCs 0.5-4 Gb RAM Linux4h CPU

FZKCOM3D

Solver coupled with turbulence & chemical models.

3D cartesian gridCell size: 0.3 m combustion 0.59 m pressure

Cluster of 7 Athlon PC - 2 CPU each.

Linux 2.4.20.≈ 14 days /with 14 processors

JRC Reacflow

Explicit scheme Second order

3D unstructured adaptive grid 0.15 m

Linux cluster26.5 to142 h CPU

NH FLACS v8

SIMPLESecond order

3D-CartesianCell size: 0.67 m

6 days CPU (1 s experiment)

TNO AutoReaGas

SIMPLEFirst order

3D Cartesian 27000 cells

UU FLUENT

Explicit method 2nd order

3D unstructured tetrahedral grid(a): 0.4 m (b): 0.2 m

2/6 CPU 4/12 Gb RAM142/197h CPU (0.32/0.63 s

experiment)

• All experimental results were known before the calculations.

Comments about results

• The influence of the polyethylene film and wire net was supposed negligible.

• Sensors at 2, 8 and 18 m have to be influenced by combustion because they do not recover ambient pressure.

Dynamics of the averaged flame front radius with time

0

5

10

15

20

25

0.00E+00 1.00E+02 2.00E+02 3.00E+02 4.00E+02 5.00E+02 6.00E+02 7.00E+02

Time (ms)

Fla

me

fro

nt

rad

ius

(m)

CEA

FzK

Gexcon(x)

Gexcon(z)

JRC

NH

TNO

TNO-1

UU(a)

UU(b)

Experimental

Dynamics of the flame front radius with time

Pressure dynamics at R = 2m

-1.00E+04

-8.00E+03

-6.00E+03

-4.00E+03

-2.00E+03

0.00E+00

2.00E+03

4.00E+03

6.00E+03

8.00E+03

1.00E+04

0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00

Time (s)

Rel

ativ

e P

ress

ure

(P

a)

CEA

FzK

Gexcon

JRC

NH

TNO

TNO-1

UU(a)

UU(b)

Experimental

Pressure dynamics at R= 2 m

Flame front reaches the sensor

Pressure dynamics at R = 5m

-1.20E+04

-1.00E+04

-8.00E+03

-6.00E+03

-4.00E+03

-2.00E+03

0.00E+00

2.00E+03

4.00E+03

6.00E+03

8.00E+03

1.00E+04

1.20E+04

0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00

Time (s)

Re

lati

ve

Pre

ss

ure

(P

a)

CEA

FzK

Gexcon

JRC

NH

TNO-1

UU(a)

UU(b)

Experimental

Pressure dynamics at R= 5 m

Flame front reaches the sensor

Pressure dynamics at R = 8m

-1.00E+04

-5.00E+03

0.00E+00

5.00E+03

1.00E+04

1.50E+04

0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00

Time (s)

Re

lati

ve

Pre

ss

ure

(P

a)

CEA

FzK

Gexcon

JRC

NH

TNO-1

UU(a)

UU(b)

Experimental

Pressure dynamics at R= 8 m

Flame front reaches the sensor

Pressure dynamics at R = 18m

-1.00E+04

-8.00E+03

-6.00E+03

-4.00E+03

-2.00E+03

0.00E+00

2.00E+03

4.00E+03

6.00E+03

8.00E+03

1.00E+04

0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00

Time (s)

Re

lati

ve

Pre

ss

ure

(P

a)

CEA

FzK

Gexcon

JRC

NH

TNO-1

UU(a)

UU(b)

Experimental

Pressure dynamics at R= 18 m

Flame front reaches the sensor

Pressure dynamics at R = 35m

-6.00E+03

-4.00E+03

-2.00E+03

0.00E+00

2.00E+03

4.00E+03

6.00E+03

0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00

Time (s)

Rel

ativ

e P

ress

ure

(P

a)

CEA

FzK

Gexcon

JRC

NH

TNO-1

UU(a)

UU(b)

Experimental

Pressure dynamics at R= 35 m

Pressure dynamics at R = 80m

-4.00E+03

-3.00E+03

-2.00E+03

-1.00E+03

0.00E+00

1.00E+03

2.00E+03

3.00E+03

0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00

Time (s)

Rel

ativ

e P

ress

ure

(P

a)

FzK

Gexcon

JRC

NH

TNO-1

UU(a)

UU(b)

Experimental

Pressure dynamics at R= 80 m

Video: FzK

Video: GexCom

Flame velocity

Pressure

Video UU

• The flame velocity is reproduced quite well in most of the calculations.

• The pressure dynamics obtained numerically are in good agreement with the experiments for the positive values.

• Negative pressures are more sensitive to far field boundary condition, this can be avoided using larger domains and finer grids.

• More benchmarks will be necessary to calibrate and improve the codes.

Conclusions