International Conference on Hydrogen Safety, Pisa, 8-10 September 2005 1 CFD modeling of large scale LH2 spills in open environment Dr. A.G Venetsanos.

International Conference on Hydrogen Safety, Pisa, 8-10 September 2005 1

CFD modeling of large scale LH2 spills in open environment

Dr. A.G Venetsanos1

Prof. J.G. Bartzis2

1Environmental Research LaboratoryInstitute of Nuclear Technology and Radiation Protection

National Center for Scientific Research DemokritosAthens, Greece

2University of Western Macedonia, Kozani, Greece

[email protected]


Introduction

• Scope– To increase our understanding and predictive ability of LH2

release and dispersion• Framework

– Work performed in the framework of the SBEP activity of HYSAFE NoE

– Also continuation of previous work (Statharas et al., 2000)• Focus

– NASA-6 large scale LH2 release in open environment, (Witcofski and Chirivella, 1984)

– ADREA-HF CFD code (inhouse, commercial) extensively applied during the EIHP and EIHP-2 EC funded projects

– Parametric study performed to investigate the effects of• the source model• the presence of the elevated pond sides around the source• the contact heat transfer between ground and nearby ambient air.


NASA-6 experimental description

Deployment of instrumentation towers and typical instrumentation array

Liquid hydrogen spill-line, valve, pond and diffuser5.11 m3 LH2 released in 38s, mass flow rate 9.5 kg s-1

2.2 m s-1 wind speed at 10m height, 15 ºC ambient temperature and 29% relative humidity


NASA-6 experiment results

Experiment, t = 21.33 s


Temperature deduced concentrations


ADREA-HF Modeling strategy

• Mean flow equations– Mixture mass (fully compressible)– Mixture momentum (3 equations)– Mixture enthalpy– Hydrogen mass fraction (1 equation for liquid + vapour)

• Turbulence modeling– Standard k- model (Launder B. E. and Spalding, 1974)– Buoyancy effects included

• Physics– Liquid phase obtained from the equilibrium phase change model– Temperature obtained for given enthalpy, pressure and hydrogen mass

fraction– Ground heat transfer modelled by solving a transient one dimensional

energy (temperature) equation inside the ground, with or without contact heat transfer

– Source modelled as a two phase jet or as pool– Earthen sides of pond modelled as a thin fence or not modelled


ADREA-HF mean flow equations

0

i

i

x

u

t

i

j

j

it

ji

ij

iji

x

u

x

u

xg

x

P

x

uu

t

u

jt

t

jj

j

x

q

Scd

xx

qu

t

q 111

j

ii

jjjt

t

jj

j

x

qdH

xxdt

dP

x

H

xx

Hu

t

H

Pr

Mixture mass

Mixture momentum

Mixture enthaply

Hydrogen mass fraction


ADREA-HF standard k- turbulence model

Turbulent kinetic energy

0.1 where, k

B

jk

t

jj

j GGx

k

xx

ku

t

k

z

g-G and z

tB

j

i

i

j

j

it x

u

x

u

x

uG

1.0 and 1.92 1.44, 1.3, where

,

321

231

CCC

CGCGCkxxx

u

t Bj

t

jj

j

Dissipation rate of turbulent kinetic energy

09.0C where,2

kCt

Eddy viscosity


ADREA-HF source models

• Two phase jet:– Horizontal surface (representing the diffuser):

• size 0.5x0.5 m2

• pointing downwards• placed in the first cell next to ground

– Boundary conditions:• vertical velocity -11.47 m s-1 from time 0 to 38 seconds• pressure 101325.0 Pa• temperature 20.4 K (boiling)• void fraction (vapour volume/total volume) 97%.

– These conditions give: • hydrogen liquid mass fraction (liquid mass/total mass) 65%• hydrogen density 3.315 kg m-3

• hydrogen flow rate equal to 9.5 kg s-1


ADREA-HF source models

• Pool:– A horizontal open area source (representing the pool)

• radius 4.45 m constant• pointing upwards• assumed in place of the ground

– Boundary conditions on this surface:• vertical velocity 0.2535 m s-1 from time 0 to 38 seconds• pressure 101325.0 Pa• temperature 20.4 K (boiling)• void fraction (vapour volume/total volume) 100%.

– These conditions give • hydrogen density 1.207 kg m-3

• hydrogen flow rate equal to 9.5 kg s-1


ADREA-HF ground heat transfer model

TTzz

ucH z

pg 0

0

*

lnPr

00 zgCpg TThcH 45.0

0**395.10

zu

uhC

0*

ln1

z

z

u

U

Contact heat transfer coefficient (Zilitinkevich, 1970)Ground heat flux

Friction velocity

z

T

zt

Tc gg

gpgg

Energy equation solved inside ground


ADREA-HF equilibrium phase change model

2

2

,1

,1

,1

,11

qqq

L

L

V

V

Mixture density Mass fractions

2,1,11 qqq LV

LV qqq ,1,11

• Phase change criterion:– Liquid phase of component-1 appears instantaneously (condensation

occurs), when the component-1 mass-fraction exceeds its saturation value in the mixture (Vice versa for evaporation)

• Distribution into phases:– The amount of component-1 in liquid phase is such that a saturation state

exists in the gaseous region:

VSqq ,11

2,11,1

1,1,1 MPPMP

MPq

SS

SVS

TPP SS ,1,1

VSV

V qqq

q,1

2,1

,1


ADREA-HF thin obstacle model

Additional resistance forces in the horizontal momentum equations (Andronopoulos et al., 1994)

i

iiii xuuF

2


ADREA-HF Solution strategy

• Domain and grid (Cartesian)– X-Z symmetry assumed– Domain 175x80x68 m. Computational cells 39600– minimum horizontal cell size 0.6 m close to the source– maximum horizontal cell size nearly 10 m close to the domain

boundaries– minimum vertical cell size 0.2 m near the ground and source– maximum vertical cell size about 9 m near the top of domain

• Discretization– Control volume approach– Fully implicit formulation– First order in time– Upwind scheme for convective terms– Automatic time step selection (0.001-0.25s)

• Three step solution sequence– One dimensional (in the z-direction) calculation of the approaching

boundary layer– Three dimensional steady state calculation of the flow over the fence– Transient hydrogen release


ADREA-HF simulations matrix

CaseSource

type FenceContact heat

transfer1 Jet Yes Yes2 Jet Yes No3 Jet No Yes4 Pool No Yes5 Pool Yes Yes


ADREA-HF Case 1 predictions versus experiment

Predicted contours of hydrogen concentration (by vol.) on symmetry plane at t = 21 s

Jet, with fence, with contact heat transfer





Jet, with fence, without contact heat transfer





Jet, without fence, with contact heat transfer





Pool, without fence, with contact heat transfer





Pool, with fence, with contact heat transfer



ADREA-HF predictions versus experiment

CaseSource





CaseSource





CaseSource





CaseSource





LAuV post calculation The shaded area corresponds to the range of measurements (taken from Verfondern and Dienhart (1997))


Conclusions

• The ADREA-HF CFD code was successfully applied to simulate the NASA trial-6 experiment. A series of CFD runs were performed to investigate on the effects of source model, fence presence and contact heat transfer.

• In all cases considered it was not able to reproduce neither the very sudden changes in cloud structure observed experimentally nor the high levels of concentrations measured at tower 7, located 33.8 m downwind and 12.9 m laterally from the source. This behaviour was attributed to wind meandering, observed during the experiments, but not modelled herein.

• Entirely different cloud structures were obtained depending on the method the source was modelled. Modelling the source as a two-phase jet, pointing downwards resulted in predicted concentrations in much better agreement with the experiments. Modelling the source as a pool resulted in overestimation of concentration levels.

• The earthen sides of the pond were modelled as a fence of infinitesimal thickness. The calculations showed that near ground concentration levels downstream the source increased when the fence was removed. This effect was more pronounced when the source was modelled as a jet.


Conclusions

• Accounting for the temperature difference between ground and adjacent air at the level of the roughness length was found to be very important

• Using a two-phase jet pointing down, contact heat transfer and including the fence results were the closest to the experimental. Still, the predicted concentration levels on tower 5, 9.4 m are underestimated and at 1 m overestimated suggesting that the predicted cloud is less lifted from ground than in the experiment. This suggests that probably a more intense heat flux from the ground would be required to obtain a better agreement with the experiment.

International Conference on Hydrogen Safety, Pisa, 8-10 September 2005 1 CFD modeling of large scale LH2 spills in open environment Dr. A.G Venetsanos.

Documents

hydrogen safety

international conference

modelled slide

adreahf standard

adreahf source models

mass flow rate

adreahf modeling strategy

relative humidity slide