On numerical simulation of liquefied and gaseous hydrogen releases at large scales V. Molkov, D. Makarov, E. Prost 8-10 September 2005, Pisa, Italy First International Conference on HYDROGEN SAFETY

Introduction of hydrogen as an energy carrier makes great demands on hydrogen safety. Development of robust and reliable risk assessment methodologies requires all-round validation of models and tools. The need to model non-uniform hydrogen-air mixture formation at real scales is important to have realistic initial conditions for subsequent modelling of partially premixed hydrogen combustion. The aim of this study is validations of the LES model in application to large-scale hydrogen release scenarios and formulation of tasks for future research in this area.

CFD models “could in principal be capable of being truly predictive tools outside of their immediate range of validation ” Lea C.J., Ledin H.S., Health and Safety Laboratory Report HSL/2002/02 LES approach may provide a route to simpler and more realistic models for technological applications. Bray K.N.C., 26th Symp. (Int.) on Combustion, 1996 LES avoids time-averaging of fluctuations and provides instantaneous field functions. Large Eddy Simulation (LES) becomes an alternative for k-  model including the field of engineering applications.

Contents Contents The LES modelThe LES model LH2 release in open atmosphereLH2 release in open atmosphere GH2 release in a closed vesselGH2 release in a closed vessel

Large Eddy Simulation (LES) model

Conservation of massConservation of mass Conservation of momentumConservation of momentum Conservation of energyConservation of energy LES model (1/2)

RNG SGS turbulence modelRNG SGS turbulence model Conservation of “H 2 ” concentrationConservation of “H 2 ” concentration LES model (2/2)

Liquefied hydrogen release in open atmosphere

Chirivella J.E., Witcofski, R.D. Am. Inst. Chem. Eng. Symp., 82, No 251, 1986, pp : - Spill 5.11 m 3 (361.8 kg) of LH 2 in 38 s - LH 2 pool radius between 2 and 3 m - Total evaporation time 43 s - Wind speed ~2.2 m/s at height 10 m NASA experiment

Calculation domain (1/2) 180 m 70 m Spill area and instrumentation towers area Cloud propagation area Characteristic size of CV:Numerical grid: CV –tower location m –cloud area m –the rest of domain up to 10 m

Calculation domain (2/2) 70 m 180 m Spill area Cloud propagation area Characteristic size of CV:Numerical grid: CV –spill area m –cloud area m –the rest of domain up to 10 m

Initial conditions atmosphere velocity profile: where (provided u=2.2 m at H=10 m) Boundary conditions velocity profile at inflow prescribed pressure conditions at outflow boundaries, p=0 Pa H 2 injection mass injection rate Run 1: injection area radius Run 2: injection area radius average injection velocity instant injection velocity Run 1: turbulence Run 2: turbulence Geomerty: Run 1 (no pool border, no obstacles), Run 2 (+) Numerical details

H2 concentration (Run 1) T exp = s T sim = s

H2 concentration (Run 2) T exp = s T sim = s

Simulated temperature (Run 1)

Simulated temperature (Run 2)

Visible cloud (Run 1)

Visible cloud (Run 2)

Cloud propagation (Run 1)

Cloud propagation (Run 2)

Phenomena to be addressed Condensation of air in temperature range K (with heat release) and evaporation above 90 KCondensation of air in temperature range K (with heat release) and evaporation above 90 K Two phase flow (gas: hydrogen-air; solid: air ice)Two phase flow (gas: hydrogen-air; solid: air ice) Detailed spill modelling (initial fractions of GH2 and LH2; heat transfer to the ground: initial violent evaporation stage, etc)Detailed spill modelling (initial fractions of GH2 and LH2; heat transfer to the ground: initial violent evaporation stage, etc)

Gaseous hydrogen release in 20-m 3 closed vessel

5.5m 2.2m Experiment 1.4m H2H2 Time of release = 60 seconds Volume injection rate: V=4.5 l/s

Non-uniform tetrahedral grid CV number: CV size: m close to place of H 2 injection CV size: up to 0.20 m in the rest of domain “Uniform” grid CV number: CV size: m s min Calculation domain

Turbulence length scales (after S Pope) L 11 = 0.7  r 1/2 – large-scale energy containing eddies r 1/2 =0.094  y – jet half-width l ei = L 11 /6– interface between energy containing and isotropic eddies L 11 r 1/2 l ei LES, VLES, super VLES…?

Initial conditions – –quiescent air, u=0 m/s, – –initial air concentration Y air =1.0, – –initial temperature T=293K Boundary conditions – –t=0-1s: V inj increased from 0 to 57.5 m/s – –t =1-59s: V inj =57.5 m/s – –t=59-60s: decrease from 57.5 to 0 m/s – –t=60s-251min: V inj =0 m/s – –Y H2 =1.0, T inj =293K Numerical details – –explicit linearisation of the governing equations – –implicit method for solution of linear equation set – –second order accurate upwind scheme for convection terms, central- difference scheme for diffusion terms – –Time steps: t=0-180 s:  t=0.01 s; t=3-251 min:  t= s Numerical details

Simulation results

2 min 50 min 100 min 250 min Hydrogen distribution 1

Hydrogen distribution 2

50min: V max =10 cm/s 100min: V max =8 cm/s 250min: V max =5 cm/s Residual velocities

Conclusions The LES model has been applied to analyse large-scale experimental LH2 and GH2 releases. The simulation of non-uniform flammable cloud formation, resulting from a LH2 spill, reproduced a characteristic structure of the turbulent eddies and the direction of cloud propagation. The simulation results were found to depend on initial and boundary conditions. The air condensation-evaporation sub-model may improve predictive capabilities of the LES model

Conclusions Good agreement was achieved with experimental data on GH2 release in 20-m 3 closed vessel up to t=250 min after the 1 minute release. The LES results demonstrated that random flow field remains in the vessel long time after the injection and this is presumably responsible for H 2 transport. Further experiments with observation of velocity field after release and simulations with higher accuracy are required to give final answer to this question.