CFD modelling of accidental hydrogen release from pipelines'

1
CFD modelling of accidental hydrogen release from
pipelines.
H. Wilkening - D. Baraldi

• Motivation
• 2-D Simulations for methane and hydrogen
• 3-D Simulation for methane and hydrogen
• Conclusion

Int. Conf. on Hydrogen Safety Pisa Sept. 8-10,
2005
2
Pipeline disaster in Belgium in 2004
Left picture is a view of the industrial area
destroyed by the explosion and the subsequent
fire. The right picture shows the fire caused by
the natural gas released from the pipeline after
being ignited by the explosion.
3
Modelling of light methane dispersion (1/2)
4
Modelling of methane gas dispersion (2/2)
5
Comparison to hydrogen
6
Comparison of thermal energy released between
methane and hydrogen
The thermal energy released is very similar for
methane and hydrogen until the first gas is
leaving the computational domain, the wind keeps
the gas within the computational domain longer.
7
Flammability limits in methane gas dispersion
Methane molar concentrations for wind (top) and
no-wind (bottom) case both 7.5 s after the
high-pressure release starts. Shown are only
those concentrations within the flammability
limit. These are the values between 5.3 and 15
all other concentrations are set to 0 .
8
Flammability limits in hydrogen gas dispersion
Hydrogen molar concentrations for wind (top) and
no-wind (bottom) case. Shown are only those
concentrations within the flammability limit.
These are the values between 4 and 74 all
other concentrations are set to 0 .
9
Comparison of thermal energy released within the
flammability limits
The thermal energy of the released flammable
gases is very different for methane and hydrogen
due to the wider flammability limits of hydrogen
(4 74 vol. conc.) compare to methane (5.3
15).
10
Numerical Setup 3-D Simulation
Grid discritisation and domain decomposition for
parallel computing Total grid with 3.5 mio. cells
run on 16 CPUs or 8 CPUs
11
3-D methane release and dispersion simulation
4 molar conc. Isosurface 10 m/s wind
121 kg/s methane release rate corresponds to 6 GW
thermal power
12
Flammability limits of methane within the release
About 10 to 30 of the released methane within
the flammability limits
13
3-D hydrogen release and dispersion simulation
4 molar conc. Isosurface 10 m/s wind
42.2 kg/s hydrogen release rate corresponds to 5
GW thermal power
14
Flammability limits of hydrogen within the release
About 70 of the released hydrogen within the
flammability limits
15
Comparison of thermal energy released between
methane and hydrogen for 3-D
The thermal energy of the released gases is
similar for methane and hydrogen also in 3-D but
again differs quite a lot for the gases within
the flammability limits.
16
Conclusions

• Accidental releases from a methane and hydrogen
  pipeline have been modelled and compared for a
  scenario with and without wind.
• Although the amount of total released energy is
  similar for methane and hydrogen, the amount of
  flammable hydrogen is larger than methane in all
  cases due to the wider flammability limits of
  hydrogen.
• Due to the larger density of methane, methane
  might be more easily accumulated close to the
  ground under certain conditions (wind and
  geometric configuration) than hydrogen.

17
Thank You
18
Governing Equations for Fluid Flow
19
Gas Dispersion in HyJet Jx7
The HyJet experiments will be used to validate a
CFD-Code for dispersion modelling. The
experiments were performed in the
Battelle-Model-Containment. Within the
experiments Helium is release through a nozzle
in one of the lower so called banana rooms.
After the release a Helium stratified atmosphere
is developed in the upper part of the
containment. In HyJet Jx7 Helium is released
through a 9.5 cm nozzle at a speed of 42 m/s for
200 s.
Iso surface of 15 Vol. He colored with turbulent
intensity at 200 s
20
HyJet Jx7 comparison with experiment
Axial profile through jet axis 190 s after the
beginning of injection
21
New High Performance Computing Hardware
The CURIE High Performance Computing Cluster is
installed with a total of 50 CPUs. The System
has a performance of 128 GFLOPS (billion floating
point operations per second) being by fare the
most powerful computing system in the JRC. The
system is used for hydrogen safety studies, for
safety studies of new innovative nuclear reactors
and for modelling of material failure on very
small scale (crystal modelling) and other
applications demanding high performance computing
power.
22
HPC load on CURIE last month (2/2)
Performance of the cluster via web interface
(individual compute nodes)
23
HPC load on CURIE last month (1/2)
Performance of the cluster via web interface