All the simulations
done with the Discrete Phase Model let us see the moving trends of the
particles in a bubble column. Indeed, we notice that when the inlet streams
are at the bottom of the column, the bubbles go preferentially near the
walls, whereas when the inlet streams are at the top, they go preferentially
to the middle of the column. In micro gravity, there is a homogeneous presence
rate of the bubbles in the column. Nevertheless, the calculations done
give us only trends since we did not have the time to run the simulations
for several seconds.

To diminish
the calculation time, we first increased the time step. But with the LES
model and the particle tracking, an important time step let Fluent loose
lots of bubbles. So we tried to use as the model of turbulence the k-epsilon
model, but it gave bad results:

To complete
this study, it would have been interesting first, to do the calculations
with a more precise mesh, especially near the walls. Besides, if we had
more time, we could have run the calculations for a longer time and let
the hydrodynamic phenomena of the column be stabilized.