[Gearb0x]

Dear Foamers,

I am trying to validate a case of Hydrogen or Helium diffusion inside air. The current setup is considering Helium properties (although the phase is called Hydrogen in the case) assuming laminar flow.

Geometry is a simple cube but solved in 2D, so a square.
Pressure is uniform, velocity field is zero everywhere.
Only the Helium fraction is set to 1.0 in the upper part of the Mesh. Setfields utility is used to set Helium fraction to 1.0 in a defined box.

Around 400 seconds, it seems an instability is appearing (at least, i don't expect a flow driven mainly by pure diffusion to behave like that) but I don't know enough in CFD to understand exactly what is going on.

Could someone maybe give me a hint on what is happening ?

[Gearb0x]

Ok the issue was apparentely the timeStep (deltaT) in controlDict being too large.

The residuals in the log file seemed ok but analyzing them through foamLog log utility and then plotting revealed an issue. Decreasing the timeStep solved the issue. Now to find how to find a way determine the timestep optimally because Courant number seemed ok.

[windscion]

Well starting at rest the courant number isn't gonna mean much.
Did you estimate the Brunt–Väisälä buoyancy frequency? That should be more helpful here.