Learning Forum

[Dominic de'Ath]

Hi all,

I am trying to use fluent to model gaseous diffusion but I can't get the models to replicate what I expect to see from a hand calc using Fick's law. 

It's looking at hydrogen diffusing in air, and whilst I recognise in most instances buoyancy will result in greater dispersion than diffusion alone, we often have cases where we need the solution to be independent of orientation hence like to rely on diffusion where we can.

The case I am trying to test is fairly simple, a hydrogen generation source (25 ml/h, 0.000025 m3/h) at one end of a pipe, which is nominally 1 m long and 50 mm diameter. The other end of the pipe is open to free air.

The analytical solution using a diffusivity of 0.263 m2/h indicates the concentration difference along the length should be ~4.8%, but no matter what I do with the CFD model (e.g. more refined mesh, running transiently etc), I cannot get close to this solution (the concentration gradient results as 20% or higher). 

Does anyone have any examples of where diffusion has been effectively modelled with the result being close to an analytical solution? Is there a validation case for how diffusion has been incorporated into the fluent code?

Thanks in advance!

[Prashanth]

Hello: There is a multicomponent species transport pipe flow verification case and one more case in Fluent help page. You can check it out.

VMFL006 Multicomponent Species Transport in Pipe Flow

VMFL042 Turbulent Mixing of Two Streams with Different Densities

[Dominic de'Ath]

Hi Prashanth,

Thanks for those links, I wasnt aware of all of those cases but unfortunately none of them demonstrate the phenomena I am looking for. I think in both of the cases you linked the main driver for mixing is from the turbulence generated by the flowing fluids. 

The case I am considering is trying to eliminate all impacts from turbulent flow to demonstrate the impacts of only Fickian diffusion (Fick's laws of diffusion - Wikipedia). 

As described above, there is a simple analytical solution for simple systems but I was hoping i would be able to use this to validate to use the model for more complex geometries.