I'm relatively new to Fluent. I've been modeling simple diffusion from a liquid droplet in water. For zero flow rate (or low, 1e-9m/s), I expect the species transport equation to yield similar diffusion to known theory.

What I've found is that my 2D model works fairly well (within a factor of 2), but the 3D model does not work well. 3D shows much less diffusion than theory, possibly by a factor of 10 or 100. These observations are based on an axial plot of species mass fraction as a function of distance from the droplet.

I started off with similar models for 2D and 3D. Same droplet size, similar meshing. But then I cut down the size of my control volume for 3D so it would run faster. Any reason the 2D & 3D species diffusion would differ so much?

Remember that an axi-symewric model has essentially an infinitely small mesh in the axial direction- so using the "same" mesh size when going to 3-D will drastically reduce accuracy. Fluent is not that good at diffusion (it's made for inclusion of convection, and that's where it is optimized, especially for higher Reynolds and Peclet numbers). Try to keep refining your grid and things usually will get better for 3-D diffusion.

Thanks for the tip Peter. I see your point about refining the 3D mesh in order to more closely approximate the "infinitely small" mesh of the 2D-axisymmetric model (normal to that axis).

I've already "adapted" my 3D mesh once. Perhaps I'll try that procedure again to further refine it near the droplet where the majority of mass fraction drops off.

When you speak of "higher" Reynolds and Peclet numbers, what order of magnitude does that imply? My early models are running near zero velocity. The plan is to raise the velocity in small steps until Pe=100 is attained. Is that still fairly low from your point of view?

In terms of the Pe, as long as the velocity and length scales are correct then a Pe of 100 is sufficiently large that it should result in considerable convection effects.