[piprus]

Hi,

I do perform a LES computation with ANSYS 19.0.0. At the moment I'd like to compare real cell dimensions (call them deltas) against mean Kolmogorov Length Scale (eta), to check if I'm able to catch the physics with my mesh resolution.

Creating eta is quite easy using Custom Filed Functions, the problem are deltas. Delta supposes to be a distance between two points storing a physical property (in a cell centroid - cell based approach), but since the Growth Ratio do not exceed 1.05, I simply assume delta to be a side of my cell. As my geometry consists of hexahedral-type elements only, the easiest way to get average delta of each cell is delta = volume^(1/3).

Now comes the tricky part. At first, the elements are varying in sizes as well as in Aspect Ratios. Secondly, the domain resembles cylindrical shape (let's say a pipe), so more than half of elements are quadrangular prisms, the rest are cuboids. Finally, due to cylindrical shape it would be nice to get deltas in cylindrical coordinates.

I believe the tricky part with obvious difficulties can be coded in UDF, but I do have a more general question. Since Kolmogorov Length Scale has no orientation (I'd call it isotropic) then is there a sens to compare it with a real cell size instead of universal approach, i.e. delta = volume^(1/3)? Or am I wrong? What's your feeling/opinion/expertise?