mesh refinement worse results
 

Hi, first I will ask you for some details:
1) What kind of mesh are you using?
2) What kind of refinements are you doing?
3) What turbulence models are you using?
4) Is it a steady or unsteady simulation?
5) Your residuals converges (goes down enough) in all the meshes you are using?
6) Did you construct a BL mesh for you tube?
Regards
PD: Regards=Saludos :)

Lots of questions!! :P

I'm using the axisymmetric model, so I have to mesh a simple rectangle (only quadrilateral mesh elements).
I'm simulating steady and laminar flow and I'm utilizing an UDF at the inlet (fully developed flow) so the refinements are made only in the "y" direction (is that correct?)

I used a size function in the y axis.

Saludos!=regards :D

Be careful with the quality of the mesh. Try to keep the aspect ratio between razonable values (under 1000 or less). I don´t know wich is the 'y' direction in your model, may be if you post an image of your mesh, I will realize that. One sugestion, if you refine your mesh in the radial direction, you have to refine in the angular an axial direstions too.

This is the simplest mesh I'm working with, I still don't know what I'm doing wrong :(

Mmm, it's strage. May be you could do some contour plots of temperature and/or velocity of the solution in the different meshes you're using. May be you can find somthing strange in the behavior of the the flow that is causing these errors in your simulations. But one thing that I know, is that if you refine the mesh, given certain BC's, you will resolve structures of the flow that you didn't before with your coarse meshes and you will tend to the exact solution for such BC's. Another thing that I sugest you is to plot the Nusselt number vs the size of the mesh an see if you get a monotonic behavior or not. Regards
Gonzalo