[pauleck]

Dear all,
I have to simulate a quite "simple" 3d laminar flow through a device with multiple flow segments. My main problem is the mesh generation, whatever I change in the meshing (pure hex, pure tet, combination of hex+tet) the resudials in fluent does not behave probably. In some cases I get severe fluctuations, in some other, the resudial stucks after a certain number of iterations.
Back to the meshing, it is really pain in the a** to modify the flow segments like I wish. Whenever I make some local edits (refining some critical areas / inflations etc...), it affects the whole device, resulting in frequently appearing errors. I am restricted to 900k cell meshing due to temporary hardware limitations. Please have a look at the attached pictures representing 3 different meshings and the impact on the convergence behaviour. The meshing is probably not really efficient, especially in the tiny flow areas but I really was not able to refine these spots (ended with alot of errors).

Fluent: 3D, laminar (viscous), pb, steady vs. transient, pbns

in Design Modeler, The fluid domain was splitted in different bodies (then grouped to one part), so the mesh connection should be fine here.

I would really appreciate your feedback, dealing with this problem for 5 full-days now, and it is just frustrating, although I am new to cfd

CASE 1



CASE 2




CASE 3