Discussion Forum

Maybe this is stupid: Have you chosen "Imcompressible Fluid" from the menu?

Differences in the outflow and inflow are probably due to inaccuracies. Meshing has a large effect on that.

BR
Lasse

Thanks for the advice. I do have incompressible flow selected. Increasing the mesh size to finer (the highest my memory constraints will allow) did not improve the flow difference either. Any other suggestions such as tolerance or time stepping settings to minimize inaccuracies?

I can't offer much help, but can confirm that it is a common problem (in my experience). I've had some help requests lodged with COMSOL support. Some of the advices have been:
- finer or better mesh (does seem to help - can you segregate the mesh to give you more detail where you eed it most?)
- integrating over inlet or outlet planes can be inacurate especially for curved wall geometries. Also something about a mesh distortion at a sudden expansion, meaning they suggested integrating over a cut plane a little back from the tube end.
- One suggestion I haven't tried (as it sounds like treating the symptom not the disease) was something about using the integrated flow rates to update the solution.
- I have read that the default tolerances can be rather coarse, so have you tried lowering the tolerance in the solver?
- Where I was struggling with mass balances with high velocities over small areas it helped to add an expansion chamber, and integrate a lower velocity over a larger area.
- Some problems (mostly with convergence issues) were helped with a parametric sweep or a step function to ramp up variables. But your pressure is sinusoidal - could you force it to start from the midvalue?

Thanks so much Ross! I was able to solve the problem by not integrating over curved surfaces and a little in from the tube outlets while increasing tolerances.