Learning Forum

[José Mantovani]

Guys, I have a little question. The more I have a more refined mesh with more cells, the slower / harder to obtain convergence. I did the tutorial available in simcafe called Steady Flow Past the Cylinder in which according to the steps I get a mesh with 18432 cells, I did it without problems and got convergence. At the end I decided to make a larger and more refined mesh with more divisional numbers and thus generating 160000 cells, practically 10x more cells than the simcafe tutorial. The viscosity is laminar and a Re = 20. In the same way in the solution process, I configured for 2000 iterations I got a Cd = 9 (desaccording to the experimental data, the Cd = 2), but the solution did not converge. And here I ask my question: how do I make the solution converge? Is it by trial and error? How can I think about it?

[Raef.Kobeissi]

Hello Jose It all depends on the mesh itself and particularly on the mesh near the cylinder wall. Have you used an inflation layer? What’s the orthogonal quality and aspect ratio? Whats the skewness etc... In the other hand you need to look at the type of mesh is it hexahedral or tetrahedral (the first one is convergence-friendly) and last but not least you need to be aware that some flow behaviour are transient in nature and you would find some difficulties solving them through a steady state model.

[José Mantovani]

Hello Raef.

First, congratulations for your videos, its great. I did with the parameters that give me Re = 20 and Re = 40 and I modeled it as stable, and for Re = 100 I modeled it as transient, but for Re=20 did not get convergence in 7000 iterations, I believe it's my mesh ... I'll check what you told me. Maybe, that do not need so many cells? And another question, with what you told me should I make hexahedral meshes to get convergence? I'm interested in validating my results, compare with experimental data. So if you could give me tips on how to proceed a mesh to get convergence would you use inflation in the region of the wall and look for greater orthogonality?

Raef if you can help me in another simple thing, whenever I import points from a file txt file of an airfoil I open in the modeler design, I create the surface everything normal but always the rope stays with 100m and not 1m. What can it be? I use units in meters, for all.

Thanks for the attention, have a g'day.