Mesh Convergence on a Thin-Thick simulation

[Julien]

Hello everyone,

I wanted to simulate a plane with the fuselage as a thick set and the wing as a thin set but when varying the mesh size I cannot get convergence on the CDi coefficient or CLtot. I think this is because I am using a thin body intersecting a thick body and as warned in the 2025 workshop the intersection between the two can cause problems but then does that mean we cannot do a mesh convergence study with a wing intersecting the fuselage and thick-thin sets ? You can find the geometry and the excel file attached 

Also, when I disable the X-Z symmetry the results go crazy, is this expected?

Thanks in advance!

[Rob McDonald]

Your mesh resolution in the supplied model is a bit of a hot mess.

Here is something that I would treat as a baseline case -- although, I would probably reduce LE/TE clustering substantially.

Start from here and see how things go.

Rob

[Julien]

Thank you for your answer and proposing a baseline case! But I am not sure to understand why your baseline (how would I choose such a baseline on another baseline) would be better than the default one I used or the second one I mentioned in the excel file and as illustrated below:

And I also don't understand why the results change even between very fine mesh configurations...

Thank you in advance!

[Rob McDonald]

You will need to develop some intuition and experience around mesh resolution requirements.

I provided the case to give you an idea of what a reasonable mesh 'looks like'.

Your original mesh had too much resolution in some respects (around the fuselage and down the straight barrel section of the fuse) -- and too little resolution in other respects (along the nose/tail of the fuselage).

The meshing operations in OpenVSP are not perfectly robust.  Sometimes they will crash or give results that result in poor VSPAERO solutions.  Usually small adjustments to the geometry or resolution can avoid those cases.

Rob

[Julien]

Good morning,

I have rerun a mesh convergence and kept every configurations with a rather uniform mesh instead of only changing the mesh near the wing and starting with a configuration close to the one you suggested:   

And here you can see all the mesh configurations I used (I don't know if I was clear but the file I provided  in the previous question, was with the finest mesh of course I haven't started with this mesh):

However I still think that CLi, CLo, CDi are not converging as you can see it in the excel file or for example with this plot on the relative error between two consecutive mesh configurations abs(CDi_Mesh_1 -  CDi_Mesh_2 ) / CDi_Mesh_2 with CDi_Mesh_1 a coarser mesh.

 

You can see all the results in the excel file and the vsp file of the initial mesh. This is why I am asking if convergence can be achieved with thin-thick intersecting sets. 

[Brandon Litherland]

You're starting much too fine for your initial case to see any convergence.  Start with something around 17 num W on the wing with 5 spanwise and increase from there.