Problems with CFD Mesh after creating Control Surface

[Avijeet Acharya]

Hello,

I am trying to create a wing with a control surface, such as a flap deflected. The idea is to create a CFD mesh and export it for a downstream volume mesher like Gmsh. I followed the video: https://youtu.be/zOoXH5xoZXo?si=F9YevhXMFjL0fD3y&t=580 as suggested by Zach Hazen in this thread: https://groups.google.com/g/openvsp/c/PyZIPWIOcaQ.
I was even successful in automating the process through the API.
This method requires creating thin slivers just before and after the control surface on the main wing, to handle the transition between the trimmed airfoil and the adjacent untrimmed airfoil.
But the problem I am facing now is when I try to mesh this geometry, the tiny slivers have poor quality trias all over (sometimes even intersecting trias). I guess this is because for the slivers, one curve is having full set of points and the other one has a lower number due to trimming. This leads to either a channel like profile or a convex bump on the slivers. Even the tutorial video mentioned above seems to have the same problem (Around 14:53 timestamp).
Any idea how do I workaround this bottleneck? Appreciate any help in this regard.

Images and the model are attached.
Mainly working on version 3.36.0 of OpenVSP.

Regards

Avijeet Acharya

[Rob McDonald]

3.36.0 is ancient.  You should update to a recent version.

I am amazed that this works as well as it does.

I believe the other user who suggested this path was taking the geometry to CAD, fixing any issues there, and proceeding.

OpenVSP is being forced to do some weird things to loft the surfaces through that rapid transition.  The mesher is then doing the best it can with an ugly surface.

I don't think you are going to have a great deal of success with this.

However, your best bet will be to break the main wing into three segments -- one before the flap, one traversing the flap, one outboard of the flap.

This will fix the problems with the abrupt transition between sections, but it will introduce its own problems with trying to make tiny perfect intersections between the wing sections.  I'm not sure what you'll be able to do there -- particularly in an OpenVSP only workflow.  This is another approach that may work best with CAD post processing involved.

Rob

[Avijeet Acharya]

Thank you for replying.

I'm forced to work with 3.36 as 3.48.2 crashes when Farfield and Symmetry (Half Mesh) are turned on because I need them.

As for your suggestion regarding the three-segmented wing, that is exactly what I tried too. But they failed because of exactly what you predicted. There were gaps in the mesh at the intersection of the inboard and midboard sections and midboard and outboard wing in some places, even though they were at the same y location. Where there were no gaps, there was no node connectivity.

But I'm happy to inform that I got it working within OpenVSP only workflow. I'm listing it here so that it may help someone in the future.

I created the flap as I was doing before, following the YouTube video cited above. But I didn't create TE trims in the main wing.
Instead, I created a secondary wing with span exactly the same as the flap (slightly higher to allow free rotation of the flap), a bigger chord than the flap and with higher t/c so that it may fully consume the flap, and placed it where the flap was. Then I turned on negative volume on this secondary wing. This gave me exactly what I wanted.
I meshed it in two rounds. First, I meshed the  Wing+Negative Volume with Farfield/HalfBody, and then meshed the flap separately.

Images attached for reference.

Regards

Avijeet Acharya

[Rob McDonald]

Negative volumes is a great way to tackle this problem -- I should have thought of that.

Instead of using an additional component, you might try creating a conformal component based on the flap with a negative offset.

I just tried it, it doesn't quite work.  It does not expand the span of the conformal, just around the foil.  So it isn't a perfect solution --  but maybe the idea will be useful to you.

Rob

[Avijeet Acharya]

Thanks for that idea. I'll see if I can use conformal components for this purpose.

Regards

Avijeet