Aero Load goes wonky near fuselage

[Josh Goates]

I'm a relatively new user of OpenVSP (though I have lots of experience with other low-order aero analysis tools). I'm modeling an aircraft with a trapezoidal fuselage and a low, swept main wing. The lift distribution is smooth and consistent with what I'd expect, except for the panels closest to the fuselage where it "blows up". I believe this is a numerical computation issue, but I'm not sure what I can do to fix it (the things I've tried haven't yielded improvement). 
Any suggestions?

[Rob McDonald]

I actually think your issue here may be more caused by the proximity of the control surface to the body.  Try deleting the control surface (or turning the radio button 'off' for that surface).

The core problem here is that VSPAERO doesn't 'know' what the local chord is.  So, it does a local mesh search to find the chord and use that in the calculations.  When you have the end of a control surface, it gets a bit confused.

There are some changes being made in the solver that will change this -- hopefully it will help calm some of this sort of thing.

I believe the total forces and moments should be unaffected by this.

Rob

[Josh Goates]

I extended the control surfaces so that they go all the way to the root (before they were just a little offset from the fuselage. This definitely helps clean up the results, but still leaves an odd low-pressure zone right near the quarter-chord of the main wing. 

[Rob McDonald]

I'm not sure what is causing that hot spot.  I would probably try to improve the nose-to-tail resolution down the fuselage - particularly over the wing.  You don't need to match the wing resolution 1:1, but there will be pressure gradients along the side of the fuselage induced by the wing that you clearly are not capturing right now.

VSPAERO works with vortex loops -- i.e. polygons with an arbitrary number of sides.  They look like quads, but those side of body panels have ~12-14 edges.  Each of those edges has a vortex strength.  The hot spot overlaps the junction of the two fuse panels that overlap the wing, I don't think that is a coincidence.

If you can post a *.vsp3 model, someone can play with it, but that is where I would start.

Your chordwise clustering on the wing is OK, but you don't want any more clustering (smaller parameter value) -- and if you increase the chordwise resolution, you may actually want less.  This is because of the way OpenVSP does clustering....

We start with a uniform spaced mesh on the real surface.  Since leading edges are round, this results in a certain amount of clustering in the thin surface representation.  Then, clustering shrinks the panels at the LE or TE.  This may be a very reasonable paneling on the thick surface, but it can be too extreme when converted to the thin surface.  Think about the first panel, with a round LE (and a small panel), that panel will be very nearly vertical.  So, the compounding of these factors can bite  you.

You're fine at this resolution, but if you were to substantially increase the chordwise resolution, you would need to change LE clustering to be closer to 1.0.

Rob

[Josh Goates]

Thank you for the guidance! I'll mess with it some more