Diverging results when having wing-fuselage intersection

[Leonardo Ungureanu]

Hello, recently I've been trying to create a model on OpenVSP. The model is simple, composed only of a wing and a fuselage. The two components taken one at the time provide good results. When I try to merge them, I get diverged results in lift distribution. I tried to use a simplified fuselage, both with VLM and panel method. I tried to refine, to make the mesh coarser, to modify the geometry in multiple ways. The result is always the same.

Here I attach the photo of the geometry, of an example of divergent result, and the file itself. The file contains more objects,  but I've been simulating the fuselage and the wing only.

Thank you for your help,

Best regards,

L. Ungureanu

[Rob McDonald]

Your screenshots did not come through.

You don't say what version of OpenVSP you're trying.  You mention VLM and Panel modes -- those are no longer labels in OpenVSP, so I assume you are using an old version.  Please make sure you are running on 3.46.0.

Your file (as attached) is not ready to run the analysis.  Your analysis was configured to use the 'Shown' Set, but no objects were shown in the saved file.  Your model includes many components -- including a second wing and eight rotors that probably aren't included in a simple analysis.

Posting an example file that is not 'ready to go' makes it difficult for anyone to help you.  We have to guess at what you are even doing.  Make it so I can exactly replicate your situation without doing any extra steps.

Your lifting surfaces all use file airfoils that appear to have blunt trailing edges -- with 'Sharp' end caps applied.  This might work, but it might also give problems.  I suggest you start working with built-in sharp TE airfoils and work up your complexity from there.

Playing around, I deleted the actuator disks, extra wing, and tails, and changed your wing airfoils to built-in NACA 4-digit -- and I was able to get an All-Thick solution to run.

The #1 advice that I say here again and again -- start simple, gain some success and experience, build up from there.

Rob

[Leonardo Ungureanu]

Hello, thank you for your answer. I'm using the 3.46.0 version, sorry for not making that clear. With VLM and panels I was referring to thin and thick solutions. Here I attach the clean file, with only the wing and the fuselage in it. In this file 4-digit airfoils are used. The problem persists, and I hope it will be clearer this time. I tried various combinations recently, with simple and complex airfoils/fuselage geometries. Nothing seems to work when it comes to having the fuselage and the wing intersecting. The solutions indeed run, but the results are full of spikes with results' values varying within different order of magnitude. Using the thin set, the solution is also wildly asymmetric (even though the geometry is symmetric). The case I submitted is only an example. I tried to increase/reduce the number of panels in different ways, I tried to make a better use of tip/root/TE/LE clustering. I also tried to move the wing and the fuselage in multiple positions. I tried also to start with a very simple fuselage which was simply a cilinder. The problem is always the same. I hope you can help me.

Thank you again,

Best Regards,

L. Ungureanu

[Rob McDonald]

The spike in load distribution at the root is an artifact of bookkeeping, not a reflection on the quality of the solution.

Your spanwise resolution is probably excessive.  Notice how short a chordwise run the root panel has.  Your load distribution there is divided by a very small chord.

Decrease the number of spanwise panels, see how it does.

Rob

[Leonardo Ungureanu]

I see, thank you. I'll try.

Best Regards, 

L. Ungureanu