Few questions on Panel Method

[Ashley M]

Hi Rob, Hi all,

I have been exploring the Panel method on OpenVSP 3.27.1. I have a few questions based on my observations.

1) I noticed that every time I rerun a panel analysis after reopening an existing vsp3 file (with an existing meshgeom ), it crashes. Is it a requirement to delete the existing meshgeom before running a new panel analysis? 

2) The load distribution outputted in the results manager is not a smooth curve (image attached).  Is there any way to mitigate this?

3)The Cp values outputted in the FEM2D file are in the order of magnitude 10^-3, which seems very low (image attached). What could be the possible reason? I need a reliable Cp distribution over each panel to run structural analysis.

Also, is there a source talking in detail about the panel method used in OpenVSP?, if available may I have the link to it.

(I have attached the vsp3 file for reference)

Thank you very much

[Rob McDonald]

Yes, you should delete the MeshGeom -- when the GUI is operating, it keeps track of the last MeshGeom and deletes it before running a second time.  After you do a start/stop cycle, it forgets that information and ends up creating a MeshGeom that is on top of the existing one -- which causes the crash.

In general, you probably don't want to save MeshGeom's to the *.vsp3 file.  They are very large and can be easily re-created next time.  Every time you update a parameter, they are out of date, so I make a habit of deleting them unless I have a specific need.

The spanwise load distribution for thick-surface (Panel) method is a very recent addition and is still getting debugged a bit.  It is likely smoother if you use the 'experimental file format' for Panel method.  Or, if you use VLM instead, it will be much better.

The case with the very low Cp is a case at zero alpha.  If it is a VLM case, then the very low delta_Cp's don't surprise me at all.  Otherwise, I would need to dig into that case.

Rob

--
You received this message because you are subscribed to the Google Groups "OpenVSP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openvsp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/5c2dc023-7ffd-449b-b097-751671c09c86n%40googlegroups.com.

[Ashley M]

Hi Rob,

Thank you very much for your input.

The spanwise load distribution data was generated with the 'experimental file format' and yet the loads did not have a smooth distribution.

The Cp values even at 10 deg alpha are very low (image attached). I would kindly request you look into it. 

Another observation is that the CDi values are negative for some alpha (image attached). What could be the possible reason for this?

I have attached the vsp3 file here again.

Thank you

[Brandon Litherland]

I've been looking into this and it seems the new pure Kutta-Joukowski theorem application in newer versions of VSPAERO require finer spacing of the grids.  I've run a case with NumW of 201 on this model with some improvements to the clustering at the wingtip and LE/TE and the CDtrefftz and CDi (KJ) are performing as expected with matched minimum drag and profiles at low magnitude alpha with KJ predicting a larger drag increase at higher angles.  Give this a try and let us know if that improves your results.

- Brandon