Import of mesh coordinates from OpenAeroStruct

[Tim Schuhmann]

Hi,

is there someone with experience on importing a mesh generated by OpenAeroStruct to OpenVSP? I have created the attached files within OpenAeroStruct, one is a .dat file for tecplot and the other is an ffd file. Unfortunately I can't import these file formats directly into OpenVSP. 

Having a look at the exported .dat format from OpenVSP and comparing it to the exported .dat file from OpenAeroStruct also didn't help me in understanding the differences in the file structure.

Best regards,

Tim

[Rob McDonald]

OpenVSP is not set up to import any of those file types.

The easiest path is probably to figure out how to write the mesh you're working with to a STL file and then to import that.  Of course such a mesh will not have any of the parameterization that a native OpenVSP file would have.

What are you trying to accomplish?

For the most part, OpenVSP wants to be the starting point for the geometry in a design process.

Rob

[Tim S.]

Hello Rob,

I want to utilize OpenAeroStruct to run a wing planform optimization. After the optimization is complete I would like to reimport the generated wing for further analysis in OpenVSP. It is possible to import the wing from OpenVSP to OpenAeroStruct, so I was suspecting it would also work in the other direction and maybe someone in the forum has already done this. I am pretty sure the developers of OpenAeroStruct will also implement this. In the mean time I will try to find a workaround.

Best regards,

Tim

[Brandon Litherland]

To come back from AeroStruct, as Rob said you'll have to either import a TRI or STL and then deform the wing using Fit Model to match up.  Depending on the complexity of the deformation e.g., pure bending, bending + torsion, bending + torsion + airfoil flex..., you could interpret the resulting deformation and then script the changes to the original OpenVSP model.  For example, if you have the LE/TE coordinates matched to sections in your OpenVSP wing, you can calculate the sweep, dihedral, and twist that would approximate the deformation reasonably well depending on the number of sections.

[Rob McDonald]

I am not sure what the OpenAeroStruct folks have planned.

If the OML is deformed through a FFD technique, then those deformations will most likely not exactly representable by OpenVSP's native parameters.  I.e. it may be impossible to recover the deformed shape in OpenVSP.

I've often thought about adding some sort of native FFD capability to OpenVSP.  It would be an interesting project if someone wanted to take it on.

Rob

[Tim S.]

Hello Rob,

for now I will use the aerodynamic planform optimization results from OpenAeroStruct and simply copy the values over to OpenVSP, for simple geometric variations like the sweep this is quite easy. It gets a little more difficult for exact twist distributions and chord length distributions, as the splines (used for setting a twist at a certain position) in OpenAeroStruct are split in equal distance. However writing a simple script to input these parameters will also solve this. 

I want to thank you for improving the FEA mesh output to calculix. This works very well if the mesh is fine enough. When the mesh is too coarse the ribs will still sometimes stick out of the surface but that can be controlled by the mesh refinement, so it doesnt pose much of a problem. Now I just need to export the pressure distributions across the wing after running VSPAERO to perform an aeroelastic analysis using calculix. 

Best regrds,

Tim

[Rob McDonald]

Are you working on this stuff solo?  Or are you collaborating with anyone -- specifically anyone at NASA?  There are some folks at NASA working through similar issues - but with somewhat different tools.  I'm not entirely sure of the scope of their work or of all of the players -- hence this question.

Dave Kinney (VSPAERO author) has developed a tool that will map pressure loads from VSPAERO to CalculiX - and deflections back.  I have not had time yet to get it working, but that is on my plate for the next few months.

Can you provide me with a test case where the mesh size determines whether the rib sticks outside of the surface or not?  I think I have an idea what is going on, but a clear test case would be a great way for me to track that down and fix it.

Let me know what works and what doesn't - now is the time to get structures stuff fixed and improved.

I haven't had time to document all of the improvements that went into 3.28.0.  If you have any questions on how things work (Trim components, selecting material orientation, etc) let me know and I'll do my best.

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/7c844ce5-afa4-4f62-96e1-f4fe3035d693n%40googlegroups.com.

[Tim S.]

I am creating a multifidelity toolchain for preliminary EVTOL design for my master thesis. So its currently just me working on this. I have not had contact with anyone at NASA regarding this topic. The only contact I had was with one of the developers of OpenAeroStruct on GitHub. Do you know with which Tools the NASA collegues are working with? 

The presentation of David Kinney regarding the mapping of the loads looks very useful for my thesis. Did I understand it right, that currently it is not implemented or is there a way to use the steps he provides for mapping the pressure loads into Calculix (in the attached presentation). The optimization he is presenting in the presentation could also replace OpenAeroStruct or could provide an alternative workflow. Would it be possible to also get in contact with David Kinney if you know him and he is fine with it?

Regarding the FEA mesh, the threshold for the ribs to get trimmed is a min edge len of 0.012, above this the ribs start to appear outside the skin (see attached mesh settings). This only happens on the top and sometimes bottom surface, not at the LE or TE. While running a half mesh I noticed an issue with the spar (attached file). Is this normal and would you say its better to always mesh the full wing?

Best regards,

Tim

[Rob McDonald]

I don't know much about the work going on at NASA.  I actually think there are at least two somewhat parallel efforts.

The code to map the aero loads to CalculiX and the deflections back to VSPAERO has not yet been released.  I hope that it will be available in the next few months.

I had not seen either of these problems with the FEA Mesh before -- it would be great if you could email me an example file with the settings to recreate that issue.

The same is true for the half mesh issue -- half meshing should work, but if you're seeing problems, I would appreciate having a clear reproducible test case.

Rob

[Tim S.]

Great, in the mean time I will try to work on a solution for mapping the pressure loads myself.

Sure I can send you the file, do you want to send me your e-mail address to my e-mail (I am not sure if you are able to see my e-mail address)? Otherwise I can also post a generic example here on the forum, where this error also occurs.

Best regards,

Tim