OpenVSP to Pointwise issue

[Jared S]

Hi,

I'm currently experiencing an issue exporting a CAD geometry and importing into Pointwise to generate a surface mesh. I've tried both IGES and STEP (untrimmed and trimmed), but there are issues with one of the components, the cowling (located at the ends of the wing) where there's a hard edge at the backend. When generating the mesh in Pointwise, this hard edge has no curve defined there (as seen below from OpenVSP in wireframe). The geometry I'm describing is the side-by-side UAM vehicle generated from one of the projects of the Aeronautics Systems Analysis Branch, NASA Langley. The OpenVSP file is provided from here: https://sacd.larc.nasa.gov/uam-refs/

OpenVSP:

Pointwise (Trimmed IGES):

Pointwise (Trimmed STEP):

 

When exporting either trimmed from OpenVSP and importing in Pointwise, with the default settings left as is, only IGES seems to work well with the exception of the cowlings at each side of the wing not rendering correctly even after modifying attributes to add additional triangles for rendering in Pointwise. Another issue with the IGES upon import is either upper or lower part of the wing doesn't render at all which I've fixed by mirroring. I've also tried changing the importing options in Pointwise with no success. When exporting an untrimmed STEP file from OpenVSP, the geometry looks fine, but the issue still lies in the hard edge at the backside (or tail) of the cowling, which creates a problem in not capturing the geometry when generating a surface mesh.

OpenVSP (Cowling):

Pointwise (Untrimmed STEP - Cowling surface and partial surface mesh):

I would prefer not having to trim any surface when exporting out of OpenVSP to reduce having to do it manually since I have four configurations (modifying OpenVSP file) of this geometry I need to generate surface meshes for. Any tips on what I can try to resolve this?

Thanks,

Jared

[Rob McDonald]

It would really be great to get Sienna, Chris, Brandon or someone else who had a hand in constructing this model chime in here.

The skinning on the nacelles was done in a rather unconventional manner.  All of the tangent strengths were set to zero -- but then curvature was used to create some shape on the left and right sides -- but not the top and bottom.

Part of me thinks that this is a stock nacelle shape that traces back to early days of ALPINE - so we may all be able to blame Travis and sit back smugly.

I suspect these surfaces -- with zero derivative, but finite 2nd derivative -- are what is giving Pointwise fits.

I took some liberties with the nacelles resulting in the attached model.  I turned 'off' all the skinning control for the nacelles except for the tip of the nose.  There, I turned off curvature control, going back to simple (symmetrical) angle and strength.  I used FitModel to tune these parameters to reasonably match the old shape.

The resulting nacelle has a curved spinner, but the remaining sections are straight-line lofted with truly sharp corners between sections.  The original model had some subtle curvature that came in to round those corners a bit -- which is why no feature line was drawn at what appeared to be 'hard' edges.

It would be great to have someone from NASA weigh in on the intent of this model to say whether the rounded corners were intentional and important for some reason.

From my perspective (not the person who built the model) -- these are early conceptual design models and there is not enough engineering knowledge about this system to have really subtle design intention in the surfaces.  So, from that standpoint, the straight-line loft is probably what was intended in the first place.  On the other hand, they may have some experience with a particular analysis tool (some mesher and CFD code probably) that does not like the sharp corners and they have devised this technique as a way to placate that tool.

If you run high-fidelity CFD on a model like this, you must realize that this nacelle is a notional nacelle meant to enclose a motor and possibly a gearbox.  The nacelle may or may not have been sized around a realistic motor and transmission.  It may or may not have been sized around some sort of cooling concept.  A CFD tool will see a lot of drag from the base end of the nacelle -- a real aircraft design would fair the aft end of the nacelle to reduce drag and remove a source of unsteadiness.  It is important to take a model like this in context -- when you move to high fidelity tools, you need to understand what the model builder really intended to be taken seriously vs. what was put in as a placeholder.  Or even if not a placeholder, perhaps it was sufficient to get some idea of frontal area, wetted area, and mass properties of a nacelle -- but it wasn't really meant to be fed into a Navier Stokes CFD tool.

I hope the attached file works better for you.  I would really appreciate it if you could test it out and report back here for both IGES and STEP -- untrimmed and trimmed.  For trimmed, please be sure to use 3.27.1, it includes some nice improvements to the trimming routines.  I know it isn't perfect -- but it is somewhat a miracle that it is working as well as it is.

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/2e8f6595-efa6-41fd-9271-8e1dfbc26945n%40googlegroups.com.

[C P]

Hi all,

Jared, i think  you don't need to trim all surfaces before meshing. Just  assemble quilts and models and then mesh. Taking Rob's model for a quick survey , the picture enclosed shows that you can have quick satisfactory results for preliminaty studies without too much work on meshing. I apply some values for domain definitiion but they were random values, just to show the results.

I imported the .igs model. Left wing had problem to be imported, but all the rest was ok. So I simply mirrored the right wing to complete the database

Hope this helps

Corrado

[C P]

I have to add some clarification , i see that my post was incomplete. When I said " you don't need to trim" I meant in Pointwise, if that was the task you made. The model was imported after using Trimmed Surfaces in  OpenVSP.

Corrado

[Jared S]

All,

Thank you very much Rob for regenerating the surfaces for the component that I had an issue with! Thanks Corrado. I was able to get a good watertight mesh in Pointwise for the nacelle similar to the images you attached. Unfortunately, I was too quick to think that was the only issue. This is regarding a gap in the split surfaces on the fuselage (also where the tail intersects the fuselage which is hard to see) upon importing a trimmed .igs file into Pointwise. The same gap shows up when importing in Rhino (v7) as well as in SolidWorks (2018 version), but it does not seem to show up when importing in FreeCAD (v0.19) unless I am not zooming close enough to see it. At first I thought it was a rendering issue, but when generating a surface mesh in Pointwise, that is not the case. I also tried importing the trimmed .stp file into Pointwise, but the fuselage split surfaces are b-line surfaces that overlap one another.

Pointwise (Trimmed IGES - gap in the aft section):

Pointwise (Surface mesh - gap in the aft section):

Rhino (Trimmed IGES -  gap in the aft section):

SolidWorks ( Trimmed IGES -  gap in the aft section):

FreeCAD (Trimmed IGES - no visible gap in the aft section):

I did however try the CFD mesh generation tool in OpenVSP and it generated a watertight mesh which I exported in .stl and imported it into Pointwise with no problems at all even at the aft section. I may end up going with this process if it works well with the solvers I am using for my CFD case studies (Helios with overflow and mstrand). I don't require a very fine mesh, but would have preferred to have it done in Pointwise since it is part of my workflow. I will try tinkering with some of the options upon import if the mesh from OpenVSP does not pan out.

Pointwise (OpenVSP generated CFD mesh):

-Jared

[Brandon Litherland]

For transparency, several of the SACD UAM Vehicle models are almost direct exports from the ALPINE process and haven't been "updated" for VSP best practices.  My understanding (before I was brought onto the project) is that the models were published to get people the geometry quickly as-is and expect that they would handle any alterations for structures, aero, mass prop, etc. on their own.  However, this doesn't help people that don't know what to look for.  One of my back-burner tasks is to update all of these models to effectively work "out of the box" to avoid these issues.  This takes some iteration with the RVLT folks who created the model because I could change a parameter to fix a VSP feature and alter some variation of the NDARC geometry.  First thing I always check = Zero Strength Skinning.

[Brandon Litherland]

Jared,

Let me know how this goes because I'm curious on how the toolchain to Helios w/ OVERFLOW and MSTRAND can be improved for future work.  

Are the gaps present on both the left and right sides of the tail?  Something to try is to use Trimmed Surfaces but only export a single component at a time.  If the fuselage comes out clean, something else is going on.

When I run into issues like this, and doing a single component export doesn't fix it, I adjust some of the features and see what is breaking the model.  If that doesn't work, I do some wireframe repair in CAD (Import Data Doctor in Creo) then export a solid version to gridding.

[C P]

Jared,

you can do all in OpenVSP. 

There is one fundamental step to do , that is, to increase the tesselation to a very fine level, before using the command Trimmed Surfaces.

I exported only iges as Surface-Shell Representation ( Step file exports a lot of geometries..).

Once imported in Pointwise it will be immediately watertight  and ready for meshing.

I have enclosed some picture to show the results of the procedure above. 

This is what I found, of course there will  be better alternatives, but it resulted to be a good time saving procedure the most of times.

Corrado

[Rob McDonald]

If you are using STEP or IGES export from OpenVSP, the on-screen tessellation has absolutely nothing to do with it.

Likewise, if you use CFDMesh to generate a surface mesh in OpenVSP, the on-screen tessellation has nothing to do with it.

If you use CompGeom to generate a surface mesh, then the on-screen tessellation dictates the resolution of the triangle mesh used by CompGeom.

Rob

To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/d6a78b6c-84e7-4e66-88fb-7565376ec32cn%40googlegroups.com.

[Rob McDonald]

I'm glad we made some progress.

Thanks for pointing out the continuing issues you have.

Please take the time to try the other export types.  I generally have the best luck with the Trimmed STEP file (when I'm using STEP files).

Please give as specific of feedback as you can for each of the export types. 

I don't have access to Pointwise, SolidWorks, Creo, or most of the other tools you're able to test the files with.

Rob

To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/6de1d45e-8d77-4c68-b5ef-ab319223d4f0n%40googlegroups.com.

[Jared S]

I really appreciate everyone chiming in.

Brandon,

Yes, the gap exists on both sides of the entire fuselage, but it is more visible at the aft region, and it also exists at the intersecting locations where the other components interface with the fuselage. I'll definitely post an update on how my cases go once it is up and running. The rotor blades in the simulation will use overflow and using mstrand for the rest of the configuration (fuselage, wing, hubs, etc.). Regarding the single component export after using trimmed surfaces, I'm doing something wrong since I don't see an option to export only a single component from all the components that were selected for trimming. I did attempt by setting each of sub-components (fuselage fairing and tail) to different sets (in the Geom Browser) including the parent (fuselage), but still no go.

Corrado,

Would you be able to share your vsp file used for the images you recently attached? I've increased the tessellation to a really fine level as well as increasing Num U for each of the cross sections especially at the fuselage, but the gap still appears for me after using trimmed surfaces, exporting a shell representation in .igs and then importing it into Pointwise. I'm using OpenVSP 3.27.1 and Pointwise 18.5 R2.

Rob,

I will continue to provide updates on my progress. I might try their other export types.

-Jared

[Rob McDonald]

On Thu, Apr 7, 2022 at 11:47 AM Jared S <jared....@gmail.com> wrote:

Corrado,

Would you be able to share your vsp file used for the images you recently attached? I've increased the tessellation to a really fine level as well as increasing Num U for each of the cross sections especially at the fuselage, but the gap still appears for me after using trimmed surfaces, exporting a shell representation in .igs and then importing it into Pointwise. I'm using OpenVSP 3.27.1 and Pointwise 18.5 R2.

WHEN EXPORTING TO STEP OR IGES, THE Num_U and Num_W TESSELLATION DOES NOT MATTER!!!!!

Rob

 

[Jared Sagaga]

Apologies on missing your previous comment Rob. I do see that tessellation has an effect when using CompGeom.

Kind regards,

Jared Delatorre Wakefield Sagaga, M.S.
Research Scientist | Science and Technology Corporation
U.S. Army Combat Capabilities Development Command Aviation & Missile Center | NASA Ames
Phone (Cell): (650) 438-1391
Email (Work-1): jareddelatorrew...@nasa.gov
Email (Work-2):
jared.d.s...@mail.mil
Email (UC): jdsa...@ucdavis.edu

--
You received this message because you are subscribed to a topic in the Google Groups "OpenVSP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openvsp/9UYWKozqstI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to openvsp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/CAEppYpFu-B5XZ8dC%3DapAeQwbEO6vNNKGinJnAz%2BBf%2B2hpAoGsw%40mail.gmail.com.

[Jared S]

All,

I just wanted to give an update. I was able to generate a watertight mesh with Pointwise (18.5 R2) using a trimmed surfaces IGES file exported from the vsp file that Rob provided earlier and I made no modifications except removing some components that I don't need for my cases. I exported an IGES file after trimming surfaces (with default settings) in OpenVSP (3.27.1). Imported it into Pointwise using default settings. The gap being there on both sides of the fuselage I think means that the trimmed surfaces model exported out of OpenVSP is not watertight. After assembling all the model databases in Pointwise with an edge tolerance of 0.01. I then generated a surface mesh and healed the gap by merging the connectors from each of edges of the surfaces that form the gap.

Pointwise (after import and before assembling the model databases):

Pointwise (assembled all model databases - red color indicates lamina boundaries or open edges (e.g. gaps) and teal indicates manifold boundaries):

Pointwise (after generating the surface mesh and deleted the domains of the right side for convenience in healing the gap):

Pointwise (side view of surface mesh before merging connectors to finish the healing process):

Pointwise (after merging the connectors to the heal the gap, the left side (purple) is mirrored to the right (green) resulting in a final watertight surface mesh):

Sorry for the long post, I wanted to make sure that I covered all the steps I took. I appreciate everyone's help on the issue. I have yet to test it in my simulation, which is the next step. The process I covered here is more hands-on then I would've liked, but I was able to get a watertight surface mesh.

-Jared

[Rob McDonald]

Instead of removing components, you should look in the use of Sets in OpenVSP.  I believe there is a video on them in the Ground School and there have been videos at past workshops.

Sets are groups of components.  A component can be in many (or no) sets.  When you do an analysis (or export) in OpenVSP, you choose which Set to use -- the default is 'Shown' and you can also use the 'All' set, but you can also define your own sets (and name them).

You might have one for mass properties, another for the OML, one for layout, etc.  Use them for whatever you want.

The gap down the side of the fuselage is certainly unexpected.  The OpenVSP geometry is watertight when OpenVSP has it, but something is happening in the export or import process.  It would be helpful if you would try the STEP output -- I generally have better luck with it.

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/c23f58fd-09ee-4b06-9e03-2d3b021a2fb8n%40googlegroups.com.

[Brandon Litherland]

Sorry folks, I'm just now opening the file and looking at it in detail.  I was using screenshots before and didn't see the problem.

The nose is made of a Cap rather than a point closeout as is best practice.  In most cases, using a cap isn't an issue but for CAD export and a few others the collapse of the left and right interpolated lines to a single point will (I think) lead to a discontinuity in the spline that doesn't resolve in export.  I'm going to recreate the fuselage with the proper nose geometry and see if that closes it up.  If you want to try it yourself in the meantime, just make a copy of the fuselage on top of itself and make the nose a point and skin to match.

[Brandon Litherland]

Here is a file with just the fuselages.  One is the old version and the other is updated to not use caps.  When I export either one of these as STP they both come into Creo as solid geometries (no gaps on the left and right) so now I'm wondering if this is really going to solve your problem.  Worth a shot.

I pulled both into Pointwise on our servers and can still see that double edge on the right and left... odd.  However, when I bring it into Creo as a solid, then re-export as a new STP from there, there aren't any gaps that I can tell.  So it seems that CAD likes the format just fine.  It's PW that's having a read issue with the format.  Out of curiosity, is there a reason you have to use STP/IGS format for this or can you export the VSP surface as TRI or P3D instead?  I assume that you want to already have the intersections so you can quickly create the unstructured grids.

[Jared S]

Rob,

I've tried with both the shell representation and BREP representation options in the trimmed surfaces feature, and unfortunately it's not importing correctly in Pointwise. I get overlapping surfaces on the fuselage. I also suspect that in the export and import process is where the model is breaking depending on the next software of choice after OpenVSP, and in this case, Pointwise. I think the way Pointwise is handling these file types might be the root of the problem. I did mention to the Pointwise support about this issue and their response was:

Please note that it seems that OpenVSP is not exporting this as a single watertight database model. It seems that surfaces are all exported separately and that is how they are coming into Pointwise when imported.

However, as you said the geometry is watertight when OpenVSP has it, and so the above statement is not what's actually happening during the export process.

Brandon,

Thank you for providing the file of the modified fuselage. As you've already mentioned, the gap still exists on the fuselage (for both .igs and .stp formats). Yes, I do prefer a IGS/STP file so that I can get to gridding from the get go. For my cases, I don't require a very fine or dense surface mesh, and so for that reason I'm utilizing the automatic surface mesh generator in Pointwise as opposed to manually doing it myself to quickly get one since I have four configurations of this geometry to generate meshes for.     

-Jared

[Rob McDonald]

The Shell representation is by definition not a single watertight BREP.  It is a bunch of separate trimmed surfaces.

The BREP should export a single watertight solid -- this is only available as a STEP file.

These files are notoriously difficult to get to work well.

They are much easier to write (good for OpenVSP) than they are to read.  Unfortunately, every side just tends to blame the other side.

True debugging of these files takes deep diving -- something vendors often can't do (because they buy their geometry engine from another vendor and do not have source access) or won't do -- because it takes a huge amount of developer time.

It isn't just a matter of time -- figuring out _why_ a particular file doesn't work is a true needle in a haystack problem.  If someone can identify a true root cause, we can usually figure out how to fix it.  Unfortunately, finding a true root cause is extremely rare.

Rob

Rob

To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/cfb8b2cd-2559-4ceb-97a5-f94514125f0bn%40googlegroups.com.