Post deleted? shell or no shell, solid not solid

[Anonymous]

Did I post in the wrong place?  or did i say something wrong or do something wrong? 

I have a shape imported from iges file, I have taken the surfaces and used patch and stitch to make a solid body. Once the last patch was stitched, it magically turned into a body.... at least it looks like a solid body... but it doesn't behave completely like a body. The faces have no plane, there is no extrude or other options available. I can't even so a section analysis because none of the faces are selectable. I can loft between 2 of these "bodies" but cannot join the result to them. It I loft and create a new body, it wont combine to these "bodies".

All I want to do is join a few bodies, and shell them to 0.5mm. Shell works... at least at some thicknesses it does, but not the shell thickness I want. I can shell it to 0.1mm but nothing else. I tried scaling up and down but the shellable thickness changes too... I read that it can help to split a body to shell the halves an recombine... but I can't get a plane thru the middle because.. the "body " faces have no planes... aaaaargh

I managed to apply a plane tangent to face at point and sliced off the face I am working with, now extrude is available on the bit of face but shell behaves the same... I split the body but shell still won't let me use anything except 0.1mm thickness...
Am I wasting my time? am I just asking too much of F360?  Please help
wing.png

[Anonymous]

I scaled the object down and can shell at 0.04mm or 1.2mm thickness, but can't find anything in between, and definately nothing close to the 0.5mm I need to achieve.

Any ideas?

[jeff_strater]

I think we'll need to see the design to be able to say for sure.  The fact that the model has no planar faces, to me, implies that the input IGES file also has no planar faces.  The success of a shell will depend a lot on the geometry.  Shell is a precise operation, and if the input data has areas of high curvature or other problems, yes, the shell can fail.

Anyway, we can see for sure if you share your design here.

Jeff

[JDMather]

Can you File>Export the *.f3d file to your local drive and then Attach it here?

It would probably be good to have the *.igs file too.

If you can't post here - I am sure Jeff can take a NDA look at the file.

[Anonymous]

Thanks,  I did share a screenshot first time I posted, but that post disappeared.  The thing that really puzzles me is the fact that shell can work with a thickness that is far too thin, or far too thick, but not find something in between.

I will try to export one of the troublesome parts.

[Anonymous]

@jeff_strater wrote:

I think we'll need to see the design to be able to say for sure.  The fact that the model has no planar faces, to me, implies that the input IGES file also has no planar faces.  The success of a shell will depend a lot on the geometry.  Shell is a precise operation, and if the input data has areas of high curvature or other problems, yes, the shell can fail.

Anyway, we can see for sure if you share your design here.

Jeff

---

The Top and bottom surfaces are from the IGES, they are of course curved and non planar.  The wide end was filled with a patch and all stitched. I assumed this would be planar. but I suppose the patched edge mat have been bowed, So that much is now understandable.  The shell behaviour is more confusing tho...

[Anonymous]

attaching f3d export of the object with no planar edges, ie before I trimmed what is probably a wonky patched edge to get at least one planar face.

Thanks for the help. 

[jeff_strater]

Thanks for sharing the design.

Yes, your suspicions were correct - that face is not quite planar.  If you need it to be planar, I would create a workplane.  There happen to be 3 points on this face through which you can construct a plane:

then, I found I needed to move that slightly in order to get it to intersect with the body all the way through.

then, I used Split Body to trim off the body, and the result is a nice planar face.

However, regarding Shell, again, your observations are correct.  This is a fact of life with Shell.  There are some models for which some range of shell values fail, where smaller and larger values will succeed.  I will pass this off to our geometry kernel guys, but my suspicion is that the geometry at the wingtip:

Is where the problems are.  See those tiny faces, and that strange edge on the left side of the image?  Those are going to cause problems, I believe.

I'll let you know what we find.

Jeff

[Anonymous]

Thanks Jeff,

I appreciate your time looking at this.  Yes that odd "dolphin-mouth" shape at the wing tip seems to be an artifact from the stitch.  Looks like I missed or deselected some the thin strip that joins the top and bottom surfaces and stitch tool bridged that gap in its own way, probably made worse by scaling later.  That said, I went back to the original iges surfaces and restitched more carefully and still get the same problems with shell.  I look forward to hear if the geometry guys find anything.  In the meantime I will look into simplifying/recreating the main geometry with cleaner edges.

Thanks

[jeff_strater]

Hi @Anonymous,

We took a closer look at this model, and my suspicions are correct:  The geometry here is the source of the problems with shell.  In addition to the areas that I pointed out above, there are also some problematic areas at the other end of the model:

If you zoom in on this area:

You can see that there is some weird extra edges there:

If you turn on curvature combs on this edge, you can see that the curvature of this tiny piece is radically different than the surrounding edges:

Also, if you turn on the curvature map, you can see that the curvature of this model is pretty uneven:

All of this adds up to challenges for the Shell feature.  

As to why some thicknesses succeed and some fail, that really is just the nature of the shell algorithm, unfortunately.  

So, what to do?  The problem is probably the imported IGES geometry is shaky.  Is there any way that you can re-model this wing in Fusion itself?

The other thing you can do is to fake the shell.  Create a solid body by patching the end, as you were doing, then copy that solid, scale it down, and subtract it from the larger one.  You won't end up with a completely uniform wall thickness, but it'll be close.  What are you planning to do with this model?  Is it OK if the thickness is not uniform?

Jeff

[Anonymous]

Thanks again for you time looking into this Jeff.

I tried the curvature analysis and the whole wing came back green with bluey accents around the edge, but that was before I scaled the model down.  I just tried again and it does indeed look messy.  I did redo the body from IGES and carefully selected several faces that kept leaving themselves out, I also found that I could slice ot tip of the wing and then shell 95% of the wing as I wanted, and left a thicker shell at the tip before combining back again. that still leaves me with the same curvature.  The comb analysis certainly looks weird, but I really don't know how to interpret that. 

The scaling down and subtract method seems it would be too difficult to line up the 2 pieces, I don't know how the alignment id automated and be eye/hand it difficult

I would love to be able to model the wing in F360, but where to begin?  Loft between a series of cross sections?  I saw there were a few other threads on that subject, perhaps I can find a method..

Cheers