Sheetmetal stuff

[rabbit]

I’d like to see an option for the extrude curves command to optionally make the end surfaces (caps?) not be in one piece, but to be split as appropriate based on the geometry being extruded…

For example, if you make a simple U shape by extruding a set of curves (inside & outside) then you get continuous end caps in the resulting 3D model.

But if you extrude just the curves describing the outside of the U for example, then make the 3D shape by offset srf the surfaces, then the 3D model has the endcaps “split” where the edges of the original faces lie.

Why would I want this?

I send a bit of stuff off to get laser cut – usually in 6mm alloy – and folded into various shapes.

If I send the 3d rhino model in .stp format, then at the shop they can often open it (in solidworks) and unfold it to get the flat pattern for cutting the sheet.

But, solidworks (or other software that unfolds sheet metal) needs the model to be built as I have described, or it wont unroll. So I have to edit the model edges manually to get this result.

What I’d really like of course is for rhino to be able to easily unroll its own models into a flat pattern, but this unfortunately doesn’t seem to be high on the list of things to look at…

cheers
rabbit

[JimCarruthers]

What can Solidworks flatten that Rhino’s unfolding tools can’t? My experience with sheetmetal is that SolidWorks and the like can only unroll maybe single-curved shapes like an extruded profile and that’s it–which Rhino can do–every other kind of detail requires rebuilding the model using parametric sheetmetal-specific features, and I’ve had the fun of watching engineers (young ones, mind,) struggling with seemingly simple assemblies complicated by the quirks of the sheetmetal features.

[rabbit]

For sure you can go a long way with rhino’s unrollsrf command - its just easier for me if i can send the shop the 3D model, so they have that reference, and since they are going to both cut and fold it they have everything they need from me to describe the job.
Recently I have just been sending even flat pieces to them from the model; It means I don’t have to document everything, convert stuff to dxf etc…
But it comes unstuck if they cant get the model to unfold…

cheers
rabbit

[ivelin.peychev]

Honestly I cannot understand what you’re asking.

[Michael_Pryor (ekim)]

There is an easy fix for that. After offsetting just use the command MergeAllFaces. Then your profile will look like the right most extrusion in this image.

[pascal (Pascal Golay)]

Hi Rabbit - - the structure of the ‘caps’ in offsetting is completely different - I’d use that for sheet metal. Extruding makes a planar cap and this is consolidated over the whole cap so that as a brep/polysurface it is a trimmed plane - in general that is the ‘right’ way to cap off a planar hole - because one curve does not know about any of the other curves that make up the boundary of the camp. On an offset, this is not the case. which is why it ‘works’ for you.

-Pascal

[rabbit]

Thank Pascal.

The thing is it took me a while to discover this was (one of the reasons) why the models wouldn’t unroll in downstream software.

The model passed all the checks I’m aware of in rhino- it’s “valid”, it has no naked edges, it Gaussian curvature is all “green”.

But because of the way it was constructed, it had properties that made it not play nice with other software that needed to be used to manufacture the part.

So to say one way is the “right” way is clearly not the case, because it depends on what the model’s purpose is to be.

So look, I am just a vagabond, a drifter on the run…I’ll never know all the intricacies and mathematics of how rhino works, let alone solidworks or inventor or autopol that shops and firms use to manipulate my models to hopefully produce the thing I want.

I can drive my hybrid car well enough to get around safely, and I have a broad understanding of how it works, but I couldn’t analyse and fix it’s rare meltdowns, nor would I want to.

I can drive rhino well enough to leverage my brain to extrude concepts, ideas and products into the real world where other people can see and interact with them; and over the years it’s allowed me to live a much enhanced and creative life, but I can’t analyse and fix it’s rare meltdowns beyond a certain point, and nor would I want to.

Nor, without some guidance, can I begin to understand or adjust internal parameters of a model that rhino says is good, but which other software third parties use and whom I need to interact with, won’t accept.

So this forum is one fine form of guidance (cheers!) but another would be some notes in the help file - perhaps a “troubleshooting” section, or an appendix “exporting rhino models to constraint based software” ; but far better still some more robust tools to analyse the model internally.

I’ve read the McNeel white paper re working with solidworks - it’s full of how accurate and versatile rhino is and how it can fix geometry that solidworks can’t, but it’s very quiet on how we might build or adjust models in rhino that will behave in sw.

During this process of tracking down why my models wouldn’t unroll, the shop sent me a .stp model built in sw from a flat pattern I’d sent them ex rhino, both a replica of the 3D part, and it’s unrolled state.

This sw model is very “clean” - it unrolls fast and easy, yet in rhino, from what I can see, it seems identical to my own model.

Where are the tools in rhino I’d need to examine both models to be able to understand the differences that are clearly there?

I can send you both models, and I’m sure you guys will be able to unpick them, but what am I able to do with my vanilla rhino?

I have no access to sw, or that kind of thing, and nor do I want to go there; I just want to be able to analyse my rhino model well enough to solve this kind of issue; for that I think rhino might need some better tools, or at least, better documentation…?

Cheers

rabbit

—Sent from mobile device & email—

[pascal (Pascal Golay)]

Hi Rabbit - if you can PM me your model or send it to my email, I’ll gladly have a look and pick out/point out the differences if I see any. But I think the upshot is that to get sheet metal friendly solids, it looks like offsetting what would be one side of the finished part is the way to get there in Rhino and not extruding closed sections of the finished part.

-Pascal

[JimCarruthers]

Of course the Solidworks model is going to unroll quickly because it was built in SolidWorks using Sheetmetal-specific parametric features that are separate from the normal part creation tools. I didn’t even know it had any ability to try to flatten imported geometry.

For parts that are laser cut and bent on a brake–is that what we’re talking about?–the closest approximation in Rhino will be to model the “mid surface” of the parts, in the cleanest manner possible with the flat areas all planes and the bends all analytic arcs. That’s not really precise enough for production though, since depending on a zillion factors the mid surface won’t be precisely the radius you want flattened out, but it’s the best you can practically do. It’s enough for quoting, and it’s not even your job to know the precise “k-factors” the shop uses, it’s up to their engineers to know what they need to input to get the result you’re asking for on their machine. It’s a pain for them to have to remodel it in SolidWorks…but frankly sheetmetal parts aren’t usually that complex…though I did have a shop make a $25000 error cutting and bending an assembly I sent them a 3D STEP to match and meticulously laid out flattened mid surface patterns…they built everything 2" too long.

[rabbit]

Thanks Jim - yeah, laser cut & bent on a brake.

I used to do exactly what you said - send the shop a 2D dxf derived from a rhino unrollsrf from the centreline of the model, plus a PDF of the part dimensioned & with any notes, 3D views etc necessary to describe the intention and critical dimensions.

Certainly the results back were good enough for me.

It’s just recently they said just send us a stp model, even of flat parts that just need to be cut…

Well, that’s potentially a lot less work for me, since I’ve already modelled all the parts & all I have to do is export them, so it seems like a good idea.

But if their software can’t eat the rhino part to unroll it, then it becomes problematical, which is where this thread comes from…

I’d like to see one click unrolling in rhino.

I’d like to see automatic 2D plans linked to the 3D model, especially for architecture and similar modelling.

I’d also like to be 18 again… so I just have to do the best I can with what I’ve got, but with an eye for continuous improvement.

cheers

rabbit

—Sent from mobile device & email—

[rabbit]

Here’s are the models Pascal.

I’d be very interested to see what differences you can find, and how you found them…

cheers

rabbit

folded_parts_rhino_solidworks_comparison.zip (1.41 MB)

[pascal (Pascal Golay)]

Hi Rabbit - so far I do not see any differences that would affect unrolling - the Rhino model is slightly cleaner in that the arcs and cylinders are rational simple arcs and not refit, but I can’t see how that makes any difference in the end. Some places the SW part has trimmed planes where the Rhino part could but has a loft style surface instead-

-Pascal

[rabbit]

Thank you for taking a look Pascal

I’ve replaced those surfaces you pointed out with trimmed planes.

While doing this I noted that the two long thin folds where the flat plate with the logo on bends up, there are four surfaces at the ends.

Trying to replace the two outer ones with the planar srf command doesn’t work since it reports the four edges don’t form planar loops.

This indicates that those four edges are a bit out of whack, but the edge srf command doesn’t care of course & just works, and the model when joined reports no naked edges.

So now the model is built, how would I know that those edges that I’d like to be planar aren’t, except by pulling the model to bits and checking with a command that itself has a built-in requirement that the condition is met?

Couldn’t / Shouldn’t there be a command that would check the model for this condition?

So you could say because of my obviously sloppy modelling I’m in a mess downstream and it’s just too bad - should have used planar srf from the get go, which is a bit harsh even if fundamentally true.

There must be a set of tools like planar srf that will accept no nonsense, and these are the tools to use when building stuff that requires this kind of rigid parameters…?

And if so what might they be…?

And wouldn’t it be useful to be able to turn on a check when modelling this kind of thing (like the bad objects check which I have on all the time) that warns if things are getting out of whack…?

Or would you say that there already is this tool, but it might need a bit of tuning up, and it’s called a brain…?

Cheers

rabbit

—Sent from mobile device & email—

[pascal (Pascal Golay)]

Hi Rabbit - I don’t think there is necessarily an advantage to using trimmed planes - I just noticed the models are different in that respect in a couple of places - it was not a value judgment! I’m not sure I see the location you mentioned but one thing to try is MergeFace - get to click on two planar faces to merge if possible, or ExtractSrf and delete the planar faces you want to consolidate and then run Cap on the part.

-Pascal

[JoergH]

From what I can see the bending radi are different. For aluminium we use 2xsheet thickness, otherwise the material starts to break. This is what was used in the sw file. Your model has an outside radius of 9mm. That is asking for trouble, and I dont´t know if your bending shop can import it correctly, at least not with the correct materials.

I have occasionally also send 3d stp files made in Rhino to laser cutters/benders, and they had no problems to open it.

[rabbit]

Thanks Pascal - No, I realise that, its just I was trying to duplicate the SW model…

Anyway, to pursue this further what I’ve done is download a trial version of Briscad Sheet Metal; briefly this is what happened:

The rhino file was exported via Autocad Solids (2004) from V5, since the trial version doesn’t have the importer for other file types module.
There is apparently a rhino converter available as an add-on.

The folds coloured yellow in the briscad view are recognised as a “bend feature”, and those coloured light blue are recognised as a “lofted bend feature” (see screen grabs)

Ideally all the bends would be yellow, and if they are, the model unfolds cleanly.
It might unfold anyway if the bends are light blue, but the red lines describing the bend radii are fragmented.

Anyway, the thing I’m interested in is the reporting of the flaws in the model. The Audit command
finds many, and fixes them automatically if possible. I have no idea what they all are, and many are probably internal to the way the program likes to see stuff, rather than actual problems in the model.
Also, it might be that the model conversion ex rhino as Autocad Solids introduces errors that were not present in the original model.
Also, if a model has zero flaws, it still doesn’t mean it will necessarily unfold.

So what’s interesting is running the Audit command BEFORE converting the part to sheet metal:

1 time(s): Warning - curve not C1
2 time(s): Error - surface irregular
1 time(s): Warning - TCOEDGE curve not G1
4 time(s): Warning - sliver face detected
4 time(s): Warning - curve near tangent, not G1
7 time(s): Warning - surface near tangent, not G1

Note these problems are persistent and don’t go away with the automatic fix option of the audit command, and remain reported throughout the process, so they are indeed real faults in the model.
Infuriatingly, the program doesn’t specifically highlight the exact places where these faults occur, but taking the blue faces as the clue, back in the rhino model I did indeed find all those problems, and was able to rebuild that section, and the next import into Briscad reported no flaws after the initial audit run, and the model unfolded easily and quickly like a flower

So what I take from this rather lengthy exercise is that while rhino does indeed have the tools to build a clean model that will be happy in downstream applications, it lacks robust reporting tools to examine the model to flag errors that are real and matter.
I think this is a grave omission. It is obviously possible to code these tools, since here is a not-very-expensive program doing just that thing.

So I’d like to put in a feature request for V6 / V7 that these kind of tools be introduced.

cheers
rabbit

[rabbit]

Yeah - good spotting!

Most of the stuff I’ve sent has been 6/12, some of it that went to a different place was 8/14; it all seems to come back 6/12 regardless…

Having said that I don’t actually know why that file I posted is that tight, I must have lost track of what I was doing somewhere - I agree it’s not right!

Cheers

rabbit

—Sent from mobile device & email—