Way to add extra material everywhere

[Pierrec]

Hello everybody,

I have a very complex part model that's full of curved surfaces, holes, mortises, etc... that I need to "puff up" 0.15mm everywhere, because I need to have it made with a laser sintering machine and I want extra material to compensate for the machine's building tolerances in all 3 axis.

Of course, I could rework all the faces one by one and move each face out by 0.15mm, but I'd love to save some time here. I tried selecting all the faces and moving them, but Solidworks won't do it. Would there be a clever way to "coat" the entire part with a uniform layer of extra material?

 

[solidmonkey]

Hi,


Have you tried using a scale factor to thosw the sintered stage? You could then suppress this to show the finished part.


Go to the drop-down menu: Insert> Feature> Scale.





Good luck!

 

[solidmonkey]

You could also select each face and create surfaces, then thicken the surfaces-that would give you direct control over the thickness.

 

[jimshaw]

Okay here goes:


1. start with your part, and create a copy of it (copy/paste in windows) called part_shell. take part_shell and insert a shell feature, the desired thickness. MAKE SURE TO SELECT "SHELL OUTWARD"
2. Create a new assembly, and insert both parts at the default location. You should now have what you want, but is in an assembly.
3. To consolidate to 1 volume, save this asembly as a "solidworks part"; this will create a part with two separate solid bodies in it.
4. Open this part, and insert a "combine" feature which will boolean add the two bodies creating a single volume, which can now be sent to your RP machine...


I haven't tried this, but in theory it should work. Please let us know of your success...


Jim

Edited by: jimshaw

 

[dgs]

I'm not sure about how to do this, but scaling is what you want. I think the part's not going to grow normal to the surfaces, it's going to grow more uniformly.

However, unless you are doing this in house o your own machine, I would have the laser sintering vendor handle it. You're contracting him to give you a part that matches your data, it's up to him to figure out how to do that.

We do this all the time with injection molding or urethane casting vendors. They determine the shrinkage and compensate the geometry accordingly.

 

[Pierrec]

Thanks for the responses everybody!

I'll try scaling the part. I'm not too sure about this one though, I think the volume will grow outward from the origin, meaning that the outer faces will move out, but center mortise in the middle of the part will widen instead of shrink. But I'll try. I'll also try the shelling procedure, thanks for the detailed instructions.

Re the laser sintering process: what I need is to figure out the geometry of the part so that, after manually ridding it of the bad SLS surface finish and mirror-polishing the part, we arrive at the correct dimension to within 0.01mm. The SLS machine's general tolerance is +-0.1mm, and we need 0.03mm to get a mirror finish manually, so 0.15mm of extra material is a safe bet.

Some companies do offer to correct the part, but in this case, this is a test part I'm having made by the SLS machine vendor to test a custom steel powder. They are in the business of selling the machines and the powders, they don't offer any extra service, they just want a model file they can plop in the machine.

Besides, the powder we're testing is custom enough that they don't know how the part will behave exactly as far as density, shrinkage, tensile strength or hardness are concerned. This is pure R&D, so the amount of extra material to leave on the part is educated guesswork at best, but they're not signing anything to guarantee it. In short, they're happy to build the part as per the model I supply, but the rest is my problem.

I'll let you know how scaling and shelling turn out.

 

[jimshaw]

Yes, when you mentioned the internal features, that's what made me think of shelling. I've done my fair share of sanding RP models


good luck!

 

[dgs]

Sounds like a 2 fold problem. First the scaling due to the shrinkage of the material (I'm assuming that it will shrink as it cures/cools) and second the add on surface to compensate for finishing. Sounds lie a fun challenge.

Oh, and yes, if you scale a part up, internal dims (say the inside of a U section) will grow, not shrink. Sounds like that won't work for you, at least for the finishing compensation part.

 

[Pierrec]

Ok, I've played around with this and here are the results:

- Scaling doesn't work for the reasons mentioned above

- Shelling doesn't work either. In fact, when moving faces doesn't work, the shelling function doesn't either. I suspect SW performs shelling by moving faces internally, or there's some shared code somewhere.

Eventually, I ended up moving faces in several steps, that is, I'd do a "face -> move" once and grab as many faces as SW would accept, then redo "face -> move" and SW would accept more faces, etc... By moving the right faces in the right order, I managed to achieve what I needed somewhat. But it was definitely a major pain in the rear, and not simple or clever at all.

 

[jimshaw]

interesting, I'm curious why shelling won't work though... can you share a picture of the geometry?

 

[design-engine]

There are many techniques for figuring out why an object will not shell. Think of it like working on your car. You narrow down the problem with logic.

For example: cut your problem part in half. if one half of the part shells then you know the problem is in the other half. Modify you cut to increase the shelled geometry to narrow down the problem area.I use/teach problem solving with a lesson plan in many of our training classes because it is a systematic approach.

 

[Pierrec]

I'm afraid I can't share the SW model because it's company policy, but I suppose I can post a couple of pictures, so here they are. The problematic areas are the complex ball-like shapes at the top of the part, as well as the intersection between some of the holes that are drilled at an angle and the exit faces of the holes which are not simple geometric faces either.

I'm not surprised shelling - or moving faces - doesn't work. This part is pushing the SW engine to its limits, so much so that it crashes very often whenever I want to modify something. SW 2009 won't even open it without crashing right away, which is why I stick to SW 2008. Unfortunately, many of the part in the assembly for this product are like that too and the whole assembly, which has only 150 parts, won't open in a machine with less than 8 gigs of RAM.

Here are the screenshot. I hope the overall shape is understandable:

 

[design-engine]

Im not sure about SW2009 or 2008 for that matter... new releases of any software tends to have some bugs. Im sure the programmers can work out those bugs.

As for shelling... that looks like a solid part. Nylon?

Lots of side and slanted pulls.

Maybe you would want to shell this part before the side action?Another trick to determine what feature is the problem... go into insert and shell the part at feature 10. if it works okay there and fails at feature 11... then you problem solve that feature 11.

 

[Pierrec]

It is a solid part, made of forged steel.

What I ended up doing to thicken the entire part is, as you advised, work on bits of the part until I hit a problem. But since I had to work on details of the part anyway, I didn't use the shelling function but simply moved groups of faces outward.

I think I have a fair idea of why and where it went wrong. It's where complex faces meet (such as lofted surfaces driven by spline guides): SW simply has no idea how to extrapolate the rest of the surfaces when they extend too far out.

The other issue might also be with the lofting function : when I created the lofted surfaces, I had to tweak the various driving sketches and guides a lot until SW finally accepted to generate a valid solution, or a volume that would accept to be combined with another. Lofting in SW is quite fragile and finicky. Sometimes I had to add 0.001mm here and there, or make a guide not-quite-tangent, just to avoid SW throwing a fit over some "geometric condition" problem. In short, SW can rebuild the part's features, but the amount of control I have over some of the surfaces' parameters is marginal at best. I suspect moving such faces is enough to change the lofting parameters enough internally that it refuses to generate a new surface.

What I'd like in SW really is a function that would add material "stupidly" instead of trying to be smart about it, meaning something that would turn the fully parametric model into a vertex model and then grow (or shrink) each polygon in a direction normal to its surface by a fixed amount, and merging/deleting polygons that end up "swallowed" by others. I reckon the best place to put such a function would be in the STL format saving options; this is where it should naturally reside. Besides, the vertex creation routine is already there, growing or shrinking the polygons wouldn't be that much of a stretch to implement really.

I think this is something that should be adressed in SW, as rapid prototyping and rapid manufacturing are getting more and more common.

 

[jimshaw]

PierreC -great part! I love challenges...


You could try cutting the whole thing in half (creating two separate models) and then work in each part individually. If you are having a complexity problem with SW, this could get around it - by simplifying the feature tree a bit. When you're done put the two halves together in an assembly and merge them back together.


Or, you could remove all of the holes/mortises/etc., scale it out, then add in those material removing features again (factoring in your .005"). This may prove quicker than troubleshotting parent/child relationships...


So this RP company won't gaurantee a tolerance range on the produced part? Sounds sketchy...
Edited by: jimshaw

 

[Pierrec]

So this RP company won't gaurantee a tolerance range on the produced part? Sounds sketchy...

They do on the materials they know, but we're asking the part made in a material they are not familiar with. So they gave me guestimates. But in any case, we need stock in order to rework the entire part by hand to make it mirror-polished and still arrive at the final dimensions.