I printed a huge batch of connectors to make a 15ft structure out of plastic | Hacker News Reader

I printed a huge batch of connectors to make a 15ft structure out of plastic

formlabs.com

100 pointsdezork18 comments

That part packing using a physics simulation is genius, very clever. I did some nesting of fabric long ago, never even thought about doing it using a physics simulation. Lovely stuff. Of course in nesting fabric there are a lot of constraints due to weave orientation (especially in that application, sails for boats), presumably this 3D printed stuff gives you at least a bit more freedom though I would not expect it to be equally strong in all directions. Even so, the parts they made are already challenging that because of the connectors protruding from all angles so presumably the re-orientation due to packing will not make the parts much less strong.

SLS parts aren't substantially weaker in the Z direction like your typical FDM parts are.

I made something similar a while back! Took somewhere between 150-200 hours to print 60 connectors. I had to design the parts with a flat side so I could print them in PLA.

https://danwink.com/blog/2016/06/19/dowel-structure-with-3d-...

Cool! I love it.

Nice! I also loved your tool cart. https://danwink.com/blog/2016/09/17/toolcart-with-3d-printed...

I use FreeCAD and Blender for a lot of my technical design/engineering work. I am happy to see a physics solution in Blender that seems suited to the task of filling the volume with the parts.

I like the fact that you can use Hylang to program both Blender and FreeCAD via their Python APIs. No need to, I just like Lisp!

Here's a SketchUp plugin that lets you make large objects using plastic bottles and 3d printed connectors: https://hpi.de/en/baudisch/projects/trussfab.html

There's a more detailed tutorial on how to use physics to do 3D part packing here:

http://amosdudley.com/weblog/Stochastic-Part-Packing

Very cool approach to scaling a project out of small components.

I wonder if the print time is 144 parts in 36 hours or 36 hours per print. Thats 15 minutes a print if it is the former!

Both. All in one print!

My understanding was that by packing the parts they could do 144 parts in one print, and that one print took 36 hours. By using SLS they don't need support structures

Using physics simulation to pack parts more closely together is such a good idea. Nothing like a little shake to get things packed together.

This works well only if your manufacturing process produces an isotropic material. SLA sort of does; it's equally strong in all directions, but there is a layering axis. Filament-welding printers don't have that property. The material is much weaker in the layering axis, because the welding between layers isn't very good. Things have improved somewhat, but strength of things made from ABS filament is much lower than the same part made by injection-molding ABS.

Isn't that offset in this case because those connectors have elements protruding at many angles? You'd be weakening some and strengthening others, unless they were designed with a particular print orientation in mind.

Ugh, what's with these obnoxious buttons? https://sr.ht/S3w1.png

I'm so tired of endless barages of SUB TO OUR NEWSLETTER and LIKE US ON FB and TWEET THE SHIT OUT OF THIS PLEASE on every goddamn website.

http://imgur.com/a/O10RH

Looks fine and unobtrusive on Chrome. Seems like your browser is rendering things weird, what are you using?

qutebrowser, which is based on QtWebEngine (Chrome).

In Firefox, if you allow some JS, they're quite narrow. But they're still there, on the left.

Reply on news.ycombinator.com