Extruders using pellets, chips, granules

I have been thinking about one of the greatest potential advantages of using thermoplastics: the ability to use plastic waste. It is a lot easier to give some guidelines here on how to find and recognize plastic types than to acquire them on the net or a store. This is specially the case in countries where the prime materials distribution network is very weak. It is always a lot easier to find discarded HDPE, PP, PET, ABS packages, containers, waste objects, even plastic bags.

The question of course is how to reuse this plastic waste once its type has been recognized. The prime material, I guess, would be shredded to chips or granules. What would be the next step then? Can we design an extruder that would load granules directly instead of the 3mm rod? Should we design a rod extruder that prepares the 3mm rods from any given material granules?

What do you all think?

I've considered this.
The problems I see are as follows.
a. A bulk powder extruder might take up significantly more room than the hot-glue gun extruder currently being used. although I am thinking about a design to dispense talc along these lines anyway.
b. The only two ways I can see to force it to extrude are one, heat it in the chamber, let it cool in the barrel enough to get a grip on it with a screw or wheel, (I'd hate to have the viscous stuff cool in a pump,) then heat it again to extrude it. OR two, use air pressure.
Method one has the disadvantage of complexity and increased size.
Method two had the disadvantage of difficulty of control.

I've been thinking it might be easier to make a purpose built extruder, using the air pressure technique, that will make the feedstock. Laying that feedstock onto a cooling drum before winding it onto a spool or turntable.

We did a considerable amount of work on this question back in February and March of 2006 and documented in in the main blog. Here are a few of the initial links there to get you started...

[blog.reprap.org]

[blog.reprap.org]

[blog.reprap.org]

There were a number of problems that we ran into. Keep in mind that we were trying to create filament for the Mk 2 rather than replace it, though we talked about that a bit. The big problem that I had was backwards migration of heat from the extruder to the polymer pump which would cause the polymer resin powder to partially fuse and jam the feed opening in the extruder.

If you want to revisit this question I'll be glad to give you whatever details of the work I did that I didn't blog. I'm sure that Adrian will as well. I still think that the notion of having a small capacity filament extruder is a great idea for recycling plastic. Recycling, however, begs the question of having a working, small capacity grinder that can reduce refuse plastic to a grit diameter that could be used in such an extruder. Nobody has done that kind of R&D yet to my knowledge.

I have done some experiments with a paper shredder. I think if I keep recycling the shards though it they will come out pretty small. It might not last long though. The motor is not rated for continuous use. I will add a fan and see what happens when I get round to having another go. I have some pictures but they are too big for this forum so they will have to wait until I blog them.

[www.willitblend.com]

I suppose it could do other plastics as well.

Hi Fernando,

I have been speculating about this as well. And I am also wondering how we can approach this (in a different way?). The thing is that there might be very creative and novel approaches to tackle this problem. I am suggesting a method to help us get to some (potentially) working prototypes. Please forgive me when rambling or making this too complex - I am new at this open source collaborative engineering. But what I was thinking about is as follows:

We want to be able to convert existing plastic parts (assumption is same plastic type, no mix of ie. ABS and HIPS) into filament i.e. plastic -> filament

Sequential "steps" in this process could be:
plastic -> shred/grind -> melt -> transport -> extrude -> filament -> wind, or
plastic -> dissolve in solvent -> transport -> extrude -> filament -> wind, or something along those lines...

What I am trying to suggest is that we try to find alternative solutions for each of these steps (or other steps) and see if they can be prototyped into a (or multiple) device(s). For example, transport can be done using an auger/worm wheel type thing, but could also be via centrigual force, pressure (as in a syringe) or another mechanism (peristaltic pump?). Heating could be done with nichrome heating a barrel and extruder, a heating cartridge, friction in the system, heated environment, etc.

This way we might be able to come up with more alternatives ( i.e. a peristaltic pump in a bath of heated oil with silicone tubing), and then from there on experiment and build prototypes. It might even be that some combination of techniques works better than we can think of now when tackling this as a whole. And it could support the creative process by breaking it down into elements and "detail" problems instead of one big one. It might also make it easier to build on top of each others creative ideas. Just my 2 cents... what do you think?


Joost

hahaha! now i have an excuse to get one!!

I think there are excellent ideas floating around here, i think we should get used to that

I for once think that the actual extruder design is quite advanced and the improvements in it's accuracy would be thrown back at least several months if it were to be redisigned.
What I propose is that people interested in using granules could start an informal idea storm (such a buzz word!) to develop a new extruder concept.

But i must agree that as far as i can see now, the size of the apparatus might be so cumbersome that it will not fit easily in our actual cartesian robots.
Solving this problem might become a sub-project on it's own. (bigger, sturdier cartesian robot? squeezing the granule extruder to a manageable size?)

Joost: I think that the subdivision of the whole process, from picking the plastics, processing them to granules and chips, melting, transport, fabricating a thread....
draws a clear picture of where some advancement is required:

1)Picking plastics and classifying them by composition
We can easily make a document where we teach people to recognize plastics.

2)Processing the plastics to granules
There is a list of shredding/grinding apparatus that we can think about.
It would be nice if we could have a range of scales for these machines.
I for one may get along with a paper shredder an electric mincer a coffee grinder, or an electric branch shredder.
A bigger operation may want to use an engine driven shredder or some heavier industrial shredder.
It really all depends on what plastic you want to feed it and to research on machines that can do the job without falling apart. (thanx for that video David, it kinda proves the point )

3)Melting
What would be the most efficient way of having a controlled melting process?
We could play with the different plastic temperatures and use water/glycerin/mineral oil... baths to melt the ones with low fusion point.
I could even find a number of fluids to safely heat up to 200-250

Edited 1 time(s). Last edit at 05/09/2007 02:48PM by Fernando.

1 ) [www.texloc.com]

there are alot of guides on the net for how to identify plastics... good to have work done already =)

Personally, I think the best route would be to make a machine that produces 3mm filament. That way it fits nicely into our current system/ecosystem. It also wouldn't be limited by the parameters of having to fit on the printer. That means we can design it to be strong and capable.

I'm imagining a system where you have your RepRap machine, a recycler/extruder machine, and then various other machines made of printed parts: eg. a gear cutter (could be eliminated if we really nail laser cutting...)

I was thinking as follows...

Old, (or new) pressure cooker, converted into a doubleboiler, (might end up using a liquid other than water for some of the plastics.)

An extruder nozzle drilled into the lower edge, and an airchuck added to the lid.

Fill with plastic, put on a hotplate. When the plastic is melted, attach lid, and air compressor.
Express plastic until the nozzle starts spitting air.

It would need...
A pressure cooker.
A drill for the pressure cooker, (and the next item.)
a pot of a larger diameter than the pressure cooker, but not more than a couple inches taller.
A piece of threaded steel pipe,
About 4 bolts, and 4 gaskets that can take the heat of direct flame. Might get by with two.
at least three insulative standoffs to place between the two pots.
an air nozzle.
One of those air chucks that can be clamped on and left.
a hotplate, or propane burner. (Tripod mounted would probably be ideal.)

As for spooling it up. I was thinking a hollow cylinder of a reasonably large diameter, (perhaps just a cut down soup can,) to spool the filament across, to dump some excess heat, and then spool it off onto a garden hose roller.

This is just off the top of my head, of course, and I don't know if it'd work or not. Or work more than once. Even if it did, prepping might involve a sharp <3mm rod to clear the nozzle before the next use.

Better would be to replace the double boiler with several heating elements, and thermostats. However, the heat would need to be taken close to the thermostat unless you're going to stir the plastic, which becomes difficult once it's time to start making the filament.

Edited 1 time(s). Last edit at 05/10/2007 12:31AM by Sean Roach.

[www.lindsaybks.com]
Heat plastic in a tube with cartridge heaters, then use a big lever to smush it into a mold.


People-powered version. Granted a machine shop is needed, but sometimes everyone just needs to settle down a little and think simple

Point is not all designs need a continuous feed. I'll give it to you yes that's the best idea for long continuous strands, but a simple joining fixture wouldn't be that hard to cook up (imagine a similar apparatus to the one that joins the z-axis belt in Darwin)

"Heat plastic in a tube with cartridge heaters, then use a big lever to smush it into a mold."

I built one of these Lindsay injection moulders early last year. It works, but the problem with it is the very limited capacity of the heated injector cylinder. You can't extrude a lot of filament with it, indeed you can't really extrude a long enough length to be worthwhile, only a few metres at most. If you increase the diameter of the injector cylinder you increase the amount of pressure the amount of force that you must apply to the cylinder.

We talked about using a cylinder driven by compressed air, compressors being very cheap, to get around this problem but never got around to building one.

We pretty much dropped all that line of inquiry when we discovered that we could buy filament from the welding rod people at very nice prices. We need to revisit that technology, though, if we seriously plan to recycle.

My day job is repairing industrial equipment (electronics side plus some mechanical). I've been in several plastic user shops. The one I can think of that is most applicable to the RepRap is one that molds bottles. This might give some perspective into the problem or even an idea.

They feed the project from a pellet hopper with shredded scraps tossed back in from rejects. These machines use a combination of heat and friction to do their job. Their heaters get up to near melting temp for the plastic. A special screw (I think!) that gets finer (or is that diameter change... check online) as it approaches the end is used. There is an injection area that holds the "shot", which should be at temperature. The final energy comes from a single cycle push that presses the liquid material out. A combination of the near-melting plastic and the turbulence of the screw threads keeps the plastic from feeding backwards. The screw is not a perfect seal, nor should it be.

Biggest problem they have is that you need enough raw power to turn the screw to mechanically liquify the plastic at startup. For the ones I normally work on, they have a 4KW AC drive running the injection cycle and a 90KW drive running the screw. Each can provide 1 minute of 200% torque/current to their motors to get the process going.

A screw based pre-extruder that is long enough screw length with heaters at the working end could easily take advantage of the same effects. However, you'll also run into the same problem of getting it going in the first place. This is still going to be a somewhat offline application since this extrusion will be FAR faster than the consumption by a RepRap.

Yep, I was going to suggest a hopper-based approach too. SOI Sentinel's system sounds even more useful -- especially the ability to throw in (possibly cut-up) offcuts, etc., and run off new items from them

Do the current materials support re-molding, by the way? It sounds like Polymorph is particularly flexible on first heating, but becomes permanently harder once cooled.