Welcome! Log In Create A New Profile


building a very large 3d printer

Posted by markomarko 
building a very large 3d printer
September 19, 2015 05:43AM
Hello everybody,

I have been playing with my little printer and dreaming of constructing a truly life-sized 3D printer. Something with a build space of some 2.6x12.0x3m high – basically the largest size that can be transported by a lorry in Europe, and equivalent to the size of a large shipping container.
This would enable the prototyping and printing of full-size boats, car bodies, micro-homes/cabins, elements of buildings, large and complex formwork and moulds, etc.

Shape and form of the printer: I think the only sensible way of constructing this printer is to take inspiration from the construction printers like this good example: [www.betabram.com] and there have been a few articles about small houses being 3D printed in China. Overall, the idea of a rolling gantry is certainly the most cost-effective and practical at this scale.

Material: Although I am huge fan of concrete, I think that its weight, lack of ductility and poor finish have their drawbacks. Printing in ABS seems to make more sense, and when purchased in bulk the cost can dip below 1$/kg. Obviously, this means that buying pre-rolled filament is out of the question and rolling your own filament is the only option. Other materials may also be feasible – any suggestions? Polyurethanes, perhaps?

Capacity: The printer has to be capable of printing at least 20kg/hour to make it viable. 50kg/hour would be even better. Anything below 20kg/hr would make little sense at this scale resulting in prints taking days or weeks.
A typical 3D printer outputs about 0.01kg/hr, with estimate based on printing about 2mm3/s through a 0.4mm diameter nozzle running at around 20-40mm/s at around 80% efficiency (since the printer is not outputting filament 100% of the time, etc etc). These are only meant to be ballpark figures based on only a few examples.

Obviously, something significantly more capacious is required. To achieve capacity a larger nozzle is required, and ideally the extruder should be moving faster (I am oversimplifying things, I know!). I have found a few links with people printing with a 1mm nozzle [www.tridimake.com] (excellent article, BTW), and a guy successfully printing with a 2.8mm nozzle [www.youtube.com] . They seem to be printing at around 100mm/s which is pretty much the max for garden variety 3d printers these days. However, the guys in Amsterdam printing a 3D house appear to be printing at 240mm/s!

I have absolutely no idea what would be the max feasible print speed, so if we fix this to 100mm/s, and to achieve 20kg/hr capacity, I think I’d need to be printing with a 8-10mm nozzle. This would achieve flow volume of some 5000-8000 mm3/s. If the speed can be increased a slightly smaller nozzle can be used (which would be good). Obviously, this is only a ballpark figure assuming 80% printing time efficiency, disregarding material expansion and probably a million other factors! I have tried creating g-code using these parameters (7mm layer height, 8mm nozzle diameter, 100mm/s max travel speed) and it vaguely corresponds to these figures in terms of build weight over time. However, please advise if I am wildly off the mark for whatever reason.
I doubt that there are extruders on the market capable of achieving this capacity so one would need to be made. I presume that it should be possible to replicate and scale up a standard stepper motor-heating element-nozzle extruder used in today’s printers. People have built these from scratch, so it should be feasible, unless I am missing something.

Filament: Not sure what would be a good filament diameter to use. Given that 1.75mm filament is used with 0.35-0.6mm nozzles, I presume 20-30mm filament should be extruded and used. Or perhaps this is wrong? Any ideas?
There is a wide selection of filament extruders available on the market and at 10-20-30kg/hr capacity these should not be prohibitively expensive. Although I am not sure how much larger dye diameter affects the cost.

For printing, probably the easiest solution is to fix a spool of filament (say 50kg or so) at the top of the gantry to move with the extruder, or it can be fed from the bottom. This is better is it would reduce the inertia forces of the print head, but then the filament needs to be fed carefully and this may not be simple at this scale!

Finish: Printing with such a large nozzle and at that speed would not result in a great finish. For the purpose of prototyping it will have to do, but otherwise decent finish would need to be achieved through post-printing. Thankfully, epoxy fillers and paints are fairly cheap where I live, as is the labour.

Potential problems: Not sure if people have tried printing ABS filament at these flow rates and with nozzles of this size. Does filament coming out of a 8-10mm nozzle behave like the one coming out of a 0.5 or 1.0mm nozzle? Probably not! Has anyone tried this?

I note that the author of the text who tried printing with a 1mm nozzle has run into problems where it was impossible for the extruded filament to span any gaps horizontally due to its weight and high viscosity – basically all horizontal ‘bridges’ were sagging badly. I can only assume that at 8-10mm nozzle this problem would only be exacerbated!

Given the size of the model, there would be considerable delay between layers. Therefore the layer below would cool and it is uncertain that the new layer would fuse with the one below properly?

At this scale, and this printing flow-rates how would the model cool? Would there be big issues with warping? Again layer-to-layer time would be decent, so perhaps that would actually help with gradual cooling? Perhaps enclosing the print area would help, but what would be a good chamber temperature? How difficult would it be to maintain it at this scale?

My experience with 3d printers is that they are sensitive to imperfections and extruders tend to collide with poorly levelled build plates or with warping printed layers underneath. With small printers this is somewhat alleviated by comfortably oversizing the enclosure, frames, runners, gears etc. At the proposed scale ‘comfortably sizing’ anything would have significant cost implications and is not really an option. Hence mechanical factors such as sagging of the gantry, lateral inertia forces, etc all come into play and are very significant.

Software and controllers – would it be possible to use standard RepRap tools? Perhaps! I sure hope so!

Cost - I seem to think that this can be done on a shoe-string budget (in relative terms, of course). However, fabricating a sturdy gantry with decent runners will not be easy. A perfectly level (and heated) build platform of this size would also be a bit of a challenge. Finally, purchasing an industrial extruder capable of producing large diameter filament is not a small investment.

Conclusions: I think most of the latter issues can be overcome one way or another. However, if ABS ‘doesn’t like’ being deposited through a 8-10mm nozzle at 100mm/s to achieve the required build speed then this will not work.

What do you guys think about it? Is there an elephant in the room that I am missing here? I haven’t read that people have done this? Why is that? If printing ABS at this capacity is not possible, what other materials would be feasible?

If it is feasible, then the first step would presumably be to build a smaller printer (or modify the one I’ve got) to test the feasibility of extruding ABS at these speeds, flow rates, nozzle and filament diameters? Then scour the internet for a few kg of large diameter filament for testing purposes.

Or is this whole thing a non-starter?
Re: building a very large 3d printer
September 19, 2015 06:55AM
I've never attempted printing with a large nozzle, but here are my thoughts. To extrude 10mm diameter at 100mm/sec would need a very long melt zone to get enough heat into the core of the extruding plastic. I think it might be more practical to have a reservoir of molten ABS and control extrusion by pressurising it, perhaps using compressed air moderated by solenoid valves. You could load the reservoir with ABS pellets, avoiding the need for a special large filament. You could have interchangeable tool heads, so that one head is being charged with pellets and melted while the other one is printing. You might even be able to use a stationary reservoir of molten ABS and feed it to the nozzle through a heated PTFE tube, so as to keep the mass of the moving parts low.

With the plastic already melted, you should be able to achieve higher speeds, perhaps 1m/s. So you could use a 3mm to 5mm nozzle and get better print quality. However, the faster you print, the more acceleration you will need to achieve fast print times, the faster you need to implement pressure changes, and the sturdier the gantry needs to be to handle the acceleration.

The whole thing would need a heated enclosure to ensure good layer bonding and control ABS warping.

However you build it, I am sure it will cost a lot of money.

Delta printer calibration calculator, mini IR Z probe, and colour touch screen control panel: [escher3d.com]

Large delta printer, and other 3D printer blog postings: [miscsolutions.wordpress.com]

Disclosure: I have a financial interest in sales of the Panel Due, Mini IR height sensor, and Duet WiFi/Duet Ethernet.
Re: building a very large 3d printer
September 19, 2015 12:16PM
Thank you for your reply - excellent comments.

Whilst having a pool of molten ABS certainly seems to have its benefits, I think researching and producing something like that would be way way beyond my capabilities. The ability to work with pellets directly omitting the production of filament sounds great and the proposed 1m/s print speed very aluring, though.

My thinking was to start with a tried and tested design of a hotend and extruder and get a fabricator to make everything 5-10x larger and provide more heaters and larger scale fans for the transition (or, better yet, water cooling). I suppose, that the size of the block and the heaters should be scaled proportionally to the volume of the material being extruded (per unit of time)? Since I would only be making one (or a few) of these and not going for mass production, the few extra grams of stainless steel or aluminium should not be a major concern.

This is something that I feel is feasible and would make it possible to test the feasibility of successfully melting 10mm or 15mm filament and testing how the molten material behaves when deposited through a large nozzle.

Or is my thinking completely off, and do you think that it will not be possible to melt 10mm filament at all, even with a fairly large block?
Re: building a very large 3d printer
September 21, 2015 11:24PM
For ideas See





Re: building a very large 3d printer
September 22, 2015 03:20AM
Good links - it is a shame that both of these use fairly standard extruders. I can only imagine how long it would take to create a
1mx1mx1m print - probably weeks.

In the first instance I will be trying to prototype a large extruder hot-end with 4-8 resistors and see how much more material I can melt in a unit of time. I've got most of the materials on order, now it's just a mater of waiting for these to arrive from China! Whilst waiting, I will try to get the metal fabricated. Will report back in a month or two!

Edited 1 time(s). Last edit at 09/22/2015 10:49AM by markomarko.
Re: building a very large 3d printer
November 09, 2015 12:42PM
It's already built somewhere in RF.

Re: building a very large 3d printer
November 09, 2015 08:31PM
This is something that I feel is feasible and would make it possible to test the feasibility of successfully melting 10mm or 15mm filament and testing how the molten material behaves when deposited through a large nozzle.

Or is my thinking completely off, and do you think that it will not be possible to melt 10mm filament at all, even with a fairly large block?

Look at the difference between a 1.75 mm extruder and a 3 mm extruder. How big a motor would you need to move 10 or 15mm filament?

How inflexible and difficult to handle would 10-15 mm dia filament be?

Edited 1 time(s). Last edit at 11/10/2015 09:49PM by the_digital_dentist.

Son of MegaMax 3D printer: [www.instructables.com]
Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]
Re: building a very large 3d printer
November 09, 2015 11:15PM
I would think you may want to use a true extruder.
Like a Filabot mechanism extruding pellets into 1.75 / 3 mm
Lyman Filament extruder

Look into true auger type extrusion.

Perhaps squirting out resin and curing it with UV light
A quick setting two part resin

So what is the budget for this printer?

Looks like $500,000 range!

confused smiley
Re: building a very large 3d printer
November 11, 2015 05:56AM
Hi guys,

Some thoughts about the project.
First, you will probably invent a lot to build any profitable giant FDM printer.
And yes, it will be realy expensive !
You need to beat actual big size prototyping process in speed and in resolution.
CNC machined foam and fiberglass molding/coating is not a long process (within a day) and it's super smooth.
It won't be easy !

I wonder if it's possible to 3D print ABS at this size without major wrapping issues.

Maybe it could be interesting to build the machine directly on a shipping container platform.
You won't have to do a new setup at each the machine moves out,
and maybe you can implement a built in heating chambre and control unit in the same volume.

Forget DC technology, a machine of this side must be powered by full AC power.
Instead of several resistors, what about an inducton heating block ?
As you will be able to extrude fast, you will need to move fast too.
The structure should be probably heavy and the moving parts too.
You will probably need AC servo motors, maybe 1hp each or more.
For the extruder maybe you will need lower torque but a 3000 rpm motor can help a lot.
As you will use a big nozzle, maybe a nozzle with a wall (like on concrete extruders) to improve edge quality can help.
Some plotter heads already have a "rotate on turn" feature.

Good luck !

Sorry, only registered users may post in this forum.

Click here to login