The Achilles’ Heel of 3D Printing
Why ‘additive manufacturing’ isn’t expected to take over large scale industrial production any time soon
We think we know what makes things expensive to make
We’ve all got a surprisingly clear idea of exactly what it is that makes something really difficult and hideously expensive to produce: serious complexity.
The cost of just about everything we make goes up exponentially as the physical functionality of its innards gets more sophisticated (big things containing motors and gears, for example, are rarely ‘as cheap as chips’: even silicon chips are only cheap because, despite their enormous complexity, we can and do make them in enormous quantities).
But there’s an exception
In 3D printing, our whole intuitive concept of ‘cost related to complexity’ is turned on its head.
The cost of creating things using a 3D printer ‘goes down with complexity’: the more complex the item being printed, the less it costs to print it.
Usually, the more complicated something is (i.e., the more parts and sub-parts it is made from and the more complex the resulting assembly process happens to be) the more it costs to make.
But, bizarrely enough, 3D printing doesn’t tend to ‘struggle’ in any cost-incurring way based upon how intricate the design of the item being manufactured happens to be: a 3D printer prints a complex 3D shape just as easily as it prints a simple one.
Complexity actually reduces 3D printing costs, are you serious?
In fact, because the cost of printing an object depends almost entirely upon nothing more than the quantity of ‘ink’ (and comparatively little electricity) the more complex the shape, the more ‘air’ there is between the parts (designers call inter-component-spaces ‘voids’) the less 3D printer ink (which is usually plastic, but can sometimes be metallic) that is required to create the printed object and so, contrarily, the more complex the object, the cheaper it is to print.
The notorious 3D printing ‘complexity paradox’
greater complexity = more + bigger voids = less ink = lower cost
The complexity paradox raises the following questions:
So if the one thing (complexity) that makes most things more expensive to make is unquestionably vanquished by 3D printing, why isn’t it ‘game over’ for all other forms of manufacturing?
How 3D printers work
A common variety of 3D printer is really just an ink jet printer (although there are many other types of 3D printer) which prints physical objects in layers by using thick ‘ink’ and by moving the print head backwards and forwards as well as from side to side (and also moving upwards as it completes printing each layer).
Ink yes, cartridge no
Unlike a conventional inkjet, the material that is used to ‘print’ the shape doesn’t come in a cartridge: the 3D printer has to be fed a very long, thin, curled-up strip of plastic that sticks out of the ‘extruder head’ at the top of the printer: it’s this strip that is the ‘3D ink’ and it’s called a ‘filament’.
The printer’s extruder head melts and then dispenses the resulting molten plastic of the filament through a nozzle, emerging as a fine, thread-like ‘dribbled squirt’ stripe of viscous, gooey ink that sets hard and fuses with the layer of ink below it as it cools (just imagine using an inkjet printer to write in frosting on a wedding cake (and yes, of course, ‘extrusion printing’ has actually been used to do this: it’s their versatility, cheapness and scope for rewarding personal inventiveness that is responsible for the current 3D printing craze).
And if 3D printing machines can currently be not just tens, but hundreds and often thousands of times cheaper to buy than their long-established industrial counterparts (which actually consist of not just single machines but of entire production lines, very often costing millions) why isn’t the manufacturing industry imploding as we all stop buying things and start making everything at home using devices that cost as little as $200?
What are the competitive advantages of conventional automated manufacturing?
Well, there are many challenges to 3D printing on an industrial scale, but the most significant and daunting of them relate to the fact that most of what is manufactured is made from parts that are ‘too many’, ‘too simple’ and made (and usually needed) ‘too fast’ for 3D printing to even come close to being competitive with existing manufacturing technologies.
How do manufacturing industry insiders view 3D printing?
3D printing, despite the ‘leading edge’ cachet associated with its technology, is ironically mostly seen by the manufacturing sector as a ‘cottage industry’ and a ‘craft’ (despite being fully accepted by them as an indispensable design and tooling resource) rather than as the destiny of of large scale manufacturing.
What sticks do they use to beat it with?
If you were wondering which unflattering terms manufacturing industry insiders use to characterise the shortcomings of 3D printing when compared to other forms of industrial production, you might want to check these out (well-informed manufacturing industry commentators are also well aware of the developments which address these issues, you can also check out the the ‘criticism responses’ sidebar, below):
- agonizingly slow operation
- niche applications only
- extremely low throughput per station
- hard to scale
- mostly small items only
- unsuited to volume production
- still in its earliest stages of development
- mostly plastic-only products
- limited range of fabrication materials
- mostly low precision output
- mostly fragile, low durability products
- mostly single fabrication material products
- mostly only for products with no moving parts
- mostly low quality surface finish
- highest spec 3D printers (fastest, most flexible, finest detail) still dramatically lower throughput than conventional production line equivalents
- cheapest 3D printers can make small, decorative knick-knacks, but not much else
- just fun to watch
- best for educational and hobby use
- just cheap toys for making cheaper toys
- just a designer’s fantasy about cutting out the middleman
- a great solution still looking for truly applicable problems
- serious cost issues on almost all large-scale applications
- unresolved technical problems on most fronts
- only really suitable for DIY or small startup usage
- only serious design role is prototyping
- only serious production role is for making molds
- suited to the desktop or garage, not factory operation
- CNC and robotics were also predicted to take over everything in manufacturing decades ago, but are still only niche
The primary challenge for 3D printing at the industrial scale is not that it is limited or ineffective (all of the above kinds of problems are rapidly yielding to extensive research).
The two differing optimisations
Produces objects with unconstrained geometric structural complexity as slowly as necessary and as cheaply as possible
Produces parts with unconstrained material composition requirements as quickly and accurately as possible
Does robotics have a role to play in the comparison?
Notice that trying to compare 3D and non-3D is asymmetric: 3D is more ‘holistic’ (except when it comes to fabrication multi-material composites, where non-3D is far less constrained) and that non-3D is more about ‘parts’ than objects, because 3D can offer the opportunity to print entire ‘multi-part objects’ (when the moving parts shortcomings of 3D printing are overcome) ‘in one go’ but in both 3D and non-3D, the ‘some assembly required’ issue can still arise, something which, in the context of automation, tends to introduce robotics, which, for the purpose of this article, I am classifying outside of both 3D printing and conventional manufacture, but at some point will need to be considered in the context of the following question:
Does ‘robotics combined with 3D printing’ change the comparison with conventional manufacturing’?
What about 3D and non-3D integration?
At this stage I can’t see how the robotic assembly issue necessarily affects the comparison, but it certainly does raise ‘integration-related process optimisation issues’ (i.e., integrating 3D printing and non-3D printing into the same manufacturing process) but I don’t see how this would introduce any ‘acceleration of 3D printing encroachment on non-3D manufacturing’ except in the probably comparatively rare situation where certain conventionally manufactured parts prove to be more suited to 3D printing (e.g. internal stress-bearing design optimisation)
3D printing’s biggest bugbears: simplicity, quantity and speed
It’s just that the unique combination of the primary advantage and disadvantage of 3D printing, namely ‘economical handling of complexity plus inherently slow operation’ are the precise opposite of (or at the very least, poorly aligned with) the imperatives that conventional manufacturing processes are optimised for: simplicity, quantity and speed.
Responses to criticisms of 3D printing
Slow for production? Yes, but it speeds up the design process tremendously
Does speed always matter? Not if the thing being printed is actually impossible (or too expensive) to make or obtain by other means
Niche only? Yes, but maybe the future is more niche that the present: mass production itself may become niche
Mostly small items only? You can actually build houses using 3D printing: also, creating giant 3D printers can be orders of magnitude cheaper than equipping any other fabrication process to produce big parts
Ok, so maybe the examples above are only exceptions, but you have to admit:
almost every criticism of 3D printing has important exceptions:
No matter how sophisticated an industrial-scale mass-produced product happens to be, most of the devices that make up its components are usually produced by manufacturing equipment which operates at tremendous rates of throughput: conventional manufacturing automation is all about making very large numbers of simple things extremely quickly (and accurately).
Volume and speed crush the complexity advantage
Hundreds of billions of plastic and metal parts are turned out in millions of industrial production lines around the globe every year, each of which is often produced in far less than a second if they are stamped out, often a few seconds or less if they are molded.
The typically noisy drumbeat of conventional manufacturing is the unmistakable sound of a completed part (but far more often, several) being ‘spat out’: the more sedate, often inaudible drumbeat of 3D printing is the sound of just one more printed ‘ink-stripe’ of just one more layer among the many layers which need to printed before the part is finished.
Even the fastest 3D printing process is typically hundreds but usually thousands of times slower (often many hours per item, very rarely much less than half an hour) than its conventional manufacturing automation counterpart and there is no sign, despite notable recent speed improvements, that it is going to come anywhere near to catching up any time soon.
In the design world, the 3D revolution has already happened
The good news for 3D printing in terms of its future role in industrial processes is that 3D is by no means rendered inconsequential by this limitation as a production line resource.
Rapid prototyping of 3D models by industrial designers (notice how ‘rapid’ for design-turnaround means something quite different to rapid on the production line) happened to be how 3D printing began, and the strong prospects for continued reduction in the costs of 3D printers and advances in quality and flexibility of their output (and also in speed, but by no means to an extent where their production line limitations as identified above are expected to be significantly overcome in the short or even medium term) mean that almost the whole of the industrial design scene has been moving towards being 100% 3D printing-capable for some time now.
3D already has an important foothold inside manufacturing
On the production side, 3D technologies (including ‘subtractive’ CNC equipment such as milling machines, a relevant but quite different 3D technology, which also suffers from quite similar but not identical speed-versus-complexity trade-offs as 3D printing) play a growing role in helping to create molds, devices and components used in non-3D manufacturing.
But that won’t tip the balance: most manufacturing still needs to address speed requirements, not complexity
But for everything else, the ‘complexity handling advantage’ and the ‘speed disadvantage’ of 3D printing compared with conventional manufacturing technology almost exactly cancel one another out.
This means that when the complexity of the product being manufactured is low enough, 3D printing’s orders of magnitude of ‘complexity advantage’ all vaporize and volume and speed requirements become decisive.
Several thousand finished units could have already rolled off the conveyor belt of a comparatively modestly equipped production line, whilst each 3D printer is still toiling merrily away, less than half way through its first unit (unsurprisingly perhaps, nobody has yet built a production process using many thousands of 3D printers in order to address the speed issue by adopting ‘massively parallel 3D printing’).
What would need to happen in order for us to switch to 3D?
So, in order for 3D printing to substantially replace existing manufacturing technology, it looks like it would need:
- 3D printing to offer us a speed of operation that was at least one, probably two, but in most cases as many as three orders of magnitude faster than it is now, or
- that the number of 3D printing machines used in manufacturing would grow exponentially and (dis?)proportionately (for one of several potential reasons, see the robotics and integration issues in the sidebar) or
- that our need for (or ability to afford?) so many mass produced items would need to diminish (due to either reduced consumption or due to a widespread and unprecedented increase in the appetite for ‘mass customisation’?)
all factored into the same equation.
Until one of these major changes occurs, but 3D printing still remains enormously slower than conventional manufacturing, it is reasonable to anticipate that 3D printing will probably extend each of its current niches and almost certainly find itself a few more, but that large scale manufacturing can be expected to continue to be dominated by other, non-3D technologies.
Is this really a big deal?
If you need to get your head around the sheer scope of the challenge, just go to your local giant supermarket or hardware store and look at the tens of thousands of products: now look at all the jars, bottles and tins and their lids: all manufactured using automation, untold billions of plastic, metal and glass items produced worldwide, every day, none produced by anything resembling 3D printing. And remember: each of those containers, made in split seconds for pennies, still needs to be filled with things like yogurt or paint and then packaged: in the real world, manufacturing technology isn’t just about making things.
And what would happen if …?
It probably goes without saying that if we suddenly discovered a way to do 3D printing (probably using 3D technologies other than those in current use) that was anywhere near as fast as conventional manufacturing, the resulting revolution could quite possibly be as big as, if not bigger than the Internet.
A useful table showing the different 3D printing technologies in Wikipedia is included below:
|Extrusion||Fused deposition modeling (FDM)||Thermoplastics (e.g. PLA, ABS), eutectic metals, edible materials|
|Granular||Direct metal laser sintering (DMLS)||Almost any metal alloy|
|Electron beam melting (EBM)||Titanium alloys|
|Selective heat sintering (SHS)
|Selective laser sintering (SLS)||Thermoplastics, metal powders, ceramic powders|
|Powder bed and inkjet head 3d printing, Plaster-based 3D printing (PP)||Plaster|
|Laminated||Laminated object manufacturing (LOM)||Paper, metal foil, plastic film|
|Light polymerised||Stereolithography (SLA)||photopolymer|
|Digital Light Processing (DLP)||liquid resin|