"The Achilles’ Heel of 3D Printing" by Peter Friedman

“The Achilles’ Heel of 3D Printing” by Peter Friedman
http://www.iijiij.com/2012/12/30/the-achilles-heel-of-3d-printing-015281

3D printing doesn’t scale to mass production. Oh, what a surprise. But why do people always focus on mass production when in reality the sweet spot for 3D printing is it’s flexibility? IMHO the future is having individual stuff - not mass produced “same for everyone”.

The technology changes faster than the for vocabulary used on manufacturing systems. The mass production should be rephrased as high volume manufacturing. It is almost certain that the future is personalized manufacturing. However for supply to meet demand the manufacturing should be in high volume. Either the printers should print fast or we should have thousands of printers to meet the demand. The second option looks more viable. However maintaining and keeping up to date of thousands of printers is not an easy task. I will write an article on this subject in my blog soon.

@Ibrahim_Kocaalioglu Agreed. We are talking about scalability in a different way - decentralized scalability that does NOT produce the same thing over and over again but rather similar stuff that gets altered a bit at every stage. When the results get shared in the open, we have something that can be called evolutionary production. With many small steps performed evrywhere to weed out the problems and thus create better resulkts in a very natural way. Not centrally managed and controlled but decentralized. Quite an interesting thought IMHO.

For it to work we need to make data exchange as simple as possible. Regardless of slicer, printer model etc we need to standardize the data exchange layer to be able to share and tinker with the specific values. Not sure if anyone is working on that - but it should be done ASAP.

Standardization of data is not easy. Data should be optimized for the type of printer, its firmware and type of material used. Else the output quality would suffer.
Additionally the data model should be complex enough to handle the future improvements on 3D printing, such as multi-material printing, multi-stage printing.
The goal is: The digital designs should yield, similar results on all printers.

Since there are many files that work on most machines, I think the bottleneck will be design automation, getting the custom product to the consumer means someone, or something has to design it digitally.

I want to comment on your phrase “Since there are many files that work on most machines”. Most of the printers are based on RepRap therefore they may result in similar output. However as the variety among printers increase it will be a problem. Multi-material, multi-color printing is hardly handled at the moment. Optimizing a digital design for 3D printing is another topic. I have short article on that http://iboman2.blogspot.com/2012/05/3d-design-tricks.html. Additionally we should have design customization wizards, where any user can easily modify them for their needs. Example putting their names on a product.

I don’t believe it will be a problem at all to produce manifold models in a common language. I can right now, for free, via open software, convert to the generally recognized “.dwg” equivelant from 10 years ago. I agree the wizars are necessary, that is the design automation component I was talking about.

I agree, as the production of designs become less of a problem. Coming up a design that is customizable for every users become a problem. Else if the designs are hard to customize for individuals, it not meaningful to produce them one by one :slight_smile:

Well, there are two easy answers, and one less easy answer to the issue of personalization, though they apply at different levels, and depend to a degree on the method of 3d design.

First, and most immediately relevant, but least easy - modify the design yourself, as the end-user - this is the current status of most 3d printing end-users who have their own printers. They either design something from scratch, or download an existing design and tweak it.

The first of the easier answers goes hand in hand with the above - parametric designs, in OpenSCAD or a similar CSG based language, where the creator of the design has pre-specified the readily adjustable values and set them up as obvious tweakable values in the design file. This could be extended, but probably won’t be for a while, as there are multiple variants competing to supplant OpenSCAD at the moment, and given OpenSCAD’s limitations, it is probably not a bad idea. Still, once the field settles and a winner emerges, even if that is OpenSCAD unchanged, the obvious extension would simply be to codify a standard for communicating the reasonable type and range of allowed values for these parameters and a description of each (XML, structured comments, something of the like), and then have a separate program or webservice, given the source file, provide a constructed interface with just those knobs and widgets exposed that were described by the source’s author, and a preview window reflecting in real-time changes to those knobs. This is theoretically equally applicable to a parametric cad such as Alibre, but would probably only survive as such when tied to a service such as Shapeways where they could control the file formats to be used.

The second easy answer is easy in its application to the problem, but somewhat less easy in its implementation, and mostly applies to services that work with raw STL files, as the most common lowest denominator, and simply involves, as @Ibrahim_Kocaalioglu mentions, ‘putting their name or logo’ on a product, by providing text or simple images, a display of the STL file with a means to select a surface region, and convolving the STL with a simple extrusion or intrusion of the text or images. In this way, basically any 3d shape that sports a flat or at least simply curved (i.e. non-discontinuous, smoothly differentiable) surface would be ‘customizable.’

What is more, to one degree or another, these already exist!

See, for example, for parametric designs:

and

and

http://i.materialise.com/creationcorner/yourkees

As to the second method, I don’t know of any services that have an easy online convolution method for any STL built-in, but there are already multiple online CAD apps, so it is certainly not beyond doable.

And if they can be done with these online ordering services, then there is no particular reason why similar simple ‘wizard’ type interfaces could not be integrated with open source 3d sites. For example, from a fellow here on G+, http://tonybuser.com/cloudscad, which already parses OpenSCAD files for parameters and creates widgets for adjusting them and then downloading the resulting STL file. No reason not to expect something similar could be done with CoffeeSCAD or others, and eventually I do expect something like that to be integrated into Thingiverse and similar sites.

@Howard_C_Shaw_III Your comment is simply wonderfull.I cannot find anything to add on. My recommendation for the 3D printing world is to come up with a website where the objectives are clearly defined (ex: to be able to manufacture accesories, toys…). The presentation of current status (on hardware, raw materials, firmware, software…) and ongoing works. You may add more to this… I want to be able to see the overall progress. I want to see points where I can contribute. There are lots of effort put on this subject. If we can syncronize and increase the synergy then we can reach our goal faster.

Why is the fact that a general solution for creating any object of any complexity is not as fast as a highly specific machine that creates only one type of object an Achille’s Heel? Oh no, I can’t create glass jars by the tens of thousands and open that glass jar business I really wanted to get into… better dismantle my RepRap… There is no 3D manufacturing, only 3D prototyping.

Every system has its own Achilles’ Heel. It is important to show a systems strengths and weakness. Therefore one can choose a system based on the need (Low volume high customization - 3D Printing, High volume no customization - injection molding). Therefore no need to dismantle your RepRap :slight_smile:

Every system has pros and cons, but not every system is doomed to failure because of a hidden flaw. Achille’s Heels lead to something’s downfall, not a well informed decision.

I hate to nit-pick on word meanings, but when it’s in the title of an article it should be used properly.

This article is saying, “nobody should own a food processor because potato chip companies will supply you with all the chopped vegetables you need”. Or rather, “potato chip companies will not invest in armies of food processors to replace their current machines so therefore food processors have no place in the world.” It’s a straw man fallacy (or a huge misuse of the term Achille’s Heel).

I certainly agree with you on your comment. Sorry it wasn’t that clear from your previous post :frowning:
It attracts more attention when you use a term such as “Achille’s Heel” compared to Pros and Cons on a title of an article :slight_smile:
Writers always do that :slight_smile: