- What Can’t You 3D Print?
- How Your Shopping Changes
With The Weather
- How Stores Follow
Every Step You Take
- What You Leave Behind
While Shopping Online
- What Will Replace
- The Birth of the
- What Is a Self-Service Government?
- Employee Training Isn’t
What It Used To Be
Consider the master glass blower. She excels at every step in her craft: dipping her long metal pipe into a hearth filled with molten glass, spinning and cooling it into a dense little button, then reheating it and blowing it out slowly like a balloon. For the construction of almost every practical type of glassware, this arduous human work has been replaced by an assembly line of molds and extruders. The segregation and automation of each task has lowered the cost of fabricating bottles and transformed a 2000-year-old art into a cutting-edge technology.
There is, however, one skill of the craftsman that has yet to be fully replicated, which is her ability to learn. Every time she spins form and structure into a glob of glowing glass, she gets better at it—faster, more consistent, more efficient. A machine in a factory does what it is told. An artist does what she knows.
Now, once again, the machines are catching up, and 3D printing is leading the way. As Dylan Reid, the chief executive at Matter Labs, a company that designs and manufactures small-batch runs for companies selling 3D printed products, explains it, machine learning is improving the way they make things at every step in the manufacturing process.
“In a closed-loop digital manufacturing system, like the one we're building, there's an opportunity not only to collect data more quickly and accurately, but to glean and share insights from data much faster, to dramatically accelerate the kind of learning that goes on in every factory using artificial intelligence,” says Reid.
In the old days, for example, efficiency in factories was often something that had to be measured manually or with a system that was separate from the machines that were doing the actual work. “At every chink in the manufacturing supply chain, a human being is interpreting their silo of data and they’re feeding that onto the next one,” says Reid. “It’s slow. It’s inefficient. It’s also not smart.”
When a product begins as a piece of data, and is constructed with self-monitoring 3D printers, it suddenly becomes possible to collect and analyze data throughout all of the design and construction phases of a product’s life. And when this information is fed back into the system, machine learning algorithms can be used to fine-tune the manufacturing process.
For example, says Reid, the most expensive part of manufacturing a product with 3D printers is the printing time itself. Having machines that can report back about their own efficiency and performance will allow companies to make crucial design modifications and choices about which printers are best at handling which projects.
Eliminating the waste of both time and materials from an intelligent supply chain has the potential to completely democratize manufacturing. In 2013, IBM sought to quantify the cost-saving potential of digital manufacturing. A team of researchers, led by Paul Brody, the company’s global industry leader for electronics, collected a handful of household items, disassembled them, compiled 3D scans of all the pieces, and created a model of how much money it would take to reproduce them with 3D printers and the next generation of flexible robotics.
“What we found was that within five years of last year, every single product we looked at was cheaper to make partly or wholly using 3D printing,” says Brody. “In the past, where you needed to have a big factory and make, say, a hundred thousand units a year in order to be cost-competitive...when you start using 3D printing and flexible robotics, you’re competitive on price at production runs of ten thousand.”
Ideally, according to Reid, Matter Labs will someday reach a point where all of the data that comes back from the production line—how long each build took, how the nozzles performed on the printers, whether the product held together well, or whether it collapsed under its own weight during construction—will be used to inform the design phase as well. Decisions that would otherwise require input from a salaried engineer will instead be made in fractions of a second by computer programs that retain insights from the successes and failures of every previous design.
“What’s cool is we don’t need to learn it as human beings. We can write scripts that automatically make it better. That’s significantly different and really important,” says Reid.
A Conversation with Paul Brody, Vice President & North America Leader, Mobile & Internet of Things at IBM
Q: People have called 3D printing the next Industrial Revolution. What makes it such a big deal?
It’s going to force everybody to rethink how they plan, manage, sell, configure, (and) design products. Every element, before and after manufacturing, is going to change with these changes in the manufacturing process.
All the things that used to be defined in hardware, that used to be driven by substantial physical constraints—like getting a mold made for a machine stamp or rearranging a physical production line—all of those things that were long lead time, high volume (and) fixed cost items are being turned into things that can be defined and executed in software at what is effectively a near-zero marginal cost.
And so the supply chain itself will be something we can redefine at will, primarily through software.
Q: How did your team arrive at, or at least confirm, those pretty bold predictions?
We did a research project and we wanted to ask ourselves, quantitatively, how big is this impact? And what we did is we bought a washing machine, a TV set, a phone, a hearing aid and a shaver. We went through these (five) items and we broke them down. Then we did a 3D scan of each part, and we figured out what it would cost to rebuild each item using 3D printing.
What we found was that within five years of last year, every single product we looked at was cheaper to make partly or wholly using 3D printing.
Q: So what does that mean for the future of manufacturing?
Today, most of these products have big centralized manufacturing facilities and then global distribution. When we modeled it out, what we found is that in case after case, manufacturing shifts from a global model to a regional model and then finally to a local model.
So in the most extreme case with the hearing aid, it’s literally more efficient within 10 years to make hearing aids in every local city than it is to have big global distribution centers. So what you’re really talking about is the localization—the democratization—of manufacturing.
In the past, where you needed to have a big factory and make, say, a hundred thousand units a year in order to be cost competitive for washing machines or shavers or hearing aids, when you start using 3D printing and flexible robotics, you’re competitive on price at production runs of five to 10 thousand.
Q: How big is that impact on business?
It’s completely transformational in terms of what kinds of companies can be competitive. I’ve grown up in an era where a lot of industries have been consolidating down to two to three players because that’s what the efficient scale supports—you just cant be competitive with less volume than that. Our estimate is that many industries that are affected by this could support between 10 to 20 times more efficient participants, which would make for a vastly more globally competitive set of industries.
For now, it seems, digital manufacturers are starting with small, relatively inconsequential products. Matter Labs specializes in jewelry. Other newly formed companies are promising to bring customized, 3D-printed insoles and earbuds to the market. But we are also beginning to see experimentation with 3D-printed prosthetic limbs and living organs. And in the future we will likely be flying on planes and washing our clothes in machines that are constructed at least partly from 3D-printed products.
As the capabilities of 3D printers increase, so too will the complexity of the designs and the load of the data that they handle. According to Brody, in order to standardize products and offer warranties, it will not be enough for companies to learn from their own mistakes. They will need to test their designs with sophisticated validation software that compiles information from many disparate sources, like databases of electrical engineering standards and Federal Aviation Administration regulatory guidelines.
“Test and verification is something that you don’t do every day," says Brody. "But when you need it you basically need the power of a super computer and the capacity to manage and store the data received from several apps and devices, whether on premises or in the cloud. What [IBM] has been offering for some time is a [...] service that allows companies to basically rent that supercomputing power to test and verify products in the cloud.”
In the future of digital manufacturing, being smart equates with saving money. And sometimes the smartest thing you can do is to know when you don’t have to learn something yourself.