How the AI Opportunity in Additive is for Parts That Were Never Meant to Be Printed
Identifying candidates for 3D printing among legacy parts is an under-appreciated challenge. AI reveals the extent to which parts in service warehouses can be transitioned to digital production.
EI Note: We're ecstatic to welcome Peter Zelinski as a guest contributor to EI! As an engineer who went into media, Peter is one of the world's most respected voices on manufacturing technology; Peter brings unparalleled insights at the intersection of physical manufacturing and digital transformation.
What This Means For You:
Deep Industry Knowledge: Benefit from Peter's years of hands-on experience exploring manufacturing facilities around the globe.
Cutting-Edge Analysis: Discover how data, ML, and AI revolutionize parts manufacturing workflows.
Forward-Looking Perspectives: Understand the technologies reshaping tomorrow's production environments.
📌 Find Peter's Exclusive Content
Peter's thought-provoking articles will be featured prominently under our new 'Guest' tab. Read his first piece below!
Additive manufacturing and artificial intelligence seem to offer a natural fit to one another, and presumably, that fit should be found in process optimization. Laser powder bed fusion, the chief additive process for metal part production, offers so many different variables potentially meaningful to the output (related to the laser, its intensity, its path; ambient conditions; the material and its state; and so) that AI should be the tool for learning and steering toward faster and better production. I first wrote about this idea in action in 2018 in research at the Colorado School of Mines, and exiting Penn State additive manufacturing professor Timothy Simpson still saw this use of AI as still the opportunity for additive yet to be realized.
But could it be the most powerful application of AI for additive manufacturing actually is to be found in parts that were never intended to be 3D printed?
When it comes to service parts—that is, legacy parts provided to service or maintain products or machinery in the field—the simple question of which of these parts is a candidate for additive manufacturing is challenging. And AI can solve this challenge now.
The Mirror Image of Prototypes
Service parts represent an enormous AM opportunity. In a sense, service parts are the mirror image of the original 3D printing application, prototyping. A prototype is the part made in low quantities when a part or product is young, so young it has not yet gone to market. A service part is the part made in low quantities when the part or product is old, so old that perhaps it has been superseded by newer models in the marketplace, but the OEM still must be prepared to support continuing users.
In the service part application, additive manufacturing is competing against real estate. That is, it is competing against the warehouses otherwise needed to organize and store lots of parts that might be ordered for some service or maintenance need. These parts originally likely were stamped, molded, machined or cast. Continuing with a process such as this means the service parts need to be made in quantities and stored in warehouses. But 3D printing, a digital process needing little setup and no tooling, offers the promise to make the part as it is needed and ship it soon after printing. A small footprint for 3D printing capacity can replace the large square footage needed to store a reserve of all the different parts the printers might otherwise be able to produce on demand.
That is the promise, anyway. The obstacle is getting there. Not every part is an additive candidate. Out of a system of warehouses storing 50,000 part numbers, which parts are candidates for translating to on-demand 3D printing, and is it worth all the engineering hours of evaluating legacy parts to find out?
Finding the Additive Wins in Legacy Drawings
That AI could replace this engineering team is one of the suggestions of capability announced by software provider 3YourMind. Its AI-enabled Technical Drawing Analysis is said to be able to “evaluate spare part manufacturability, determining whether parts can be produced [via] powder bed fusion, material extrusion, vat polymerization” and other additive processes. In other words, legacy part drawings fed to this system can be analyzed not according to the manufacturing possibilities of their time, but according to the options now. And additive is the most striking difference between that past and now.
The full list of additive processes in the 3YourMind release suggests and addresses another challenge in the service parts application. “Additive manufacturing” is, in fact, the broadest category of part-making operation we have, covering many different digital layering processes that fill different application needs. Few AM professionals are expert in most AM processes. But if a company can discover through AI that a large share of its legacy parts can be made to specification through a single AM process (vat photopolymerization, say), then that company has discovered something powerful. Presumably, a single industrial 3D printer, or maybe two for redundancy, can meet the need for service parts that maybe an acre of warehouse space is addressing today.
Photo: AI-generated.
Peter Zelinski is a writer who reports on technology advances in discrete part manufacturing. His views are his own.