NASA Is Testing 'Born Qualified' Metal 3D Printing — 50,000 Layers at a Time

Key takeaways

NASA’s Marshall Space Flight Center has awarded Chicago-based Phase3D a role in a program to test whether metal 3D printed parts can be qualified while they are being built, rather than months after.
The program will capture more than 50,000 individually inspected build layers on a production-scale, quad-laser EOS M300-4 — one of the largest in-situ inspection datasets ever assembled for metal AM qualification.
The test article is a set of topology-optimised Invar 36 brackets, built with an unnamed US aerospace prime and space propulsion manufacturer.
The problem being attacked is time and waste: qualifying one flight-critical metal AM part currently takes more than 18 months, and Phase3D cites industry rejection rates as high as 30%. The stated goal is a 2–3× faster qualification timeline.
If layer-by-layer measurements can be reliably correlated with post-build CT scans, manufacturers could define real-time go/no-go thresholds — what NASA calls “born qualified” manufacturing.

Metal additive manufacturing has a dirty secret that has nothing to do with printing. The printing part largely works. What does not work is proving that it worked. Every flight-critical metal part that comes off a powder bed fusion machine still has to survive a long tail of CT scanning, destructive testing and paperwork before anyone will bolt it to a rocket — and that tail, by NASA’s own estimate, runs longer than 18 months per component.

NASA’s Marshall Space Flight Center wants that number cut by a factor of two or three, and it has just handed Phase3D a program to find out whether that is possible.

What the program actually does

Phase3D will deploy two products on the job. Fringe Inspection is the hardware layer: it projects structured green light into the build chamber and captures calibrated 3D height measurements after every powder spread and every laser exposure. Fringe Qualification is the software layer that aggregates all of that data across machines, facilities and production programs into a single qualification workflow.

The measurements are aimed at the failure modes that normally stay invisible until a part is cut open or scanned: powder spreading irregularities, recoater blade interactions, layer shifts, melt pool abnormalities, spatter accumulation, delamination and unexpected surface height variation. Because every layer is measured, engineers can see not only that a deviation happened but exactly when and where in the build.

The test case is deliberately unglamorous: topology-optimised brackets in Invar 36, the iron-nickel alloy prized in space hardware for its near-zero thermal expansion. Brackets are representative structural components, they fly in quantity, and they are exactly the kind of part where an 18-month qualification cycle is economically absurd. The build partner is a confidential US aerospace prime and space propulsion manufacturer, according to 3D Printing Industry and 3DPrint.com.

Why the machine choice matters

The EOS M300-4 is not a lab curiosity. It is a large-format, four-laser production system, and it is increasingly the platform aerospace and defence contractors install when they move from demonstrator parts to actual serial output. Multi-laser machines are also, not coincidentally, where in-situ monitoring gets hard: four beams writing into overlapping regions of the same layer produce interactions that a single-laser research rig never shows you.

That is the real step up here. Phase3D’s earlier results were promising but narrow. In work funded through an AFRL Phase I STTR contract in 2024, the company validated its Fringe Research software on Ti64 for the US Air Force on an EOS M 290, and on GRCop-42 for NASA on a Colibrium Additive M2. The Air Force tests showed 81% of in-situ anomalies correlating to CT-detected defects for depressions larger than 47 µm; the NASA tests showed 83% correlation, rising to 100% for depressions larger than 42 µm.

Those are strong numbers — on test geometries, on single-laser machines. The 50,000-layer dataset is designed to answer whether the same correlations survive contact with a quad-laser production system printing flight-representative hardware.

“Qualified as it is built”

Phase3D founder and CEO Dr. Niall O’Dowd framed the pitch bluntly in the company’s announcement: “For decades, qualifying a 3D-printed part for spaceflight has meant months of destructive testing and CT scanning, an approach that does not scale. With Fringe Inspection, the part is qualified as it is built. Every powder layer, every weld, every anomaly is captured in calibrated, defensible data.”

He added that “real-time inspection is the missing piece in the [AM] ecosystem. Powders, lasers, machines, and process parameters have all matured. What has been missing is a way to prove, in real time, that the part you built is the part you designed, on every layer, every time.”

It is a confident claim, and worth holding at arm’s length until the data lands. Correlation between a height map and a CT scan is not the same thing as a certification authority accepting that height map in place of the CT scan. The program’s actual deliverable is narrower and more useful than the marketing: quantitative go/no-go thresholds that NASA, its primes and its suppliers can point at.

Why it matters beyond NASA

The work is structured around NASA Civil Space Shortfalls 1490 through 1494 — a formally identified set of capability gaps covering in-situ monitoring, process qualification and the qualification of complex additively manufactured geometries. It also aligns with NASA-STD-6030, NASA-STD-6033 and SAE AMS7032, the standards that already govern how metal AM parts get cleared for flight.

That standards alignment is the part with teeth. Qualification frameworks are the slowest-moving, highest-leverage layer of industrial additive manufacturing. Every serious aerospace AM story of the past year has really been a qualification story — Airbus opening its polymer AM specification to a second qualified ecosystem, Pratt & Whitney and GKN taking additive into the F135 engine case, ESA grinding through sample after sample on orbit. Printing the part was never the bottleneck.

If in-process data can carry even part of the evidentiary load currently borne by post-build inspection, the economics shift for everyone downstream — defence, energy, medical — not just for spaceflight hardware. And if the correlations fall apart at production scale, that is worth knowing too, before another generation of machines ships with monitoring dashboards nobody is allowed to certify against.

FAQ

What does “born qualified” mean in 3D printing?

It means a part is certified during production rather than after it. Instead of printing a component and then proving its integrity through CT scanning and destructive testing, sensors capture enough calibrated data during the build that the part’s quality is established the moment it finishes.

How does Fringe Inspection detect defects?

It projects structured green light onto the powder bed and measures the distortion of that pattern to build a calibrated 3D height map after every powder layer and every laser exposure. Deviations in that height map — depressions, layer shifts, spatter, recoater streaks — flag where a defect is likely to have formed.

Why is Invar 36 used for the test brackets?

Invar 36 is an iron-nickel alloy with an unusually low coefficient of thermal expansion, which makes it valuable for structures that must hold their geometry across the temperature swings of orbit. Topology-optimised brackets in Invar 36 are representative of real structural spaceflight hardware.

Does this replace CT scanning?

Not yet, and that is precisely what the program is testing. The 50,000-layer dataset exists to measure how closely in-situ readings track CT results at production scale. Only if that correlation holds can regulators and primes justify substituting real-time thresholds for some post-build inspection.

NASA Is Testing ‘Born Qualified’ Metal 3D Printing — 50,000 Layers at a Time