Here is a number that explains a lot about why metal 3D printing has been slow to reach real production: NASA estimates that qualifying a single flight critical additively manufactured component currently takes more than 18 months. Not to design it, not to print it, but to prove it is trustworthy enough to fly.
NASA Marshall Space Flight Center just funded a program aimed squarely at that number. Phase3D, an inspection company based in Chicago, Illinois, has been awarded a role in a research effort to test whether a metal printed part can be qualified while it is still being built, rather than months after it comes off the plate.
What the program actually does
The plan is to generate more than 50,000 individually inspected build layers on a production scale EOS M300-4 quad laser system, which reporting on the award describes as one of the largest in situ inspection datasets ever assembled for metal additive manufacturing qualification research. The test parts are topology optimized Invar 36 brackets, chosen because they represent structural components that actually fly on spacecraft.
Phase3D is deploying two systems on the machine. Fringe Inspection uses structured green light projection to capture calibrated 3D height measurements after every powder layer and every laser exposure. Fringe Qualification aggregates that layer data across machines, facilities, and production programs into a single qualification workflow.
Together they produce a continuous measurement record of the entire build. The question is whether that record can be reliably correlated with post build CT scanning. If it can, manufacturers could set quantitative go and no go thresholds and make qualification decisions in real time. NASA calls that "born qualified" manufacturing.
The defects being hunted are the ones that stay invisible until post build testing: powder spreading irregularities, recoater blade interactions, layer shifts, melt pool abnormalities, spatter accumulation, delamination, and unexpected surface height variation. Because the measurements happen at every layer, an engineer can see when and where a deviation happened, not just that something somewhere is wrong.
How do I know a 3D printed part is actually solid inside before I bolt it in?
You either inspect it after the fact with CT scanning or destructive testing, or you buy from a shop that watches the build as it happens and can tell you which layer went sideways, and for most everyday parts the practical answer is a shop that controls its process and tells you honestly what the part is rated for. A finished print is opaque. You cannot see a void, a delamination, or a layer that did not bond by looking at the outside, which is exactly why aerospace has been stuck paying for CT scans and coupon testing on every batch.
For plastic parts the same logic applies at a smaller scale. Stiffness in a printed polymer part comes from perimeter count and wall thickness far more than from infill percentage, and a part printed with three perimeters looks identical to one printed with five. That is why we tell customers what a part is printed in and how, rather than just quoting a price. Our build and ship process page walks through it.
The evidence trail behind the award
This is not a cold start. In 2024, Phase3D validated its Fringe Research software on Ti64 for the US Air Force, printed on an EOS M 290, and on GRCop-42 for NASA, printed on a Colibrium Additive M2. The Air Force tests showed 81 percent of in situ anomalies correlating to CT detected defects for depressions larger than 47 micrometers. The NASA tests showed 83 percent correlation, and 100 percent for depressions larger than 42 micrometers. That work was funded through an AFRL Phase I STTR contract.
Those are strong numbers, but they came from test geometries on single laser machines. The new program exists to find out whether the same correlations hold on a four laser production system printing flight representative brackets. That is a harder problem, and it is the honest reason the research is still research. The work is structured around NASA Civil Space Shortfalls 1490 through 1494 and aligns with existing standards including NASA-STD-6030, NASA-STD-6033, and SAE AMS7032.
My printed part looked perfect and still broke. What went wrong?
Almost always the failure is layer adhesion, wall thickness, or print orientation, not the material, because a printed part is strongest along the layers and weakest between them, so a part loaded in the wrong direction snaps along a layer line while looking flawless right up until it does. The second most common cause is a geometry that needed supports and did not get them, leaving a sagging bridge or a drooped overhang buried inside an otherwise clean looking part.
The fixes are unglamorous. Reorient the part so the main load runs along the layers instead of across them. Add perimeters before you add infill. Thicken the wall at the stress riser. If the part sees heat, move up a material class rather than hoping a tougher print profile will cover it. And if you are recreating a part that already failed once, scanning the original and rebuilding it as real CAD lets you fix the weak spot instead of faithfully reproducing it, which is the whole point of our 3D scanning and reverse engineering service.
Why a spaceflight program matters to a shop with five printers
Nobody is putting structured light inspection on a desktop FDM machine this year. What is worth noticing is the direction of travel: the industry is moving from "print it and test it afterward" toward "measure it while it happens," and that changes what buyers are allowed to expect from a supplier.
At our scale the equivalent is boring discipline. Verify geometry in code before anything prints, check that a model is a single solid, assert that every cut lands where it was supposed to, weigh the finished part against the predicted mass, and treat photos of a first article as inspection data rather than as marketing. None of that requires a green light projector, and all of it catches the same category of problem: a defect you would never have found by looking.
Frequently asked questions
Can someone 3D print a replacement part if I only have the broken original and no drawings?
Yes. The original gets 3D scanned, the scan is rebuilt as a proper CAD model, and the part is printed from that model. Having the broken piece is usually enough as long as the mating surfaces are intact. If a critical dimension is destroyed, a caliper measurement of the part it bolts to fills the gap.
How do I know a 3D printed part will hold up under load?
Tell the shop the actual load, the direction it acts in, and the service temperature, then ask how the part will be oriented on the plate and how many perimeters it will get. A shop that can answer those three questions specifically is a shop that has thought about the load path. Printed parts are strongest along the layers and weakest between them, so orientation is the single biggest lever.
What does it cost to get one custom part 3D printed?
For a single small part in a common engineering plastic, the cost is usually a modest setup plus machine time, with no tooling charge at all, which is why one off printing beats molding at low quantity. The honest variables are part size, material, and how much design work is needed before printing starts. Send the file or the photos and ask for a flat number.
Do I need a CAD file to order a 3D printed part?
No. A STEP or STL file is the fastest path, but a physical sample, a dimensioned sketch, or clear photos with a few caliper measurements are all workable starting points. Expect the design step to add time to the quote if you are starting from an object rather than a file.
How we handle claims
The 18 month qualification timeline, the 50,000 layer target, the machine models, and the correlation percentages are verified research reported by additive manufacturing trade press covering the NASA Marshall award and Phase3D's 2024 Air Force and NASA validation work. We do not fill gaps with estimates. If a number is not in a published source, it does not appear here.
We also review products. If you make filament, tooling, print accessories, or shop equipment and want it evaluated and written up honestly, including the parts that do not work, reach out at info@dcadditivepros.com. We are the reviewer here, and reviews stay independent of whether a sample was provided.
DC Additive Pros is an independent aftermarket manufacturer in Rockville, Maryland. We are not affiliated with, authorized by, or endorsed by Phase3D, NASA, NASA Marshall Space Flight Center, EOS, Colibrium Additive, the US Air Force Research Laboratory, or SAE. Brand names are used nominatively for identification and reference only.