The Air Force Reverse-Engineered Its Own F-35 Parts
A maintenance squadron at Hill AFB couldn't get a training part delivered on time. So they took a real one apart, scanned it, rebuilt it in CAD, printed it — and broke it. Then they fixed it and broke it again, until it held three times the load for a month straight. That workflow is not exotic. It's the same one we run in a van in San Diego.
Figures from the U.S. Air Force Materiel Command release, 28 May 2026.
Everybody who has ever fixed anything knows this feeling. The part exists. The part is made. Somebody, somewhere, has a bin of them. And you are going to wait eleven weeks for it, because that's just how long it takes, and no amount of phone calls changes the number.
We hit this constantly. A discontinued hotend shroud. A Creality bed-leveling bracket that's out of stock everywhere. An Elegoo vat lid nobody imports anymore. Somebody's Prusa has a cracked extruder idler and the replacement is three weeks out, which means their print farm is three weeks out.
Turns out the U.S. Air Force has the exact same problem, except the machine on the other end of it costs somewhere north of $80 million.
What actually happened at Hill
809th Maintenance Support Squadron · Hill AFB, Utah
In late May 2026, Air Force Materiel Command published a short release about a lab most people have never heard of. The 809th Maintenance Support Squadron runs something called the RAPID Lab at Hill Air Force Base in Utah. Their problem: F-35 maintainers have a mandatory, safety-critical task they have to practice — installing the canopy's emergency severance system — and the training frames they practice on weren't showing up. Supplier deliveries were late. Training stopped.
So the lab reverse-engineered the canopy frame itself. A team from the 309th Aircraft Maintenance Group disassembled a real frame so it could be scanned. RAPID's engineers modeled it, printed it, and iterated. The result: additively manufactured canopy frames now feeding training sites, with a second variant in progress so all three F-35 models are covered.
Reverse Engineering · Advanced Manufacturing · Prototyping · Innovation · Design
Worth noting that "reverse engineering" is the first word in the name of an official Air Force capability. This isn't a skunkworks side project. It's a funded function with a flight chief.
Read the headline carefully
Honest trade-off #1
The version of this story going around social media is "Air Force 3D prints F-35 parts." That's true, but it's doing a lot of work in a small number of words, and the distinction matters if you're going to take any lesson from it.
These are training frames. They are not bolted to a jet that flies. They live on a stand in a training facility so that Airmen can put their hands on the real geometry and practice a procedure they cannot afford to get wrong on a live aircraft. The Air Force validated the printed frames for that mission specifically — the reverse-engineered geometry and the printed materials were checked against the training requirement, in coordination with Joint Strike Fighter engineers at the Ogden Air Logistics Complex.
What's genuinely interesting is the door they left open. The same release says the lab is now hunting for small, high-impact components that can be safely printed for ground use, or installed on the aircraft when the cognizant engineer approves it. That phrase — cognizant engineer — is the whole ballgame. Somebody with authority and liability has to sign. High-performance polymers are what make the conversation possible at all, particularly for cockpit environments, where you're fighting temperature limits and off-gassing rules at the same time.
The first one broke. That's the useful part.
Load paths, torque, and why iteration isn't failure
Here's the detail I'd underline if I were teaching this: an early printed component fractured under torque when it was mounted to the training stand.
Not "the print failed." Not "the material was wrong." It cracked when a real load went through it in a real fixture — which is exactly when parts crack, and exactly why you cannot evaluate a functional print by looking at it. The team's response was to strengthen the structure and redistribute the load paths until it comfortably beat the requirement. Then they proved it: the printed frame carried three times the canopy's weight, inverted, for more than a month.
That last bit is the professional move. A one-hour pull test tells you about ultimate strength. A month-long inverted hold tells you about creep — the slow deformation thermoplastics do under sustained load, especially when it's warm. Creep is the thing that quietly ruins printed brackets six weeks after everybody declared victory, and almost nobody tests for it.
When a printed part cracks under torque, the fix is usually geometry before material. Reroute the load so it runs along the layers instead of trying to peel them apart, add a generous fillet at the stress riser, and thicken the wall count before you thicken the whole part. Swapping to a more expensive filament to rescue a bad load path is the most common and most expensive mistake in functional printing.
Wait for the vendor, or scan it yourself
Honest trade-off #2
Reverse engineering isn't automatically the right answer. It's the right answer in a specific set of conditions, and knowing which is most of the skill.
| Order the original part | Scan it, model it, print it | |
|---|---|---|
| Lead time | Weeks to months, entirely outside your control | Days — and the delay is your own iteration, which you can attack |
| Cost | Whatever the supplier says it is | Material plus machine time, plus real engineering hours up front |
| First attempt | Usually correct | Usually wrong. Budget for three |
| Changing the design | Not an option | The entire point — you can fix what annoyed you about the original |
| Certification | Comes with the part | Falls entirely on you |
| Best fit | In stock, cheap, or flight/safety-critical | Obsolete, backordered, tooling, fixtures, training aids, one-offs |
Notice that the Air Force picked a target that sits squarely in the right-hand column. Training aids and tooling are the ideal first reverse-engineering project because a failure costs you a redo, not a life. Start there. Everybody who has gotten good at this started there.
The workflow, honestly described
Six steps — order matters
Get a known-good reference
You cannot reverse-engineer a broken part and expect good geometry. Hill disassembled an intact frame first. If yours is snapped, find an unbroken one, borrow one, or measure the mating surfaces on the assembly it bolts into.
Scan — and know your tolerance before you start
A structured-light scanner gets you into the tens of microns. A phone photogrammetry app does not. Neither is wrong; they're answers to different questions. Decide what fit actually needs to be accurate before you pick the tool, because that decision sets the budget.
Rebuild it, don't sculpt it
The scan is reference, not geometry. Bring the mesh into Fusion 360 or your CAD of choice and rebuild the part with real sketches, real constraints, real datums. A cleaned-up mesh is unparametric and unfixable. A rebuilt model can be adjusted 0.15 mm at a time when the fit is tight.
Print the cheap version first
Fit check in PLA. Confirm every hole, boss, and mating face before you spend nylon or polycarbonate on something with a hole in the wrong place. Cheap iterations are the reason this is faster than the vendor.
Load it until it breaks
Then look at where it broke and why. A part that survives your test bench for an hour has told you almost nothing about how it lives for a year. Leave it loaded. Leave it in the heat. That's what the month-long inverted hold was for.
Get someone else to check it
Hill ran their scans and design work past JSF engineers before calling it done. You probably don't have a Lightning Support Team, but you do have somebody who'll tell you the truth. Use them.
The part nobody said out loud
Technical data rights
Worth flagging, because you'll see it in the comments on every repost of this story: the Air Force's release doesn't name a single vendor. Not Lockheed Martin, not anybody. Aerospace trade coverage and defense commentators read that omission as pointed, framing the whole exercise against a long-running fight over technical data rights on the F-35 program. The Air Force itself made no such claim — the release is about a training gap and how a lab closed it.
The underlying issue is real and much older than the F-35, though. When the government buys an aircraft without buying the drawings, it eventually finds itself unable to make a part for a plane it owns. It happened with the B-2. It happened with the B-52, where engineers at Tinker AFB reverse-engineered an anti-icing gasket for an engine designed in the 1950s because the original supply chain had simply evaporated. Reverse engineering in defense sustainment is usually less about spite and more about the fact that nobody has produced that part since the Carter administration.
Either way, the direction is clear. Stitzer, RAPID's flight chief, put the plan in four words: "We're just getting started."
Why this lands differently in San Diego
Local context
San Diego isn't an abstract observer here. MCAS Miramar hosts two Marine F-35C squadrons — the Black Knights of VMFA-314 and the Tomcats of VMFA-311 — and the county's aerospace and defense supply chain runs from Kearny Mesa machine shops through Rancho Bernardo and Poway out to the contract manufacturers in Otay Mesa. If you work anywhere in that ecosystem, you already know the sustainment math: the part is late, the schedule is not, and somebody has to bridge the gap.
What's changed in the last few years is that the bridging tools got cheap. A modern enclosed printer running PA6-CF or polycarbonate costs less than a decent used truck. Handheld structured-light scanners that cost $40,000 in 2015 now cost less than a laptop. The gap between what Hill AFB did and what a competent small shop in North County can do is now mostly process discipline — scanning method, CAD rigor, honest load testing — rather than equipment.
That's the actual takeaway. They didn't have magic. They had a scanner, CAD, a printer, and the patience to break their own part twice before trusting it.
Questions people actually ask
Did they print a part that flies on an F-35?
No. The canopy frames are training articles, validated for training use. The Air Force did say it's identifying small components that could be printed for ground use or, with engineering approval, installed on aircraft — but that's future intent, not what shipped here.
Why does the Air Force have to reverse-engineer a jet still in production?
Because owning an aircraft and owning its technical data are two different purchases. Contracts written to be cheap up front often leave the government without the drawings, which means only the original manufacturer can make replacements. The F-35 case is contested and I'd point you at the primary sources rather than the hot takes, but the general pattern is well documented across programs.
Can I reverse-engineer a replacement part for my car, appliance, or drone?
For non-structural, non-safety parts — trim clips, brackets, housings, knobs, fixtures — yes, and it's one of the most satisfying things you can do with a printer. For anything where failure means somebody gets hurt, no. Suspension components, load-bearing mounts, pressure vessels, anything in a braking or restraint system: buy the real part. The Air Force has a cognizant engineer for exactly this reason and you probably don't.
What material would you actually use for something like this?
For functional load-bearing prints, PA6-CF or PA12-CF for stiffness and heat resistance, polycarbonate blends when you need toughness over rigidity, PPS-CF or PEI-class polymers if you're chasing high temperature and low off-gassing. All of them want a heated chamber, a hardened nozzle, and dry filament — nylon that's been sitting out is not nylon anymore. For most people, ASA or PETG-CF is the realistic sweet spot between "actually works" and "actually printable on the machine you own."
Does Dreaming3D do this kind of work?
Yes — reverse-engineering a discontinued or backordered part is a regular request. We can work from a physical sample, a broken part plus its mating assembly, or dimensioned photos, and produce both the CAD file and the printed part. And because we're mobile, we come to your shop across San Diego County rather than asking you to ship a fixture across town.
What's the part you've been waiting on the longest — and is it something a scanner and an afternoon in CAD could have solved by now? Tell us in the comments. If it's interesting enough, we'll try to reverse-engineer it and write up what happened, including the parts that break.
Sources: Air Force Materiel Command, "RAPID Lab delivers additively manufactured F-35 canopy frame," 28 May 2026; F-35 Joint Program Office statements, June 2026; Aerospace Global News and Simple Flying coverage of the release; Air & Space Forces Magazine reporting on Air Force reverse-engineering and additive manufacturing programs. Photo suggestion: the official U.S. Air Force images by Cynthia Griggs are public domain and available through Defense Visual Information Distribution Service.