The Hardest Part of Ownership
We already wrote the rule down once: not never-fail, but predictable failure. Every certified composite structure in the sky today earned its place by that standard — proven strong, then proven legible, one bounded, inspectable step at a time, over decades, by the two most conservative institutions in aviation. HELIOS is where we find out if we meant it.
Airbus's HELIOS program is investigating weight-optimized CFRP fuselage structures using rivet-free joining — bonded longitudinal seams, novel frame-to-skin connections. The pitch is familiar and, on its own terms, reasonable: fewer fasteners means fewer stress concentrations, less part count, less mass. Lightweighting hides in the interfaces, and the big gain comes from redesigning the interface, not just the part. That's true. It's also incomplete, because it answers one question — how do we make this joint lighter — without asking the one that actually governs a thirty-year airframe: how will we know when this joint is dying?
The lesson we already paid for, in metal
Aviation has been here before, in a different material, at a terrible price. The De Havilland Comet broke apart three times in 1953–54 — fatigue cracks starting at rivet holes near the window corners. Aloha Airlines Flight 243 lost eighteen feet of fuselage skin in 1988 when corrosion-driven fatigue cracked a whole row of rivets that a maintenance program failed to catch in time. Japan Airlines 123, the worst single-aircraft accident in history, traced back to a botched repair that funneled all the joint's load through one center rivet row instead of two — fatigue cracks nobody found until 520 people were dead.
Every one of those disasters has the same shape: a known, physically bounded flaw type — a fatigue crack at a fastener hole — that inspection should have caught and, for one reason or another, didn't. The lesson the industry drew wasn't "eliminate the fastener." It was "build the inspection regime the fastener deserves." Eddy-current, torque checks, crack-length trending across service intervals — decades of accumulated capability, built specifically around a flaw type that grows slowly and shows itself if you know where to look.
The flaw that doesn't show itself
A bonded joint's worst failure mode isn't a smaller version of a rivet crack. It's a different kind of thing. The kissing bond — two composite surfaces in full mechanical contact, transmitting compressive load, with next to zero adhesive strength between them because of a contamination film, a cure defect, moisture at the interface — can be nearly invisible to the standard inspection tool. Ultrasonic C-scan finds bonds by detecting a gap or an impedance mismatch, and a kissing bond has neither: the surfaces are touching. It can carry flight loads for years and then simply stop, with no crack-length trend, no warning inspection window, no schedule to trust.
That's not a harder version of the rivet-hole problem. It's the failure mode damage-tolerance certification was built to prevent, arriving in a form the primary detection method may not be able to see. Which is why, historically, bonded primary joints in aircraft have almost never gone in without a sparse row of mechanical fasteners riding alongside — not for strength margin, but as a crack-arrest fallback, the thing that keeps a silent disbond from unzipping the way JAL 123's repair did. Remove that row in the name of weight, and you haven't removed unnecessary hardware. You may have removed the one component whose entire job was to fail safely when the primary joint didn't.
What seven thousand X-rays teach a designer
I have my own testimony on this, not a line from a report. My early composite years were spent in layup clean rooms — good air handling, the layup and debulk process a strange kind of zen, concentration and relaxation running at the same time. But as the work pushed into more complex parts, my later years moved into the CT and UT stations. Complex parts need complex inspection technology, and over roughly twenty years I put somewhere north of seven thousand X-rays and contour plots in front of my eyes, hunting for flaws — by my own rough count, about twenty percent fewer than the full-time inspection specialists I worked alongside. I never sat for a Level II license, but I was doing the work.
What that many hours in front of a flaw teaches you isn't what you'd expect. It doesn't make you a better stress engineer. It makes you a better simplifier — because after enough time spent trying to find a defect a part is hiding, you stop asking "can this carry the load" and start asking "can this be inspected at all." The best designs I ever signed off on weren't the ones that squeezed out the last percent of margin. They were the ones a technician could actually see into.
And R&D doesn't give you the luxury of a clean answer. You build maybe ten percent over your production run for spares, and when a part comes back questionable and you're out of spares, you don't get to demand a fresh one. You pick the best of the worst, run the test, and keep your fingers crossed. That's not a flaw in the process — it's what happens when the part you built can't tell you clearly enough what's wrong with it. A joint that's easy to inspect doesn't just save money downstream. It saves you from that exact moment, the one where the whole test program is riding on a part you already knew was borderline.
Some inventions are short-term solutions
None of this makes rivet-free bonding a bad invention. It might be exactly right for a five-year-life component, a part swapped on schedule regardless of condition, an airframe whose ownership horizon never reaches far enough to meet the failure mode's blind spot. The mistake isn't the technology. It's the application window. A joining method that trades a legible, decades-long failure story for an illegible one belongs on products short enough that the illegibility never gets tested — not on primary structure carrying passengers for thirty years.
Don't make maintenance the hardest part of ownership
Pull the aerospace off it and the principle is one anyone who's owned anything for a long time already knows. What determines whether you regret a purchase was never day-one performance. It's whether year fifteen is a phone call or a demolition. For a fleet of airliners that isn't a metaphor — MRO cost over an airframe's life dwarfs what it cost to build. A part optimized in isolation looks like a win on delivery day. Whether it is a win gets decided by the system it lives inside for the next three decades — the inspection regime, the depot schedule, the line mechanic who either can or can't find the problem before it finds the passengers.
That's the whole argument, in the end. We didn't spend seventy years building the eyes to catch a rivet crack just to trade them for a joint whose worst failure was designed, however unintentionally, to stay invisible. Predictable and measured failure was never a slogan. It was the price of trust. Any invention that can't pay it doesn't belong on a product built to last thirty years — no matter how much weight it saves on the day it leaves the factory.
Herbert Roberts, P.E. is a licensed professional engineer with 30+ years in aviation research and development across two companies, and has spent eight years analyzing accidents for attorneys under his P.E. license