An airliner fuselage contains something in the order of a million rivets. Each one is a hole drilled through the skin, which is to say a stress concentration deliberately introduced into a pressure vessel, repeated a million times.

A welded fuselage would have none of them. It would be lighter, smoother and, joint for joint, stronger. Nobody builds one.

The reason is the best idea in aircraft structures, and it is not about strength at all.

Rivets do not make it stronger

Take the strength argument at face value, because it is correct. A properly executed weld between two pieces of metal is continuous parent material. A riveted lap joint is two overlapping sheets clamped by fasteners, with load transferring through the fasteners in shear and through friction between the faying surfaces. Per unit of joint, the weld wins.

Now put a crack in each.

A crack in a pressurised fuselage does not sit still. It grows a little on every cycle as the cabin pressurises and depressurises, and the longer it gets the faster it grows, because the stress at the tip rises with crack length. Beyond a certain length it stops being a fatigue crack and becomes a fast fracture running at a fair fraction of the speed of sound through the material.

In continuous metal there is nothing in its path. In a riveted structure the crack reaches a line of rivet holes and a frame or a tear strap, and the discontinuity that a designer would otherwise hate is now the thing that blunts the crack and dumps its energy. The skin is divided into panels, and a failure in one panel is meant to stay in that panel.

That is the whole idea: fail-safe structure, and its modern successor, damage tolerance. The assumption is not that the aircraft is free of cracks. The assumption is that it has them, that they will grow, and that the structure must be arranged so they are detectable long before any one of them becomes critical.

The aircraft that proved it

On 28 April 1988 a Boeing 737 of Aloha Airlines lost roughly eighteen feet of upper fuselage at 24,000 feet, from just behind the cockpit back to the wing. One flight attendant was swept out and died. The aircraft landed at Kahului.

It is usually told as a story about corrosion and an ageing fleet, and it is. But the structural detail is the one that matters here, and one of the commenters under the original post got it right.

That aircraft’s lap joints used an early cold-bonding design: the skin panels were glued with an epoxy-impregnated scrim cloth, and the bond was supposed to carry the pressurisation load, with rivets largely locating the panels. The scrim could wick moisture into the joint. When the bond let go, the load it had been carrying went into the rivet line instead, and fatigue cracks started at the knife edge of the countersunk rivet holes, at many holes at once.

That is called multiple-site damage, and it is the specific thing that defeats a fail-safe design. One crack runs into a tear strap and stops. A row of small cracks at adjacent rivet holes can link up into one long crack that arrives at the tear strap with far more energy than the strap was designed to absorb, and on that aircraft the tear straps had disbonded too.

The point for this argument is what happened next: the aeroplane landed. A cabin was opened to the atmosphere at 24,000 feet and the structure that remained still carried the loads. That is damage tolerance doing precisely what it exists to do, at the far edge of what anyone would design for.

And you could not weld it anyway

The crack-arrest argument is the interesting one, but there is a blunter reason sitting underneath it: the metal will not take a weld.

Airframe skins are typically 2024 in the T3 temper, an aluminium-copper alloy, and 7075, aluminium-zinc-magnesium-copper. Both get their strength from precipitation hardening, a heat treatment that puts fine particles into the metal in a controlled way. Both are classed as unweldable by fusion methods, for three compounding reasons.

The weld pool solidifies over a range of temperature, and the last liquid to freeze sits in films between grains. Shrinkage pulls those films apart, which is hot cracking, and it happens as the joint cools, before it has carried any load. Around the weld, the heat-affected zone gets a thermal cycle nobody designed, so the careful precipitation structure is coarsened or dissolved and the temper is gone. And 7075 in particular is susceptible to stress corrosion cracking, so a zone with altered metallurgy and locked-in residual stress is precisely where a crack will start in service.

Add the geometry. A fuselage skin can be under a millimetre thick. Melting a joint into something that thin, repeatably, across kilometres of seam, with no way to inspect the result except by dragging equipment over every inch of it, is a manufacturing problem nobody wants when the alternative is a fastener with a specification, a visible head and a gauge to check it.

Where aircraft are welded

Plenty of an aeroplane is welded. Engine mounts, pylons, trusses, firewalls, landing gear components, ducting, tanks: thicker parts, different alloys, steel and titanium, and places where a crack has somewhere to go that is not a pressure hull.

And there is one aircraft that makes the whole argument by exception. The MiG-25 is roughly 80 per cent nickel steel, 11 per cent aluminium and 9 per cent titanium, and its steel structure was put together with spot welding, automatic welding and a great deal of hand arc welding. Mikoyan would have preferred titanium. They could not make it work, for a reason that should sound familiar by now.

A heavy welded steel aeroplane, designed for short high-speed sprints rather than tens of thousands of pressurisation cycles, is a different engineering problem from an airliner that must survive sixty thousand flights. Both answers are correct for their question.

The thing that is quietly changing

Several commenters raised friction stir welding, and they are right to.

FSW is not welding in the sense above. A rotating shouldered tool is plunged into the joint and dragged along it; friction softens the metal and the tool stirs the two sides together plastically. Nothing melts. Because nothing melts, there is no weld pool to hot-crack and no solidification structure to go wrong, which is exactly why it works on the alloys a torch cannot touch.

It has flown. The Eclipse 500 was the first aircraft to use friction stir welding in primary structure, in the cabin, aft fuselage, wings and engine mounts, certified with the FAA.

Note what is missing from that sentence. It is an argument about cost, quality and weight, which are the arguments a manufacturer makes. It is not an argument that the joint is more damage tolerant, and that is still the question a large pressurised fuselage has to answer. FSW has spread through aerospace in panels, stiffeners, tanks and spacecraft structures, and it has not yet replaced the riveted lap joint on a large airliner.

The bigger shift is sideways. On a carbon fibre fuselage the question stops being rivets or welds at all, because the barrel is laid up in one piece and the failure mechanics are completely different. The 787 and the A350 did not solve the riveting problem. They left it.

Why the original claim is nearly right

The post that prompted this put it in one line: rivets do not make the joint stronger, they make the failure survivable. That is a good sentence and it is essentially correct.

Two things are worth adding to it.

The first is that crack arrest is not the only reason, and on a factory floor it is probably not the first reason. You cannot fusion weld the alloys, you cannot reliably melt a joint into sub-millimetre skin, you cannot inspect the result cheaply, and you cannot repair it in a hangar overnight the way a mechanic drills out a bad rivet and squeezes a new one. Damage tolerance is the elegant answer. Manufacturability and maintenance are the ones that actually decide it.

The second is that the fail-safe promise is conditional, and Aloha is what it looks like when the condition is not met. Crack arrest works if cracks arrive one at a time at a structure that is still bonded and still capable of absorbing them. A row of small cracks at adjacent holes, in a joint whose adhesive has let go, defeats it. Which is why the regulatory response to 1988 was not a change of material but a change of inspection: find the cracks.

The riveted fuselage is not a compromise engineers are waiting to escape. It is a deliberate choice to build a structure that fails slowly and visibly rather than one that fails well until it fails completely.

The post

This came round from the account gaugehow, and Instagram labels it as AI-assisted content, which is worth knowing before quoting it. The engineering holds up. The comment thread under it is unusually good: readers correctly raised friction stir welding, the loss of heat treatment, the ease of sectional repair, and the bonded-joint detail on the Aloha aircraft. All four are in this article because they were right.

Sources: National Transportation Safety Board report on Aloha Airlines Flight 243; TWI; NASA; Assembly Magazine (Eclipse Aviation); Airplane Academy; Wikipedia entries for the Mikoyan-Gurevich MiG-25 and Aloha Airlines Flight 243; gaugehow on Instagram; Afterburner reporting