Replacing airframe panels and inspecting hard-to-reach parts would be easier with screws, I presume. So why are rivets used?
Rivets need to press the two parts together AND to transmit shear. The pressure results in friction which is responsible for a considerable part of the load transmitted between two riveted panels. In order to hold up to the maximum possible amount of shear, the rivet needs a smooth surface. A thread would make it much more vulnerable due to the notch effect of the thread.
Since skin panels and flanges vary in thickness, a LOT of different screw sizes would need to be kept on stock so that every combination of panel thicknesses is covered with a fitting screw which has just the right amount of smooth shank length. Rivets, on the other hand, will be hammered to the right length during manufacture, so only a few sizes are needed to cover all those panel thicknesses.
Historically, metalworking tools were much less precise than today. For optimum shear transmission, the bolt must sit tightly in its hole. In the past, this could not be done with screws, but only with rivets which grow thicker when set in place. And since aviation regulation authorities are VERY conservative, the easiest to certify joints are those proven rivet joints.
Also, repairability is not so bad with rivets, especially if you can reach only one side. The rivets get drilled out (with a slightly larger diameter drill) and after inspection the parts are riveted together again. And if you need the lowest mass for a given strength, again rivets win over screws. The difference might be small for the individual rivet, but it adds up for the whole airframe.
- They are cheap and simple, especially in the volumes needed.
- They are impossible to open, a useful safety consideration. You only have to check that they are properly inserted from the beginning. They won't shake loose.
- Flush rivets are aerodynamically good since they can be made flush with the fuselage, which is difficult (or possible?) with bolts I believe.
- (Blind) rivets are great for complex structures, since you can apply them entirely from the one side. You might be able to use screws if you pre-inserted them and held the thread and rotated the nut, but the nuts would still be outside the skin in the end.
- Furthermore, I believe a protective coating is applied between many (commercial) aircraft panels for moisture and corrosion protection, so you wouldn't want to take them off to start with.
- There's very seldom any reason to start taking apart aircraft. If it's bad, you can rivet over it or drill out the rivet.
- Rivets can be made a bit smaller than bolts, saving weight. (Not entirely confident on this one)
Replacing airframe panels and inspecting hard-to-reach parts would be easier with screws, I presume.
Every time you make a hatch it introduces a weak spot into the loaded airframe, such as the skin covering the wings and the cabin, it has to be reinforced to bring it up to the required strength, adding weight. That's why these are kept to a bare minimum.
That's not to say aviation doesn't use screws or has its own versions. One screw derivative is called a hi-lok and is a permanent joining like a rivet. It works much like a classic bolt but has a calibrated diameter between the top hex nut and the bottom nut, so it snaps off when it reaches the correct torque. That way it's got a safety guarantee being attached well.
Vibrating loose is the major concern.
The panels are generally made from two sheets of aluminum skin (.030 - .060 thick) with a paper honeycomb bonded between them and solid aluminum rails from 1" - 3" wide at the edges. This makes the panels incredibly resistant to flex or compression spread over the surface and extremely lightweight. (You could easily, however shove a pencil through the middle of it (not recommended to try in-flight))
Using screws would a) easily distort the surface of the panels (consider the way a screw is applied (torsion, which causes it to pull itself in) vs a rivet (which only applies force against itself until near-full compression is reached) or b) be highly susceptible to vibrating loose ( particularly since the skin is too thin to contain even a single thread )
It is possible to use screws into the solid edges, but you still have the risk of vibrating loose, plus the manufacturing costs of threading holes or using capture-nuts.
What differentiates a riveted joint from a bolted joint is that the rivet expands when bucked to completely fill the hole into which it has been installed. Bolts/screws require that the pattern of holes be drilled very close to perfect locations, but still the required tolerance of the hole over the fastener size means that the joint must either be secured by sufficient clamping force to hold by friction, or one must accept that the joint must move to load the fasteners in shear. Of course, that means when the load reverses (as do many airframe loads) the joint can only load in shear after it has moved into the new load direction. Also to BE in shear loading, the bolt pattern has to be perfect, otherwise, loading of ALL of the applied shear joints can not be shared until the fastener or the structure has deformed sufficiently to allow the tolerance of hole location to equate and shared between all fasteners.
The next issue is weight, and it is another big one. Bolts are much heavier than rivets, and there can be thousands of them even on a small airplane.
Then there is load transfer: To have an actual shear loaded joint with a threaded fastener, the threads should not be pressing on the side of the hole, so that means a washer under the nut so the shank can load against the hole - and usually a washer under the bolt head, as there is a radius under there that should not go into the hole unless there is enough thickness for a countersink for it to fit in.
Alternative: These problems can be (and are) dealt with in some situations by using countersunk screws, that take up the tolerance for shear loading by cocking slightly in their bore - but not really meant for serious shear loads but can give a good fit on removable panels.
Someone mentioned glued joints: common as dirt. The first Genav aircraft to do so was Jim Bede's BD1 - known to the rest of us a American Aviation Yankee (AA-1). Bonded joints are what made the skin of all AA1/A/B/C and AA5/A/B and AG5B as well as AG7 Cougars so clean that they fly a lot faster than their contemporary riveted competitors on the same power.
When comparing rivets and bolts/nuts for fastening sheet metal, riveting comes out as the superior method.
Rivets are forged
From this site:
Riveting is a forging process that may be used to join parts together by way of a metal part called a rivet. The rivet acts to join the parts through adjacent surfaces.
The rivet is inserted in the pre-drilled hole, and then the ends are pressed or hammered together in order to expand the rivet into the hole. At the end of this process the rivet fills the hole snugly and completely, and the contact area helps in carrying stresses from one plate into the other. Clamping the plates together enables friction to take part of stress transfer as well. The forging process actually strengthens steel rivets.
Bolts don't fit
Bolting down two metal sheets works a bit differently: a hole is drilled in the sheets and a bolt inserted, the bolt usually having a smaller diameter than the hole. The bolt itself does not contribute to stress transfer, that is purely done by the friction from clamping. It is possible to drill a tightly tolerance hole and insert a bolt that fits exactly, but this is an expensive process that requires perfectly perpendicular drilling and care in handling. Expensive and slow, best done on straight surfaces, not the bent shape of an aircraft fuselage.
Steel and aluminium corrode
Bolts are loaded under tension and are best made from steel: aluminium bolts are stripped of their threads easily and stretch a lot. But if we use steel bolts on aluminium sheet, we get galvanic corrosion. We must either use steel on steel or aluminium on aluminium, and aluminium bolts are just not very good.
Nuts unfasten from fatigue
Nuts work themselves loose under alternating loads, and an aircraft has a lot of that. Every landing the wing slams down, bends upward again at take-off. The fuselage is pressurised at altitude, then de-pressurised at approach. The engines and hydraulic pumps introduce vibrations into the aircraft skin.
In order to stop the nut from unfastening, a locking washer needs to be installed or a compound such as loc-tite must be applied - at every one of the thousands of bolts that would fasten the aircraft skin plates.
Rivets are airtight
A point missed by the previous answerers: the fuselage is a pressure vessel, drill it full of holes for bolts, even tightly fitting ones, and it will leak like a sieve. Fill the hole up with forged aluminium and it will remain airtight.
Why is riveting not used in more applications then
Bolting metal sheets together has the advantage that the fastening can be undone without any specific tools. A rivet needs to be drilled out, a nut is just loosened. Great for performing maintenance, but in permanent constructions like bridges there is no need to unfasten bits of it. Long life aircraft do need their skin panels to be replaced every once in a while, but even that is a bit of a non-event with aluminium rivets, they drill out easily.
Why is riveting not used more then? It used to be, providing strong structures for bridges for instance. The method that largely replaced it is welding:
- Only one welder is required, instead of at least two riveters for each side of the rivet.
- No need to heat rivets to a red-hot state in an oven and haul them up to the point of application.
- Welding is a line joint, riveting a point joint: stress tubes need to dodge the rivet holes and are concentrated around them, while they flow over the complete width of the weld into the next plate.
Bonding is even better
The best joining method for metal sheets is the one that joins surfaces: glueing, although for construction purposes the proper word is bonding. Fokker pioneered the use of bonded aluminium sheets in the F-27 and F-28: no holes, perfectly smooth skin, large stress conversion area with relatively low stresses. Not an easy method to implement: large autoclaves are required for the bonding process, and it is not possible to visually inspect the quality of the bond, like it is with a weld. For that purpose an ultrasonic bond tester was developed which would reveal bonding imperfections.