Screw, rivet, weld or adhesive: what a threaded joint is for

What the threads actually cost

Worth listing plainly, because each item is a whole engineering topic elsewhere on this site and it is easy to meet them one at a time without noticing they share a cause.

• A hole, and an open hole concentrates stress. Kirsch’s 1898 solution gives a factor of three at the edge of a circular hole in an infinite plate under uniaxial tension, measured against the remote stress. Be careful with that number: referred to the net section instead, it falls towards two as the hole grows relative to the width, so quoting “three” without saying which stress it is relative to means very little. What survives is the shape of the problem — an open hole is a place where the stress is higher than the average, and a joint that needed no hole never created one
• Preload is stored energy, and it leaks. Embedding and transverse movement both remove it, and how fast depends on things you may not control, which is why clamped length turns out to be a design variable rather than a leftover
• Assembly becomes a controlled process. A torque figure, sometimes a sequence, sometimes a re-torque — and torque is a poor proxy for the clamp force you actually wanted
• Cost per joint is parts plus insertion plus inspection, repeated for every unit you will ever build

What each alternative buys, and what it charges

The rivet row is where a tidy comparison usually goes wrong. A rivet needs a hole too, but it fills it, and setting the rivet develops both a clamping force and residual stress around the hole — so it is not a free hole and it is not a joint without preload. What it lacks is a thread that can unwind, and a number you can put on a drawing.

The question that decides it

Strength narrows the list before anything else does — an adhesive loaded in peel, a weld on a material that does not weld, a snap-fit at a temperature where the polymer relaxes are all ruled out on capability, not preference. But once more than one method survives that screen, strength usually stops deciding, and the question becomes:

• How many times will this joint be opened, across the product’s life?
• Is the clamp force doing a job? Sealing a gasket, carrying a sideways load by friction, keeping an electrical contact, holding a joint together against a fluctuating load. It does not have to be written on the drawing to be working — but if nobody can say what it is doing, that is worth knowing before arguing about torque values
• By whom, and with what tools? A joint opened once in a factory with a press is a different problem from one opened by a technician on a ladder
• What happens at end of life? If the answer is a recycler separating materials, that is a real opening, and it may be the only one

And then the two reasons you chose it start working against each other

A screw is chosen for demountability, and then the joint loosens, and the response is threadlocker. That is worth pausing on, because a permanent-grade adhesive on a screw gives back part of the property that was paid for.

Sometimes that is exactly right — a removable grade holds against vibration and still comes apart with hand tools. A permanent grade is not the end of the world either; manufacturers specify localised heat and hand tools to remove them. But it is worth noticing what has happened: the joint is being made harder to open, which was the thing the threads were bought for, and threadlocker only addresses one of the two ways a screw comes loose in any case. If the answer keeps drifting toward permanence, the question worth revisiting is whether it needed to open at all.

Where the screw genuinely wins

• Service access — anything a technician opens, anything under warranty repair
• Adjustment — alignment that has to be set after assembly, or reset later
• Mixed materials that cannot be welded, where an adhesive would need a cure the line does not have time for
• Regulated disassembly — batteries and other components that must be separable at end of life
• Low volume, where tooling for snap-fits or fixturing for welds does not amortise

In every one of those the threads are buying something specific. That is the test: name what the demountability is for before working through the specification, because the rest of that decision order assumes this one was answered.

This page covers step 1, the substrate. The whole order is substrate, thread, head, drive, finish, documentation, and why doing it out of order is rework rather than a tweak is in specifying a screw.

Common questions

When should you not use a screw?

When neither of the things threads are good at is being used. Threads let a joint be taken apart and put back with ordinary tools, and let the clamp force be set to a figure somebody chose. Most of the cost is incurred either way: a hole, a preload that has to be established and then maintained against embedding and movement, and a tightening process that has to be controlled. If the joint is assembled once and never serviced, and nobody can say what its clamp force is doing, the threads are being paid for and not used.

Does a rivet avoid the stress concentration a screw creates?

Not the same one. A rivet needs a hole, but it fills it, and setting the rivet develops a clamping force and residual stress around the hole — so the free-hole result does not transfer directly. Kirsch's 1898 solution gives a factor of three at the edge of an open circular hole in an infinite plate under uniaxial tension, relative to the remote stress; against the net section it falls towards two as the hole grows relative to the width. A riveted joint does carry clamp force. What it does not have is a thread that can unwind, or a preload figure you can specify and verify.

Why does welding or bonding avoid the hole problem?

They avoid the fastener hole, which is not the same as avoiding stress concentration. A weld has its own, at the toe and the root, and fatigue cracks in welded joints almost always start there — plus a heat-affected zone, distortion, and limits on which materials and geometries can be joined. An adhesive spreads load over an area rather than into a hole edge, and charges for it in surface preparation, cure time, temperature, moisture and creep limits, weakness in peel and cleavage, and a bond that is difficult to inspect once made.

Is using threadlocker a sign the joint should not have been screwed?

Sometimes, and it is a question rather than a verdict. A removable grade holding a joint against vibration is a reasonable design, and even permanent grades can be removed — manufacturers specify localised heat with hand tools. But adding one makes the joint harder to open, which is one of the two things the threads were bought for, so if the specification keeps drifting toward permanence it is worth asking whether it needed to open at all.

What decides between the methods?

Strength narrows the list first — an adhesive loaded in peel, a weld on an unweldable material, a snap-fit at a temperature where the polymer relaxes are ruled out on capability. Once more than one method survives that, it comes down to how many times the joint is opened over the product life, by whom and with what tools, and whether the clamp force is doing a job such as sealing or carrying load by friction. A joint opened once in a factory with a press is a different problem from one opened by a technician on a ladder, and end-of-life separation for recycling counts as an opening even if it is the only one.

References

• Stress concentrations at holes — the Kirsch (1898) solution
• TWI — fatigue cracks in welded joints initiate at the weld toe or root
• VDI 2230 Part 1 — Systematic calculation of highly stressed bolted joints