Bolt preload — the bolt barely feels the load, until the joint lets go

The same joint, the same bolt, the same load cycling on and off. The only thing you are changing is how tight the bolt was done up. Drag upward to tighten, down to slacken; the arrow keys do the same. End is tightened to near proof load; Home is barely finger tight. Watch the orange bracket — it is everything the bolt feels. Sound is off until you turn it on.

What you are looking at

Tightening a bolt stretches it and squeezes the parts it holds together. The bolt ends up pulling with some force and the members end up pushing back with exactly the same force, and nothing moves. That force is the preload, and the control here is nothing but the preload.

Now pull on the joint. The instinct is that the bolt takes what you apply, because the bolt is what is holding it together. It does not. Pulling the joint apart relieves the squeeze in the members at the same time as it stretches the bolt a little more, and the members are far stiffer than the bolt — a short fat column of steel against a long thin one. Most of what you apply is spent un-squeezing the members. The bolt takes only its share of the stiffness, which for this joint is about a sixth.

That lasts exactly as long as there is squeeze left to relieve. When the compression in the members reaches zero the joint opens, and from that moment the members are carrying nothing whatever and the bolt is carrying all of it. On the plot the line stops being nearly flat and becomes the diagonal — the bolt now feels every pound you apply, and more importantly it feels every pound you take away again.

Bolts almost never fail by being pulled apart once. They fail in fatigue, from the swing rather than the size, which is the orange bracket. Keep the joint shut and the swing is a sixth of the load; let it open and the swing is the whole of it, several times a second, for as long as the machine runs.

The two things worth taking away

A broken bolt is usually a loose-joint problem, not a weak-bolt problem. Fitting a stronger bolt at the same torque raises the preload hardly at all — the grade changes what the bolt can survive, not how hard somebody pulled the spanner. If the joint was opening before, it opens now, and the stronger bolt fails in the same place for the same reason. The fix is more preload, a bigger bolt, or a stiffer joint.

Past the point where the joint stops opening, tightening further does nothing for fatigue. The bracket stops shrinking, exactly and completely, and the rest of the tightening goes into the bolt's mean stress and into margin against everything else — slip, loosening, a bigger load than the one designed for. Those are good reasons to keep tightening to a proper figure. Reducing what the bolt feels from the load is not one of them, past that point.

What is not verified

Stated plainly, because a piece whose value is trustworthiness should not overclaim. The first list is the part that would change how you read the result. The second is scope — true, worth saying, and not load-bearing.

• No engineer has reviewed this. If you are one and would check it, REVIEW.md says what would actually help and is meant to cost about fifteen minutes. The load factor comes from the standard stiffness relation with Wileman's published member stiffness, and the built-in checks confirm the arithmetic — but that check is the author's own, not a fastener engineer's.
• The load is concentric and axial. It is applied along the bolt and at the bolt. Real joints are usually loaded off to one side, which prises the joint open on one edge at a far lower load than this drawing shows, and is the commonest way a joint that "cannot" open does.
• Member stiffness is an empirical fit. Wileman's expression is a curve fit to finite-element results for a particular geometry. The frustum methods in VDI 2230 disagree with it, and with each other, by enough to move the load factor noticeably — so treat a sixth as the right order, not a number to design to.
• Preload is drawn as a number you simply choose. Getting it is the hard part: torque is a poor proxy, with most of it lost to friction under the head and in the threads, and scatter of plus or minus a quarter is normal for a plain torque wrench. That scatter matters most exactly where this piece says it does — near the threshold.

Scope, for completeness. True of the model, and none of it changes the answer above.

• Nothing here relaxes, loosens, or bends. Embedding, gasket creep and vibration all lose preload over time, and this joint holds whatever it was given forever. The bolt is also assumed to stay elastic.
• One bolt. Real joints have several, sharing unevenly.

Commissions

The subject changes; the method does not. If you have a failure mode, a mechanism, or a process that people keep getting wrong because the diagram cannot move, that is the shape of problem this suits. You get one file, the checks that hold it honest, and the notes explaining every decision including the ones that were reversed. Commissions start at £4,000 and take three to four weeks from an agreed brief. yinxueshi1017@gmail.com

Not commissioning anything? Still worth an email. If your version of this failed differently — other geometry, other numbers, a mechanism this piece leaves out — tell me in one line. Corrections are the main thing that decides what gets built next.