Bolted joints: why the preload does the work

A bolt is a stretched spring

Tighten a nut and two things happen at once. The bolt stretches — only a fraction of a millimetre, but it is steel pulled hard, so the tension is large — and the parts between the head and the nut are squeezed together. The bolt's tension and the clamping force on the parts are the same force, called the preload.

Everything a bolted joint does well comes from that preload. A loose bolt rattles, wears its hole and breaks from fatigue. A properly preloaded bolt holds the parts as one piece, and the external load has to fight the clamp before it can do much to the bolt.

Why preload protects the bolt

Pull on a preloaded joint with an external force P and both springs respond. The bolt stretches a little more, so its tension rises; the clamped parts relax a little, so their compression falls. Because they move by the same small amount, the load splits in proportion to their stiffness: the bolt gains C·P and the parts lose (1 − C)·P, where C = k_b/(k_b + k_m) is the joint constant.

The parts are usually much stiffer than the bolt, so C is small — often around 0.2 for a steel bolt clamping steel. The bolt then feels only about a fifth of the external load, and only a fifth of any load that fluctuates, which is why preloaded bolts survive vibration and cyclic loads that would break a loose one.

That protection ends when the clamp runs out. Once (1 − C)·P reaches the preload, at P = F_i/(1 − C), the parts separate, and from then on the bolt carries the whole external load on its own.

Stiffness decides the split

A bolt's stiffness is its threaded and unthreaded lengths acting as springs in series, set by the area and the grip length. The clamped parts are stiffer than their thickness alone suggests: the clamping pressure spreads outward from under the head in a cone, and that cone of material is what resists.

Anything that softens the parts raises C and sends more load into the bolt — aluminium instead of steel, a gasket, a stack of thin plates. Anything that softens the bolt lowers it: a longer bolt, or a necked-down shank, which is why bolts meant to survive fatigue are sometimes made deliberately slender between the head and the thread.

How much preload, and what 8.8 means

The markings on a metric bolt head are its ISO 898-1 property class. In 8.8, the first number times 100 is the nominal tensile strength, 800 MPa, and the second number times 10 is the percentage of that at which it yields, 80%. A 10.9 bolt is nominally 1000 MPa with a yield at 90% of it; a 12.9 bolt, 1200 MPa and 90%.

Each class also has a proof strength, a stress the bolt must carry without permanent stretch. A common recommendation, from Shigley's machine design textbook, is to preload to about 75% of the proof load for joints that will be taken apart again and about 90% for permanent ones. The preload is only useful if it stays: bolts that are too lightly tightened lose their clamp to vibration and settling long before they are overloaded.

Torque is a rough guide to tension

Most joints are tightened with a torque wrench, and the torque is related to preload by T = K·F·d, where d is the bolt diameter and K is the nut factor. K lumps together all the friction in the thread and under the head, and friction takes most of the effort — commonly quoted as around 90% of the torque, leaving about a tenth to stretch the bolt.

So the same torque gives very different preloads depending on the finish and lubrication. Shigley's table runs from K = 0.3 for a plain black finish to 0.12 with anti-seize compound: the same torque on a lubricated bolt can give more than twice the preload of a dry one, sometimes enough to break it. Even with the right K, torque control alone typically scatters preload by around ±25%. Where preload really matters, engineers tighten by turning the nut through a set angle, use washers that show the tension, measure the bolt's stretch ultrasonically, or use bolts designed to be tightened just past yield.

Separation and slip

Many joints carry shear rather than tension, and a preloaded joint can carry it by friction alone: the slip resistance is the friction coefficient times the clamping force on each sliding interface. Steelwork standards grade the faying surfaces for this — from about 0.5 for grit-blasted steel down to 0.2 for surfaces left as rolled — and friction-grip joints are designed so that they never slip at all.

External tension eats into that clamp, so a joint pulled and sheared at once slips sooner than either load alone would suggest, and a separated joint has no friction left. Clamp is also lost over time as surfaces settle into each other, which is why wheel nuts are commonly re-checked after the first 50 to 100 kilometres following a wheel change.

Check yourself

Read next