Bolt Torque and Preload: Why Tightening Torque Is Only a Means to Clamp Load

Nobody actually wants torque. What a bolted joint needs is preload — the clamping force that holds the joint faces together, carries the working load and keeps the bolt itself from fatiguing. Torque is simply the most convenient way to produce preload on an assembly line: you cannot put a force gauge inside a tightened joint, but you can measure how hard the wrench is turning.

That convenience has a price, and the price is accuracy. Torque measures resistance to turning, not tension in the bolt — and most of the resistance is friction, which varies with every surface the bolt touches. This article works through where the torque actually goes, the standard formula that connects torque to preload, and why the same bolt, at the same wrench setting, can end up dangerously loose or stretched past yield depending on conditions the torque table never mentioned.

Where the torque goes: the 50 / 40 / 10 rule

A widely used rule of thumb for a plain, unlubricated steel joint breaks the applied torque into three parts:

• About 50% is consumed by friction under the bolt head or nut face, where the rotating part drags against the clamped surface.
• About 40% is consumed by friction in the thread flanks (the thread geometry itself — pitch, fit and surface — is covered in our metric thread guide).
• About 10% is left to do the useful work: stretching the bolt and generating preload.

Read those proportions again: the quantity you want is the smallest slice of the quantity you control. A 20% swing in friction — a different finish, a trace of oil, a rougher washer face — barely moves the wrench reading but moves preload a great deal, because it is subtracted from the 90% that friction already takes. This is not a defect of torque wrenches; it is the physics of the method, and it is why torque values can never be separated from the friction conditions they were written for.

The formula: T = K · F · d

The standard short-form relationship between tightening torque and preload is:

T = K × F × d

• T — tightening torque (N·m)
• F — target preload (N)
• d — nominal bolt diameter (m, in consistent units)
• K — the "nut factor", a dimensionless coefficient that bundles all the friction in the joint into one number

K is where the honesty of the formula lives. It is not a material constant and it is not printed on the bolt. For plain, dry steel surfaces it is commonly taken as roughly 0.20; lubricate the same joint and K can fall toward 0.15 or below; rough, dry or damaged surfaces push it higher. Every one of those numbers is an estimate for a family of conditions, not a property of your specific joint — which is why serious specifications state the surface and lubrication condition alongside the torque, and why copying a torque value from a generic chart, without its conditions, is copying only half the instruction.

A worked example: M10 class 8.8, and how friction moves the answer

Take the M10 class 8.8 bolt from our strength concepts guide: tensile stress area 58.0 mm², nominal yield load 37.1 kN. A common design practice is to assemble to roughly 70% of yield — high enough to keep the joint clamped under working load, low enough to keep margin against the scatter of the tightening method. That gives a target preload of about 26 kN. (These are illustrative figures from the standard designation values, worked here to show the method — a real joint is designed from its full specification.)

With K = 0.20 (plain, dry) and d = 0.010 m:

T = 0.20 × 26,000 × 0.010 ≈ 52 N·m

Now hold the wrench setting at 52 N·m and change only the friction:

Same bolt, same torque wrench, three different joints. The middle row is the one that breaks bolts: an assembler adds oil "to help things along", the wrench still clicks at 52 N·m, and the bolt is quietly stretched to the edge of permanent deformation — or past it, into the failure modes our strength guide describes. The bottom row is the one that comes loose in service and gets blamed on vibration.

This scatter is intrinsic to torque-only control — ±25% or more around the target is commonly cited even with a calibrated wrench and consistent parts. Joint designers live with it by keeping the target comfortably below yield and comfortably above the load the joint must resist. It is also why critical joints move to torque-plus-angle or direct tension measurement: once the head seats, angle stretches the bolt by a geometric amount that friction cannot distort.

Too tight and too loose: the two failure directions

Over-tightening spends the bolt's strength at assembly, before the joint ever sees service:

• The bolt yields or fractures during tightening — sometimes immediately, sometimes as a delayed crack in a stretched, embrittled part.
• Threads strip in the bolt or the tapped hole — the risk our thread guide flags where engagement and fit were marginal to begin with.
• The clamped parts are damaged instead: gaskets extruded, soft materials crushed, flange faces bent — preload the joint cannot hold is not reserve strength, it is damage.

Under-tightening leaves the joint to work the way it was never meant to:

• The faces slip under load, and the bolt — designed to be a stiff spring in a clamped stack — starts taking the full alternating load in bending and tension. That is the classic path to fatigue failure.
• The joint leaks, frets, or backs off: a bolt below its working preload has little resistance to self-loosening under vibration.

Both directions start from the same root cause: the torque was set, but the preload — the thing the design actually specified — was never where it was assumed to be.

Why a specification states the class and the torque condition

A complete tightening instruction has three legs, and removing any one of them makes the other two meaningless:

1. The property class sets the ceiling. Preload targets are a fraction of yield, and yield is the class: an M10 in 8.8 and the same M10 in 12.9 have different yields, different targets and different torques. Choosing the class is a design decision in its own right — our property class guide walks through it.
2. The friction condition sets the conversion. Finish and lubrication decide K. This is where the fastener manufacturer enters the tightening conversation: black oxide, Dacromet, colour zinc and zinc plating — the four finishes we produce — do not present the same surface to the nut face and thread flanks, so a torque written for one finish does not transfer unchanged to another. We state this as a principle rather than a table, deliberately: the coefficient of a specific finished, lubricated, assembled joint is something to establish for that joint, not to copy from a blog.
3. The torque is then an instruction for one defined combination of the first two — plus the method (hand wrench, calibrated tool, torque-plus-angle) and the condition of the mating surfaces.

When a drawing or assembly specification gives you all three, tightening is reproducible. When it gives torque alone, the missing conditions will be invented on the shop floor, one assembler at a time.

A note on where our side of this ends: we manufacture the bolt to its class, material and finish — 40Cr as the standard material route to the higher classes, across cold-headed M5–M24 and machined M8–M64 production, verified by composition, metallography, hardness, dimensional, contour and roughness inspection, with reports available to accompany an order. The preload target for your joint belongs to the joint's design; the class and finish on your drawing are what let that target be hit consistently. If the tightening condition affects what you need from the part — a finish chosen with friction in mind, a tolerance class that keeps thread friction consistent — that is worth raising at inquiry stage, not after the first assembly trial.

Common mistakes

• Copying a torque chart without its conditions. Charts assume a finish and a lubrication state. Change either and the chart is a different joint's answer.
• Adding lubricant without reducing torque. Lubricant lowers K, so the unchanged torque produces far more preload — the worked example above is this mistake, quantified.
• Treating the wrench click as proof of preload. The click proves torque was applied; preload is inferred, with the full scatter of the friction conditions that day.
• Tightening a high-class bolt to a low-class habit. Higher classes need their own targets; "a bit more for safety" past yield is how 12.9 bolts get quietly ruined at assembly.
• Reusing a bolt that has been into yield. A permanently stretched bolt has spent its clamp reserve; its next tightening starts from a different length and an unknown preload.

FAQ

Is more preload always better?
No. Preload has a correct band: high enough that the joint faces never separate or slip under the maximum working load, low enough to stay clear of yield under the worst friction scatter at assembly. Above the band you are damaging the joint at installation; below it, service does the damage.

Why do torque tables for the same bolt size disagree with each other?
Because they assume different things: different K values, finishes, lubrication states and target fractions of yield, often printed in footnotes nobody reads. A table without stated conditions is incomplete — ask what K and what surface it assumes before using it.

Should torque change if the finish changes?
In principle, yes: the finish is part of the friction condition, so the same target preload converts to a different torque. How much different is joint-specific and has to be established for the actual combination — which is why the finish and the tightening requirement should be specified together, on the same drawing.

Can you tell us what torque to use for your bolts?
We can tell you the class, material and finish of what we supply, and verify them by test — the inputs a tightening specification is built from. The torque itself depends on your joint: its materials, surfaces, lubrication and target preload. Send the drawing and the assembly conditions through the custom fasteners page and the specification can be settled before production, rather than discovered at assembly.

HyHaven manufactures custom fasteners to your drawing or sample, in property classes 4.8–12.9, from 2,000 pieces. Request a quote.