What a Torque Wrench Cannot Tell You

A torque wrench measures torque. It does not measure the thing you actually care about, which is how hard the bolt is clamping. The gap between those two is larger than the tool's clicking suggests.

Most of the torque never reaches the bolt

Of the torque applied to a fastener, the large majority is spent overcoming friction — under the bolt head or nut face, and in the thread flanks. Only a modest share goes into stretching the bolt, which is where the clamp force comes from.

That is why the nut factor dominates the torque calculation so completely. The same M10 bolt at the same 25.2 kN preload wants 50.45 N·m dry and 30.27 N·m waxed — and the difference is entirely friction that never became clamp force.

What that means for repeatability

Because friction varies between fasteners, a torque figure applied consistently produces preload that is not consistent. Published estimates of the scatter in preload from torque control alone are wide — it is routinely treated as one of the least precise common engineering methods, which is why critical joints often use something else.

Sources of the scatter, all ordinary:

• Surface condition. Plating, oxidation, machining marks and residue all change friction.
• Lubrication, intended or accidental. Oil from handling counts.
• Reuse. Thread condition changes after the first tightening.
• Dirt and damage. A gritty thread can absorb a large share of the applied torque.
• Speed and smoothness of tightening, which affect how friction behaves.

The methods that measure something closer

Where the joint matters, other approaches reduce the dependence on friction. Angle control tightens to a snug torque and then a specified further rotation, which relates more directly to bolt stretch. Yield control detects the point where the bolt starts to yield. Direct measurement — ultrasonic length measurement, or load-indicating washers — measures the stretch or the load itself.

Those exist precisely because torque is a proxy. If a specification calls for torque-plus-angle rather than a torque figure, that is why, and the angle part is not optional.

Looking after the wrench

A few practical points that undermine otherwise careful work:

• Click wrenches drift and need periodic calibration. An uncalibrated wrench is an unknown, not a conservative one.
• Wind the setting down after use if the manufacturer says to — leaving a spring compressed at a high setting shortens its life.
• A torque wrench is not a breaker bar. Using one to loosen fasteners is a common way to damage the mechanism.
• Extensions and adaptors change the applied torque if they extend the wrench's effective length. A crow's foot at 90° does not; one in line does.
• Pull smoothly to the click and stop. Continuing past it, or jerking, both defeat the measurement.

And the boundary worth restating

The calculator on this site estimates a torque from a target preload using a general model. It is for understanding the relationship — why lubrication matters, why small bolts have small numbers, why a spec has a condition attached to it.

It is not a source of torque figures for a joint that matters. Wheels, brakes, steering, suspension, lifting equipment, pressure containment and anything structural take their figures from the manufacturer or the designing engineer, along with the tightening sequence, the condition assumed and any re-check interval. Where the two disagree, the specification wins — and a bolt whose spec you cannot find is a bolt to look up.

Sequence and passes

On a multi-bolt joint, tightening one fastener changes the load on its neighbours, so the order matters as much as the figure. A flange tightened round the circle in sequence ends up with a very uneven preload distribution; the crossing patterns exist to spread it.

Staged passes do the same job across time — a first pass at a fraction of final torque, then a second, then the final — letting the joint settle progressively rather than deforming under the first bolt to be fully tightened.

Where a specification gives a sequence and a set of passes, those are part of the specification and not optional extras.

Relaxation, and why a re-check exists

Joints lose preload after tightening. Surface asperities flatten, coatings bed in, and gaskets compress — the losses are largest in the first hours and largest on joints with soft or numerous interfaces.

That is why some specifications call for a re-torque after a period or after a heat cycle, and why the instruction to re-check wheel fasteners after a short distance exists. Skipping it is skipping part of the tightening procedure rather than performing an optional extra.

What to write down

If you are doing your own work, recording the figure used, the condition assumed and the date takes seconds and answers the question that always comes up later: was this bolt tightened dry or lubricated, and to what?

Without that, a joint being re-worked a year on has an unknown history, and the safest assumption is the conservative one — which usually means new fasteners.