Torque Is a Proxy for Preload — and a Poor One
Torque is not what holds a bolted joint together. Preload does — the tension in the stretched bolt clamping the parts. Torque is just the most convenient thing to measure, and it is a poor proxy, because most of what you apply is spent on friction rather than on stretch.
The model
The industry short form is T = K · F · d: torque equals a nut factor, times preload, times nominal diameter. The bolt torque calculator works backwards from a target preload, which is itself a fraction of the bolt's proof load.
Preload comes from F = preloadFactor · proofStrength · As, where As is the tensile stress area — the effective cross-section that actually carries load, which is smaller than the shank because the thread cuts into it. The tool derives it from the ISO 898 formula rather than reading a table, so it always agrees with the thread reference.
Worked, for an M10×1.5 property-class 8.8 bolt at the default 75% of proof:
- Stress area: 57.99 mm²
- Preload: 0.75 × 580 MPa × 57.99 = 25,225 N (25.2 kN)
- Torque at K = 0.2 (dry): 50.45 N·m (37.21 lb·ft)
The nut factor is where the uncertainty lives
K bundles all the friction — under the head, in the threads, and any variation in surface finish — into one dimensionless number. It is the least certain term in the equation and it dominates the result.
Take that same M10 and change only K:
| Condition | K | Torque for the same 25.2 kN preload |
|---|---|---|
| Dry | 0.20 | 50.45 N·m |
| Zinc plated | 0.22 | 55.50 N·m |
| Lightly oiled | 0.18 | 45.41 N·m |
| Lubricated | 0.15 | 37.84 N·m |
| Waxed or moly | 0.12 | 30.27 N·m |
The same bolt, the same clamp force, and the torque figure ranges from 33.4 to 61.2 N·m depending on nothing but friction.
Which is why lubricating a bolt is dangerous
Read that table the other way. If a specification says 50.45 N·m for a dry bolt and you apply it to a waxed one, K has dropped from 0.20 to 0.12 — so the preload you actually generate is two-thirds higher than intended, taking a bolt targeted at 75% of proof well past it.
That is a real failure mode and it is the single most common way a correctly-followed torque figure yields a wrong joint. A torque spec is only meaningful alongside the condition it assumes. If the specification does not say, that is a question rather than a licence to guess.
Where this estimate should not be used
Being blunt, because this is the tool on this site with the most direct physical consequences.
Anything structural or safety-critical — wheels, brakes, steering, suspension, lifting equipment, pressure containment, anything holding a person up — has a specification from the manufacturer or the designing engineer, and that specification wins over any general model. Those figures account for the joint's materials, the gasket, the tightening sequence, the reuse history and the failure mode, none of which appear in T = K · F · d.
The calculator is for understanding the relationship and for sanity-checking a figure you already have. It is not a source of torque specs for a joint that matters, and a bolt whose spec you cannot find is a bolt to look up rather than to estimate.
What else the model ignores
- Relaxation. Joints lose preload after tightening as surfaces bed in — which is why some specifications call for a re-check.
- Sequence. On a multi-bolt joint, tightening order and staged passes change the final preload distribution substantially. The model handles one bolt in isolation.
- Reuse. A bolt taken past yield once is not the same bolt. Some are single-use by specification.
- Joint stiffness. Gaskets, soft materials and long grip lengths all change how preload behaves in service.
None of that makes the arithmetic useless — it makes it a way of understanding why torque figures are the shape they are, and why two apparently similar bolts carry different numbers.
Property class, and where the proof strength comes from
The proof strength in the calculation is set by the bolt's property class, marked on the head. The tool carries the standard set: 4.6 at 225 MPa, 4.8 at 310, 5.8 at 380, 8.8 at 580, 10.9 at 830 and 12.9 at 970.
The class markings encode the strength directly — the first number is a tenth of the tensile strength in MPa, the second the ratio of yield to tensile — which is why an 8.8 and a 10.9 of identical size carry very different torque figures. Substituting a lower class into a joint specified for a higher one is not a small change.
Unmarked fasteners are the practical problem. A bolt from a mixed tin has an unknown class, and there is no way to guess it from appearance. On anything that matters, that is a reason to use a known fastener rather than to assume a low class and hope.
The 75% figure is a convention
The default targets 75% of proof load, which is a common general figure and not a universal one. Some specifications work at 90% for a well-controlled joint; some work far lower where the joint has to survive fatigue, or where the clamped material cannot take the load.
Changing that fraction moves the torque proportionally, so it is worth knowing what your specification assumes rather than accepting the default silently.