When you tighten a bolt, what actually holds the joint together is tension (clamp force) stretching the bolt — but a wrench measures torque, the twisting effort. The short-form relationship T = K · D · F connects them through a friction term called the nut factor (K). The catch: K varies with lubrication, plating, and surface condition, so the same torque can produce noticeably different tension. That's why the dial above shows a ±25% band, not a single exact value.
An M10 class 8.8 bolt, dry (K ≈ 0.20), targeting 75% of proof load: the tensile stress area is about 58 mm², and class 8.8 at that diameter is a 580 N/mm² proof stress, giving a proof load near 33,600 N.
Target preload is 75% of that, and T = K · D · F works out to roughly 50 N·m. Lubricate the same bolt (K ≈ 0.13) and the same clamp force needs only about 34 N·m — which is exactly why applying a dry torque spec to a lubricated bolt can over-tension it.
Not for two of them. ISO 898-1 gives class 8.8 two proof stresses — 580 N/mm² up to and including M16, 600 above it — and SAE J429 gives grade 2 two, 55,000 psi to ¾" and 33,000 above. Torque is directly proportional to that figure, so a single value would mis-specify most of the range: using 600 everywhere over-tightens every size up to M16 by 3.4%. This page resolves the figure for the size you picked, and the grade selector shows both where a grade steps.
It leaves margin. Proof load is roughly the bolt's elastic limit; staying below it means the bolt holds tension without permanently stretching, with headroom for the service load on top.
Lubrication lowers friction (a lower K), so more of your twisting effort becomes useful tension instead of being lost to friction. Less torque is needed for the same clamp force.
Modest. Torque-to-tension scatter of ±25–30% is normal because K is hard to pin down. For critical joints, angle-control or direct tension measurement is far more precise.
No — use the fastener or equipment manufacturer's stamped spec. This tool is for estimation and understanding, and its results should be checked against an authoritative source.
Releases in which this page changed, newest last. Derived from the archived copy of every release, not from notes written afterwards — so it reflects what actually shipped. Site-wide passes are left out; they are in the full changelog.