Torque is not preload
A torque wrench measures how hard you are twisting. It does not measure what the joint actually cares about, which is the clamp load in the bolt. Almost all of the torque is spent overcoming friction — and friction changes with the finish, the lubricant and the state of the threads. Press play to see where the torque goes, then move the friction slider and watch the same torque give a very different clamp load.
What to specify instead
- The tightening method — turn-of-nut, DTI washers, or direct tensioning — not a torque figure alone.
- If torque is used, the lubrication condition it assumes, and that the same condition is reproduced on site.
- The target preload in kN and the basis for it, so the figure can be checked.
- Who verifies, and how many of the joint's bolts get checked.
Torque from T = K × F × d, with F the target preload, d the nominal diameter and K the nut factor for the finish shown. Preload basis 70% of proof stress. Expect roughly ±25% scatter around any torque-derived preload even with the friction condition controlled.
Where these figures come from. Proof stresses and stress areas are the ISO 898-1 values, and the stress area is calculated by the ISO 898-1 formula. K is not from a standard — no standard publishes one, because it is not a material property. The values here are the commonly used industry figures for each finish, the same set used on our torque chart, and they are approximations by nature. If preload matters, measure it rather than assume a K.
Where the torque actually goes
The split drawn in step 3 is the approximation used throughout the fastener industry: roughly 50% of the applied torque is consumed by friction under the turning face, roughly 40% by friction in the threads, and roughly 10% does the work of stretching the bolt. The exact split depends on the geometry and on both friction coefficients, but the message does not change with the detail: the number on your wrench is mostly a friction measurement.
That is also why the nut factor K is not a physical constant. It bundles thread friction, face friction and the thread geometry into one number so that T = K × F × d can be written down. Change the finish, the lubricant, the surface condition, whether the bolt has been tightened before, even how fast you pull — and K moves.
The same torque, four finishes
Pick a size and grade above; this compares the clamp load that one torque figure produces across the finishes, holding everything else equal.
Torque held at the value for the finish selected above. Clamp load shown as a percentage of the bolt's proof load, so you can see where a finish change pushes you.
Questions we are asked
Why does the same bolt need a different torque when it is galvanized?
Because the friction is different. A dry hot-dip galvanized thread has a higher nut factor than a plain or oiled one, so more of the torque is eaten by friction and less reaches the bolt as clamp load. To land on the same clamp load you need more torque. Use the figure for the finish and lubrication you actually have in your hand.
Can I just lubricate everything to make it consistent?
Lubrication does reduce the scatter, which is why it is common on flanges and studs. But it also lowers the nut factor sharply, so every torque figure has to be recalculated for the lubricated condition — applying a dry figure to a lubricated bolt is how bolts get taken past yield. Lubricate by decision and state it in the procedure, never as an afterthought on site.
What is turn-of-nut and why is it better?
Once the joint is snug, the bolt behaves like a stiff spring: a known rotation of the nut produces a known stretch, and stretch is clamp load. Turning the nut a specified fraction of a turn past snug therefore controls the preload directly, without depending on friction at all. It is the reason structural codes lean on it rather than on torque.
How accurate is a torque wrench in practice?
The wrench itself can be good; the joint is what introduces the error. Even with the friction condition controlled, preload from a torque-controlled tightening is commonly taken as roughly ±25%. With the friction condition uncontrolled — mixed finishes, site dirt, reused fasteners — the spread is considerably wider. That is the whole argument for specifying a method rather than a number.
Related
Torque & preload chart — every size, printable · Installation methods · Galvanizing & thread fit · Back to the Bolt Lab