Bolted joints

Bolt Torque Chart, and What It Cannot Tell You

Proof load is standardised by ISO 898-1. The torque column is not: change the nut factor from 0.20 to 0.14 and the same row moves about 30%.

Topic
Bolted joints
Reading time
6 minutes
Published
Revised
Editorial author
WindInspectTech · Source policy
Technical reviewer
Not attributed — no named technical reviewer is claimed
Source basis
ISO 898-1:2013 (Table 7 proof loads and the proof stresses behind them), ISO 68-1:1998 for thread geometry, ISO 16047:2005 for torque/clamp-force testing, VDI 2230 Part 1 and the NASA Fastener Design Manual.

The table everyone publishes

Proof load is standardised. The torque column is an example.

Two different kinds of number sit in every bolt torque chart, and they are usually printed in the same typeface. The proof load Fp is standardised: ISO 898-1:2013 Table 7 tabulates it for each size and property class, and it does not depend on who is tightening the bolt. The torque is a calculation — T = K · F · d — and it depends entirely on two values the chart chose for you.

The columns below declare both: the target is F = 0.7 · Fp and the nut factor is K = 0.20. They are shown so the order of magnitude is visible and so anyone can recalculate with their own figures. They are not recommended values.

ISO 898-1:2013 Table 7 proof load, with an example torque at F = 0.7·Fp and K = 0.20
SizeStress area AsFp class 8.8Example torque 8.8Fp class 10.9Example torque 10.9
M1284.3 mm²48 900 N82 N·m70 000 N118 N·m
M16157 mm²91 000 N204 N·m130 000 N291 N·m
M20245 mm²147 000 N412 N·m203 000 N568 N·m
M24353 mm²212 000 N712 N·m293 000 N984 N·m
M30561 mm²337 000 N1 415 N·m466 000 N1 957 N·m
M36817 mm²490 000 N2 470 N·m678 000 N3 417 N·m

Check the first row by hand: 0.20 × (0.7 × 48 900 N) × 0.012 m = 82 N·m. Every other cell in the torque columns is the same arithmetic. That is the point of showing the assumptions rather than only the result.

The argument, made operable

Change the nut factor and watch the answer move.

Set the size, the property class, the fraction of proof load you are targeting and the nut factor you believe applies to your assembly. The preload and the torque update together, and the comparison row shows what the same target costs under a drier or better-lubricated condition.

Proof load Fp
—
Target preload F
—
Torque T = K · F · d
—

—

Proof stress is taken as 580 MPa for class 8.8 up to and including M16, 600 MPa above it, 830 MPa for 10.9 and 970 MPa for 12.9, which is what reproduces the tabulated values above. The result is a calculation from stated inputs, not a specification, and it says nothing about the force in a bolt that is already installed.

The number doing the hiding

Everything the chart does not know is inside K.

In T = K · F · d, the diameter is geometry and the preload is a decision. K is where the rest of physics goes: friction at the thread flanks, friction under whichever face turns, and the helix angle, collapsed into a single dimensionless number. The NASA Fastener Design Manual names coating, lubricant, surface finish, repeated use and tightening speed among the things that move it.

Most of the applied torque never becomes clamping force. That is not a defect of the method; it is what the method is. It does mean that two bolts in the same flange, tightened to the same torque on the same day, can end up at materially different tensions — and a chart cannot tell you which is which. ISO 16047:2005 exists precisely because the torque/clamp-force relationship is meant to be established by test for the actual assembly rather than read from a table.

A hot-dip galvanised fastener is the case where this is least forgiving, which is why it turns up so often on wind-turbine tower flanges. The five ways to verify bolt tension sets out what each one evidences, and the comparison of pretensioning methods sets out what each approach controls and what it leaves inferred.

Showing the working

Where the calculator and the standard disagree, and by how much.

ISO 898-1 Table 7 lists proof loads as rounded values. The calculator does not have that table — it multiplies the stress area by a proof stress. The two therefore differ slightly, and the honest thing to do with that difference is show it rather than hide it behind a decimal place.

Calculated As × Sp against the tabulated proof load, class 10.9
SizeTabulated FpCalculated As × 830 MPaDifference
M1270 000 N69 969 N-0.04%
M16130 000 N130 310 N+0.24%
M20203 000 N203 350 N+0.17%
M24293 000 N292 990 N-0.00%
M30466 000 N465 630 N-0.08%
M36678 000 N678 110 N+0.02%

The largest disagreement is under half a percent, and it is a rounding convention rather than a physical claim. It is also several times smaller than the spread the nut factor introduces, which is the reason this page spends its space on K rather than on decimal places.

The limit

A chart answers tightening. It does not answer inspection.

There are two different questions that both sound like "what is the torque". The first is how should this joint be tightened, which a design specification answers and which a chart can illustrate. The second is what force is in this bolt now, on a flange that was assembled some years ago by someone else — and no torque value answers that, because the relationship that produced it was never recorded for those fasteners.

Where the second question is the one that matters, the force has to be observed on the bolt itself. Ultrasonic bolt tension measurement sets out how that is done and what it needs; ultrasonic bolt load verification as a service sets out what an engagement produces and what the enquiry has to state. The WIT-BW10 is the documented system where the requirement is an instrument rather than a service.

Provenance

Sources and limitations

  • ISO 898-1:2013 — mechanical properties of fasteners of carbon steel and alloy steel. Table 7 is the source of the six tabulated proof loads; the standard itself is paywalled and is cited, not reproduced beyond those values.
  • ISO 16047:2005 — torque/clamp-force testing, which specifies establishing the relationship by test rather than assuming it.
  • VDI 2230 Part 1 — carries assembly preload through a tightening factor that budgets for the imprecision of the method rather than treating it as exact.
  • NASA Fastener Design Manual — an open engineering reference describing the preload and torque-control variables that the nut factor absorbs.
  • Limitation. Every torque figure on this page is an example calculation from declared inputs. None is a recommended tightening value, none is specific to any joint, and none describes the state of a fastener that is already installed. Sizes outside the six tabulated rows are computed rather than taken from the standard, with the size of that discrepancy shown above.

Decision support

Frequently asked questions

What torque should I use for an M36 class 10.9 bolt?

This page will not give you one, and neither should a chart. The example column shows 3,417 N·m for M36 at 10.9, but that number exists only because two assumptions were declared: a target of 70% of proof load, and a nut factor of 0.20. Change the nut factor to 0.14 — plausible with a good lubricant — and the same preload needs about 2,390 N·m. The torque a joint needs comes from its own design specification, not from a chart.

What is the nut factor, and why does every chart use 0.20?

The nut factor K is the whole of T = K·F·d that is not geometry: thread friction, friction under the turned face, and the thread helix, collapsed into one dimensionless number. 0.20 is a convention for plain steel, not a measurement of your fastener. ISO 16047:2005 exists because the torque/clamp-force relationship is supposed to be established by test for the actual assembly rather than assumed from a table.

Does a torque chart tell me the tension in an installed bolt?

No. It tells you what torque a calculation says would produce a chosen preload under an assumed friction condition. Whether that preload is present in a bolt already in service is a separate measurement question, and the two are routinely confused. Ultrasonic measurement answers it for an identified fastener; a torque value does not.

Why does the chart stop at M36?

ISO 898-1 tabulates to M39, and the six rows here are the sizes already cell-checked on this site. Beyond them the calculator computes from the stress area and a stated proof stress, which lands within about 0.4% of the tabulated values — shown on this page rather than claimed. For large tower-flange and mill-liner fasteners above M39, the proof load comes from the bolt specification, not from ISO 898-1.

Engineering review

If the question is what force is in the bolt now, a chart cannot answer it.

Tell us the fastener, the joint, the target preload and its range, whether a baseline exists and the procedure that governs acceptance. We will tell you what can be established before anyone quotes a day rate.

Technical enquiry

sales@WindInspectTech.comWhatsApp sales
At least 20 characters. Do not include passwords, credentials, export-controlled files or confidential design data.
Company and destination (optional)
Useful technical inputs (optional)

Component, operating condition, required output and preferred data interface.

Inquiry route
Use email insteadBy submitting, you confirm that you have read the Privacy Notice. Product values remain configuration-specific and require application review.