Bolt tightening torque and preload: methods, k-class, inspection

Preload Fp,C: what it is and how it is determined

A bolt with controlled tension works like a spring pressing the stack of plates together. The force with which the shank is stretched after tightening is called the preload. It has to come out as the project states: the friction between plates in a slip-resistant connection, the behaviour of a flange under variable load and whether the joint opens all depend on it.

DSTU EN 1090-2:2019 takes the nominal minimum preload as Fp,C = 0.7 · fub · As. Here fub is the nominal tensile strength of the bolt material and As is the cross-section area of the bolt at the thread. This preload is used for all slip-resistant and other preloaded connections unless the project gives a lower level. If it does, the project must name the assemblies, the tightening method, the preload parameters and the inspection requirements.

For M20 class 10.9 we get 0.7 × 1000 MPa × 245 mm² = 171.5 kN, and Table 18 of the standard says 172 kN. For class 8.8 it is 0.7 × 800 × 245 = 137.2 kN, and the table gives 137 kN. The full Fp,C table for M12–M36 is in the article on bolted connections of steel structures and is not repeated here.

The standard also names the reasons for specifying preload at all: resistance to slip, seismic connections, resistance to fatigue and execution quality, durability among them.

Why preload is specified as force, not as torque

Torque gives the force only indirectly. Part of it goes into friction in the thread and under the nut (or the head), and only the rest stretches the bolt. How much goes into friction depends on the lubricant, the coating, the state of the thread and the washers. So the force Fp,C is fixed first, and the torque is chosen for it for each combination of bolt and nut.

Torque Mr and the factor k

DSTU EN 1090-2:2019 defines the reference torque Mr that produces the force Fp,C for each type of bolt and nut combination, and there are two ways to find it. The first is to take the value declared by the assembly manufacturer for its K-class. The second is to obtain it by calibration under site conditions according to Annex H.

For K2 assemblies the torque is Mr,2 = km · d · Fp,C. Here km is the mean value of the factor k that the manufacturer declares for the lot and d is the nominal bolt diameter. For class K1 the standard gives a ready factor: Mr,1 = 0.125 · d · Fp,C.

The factor k itself is a dimensionless ratio of the torque to the product of the diameter and the force: k = Mr / (d · Fp,C). It collects all the friction losses. The slicker the lubricant and the smoother the thread, the smaller k and the smaller the torque needed for the same preload.

If the diameter is put in mm and the force in kN, the torque comes out directly in N·m: 1 kN × 1 mm = 1 N·m.

After cl. 8.5.2 and Table 19 of DSTU EN 1090-2:2019. The K-class is the calibration state of the assembly in the delivery condition (EN 14399). For the torque method and for HRC the coefficient of variation Vk of the assembly must not exceed 0.06.

Why k cannot be taken from a table found on the internet

The value km is the result of tests on a particular lot of assemblies, that is bolts, nuts and washers in the lubricant and coating they left the factory with. The standard directly requires such assemblies to be used without changing the lubricant received with the delivery. If the bolt is lubricated or the nut is taken from another lot, k is already different, and so is the preload at the same torque.

The calibration state of the assembly applies if it is installed by turning the nut. When the bolt head is turned, the calibration is carried out according to Annex H or from the results of an additional test.

Example: torque for M20 class 10.9

Take an M20 class 10.9 assembly of the HV system, K-class K2, with a declared km = 0.13. This number is conditional, for the example: in a project the value is taken from the assembly manufacturer’s documents. The tightening method is by torque.

Arithmetic: 0.13 × 20 × 172 = 447.2 N·m; 0.75 × 447.2 = 335.4; 1.10 × 447.2 = 491.9; 1.05 × 447.2 = 469.6 (N·m). For another diameter, class or k the calculation is the same, and only Fp,C, d and k change.

What happens if the real k is different

Here is why k matters. The same torque of 447 N·m applied to an assembly with another factor gives another force: F = Mr / (k · d). The values of k in the table are conditional, taken for illustration.

At k = 0.10 the force of 224 kN exceeds 220 kN, which is the proof load of an M20 class 10.9 shank (900 MPa × 245 mm² = 220.5 kN). The estimate is approximate: twisting of the shank is not taken into account. At a larger k the bolt stays under-tightened and a slip-resistant connection does not have the design friction.

The same bolt in the combined method

If the assembly is class K1, the first-step torque is taken from Mr,1 = 0.125 × 20 mm × 172 kN = 430 N·m, so 0.75 · Mr,1 = 322.5 N·m. Table 20 of the standard gives 322 N·m for M20 class 10.9.

The second step is a further turn of the nut by an angle that depends on the thickness t of the stack, which includes packing and washers. Let t = 60 mm. For M20, 2d = 40 mm and 6d = 120 mm, so 2d ≤ t < 6d, and the nut is turned a further 90°, a quarter turn. The position of the nut on the thread after the first step is marked with a marker or paint.

Tightening methods: what sets the preload in each

DSTU EN 1090-2:2019 permits four methods, and the project may limit the choice. A summary table with K-classes, steps and checks is given in the article on bolted connections. Here we cover what happens in each method and why their requirements for the assembly differ.

• torque method: the wrench sets the preload. The first step is about 0.75 Mr on all bolts of the connection, the second 1.10 Mr. The factor 1.10 corresponds to a scatter of k with a coefficient of variation of 0.06 (1 / (1 − 1.65 × 0.06) ≈ 1.11). The torque is applied smoothly and held at a constant level. That is why the method needs K2 assemblies;
• combined: the wrench brings the bolts to about 0.75 Mr, and after that the preload is set by the angle of the nut turn, not by torque. For that reason the scatter of k matters less, and K1 assemblies will do. If the bearing surface under the head or the nut is not perpendicular to the bolt axis, the angle is determined by a test rather than from the table;
• HRC: the bolt limits the torque itself. A two-socket non-impact electric wrench turns the nut while holding the splined end of the bolt until the neck shears. The preload level is set by the bolt geometry and the lubricant, and the wrench does not need calibration. The first step is completed no later than the moment the outer socket stops turning. When there is no room for the wrench, they switch to the torque method with K2 or to DTI;
• DTI: indicator washers according to EN 14399-9. The first step gives a uniform tight clamp (the first deformation of the protrusions appears on the indicator), the second compresses the protrusions to the value set by that standard. Gap readings may be averaged. Indicators are not used with weathering and stainless steels.

The order is the same for all methods. Tightening is done gradually, from the part of the connection with the greatest stiffness to the least stiff, and to even out the preload more than one cycle may be needed. Before preloading, the remaining gap between the steel parts must not exceed 2 mm. Tightening is done by turning the nut, except where there is no access from the nut side.

First-step torque of the combined method for K1 assemblies, 0.75 · Mr,1 ≈ 0.094 · d · Fp,C (Table 20 of DSTU EN 1090-2:2019). For K2 assemblies the first step is 0.75 · km · d · Fp,C for the declared km.

Wrenches and tools

All steps of the torque method need a torque wrench with an accuracy of ±4 % according to EN ISO 6789 (all parts), maintained under that standard. For pneumatic wrenches the check is repeated every time the hose length changes. A wrench must also be checked after any unforeseen event: a heavy blow, a drop, an overload.

At the 447 N·m of the example, ±4 % is ±18 N·m, that is a range from 429 to 465 N·m. This scatter of the wrench adds to the scatter of k, so a wrench without a calibration certificate is not suitable for a preloaded connection.

For the first step of the combined method the requirements are softer: accuracy up to ±10 % and a check once a year. An impact wrench is acceptable only for the first step of the torque method. Torque is applied smoothly. For HRC no calibration of the wrench is needed.

Calibration of other ways of installation, for example axial tension with hydraulic devices or applying tension with ultrasonic monitoring of the force, is carried out according to the recommendations of the equipment manufacturer.

Lubricant, coatings and galvanizing of the assembly

The lubricant applied to the assembly at manufacture is part of its calibration. High-strength assemblies for preloading are used without changing the lubricant, unless the DTI method is chosen or the procedure of Annex H is carried out. Annex H reminds that the state of the lubricant may change when assemblies are exposed to the environment or stored for a long time.

There is one exception: if because of unforeseen climatic effects preloading is carried out with delays, the lubricant may be replaced with another one after its suitability has been checked.

The standard describes coatings on fasteners as follows: hot-dip galvanized ones must comply with EN ISO 10684 and electroplated ones with EN ISO 4042. For class 10.9 assemblies the note to cl. 5.6.1 separately mentions the risk of hydrogen embrittlement in electroplating and hot-dip galvanizing. What this means for threads and strength classes is covered in the article on fastener strength classes. For the joint something else matters: the calibration of k applies to the assembly in the delivery condition, so it is not taken apart and the lubricant is not changed.

Annex I of DSTU EN 1090-2:2019, Table I.1 (informative). The loss of preload must be checked if the coating on the contact surfaces is thicker than 100 µm per surface or the material is permeable. In connections of categories B, C and E under EN 1993-1-8 it may be necessary to put 0.9 Fp,C into the design or to set the preload and tighten the bolts again after a few days.

Coating between the washer and the stack

There must be no thick-film coatings between the washers and the contact surfaces. If they cannot be avoided, the project has to state measures that compensate for the possible loss of preload over time.

Reusing bolts and assemblies

The rule in the standard is short. An assembly that has been tightened to the minimum preload and then loosened is dismantled, and the whole assembly is rejected: the bolt, the nut and the washers. The reason is clear: the calibration of k refers to the delivery state, and after tightening the lubricant and the thread are no longer the same.

Bolts that were only used to fit the connection up before final assembly were not brought to the minimum preload and are suitable for final tightening to the specified force. Assemblies used in calibration tests under Annex H are not put into the structure and are not reused in further tests.

Inspection has its own rules as well. An over-tightened assembly is removed and rejected, and in the combined method an assembly where the angle is exceeded by more than 30° or the bolt or nut is damaged is replaced with a new one.

Reference torque and checking the tightening

The torque check shows whether a bolt has been left under-tightened. A wrench set to the check torque is applied to the nut, and the value needed to make the nut start turning must be at least 1.05 of Mr,i, where Mr,i equals Mr,2 (from the declared km) or Mr,test (the torque found by calibrating the assembly on site under Annex H). For the M20 example that is 470 N·m.

The check is made between 12 and 72 hours after the bolts of the subgroup were tightened. The wrench must be calibrated and have an accuracy of ±4 %. The nut must not turn by more than 15°: otherwise the bolt is under-tightened and is tightened again to the full torque. If a bolt is rejected in the check, the accuracy of the wrench used to preload the bolts is checked as well.

For the combined method the first step is checked with the same torque that gave 75 % of the preload, and the second by the marks on the nuts: a shortfall of more than 15° is corrected by a further turn, an excess of more than 30° leads to replacement of the assembly. HRC is checked by inspecting 100 % of assemblies (the splined end must be sheared off), DTI by inspecting all indicators for full compression.

How many assemblies to inspect for the execution class EXC and how sequential sampling works is covered in the article on bolted connections. A check for over-tightening is made only where the project specifies it.

Common mistakes

• a torque «taken from a table»: someone else’s torque is taken for another assembly, lubricant or coating. It gives the wrong preload even on an identical bolt;
• lubricating «to make it run easier»: k falls, and at the same torque the bolt stretches more, up to exceeding the proof load (see the calculation above);
• the second step with an impact wrench: an impact tool is accepted only for the first step of the torque method;
• one step instead of two: the bolts are tightened one by one to full torque, and the first ones relax while the next are tightened. First all bolts are taken to 0.75 Mr, then all to 1.10 Mr;
• a wrench without a calibration certificate or after a drop, a 100 N·m wrench used for 450 N·m at the end of its scale;
• the nut not marked after the first step of the combined method: the angle of the further turn can then not be measured;
• a gap in the stack of more than 2 mm before preloading: bolts cannot close it, and the connection is corrected first;
• the minimum torque for erection bolts from installation guidance taken as the preload torque: it is a lower limit for ordinary bolts, while for a preloaded assembly the torque is calculated from k. For M20 class 10.9 in the example above it is about 450 N·m, and 15 kgf·m is only about 147 N·m;
• a K0 assembly tightened by the torque method: this method needs K2, and for K0 the factor k is not specified.

What to put in the KMD

Erectors do not pick the torque by feel: they read it from the drawing and the assembly documents. So the KMD and the fastener request should carry these items.

What of this lies with us

We fabricate steel structures (steel fabrication) and prepare KM and KMD drawings (KM and KMD design). We supply the fasteners for the connections together with the structures, so it is convenient to name the type of assembly, the K-class and the tightening method already in the request. For the rules of drawing up the fastener schedule see the article on KM and KMD drawings, and for a sample sheet see the KMD sheet and specification example.

Erection is done by our partner construction company, and tightening of the bolts and checking of the preload stay with it, under the work execution project. So the torque, the method and the K-class have to be in the project and not worked out on site. What to hand to the fabricator along with the drawings is covered in the article package for the fabricator.

Standards and validity

The execution of preloaded bolted connections, the preload Fp,C, reference torques, tightening methods, calibration and inspection are covered by DSTU EN 1090-2:2019 «Execution of steel structures and aluminium structures. Part 2. Technical requirements for steel structures» (EN 1090-2:2018, IDT). It has been in force since 01.07.2020 (order of SE UkrNDNC of 10.05.2019 No. 117). All clauses and tables in this article are given from that text: cl. 5.6, 8.2.4, 8.5, 12.5.2, Tables 18–21 and 25, Annexes H and I.

Assemblies for preloading are covered by the EN 14399 series, to which DSTU EN 1090-2:2019 refers. In Ukraine DSTU EN 14399-1:2019 (general requirements, including the k-class), DSTU EN 14399-3:2024 and DSTU EN 14399-4:2024 (systems HR and HV, in force from 01.07.2025) are current. The value of k and the coefficient of variation for a particular lot are set by the assembly manufacturer, and a project uses the documents for the lot.

The calculation of the connection itself, including the friction and safety factors, is made under DBN V.2.6-198:2014 with Amendment No. 1; this is covered in the article on bolted connections. The figures in this article are for orientation; a project uses the current text of the standards and the documents of the assembly manufacturer.