Bolted connections in a steel hall: why the same M20 holds 27 or 69 kN

The bolt is the cheapest part of a hall and the only one that somebody tightens by hand on site. That is also why it is the place where most goes wrong — not in the calculation, but in what happens after it. The same M20 of grade 10.9 in the same connection transfers 27 kN or 69 kN, depending solely on how the faying surfaces of the plates have been prepared. The difference is not in the bolt, not in the steel and it costs next to nothing — and nobody mentions it. This is a guide through what a hall quotation usually states as a single word: “bolted”.

Five connection categories, two philosophies

EN 1993-1-8 divides bolted connections into five categories. Three are for shear — A, B and C — and two for tension, D and E. The difference between them is not in the number of bolts but in how the connection transfers the force at all.

The practical consequence: category A allows a movement equal to the hole clearance. For M20 the normal clearance is 2 mm, so a connection that transfers the force only after the plates have moved 2 mm behaves differently from a preloaded one. On a single connection that bothers nobody. On a frame with ten connections in a row the movements add up and the hall becomes softer than the designer had calculated.

Preloading: a bolt that does not transfer the force but clamps it

A preloaded bolt does not work like a pin in a hole. It clamps the plates together so hard that the force is transferred by friction between their faying surfaces. The bolt is only the clamp. The force with which it clamps is prescribed by the standard: Fp,C = 0.7 × fub × As, where fub is the tensile strength of the bolt and As the stress area. For an M20 of grade 10.9 that is 0.7 × 1,000 N/mm² × 245 mm² = 171.5 kN — about 17 tonnes per bolt.

From that follows what investors rarely hear: the resistance of such a connection depends on the roughness of the plates, not on the bolt. In Table 18, EN 1090-2 prescribes four friction classes, from A with a factor of µ = 0.5 (blast-cleaned, free of pitting) to D with µ = 0.2 (untreated as-rolled surface). The difference between the best and the worst class is a factor of 2.5 — on the same bolt, the same plate and the same price of steel.

The most expensive error: the painted faying surface

On site this happens in the most banal way: the structure arrives painted in the full system, including the surfaces that are going to overlap. Somebody judges that it is “better if everything is protected” — and thereby turns class A into something indeterminate, well below class D. The connection still looks the same. The difference shows only when the full load appears.

The solution is banal if it is agreed in advance: the faying surfaces are masked before painting and prepared to the friction class from the design, and after erection the edge of the connection is sealed and overpainted. That is an item that has to exist in the quotation — if it is not there, somebody will improvise it.

Why a tightening torque table from the internet does not apply

EN 1090-2 permits several procedures for achieving the preload — the torque method, the combined method, the method with HRC bolts and the method with direct tension indicators. All of them have a common condition: they use assemblies of fasteners of class K2 to EN 14399-1, for which the manufacturer declares the mean value of the tightening coefficient k_m and its scatter.

The torque is then calculated from that declared value, for that delivery. That is why a generic table of torques in newton metres is not applicable — the same size and the same grade of bolt from two deliveries may call for a different torque because their factory lubrication differs. For the same reason additional lubrication of the bolts on site voids the declared k_m and with it the whole tightening procedure.

HR and HV: two systems that look the same

Preloaded assemblies come in two systems. HR has a longer nut and is designed so that under overload the bolt itself fails — more slowly and with visible elongation. HV has a shorter nut and fails by thread stripping, which is more compact and cheaper, but gives less warning. Both are standardised and both are correct; they are not interchangeable piece by piece, because they differ in the geometry of the nut and in the behaviour at failure.

For the investor only one thing matters: the documentation must state which system, and the whole assembly from the same delivery has to arrive on site. A mixture of the bolt of one system and the nut of another looks tidy and is not tidy.

Holes, clearances and why connections do not line up

Normal holes have a clearance of 1 mm for M12 and M14, 2 mm for M16 to M24 and 3 mm from M27 upwards. That clearance exists so that the structure can be put together at all — without it nothing would line up, because fabrication itself has its own tolerances.

When a connection does not line up, there are two routes. The correct one is that the deviation is measured and, if it is within what was designed, the hole is reamed with the designer’s consent. The usual one is that somebody picks up a flame. A hole enlarged with a flame changes the properties of the steel around it, does not respect the geometry, and on a category C connection it practically invalidates the calculation — and after painting it can no longer be seen.

That is why the execution class and inspection are not bureaucracy. From EXC2 upwards, the contractor must have a procedure for deviations and a record of who executed the connection and how. More on what the execution class actually means is in the article on EN 1090-2.

The bolted connection is the only place on a hall where the whole calculation rests on what somebody will do in a single afternoon at a height of ten metres. Everything that is not agreed about it beforehand will be agreed then — and in favour of whoever is holding the spanner. How what gets scratched and welded at those connections is protected we have set out in the article on corrosion protection, and how what those connections hold is sized in the article on the hall column.

We make steel structures and prefabricated halls from our own production in Đakovo. If you have the dimensions and the use of the building, we can start from that: about steel structures and an enquiry with dimensions.