Erection tolerances: why parts “within tolerance” still do not fit together
The hall is finished and the sectional door does not fit the opening. The steel contractor says everything is within tolerance. The building contractor says the foundation is as designed. The door manufacturer says the opening is out of true. And — worst of all — all three of them can be right. Deviations add up, and nobody contracts who is answerable for the sum. This is a guide through the only part of building a hall in which everybody keeps to the rules and the outcome is still no good.
Two kinds of tolerance that are constantly confused
EN 1090-2 separates tolerances into two groups, and that difference is not a formality:
Essential tolerances
Deviations that affect the resistance and stability of the structure — the curvature of a column, the eccentricity of a connection, the position of a support. The standard prescribes them unconditionally and they are not a matter for negotiation. Exceeding them means that the structure no longer corresponds to the calculation.
Functional tolerances
Deviations that affect assembly, appearance and the installation of the other components — the flatness of a wall, the position of an opening, the alignment of the purlins. They do not bring the hall down, but they decide whether the door, the panel and the flashing will seat.
Functional tolerances have two classes, and class 2 is the stricter. If the design does not say which of them is required, the more lenient class 1 applies. A contract that says only “to EN 1090-2” thereby chooses the more lenient class — entirely in accordance with the standard, and entirely unexpectedly for a client who thought he had tightened something up.
On top of that, the standard prescribes fabrication tolerances and erection tolerances separately. A component can be flawlessly fabricated and still end up out of position because the deviation happened during setting — and the other way round.
Deviations add up, and nobody contracts for it
The path from the foundation to the edge of the envelope has at least five or six links: the position of the foundation, the position of the anchor in it, the fabrication of the column base, the plumb of the column after grouting, the length of the girder and the position of the purlin. Every link has its own permitted error, and every one of them can be properly within it.
The problem is that those errors do not cancel out automatically. In the worst case they all go the same way and add up. In the more likely case they partly cancel out, so the actual deviation is calculated as the square root of the sum of the squares — which is considerably less than a plain sum, but still more than anyone expects.
How much deviation adds up by the time you reach the envelope
Worst case — everything the same way
±39 mm
a plain sum
The more likely outcome — statistical
±17.6 mm
the square root of the sum of the squares
Every link in the chain can be within its own tolerance and the sum still be 39mm. That is why “everything is as designed” and “the door does not fit” are not contradictory statements — both can be true at the same time.
* All the values are user input, not values from the standard. The actual tolerances are taken from the design and from the tables of EN 1090-2 for the agreed class, and for concrete work from EN 13670. The statistical sum applies when the deviations are mutually independent; when they are caused by the same systematic reason (a wrong reference point, for example), reality moves towards the worst case.
Where the chain really breaks: concrete and steel do not read the same standard
The foundations are executed by the building contractor and he works to EN 13670 and to the design of the concrete works. The steel structure is executed by another contractor, to EN 1090-2. Those are two standards with their own tables of deviations, written for two different trades — and nowhere is there a document which says that their sum has to satisfy anything.
The practical consequence: the foundation can be executed to its own standard, the anchors set to their own tolerance, the column erected to its own — and nothing fits. That is why anchors are set in sleeves or pockets whenever possible: their permitted deviation is then ±10 mm instead of ±3 mm for a rigidly cast-in anchor, so the difference can be adjusted before grouting instead of being cut with a flame afterwards.
Who measures what, and when
- Before the steel is delivered: a setting-out survey of the executed foundations and anchors. This is the only point at which a deviation can still be corrected cheaply.
- After the frames have been erected: a survey of the actual position of the columns and of the heights, before the made-to-measure envelope is ordered.
- Before the doors and windows are ordered: a measurement of the actual openings on the finished structure, not copying from the design.
- Before the flashings: a measurement of the actual edges. A flashing is folded to measure and it is what conceals the deviation — but only if it has been measured and not calculated.
That last item is the reason why flashings are made last and to a measurement taken on the building. The sheet-metal work is the only layer that adapts to the actual condition instead of to the drawing — and that is why in practice it swallows the sum of all the deviations that have accumulated beneath it. There is more on how a flashing is measured and calculated in the guide to flashings.
What to agree before anything reaches the site
- Which class of functional tolerances is required — 1 or 2? If it does not say, you get the more lenient one.
- Who pays for the setting-out survey of the foundations, and does it have to arrive before the steel goes into production?
- Do the anchors go into sleeves or are they rigidly cast in — and who bears the cost if they miss?
- Are the doors and the envelope ordered from the design or from a measurement of the finished structure?
- Who is answerable for the sum of the deviations of two contractors — because under the standards nobody is?
- How is a deviation dealt with when it appears: by rework, by grouting or by a change — and with whose consent?
Tolerances are the only subject in building a hall where nobody makes a mistake and the result is still no good. The solution is not a stricter standard but one sentence in the contract: who measures, when the measurement is made and who pays the difference. Everything else is an argument on site in which each side has a piece of paper that proves it right. How the anchors and the base plate are executed is in the article on the hall column, and what the execution class actually means in the article on EN 1090-2.
We make steel structures, prefabricated halls and made-to-measure flashings in our own production in Đakovo — and we fold the flashings to a measurement taken on the building, not to the drawing. If you have the dimensions and the use: about steel structures and an enquiry with dimensions.
Frequently asked questions
Because deviations add up. The path from the foundation to the edge of the opening has five or six links — the position of the foundation, the position of the anchor, the fabrication of the column base, the verticality of the column, the length of the girder and the position of the purlin. Each of them can be properly within its own permitted tolerance, and the sum still comes out in double figures in millimetres. That is why “everything is as designed” and “the door does not fit” can be true statements at the same time.
English version of the Croatian original: Montažne tolerancije: zašto se elementi „unutar tolerancije“ ne poklope.