Tolerances and limit deviations: why they exist and how to read them

Why no dimension works without a range

An elbow with an outside diameter of 114.3 mm never measures exactly 114.3 mm. No manufacturing process hits a number; tools wear, materials spring back, heat expands. That is not a defect but the normal case – and precisely why a standard must say how far reality may deviate from the number.

A dimension without a limit would be doubly useless: not orderable, because the manufacturer does not know what to hit, and not inspectable, because the inspector does not know what to let through. The limit is therefore not a restriction on a promise; it is the promise.

The second reason is interchangeability. When a fitting is welded in, it meets a pipe made to a different standard. For the two to fit together without anyone matching them beforehand, both standards have to apply the same yardstick at the joint – and that is exactly what they do.

Four words that are constantly mixed up

On the shop floor everything is a “tolerance”. In technical language these are four different things, and the difference matters as soon as someone starts calculating:

Why we show limit sizes and not limit deviations: because nobody can read a deviation off a calliper. Whoever measures has a number and wants to know whether it lies between two others – not whether a difference they first have to form lies in an interval they also first have to form. In the drawing the short word “tolerance” is used all the same: it is the word everyone uses, and space on a sheet is precious.

How a number gets into the standard

A limit is always a trade-off between two costs: a tight tolerance makes manufacture expensive, a wide one makes installation expensive. The standards committee places it where usual manufacture holds it without special effort and usual installation tolerates it without rework.

Why large parts get a percentage and small parts millimetres. A fixed millimetre limit would be generous at ⌀ 21.3 and unattainable at ⌀ 1219 – the manufacturing deviation grows with the part. That is why EN 10253 states a three-part rule for the outside diameter:

• ±1 % of the nominal size – the actual statement,
• at least ±0.5 mm – otherwise the percentage rule would fall below what is measurable and manufacturable at small diameters,
• at most ±5 mm – otherwise it would allow gaps at large diameters that no weld can bridge any more.

These three lines are not formula clutter but three different worries, dealt with one after the other. Whoever has seen this once reads every similar rule faster.

The ways of setting a limit

A tolerance is not always a number. We went through the limit deviations of our components, and several different ways came together in which a standard sets a limit:

Counted on the whole rule base for butt-welding fittings and flanges in our data layer; the German page shows the number of rule lines per kind, rebuilt from the data with every change.

The three highlighted ones are the reason for this page. “No limit value” means three different things in a standard, and these three cases are confused every day:

• Contained means: the limit is somewhere else. Below ⌀ 273, EN 10253 names no value of its own for the roundness – but the largest and the smallest diameter must each lie within the diameter limit deviation, and that limits their difference by itself. Whoever reads “no requirement” here reads wrongly.
• By agreement means: the standard considers the case too varied to regulate and expressly pushes it into the order. Whoever agrees nothing here gets what the manufacturer considers right – in accordance with the standard.
• None means: the limit does not exist, and that is a decision. On the body of a fitting the wall thickness is not limited upwards, because a thicker wall makes nothing worse there – it carries more. Downwards it certainly is limited.

Why measurement is stricter at the welding end than on the body

Almost all fitting standards separate two places: the welding end and the body. The tight limits apply at the end, the wide ones or none at all on the body. The reason is simple: the end is where the welding happens. There the fitting meets a pipe, and there it is decided whether the joint fits. On the body the part touches nobody.

That is why the pipe standards of the EN 10216 series say the same sentence for the outside diameter as the fitting standard, word for word. That is no coincidence but intent – what comes together at the joint has to fit together.

What this means at goods inward

• Measure where the standard measures. A diameter measured on the body and held against a limit deviation for the end yields a complaint that is none.
• Roundness needs two measurements in the same cross-section – the largest and the smallest outside diameter. A single value says nothing about it.
• For the wall thickness the lower limit deviation is the one that carries. It decides the permissible internal pressure; upwards, more wall is no defect.
• A non-binding value in a standard is not a promise. It is one solution, not the solution – what governs is the verification the standard provides for it.

Where the limit deviations of our parts are

For every size, with a dimensioned drawing and the permissible range for every dimension: our dimension finders for elbows, tees, reducers and caps to EN 10253-2, each with its sister page for stainless steel to EN 10253-4 (elbows, tees, reducers, caps and dished ends). Further tables in English: