EN 13155 and ASME BTH-1: two philosophies for the same beam

The American argument is about material. Take the steel's strength, divide by a design factor that follows from a design category, and check every member and connection against the result. The conservatism lives in one number, that number is traceable to a duty decision about how well the loads are known, and everything downstream is an ordinary structural check.

The European argument is about the device. Require the whole assembly to satisfy stated conditions under multiples of its rated load: elastic behaviour at one multiple, no collapse at a higher one. Then, separately, check the members against a structural code with its own partial factors. The conservatism lives in the multiples, and the multiples are about the device as an object rather than about any member in it.

Three practical consequences follow.

They can disagree about which check governs. A member check and a whole-device condition are different questions, and on the worked beam below the European route's governing check is not a member at all.

They optimise differently. Under the American scheme you improve a device by improving its worst member. Under the European one you may have to improve the device's overall behaviour under overload, which is not always the same intervention.

Neither is generally more conservative. Anybody who tells you one is has not run enough devices through both.

1. 01What question is the scheme asking?

   American: can each member carry the design load?

   Allowable stress from a design factor, tied to a design category. Plus a service class governing the fatigue side. Both are declarations somebody makes and owns.

   European: does the whole device survive multiples of its rated load?

   Proof conditions on the assembly, plus member resistances from the structural code. Two documents doing two jobs. Neither is complete alone.

2. 02What decides the margin?

   American

   The design category, from how well the loads are known. One number, traceable, and it moves every check by the same ratio.

   European

   The load multiples in the proof conditions, plus the national partial factors. The multiples are fixed by the standard; the partial factors are a national decision.

3. 03What happens with a lot of load cycles?

   American

   A service class, and a fatigue check inside the same standard.

   European

   Above a stated cycle threshold, the proof moves to a different standard family entirely. Not a harder version of the same calculation - a different set of documents, and the point at which a tool should stop rather than guess.

It is tempting to read "the device shall behave elastically at twice its rated load" as "apply a factor of two to the load and check the members". It is not the same thing, for three reasons.

It is about the device, not a member. The condition is on the assembly's behaviour. A device that yields locally somewhere harmless at twice load may still fail the condition if that yielding matters to the whole.

There is more than one multiple. Elastic behaviour at one, no collapse at a higher one. Those two conditions can be governed by different mechanisms in the same device, which a single load factor cannot express.

It can be satisfied by test. Where the standard allows verification by static test rather than by calculation, the device is proved rather than argued for. That is a different kind of evidence and it is available where a calculation is awkward.

The practical consequence is about how you fix a failure. A member check that fails is fixed by making that member stronger. A proof condition that fails may be fixed that way, and it may instead need a change to how the device redistributes load at overload, or to a detail whose behaviour under yield is the problem.

Two things to take from that.

The threshold is a design input, not a formality. Declaring the cycle count honestly can change which standard family your device is proved under, and that is a schedule and scope question rather than a calculation one.

A tool that stops is telling you something. An engine that produced a number here would be substituting a calculation it can perform for the one the standard requires. The refusal is the correct output, and it is the same behaviour as declining to invent a load factor for a jurisdiction that has none.

If you are on the American route. Get the design category and service class right first, because they move everything by a constant ratio and the whole margin traces to them. Then improve the worst member, and expect the worst member to be the answer.

If you are on the European route. Expect two families of check and expect the proof conditions to be competitive with the member checks. Decide early whether you are verifying by calculation or by test, because that decision changes what evidence has to exist and when.

If you are designing for both markets. Do not average them. Run both, take the worse of each check, and state that the device satisfies both schemes with the governing check named under each. A device sized to the maximum of two schemes is heavier than either requires, and that is an honest cost of selling into two markets.

If somebody asks which is more conservative. The correct answer is that it depends on the device, and the worked example is a fair illustration: the American route gave 64 percent and the European one 86 on the same beam, and swapping the section for a stockier one would move both, not necessarily in the same direction.

1. A proof condition read as a load factor. The condition is on the device's behaviour, and fixing a failure may not be a member change.

2. The harmonised standard cited alone. It carries no member resistances. A calculation citing it alone has checked nothing against a capacity.

3. A design factor carried across. It is tied to a design category defined in its own standard, and it means nothing outside it.

4. Partial factors assumed. They are a national decision, and the same beam can be accepted differently in two countries.

5. The cycle count declared casually. On the European route it can move the whole proof to a different standard family.

6. The two schemes averaged. A device for both markets satisfies both, with the governing check named under each. There is no midpoint that means anything.