[brudgers]

While the statement seems technically correct, it smells a bit disingenuous to me.

Most sophisticated structures are engineered using load factors in combination rather than traditional strength design methods because it leads to more efficient use of materials and hence lower cost.

Yes, strength design might require a 4x safety factor. That’s why engineering practice employs statistical methods and combined loads.

Keep in mind that the safety factors are never supposed to be the difference between stability and failure. If the structure relies on safety factors there’s a problem.

[dylan604]

> leads to more efficient use of materials and hence lower cost.

I know that isn't always a bad thing, but this is the kind of sentence that always gives me pause. In the case were over-engineering something to 10x minimum tolerance can be re-configured to be only be 8x minimum, then sure, that sounds like there's some room to safely redesign for efficiency/cost sake.

My pessimism for the world today tends to make my first impression the "lower cost" phrase being used is always the "cheap bastards looking to cut corners at the expense of safety".

[brudgers]

Most choices are not cake or bread.

Most choices are bread or going hungry.

Or a new bridge instead of an old failing bridge or no bridge at all.

The fact that the bridge failed catastrophicly rather than ductally suggests the fundamental problem was systemic in the design and the mode of failure was from unanticipated forces.

What I mean that if the force was 8x the design load 2x vs 4x safety factors would be irrelevant.

[yunohn]

> If the structure relies on safety factors there’s a problem.

Maybe I’m misunderstanding your point, but I don’t think it’s unreasonable to have actual (large) limits on what a structure can handle? Otherwise, everything will cost X times more time/money to build.

[MarkMarine]

The idea is to work out the maximum design limit, then add a factor of 2x or 4x to that to allow for things you can’t predict or model well. If the design is relying on some part of the above of the 2x to work, it’s not really a safety factor.

I’m a mechanical engineer and figuring out the appropriate strength a part needs, then making it twice as strong, is a common tactic. You can’t account for material differences, little casting voids, or just your math being off a little by the approximate methods we use to calculate stresses. The more perfectly we can model and then repeatedly destructively test a part, the closer we can get to a safety factor of 1.

[yunohn]

Ah, gotcha. That makes sense!