What really is the difference between allowable stress design and ultimate limit state design?

Question

I'm trying to understand this difference. From what I've read, ASD basically deals with making sure the stress in a structure does not exceed a certain limit, the yield stress of the material. It ensures that structures remain in their elastic region while subject to designed structural loads.

Limit state design on the other hand defines a set of ultimate limit states against which the structure has to be checked. These states can be static equilibrium of the structure (EQU), the strength of a structure (STR) and fatigue damage of the structure (FAT).

But don't these two design philosophies in the end boil down to the same thing, making sure the structure (steel structure for example) does not reach a stress too high? For example, we might check a steel beam against loads by calculating the expected characteristic loads such as wind and dead load, combine these according to different defined ultimate states and then check that the stress resulting from these does not exceed the yield stress of the beam. But isn't this in the end what allowable stress design does? What is the practical difference between the two?

EDIT: Making further research, I noticed that limit state codes do sometimes allow yield to occur. For example, in Eurocode 3, part 1-1, 6.2.5, it is stated that the resistance of a cross section against bending is determined using the plastic section modulus.

In this Youtube video, the differences are given:

Here the difference is given to be that in ASD, we use service loads which are the expected loads without any safety factors applied. Aren't these what are known as characteristic loads in limit state design? In contrast, it is stated that limit state design uses factored loads. So why is it that in ASD we use loads without any factors? Aren't safety factors there to take uncertainties in things like material properties and modeling theories into account? Why don't they exist in ASD?

Answer 1

I would say you can design anything using both methods, although I would consider LRFD more general.

In some areas, for example pressure vessel design, the maximum pressure in the vessel is basically known, also because the vessels have a safety valves. So using the actual load without factor makes sense. Then you have just one factor for calculating allowable stress and factor for weld efficiency, both of which can be multiplied, so at the end, there could be just one safety factor. So it is more practical to use ASD. And yes, partial plasticity is no problem for a safe design, so that is not the difference.

In Eurocodes, it makes more sense to use LRFD because many of the loads and their combinations are probabilistic. So using LRFD, you can give a safety factor to each load in a particular load combination. At the end, you could also transform all the uncertainties into one safety factor, but there would be no advantage in that, since this factor would be good only for this particular geometry, materials and load combination.

Every calculation has many uncertainties and LRFD allows you to decompose the calculation into individual components, which can have associated uncertainty factors (material properties, limit state model, geometry, loads). While this makes sense for some calculations, it is excessively complex for others.