Structures, water, computers, languages and people (not necessarily in this order)

From Allowable Stresses to Limit States

My geothecnical colleagues get annoyed when I say that foundation engineering is a step behind structural analysis because they still use a permissible stress approach instead of a limit state approach.

For a working stress formulation, the real magnitude of service loads is used in the whole design and the strength of the material (i.e. soil, concrete, plastic...) is divided by a huge coefficient. Under such scenario, the resistance is not used in the full extent and the cheapest design can hardly be obtained.

For a limit state conception, for each material and load, a partial load factor is assigned individually depending on the material properties and the likehood of occurrence. The material strength can be utilized to its maximum design value during its lifespan. Therefore, the calculations are more refined and the construction cost of the structure is lower. In the future calculations should be based on fully probabilistic models.

As far as I know, (correct me if I am wrong), most of the American codes, the Italian ones and the British standards still use the allowable stress approach even for some structural analysis. My advice is to have always a critical eye on the codes.


  1. I must confess myself as a determist, so I am a little reticent to probabilistic methods. But, in the end, I calm down myself thinking that any kind of modelling or calculation must be based on experience and observation, and so statistics are always, implicit or explicitically, involved in calculus and design.
    That's only my point of view, the "patch" I gave myself to feel not unsafe when handling probabilistic methods.

  2. Engineering is not perfect. Many US engineers still prefer the deterministic tecnique combined with the experience rather than the pseudoprobalistic method and their old fashioned calculations lead to nice and safe designs.