As far as LRFD and ASD go...I've got a few comments. First, I was a little surprised by Focht3's seemingly very quick response to a question with so many sides like this by saying, "Advantages?! There aren't any!"...Don't you know that geotechs are forbidden from using such extreme words/phrases?? Ha Ha! Don't let John Bachner catch you using an extreme phrase like that. Hey, just kidding there! Maybe it's ok when not writing a report or something. It's strange because I tend to not use extreme words or phrases even in my personal life now. Some situations are very serious-see below:
My wife; "Hon, the car only has an 1/8 of a tank of gas left. Will we make it to the next city in 55 miles?"
Me; "Well Hon, we MAY make it."
I’m sorry, I digress in a feeble attempt to be funny.
Surely there must be something good about LRFD when applied to geotechnical engineering, right? I suppose there are also downfalls too. Both sides were discussed in the link given.
To me, the bottom line is that the design parameters we give to structural engineers to use, or our own interpretation of the behavior of a soil-structure system has to pass the final step; that is, we need to be confident enough to put our name on the design in the end and to uphold the standard of care in the development of the design with the end design usage in mind. The method used to do that apparently varies quite a bit from person to person.
Personally, I've always believed that because soil parameters vary so widely, relative to other construction materials, that a concept such as LRFD would lend itself quite well to soil mechanics. A more statistical approach, or a probability-based approach makes sense to me. As opposed to ASD where the factor of safety tends to be applied at the end of a design and it acts as a single blanket reduction factor for all parameters involved regardless of how confident we are in each. The factor of safety is very subjective.
If we had a way to deal with soil parameters independently, it seems as though that would make the most sense (unless, of course, if this method adds lots of additional time or effort to the work). We can usually estimate within a reasonable range the unit weight of a soil, particularly if we construct a quick phase diagram. On the other hand, hydraulic conductivity is much more elusive and with our best measurements and estimates the actual hydraulic conductivity insitu may vary substantially from what we thought it was. A calculation such as critical gradient in a dewatered cellular excavation, in which both parameters must be considered (unit weight and hyd. cond.) could be treated appropriately with an LRFD approach...in theory. In reality, no one will balk at adding 3 more feet of sheet pile in the ground so that we can make "extra sure" there won't be a problem with boiling at the base. Because of the reality of the situation, slicing the resistance factors so closely may be a waste of time, when it is economical just to add a little extra in the way of sheet pile.
I also believe that just about any soil analyses (for example allowable bearing pressure, i.e. settlement and bearing capacity, or slope stability, etc.) will benefit much from using a type of parametric analysis. Duncan's method mentioned in one of the other posts in the link given is very nifty. It’s always been beneficial for me to observe which soil parameters are affecting the output the most and how. Of course, the results you get are only as good as the level of knowledge you have in inputting parameters and checking different scenarios. What is the “probability” of unacceptable performance using this design over that design?” The question of probability, quantitative or even qualitative, can’t really be satisfactorily answered without a parametric study. Yet, this is usually the type of question the client is most interested in asking!
I look forward to comments!