Calibrating and comparing reliability analysis procedures for slope stability problems

Jie Zhang,Wilson H. Tang,Limin Zhang,Yinren Zheng
2007-01-01
Abstract:Reliability analysis, in most cases, can only be implemented in an approximate way. This produces two subsequent questions: (1) how approximate is the approximate procedure is? (2) how the relative validity of different approximate procedures can be compared? In this paper, the practical difficulties for rigorous reliability analysis of Hong Kong slopes are addressed. Two types of approximate reliability analysis procedures for the analysis of Hong Kong slopes have been calibrated and compared using field observation data. It is found that the procedure which does not consider soil suction (P1) is highly biased towards the conservative side, while the procedures considering suction (P2-1 and P2-2) are biased towards the unsafe side. These approximate procedures are calibrated with field observation data to estimate the actual reliability index and the actual failure probability of the system. A comparison among different procedures shows that the calibrated P1 is the best procedure in terms of prediction accuracy. As discussed in this paper, the reliability analysis procedure with a more rigorous soil mechanics background may not always result in better results in terms of prediction accuracy. Whether a procedure with a more rigorous theoretical background could produce a more accurate prediction depends heavily on the quality and quantity of information available.
What problem does this paper attempt to address?