Reliability analysis of deep excavations by RS and MCS methods: case study

Abstract

In recent years, uncertainty inherent in geotechnical properties has caught great attention of researchers and engineers; thus, reliability analyses and non-deterministic methods have been applied widely. This paper presents the reliability analysis of a case study on deep excavation in a cemented coarse-grained alluvium adopting random set (RS) and Monte Carlo simulation (MCS) methods to tackle the inherent uncertainty associated with soil properties. The horizontal displacement of the excavated wall and safety factor of stability were adopted as a basis for assessing the system response. Pertinent codes were written by employing RS and generating random numbers of probability distribution functions in FLAC2D finite difference program. Introducing RS code into the numerical analysis process drastically reduced the computational effort required for this method. It was found that in cases of deep excavations with low target probability of failure, RS presents satisfactory performance and thus is preferable to MCS, while MCS can be employed as a beneficial tool for stability analysis and some special cases of deformation analysis in excavation problems. The results are presented and compared in terms of sampling procedure, as well as statistical moments, probability of occurrence and most likely values at every stage of construction.