System safety assessment with efficient probabilistic stability analysis of engineered slopes along a new rail line.

A proper stability assessment of engineered cutting and embankment slopes is crucial to safe train operation and the performance management of rail infrastructure. Through the presentation of a case study, this paper develops a methodology for the safety evaluation of large-scale slope systems incorporating efficient probabilistic stability analysis of engineered slopes for a long rail line under construction. A long geotechnical slope is equally segmented into multiple consecutive sections according to the representative value of the local failure lengths of three-dimensional slopes, and each section is assessed for its probability of failure ( P f ). Soft computing by multivariate adaptive regression splines (MARS) is incorporated into the reliability analysis of a batch of slope segments. The construction of a MARS model requires a subset of data samples obtained from a limit equilibrium slope stability program; the validated MARS model is then used to generate the probability of failure of the rest of slope sections. The P f values for 2691 sections in total, determined by either direct probabilistic stability analysis or the MARS-derived predictive model, is subsequently introduced into a reliability-based performance evaluation of the long railway geotechnical slope system. The k-out-of-n system model is adopted to characterize the relationship between the entire system and its components concerning safety. The effect of remediation on the reliability of geotechnical system is finally explored by examining the variation characteristics of P f with a tolerable number of failure segments in the system. The proposed methodology can be readily extended to assess large-scale geotechnical systems for an operational rail line. [ABSTRACT FROM AUTHOR]