岩石邊坡可靠度設計之探討

Abstract

本研究探討岩石邊坡穩定之可靠度設計，分別探討平面滑動與岩土界面滑動兩種邊坡破壞機制之岩石邊坡可靠度問題。傳統容許應力設計並無明確對應之破壞機率，本研究探討一、二、三級可靠度設計之可行作法，比較其準確性與簡易性，以期能對於大地工程實務推動可靠度設計有所助益。針對平面滑動破壞機制，研究目標在於提出擬定LRFD所需部分係數之可行流程，令其適用於不同參數條件，使所提出之部分係數標定方法有望能用於未來之規範訂定。部分係數率定是以控制可靠度為前提，因此先將率定案例以第二級可靠度方法AFOSM調整至目標可靠度，再進行第一級可靠度方法LRFD所需之部份係數率定，並以第三級可靠度方法之蒙地卡羅模擬驗證其正確性。針對沿岩土界面滑動破壞機制，本研究運用三種切片分析方法，分別為傳統切片法、簡化Bishop法及Wedge法，以蒙地卡羅模擬與AFOSM探討破壞機率，檢視其破壞模式之特性與參數敏感性，並比較AFOSM分析與蒙地卡羅模擬結果之差異。最後結果包含各種可靠度分析之準確性與簡易性的比較。本文之研究顯示：平面滑動破壞模式之參數式部分係數率對允許風險之均一性有所提升；岩土界面滑動破壞模式之AFOSM分析之破壞機率與蒙地卡羅模擬所得結果相仿。

This thesis explores the methodology of reliability design for rock-slope stability involving the plane-slide and rock-soil-interface slide mechanisms. Conventional slope stability design based on factor of safety does not necessarily provide a consistent probability of failure. The applicability of the first, second and third level of reliability design on rock-slope stability problems was studied; the accuracy and relative simplicity of these methods were compared. This study attempted to promote the reliability design in geo-engineering practices. For the plane-slide failure mechanism, the aim is to develop a methodology for the calibration of the required partial factors in LRFD, applicable in code calibration in the future. The partial factors were calibrated against target reliability. First, the reliabilities of various cases were adjusted to the target reliability by the second level reliability method, AFOSM. Next, the partial factors for the first level reliability method, LRFD, were calibrated. After that, the consistency of reliability by LRFD was verified by the third level reliability method, the Monte Carlo Simulation. For the rock-soil-interface slide mechanism, three types of slice methods, including the conventional slice method, the simplified Bishop method and the wedge method, were implemented. The probability of failure was analyzed by the Monte Carlo Simulation and AFOSM to examine the characteristics of failure mode and the sensitivity of parameters, and to compare the accuracy and simplicity of these reliability methods. This study reveals that the consistency of allowable risk can be achieved by using the parametric LRFD in routine designs for plane-slide mode. For the rock-soil-interface slide mode, the failure probability obtained by AFOSM and the Monte Carlo simulation are almost identical.