Comparison of Informed and Un-Informed Shear Strength Reduction Procedures for Finite Element Method Slope Stability Analysis
Shear Strength Reduction , Slope Stability , Finite Element Method , Geological Engineering , Geotechnical Engineering
Geotechnical engineers commonly rely on numerical models such as the Finite Element Method (FEM) to analyze natural, civil, and mining slope stability problems. One of the goals of these analyses is typically to determine a Factor of Safety (FoS) like that obtained using Limit Equilibrium Methods (LEM). To compute the FoS, Shear Strength Reduction (SSR) techniques are commonly employed. The SSR procedure involves the systematic computation of elastic-plastic FEM slope models with differing strength parameters to solve for the FoS within a specified tolerance. This is generally achieved using a monotonically increasing or a bracketing and bisecting technique. These procedures are standard in most geotechnical FEM software packages. The advances in personal computing technology have aided the adoption of the SSR technique into engineering practice. The technique provides significant advantages over LEM techniques primarily pertaining to the lack of assumption about failure surface and forces, and the ability to compute deformation. One of the major drawbacks of the FEM compared to LEM is the time required to complete the analysis. A single FEM SSR analysis can take hours to days to complete if using a very fine mesh. There has been limited academic literature on the topic of improving SSR procedure runtime, structurally. This thesis presents a framework for a technique called “informed” SSR. Informed SSR procedures utilize previously computed SSR stages to provide information to subsequent SSR stages to develop a better initial guess for the FEM solution. These informed procedures are compared to “un-informed” procedures included in Rocscience’s RS2 which solve each SSR stage independently of all other SSR stages. This thesis also assesses optimization of informed SSR procedures for modified Strength Reduction Factor (SRF) step sizes and constant strength parameter reduction techniques. The cases analyzed in this thesis show that informed SSR results in drastically decreased run times when compared to analogous un-informed procedures. This is achieved whilst not drastically changing the values, such as FoS and nodal displacement, output by the un-informed procedures.