Probabilistic assessment of the seismic performance of earth slopes

• Published in: Bulletin of earthquake engineering 2014-06, Vol.12 (3), p.1071-1090
• Main Authors: Rathje, Ellen M., Wang, Yubing, Stafford, Peter J., Antonakos, George, Saygili, Gokhan
• Format: Article
• Language: English
• Subjects: Blocking; Civil Engineering; Earth and Environmental Science; Earth Sciences; Earthquakes; Environmental Engineering/Biotechnology; Geophysics/Geodesy; Geotechnical Engineering & Applied Earth Sciences; Ground motion; Grounds; Hazards; Hydrogeology; Mathematical analysis; Original Research Paper; Plate tectonics; Probabilistic methods; Probability; Probability theory; Seismic activity; Seismology; Sliding; Slope stability; Structural Geology; Vectors (mathematics)
• ISSN: 1570-761X; 1573-1456
• DOI: 10.1007/s10518-013-9485-9

Sliding block displacements are used to evaluate the potential for seismic slope instability. Deterministic approaches are typically used to predict the expected level of sliding block displacement, although they do not rigorously account for uncertainties in the expected ground shaking, dynamic response, or displacement prediction. As a result, there is no concept of the actual hazard associated with the displacement computed by the deterministic approach. This paper summarizes and extends recent developments related to the probabilistic assessment of sliding block displacements. The probabilistic approach generates a hazard curve for displacement in which the annual rate of exceedance for a range of displacement levels is computed. The probabilistic approach is formulated both in terms of scalar hazard analysis (i.e., using one ground motion parameter, peak ground acceleration) and vector hazard analysis (i.e., using two ground motion parameters, peak ground acceleration and peak ground velocity), and applied both to rigid and flexible sliding block conditions. Generally, the vector probabilistic approach predicts displacements that are 2–3 times smaller than the scalar probabilistic approach, revealing the value of characterizing frequency content via peak ground velocity. Comparisons between the deterministic and probabilistic approaches, in either a scalar or vector context, indicate that the deterministic approach can severely underestimate displacements relative to the probabilistic approach because it ignores the aleatory variabilities in the dynamic and sliding responses of the sliding mass. This under-prediction is most significant for longer period sliding masses. Modifications to the deterministic approach are proposed that provide displacements that are more consistent with the probabilistic approach.