Practitioners and researchers have an array of deformation-based methods from which to choose when assessing the seismic stability of earth structures. However, with this variety comes confusion in the seismic hazard community on how to select an appropriate method for a given problem. Basic questions center on issues of method accuracy, precision and uncertainty. The validity of certain assumptions associated with these methods is also in question.;In this study, a database of case histories of earthquake-induced deformation in slopes was compiled to address the issues of method accuracy and precision. These case histories were analyzed with 20 deformation-based methods using Monte-Carlo simulation to propagate uncertainties in the seismic resistance and demand. Accuracy of the methods was assessed by comparing the median predictions to the field-measured displacements. Precision was assessed through statistical analyses and systematic inter- and intra-method comparisons. In a separate study, numerical analyses were performed on several hypothetical slope models to assess the validity of the limit-equilibrium pseudostatic surface assumption.;The predictive accuracy was found to be remarkably consistent among methods evaluated with the majority under-predicting the actual displacement. About 50% of the total number of predictions fall within +/-20 cm of the measured amount. Overall, a variety of methods can be expected to yield predictions that are within +/-100% of the actual amount.;Non-linearity of the displacement-acceleration ratio relationship was found to have an amplifying effect on input parameter uncertainties and significantly influences precision. In addition, method precision is dependent on: (1) location along displacement-acceleration ratio relationship and (2) characteristics of the seismic loading which control the shape of this relationship. Both of these suggest precision is dependent on site-specific information.;Comparisons made between the pseudostatic and fully-coupled surfaces suggest that dynamic response affects the geometry of the failure mechanism for wavelength ratios (lambda/H) between 1 and 4 where surfaces can be up to 20% smaller in volume. For these conditions, the pseudostatic surface assumption is not appropriate and coupled analyses are recommended. For lambda/H greater than 4, dynamic response has less influence on the failure mechanism geometry and methods based on the pseudostatic surface assumption (rigid-block and decoupled) are appropriate.
展开▼
