Physical model studies of seismically induced deformations in slopes

Physical model experiments were conducted on a 1-g shaking table with the goals of: (1) investigating the mechanisms of seismically induced permanent deformations in slopes and embankments, (2) assessing the accuracy and applicability of the popular "Newmark-type" procedures for estimating deformations in slopes, and (3) developing a suite of fully defined "model-scale" case histories for calibration of existing numerical procedures for predicting seismic slope deformations and for the future development of advanced numerical analyses. The inclined plane experiments indicated that the Newmark (1965) sliding block procedure generally provides unconservative estimates of deformation when the predominant frequency of the input motion is 0.2 to 1.5 times the natural frequency of the sliding mass. Conversely, the procedure was found to be generally conservative when the frequency ratio is in the range of about 1.5 to 2.2. The inclined plane tests also suggest that one-dimensional decoupled deformation analyses are generally conservative, with decoupled analyses overpredicting measured deformations by about 10% to 20% for the tests analyzed. The model slopes were commonly observed to displace along two or more localized shear surfaces. The multiple shear surfaces were typically of the same orientation and generally located within relatively close proximity of each other. Surface deformations varied over the length of each model, with the largest displacements occurring at the toe or along the face of the slope. Newmark's (1965) assumption that deformation occurs along a single surface reasonably approximated the actual deformation behavior for tests where multiple shear surfaces developed in close proximity to each other. For approximately half the tests, however, the single surface assumption proved to be an oversimplification. Accuracy was improved when the Newmark (1965) procedure was modified by using the acceleration-time history recorded in the soil immediately below the shear surface in lieu of the base acceleration to compute deformations. The accuracy of both the original (1965) and modified Newmark formulation was greatest when a degrading yield acceleration was used to model the soil's transition from peak to residual shear strength. Back-analyses of the slope tests indicate that the Newmark analysis is best performed: (1) sing shear rate-corrected soil or geosynthetic shear strengths; (2) by properly modeling the soil or geosynthetic displacement versus shear strength relationship; (3) using an acceleration-time history that is appropriate for the base of the slip surface; (4) recognizing that sliding block procedures only account for localized deformation. (Abstract shortened by UMI.)