Engineering evaluation of permanent ground deformations due to seismically-induced liquefaction
[Thesis]
M. H. Baziar
R. Dobry
Rensselaer Polytechnic Institute
1991
327
Ph.D.
Rensselaer Polytechnic Institute
1991
A study is presented on the engineering evaluation of permanent lateral ground deformation associated with liquefaction and lateral spreading during earthquakes. Newmark's (1965) rigid sliding block model is used as main analytical technique, based on the reasonable assumption that generally the deformation occurs during shaking in the downhill direction of the ground surface slope. The study involves laboratory tests, analytical work, and comparisons with case histories and field observations. It culminates with a simplified method to predict lateral deformation using regional attenuation relations for peak ground acceleration and velocity. The laboratory work includes two series of tests. In the first series, it is concluded that the Newmark's method can be applied to dry sand but not to saturated medium dense or dense sand, due to their dilative character and to the shape of their dense stress-strain curves, which do not exhibit the strength plateau required by the method. In the second test series, undrained monotonic and cyclic triaxial/torsional experiments were conducted on very loose, layered, pluvially deposited silty sand SF-7 obtained from the hydraulic fill in the Lower San Fernando Dam. It was found that the stress-strain behavior is contractive in all tests and for a range of Kc = 1.0 to 1.5, the steady-state strength on 45 planes is qus = Su = 0.145 usd\bar{\sigma}\sb{\rm 1c}usd. This is interpreted as Su = 0.145 usd\bar{\sigma}\sb{\rm v}usd, in the field, where usd\bar{\sigma}\sb{\rm v}usd = static effective vertical overburden pressure, and it is used in subsequent calculations as representative of very loose, mildly sloping in-situ Late Holocene natural fluvial deposits typically affected by lateral spreads. The sliding rigid block model in conjunction with available accelerograms and Su/usd\bar{\sigma}\sb{\rm v}usd = 0.145 is successfully applied to predict measured lateral displacement d in the Western United States. The lateral displacements obtained using these accelerograms are compared with Youd and Perkins' chart and empirical equation, with very good agreement. A general, practical engineering procedure is developed to evaluate lateral spreads at different sites and regions around the world. This method is calibrated with the LSI measured in the Western United States, with excellent agreement, and it is found that it predicts lower values of LSI than Youd and Perkins' empirical expression at short distances from the fault, for which no LSI data are available.