Centrifuge modelling of liquefaction and lateral spreading of virgin, overconsolidated and pre-shaken sand deposits

This paper presents the results of five centrifuge model tests of earthquake-induced liquefaction and lateral spreading in sand using a laminar box. These centrifuge experiments simulate a gently sloping, H = 6 m thick stratum of saturated homogeneous sand of infinite lateral extent and relative densities of 45% and 75%. While four of the tests were conducted at a centrifugal acceleration of 30 g, the fifth experiment was a “model-of-the-model” verification at 50 g. Each deposit was subjected to the same lateral base shaking having a prototype peak acceleration of approximately 0.2 g, frequency of 2 Hz, and duration of about 22 cycles. In all models the soil was saturated with a viscous fluid such that under the increased gravitational field the deposit had the prototype permeability of fine-grained sand. Detailed discussions and comparisons of the five centrifuge tests are included. In two of the models, the sand was respectively overconsolidated (overconsolidation ratio, OCR = 4) and pre-shaken, in order to study the effects of these factors on liquefaction and lateral spreading; neither the pre-shaking nor the overconsolidation changed much the density of the soil. The study of effects of sand relative density, overconsolidation and pre-shaking is performed by comparison of three key measured parameters: thickness of liquefied soil H_l permanent lateral displacement D_H, and ground surface settlement S. It is found that either overconsolidation or pre-shaking of 45% relative density sand reduces H_l D_H and S as much as increasing the relative density of the soil to 75%. In addition, comparison with three similar centrifuge tests reported elsewhere which modeled a 10 m thick stratum of virgin sand, allows evaluation of the influence of deposit thickness. It is found that increasing the deposit thickness increases H_l, D_H and S.