The effect of seismic preshaking history on the liquefaction resistance of granular soil deposits

Abstract

The results of the experiments showed that the combination of mild/moderate (Events A) to mildly strong (Events B) shakings resulted in a significant increase in liquefaction resistance of the deposits over time. However, the occurrence of extensive liquefaction, caused by an Event C or D, resulted in a dramatic immediate reduction in liquefaction resistance to a level comparable to, or even lower than, that of the young recent deposit before preshaking had started. This suggests a complex phenomenon, with most of the shaking events strengthening the liquefiable layer but some of them weakening it, sometimes dramatically when extensive liquefaction was involved.

The estimation of the liquefaction potential of granular soil deposits is mainly based on field liquefaction triggering charts linked to the Seed and Idriss Simplified Method. Currently these field liquefaction triggering charts seem to be generally conservative and do not take into account factors that tend to increase the liquefaction resistance of many natural sand deposits. Specifically, there are some field evidences suggesting that the Andrus and Stokoe (2000) liquefaction chart - based on field measurement of soil shear wave velocity - separates well liquefaction from no liquefaction of uncompacted artificial fills, but may underpredict the liquefaction resistance of natural sands in some seismic areas by as much as 100-200%. Most case histories supporting this conclusion are silty sand sites located in the Imperial Valley of Southern California. These deposits are only one to a few centuries old but they have been subjected to unusually high seismic activity. This effect of preshaking on the liquefaction behavior of sandy soils is systematically studied in this dissertation.

The research incorporated data from both the field, and centrifuge and full-scale experiments. State of the art tools and sensors were used in the tests in order to capture as precisely as possible the response of the soil and connect it to the field. The work included three centrifuge tests performed at RPI (Experiment 1-3) and a full scale test performed at University at Buffalo, UB (Experiment 4). The tests simulated the effect of several decades to several centuries of earthquake events on a 5-6 m uniform clean or silty sand horizontal deposit, including both events that liquefied the deposit and others that did not liquefy it. The experimental program involved testing different shaking sequences, in order to explore the complex relation in the field between preshaking seismic events that increase the liquefaction resistance of young deposits, and liquefying events that may decrease it totally or partially. Different shaking sequences were applied in the four experiments. However, all deposits were subjected to three or four basic event types: Events A, B and C (or D). An Event A was defined as 5 sinusoidal cycles of a peak base acceleration, apb ≈ 0.035-0.045 g; an Event B was defined as 15 sinusoidal cycles of a peak base acceleration, apb ≈ 0.04-0.05 g; and an Event C (or D) was defined as 15 sinusoidal cycles of a peak base acceleration, apb ≈ 0.1-0.25 g, all in prototype units. The prototype frequency in all cases was 2 Hz. The 15-cycle duration of Events B, C and D corresponds approximately to an earthquake of moment magnitude, Mw ≈ 7.5; while the 5-cycle duration of the Events A correspond to Mw ≈ 6. An Event A represents a mild to moderate earthquake shaking; an Event B represents a mildly strong earthquake shaking; and an Event C (or D) represents a strong to very strong extensive liquefaction shaking in the field.