Effect of soil gradation on embankment response during liquefaction: A centrifuge testing program

This paper describes a centrifuge study undertaken to investigate how sand gradation affects the system-level performance of embankments subjected to strong shaking. Current analysis and design practices are primarily based on knowledge from case history records of liquefaction, with the majority of those from sites consisting of clean, poorly graded sands. The narrow range of gradation characteristics represented in the case history database poses a challenge during the analysis of embankment structures traditionally constructed with, or founded on, more broadly graded soils. The tests herein were designed to elucidate how embankments uniformly constructed with a well graded and poorly graded sand perform differently during earthquake shaking. A centrifuge experiment test program was developed and conducted using the 9-m-radius centrifuge at the UC Davis Center for Geotechnical Modeling. The experiment design consisted of two submerged 10-degree embankments positioned side-by-side in the same rigid model container, with one embankment constructed with poorly graded sand and the other with well graded sand. The embankments were dry pluviated to the same relative density, but the absolute densities of the sands were different. The embankments were identically instrumented with dense arrays of in-situ sensors beneath the level ground above the slope and in the mid-slope to measure the dynamic response during liquefaction. Results showed that embankments constructed at equal relative densities would both liquefy (i.e., ru reach 1.0), but deformations were less severe for the embankments constructed with the well graded sand. Greater resistance to the generation and faster dissipation of excess porewater pressures, coupled with stronger dilatancy of the well graded sand increased embankment stability, curtailing liquefaction-induced deformations.