| description abstract | A broad range of coarse-grained soils that vary in gradation uniformities, maximum particle sizes, and absolute densities are prone to liquefaction during earthquake shaking. However, clean, poorly graded sands form much of the liquefaction case-history database, with these soils often serving as the basis for analysis procedures. This paper presents a centrifuge testing program studying if the liquefaction triggering and deformation performance of embankment systems constructed with poorly graded sands universally applies to well-graded soils. Two soils were used in this study, named 100A and 25ABCD, had variations in maximum particle sizes, coefficients of uniformity (Cu), and void ratio extremes. Dense arrays of in-situ porewater pressure transducers and accelerometers indicate that liquefaction was triggered at near unity in the different soils. The 25ABCD soil, with its larger Cu and lower void ratio indices, exhibited stronger dilative tendencies, better preservation of the long period energy of the input motion, and more rapid dissipation of excess porewater pressures. The 25ABCD embankments had lower overall levels of strain at initial liquefaction triggering and accumulated less strain during cyclic mobility. While the two soils were pluviated to the same relative density and subjected to the same level of shaking, the slope surface displacements in the 25ABCD embankments were 60%–70% less than the displacements measured in the comparable 100A embankments. These results support the hypothesis that liquefaction and deformation behaviors depend on soil gradation, and findings from this experimental program can be leveraged for more accurate performance predictions of levees and earthen dams. | |
