Performance of Embankments on Liquefiable Soils Improved with Dense Granular Columns: Observations from Case Histories and Centrifuge Experiments

Dense granular columns (DGCs) are generally known to mitigate the liquefaction hazard through a combination of (1) installation-induced ground densification, (2) enhanced drainage, and (3) shear reinforcement. However, the relative contribution of these mitigation mechanisms remains poorly understood. A recent case history of successful embankment performance on a liquefiable site treated with DGCs that had relatively low area replacement ratios ($A_r$) (where drainage is not notably enhanced) suggested that shear reinforcement and installation-induced ground densification may be the two dominant mitigation mechanisms provided by DGCs. In this paper, we present a series of four dynamic centrifuge experiments designed and conducted to test this hypothesis under controlled conditions. Consistent with case history observations and supporting our initial working hypothesis, densification combined with shear reinforcement was shown to be primarily responsible for limiting the embankment’s seismic deformations. Additional drainage led to minor improvements in terms of embankment settlement, while increasing its permanent lateral displacement. The results suggest that the combined effects of $A_r$ and ratio of maximum shear modulus of the DGCs to that of the surrounding soil ($G_r$) can play a key role in the distribution of stress between DGCs and soil prior to shaking and the extent of softening and strain accumulation in various layers during shaking. For example, it was observed that densification of the liquefiable sand layer around DGCs shifted the generation of larger excess pore pressures to greater depths compared to the DGC-treated test without densification. This led to a base isolation effect that reduced accelerations, degree of softening, and accumulation of shear and volumetric strains at shallower depths, producing a notably improved performance for the soil–embankment system even when the DGC’s drainage capacity was inhibited. These observations were attributed to the reinforcement effect of DGCs, the simultaneous reduction in $G_r$ due to densification, and a more even transfer of the embankment load onto the soil–column matrix, increasing the stiffness and strength and reducing shear strains in the shallower and looser layer. The presented experimental results point to the importance of accounting for pre- and postinstallation soil density and stiffness in relation to DGCs, confining pressure distributions, kinematic constraints, and activity of various mitigation mechanisms when evaluating the potential influence of DGCs on seismic demand, liquefaction triggering, and deformations near embankment structures.