Effects of void redistribution on post-earthquake residual strengths for liquefiable soils

1. The post-liquefaction strength of soil can be complicated by void-redistribution, whereby water flow can lead to localized loosening and weakening of soils at interfaces with lower permeability layers. 2. This complicated pattern of water flow and migration through a soil profile cannot be recreated in an undrained lab test, which is why shear strength evaluations based on lab data for samples at their pre-earthquake void ratios can yield higher strengths than those suggested by the case-history based empirical correlations. 3. The recent modification by Idriss and Boulanger [4] to the commonly used case-history based correlations allows for an explicit consideration of void redistribution when evaluating shear strength. However, the challenge of determining whether the conditions at the site will lead to void redistribution or not still remains. 4. Previous studies of void redistribution have shown that void redistribution and shear localization depend on all the factors that affect the generation and diffusion of earthquake-induced excess pore water pressures; e.g., initial relative density (Dr),geometry (slopes and stratigraphy), permeability (magnitudes and contrasts), and earthquake loading characteristics (intensity, duration, and frequency content). 5. A recent series of centrifuge tests of mildly sloping ground with a clay crust overlying loose saturated sand suggests that: a. Prefabricated drains can prevent shear localization even when ru=100% has been triggered in the treated zone. b. Cracking of the lower-permeability surface layer can facilitate boiling and bleeding of the accumulated excess pore pressures, thereby allowing the soil to regain its shear strength and bring shear localization and lateral spreading to a stop. 6. These and other centrifuge model test results can be used as a basis for evaluating the ability of coupled numerical models [e.g., 8, 9] to distinguish between cases where liquefaction will lead to void redistribution and localized shear strength loss and those cases where it will not. Systematic evaluations of coupled numerical models against sets of detailed experimental data that represent a range of conditions are necessary for guiding further developments and improving our ability to account for void redistribution effects in practice.