UC Berkeley

The likely amount of free-field ground settlement following soil liquefaction can be used as an engineering demand parameter to assess the ground damage due to liquefaction. Its estimation can be improved using recently collected laboratory test results from cyclic tests performed on uniform clean sand, uniform nonplastic silty sands, and uniform nonplastic silts, and from new liquefaction-induced ground settlement field case histories developed from the reconnaissance efforts conducted in Christchurch and Wellington in New Zealand after the 2010-2011 Canterbury earthquake series and the 2013-2016 northern South Island earthquakes, respectively. This new information has been incorporated in the development of a probabilistic procedure for estimating liquefaction-induced free-field ground settlement. The strain potential of liquefied soil is studied using a large database of cyclic laboratory test results on nonplastic uniform clean sands, silty sands, and silts. This database enabled exploring the cyclic maximum shear strain and the post-liquefaction volumetric strain responses using different measures of state (i.e., relative density, void ratio, and state parameter). In contrast to current state-of-practice approaches that are based on clean sand, this study developed a series of probabilistic models that capture the post-liquefaction volumetric response of a wide range of nonplastic uniform soil. To complement the insights gained from the laboratory testing, a comprehensive database of 205 well documented case histories of liquefaction-induced free-field ground settlement was developed. Well-documented field case histories provide valuable information about the interacting effects of variable soil properties within sites with differing stratigraphic profiles, and multi-directional ground shaking of differing intensities and duration. This information is key for developing robust empirical models. A case history results from the combination of a site with laterally consistent stratigraphy with at least one cone penetration test, an earthquake event, and post-liquefaction ground settlement measurements. In this thesis, case histories are classified into hydraulic fills and natural soil deposits. Hydraulic fills are relatively uniform hydraulically deposited structures whereas natural soils are heterogeneous layered systems that result from complex geological processes. The proposed probabilistic model for estimating liquefaction-induced free-field ground settlement is based on the laboratory-based relationships for post-liquefaction volumetric strain developed through this research. Comparison against high-quality case histories led to the inclusion of adjustment factors in the model to capture field observations. As a result, the model captures the influence on settlement of the ground motion intensity and duration and the site’s compressibility. The amount of ground settlement largely depends on the subsurface soil state. Hence, as part of this study, correlations to estimate relative density and the state parameter are developed to enable use of the laboratory-based volumetric strain models for a wide range of different soil types. In engineering practice, the estimation of liquefaction-induced ground settlement is separated from the ground motion intensity measure estimate. Hence, settlement is typically estimated based on a single ground motion intensity measure. A performance-based engineering procedure is employed so the full range of the ground motion intensity measure is considered when estimating annual rates of exceedance of liquefaction-induced ground settlement. The procedure convolves the hazard calculation for the ground motion intensity measure with the probabilistic model for liquefaction-induced free-field ground settlement to produce hazard curves for liquefaction-induced ground settlement. The uncertainty in the inputs to the seismic hazard component and in the inputs to the ground settlement estimation are explicitly incorporated in the procedure using a logic tree approach. The proposed procedure can be used in practice to perform a performance-based assessment of liquefaction-induced ground settlement and its resulting damage to infrastructure. Lastly, a soil specimen subjected to the constant shear drained stress path can transition suddenly from a stable drained shear condition to an unstable undrained mode of shear. It is a dangerous triggering mechanism because instability is triggered without warning and at small deformation levels (typically < 1% in shear strain). The constant shear drained-to-undrained mode of shear is thought to be a primary mechanism of static liquefaction flow failures. For example, it is thought to be the triggering mechanism leading to the Aberfan coal tip failure in 1966 and the Fundao mine tailings failure in 2015. However, the data available from tests performed with this stress path are largely limited to clean sands. Few tests have been performed on fine sand and silt tailings. Due to the relevance of this stress path as a triggering mechanism leading to instability of mine tailings materials, a series of dense-of-critical and loose-of-critical state constant shear drained stress path triaxial tests have been performed on test specimens of a tailings silty sand. Supporting laboratory tests that provide a thorough mechanical characterization of the material are provided (e.g., isotropically consolidated drained and undrained triaxial tests, and one-dimensional compression tests). The formulation of a generalized critical state constitutive model (i.e., NorSand) is extended so it reproduces key responses observed during the laboratory tests performed in this study. The extended NorSand constitutive model can be used to examine the field response of this material and similar tailings materials.