Quantitative Soil Behavior Type Assessment from Cyclic Test Results

The determination of soil behavior in response to cyclic loading is the first step in evaluating seismic ground failure. Cyclic soil behavior is generally categorized as “sand-like” or “clay-like.” This classification is related to the excess cumulative and within-cycle pore pressure development under undrained (constant volume) conditions and its resulting effect on strength and stiffness. Soils exhibiting sand-like behavior may undergo initial liquefaction (within-cycle pore pressure ratios approaching 1.0) depending on their state and loading conditions, which leads to “pinched” stress–strain loops of limited thickness near the strain origin. Conversely, soils with clay-like behavior typically generate limited excess pore pressure, with wider stress–strain loops, but can experience progressive strength loss through cyclic softening. Over time, researchers have shifted from using field case history datasets to laboratory test datasets for developing liquefaction susceptibility models. The interpretation of cyclic laboratory data to develop liquefaction susceptibility models has included analysis of cyclic stress–strain loops, rate of strain accumulation, and excess pore pressures. Interpretation of monotonic undrained tests has focused on the degree to which critical state lines are parallel to isotropic consolidation lines. Such interpretations have been based on judgement, because systematic criteria for interpreting susceptibility from lab test results were not available until recently. This paper presents a quantitative model describing the extent to which a soil is sand-like or clay-like based on cyclic laboratory tests contained within the Next Generation Liquefaction database. We selected a set of 64 cyclic direct simple shear (CDSS) test specimens that span a broad behavioral range. Each analyst assigned a cyclic behavior rating distribution between 0 (clay-like) and 100 (sand-like) based on CDSS results alone, i.e., in the absence of any other behavioral or classification information. Expert evaluations are used to develop relationships between interpreted soil behavior and various quantitative metrics derived from cyclic tests (e.g., pore pressure ratio). Moreover, we assess the degree of between-analyst differences in test interpretation. While not the subject of this manuscript, the ultimate aim of this research is the development of probabilistic liquefaction susceptibility models dependent on various independent variables, including soil index properties and cone penetration test results by using laboratory test results directly.