An Investigation of Liquefied Shear Strength Using Novel Centrifuge Tests, Direct Simple Shear Tests, and Field Case Histories
Advisor: Professor Scott M. Olson
In person and virtual
Abstract
Recent high-profile catastrophic ash pond and tailings dam failures, which were classified as flow liquefaction cases, renewed the interest in the mobilized shear strength during flow liquefaction, usually termed liquefied shear strength, su(liq). Statistics indicated that although the number of all failures and incidents decreased in the past 30 years or so, the number of “very serious” and “serious” failures actually constituted an increasing percentage since 1960. The average loss for each “very serious” failures was over $500 million. The reality in the waste material industry indicates an urgent need for safety improvement, which highly relies on the strength of the deposited material within the impoundment.
To improve the understanding of the behavior of soils within a tailings or coal ash impoundment and to better quantify the shear strength of the soil if liquefaction happens, three research methodologies were adopted in the current study: (1) field flow liquefaction failure back-analysis; (2) element-wise direct simple shear testing; and (3) centrifuge modeling.
The field flow liquefaction cases served as valuable information for liquefied shear strength study because those failures provide the most realistic estimates of su(liq). A maximized database of liquefaction-induced flow failures (71 cases) was built in the current study, including 13 new cases identified in addition to the database compiled by past investigators. Previously analyzed cases were reviewed, and adjusted if needed. Two predictive models based on field test indices, cone penetration test (CPT) tip resistance and standard penetration test (SPT) blowcount were developed based on the cases with the best quality and analyzed using the most rigorous procedures. The predictive models developed in the current study highlighted the importance of compressibility of the soil in the field.
Element-wise, laboratory direct simple shear (DSS) tests were used to characterize the critical state properties of three nonplastic soils, including the one used in the centrifuge modeling. Other than that, various strain rates were used in DSS tests to supplement the strain rate range used in the centrifuge modeling. Those DSS tests under controlled drainage conditions indicated a negligible or minor (if existed) strain rate effect. Moreover, various complicated loading scenarios were adopted in DSS tests to closely mimic the loadings in the centrifuge modeling to provide insights for interpreting centrifuge test results.
A novel centrifuge model, which involved pulling a thin metal plate (i.e., coupon) through the liquefied soil horizontally at a constant initial effective vertical stress with a constant velocity, was developed to mimic the shearing along a critical slip surface in a flow liquefaction failure. The force needed to pull the coupon through the liquefied soil was converted to stress to reveal the mobilized shear resistance along the coupon-sand shear surface. By varying the velocity of the coupon, two distinct shear surface behaviors were identified via interpreting the porewater pressure response at the coupon-sand shear surface, indicating that the behavior of the soil around the coupon was greatly affected by the velocity (i.e., strain rate). Pulling forces from low-velocity tests were used to extract su(liq). In addition, free-field su(liq) were obtained using inverse analysis procedures when the shaking was intense enough to trigger liquefaction within ¼ of the first cyclic loading cycle. All obtained su(liq) from centrifuge tests were related to cone tip resistance measured in-flight. In contrast to DSS tests, su(liq) obtained from the centrifuge model exhibited strain rate dependency, which was successfully captured by a Bingham plastic model. Dimensionless parameters were introduced to reveal the physics of the liquefied soil in the centrifuge model. su(liq) obtained from the centrifuge model were compared to the values back-calculated from field flow failure cases. The back-calculated su(liq) from field flow failures agreed well with the centrifuge su(liq) where the soil could be classified as Bingham plastic model according to the dispersive-viscous ratio (Dv*, the ratio between the grain inertial stress and the fluid viscous stress). This indicates that the Bingham plastic model parameters might be useful for describing su(liq) mobilized in the field.
