Staged construction of embankments on soft ground remains one of the most challenging topics in geotechnical engineering due to the complex shear and consolidation behavior of clays. This thesis presents a case study on the performance of the New Hamilton Partnership (NHP) levee in Novato, California. This 11ft high levee was constructed over 30 ft - 40 ft thick layer of San Francisco Bay Mud during a six-month period in 1996. Settlements along the levee crest were monitored over a period of 5.2 years after the end of construction (until early 2002), at which time URS installed piezometers to measure the existing consolidation stresses (s'vc) within the Bay Mud. URS also conducted state-of-the art field and laboratory test programs to develop well-defined values of preconsolidation stress (s'p) and compressibility parameters for the Bay Mud. However, conventional 1-D consolidation analyses greatly underestimated the measured levee settlements. Hence URS reduced s'p by 20% for the Plaxis FE analyses with the Soft Soil Model (SSM) used to replicate the performance of the existing NHP levee and then to design an expanded levee system. This thesis presents a detailed re-evaluation of the NHP levee performance and of the stress history, strength, and consolidation properties of the Bay Mud obtained during the URS geotechnical site investigation.(cont.) New conventional 1-D consolidation analyses with higher values of the recompression ratio and revised profiles of s'vc indicate that the measured levee settlements at 5.2 years can be matched when s'p is reduced by 10% to 15%. The thesis also presents two series of Plaxis analyses with the Soft Soil Model. The first evaluated SSM parameters to better model results from the laboratory consolidation and K0-consolidated undrained shear tests on the Bay Mud. The second series conducted 2-D FE analyses to identify the most important variables effecting the predicted performance of the levee during and after construction. These parametric analyses show that the measured settlements during the 5.2 year period and the excess pore pressures measured in early 2002 can be consistently described only after careful definition of four major variables: the recompression ratio, RR, the normally consolidated coefficient of consolidation, cv(NC), and the preconsolidation stress, s'p, of the Bay Mud; and the boundary drainage conditions. The measured performance is best matched by using values of cv(NC) and s'p that are less than measured by the laboratory CRSC tests.(cont.) Analyses with more sophisticated soil models are needed before definitive conclusions can be reached regarding the in situ properties of the Bay Mud and whether no not secondary compression (creep) plays an important role during primary consolidation (i.e., Hypothesis A versus Hypothesis B).
