|Residual soil behavior can be difficult to predict using current geotechnical formulas, because it exhibits properties that are not common to transported soils. These unique properties are mainly influenced by the fabric structure of the soil and cemented nature of weathered-in-place soil, which can result in increased amounts of strength when compared to transported soils. Proper characterization and modeling is further complicated due to the fact that residual soils often contain a high percentage of silt and clay sized particles which can introduce large amounts of apparent cohesion when the soil is unsaturated. This apparent cohesion can be detected by common insitu tests, but is often times incorrectly identified as being a result of cementation or fabric structure. Such classification can be dangerous for use in designs, because the apparent cohesion of the soil is reduced as the water content of the soil increases. Conversely, it is also common practice to ignore the cohesive component of the soil completely, which can lead to designs that are inefficient.
In this research, two excavations were instrumented at the Auburn National Geotechnical Experimentation Site (NGES) in Opelika, AL. Preliminary modeling was used to determine the depth of excavation that would remain stable when unsaturated, but would become unstable as the surrounding soil neared saturation. The excavations were constructed approximately 6m deep x 30m long with a vertical face. The primary goal of this experiment was to determine the boundary conditions that resulted in failure of the excavation, and compare the results to a finite element model with the same boundary conditions. In doing so, conclusions could be drawn regarding the accuracy of common laboratory test methods for estimating the strength properties of residual soil.
The instrumentation plan was designed to monitor real-time pore water pressures (positive and negative) surrounding the excavation, as well as the deflection throughout the course of each 1-year test period. Time-lapse cameras were used to identify when failures had occurred, and the approximate geometry of the failure planes. Although the Auburn NGES is a highly characterized site, undisturbed soil samples were taken and used in conjunction with previous soil test results to accurately define the material properties and layering based on common laboratory test methods. In addition to common laboratory tests, unsaturated triaxial tests were also conducted, and soil-water characteristic curves were measured to further define the unsaturated properties of the soil.
In both excavations, failure was observed along a similar plane, which began at the bottom of the excavation, and propagated to the surface (approximately 2m behind the face of the excavation) along existing tension cracks that were developed during the construction of the excavation. The boundary conditions and the laboratory soil properties were input into a finite element model, and the model predicted a factor of safety of approximately one at the critical state, with the factor of safety being significantly higher when dry conditions were simulated. Based on these results, recommendations were made regarding the most appropriate test methods for determining the strength properties of residual soil for use in geotechnical design.