Reliability of Spread Footing Performance in Desiccated Clay

To advance the use of reliability-based design procedures, it is necessary to evaluate the sources of the design parameter uncertainty including inherent variability, measurement error, and transformation uncertainty. The results of a probabilistic evaluation of undrained footing bearing performance are discussed in the context of an extensive test site characterization described in a companion paper. Kriged cone tip resistance values are transformed into design parameters using a second-moment probabilistic approach and compared with the parameters obtained from the laboratory test analyses on specimens retrieved from the test site. The spatial, measurement, and transformation uncertainty are incorporated into probabilistic finite-element and bearing capacity analyses in which the results are compared against a full-scale load test performed at the test site. The results indicate that the reliable assessment of the spread footing response depends to a large degree on the assumed strength anisotropy and soil layering. A comparison of the probabilistic estimates of the bearing capacity to deterministic estimates developed from the correlations and average design parameters indicate that the consideration of uncertainty results in a desirable prediction accuracy. DOI: 10.1061/(ASCE)GT.1943-5606.0000706. (C) 2012 American Society of Civil Engineers.