Probabilistic Assessment of Model Uncertainty for Prediction of Pile Foundation Bearing Capacity; Static Analysis, SPT and CPT-Based Methods

Abstract

Geotechnical designs, like other engineering disciplines, are always accompanied by uncertainties. Supplying continuous and reliable records and reducing the uncertainty associated with measurement errors, the cone penetration test (CPT) enhances the geotechnical designs to a more decent level. Deep foundation design, as an essential challenge of foundation engineering, is also involved with different sources of uncertainty. Moreover, the presence of various design methods, relying on different assumptions and requirements, introduces further complications to the selection of an appropriate method, which leads to the broad spectrum of the predictions. Hence, a database, including 60 driven pile load test results and CPT records in the vicinity of them, was compiled to investigate the model uncertainties embedded in various predictive approaches. Investigated approaches include two static analyses, five SPT, and five CPT-based methods, and were implemented to predict axial pile bearing capacity. The model parameters for these methods are investigated through seven statistical, probabilistic, and reliability-based criteria. Performance of the methods under these criteria is assessed by the use of radar charts. Moreover, the resistance factor, adopted in this study to estimate the efficiency ratio and actual factor of safety, is calibrated by four different prevailing methods. Eventually, among conventional available predictive methods, CPT-based methods perform better than others and result in cost-effective and optimized trends.