Characterization of Method Uncertainty in Axial Pile Capacity Design Methods

The paper describes several approaches and procedures for deriving the mean, standard deviation and probability density function of the method uncertainty of five axial capacity pile design methods, i.e. the API, NGI, ICP, Fugro and UWA methods. A new unified database of high quality field tests recently developed in a Joint Industry research project (Lehane et al., 2017) is used for the quantification of method uncertainty. The focus of this paper is on the statistical description of the method uncertainty parameter for the API, NGI, ICP, Fugro and UWA methods for predicting the axial capacity of piles in sand and in clay. Probabilistic calculations of the axial pile capacity for typical offshore piles using the above five design methods (API, NGI, ICP, Fugro and UWA methods) showed that method uncertainty is a major contributor to the uncertainty in foundation performance. It can have a strong influence on the calculated annual probability of failure of a pile foundation, and thus on the safety level associated with the axial capacity of a pile. Establishing the capacity model statistics based on the measured (Qm) and calculated (Qc) values of pile capacity in the database requires careful consideration of several factors. Case studies demonstrate that, for a pile resistance factor calibration exercise, only the cases where the method overpredicts the actual (measured) capacity (Qm/Qc < 1) are of interest. Therefore, the part of the cumulative distribution function that represents those data should be fitted as well as possible. Another issue in the derivation of method uncertainty statistics is the possible dependency of the mean and standard deviation of method uncertainty on pile length and/or pile diameter. The paper will present the procedures for the characterization of method uncertainty and demonstrate its application through a case study for offshore piles. The effect of different interpretations of method uncertainty on the calculated annual probability of failure and the calibrated resistance factor will be presented in the paper.