A database of 156 prestressed concrete piles was gathered to evaluate load and resistance factor design (LRFD) phi factors for eight AASHTO axial pile prediction methods. The majority of the piles had standard penetration test (SPT) data, with limited cone penetration test and laboratory information. In estimating the LRFD phi factors, variability associated with load test capacity estimation, as well as soil parameter assessment by means of in situ tests, was considered. Using an advanced first-order second-moment method, the LRFD phi factors were computed considering all variability (model, load testing, and soil parameters), as well as just the model (i.e., one of the eight AASHTO methods). Current AASHTO LRFD phi factors are shown to be high for some methods (i.e., alpha-Tomlinson and Meyerhof) and low for others (i.e., Schmertmann SPT). A number of prediction models are recommended for cohesive and cohesionless soils, as well as methods for determining soil parameters from in situ tests.
Abstract