Statistics of Model Factors and Consideration in Reliability-Based Design of Axially Loaded Helical Piles

Abstract

Geotechnical design codes have been migrating toward reliability-based design (RBD) concepts. ISO 2394 identified the characterization of model uncertainty as one of the critical elements in the geotechnical RBD process. This paper collects a large number of field axial load tests on helical piles for model uncertainty assessment as required in developing RBD. At the ultimate limit state (ULS), the model uncertainty is represented by a model factor, which is defined as a ratio of measured resistance over calculated resistance. The measured resistance is interpreted from load test data using the methods recommended in the helical pile industry, while the existing empirical or semiempirical methods are applied for resistance calculations. A hyperbolic model with two parameters is adopted to fit the measured load-settlement data. The uncertainties within the load-settlement curves are captured by a bivariate random vector containing the hyperbolic parameters as its components. Statistical properties of the model factors such as means, coefficients of variation, and probability distributions are determined from the database. Moreover, several copulas are evaluated to establish the correlation structure within the hyperbolic parameters. Finally, the ULS model statistics are incorporated into the calibration of the resistance factor in the load and resistance factor design (LRFD) of axially loaded helical piles.