Uncertainty and Bias in Evaluation of LRFD Ultimate Limit State for Axially Loaded Driven Piles-Article

Article

Pdf Summary

The paper examines how uncertainty and bias affect LRFD resistance factors for the ultimate axial capacity of driven piles. It argues that bias—systematic error caused by design methods omitting influential factors—can be more important than statistical uncertainty, especially when applying correlations beyond the conditions represented in their test databases.<br /><br />Typical databases contain relatively small piles, while high-capacity offshore piles are much larger and longer and operate under different stress and installation conditions. Applying methods calibrated on smaller piles therefore involves substantial extrapolation. The paper reviews reliability-based resistance-factor calibration and recommends an approach for combining dead- and live-load uncertainty that better reflects their relative contributions.<br /><br />It identifies important factors that should be considered when extrapolating pile-capacity methods, including soil strength and compressibility, effective stress, pile dimensions and end condition, interface properties, friction fatigue, and time between installation and loading. Although extrapolation uncertainty generally increases as design conditions depart from database conditions, the paper finds that extrapolation bias can have the greater effect on reliability.<br /><br />Examples illustrate potential bias in shaft-friction estimates. In sand, predictions from several design methods diverge for piles with slenderness ratios beyond those typical of calibration databases; some simplified correlations may become unconservative. In clay, predicted shaft resistance can differ by as much as a factor of 2.5 depending on soil state and method, with sensitivity, plasticity, and consolidation timing among the possible explanations. Pile capacity can also increase over time through consolidation and ageing, but these effects vary with soil and pile conditions.<br /><br />The paper concludes that reliable resistance factors require uncertainty and bias to be assessed for the specific load combination and design situation. Design methods should account for the mechanisms governing pile resistance and clearly represent the conditions to which their correlations apply.

Keywords

LRFD

resistance factors

driven piles

axial capacity

uncertainty

bias

reliability-based calibration

pile capacity extrapolation

shaft friction

offshore piles