Performance-Based Reliability Design for Deep Foundations Using Monte Carlo Statistical Methods-Article

Article

Pdf Summary

The paper presents a performance-based reliability design (PBRD) method for drilled shafts under lateral and axial loading. Unlike the deterministic serviceability checks in current AASHTO LRFD practice, the method defines failure as foundation displacement exceeding a specified allowable limit and estimates the probability of failure using Monte Carlo simulation.<br /><br />Soil properties are modeled as spatially correlated random fields, while loads and computational-model bias can also be treated as uncertain. The method uses the p-y model for lateral response and load-transfer curves for axial response. Foundation dimensions are adjusted until the calculated failure probability meets a target reliability level. Three examples examine lateral loading, uplift, and compression.<br /><br />The results show that soil variability—especially the coefficient of variation and correlation length of undrained shear strength—can substantially affect failure probability and design dimensions. Variability in soil strain parameters has a smaller effect, while unit-weight variability has little influence in the examples. Accounting for uncertain loads can require larger shafts, and greater epistemic uncertainty in the design model increases estimated failure probability.<br /><br />The authors conclude that PBRD offers a practical way to incorporate uncertainty into serviceability design and can support safer, more economical foundations. Reliable application depends on accurately characterizing soil spatial correlation and model uncertainty. Construction-related uncertainty, which the proposed method does not address, must also be considered.

Keywords

performance-based reliability design

drilled shafts

lateral loading

axial loading

probability of failure

Monte Carlo simulation

spatially correlated random fields

undrained shear strength

serviceability design

AASHTO LRFD