Probabilistic analysis and resistance factor calibration for deep foundation design using Monte Carlo simulation

Thuy Vu,Erik Loehr,Douglas Smith

doi:10.1016/j.heliyon.2018.e00727

Thuy Vu, Erik Loehr + Show 1 more

Open Access

https://doi.org/10.1016/j.heliyon.2018.e00727

Copy DOI

Abstract

The method of incorporating the sources of parameter uncertainty is crucial when conducting probabilistic analysis for service limit state (SLS) design of a deep foundation. This paper describes the method of using Monte Carlo simulation for probabilistic analyses and for calibration of resistance factors of drilled shafts at SLS. The paper presents discussions on the finding of an impossible case, where the different combinations of load, variability of soil strength and target probability of failure made it impossible to calibrate the SLS resistance factors. Resistance factors for drilled shafts in shale are introduced, and were found to be responsive to load levels. The higher load level, the lower the resistance factor. These findings help smooth the transition from allowable stress design to load and resistance factor design for geotechnical engineers.

Highlights

Geotechnical engineers have been working to transition from allowable stress design, which has been used for many years, to load and resistance factor design (LRFD)
The ultimate limit state relates to geotechnical strength failures; for example when the applied load is equal to the resistance
This paper describes a proposed procedure that allows service limit state (SLS) design to be performed to achieve some desired target reliability without requiring case-specific calibration or more rigorous reliability-based design

Summary

Introduction

Geotechnical engineers have been working to transition from allowable stress design (or working stress design), which has been used for many years, to load and resistance factor design (LRFD). The design reaches a limit state when a component of the structure does not fulfill its prescribed function. The ultimate limit state relates to geotechnical strength failures; for example when the applied load is equal to the resistance. Several probabilistic approaches are used in reliability-based design and in the LRFD resistance factor calibration. FOSM is based on a Taylor series expansion of a performance function (Baecher and Christian, 2003). The approach is based on assumptions that all input parameters are normally distributed, and that the limit state is a normally distributed variable. The Monte Carlo simulation method utilizes random number simulation to extrapolate probability density function values (Baecher and Christian, 2003; Harr, 1987).

Methods

Results

Conclusion