Evaluating the Site Variability Using Bayesian Analysis

Document Type

Conference Proceeding

Publication Date

1-1-2023

Abstract

This study proposed a robust two-level Bayesian framework to incorporate the site specific variability for estimating the pile capacity from cone penetration test (CPT) based on LCPC method for LRFD design of pile foundations. The first level of framework considered a prior data for the bias factor and updated it (posterior level 1) using regional test data. In the second level, the posterior of the bias was further updated using the site specific data (posterior level 2). In order to update the Bayesian model in this study, the mean bias (λ) of measured/predicted pile capacities and standard deviation of bias (σ) were obtained from the pile load test database of 33 sites in Louisiana. The Houma Bridges site was selected to implement the site specific variability into LRFD design of piles. The results show that the local values of λ and σ for the Houma Bridge site are (1.09, 0.315). However, the updated λ and σ (posterior level 2) using Bayesian analysis are (1.04, 0.32). The updated posterior parameters for the Houma Bridges site lie between the prior level 2 parameters and the likelihood level 2 parameters, taking into consideration the site specific variability. The posterior of level 2 parameters (λ, σ) can be used to calibrate the resistance factor (φ) for LRFD design of piles based on LCPC design method for the specific site.

Publication Source (Journal or Book title)

Geotechnical Special Publication

First Page

205

Last Page

211

This document is currently not available here.

Share

COinS