Reliability-based design of large-diameter bored piles in Hong Kong―Part II: Bayesian updating using load tests

摘要

Several design methods are widely used in Hong Kong for estimating the shaft resistance of large-diameter bored piles in saprolites and the toe and shaft resistance in rock-socket piles. The methods being considered include "rule of thumb" methods and correlations with the Standard Penetration Test (SPT) blow count or the Unconfined Compressive Strength (UCS) of rocks. In the accompanying paper, it is found that the "rule-of-thumb" methods for the shaft resistance in saprolites and the toe resistance in rock are associated with values of reliability index of 2.59 and 2.85, respectively, which are in the range of the recommended target reliability index of 2.5-3.0. However, the reliability associated with the correlations for determining shaft resistance in saprolites and shaft resistance in rock socket are very low if no proof tests are conducted. This paper aims to update the values of the mean and standard deviation of resistance components and the total pile capacity using Bayes' method and to investigate the impact of load tests on the reliability of bored piles. For a particular new site where proof load tests are conducted, the reliability of piles in that site can be updated using the Bayesian approach. If favorable test results are obtained, the p obability of exceedance will be greatly decreased and the factor of safety (SF) could be reduced. On the contrary, if a design method has a sufficiently large bias, the reliability of the pile could be more accurately predicted by the updating and might not always be improved. With load tests, the "rule-of-thumb" methods for estimati !g the shaft resistance in saprolites and the toe resistance in rock are found to be reliable or somewhat conservative. Based on the updating analysis, suitable factors of safety for design of bored piles in Hong Kong are proposed and a SF value of 2.0 with proof tests is found to be on the un-conservative side.