Bayesian Estimation And Reliability Analysis Of Vertical Bearing Capacity Of Single Piles

High-rise buildings and major bridges are increasingly constructed with rapid development of economy. Accordingly, reliability of structures plays a significant role in engineering practice. Pile foundation is widely useed as the foundation of the structures due to its high bear capacity, small settlement and convenient contruction. Therefore, reliability analysis is highly important in engineering practice. However, there are two problems in existing researches, they are:(1) the data from similar projects are discarded, they are not adopted to determine the bearing capacity of pile; (2) the error of reliability index may be relatively large due to the great discreteness of data from the static load tests, which may cause relatively significant loss to the project. Considering the above two problems, following studies have been conducted in this dissertation.Due to high cost and long construction period of the project, the data from static load tests of piles are often scarce and their dispersion degree is very large, which brings more difficulties to reliasbility analysis of bearing capacity. In this dissertation, the mean and variance of population sample are calculated based on the concept of range extremity difference and the results from static load tests. Then the calculation formula of reliability is presented on the condition that the vertical bearing capacity of pile follows log-normal distribution. The results from case studies show that the calculation accuracy is satisfactory and it can provide good references to designers to determine the design parameters according to testing piles.The design value for bearing capacity of pile is usually determined from the static load tests in pile foundation engineering. However, the error of the design value obtained from static load tests is large due to limited number of testing piles and great discreteness arising from many adverse factors including possible delays and high cost. The value of a random variable can be more accurately evaluated using Bayesian statistic because the general information, sampling information and prior information are systematically associated in Bayesian framework. Based on Range theory and Bayesian statistics principle, the prior distribution of the mean of bearing capacity of pile is presented to be normal distribution, and posterior distribution also follows a normal distribution according to Bayesian statistics. The calculation formula of design value is obtained using the abundant information of soil exploration and the results from the static load tests in the Bayesian framework,The achievements in this dissertation not only can meet the requirements of safety and economy for pile foundation, but also provide a theoretical basis to determine the design value and references for designers.