Interpretation of Site-Specific Reliability Index for Piles Using Multiple Bidirectional Tests

Abstract The Osterberg cell test is an effective tool to evaluate the bearing behavior for piled foundations. Because of the significant intrinsic variabilities in material properties of piles, soil-pile intermediate, and soils, the responses of multiple piles under loading are uncertain. These uncertainties should be considered to evaluate the bearing characteristics of piles, in this work, a methodology for estimating the geometric reliability index of piles using the Osterberg cell testing is proposed. A dataset of load−displacement curves including 26 Osterberg cell-tested drilled shafts at the Los Angeles Memorial Coliseum site is used to illustrate the proposed method. Regression analyses with a two-parametric power law curve-fitting model are firstly used to construct the best-fit model for each load-displacement curve along an upward or downward direction individually. As a result, the regression parameters for this site are combined into a random vector. Statistical analyses are then performed on these regression variables, and the bivariate normal and non-normal models are developed. These bivariate models are used to predict the reliability index of these drilled shafts via the geometric reliability method. Such an interpretation of reliability index by analyzing the scatter in load test data can be helpful to design on pile foundations.