Performance Reliability of Offshore Piles

Abstract

ABSTRACT This paper describes an assessment of the bias, error and reliability associated with capacities and performances of axially and laterally loaded piles, designed according to API Recommended Guidelines [1,2]. We first identify the major sources of uncertainties affecting axial and lateral performances of piles in clay and sand, which are then assessed using available field and laboratory test data. These individual components of uncertainties are then integrated through a simple probability model for further reliability evaluation. Results from this research will help to direct future research efforts for reducing major uncertainties affecting pile performances, and to assist designers to better assess differences between various editions of API-RP2A Recommended Guidelines. In the long term, it will facilitate the development of design procedures which maintain a consistent level of safety between different site conditions, alternative pile configurations and structural and foundation subsystems. Only single piles are considered in this paper. Reliability of pile system is currently under investigation. INTRODUCTION Traditionally, the axial bearing capacity of a pile is calculated using a given pile capacity prediction method and a chosen set of soil parameters. In reality, the capacity of an offshore pile subject to storm loadings could differ significantly from the calculated capacity due to various factors and sources of uncertainties. First each pile capacity prediction method rests on some simplifying assumptions. Even if the institute soil properties at the pile location are perfectly known, some scatter is still expected between the measured and predicted capacities. This model error would vary among different prediction methods. For example, to evaluate the model error associated with a given pile capacity prediction method, untrained shear strengths are needed in each site to predict the capacity in clay. However, such strength values are based on a large range of sampling and testing techniques. In order to maintain consistency, a commonly available sampling/test method is generally adopted as the standard strength. Hence, the statistics of the model error assessed according to the standard strength values would only be applicable to a design situation where the same standard sampling/test method is used for its soil strength determination. If any other sampling/test method were used, a correction factor would be required to account for the discrepancy in soil strength values expected between that and the standard method. Second, a pile subject to storm loading does not necessarily have the same capacity as that measured at a conventional load test. For instance, load tests are generally performed within 50 days of pile installation; whereas the maximum load applied to a pile during a structure's lifetime may occur years after installation. For most normally consolidated clay where reconsolidation occurs around a pile, the capacity measured during load tests could underestimate the actual pile capacity. Third, the rate of load application during load test is generally much slower than the loading rate from waves. Since soil strength generally increases with loading rate, the capacity measured during load tests would likely underestimate the actual pile capacity. Correction factors are thus needed to account for each of these and other biases.