Minimizing hydrocarbon release from offshore piping by performing probabilistic fatigue life assessment

Abstract

Abstract Topside piping is the major source of hydrocarbon release (HCR) on offshore oil and gas (OOG) platforms in the North Sea region. Since 21% of piping failures are caused by vibration induced fatigue (VIF), an accurate remnant fatigue life (RFL) assessment has the potential to minimize the chances of HCR from an operating piping system. BS-7910 gives two possible approaches for performing a RFL assessment: the S–N curve approach and the fracture mechanics (FM) approach. Since there are large number of uncertainties (such as uncertainty due to the crack growth model, future loading, material and geometric properties, etc.) involved in the RFL calculation process, therefore it is vital to consider the aforementioned sources of uncertainty in order to arrive at an accurate RFL estimate. Nevertheless, BS-7910 provides limited guidance on how to handle uncertainty in RFL assessment. The most common way of dealing with the aforementioned uncertainty is to evaluate RFL probabilistically. This manuscript thus explains the procedure of the probabilistic RFL assessment of offshore topside piping, with an emphasis on uncertainty quantification, propagation and management. Uncertainty quantification handles the identification and characterization of the different sources of uncertainty that may influence the future behavior of the piping component and, in turn, the RFL estimate. Thereafter, uncertainty propagation employs the formerly quantified uncertainties and utilizes the aforementioned information to estimate the RFL. Finally, uncertainty management deals with performing sensitivity analysis to find the individual contributors to uncertainty in the estimated RFL. A numerical case study illustrating the deterministic and probabilistic RFL assessment of topside piping is presented. Afterwards, probabilistically predicted RFL is used to demonstrate the calculation of an inspection interval. Finally, the implications of probabilistically estimated RFL on HCR from process piping is discussed.