Assessing component and system reliability of offshore structures

The demand for higher reliability and lower costs is becoming the major requirement for offshore structures. To meet these competing demands, design and optimization methods capable of considering all relevant aspects of the system lifecycle are required. Concerning design procedures, despite the Load and Resistance Factor Design methods (LRFD) have proved to be more rational than the Working Stress Design (WSD), two main limitations still affect present LRFD codes. Firstly these procedures are generally based on a semi-probabilistic approach, in which a pre-established target reliability level is used to check each load combination. From a practical perspective the application of partial safety factors allows straightforward design format, which turn in a simplified engineering interfacing with reliability methods. Such parameters are not always meaningful from a reliability-costs optimization viewpoint. Secondly, LRFD methods are mainly component-based with the underlying hypothesis that the structural system will be safe as long as all its component satisfy the codes. Many efforts have been made to deal with systems reliability i.e., analyzing the load redistribution after member failures, identifying multiple failure paths and relating the environmental load factors to reserve strength ratio (RSR). In this paper, dominant collapse modes are identified through modern research methods and strategies for quality assurance based on critical members are provided in order to prevent failures due to accidents, fabrication mistakes and human errors that are not reflected in the present code checks. Finally, the tubular joint failure comparison under API RP2A WSD code and ISO 19902 (LRFD-type code) is carried out. As to the latter, both component-based and system-based approaches are considered . Differences in the results are discussed in the reliability-costs optimization context, emphasizing the impact of the uncertainty-based multidisciplinary design optimization on the failure costs.