DOMENIC JOSEPH DI FRANCESCO

Attempts to formalize inspection and monitoring strategies in industry have struggled to combine evidence from multiple sources (including subject matter expertise) in a mathematically coherent way. The perceived requirement for large amounts of data are often cited as the reason that quantitative risk-based inspection is incompatible with the sparse and imperfect information that is typically available to structural integrity engineers. Current industrial guidance is also limited in its methods of distinguishing quality of inspections, as this is typically based on simplified (qualitative) heuristics. In this paper, Bayesian multi-level (partial pooling) models are proposed as a flexible and transparent method of combining imperfect and incomplete information, to support decision-making regarding the integrity management of in-service structures. This work builds on the established theoretical framework for computing the expected value of information, by allowing for partial pooling between inspection measurements (or groups of measurements). This method is demonstrated for a simulated example of a structure with active corrosion in multiple locations, which acknowledges that the data will be associated with some precision, bias, and reliability. Quantifying the extent to which an inspection of one location can reduce uncertainty in damage models at remote locations has been shown to influence many aspects of the expected value of an inspection. These results are considered in the context of the current challenges in risk based structural integrity management.

Engineers perform fatigue assessments to support structural integrity management. Given that the purpose of these calculations is linked to problems of decision making under various sources of uncertainty, probabilistic methods are often more useful than deterministic alternatives. Guidance on the direct probabilistic application of procedures in existing industrial standards is currently limited and dependencies between marginal probabilistic models are generally not considered, despite their potential significance being acknowledged. This paper proposes the use of Bayesian data analysis as a flexible and intuitive approach to coherently and consistently account for uncertainty and dependency in fatigue crack growth rate models. Various Bayesian models are established and presented, based on the same data as the existing models in BS 7910 (a widely used industrial standard). The models are compared in terms of their out of sample predictive accuracy, using methods with a basis in information theory and cross-validation. The Bayesian models exhibit an improved performance, with the most accurate predictions resulting from multi-level (hierarchical) models, which account for variation between constituent test datasets and partially pool information.