AbstractStructural integrity practitioners can encounter countless, complex and often unknown uncertainties when dealing with myriad applications. The existence of such uncertainties can span the lifecycle of engineering assets: commissioning, design, prototyping, material testing, manufacturing, physical testing, in-service data, inspection data, fitness-for-service assessments and decommission. Traditionally, a deterministic mindset has prevailed amongst the engineering community at large. However, given the widening acceptance of data science and machine learning across a plethora of industries, trends towards accepting more probabilistic and data-driven solutions have emerged in engineering. Structural integrity is facing a similar change, with the nuclear sector now slowly recognising that there are strong needs that can be fulfilled by probabilistic paradigms. Presently the use of probabilistic approaches in the nuclear sector has been limited, bespoke and mainly summoned when traditional deterministic approaches fail to deliver business targets due to over conservatism. This work formulates a complete methodology based on the Monte-Carlo approach for conducting probabilistic calculations, focusing on applications considering the R5 Volume 2/3 procedure for high-temperature Advanced Gas Reactor (AGR)
components. The presented methodology is generic enough to be applicable to a host of structural integrity applications. A number of case-studies are presented in Chapter 3-6 which consider specific implementation issues including the probabilistic representation of input parameters, treatment of correlations, loading uncertainties, conducting post-assessment sensitivity analyses, the extrapolation of assessment location probabilities to component-level and, thereafter, to population-level estimates. With the presented methods having implications for structural integrity applications in general, one of the aims of this work is to bridge the gap between the knowledge of statistical and probabilistic methods on one side, and the general structural integrity community on the other end. Consequentially, this work is intended to promote further implementation and engagement, aiding further acceptance within a wide range of structural integrity fields, and ultimately encouraging the creation of a unified probabilistic framework for structural integrity.
|Date of Award||25 Jun 2019|
|Supervisor||Julian D Booker (Supervisor) & Christopher E Truman (Supervisor)|