Bayesian Updating of Fatigue Crack Growth Parameters for Failure Prognosis of Miter Gates

Navigable waterways play a vital role in the efficient transportation of millions of tons of cargo annually. Inland traffic must pass through a lock, which consists of miter gates. Failures and closures of these gates can significantly disrupt waterborne commerce. Miter gates often experience fatigue cracking due to their loading and welded connections. Repairing every crack can lead to excessive miter gate downtime and serious economic impacts. However, if the rate of crack growth is shown to be sufficiently slow, e.g., using Paris' law, immediate repairs may be deemed unnecessary, and this downtime can be avoided. Paris' law is often obtained from laboratory testing with detailed crack measurements of specimens with relatively simple geometry. However, Paris' law parameters for an in situ structure will likely deviate from those predicted from physical testing due to variations in loading and materials and a far more complicated geometry. To improve Paris' law parameter prediction, this research proposes a framework that utilizes (1) convenient vision-based tracking of crack evolution both in the laboratory and the field and (2) numerical model estimation of stress intensity factors (SIFs). This study's methodology provides an efficient tool for Paris' law parameter prediction that can be updated as more data become available through vision-based monitoring and provide actionable information about the criticality of existing cracks.