Quantification of extreme event statistics in ship design

Abstract

Increased operational demands on Navy vessels extend time at sea and service life, making the accurate prediction of catastrophic failures increasingly challenging. The high value of these capital assets puts great pressure on designers and decision-makers as they work towards preventing such failures while balancing both engineering and material cost. The current method for the quantification of extreme events is direct Monte Carlo simulation supplemented by complex statistical models. When such models are not sufficiently bound by physics-based simulation, the noise of statistical uncertainty quickly overpowers the response predictions for rare events. This thesis builds on previous work at the MIT Stochastic Analysis and Non-linear Dynamics (SAND) lab for the quantification of extreme events using wave groups. By separating the event probability from the physics models, we are able to capture rare events in ship motion and loading conditions for a modest computational cost. Improvements to the wave groups methodology ensured the slope and amplitude of the incident waves reflected the waves encountered in a given wave spectrum. The remaining discussion explores the value of a near-real-time risk analysis tools in reference to ship design and ship operations, with unique application to Navy and commercial vessels.