December 2, 2024-Bryan Tomer, Carderock

December 2,  2024
4:00 p.m.
Room:  Torgersen Hall 2150
Speaker: Bryan Tomer, Carderock
Faculty Host: Dr. Pat Artis

"Predicting Extreme Motions and Loads in Early Stage Ship Design"

Abstract:  The process of designing a safe vessel does not just involve ensuring that it is stable under both intact and damaged conditions, whether that be in calm water or in waves. That process begins with the structural design of the vessel which must ensure that over the lifetime of the vessel, the structure is not over stressed even under the worst weather conditions that it could be expected to experience. A structural failure can lead to a vessel experiencing flooding resulting in a damaged stability issue or in the case of a total structural failure the catastrophic loss of the vessel. Adequate structural design depends on an accurate characterization of the maximum lifetime primary and secondary loads that the vessel might be expected to experience, the result of a superposition of multiple nonlinear processes.

Today, those loads are determined using empirical methods found in classification society rules, or from loads measured during model tests or predicted using a computational tool. As physical model tests and computational modeling are limited to relatively short periods of time relative to a vessel’s lifetime, the measured or predicted loads must be statistically extrapolated to determine their maximum values. This extrapolation is often performed using a Weibull distribution fit, and when an acceptable fit to the Weibull distribution cannot be obtained, the measured/predicted value is often just doubled. In the absence of a reliable statistical extrapolation method, a method of determining maximum loads is required. 

Determining an accurate approximation to the exact single significant amplitude (SSA) motions for a vessel in an extreme seaway requires hundreds of thousands of hours of full-scale motion data. This is obviously impractical experimentally, as this amount of time exceeds the operational life time of a vessel and there is no way that “Mother Nature” could produce a statistically stationary seaway for even a tiny fraction of that amount of time. This leaves simulations as the only means by which such an extended period of time’s motion data can be produced in a stationary seaway. However, the fastest accurate motion- prediction tools run roughly one to ten times slower (CPU time/real time) than real time on high performance computer systems and the most precise prediction tools can run 10,000 or more times slower than real time. Therefore, it is impractical to use a high-fidelity motion prediction model to produce data for use in statistical validation. The solution of this dilemma is a reduced-order model that captures the most critical aspects of the physics of a ship operating in a seaway—particularly the geometric nonlinearities.