Session: 02-04-02 Extreme Loads and Response 2

Submission Number: 176327

Nonlinear Time-Reversal Method for Efficient Probabilistic Design Wave Identification 

This paper presents an efficient methodology for the probabilistic identification of critical design irregular waves, a crucial preparatory step in multi-fidelity workflows that culminate in analysis using high-fidelity solvers such as computational fluid dynamics. This study introduces a simplified approach that combines the Higher-Order Spectral Method (HOSM) with a nonlinear time-reversal (TR) technique. By initiating the analysis with a linear target response and employing a single backward simulation, the proposed method determines both the design wave and its probability without the iterative optimization required by the First-Order Reliability Method (FORM).

Systematic validation against Monte Carlo Simulations (MCS) and the direct combination of HOSM and FORM demonstrates close agreement in the probabilistic predictions obtained with the TR method. For crest-dominated responses in highly nonlinear seas, the wave profiles derived from TR-RCW and FORM-RCW exhibit only minor differences (Improved Surface Similarity Parameter, ISSP ≈ 0.1). This slight divergence arises from their fundamentally different principles: while FORM identifies a wave profile in which free and bound waves are in physical equilibrium, the TR method imposes a linear target profile at the final time—a profile that does not inherently incorporate equilibrium bound waves—and then calculates its nonlinear time propagation.

Furthermore, the non-iterative nature of the TR method proves more robust, successfully delivering a solution in a highly nonlinear case where the iterative FORM analysis failed to converge. Overall, the proposed TR approach yields a computationally efficient alternative for the probabilistic component of multi-fidelity analyses against realistic, nonlinear extreme seas.

Presenting Author: Wataru Fujimoto Nippon Kaiji Kyokai

Presenting Author Biography: Wataru Fujimoto is a marine data scientist at the ClassNK Research Institute, where he focuses on the advanced utilization of metocean information for maritime safety. He received his Ph.D. in Ocean Engineering from the University of Tokyo, with a dissertation on the nonlinear dynamics of freak waves and data assimilation techniques. His doctoral work was recognized with the Dean’s Scholar Award.
Prior to joining ClassNK in 2020, he worked at MS&AD InterRisk Research Institute as a risk analyst specializing in natural disaster modeling, particularly flood risk assessment and post-disaster damage surveys. His current research interests include probabilistic wave modeling, nonlinear wave dynamics, and the integration of observational data into simulation frameworks.
Dr. Fujimoto has published in peer-reviewed journals and presented at international conferences on ocean engineering and ship safety. He is proficient in numerical modeling using tools such as MATLAB and Fortran, and actively contributes to the development of multi-fidelity workflows for ship response prediction.

Authors:

Wataru Fujimoto Nippon Kaiji Kyokai
Tomoki Takami Graduate School of Maritime Sciences, Kobe University