A benchmark study of linear seakeeping predictions and uncertainty in global wave loads

This paper presents a benchmark study of linear wave-induced motions and global loads, organised by the MARSTRUCT Virtual Institute with the objective of quantifying the hydrodynamic modelling uncertainty and its consequences for extreme load prediction across different hull forms. This is a follow-up to a benchmark study Parunov et al. (2022) that addressed this problem for a container ship tested many years ago and is now extended to several recent hull shapes to determine how case-dependent the earlier results were. Fifteen institutions participated in the benchmark, submitting twenty-four sets of calculations obtained using twenty-two numerical codes, including strip theory, three-dimensional boundary element methods and CFD. Three modern ship types, namely a chemical tanker, an LNG carrier and a cruise ship, are analysed using experimental data from the EU Extreme Seas project. The study focuses on head-sea conditions at zero or low forward speed and examines heave, pitch and the vertical wave bending moment at midship. Uncertainty in transfer functions is quantified using a frequency-independent model error, and its impact on ultimate limit-state verification is assessed through extreme-value analysis of the vertical wave-induced bending moment. The benchmark results provide quantitative support for commonly adopted model uncertainty assumptions in structural reliability analysis. For strip theory, the observed bias and dispersion of predicted bending moments are consistent with a mean correction factor of approximately 0.9 and a coefficient of variation of about 0.15. For three-dimensional panel methods, the results support an essentially unbiased model with a coefficient of variation close to 0.1.