Abstract This study applies a suite of novel methodologies for simulating extreme waves and their associated wave loads. This improves significantly the efficiency of generating appropriate data required for reliability analyses, that are typically required as part of the design and/or reassessment of marine structures. This is demonstrated on the derivation of both the statistical properties of extreme wave crests and the applied wave-in-deck (WID) loading. The proposed methodologies are developed on the basis of extensive physical model testing conducted at the Hydrodynamics Laboratory at Imperial College London. The experiments relate to random wave design conditions that are typically employed in intermediate and deep water depths. Addressing this issue within the framework of stochastic (random) waves typically involves prohibitively long (experimental or numerical) simulation times. The proposed methodology addresses this challenge by enabling the theoretical pre-selection of extreme wave crests that generate extreme WID loads. This approach reduces physical modelling time from days to hours. In adopting this approach, an extensive database of waves and loads is generated. A data-driven approach is then adopted to define the wave variables that best predict the applied WID loads. Using these variables, alongside related physical insights, new data-driven models are derived to predict WID loading. These models are shown to be accurate in defining WID loading arising in random seas. The accuracy of the proposed methodology is established for both the present datasets and independent experiments. As a whole, the proposed framework is shown to efficiently and accurately provide extreme crest heights and wave-in-deck loads. These predictions are accompanied by robust estimation of the associated uncertainties which are crucial for reliability analyses.
Huo et al. (Sun,) studied this question.