This work concerns the quantification of numerical accuracy for focused wave interactions with floating structures, using results obtained via a commonly used computational fluid dynamics (CFD) and linear wave superposition approach. It represents an individual contribution to the CCP-WSI Blind Test Series 3, where numerical predictions for a structure's motion are submitted for comparison with physical data, without prior access to this data. An estimation of accuracy based on the reproduction of physical empty tank data, and previous experience, is compared with the observed error in the structure's motion. The results imply that the error in peak values of heave and empty tank surface elevation are comparable, but the peak surge and pitch are substantially larger. This is likely due to a combination of numerical modelling errors, although numerical uncertainty must also be reduced in order to fully assess the problem.
Two key issues that are limiting the routine use of computational fluid dynamics (CFD) are the uncertainty in its accuracy and the time required to obtain numerical results. The time taken to run a simulation is notoriously long, but this can be decreased through use of a larger computational resource. However, an often overlooked factor is the man-hours required to set up a case through processes such as mesh design; this has the potential to be considerably more time-consuming than the simulation time. For industry to benefit from the strengths of CFD models, the efficiency of the setup process needs to be increased, and this could be achieved through increased confidence in prediction by parametrically understanding numerical accuracy and providing standardised, “best-practice” implementations. An ever expanding use of CFD simulations for wave-structure interaction (WSI) applications (Windt, Davidson, and Ringwood, 2018; Palm et al., 2016; Devolder et al., 2018) has led to preliminary studies seeking to set the foundations for standardisation, such as the expansion of mesh convergence schemes to estimate uncertainty (Eskilsson et al., 2017; Wang et al., 2018), assessment of available wave generation methods (Windt et al., 2019a), the influence of mesh deformation scheme (Windt, Davidson, Akram, and Ringwood, 2018), and turbulence modelling under breaking waves (Brown et al., 2016). However, in general, there are very few established guidelines for design of WSI CFD simulations. Bearing in mind the enormous number of techniques and settings available to a user, it is therefore neither uncommon nor unexpected to see a wide range of solutions for a single problem where the desired solution is not known a priori, even when applying the same base CFD code (Ransley et al., 2019, 2020).