This paper documents the predictive capability of rotating blade-resolved unsteady Reynolds averaged Navier-Stokes (URANS) and Improved Delayed Detached Eddy Simulation (IDDES) computations for tidal stream turbine performance and intermediate wake characteristics. Ansys/Fluent and OpenFOAM simulations are performed using mixed-cell, unstructured grids consisting of up to 11 million cells. The thrust, power and intermediate wake predictions compare reasonably well within 10% of the experimental data. For the wake predictions, OpenFOAM performs better than Ansys/Fluent, and IDDES better than URANS when the resolved turbulence is triggered. The primary limitation of the simulations is under prediction of the wake diffusion towards the turbine axis, which in return is related to the prediction of turbulence in the tip-vortex shear layer. The shear-layer involves anisotropic turbulent structures; thus, hybrid RANS/LES models, such as IDDES, are preferred over URANS. Unfortunately, IDDES fails to accurately predict the resolved turbulence in the near-wake region due to the modeled stress depletion issue.

Collaboration is described on assessment of computational fluid dynamics (CFD) predictions for surface combatant model 5415 at static drift β = 0 deg and 20 deg using recent tomographic particle image velocimetry (TPIV) experiments. Assessment includes N-version verification and validation to determine the confidence intervals for CFD solutions/codes, and vortex onset, progression, instability, and turbulent kinetic energy (TKE) budget analysis. The increase in β shows the following trends. Forces and moment increase quadratically/cubically, and become unsteady due to shear layer, Karman and flapping instabilities on the bow. Wave elevation becomes asymmetric; its amplitude increases, but the total wave elevation angle remains same. The vortex strength and TKE increase by about two orders of magnitude, and for large β, the primary vortices exhibit helical mode instability similar to those for delta wings. Forces and moment for both β and wave elevation for β = 0 deg are compared within 4% of the data, and are validated at 7% interval. Wave elevation for β = 20 deg, and vortex core location and velocities for both β are compared within 9% of the data, and are validated at 12% interval. The vortex strength and TKE predictions show large 70% errors and equally large scatter and are not validated. Thus, both errors and scatter need reduction. TKE budgets show transport of turbulence into the separation bubble similar to canonical cases, but pressure transport is dominant for ship flows. Improved CFD predictions require better grids and/or turbulence models. Investigations of solution-adaptive mesh refinement for better grid design and hybrid Reynolds-averaged Navier-Stokes/large eddy simulation models for improved turbulent flow predictions are highest priority.

A stand-alone machine learned turbulence model is developed and applied for the solution of steady and unsteady boundary layer equations, and issues and constraints associated with the model are investigated. The results demonstrate that an accurately trained machine learned model can provide grid convergent, smooth solutions, work in extrapolation mode, and converge to a correct solution from ill-posed flow conditions. The accuracy of the machine learned response surface depends on the choice of flow variables, and training approach to minimize the overlap in the datasets. For the former, grouping flow variables into a problem relevant parameter for input features is desirable. For the latter, incorporation of physics-based constraints during training is helpful. Data clustering is also identified to be a useful tool as it avoids skewness of the model towards a dominant flow feature.

This study proposes and validates a novel machine-learned (ML) augmented linear Reynolds averaged Navier Stokes (RANS) model, and the applicability of model assessed in both interpolation and extrapolation modes for periodic hill (Hill) test case, which involves complex flow regimes, such as attached boundary layer, shear-layer, and separation and reattachment. For this purpose, the ML model is trained using direct numerical simulation (DNS)/LES datasets for nine different cases with different flow separation and attachment regimes, and by including various percentages of the Hill DNS dataset during the training, ranging from no data (extrapolation mode) to all data (interpolation mode). The predictive capability of the ML model is then assessed using a priori and a posteriori tests. Tests reveal that the ML model's predictability improves significantly as the Hill dataset is partially added during training, e.g., with the addition of only 5% of the hill data increases correlation with DNS to 80%. Such models also provide better turbulent kinetic energy (TKE) and shear stress predictions than RANS in a posteriori tests. Overall, the ML model for TKE production is identified to be a reliable approach to enhance the predictive capability of RANS models. The study also performs (1) parametric investigation to evaluate the effect of training and neural network hyperparameters, and data scaling and clustering on the ML model accuracy to provide best practice guidelines for ML training; (2) feature importance analysis using SHapley Additive exPlanations (SHAP) function to evaluate the potential of such analysis in understanding turbulent flow physics; and (3) a priori tests to provide guidelines to determine the applicability of the ML model for a case for which reference DNS/LES datasets are not available.

This study evaluates the capability of Navier–Stokes solvers in predicting forward and backward plunging breaking, including assessment of the effect of grid resolution, turbulence model, and VoF, CLSVoF interface models on predictions. For this purpose, 2D simulations are performed for four test cases: dam break, solitary wave run up on a slope, flow over a submerged bump, and solitary wave over a submerged rectangular obstacle. Plunging wave breaking involves high wave crest, plunger formation, and splash up, followed by second plunger, and chaotic water motions. Coarser grids reasonably predict the wave breaking features, but finer grids are required for accurate prediction of the splash up events. However, instabilities are triggered at the air–water interface (primarily for the air flow) on very fine grids, which induces surface peel-off or kinks and roll-up of the plunger tips. Reynolds averaged Navier–Stokes (RANS) turbulence models result in high eddy-viscosity in the air–water region which decays the fluid momentum and adversely affects the predictions. Both VoF and CLSVoF methods predict the large-scale plunging breaking characteristics well; however, they vary in the prediction of the finer details. The CLSVoF solver predicts the splash-up event and secondary plunger better than the VoF solver; however, the latter predicts the plunger shape better than the former for the solitary wave run-up on a slope case.