Large-eddy simulations of interaction between surface waves and a tidal turbine wake in a turbulent channel

Abstract

Tidal stream turbines are now being developed for array deployments, largely at sites with relatively shallow water depths on either bed-supported, or floating support structures. Proximity to the free-surface presents design challenges with increased exposure to wave-induced kinematics leading to potential for increased peak- and fatigue-loads. Free surface proximity can also alter wake recovery rates which can influence the siting, and operation, of further turbines. To-date the impact of waves on turbine loading and on wake recovery has receivedlimited attention, generally for specific combinations of conditions reproducible in experimen tal facilities [1-2] or numerical models [3]. Improved understanding of how waves affect both turbine loading, and wake dynamics is necessary to inform the development of appropriate load prediction and mitigation methods and to further parameterise wake recovery to inform array siting.
 This work presents CFD analysis of the loading and wake of a three-bladed horizontal axis tidal turbine within a wide, shallow channel with a turbulent inflow developed using a synthetic eddy method and free surface waves modelled using a Level-Set Method (LSM). The turbine and wave conditions considered are based on prior experimental studies [1] within a channel depth h = 0.45 m and mean flow U = 0.47 m/s. The simulated channel is width 3.67h and length 88h and long-crested regular waves are imposed, with the focus of this study on waves with non-dimensional wavenumber kh = 2.8, comparable with experimental conditions. Simulations are accomplished using the in-house large-eddy simulation (LES) code DOFAS(Digital Offshore Farms Simulator) [4], which adopts an Actuator Line Method (ALM) to resolve the turbine blades. LSM is used to model the air-water interface in an accurate manner [5] with waves can be generated in DOFAS using linear or second-order Stokes theories with an absorption layer placed at the outlet of the domain to avoid wave reflections. At the inlet a mean logarithmic velocity profile is imposed over which artificial turbulence is added. Simulations were run for 400 s of physical time on 8,000 cores using ARCHER2.
 The waves studied are shown to alter the rate of wake recovery, increasing the rate in the near-wake of the turbine and slightly reducing the rate beyond eight diameters downstream. Analysis of turbulence characteristics indicates a significant variation across the water column due to wave action. Propagation over the turbine wake also introduces directionality to the wave field, which is associated with the change of wave-speed over the wake region. Modal decomposition analysis of the velocity fluctuations with Proper Orthogonal Decomposition(POD) reveals that the wake dynamics behind the turbine change due to the waves.