Development of an interface to introduce stationary LES data to the URANS solver THETA for HAWT performance prediction

Abstract

The impact of different sheared velocity profiles on the performance prediction of a horizontal axis wind turbine in the atmospheric boundary layer is investigated. Firstly, the wall roughness in the analytical logarithmic description of the atmospheric boundary layer is varied to obtain different velocity profiles. Subsequently, it is proposed to replace the analytical logarithmic description of the atmospheric boundary layer by the time-averaged velocity data of a precursor large eddy simulation (LES) and to reconstruct the turbulence of the velocity fluctuations. The LES data are introduced as inflow condition through a LES-RANS interface in a one-way coupling approach. Three different methods to reconstruct URANS turbulence values out of the velocity fluctuations are investigated. It is shown that the reconstruction method has an impact on the development of the velocity profile, turbulent kinetic energy, and the turbulent dissipation during the transport through the URANS domain. The different inflow data, which the horizontal axis wind turbine experiences, are responsible for changes in the overall rotor thrust (up to 2.7%) and rotor torque (up to 2.4%). Conversely, the induction factors and effective angles of attack hardly change and can well be compared with a blade element momentum method. Finally, the results of both approaches to prescribe the atmospheric boundary layer are compared. The thrust and power coefficients, and wake recovery are close to each other. Simulations are carried out on an industrial 900 kW wind turbine with the incompressible URANS solver THETA. KEYWORDSatmospheric boundary layer, horizontal axis wind turbine, incompressible URANS solver, LES-RANS interface, logarithmic velocity profile, turbulence model boundary condition, turbulence reconstruction INTRODUCTIONComputational fluid dynamic (CFD) tools that are used to analyse the aerodynamic behaviour of horizontal axis wind turbines (HAWT) can be classified in three categories. In the field of HAWT, the first CFD computation with a Reynolds-averaged Navier-Stokes (RANS) solver dates back to 1998. 1 Since then, RANS computations have become the working horse in industrial context because of its comparatively low costs. The multiple fields of application in wind energy range from the analysis of wake characteristics, 2 interaction between terrain and HAWT, 3 transition behaviour, 4 and aeroelastic effects. 5,6 Large eddy simulations (LES) are nowadays used to analyse the interaction between the atmospheric boundary layer (ABL) or terrain and the HAWT wake. 7-9 During the simulation, the HAWT is modelled by an actuator line model, while the grid spacing of the LES grid reaches up to one rotor diameter. The third approach comprises all variations of detached eddy simulations (DES) that are usually used to link the scales between the LES and the RANS for specific wind turbine configurations, for example, the analysis of unsteady loads during stall in rotor-upwind configurations 10 or standstill. 5 In order to improve the performance a...