This thesis describes the further development and validation of the dynamic meandering wake model for simulating the flow field and power production of wind farms operating in the atmospheric boundary layer (ABL). The overall objective of the conducted research is to improve the modelling capability of the dynamics wake meandering model to a level where it is sufficiently mature to be applied in industrial applications and for an augmentation of the IEC-standard for wind turbine wake modelling. Based on a comparison of capabilities of the dynamic wake meandering model to the requirement of the wind industry, four areas were identified as high prioritizations for further research: 1. the turbulence distribution in a single wake 2. multiple wake deficits and build-up of turbulence over a row of turbines 3. the effect of the atmospheric boundary layer on wake turbulence and wake deficit evolution 4. atmospheric stability effects on wake deficit evolution and meandering The conducted research is to a large extent based on detailed wake investigations and reference data generated through computational fluid dynamics simulations, where the wind turbine rotor has been represented by an actuator line model. As a consequence, part of the research also targets the performance of the actuator line model when generating wind turbine wakes in the atmospheric boundary layer. Highlights of the conducted research: 1. A description is given for using the dynamic wake meandering model as a standalone flow-solver for the velocity and turbulence distribution, and power production in a wind farm. The performance of the standalone implementation is validated against field data, higher-order computational fluid dynamics models, as well as the most common engineering wake models in the wind industry. 2. The EllipSys3D actuator line model, including the synthetic methods used to model atmospheric boundary layer shear and turbulence, is verified for modelling the evolution of wind turbine wake turbulence by comparison to field data and wind tunnel experiments. 3. A two-dimensional eddy viscosity model is implemented to govern the distribution of turbulent stresses in the wake deficit. The modified eddy viscosity model improves the least-square fit of the velocity field in the wake by ~13% when compared to higher-order models. 4. A method is proposed to couple the increased turbulence level experienced by a turbine operating in waked conditions, to the downstream wake evolution of the wake-affected turbine. The intraturbine turbulence coupling improved the fit of the turbulence distribution by ~40% and the wind speed distribution by ~30% over a row of eight turbines. 5. The effect of the atmospheric shear on the turbulent stresses in the wake is captured by including a local strain-rate contribution for the ambient shear gradient. This results in more realistic turbulent stress levels in regions of small wake deficit gradients; this is particularly important in the far-wake region where atmospheric shear gradients are an important contribution to the local strain-rate. 6. A method to include the effect of atmospheric stability on the wake deficit evolution and wake meandering is described. Including the atmospheric stability effects improved the model prediction of the mean velocity field by ~19% and of turbulence distribution by ~28% in unstable atmospheric conditions compared to actuator line results. The power production by a row of wind turbines aligned with the wind direction is reduced by ~10% in very stable conditions compared to very unstable conditions at the same turbulence intensity. This power drop is comparable to measurements from the North Hoyle and OWEZ wind farms.
DTU Wind Energy PhD-0012; DTU-Wind-Energy-PhD-0012