High-fidelity aeroelastic analyses of wind turbines in complex terrain: FSI and aerodynamic modelling

This paper shows high-fidelity Fluid Structure Interaction (FSI) studies applied on the research wind turbine of the WINSENT project. In this project, two research wind turbines are going to be erected in the South of Germany in the WindForS complex terrain test field. The FSI is obtained by coupling the CFD URANS/DES code FLOWer and the multiphysics FEM solver Kratos, in which both beam and shell structural elements can be chosen to model the turbine. The two codes are coupled in both an explicit and an implicit way. The different modelling approaches strongly differ with respect to computational resources and therefore the advantages of their higher accuracy must be correlated with the respective additional computational costs. The presented FSI coupling method has been applied firstly to a single blade model of the turbine under standard uniform inflow conditions. It could be concluded that for such a small turbine, in uniform conditions a beam model is sufficient to correctly build the blade deformations. Afterwards, the aerodynamic complexity has been increased considering the full turbine with turbulent inflow conditions generated from real field data, in both a flat and complex terrains. It is shown that in these cases a higher structural fidelity is necessary. The effects of aeroelasticity are then shown on the phase-averaged blade loads, showing that using the same inflow turbulence, a flat terrain is mostly influenced by the shear, while the complex terrain is mostly affected by low velocity structures generated by the forest. Finally, the impact of aeroelasticity and turbulence on the Damage Equivalent Loading (DEL) is discussed, showing that flexibility is reducing the DEL in case of turbulent inflow, acting as a damper breaking larger cycles into smaller ones.

The effect of oscillating turbulent inflow on the shape of downstream turbulence

Large Eddy Simulations (LES) has been widely applied and used in several decades to simulate a turbulent boundary layer in the numerical domain. In this study, we aimed to make a synthetic inflow generator (SIG) yielding an appropriate property of turbulent boundary layer in the inlet section and making quick development in the downstream of a three-dimensional domain. In order to achieve turbulent boundary layer quickly in a limited domain, the oscillating term was implemented in the well-defined boundary layer, which was expected to make faster convergence in the calculation. Cholesky decomposition was also applied to possess turbulent statistics such as the randomness and correlation of turbulent flow. In a result, the oscillating inflow did not show the faster convergence, but it indicated a possibility to improve statistical quantities in the downstream. In addition, regarding the mean flow characteristics were very close to the calculation without the oscillating flow. On the other hand, the turbulent statistics were improved depending on the oscillating magnitude.

CFD analysis of a Darrieus Vertical-Axis Wind turbine installation on the rooftop of a buildings under turbulent inflow conditions

The behavior of a rooftop mounted generic H-rotor Darrieus vertical axis wind turbine (H-VAWT) is investigated numerically in realistic urban terrain. The interaction of the atmospheric boundary layer with the different buildings, topography, and vegetation present in the urban environment leads to the highly turbulent inflow conditions with continuously changing inclination, and direction. Consequently, all these factors can influence the performance of a VAWT significantly. In order to simulate a small H-VAWT at rooftop locations in the urban terrain under turbulent inflow conditions, a computational approach is developed. First, the flow field in the terrain is initialized and computed with inflow turbulence. Later, the wind turbine grids are superimposed for further computation in the turbulent flow field. The behavior of the H-VAWT is complex due to the 3D unsteady aerodynamics resulting from continuously changing the angle of attack, blade wake interaction, and dynamic stall. To get more insights into the behavior of a rooftop mounted H-VAWT in turbulent flow, high fidelity DDES simulations are performed at different rooftop positions and compared the results against the behavior at uniform inflow conditions in the absence of inflow turbulence, built environment. It is found that the performance of wind turbine is significantly increased near the rooftop positions. The skewed flow at the rooftop location increases the complexity. However, this effect contributes positively to increasing the performance of wind turbines.

A Validation Study of the Aerodynamic Behaviour of a Wind Turbine: Three-Dimensional Rotational Case

Numerical modelling and simulation of a rotating, tapered, and twisted three-dimensional blade with turbulent inflow conditions and separating flows is a challenging case in Computational Fluid Dynamics (CFD). The numerical simulation of the fluid flow behaviour over a wind turbine blade is important for the design of efficient machines. This paper presents a numerical validation study using the experimental data collected by the National Renewable Energy Laboratory (NREL). All the simulations are performed on the sequence S of the extensive experimental sequences conducted at the NASA/Ames wind tunnel with constant RPM and variable wind speeds. The results show close agreement with the NREL UAE experimental data. The CFD model captures closely the totality of the defining quantities. The shaft torque is well-predicted pre-stall but under-predicted in the stall region. The three-dimensional flow and stall are well captured and demonstrated in this paper. Results show attached flow in the pre-stall region. The separation appears at a wind speed of 10 m/s near the blade root. For V>10m/s, the blade appears to experience a deep stall from root to tip.

Numerical Investigation of Droplet Properties of a Liquid Jet in Supersonic Crossflow

The atomization process of a liquid jet in supersonic crossflow with a Mach number of 1.94 was investigated numerically under the Eulerian-Lagrangian scheme. The droplet stripping process was calculated by the KH (Kelvin-Helmholtz) breakup model, and the secondary breakup due to the acceleration of shed droplets was calculated by the combination of the KH breakup model and the RT (Rayleigh-Taylor) breakup model. In our research, the existing KH-RT model was modified by optimizing the empirical constants incorporated in this model. Moreover, it was also found that the modified KH-RT breakup model is applied better to turbulent inflow of a liquid jet than laminar inflow concluded from the comparisons with experimental results. To validate the modified breakup model, three-dimensional spatial distribution and downstream distribution profiles of droplet properties of the liquid spray in the Ma = 1.94 airflow were successfully predicted in our simulations. Eventually, abundant numerical cases under different operational conditions were launched to investigate the correlations of SMD (Sauter Mean Diameter) with the nozzle diameter as well as the airflow Mach number, and at the same time, modified multivariate power functions were developed to describe the correlations.