Abstract
Large-scale wind turbine installations are sited using layouts based on site topology, real estate costs and restrictions, and turbine power output. In order to plan efficient wind farm deployments, computationally efficient CFD models of wind turbine wakes are needed to predict the performance of individual turbines in the field. Given the large computational demands to calculate turbine wakes using traditional CFD methods, many research efforts have focused in recent years on the development of blade modeling techniques. These blade models are combined with CFD field computations and are intended to model the forces on turbine blades and their resulting effects on the flow, without actually including a physical model (grid) of the blades in the CFD simulation. The most promising method, the Actuator Line (AL) method, creates a physical model of the turbine blades, and has been found to result in realistic wake flow simulations with accessible compute resources. The AL method typically uses an exponential function to spread blade forces over a cloud of CFD grid points to prevent singularities from occurring when large force magnitudes are applied to individual grid points. Yet, in typical implementations of this method, the geometric extent to which this distribution occurs and the applied force directions do not appear to be related to the physics of the blade force calculations or geometry. Moreover, the distribution function reduces the force magnitude as distance from the actuator line increases, but this does not necessarily match the force loading on an actual turbine blade.
In order to address some of these issues, the current research study has focused on developing an evolved actuator line method (ALEvo) in combination with the advanced parallel Spectral Element Method (SEM) open-source CFD code NEK5000, to model the wake flows of large wind turbines. This alternate AL implementation creates a geometric representation of each rotating turbine blade, and the aerodynamic forces are calculated using traditional blade element equations. Turbine blades are rotated in time with progressing CFD field calculations. Blade forces are then input as body forces into the Navier Stokes equations in the host CFD program without the use of an exponential spreading function. In this paper, we compare results for the new actuator line method with those of the traditional AL method, as well as with published experimental results. Comparison with the traditional method shows better agreement of ALEvo with experimental data, and confirms overestimation of power output with the traditional method. Furthermore, significant computing time savings are achieved with the proposed new method.