Actuator turbine models and trailing edge flow: implementation in an in-house code

Abstract

To be able to simulate a trailing edge flow in CALC some modifications needs to be done in the code. The main reason for this is that CALC only can handle a computational domain consisting of one block. Modifications are made both in the multigrid solver that solves the pressure field and the flow solver. Some of the modifications that are made in the multigrid solver can be recognized from the implementations in the flow solver but the multigrid solver uses 1D arrays compared to 3D arrays in the flow solver. This implementation allows CALC to run a flat plate simulation and this was one of the validating cases run. One laminar flow and one turbulent flow using the Reynolds Averaged Navier Stokes (RANS) turbulent model and the results showed good agreement. One more validating RANS case was used here representing an airfoil instead of a flat plate. This simulation showed that the flow fulfilled the no slip condition at the surface of the airfoil which was the focus of the validation. The second part of this master thesis consisted of implementing two Actuator turbine models named Actuator disk model (ADM) and Actuator line model (ALM). The implementation of ADM which is the less accurate model of the two was validated by an axisymmetric flow using the 5-MW National Renewable Energy Laboratory (NREL) wind turbine. The ALM model was validated using the same turbine but with a 3D flow due to its 3D behaviour. The results were acceptable when comparing to the NREL data and results in [1]. The results from the ADM simulation obtained by restricting the "i variable in the Gaussian function showed that more consistent predictions of the rotor thrust and power between different meshes could be obtained.