ebranlard
commented
1 year ago This feature would indeed be great.
As a note, on top of the careful interpolation of polars (to avoid double stall), we should implement 3D polar corrections into AeroDyn, because 3D polar corrections typically depend on the radial location (c/r).
The Snel and Du-Selig implementation can be found here together with a fairly standard blending procedure (to blend the corrected polar with the uncorrected polar). This will require:
- new AeroDyn input file parameters to choose which 3D correction to apply to the polar.
- Possible knowledge of the tip-speed ratio (either from ElastoDyn initial rotor speed), or a user input.
- Robust routines to calculate $C_{l,\alpha}$ and $\alpha0$. I believe we already have some in AFI, but we should test the robustness of the methods.
- Possibly some new AeroDyn inputs for the blending of the 3D correction.
It's likely fine to first implement the interpolation of polars, and then implement the 3D correction. The error introduced by omitting the 3D correction are likely small.
jjonkman
ietqlw
Feature overview For coupling AeroDyn to CFD codes, it would be really useful to have a method built directly into AeroDyn 15 to refine the aero mesh. In issue #1232, there was a mismatch in position between the aero position mesh and the force mesh passed to CFD codes such as NALU-Wind, AMR-Wind, and SOWFA. This issue had come about as a result of including the structural mesh in the mesh interpolation within the OpenFOAM module (OpFM). While the structural mesh was removed with PR #1324, the interpolation of the aero mesh to a finer mesh that the CFD requires is still handled within the OpFM module. This refinement might be giving rise to discontinuities in the aero loads passed back to the CFD as shown in the following figure.
Figure: Comparison of CFD couplings with blade resolved (BR) CFD. Note the discontinuities in the normal force in the BEM, Vortex-Method, and ALM curves . These discontinuities are likely due to a combination of changes between airfoils in AD15 and mesh mapping from the course aero mesh to the finer CFD mesh near the airfoil change.
In addition to possible discontinuities from the mesh mapping, it would be very convenient to include a simple mesh refinement in AD15 directly. This would greatly simplify using OpenFAST with a CFD code.
Describe the solution you'd like We would like to include a mesh refinement parameter in the AD15 input file. This would trigger AD15 to interpolate additional nodes along the blade span, and would require some interpolation of airfoils.
From @mchurchf (comment https://github.com/OpenFAST/openfast/issues/1232#issuecomment-1317119119 from #1232)
To summarize, AD15 would get the following changes:
In addition, the following changes would be made to the OpFM module:
ActForceMotionsPointsandActForceMotionsPointsmeshes and all interpolation to generate themAD15%u%BldMotionmesh positions to CFD (this is now a refined mesh)AD15%y%BldLoadsmesh positions and values to CFD (this is now also a refined mesh)And some API changes for communication with CFD codes:
NumActForcePtsBladepassed from CFD would be removedNumActForcePtsTowerpassed from CFD would be removedNumPtsBladenew parameter to pass to CFDNumPtsTowernew parameter to pass to CFDOpenFOAM_Types.hfileDescribe alternatives you've considered The current best practices method of refining the AD15 mesh is to preprocess the AD15 blade file and polars to create a more refined blade file with modified polars. This method still leaves some room for error as the CFD mesh is linearly spaced along the blade span or tower, but the AD15 meshes are not uniformly spaced. As show in the figure above, this may be leading to artifacts.
Those who may want to weigh in on this proposal @tonyinme @mchurchf @michaelasprague @gantech @jjonkman @ebranlard