CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FloEFD, FloWorks & FloTHERM (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/)
-   -   Best practices for modeling a rotor/propeller (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/242833-best-practices-modeling-rotor-propeller.html)

g7sznAc4 May 14, 2022 12:57

Best practices for modeling a rotor/propeller
 
Hi, I have a 2m span drone propeller/rotor that I'd like to simulate in SolidWorks Flow Simulation 2022. I have experimental data for validation: thrust and torque at various RPMs, as well as some pitot probe data from 15" upstream and downstream - so I'd like to resolve the wake as well as aerodynamic forces.

I have lots of experience with other solvers using body-fitted meshes, but not much experience with SolidWorks. It seems like most of my previous experience does not translate well to FloEFD.

I've spent quite a bit of time examining sensitivity to parameters such as computational domain size, rotating region size (I'm using local regions/averaging), local mesh refinement on the propeller, global mesh refinement, and calculation control refinement settings.

I'm having a very hard time matching the experimental data - my best prediction so far overpredicts torque by 10% and underpredicts thrust by 15%. I've found some literature using FloEFD on similar simulations where they were able to get a much better match to experiment, so I'm hoping that I'm overlooking something and it's possible to improve my predictions.

What I'm finding so far:
  • Computational domain size doesn't seem very important, unlike other solvers I've used. A small domain is fine.
  • Global rotating does not seem to work - even with a fairly tight computational domain, there are large bubbles of supersonic flow in the farfield and the predicted forces are nowhere close to accurate.
  • A smaller rotating region seems best - only just large enough to encompass the propeller. I thought that a larger rotating region might provide a more realistic flowfield around the blades and give better predictions, but with a larger region I have convergence problems and the simulation is somewhat unsteady. In general, I see some odd transients near the radial boundary of the cylindrical rotating region.
  • In order to resolve the wake, it seems I must have a very small rotating region - otherwise there seems to be no wake downstream of the propeller.
  • Initially I thought that using a local mesh to refine the propeller would be best, but it seems that having a fine propeller mesh actually hurts predictions - I get the best convergence and predictions if I use no local mesh at all - just a simple automatic global mesh, and let FloEFD refine as it wishes. However, the mesh on the propeller looks coarse to me so I'm wondering if this is why my predictions are poor.
  • For calculation control options, it seems I must start with a very coarse mesh and run for many travels (~10) to let the wake develop. Only once the wake has developed somewhat can I begin refining the mesh, otherwise there is very little wake.
  • Surface goals don't seem to work correctly - if I create a surface 15" downstream of the propeller and try to add goals looking at e.g. min/max/average static pressure, I get an error about an invalid body and have to delete the goals to proceed.
  • It really doesn't take much mesh to get a converged solution - after around 500k total cells and 50k fluid/solid cells the answer doesn't change much. I've run as high as 40m cells. However, FloEFD doesn't seem to want to refine the fluid/solid cells - it's happy with around 10k, I have to add more manually if I want further refinement of the propeller.

Is this consistent with what others have found? If anyone has any recommendations for improving my thrust/torque predictions I would greatly appreciate it.


All times are GMT -4. The time now is 06:13.