Difference in Cp between BEMT and OLAF

Hello everybody,

I am an MSc student currently doing my thesis at the TU Delft in aerodynamics and wind energy. I started using OpenFAST a couple of weeks ago and wanted to compare the results from two different “WakeMod” types in the AeroDyn15 input file, namely BEMT and OLAF. What I found was that the BEMT results coincide with the literature but the OLAF results are way off and even break the Betz limit. Some details about the simulation are shown below:

  • It is a static turbine, not floating (the baseline 5 MW from the NREL)
  • It is below rated conditions with a prescribed rotor speed (TSR 7.5 with 9 m/s input speed)
  • The input is not turbulent
  • The circulation is not prescribed, currently, the CircSolvingMethod is cl-based
  • I have run the simulation for 30 seconds since the results are already converged at that point (see the plots in the attachments)

Under these conditions, BEMT gives a Cp of 0.48 while OLAF gives a Cp of 0.60. The only change in the two simulations is the change in WakeMod and therefore also the addition of the OLAF input file. I have played around with the OLAF input variables quite a bit to see what would happen in the output but nothing significant seems to change. Do you have any idea what the mistake could be?

For reference, I have added the Cp plot of the BEMT and OLAF simulation in the attachments.

I look forward to hearing from you.

Kind regards,

Carlos Malveiro

Cp_BEMT_vs_OLAF

Hi Carlos,

We generally observe higher Cp with OLAF, but my guess is that in your case, it’s likely that you wake is not long enough (not enough wake panels). You can find guidelines on how to set OLAF parameters based on the time step and rotational speed:

https://openfast.readthedocs.io/en/dev/source/user/aerodyn-olaf/RunningOLAF.html#guidelines

Let us know if that works.

It’s also a good idea to visualize the wake (using WrVTK in the OLAF input file) when setting up a new model. The VTK outputs do increase the computational time, so it’s best to use low FPS.

Cheers,

Emmanuel

Hi Emmanuel,

Thank you for the quick response. I updated the OLAF input file based on the link that you sent and it turns out that it was the DTfvw value that had to be increased (it was still at default which equals DTAero). This new DTfvw corresponds to 6 degrees of one rotor revolution as explained in the guidelines, and gives a more realistic Cp of 0.55. But why should this DTfvw correspond to 6 degrees and can it not be lower?

Besides that, it seems strange to not have the same time step in both AeroDyn and OLAF. Does it not matter that the DTAero is not equal to the DTfvw? If I increase the DTAero to the same value as DTfvw then the warning “some vortices were found below the ground” occurs after which a fatal error pops up.

I also wonder why it is that we generally observe a higher Cp with OLAF as compared with BEMT? The reason I am asking is that I am also running a CFD actuator line model simulation of this wind turbine and there I observe a lower Cp value. I would assume that the OLAF Cp would be somewhere in between the CFD simulation and the BEMT simulation since it is in between the two in terms of fidelity.

I look forward to hearing from you.

Kind regards,

Carlos Malveiro

Hi Carlos,

Using 6 degrees of discretization is only a recommendation, you can use the same time step as AeroDyn, but you should then change the number of near wake and far wake panels accordingly so that you have a sufficiently long wake. The feature of having different time step for the glue code and OLAF is mostly useful when small time steps are needed by the structural solver (e.g. BeamDyn). Otherwise, it’s fine and simpler to use the same time step for all.

Vortices entering the ground can occur at times (with high yaw/shear or turbulence), but often it is a result of the wake not being regularized enough. Visualizing the wake may help indicate if indeed the wake is not “smooth” enough, in which case the wake regularization and the corespread eddy viscosity can be increased.

CPs are typically higher because the induction is lower than the ones obtained with BEM. Increasing the wake length may increase the induction. Having more spanwise discretization might allow for more expansion and rollup of the wake, which can potentially increase the induction as well. So far we don’t have strong guidelines on these matters, and it could be that some improvements are needed for OLAF. Since CFD, BEM and Vortex methods have different physics and shortcomings, it’s not easy to say which one to trust. I’ve heard that CFD’s CP tend to be too low compared to CPs obtained in the field. But I agree that the CPs obtained with OLAF are often on the high end.

I hope that helps,

Emmanuel

Hi Emmanuel,

Thank you for the information. It all makes sense now.

Kind regards,

Carlos Malveiro