Abnormal Rotor Speed without ServoDyn, NREL 5MW WT

Hi All,

I have tried several cases based on Test26.fst. I changed the wind input to steady wind condition with 12 mps and 0.2 ShearExp at 90 m (reference hub height).

Since my primary interest is the blade deflections under a steady wind condition, so I disabled the ServoDyn in my simulations. I was assuming that FAST would couple the InflowWind, BeamDyn, and AeroDyn only. However, I found out that the rotor speed would keep increasing without the ServoDyn module. It could reach more than 20 rpm in a 20 sec simulation.

I cannot figure out what is going on. I also read the ServoDyn workshop PDF, but it does not address this issues. Therefore, I am wondering why the rotor speed would not reach steady-state and how the FAST coupled with ServoDyn.

Below images are results from the FAST. Note that both cases have 20 sec simulation (0.0005 time step size) and steady wind condition at 12 mps.


Dear Daniel,

The ServoDyn module of FAST v8 only models the physics of the controller and electrical drive. The actually structural degrees-of-freedom (DOFs) of the turbine, including the generator rotation and shaft torsion, are modeled by the ElastoDyn module (except for the structural dynamics of the blades, which are modeled in BeamDyn in Test26).

So by disabling ServoDyn, you’ve eliminated the controller, including eliminating the generator torque and active blade-pitch commands; if you run the simulation long enough, the rotor speed will eventually reach steady state at the rotor speed at which the aerodynamic torque drops to zero.

If you want to simulate without the controller, I suggest at least disabling the generator DOF in ElastoDyn (GenDOF = False) so as to simulate with the generator spinning at a fixed speed.

I hope that helps.

Best regards,

Hi Jason,

Thank you for the answer. I will try some cases with your suggestion.

In order to understand FAST more properly, is there any document about how FAST couple each module, especially between InflowWind, AeroDyn, and BeamDyn?

I am thinking to write my own driver file to couple InflowWind, AeroDyn, and BeamDyn, so I can focus on the rotor design itself.

Please let me know if my idea is feasible and heading right direction.

Again, thank you for your help.

Dear Daniel,

The general coupling algorithm implemented in the FAST glue code is explained in the following papers:

nrel.gov/docs/fy14osti/60742.pdf
nrel.gov/docs/fy16osti/63203.pdf

However, the mixed time-step algorithm and the use of MUMPS documented in the second paper have not been fully implemented and released. Also, a number of liberties (deviations from the algorithm described above) have been taken in the actual FAST glue-code implementation to speed-up the algorithm based on experience and knowledge of the modules. You can review the FAST source code to see exactly how the modules are coupled.

You could develop your own glue code, but this will not be easy.

Best regards,