Effects of the time step on AoA for FAST v8.16

Hi All,

I ran two cases of FAST simulations on NREL 5MW at the rated configuration ( -5 tilt, -2.5 pre-cone, etc.). The wind condition is nonuniform wind with the power-law profile, 10 m/s at 100 m. All the DoFs in the ElastoDyn are disabled so that the rotor constantly rotate at 12 rpm. The BeamDyn uses the reference blade input file and has 7 nodes. The AeroDyn blade input file is also from the reference file.

The AeroDyn setting is the following,

One of the cases has time step, 0.01 sec and another one has time step 0.005 sec. The AoA responses at the Blade 1 tip from these 2 cases have about 0.2 deg differences. A full-time history plot is below

I am wondering that is it the smaller time-step yields more accurate result? or for the wind turbine aerodynamics, that 0.2 deg differences do NOT have significant effects. Is there a way to explain why different time-step causes such variation? I am guessing its the AeroDyn-BeamDyn coupling.


Dear Daniel,

Normally the simulation time step should be set so as to get a converged solution (and to maintain numerical stability). While both of your solutions look numerically stable, it doesn’t appear that the 0.01-s simulation is converged. Is the solution converged with the 0.005-s time step i.e. does decreasing the time step further change the result?

FYI – Test26 in the FAST CertTest is a model of the NREL 5-MW turbine using BeamDyn to model the blade structural dynamics. This model was set up to have a time step of 0.001 s, of course, the time step could be increased when you disable all structural DOFs in ElastoDyn, as you’ve done.

Best regards,