Question about the Selection of Simulation Time Step

[axxyyyyyyyy378]

Hi everyone,

I came across the empirical rule for choosing the OpenFAST simulation time step in a previous post: DT ＜ 1 / (10*f) where f is the highest full-system natural frequency of the model. In principle, this frequency should be obtained via a linearization analysis followed by an eigenvalue decomposition (and if blade degrees of freedom are enabled and the rotor is spinning, the MBC3 transformation must be used). I only understand the theory but don’t know how to carry it out in practice. So, can the current OpenFAST linearization module be used to extract the maximum full-system natural frequency? I’d like to compute this frequency for the 5MW_OC4Semi_WSt_WavesWN baseline model. Could you share some practical tips on how to do that? Thank you!

[jjonkman]

Dear @axxyyyyyyyy378,

The OpenFAST linearization functionality can be used to derive a linear-equivalent state-space model of the nonlinear OpenFAST solution about an operating point. Post-processing this linear model can be performed to derive the full-system natural frequencies, damping, and mode shapes (i.e., application of MBC3 followed by Eigenanalysis on the state matrix, A).

There are many posts on our forum (https://forums.nrel.gov/) and on OpenFAST GitHub that discuss the process, which I will not repeat here. I'll just say that the Automated Campbell Diagram Code (ACDC) is the easiest way to execute such an analysis: https://github.com/OpenFAST/acdc.

Best regards,