Issue with large force time-history fluctuations in OpenFAST output

[birdie888]

Hello everyone,

I am currently working with the official OpenFAST model 5MW_OC4Jckt_DLL_WTurb_WavesIrr_MGrowth and comparing my simulation results with the reference case LC2.1 described in the paper “Offshore Code Comparison Collaboration Continuation (OC4), Phase I – Results of Coupled Simulations of an Offshore Wind Turbine with Jacket Support Structure”.

In this load case, the offshore wind turbine is assumed to be fully rigid, with no wind and still water conditions. I followed the LC2.1 configuration as closely as possible and compared my results with the published benchmark data.

From the OpenFAST outputs, the mean values of the axial force in member K1L2, the axial force in Brace59, as well as the structural displacements are in good agreement with the reference results. However, I observe relatively large fluctuations in the force and displacement time histories around the mean values, even under these no wind and still water conditions.

I would like to ask whether such fluctuations are expected for LC2.1, and whether they could be related to numerical aspects such as time step size, damping settings, or solver tolerances. I am also wondering if there are recommended ElastoDyn or SubDyn settings that can help reduce these oscillations for this case.

Any insights or suggestions would be greatly appreciated. Thank you very much for your help.

[birdie888]

[andrew-platt]

Which version of OpenFAST are you using?

[birdie888]

Thank you very much for your reply. I am using OpenFAST V4.1.2 released

[jjonkman]

Dear @birdie888,

These time-dependent responses are the initial condition solution from the structural DOFs that are enabled. You have the six Guyan modes of the support structure (enabled through ElastoDyn DOFs PtfmSgDOF - PtfmYDOF), as well as eight Craig-Bampton modes (Nmodes = 8 in SubDyn), enabled.

If you track where the ElastoDyn platform displacements (ElastoDyn outputs PtfmSurge - PtfmYaw) are setting out to, you can set these as initial conditions for a subsequent simulation to minimize the start-up transient associated with the Guyan modes (it looks like this is likely already done for PtfmHeave, given its nonzero initial condition set in ElastoDyn, but the other initial platform displacements are zero). However, there is no way through the SubDyn input file to set initial displacements of the Craig-Bampton modes, so, these will still settle out over time. That said, you could disable these Craig-Bampton modes (leaving only the Guyan and static-improvement modes) to eliminate the transient resulting from the Craig-Bampton modes).

Best regards,