Help with Campbell diagram: unexpected jumps in mode frequencies

[jefvar]

Hello @jjonkman,

Last time I reached out to you with a question about a linearization problem I was having. I forgot to mention that it is already solved:
#3391 (comment)
Thanks again for your help with that.

Now, after performing the linearization and the MBC transformation, I am trying to obtain the Campbell diagram. I am working with the DTU 10MW RWT, linearized at 10 wind speeds (4–25 m/s) with CompAero = 2 and DBEMT_Mod = -1, using the steady-state rotor speed and pitch angle at each operating point. I then post-processed the resulting .lin files with the Campbell scripts from the matlab-toolbox (campbell_diagram_data, identifyModes, plotCampbellData).

I expected each mode frequency to stay reasonably smooth across wind speeds, but my diagram shows some "jumps" above rated wind speed, as seen in the picture:

From this point I am not sure where the problem comes from. Do you know where I should double-check? If it helps, I attach some files related to this case.
CampBellDTU_10MW_RWT.zip

Thank you in advance!

[jjonkman]

Dear @jefvar,

I'm glad to hear your question in discussion #3391 has been solved.

Your Campbell diagram looks close to correct, except that a few of the modes are misinterpreted/mislabled at a few operating points (where the jumps are), e.g.,:

• 1st blade flap (collective) at 17 m/s
• 1st blade edge (progressive) at 17 m/s
• 1st drivetrain torsion at 15 m/s

I see that you are using the old Campbell scripts from the MATLAB Toolbox, but the mode identification approach used in these scripts was not robust. I'm not exactly sure what is causing the misinterpretation by the script, but you could try to identify the modes manually instead. That said, we now recommend using the Automatic Campbell Diagram Code (ACDC) in place of the old MATLAB Toolbox scripts: https://github.com/OpenFAST/acdc. While still not foolproof, I would hope ACDC does a much better job at identifying the modes, but it also makes it easier to review each mode and label them properly if not automatically identified correctly. Does ACDC solve your problem?

Best regards,

[jefvar]

Dear @jjonkman,

Thank you for the recommendation — I would say yes, ACDC solved my problem.

One difference with my original setup: as far as I can see, ACDC expects VSContrl = 1 in the ServoDyn file (I could not change it), while my linearizations were run with VSContrl = 0 using a Simple Induction Generator. Following your earlier suggestion of keeping the generator torque constant during TrimCase = 3, I had set the SIG synchronous speed to half of the computed operating generator speed (I compute the initial conditions of my simulations beforehand), so that with a pull-out ratio of 1 the generator always operates beyond the pull-out point and its torque remains constant at that speed. In any case, I set up roughly equivalent parameters with VSContrl = 1 and obtained a good result, as seen in the picture:

For context, my end goal is to use the Campbell diagram to identify the resonant frequencies that show up in the Bode plots of my linear models and to know which turbine component each one belongs to as a preliminary step before tuning the pitch and torque controllers. In the ACDC plot the modes are currently labeled "Line1, Line2, …": is there a way to have ACDC display the identified mode names instead, or would you suggest cross-referencing them with the MATLAB Toolbox diagram?

Best regards,

[jjonkman]

Dear @jefvar,

I'm glad ACDC solved your issue.

We typically use VSContrl = 1 to set a constant generator toque, but I could see how VSContrl = 0, GenModel = 1 (SIG) could be set-up to achieve the same thing.

ACDC attempts to connect similar modes across operating points with lines, but doesn't try to label the lines as specific modes, which the MATLAB Toolbox tried to do (but was not always robust). But ACDC does make it easy to visualize the modes by clicking on the Campbell Diagram, which should make it easier for you to label the lines yourself. You could certainly use the MATLAB Toolbox in combination, at least on the first pass.

Best regards,