interpretation of MBC-based modal results in large, highly flexible turbines

[TaoWan22]

Dear OpenFAST team,
Thank you very much for developing and maintaining OpenFAST and the linearization/MBC tools.
I am currently using OpenFAST linearization together with MBC to analyze aeroelastic modes of a large, flexible wind turbine (IEA 15 MW).
I would like to ask a question regarding modal identification after the MBC transformation when higher-order blade DOFs are included.
In my current setup:

• I use multiple .lin files at a single operating point (same wind speed, rotor speed, and pitch), covering many azimuth positions.
• MBC is applied using openfast_toolbox to obtain a time-invariant linear system.
• The structural model includes higher-order blade flapwise and edgewise DOFs (in addition to first-order modes).
  After MBC, I observe that:
• Tower-dominated first-order modes (FA/SS) are clearly identifiable.
• However, classical first-order blade system modes (e.g. collective / pitch-wise / yaw-wise flapwise and edgewise modes) do not appear as single, clearly separated modes.
• Instead, they seem to split into groups of closely spaced blade-dominated modes, with contributions from multiple higher-order blade DOFs (e.g. Bld sin/cos N8–N11 FLAP or EDGE).
  This makes it difficult to directly associate the numerical modes with the traditional “first-order system modes” commonly shown in the literature (e.g. collective, pitch-wise, yaw-wise blade flap/edge modes).
  My questions are:

1. Is this modal “splitting” an expected and physically correct consequence of including higher-order blade flexibility in the linearized model?
2. From the OpenFAST developers’ perspective, what is the recommended approach for modal identification in this case?
   • Should users reduce the model (e.g. freeze higher-order blade DOFs) when they want to recover the classical first-order system modes?
   • Or is it preferred to interpret the results in terms of modal families rather than one-to-one correspondence with simplified textbook modes?
3. Are there any recommended post-processing or diagnostic quantities (e.g. specific modal energy measures or state groupings) that can help identify representative blade flap/edge system modes in a high-DOF model?
   I would greatly appreciate any guidance on the intended interpretation of MBC-based modal results in large, highly flexible turbines.
   Thank you very much for your time and for the excellent tools.
   Best regards

[jjonkman]

Dear @TaoWan22,

Can you provide more information, e.g., a figure, explaining what you mean by modal "splitting". I'm not sure I understand what you are referring to to comment.

When you refer to higher-order blade DOFs, are you referring to the use of BeamDyn within OpenFAST, or something else?

After linearizing about multiple azimuth angles, are you azimuth-averaging the state-space matrices to obtain the LTI model?

Best regards,

[TaoWan22]

Thank you for your response. Below I provide more details on the current setup and results.

This case corresponds to a standstill operating condition, with both the blade pitch angle and nacelle yaw angle held fixed. The .fst initial file：

the linearization settings：

, and the relevant parameters in the IEA-15-240-RWT-Monopile_ElastoDyn.dat file are kept consistent throughout the simulations：

After completing the simulations, I used the OpenFAST Toolbox to extract modal information from 41 linearization files obtained at the same operating point, and the modal results were subsequently averaged.

The resulting table contains the following five columns:

• Mode Index
• Natural Frequency (Hz)
• Damped Frequency (Hz)
• Damping Ratio
• Mode Description

Taking the first mode as an example

the corresponding mode description includes contributions such as
ED 1st tower SS and
Bld cos(N·ψ) EDGE/FLAP-DISP.How are these coupled modes post-processed, and how is this description interpreted in the literature reports?
Best regards,

[jjonkman]

Dear @TaoWan22,

Thanks for clarifying.

I see that you've parked the rotor (GenDOF = FALSE with RotSpeed = 0 rpm in ElastoDyn). As such, I don't see the point of linearizing at different time instances and averaging. I suggest ensuring the solution is in steady state before linearizing, by setting CalcSteady = TRUE. Then--given that the rotor is not spinning--it should be sufficient to linearize only once and not have to average at all.

Regarding the modes, the modal interpretation from MBC3 you are using is now outdated. I recommend using our Automatic Campbell Diagram Code (ACDC) instead to drive the linearization and modal interpretation process: https://github.com/OpenFAST/acdc. ACDC provides a convenient way to visualize the full-system mode shapes, which will allow you to more easily interprate what each mode is.

Best regards,