Using OpenFAST/ROSCO for supervisory derating and jacket substructure load assessment

[YuhengShey]

Hello everyone,

I am a PhD researcher studying ageing jacket-supported offshore wind turbines, with a focus on whether supervisory derating / load mitigation strategies can help manage fatigue and extreme-load demand in the support structure.

My current workflow is based on ROSCO + OpenFAST, using the OC4 jacket as a baseline model. I would like to better understand the appropriate role of OpenFAST/SubDyn in this type of study, especially when the interest is the jacket substructure, rather than only the rotor–nacelle–tower system.

I would greatly appreciate guidance on the following points:

I am a reseacher who are investgating in the how to use the wind turbine to derating and management about jacket type substructure fatigue or extreme load. Currently I simulate using the Rosco & Openfast simualtion workflow. I have several question want to ask:

1. Use of SubDyn outputs for substructure assessment
   If I apply a derating strategy such as power derating or peak shaving, can I use SubDyn outputs (for example, mudline bending moments, interface loads, or member/joint load-related outputs) to assess how the jacket substructure is affected? More specifically, are these outputs suitable for identifying changes in substructure fatigue or extreme-load demand, or are they mainly useful as global indicators?

2. Effect of supervisory derating on jacket fatigue demand
   Many load-mitigation studies seem to focus on the rotor, nacelle, and tower. For a fixed-bottom jacket-supported turbine, is it technically reasonable to expect that supervisory derating or soft-yaw-type strategies can also produce meaningful reductions in jacket substructure fatigue demand? Or is a reduction in upper-structure loads not necessarily representative of a reduction in support-structure fatigue?

3. Transfer of OpenFAST loads to higher-fidelity FE models
   If I want to use OpenFAST for wind-wave-current coupled simulation and then perform more detailed structural analysis in another FE software such as ANSYS, what is the recommended practice?
   In particular, how should one select and transfer OpenFAST outputs as interface loads or equivalent load time histories for a support-structure FE model?

4. Prescribing simplified operating conditions
   For some parametric studies, I would like to impose fixed operating parameters such as pitch angle, rotor speed, or yaw angle, instead of relying entirely on the closed-loop controller response. Is this possible in OpenFAST in a physically consistent way, and if so, what approach would you recommend?

5. Can OpenFAST be directly relied upon for jacket substructure (joint) fatigue studies, or should it mainly be used as a global load generator?

My overall goal is to understand whether OpenFAST can be used as a reliable first-stage tool for support-structure health/load assessment under operational mitigation strategies, and where a higher-fidelity structural model becomes necessary.

Thank you very much for your time and advice.

[jjonkman]

Dear @YuhengShey,

While I have not personally done the work you are proposing, here are my initial responses to your questions:

1. I don't see why not. Just like derating or peak shaving can reduce tower loads, so can they reduce loads in the jacket substructure and the outputs you've identified are representative.
2. Yes, as in (1).
3. Can you clarify what specifically you want to calculate in ANSYS instead of SubDyn? Are you referring to, e.g., calculations of stress at joints?
4. It is certainly easy to disable specific degrees of freedom in ElastoDyn (such as the generator speed, GenDOF = FALSE, and the nacelle-yaw DOF, YawDOF = FALSE), and to prescribe fixed values of pitch, speed, and yaw instead. But to define time-varying values of these would be difficult without active control.
5. The SubDyn module of OpenFAST will output beam-level reaction loads like shear and axial forces and bending and torsion moments. You can often use loads as a proxy for fatigue. But if you want to process fatigue in terms of stress, including stress concentration factors at joints, you'd have to post-process the loads in FEA like ANSYS, as in (3).

Best regards,