Questions about Dynamic Yaw Control with ROSCO under Time-Varying Wind

[6Mr-Y]

Hello,I am currently researching the yaw control of floating offshore wind turbines with the IEA 15MW turbine as my research object. I aim to implement dynamic yaw control under time-varying wind conditions using the ROSCO controller, but I am unsure how to achieve this. I have two main questions:
How can I generate a time-varying wind field where both wind speed and wind direction change over time? I am not sure which tools or methods to use for this purpose.
How can I implement dynamic yaw control of the turbine using the ROSCO controller? I have checked the example files provided with ROSCO but am unsure which example to refer to.
Any guidance would be greatly appreciated. Thank you very much!

[jjonkman]

Dear @6Mr-Y,

I'll respond regarding the wind input, but let others (such as @dzalkind) respond regarding ROSCO.

The synthetic turbulent wind-inflow models like Kaimal and Mann, available in TurbSim, Bladed, and HAWC wind data file formats consist of time-varying full-field turbulent inflow, but are assumed to propagate at fixed mean wind speed and direction. To create both time-varying and direction-varying wind inflow, you could use simple uniform wind data files (time varying, but uniform in space, not full-field), or you could use a high-fidelity precursor such as AMR-Wind to generate 4D wind data files (3 spatial dimensions plus 1 time dimension).

Best regards,

[6Mr-Y]

Thank you for your guidance. So, if I want to continue using TurbSim to generate wind, do I need to manually code the values of UX, UY, UZ at each time step? Thank you for your answer.

[jjonkman]

Dear @6Mr-Y,

TurbSim generates 3D wind data files (2 spatial dimensions plus 1 time dimension) and InflowWind uses Taylor's Frozen turbulence assumption to make this 4D (3 spatial dimensions plus 1 time dimension) based on a given mean wind speed and direction. So, TurbSim and InflowWind are not set up to support time-varying mean wind speed and direction. This topic has been discussed several times on our forum, e.g.:
https://forums.nlr.gov/t/turbsim-wind-direction-change/2129
https://forums.nlr.gov/t/stitch-together-turbsim-timeseries/2245
https://forums.nlr.gov/t/dynamic-wind-direction-variation-in-fast-farm-using-turbsim-or-user-defined-time-series/8365

Best regards,

[6Mr-Y]

Thank you for your guidance. So, are you recommending that I use AMR-Wind to generate the time-varying wind fields I need?
My goal is to simulate a floating wind farm under changing wind directions, where the turbines can perform dynamic yaw control in response to direction shifts. Could you confirm if AMR-Wind is the right tool for this scenario? Thank you again for your advice.

[jjonkman]

Dear @6Mr-Y,

As mentioned in the forum posts I linked to, it is common in wind turbine simulations to consider fixed mean wind speeds and wind directions within a given simulation and to consider consider other mean wind speed and other wind directions in separate simulations.

If that approach is not valid for your application, and you wish to run long simulations with time-varying mean wind speed and wind direction while including full-field turbulence, than using AMR-Wind to generate wind inflow is a possible pathway. And FAST.Farm, which uses OpenFAST to model an individual turbine, is set-up to directly support the use of 4D wind data generated through AMR-Wind.

Best regards,

[dzalkind]

I would start with the open loop control demo: https://github.com/NatLabRockies/ROSCO/blob/main/Examples/14_open_loop_control.py

It's the only example that sets up yaw control in both ROSCO and OpenFAST; there are a number of settings required to make it work.

It might also be a good proving ground for testing the feasibility of your approach, before using closed-loop methods. Similarly, you might want to consider uniform wind inputs from InflowWind, with deterministic direction changes, to assess feasibility before setting up higher fidelity options.

[dzalkind]

Yes, your understanding is correct about open loop control.

The closed loop yaw controller is set up in 33_yaw_control. You'll want to enable the platform DOFs for floating turbines.

The wind_case_fcn will create wind inputs for you, but you'll eventually want to learn how to generate your own, possibly using those functions. I suggest you review the code and gain an understanding for what's happening in the software; that will be your best guide.

[dzalkind]

Additionally, I would like to confirm: Can this yaw control example be run and implemented directly in FAST.Farm?

If you use the same settings in ROSCO/OpenFAST. We can only provide the building blocks. Good luck.

[6Mr-Y]

Thank you for your patient guidance @dzalkind! My current research focus is on controller design for floating wind farms, and I noticed that FAST.Farm provides the Super Controller module.
Therefore, I would like to confirm with you:Can I directly implement wind-farm-level closed-loop control based on 33_yaw_control.py in the future?
My goal is:Under time-varying wind directions,Upstream turbines can yaw according to the wind direction;Downstream turbines can yaw based on the wake positions from upstream turbines.Could you please tell me whether this goal can be achieved in FAST.Farm using Example 33 + ROSCO?Thank you again for your patient and professional guidance!

[dzalkind]

Therefore, I would like to confirm with you:Can I directly implement wind-farm-level closed-loop control based on 33_yaw_control.py in the future?

No farm-level control is in that example. We maintain the ZeroMQ-based interface for farm-level control: https://github.com/NatLabRockies/ROSCO/blob/main/Examples/17c_zeromq_fastfarm.py

My goal is:Under time-varying wind directions,Upstream turbines can yaw according to the wind direction;Downstream turbines can yaw based on the wake positions from upstream turbines.Could you please tell me whether this goal can be achieved in FAST.Farm using Example 33 + ROSCO?

I like to say: anything is possible in software. We have provided the building blocks, but you will need to drive the development effort. Good luck.

[6Mr-Y]

Thank you for your guidance. As a new user of FAST.Farm, I will keep learning.