Tum misalignment model

[sTamaroTum]

TUM misalignment model

This is a proposal to include the misalignment model developed by TUM to floris v4. The methodology is based on the preprint available currently at https://wes.copernicus.org/preprints/wes-2023-133/ .
The model is coded into simulation/turbine/operation_models.py . It consists of the usual power(), thrust(), axial_induction() functions. There are two extra functions called compute_local_vertical_shear() and control_trajectory().

• compute_local_vertical_shear() is called to evaluate the vertical (linear) shear that each rotor experience, based on the inflow velocity. This allows to make the power curve asymmetric w.r.t. yaw misalignment.
• control_trajectory() determines the tip-speed-ratio and pitch angles that occur in operation. This routine assumes a standard region 2 control approach (i.e. k*rpm^2) and a region 3. Also region 2 1/2 is considered. In the future, different control strategies could be included, even user-defined.

The model requires a few new inputs that were added to the power_thrust_table in the yaml file. It also requires the cp-pitch-tsr curves for the turbine model considered. An .npz file is added to the turbine_library folder, as well as a file iea_3mw.yaml file ready for use with the model. This file refers to the IEA 3 MW reference turbine by Bortolotti et al. (2019). In the future, more wind turbine models could be added.

The model accepts power_setpoints inputs in watts to simulate derating.

The model allows also to obtain the pitch and tip speed ratio of the wind turbine. Right now the code is printing the values of tip speed ratio and pitch to screen. This could be removed for clarity, but get_pitch() and get_tsr() functions could be coded (?).

Related issue

Impacted areas of the software

simulation/turbine/operation_models.py

Additional supporting information

The model seem to work well with the examples, but it crashes at wind speeds higher than 18 m/s, because the operating tip-speed-ratio gets too low.

In the tum model, the tilt is negative when tower clearance is created. This seems conflicting with floris convention, so there is a sign change at the start of the power(), thrust() and axial_induction() functions.

Sorry if something is wrong, this is the first time I work with github.

Test results, if applicable

[rafmudaf]

Just a heads up that these comments are used to automatically generate this page in the docs: https://nrel.github.io/floris/input_reference_turbine.html. The structure (number of hashes) and location relative to the key/value pairs does matter. If you haven't it's worthwhile to build the docs locally and see the result.

[misi9170]

Oh, my bad, thanks---I'll do that and adjust as needed

[rafmudaf]

More and more, having yet another version of the inputs defined in this standalone class only for testing seems wasteful. I'm suggesting it be addressed here, but just pointing it out.

[misi9170]

Tests that are currently failing are expected; these are comparisons between the Cp/Ct data for the reference turbines and the basic power/thrust curves. We are working on a fix for this.

[misi9170]

Test are now fixed. The Cp/Ct surfaces for the reference wind turbines are not perfect matches to the reference thrust and power curves, so the tests do not attempt to assert this claim. However, the discrepancy is now clearly described in the documentation, and a warning is now raised when running with the "controller-dependent" model:

floris.logging_manager.LoggingManager WARNING Cp/Ct data provided with FLORIS is for demonstration purposes only, and may not accurately reflect the actual Cp/Ct surfaces of reference wind turbines.

[misi9170]

Thanks @sTamaroTum for the updated description---the documentation for the model is now complete.

I've uploaded a couple of screenshots of compiled documentation here, but for further details, see the Controller-dependent Model section of docs/operation_models_user.ipynb.