Robust yaw control algorithm

[prith-gs]

Hello,

I came across this paper by Andreas Rott on Robust active Wake control (https://wes.copernicus.org/articles/3/869/2018/wes-3-869-2018.pdf).

This paper presents a method for optimizing yaw control in wind farms by accounting for wind direction variability and measurement uncertainties. Using an older version of FLORIS, the study investigates how dynamic wind direction changes impact active wake deflection.

I was wondering if this can be run using the current version of FLORIS (v4.2.2) by combining the Yaw Optimization package and the FLORIS Uncertainity model?

I quickly went through the examples, but I wasn't clear if this can be done.

Looking forward to your response.

Best,
Pritheesh

[misi9170]

Hi @prith-gs ,

Yes, you can use the UncertainFlorisModel with the yaw optimization tools. Please see examples/examples_control_optimization/002_opt_yaw_single_ws_uncertain.py.

Note that this is based on the work of Simley et al. (2020) rather than Rott et al. (2018), but I understand the approaches to be similar (Simley et al. refers to Rott et al. throughout).

If you run that example, you'll see a comparison between yaw angles optimized with and without uncertainty. You can of course try different values to of wd_std to see the effects on the optimized yaw angles. Also, consider setting the maximum_yaw_angle parameter of YawOptimizationSR to something high to highlight the distinction between the uncertain optimization and the "certain" optimization (e.g. yaw_opt = YawOptimizationSR(fmodel, maximum_yaw_angle=35), yaw_opt_u = YawOptimizationSR(ufmodel, maximum_yaw_angle=35)) since the default maximum of 25 degrees limits the largest yaw angle that the certain optimization will select.

Running in that configuration, the differences in optimal yaw angles are fairly clear:

Cheers,
Misha

[prith-gs]

Hi Misha,

Thanks for the quick response.
Are there any plans to include other uncertainty models such as the model from Rott et al. (2018) in the future?

Best,
Pritheesh

[misi9170]

Hi Pritheesh,

No, not at this time. If you are finding particular needs for using other uncertainty models to achieve functionality that isn't currently implemented in FLORIS, you're of course very welcome to develop new features and submit them to FLORIS via a pull request (see our Developer guide), but we don't have plans to add other uncertainty models ourselves.

Misha