Installation issue

[Giorgio-C-30]

Hi all,

I have a very easy installation problem. However I am an amateur and I can not solve it.
I have installed floris following the guide with the command "pip install floris". However, when I am trying to run the first script
"from floris import FlorisModel
fmodel = FlorisModel("gch.yaml") "

I get this error: " No such file or directory: 'gch.yaml' ".
Verifying on the lib folder, I see that only the core folder of floris has been installed, instead of the whole github main folder. Hence the examples folder is missing. I have tried to copy and paste it, but it is not working.

Any suggestion is very welcomed.
Thank you in advance.
Giorgio

[Bartdoekemeijer]

Hey Giorgio,

If you installed FLORIS through a pip install floris, then indeed you may not have the example scripts and files on your system. You can find them here: https://github.com/NREL/floris/tree/main/examples.

You get the error No such file or directory because, indeed, you do not have the file gch.yaml on your machine (or at least, not in the folder that you are trying to run the Python script out of). You can download gch.yaml here: https://github.com/NREL/floris/tree/main/examples/inputs.

Download the gch.yaml file and place it in the folder next to your Python script. It should work now.

[Giorgio-C-30]

Thank you so much for the quick answer.
I have copied it in my folder next to my Python script, but i get this error now:

"The filename, directory name, or volume label syntax is incorrect: attrs generated init floris.core.farm.Farm ".
Any suggestions?

Many thanks,
Giorgio

[Bartdoekemeijer]

Hi Giorgio,

See #951 for a more elaborate discussion and a solution to the <attrs generated init floris.core.farm.Farm> error.

Cheers,
Bart

[javadi97]

hello friends
I have data for a wind farm with six 2 MW turbines, but there is no 2 MW turbine in the Floris library.
Is there a way I can redefine this turbine for Floris?
I want to compare the data obtained from Floris in the field of wake steering with the real wind farm data

[misi9170]

Hi @javadi97 , do you have power and thrust curves for your 2MW turbines, and also information on the hub height, rotor diameter, etc? If so, you can generate dictionary representations for them using build_cosine_loss_turbine_dict(), and then pass the created dictionaries into the turbine_type argument of FlorisModel.set()

[misi9170]

examples/examples_turbines/003_specify_turbine_power_curve.py demonstrates this functionality.

[noorhj]

hello hello friends, "I am working on a project where I plan to use the RL algorithm to optimize wake steering in a wind farm, and I am using FLORIS for the modeling. I am trying to determine the appropriate temporal resolution for the wind condition data. Would hourly data over the course of a year be sufficient, or would it be better to use ten-minute data for the same period? What is the recommended time interval for accurate wake modeling and optimization in FLORIS?"

[misi9170]

Hi @noorhj,

Thanks for the question--the answer is a bit nuanced. As you know, FLORIS is a steady state model, so does not inherently have a temporal resolution. However, it is also useful to think of FLORIS as modeling the time-averaged flow---but the question remains, what is the time window that the averaging is working over?

Unfortunately, there is not a straightforward answer. We often compare FLORIS to SCADA records from wind farms with ten-minute averaging. This may be more appropriate for small farms, where ten minutes is in the ballpark of the time-scales of the propagation of wind through the farm (for example, if the wind speed is 8 m/s and the wind farm is 2km wide, then the wind may be expected to "traverse" the farm in around 4 minutes). For much larger farms, longer averaging times may be more appropriate.

We also often use the UncertainFlorisModel, rather than the FlorisModel, when we want to consider time averaging. This allows the user to specify standard deviations in the wind direction that may be expected to occur over the averaging window. 3 degrees of standard deviation is usually a good starting point for 10-minute averaging, and I would expect a value slightly higher than that to be best for an hour (maybe 4 or 5 degrees).

In the end, I think you could use either hourly or ten-minute data. If you want to know more about the UncertainFlorisModel, you can take a look at these examples.

I hope this helps!

Misha