Adds classes to structure wind energy data for FLORIS #775

paulf81 · 2024-01-12T21:28:32Z

@rafmudaf , @misi9170 , @ejsimley, editing this now that it is ready for review

Wind Rose Refactor

Refactor wind rose class to work with new v4 framework. Main points of inclusion in this pull request:

Split wind_rose class into two classes, WindRose and TimeSeries, within one module to avoid circular imports
Defines a super class WindData which both TimeSeries and WindRose inherit from
Provide _unpack() functions that provide the information within either class to floris' reintialize function
Provides testing for the new classes and functions in wind_rose_time_series.test.py
Ruff formant floris_interface and add capability to pass WindData objects to reinitialize and get_aep functions
Adds an example 34 showing new syntax of WindRose and TimeSeries

Context

Originally this pull request was described as addressing more capabilities for the new WindRose TimeSeries but we decided to limit this to just the the above points and upcoming issues/pull requests will address turbulence_intensity based on wind rose

paulf81 · 2024-01-13T06:01:09Z

Thanks for the chat today @misi9170 and @rafmudaf , just logging this for when we're back next week. I tried to implement the points we discussed and I added some capability to pass the objects directly to reinitialize (I think it works? There's an example now testing it). One thing though, can reinitialize accept an array of turbulence intensity? Otherwise have a good weekend!

misi9170 · 2024-01-13T18:28:05Z

One thing though, can reinitialize accept an array of turbulence intensity? Otherwise have a good weekend!

At this stage, the turbulence_intensity keyword argument to reinitialize() is intended to be scalar, I believe. However, this line broadcasts that scalar value to have size n_findex x n_turbines x 1 x 1; so, it is seems that it could be straightforward to provide a different turbulence intensity for each findex, we'd just have to make sure to pass everything through correctly.

paulf81 · 2024-01-16T13:59:55Z

One thing though, can reinitialize accept an array of turbulence intensity? Otherwise have a good weekend!

At this stage, the turbulence_intensity keyword argument to reinitialize() is intended to be scalar, I believe. However, this line broadcasts that scalar value to have size n_findex x n_turbines x 1 x 1; so, it is seems that it could be straightforward to provide a different turbulence intensity for each findex, we'd just have to make sure to pass everything through correctly.

That sounds good, I think we should move towards a number of scalar inputs being allowed to be specified for each findex, but we can discuss more

rafmudaf · 2024-01-18T23:14:12Z

Thanks @misi9170 and @paulf81, my comments are mostly addressed. Let me know when the remaining comments are addresses, and I'll resolve the review and approve.

paulf81 · 2024-01-18T23:17:10Z

Thanks @misi9170 and @paulf81, my comments are mostly addressed. Let me know when the remaining comments are addresses, and I'll resolve the review and approve.

Thank @rafmudaf !

paulf81 · 2024-01-18T23:24:28Z

@rafmudaf I think all comments now addressed, thank you!

ejsimley

See the comments I left, but this looks good overall!

ejsimley · 2024-01-19T22:38:56Z

floris/tools/floris_interface.py

@@ -838,6 +851,76 @@ def get_farm_AEP(

        return aep

+    def get_farm_AEP_with_wind_data(


Is it necessary to have a separate function for wind data? Or can wind_data just be an optional argument to get_farm_AEP?

This function is a light wrapper around the other (it calls get_farm_AEP), @misi9170 and I thought this was pretty clean, the other way works too but then you have ambiguous situations depending which values are passed in

OK, I think that makes sense. But I think this function would be a little easier to use if it didn't require wind_data to "match the wind_data object passed to reinitialize()" as mentioned in the docstring. What if you added the line self.reinitialize(wind_data) so users wouldn't have to call that separately beforehand if they wanted to update the wind conditions?

I definitely could! But I think that would then mis-match the behavior of get_AEP, which doesn't accept new wind speeds/directions, it just requires that freq have the same dimensions. The analogy here is that it doesn't use ws/wd but does again require freq to match. I think we'd have to change both functions if we make this change or they will be subtly different if that makes sense? Or I'm overthinking it...

ok, no worries! Since that's the convention in get_farm_AEP, no need to change anything.

ejsimley · 2024-01-19T22:42:56Z

floris/tools/floris_interface.py

+        # turbulence intensity using the unpack_for_reinitialize
+        # method
+        if wind_data is not None:
+            wind_directions, wind_speeds, turbulence_intensity = wind_data.unpack_for_reinitialize()


Since floris_interface is expecting a scalar turbulence intensity input, how will it handle an array provided by wind_data.unpack_for_reinitiaize()?

ejsimley · 2024-01-19T22:45:22Z

floris/tools/wind_data.py

+from floris.type_dec import NDArrayFloat
+
+
+class WindDataBase:


Minor point: should this inherit from LoggingManager? Or is there no plan to do any logging?

I actually don't know! @rafmudaf or @misi9170, should I it inherit from LoggingManager?

It wouldn't hurt to have the logging functionality available. One immediate use case might be to give some additional information when the length of the WindRose or TimeSeries don't match the length of findex.

ejsimley · 2024-01-19T22:47:03Z

floris/tools/wind_data.py

+        wind_speeds: NumPy array of wind speeds (NDArrayFloat).
+        freq_table: Frequency table for binned wind direction, wind speed
+            values (NDArrayFloat, optional).   Must have dimension
+            (n_wind_directions, n_wind_speeds).  Defaults to None.


Suggest describing what happens by default when freq_table, ti_table, and price_table are None. For example, uniform frequencies and prices are assumed.

That is a good suggestion, implementing now

ejsimley · 2024-01-19T22:50:16Z

floris/tools/wind_data.py

+        ti_table: Turbulence intensity table for binned wind direction, wind
+            speed values (NDArrayFloat, optional).   Must have dimension
+            (n_wind_directions, n_wind_speeds). Defaults to None.
+        price_table: Price table for binned binned wind direction, wind


Here and elsewhere in the module, what do you think about calling this something more generic than "price"? Maybe "value" or "value_of_power"? This could convey that the table could potentially be used to weight the power in each bin to account for a variety of things like environmental or social value, etc.

Also, it would be good to mention how this input is used, for example, that it can be used to weight the power in each bin to compute the total value of the energy produced.

Good call! Implementing now,

floris/tools/wind_data.py

ejsimley · 2024-01-19T22:56:21Z

floris/tools/wind_data.py

+            )
+
+            # Only keep the range with values in it
+            wd_edges = wd_edges[wd_edges + wd_step >= wind_directions_wrapped.min()]


Based on how pd.cut is called with right=False, I think the ">=" should be ">" here (but ">=" is correct in the next line). Is that right? I just considered the case where wd_step = 4 and the minimum wind direction is 2. Then I don't think we need the "-2" bin edge. Same comment for the wind speed case below.

You are right, that is a good catch, thank you!

paulf81 · 2024-01-20T05:15:25Z

Thank you @ejsimley and @rafmudaf for the round of comments! I think all (except for the one question for you @rafmudaf and @misi9170 on whether WindData should inherit LoggingManager), this one should be good to go? I'll hit request re-review now...

rafmudaf · 2024-01-22T20:37:06Z

floris/tools/__init__.py

-from .wind_rose import WindRose
+from .wind_data import (
+    TimeSeries,
+    WindDataBase,


I don't think users will be using this base-class, they'll probably only use the derived classes. So we can remove WindDataBase from the user-facing imports.

paulf81 · 2024-01-24T03:59:36Z

I fixed the issue that cropped up after not importing Base, let me know if ready to merge, thank you @rafmudaf !

ejsimley

Thanks for updating Paul. I just have one comment on get_farm_AEP_with_wind_data to consider.

ejsimley · 2024-01-24T18:17:10Z

floris/tools/floris_interface.py

@@ -838,6 +851,76 @@ def get_farm_AEP(

        return aep

+    def get_farm_AEP_with_wind_data(


OK, I think that makes sense. But I think this function would be a little easier to use if it didn't require wind_data to "match the wind_data object passed to reinitialize()" as mentioned in the docstring. What if you added the line self.reinitialize(wind_data) so users wouldn't have to call that separately beforehand if they wanted to update the wind conditions?

paulf81 · 2024-01-24T20:16:19Z

thanks @ejsimley ! @rafmudaf, merge city and I'm the mayor?

paulf81 added 6 commits January 11, 2024 12:58

move files around to get started

5b6abad

move files around to get started

b515274

ignore unused in __init__ files

f595720

Consolidate wind rose and time series into one module

2096d9c

Update init

9f94b01

Update tests

a9bb92c

paulf81 added floris.tools in-progress Work is actively in progress v4 Focus of FLORIS v4 labels Jan 12, 2024

paulf81 self-assigned this Jan 12, 2024

This was referenced Jan 12, 2024

Support for turbulence intensity as a wind condition #666

Closed

Feature request: Ambient TI as a function of wind direction and wind speed #713

Closed

paulf81 added 5 commits January 12, 2024 15:22

Add unpack functions

f70c411

Add grid and unpack tests

1fbb6b5

Small refactor

1c54853

Add resample function

96b477b

Test resample

d9be4ac

paulf81 mentioned this pull request Jan 12, 2024

TI as part of wind rose for farm power calculation #307

Open

paulf81 added 7 commits January 12, 2024 22:28

Add plot wind rose function

bd0e7f3

Add new wind rose usage example

c76f479

Delete old code

18a4884

add super class to import

cb3a7e1

Add a super class and inheritance

8685b5f

Add wind_data to reinitialize (also ruffing)

d087abd

Show example of reinit off wind_data objects

c75c23c

paulf81 added 2 commits January 16, 2024 13:01

Update how compute 0 freq works

d419e0d

Test computing all cases

f6df165

Add dimensions to doc string

2e765b1

paulf81 added 2 commits January 18, 2024 16:19

Add context to to_wind_rose test comments

41af717

Add error to reinitialize

1d07c57

paulf81 mentioned this pull request Jan 19, 2024

Convert turbulence intensity from single value to n_findex length array #782

Merged

6 tasks

ejsimley requested changes Jan 19, 2024

View reviewed changes

paulf81 added 4 commits January 19, 2024 22:00

Explain what happens in default cases for WindRose

2053c63

Rename price to value

126f4ad

Add check on ti and value

4dba98d

Fix bin minimum

6049894

paulf81 requested review from ejsimley and rafmudaf January 20, 2024 05:15

rafmudaf reviewed Jan 22, 2024

View reviewed changes

remove wind data import

04d4862

paulf81 requested a review from rafmudaf January 24, 2024 02:14

Import WindDataBase correctly

caa8d42

ejsimley reviewed Jan 24, 2024

View reviewed changes

ejsimley approved these changes Jan 24, 2024

View reviewed changes

Spell check and formatting

b7dd720

rafmudaf changed the title ~~Wind rose refactor~~ Adds classes to structure wind energy data for FLORIS Jan 25, 2024

rafmudaf approved these changes Jan 25, 2024

View reviewed changes

rafmudaf merged commit 420fb0f into NREL:v4 Jan 25, 2024
8 checks passed

paulf81 deleted the wind_rose_refactor branch January 25, 2024 18:04

rafmudaf mentioned this pull request Feb 26, 2024

Drive layout optimizations using WindData objects #822

Merged

misi9170 mentioned this pull request Apr 8, 2024

FLORIS v4 #870

Merged

4 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adds classes to structure wind energy data for FLORIS #775

Adds classes to structure wind energy data for FLORIS #775

paulf81 commented Jan 12, 2024 •

edited

Loading

paulf81 commented Jan 13, 2024

misi9170 commented Jan 13, 2024

paulf81 commented Jan 16, 2024

rafmudaf commented Jan 18, 2024

paulf81 commented Jan 18, 2024

paulf81 commented Jan 18, 2024

ejsimley left a comment

ejsimley Jan 19, 2024

paulf81 Jan 20, 2024

ejsimley Jan 24, 2024

paulf81 Jan 24, 2024

ejsimley Jan 24, 2024

ejsimley Jan 19, 2024

ejsimley Jan 19, 2024

paulf81 Jan 20, 2024

rafmudaf Jan 22, 2024

ejsimley Jan 19, 2024

paulf81 Jan 20, 2024

ejsimley Jan 19, 2024

paulf81 Jan 20, 2024

ejsimley Jan 19, 2024

paulf81 Jan 20, 2024

paulf81 commented Jan 20, 2024

rafmudaf Jan 22, 2024

paulf81 Jan 24, 2024

paulf81 commented Jan 24, 2024

ejsimley left a comment

ejsimley Jan 24, 2024

paulf81 commented Jan 24, 2024

		@@ -838,6 +851,76 @@ def get_farm_AEP(

		return aep

		def get_farm_AEP_with_wind_data(

Adds classes to structure wind energy data for FLORIS #775

Adds classes to structure wind energy data for FLORIS #775

Conversation

paulf81 commented Jan 12, 2024 • edited Loading

Wind Rose Refactor

Context

paulf81 commented Jan 13, 2024

misi9170 commented Jan 13, 2024

paulf81 commented Jan 16, 2024

rafmudaf commented Jan 18, 2024

paulf81 commented Jan 18, 2024

paulf81 commented Jan 18, 2024

ejsimley left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

paulf81 commented Jan 20, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

paulf81 commented Jan 24, 2024

ejsimley left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

paulf81 commented Jan 24, 2024

paulf81 commented Jan 12, 2024 •

edited

Loading