Changelog¶

v0.1.0-alpha¶

This is the initial release of foxes - ready for testing.

So far not many models have been transferred from the Fraunhofer IWES in-house predecessor flappy, they will be added in the following versions. Also optimization is not yet included. We are just getting started here!

Enjoy - we are awaiting comments and issues, thanks for testing.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.1.0

v0.1.1-alpha¶

• New code style, created by black

• Small fixes, discovered by flake8

• Tests now via pytest instead of unittest

• Introducing github workflow for automatic testing

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.1.1

v0.1.2-alpha¶

• Adding support for Python 3.9, 3.10

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.1.2

v0.1.3-alpha¶

• Adding docu: https://fraunhoferiwes.github.io/foxes.docs/index.html

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.1.3

v0.1.4-alpha¶

• Fixes

  • Static data: Adding missing data wind_rotation.nc to manifest

• Models

  • New wake model added: TurbOParkWake from Orsted

  • New turbine type added: PCtSingleFiles, reads power and thrust curves from two separate files

  • New turbulence intensity wake model added: IECTI2019/Frandsen and IECTI2005

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.1.4

v0.2.0-alpha¶

• Core

  • Introducing the concept of runners

• Opt

  • New sub package: foxes.opt, install by pip install foxes[opt]. This introduces the capability to run wind farm optimizations. As examples layout optimization for simple cases are given, see below.

• Models

  • New turbine model: Thrust2Ct, calculates ct from thrust values

  • New turbine type: NullType, a turbine type with only rotor diameter and hub height data

  • Renamed PCtSingleFiles into PCtTwoFiles

• Output

  • New output: WindRoseOutput, providing a plotly figure that shows a rose-type histogram

  • New output: AmbientWindRoseOutput, providing rose-type plotly figures of ambient data (no wake calculation involved)

• Algorithms

  • Improved Downwind: Now with option for ambient runs (no wakes)

• Utils

  • New utility: show_plotly_fig, opens a window that shows a plotly figure (instead of browser)

  • New runners: DefaultRunner, DaskRunner. The latter features parallel runs via dask

• Examples

  • Introducing two sub-folders of examples: foxes and foxes.opt

  • New example: wind_rose, calculation of wind rose states data

  • New example: layout_single_state, wind farm layout optimization for a single wind state

  • New example: layout_wind_rose, wind farm layout optimization for wind rose states

• Notebooks

  • New notebook: layout_opt.ipynb, demonstrating a simple layout optimization case

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.2.0

v0.2.1-alpha¶

• Input:

  • New input: MultiHeightStates and MultiHeightTimeseries, for input data at several height levels, e.g. WRF results at one point

• Output:

  • New output: FarmResultsEval, calculates sum, mean, min, max over states or turbines for the whole wind farm. Also calculates capacity, efficiency, yield, P75, P90.

  • New output: ResultsWriter, writes farm results or pandas data to csv file

  • Renaming: AmbientRosePlotOutput is now called StatesRosePlotOutput

• Notebooks:

  • New notebook: multi_height.ipynb, demonstrating the usage of multi-height wind input data

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.2.1

v0.2.2-alpha¶

• Bug fixes

  • Bug fixed in MultiHeightStates for wind veer cases

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.2.2

v0.2.3-alpha¶

• Input states

  • Improved: SingleStateStates now supports profiles

• Turbine models

  • New model PowerMask: Derates or boosts a turbine

• Examples

  • New: power_mask, demonstrates derating and boost

• Utils

  • New: cubic_roots, solves a cubic equation

• Output

  • New: StateTurbineMap, creates heat maps for state-turbine data

  • New: TurbineTypeCurves, creates power and thrust curve plots

  • Improved: FarmLayoutOutput now supports scatter color by variable

• Documentation

  • Adding forgotten foxes.opt to API

• Notebooks:

  • Now including results as colored layout plots

  • New notebook: wind_rose.ipynb, demonstrating how to calculate wind roses

  • New notebook: power_mask.ipynb, showing derating and boost via a power mask

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.2.3

v0.2.4-alpha¶

• Bug fixes

  • Hotfix for bug in TurbineTypeCurves output

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.2.4

v0.2.5-alpha¶

• Core:

  • New: ExtendedStates, allows for adding point data models to states

• Input:

  • Improved: FieldDataNC, now with support for spatially uniform variables

• Turbine types:

  • New: WsRho2PCtTwoFiles, air density dependent power and ct curves

• Turbine models:

  • New: SectorManagement, sets variables by range rules on other variables

• Point models:

  • New: SetUniformData, set uniform variables (optionally state dependent)

• Examples:

  • New: sector_management, demonstrates how to model wind sector management

• Notebooks:

  • New: sector_man.ipynb, demonstrates how to model wind sector management

  • New: data.ipynb, lists and shows the static data

• Bug fixes:

  • Fix for bug with option col_models in farm layout from csv

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.2.5

v0.2.6-alpha¶

• Output:

  • Improved: FarmResultsEval can now handle NaN data in Einstein sums

  • Improved: ResultsWriter can now also write turbine names

• Turbine models:

  • New: TableFactors, multiplies variables with data from a two dimensional table

  • New: YAW2YAWM, calculates yaw deltas from absolute yaw and wind direction

  • New YAWM2YAW, calculates absolute yaw from yaw deltas and wind direction

  • Generalized: kTI, now includes optional offset kb

• Turbine types:

  • Changed: Now yaw correction of P and CT switched on by default

• Point models:

  • New: TKE2TI, calculates turbulence intensity from TKE

• Wake models:

  • New: PorteAgel, calculates wakes based on the Bastankhah PorteAgel 2016 wake model

• Wake frames:

  • New: YawedWake, bends wakes in yawed conditions

• Wake superposition models:

  • Improved: LinearSuperposition, now includes options for lower/higher limits of total wake deltas

• Examples:

  • New: compare_wakes, compares wake models along horizontal lines

  • New: yawed_wake, demonstrates wake bending by yawing a rotor

• Notebooks:

• New: yawed_wake.ipynb, demonstrates wake bending by yawing a rotor

• Bug fixes:

  • Fix for bug with ExtendedStates, now it is actually working

  • Fix for bug with wake width in CrespoHernandezTIWake

  • Fix for bug with YAW and WD when using the YAWM2YAW turbine model

  • Fix for bug in TurbOPark wake model, double counting constant offset in sigma

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.2.6

v0.3.0-alpha¶

• Core:

  • Models now with simplified structure, reduces memory load during calculations

• Algorithms:

  • New: Iterative, iterative wind farm calculation until convergence has been reached

• Input:

  • Improved: FieldDataNC now accepts xarray Dataset or file pattern str as input

  • New: ShearedProfile, Vertical WS profile can be determined with shear exponent

• Output:

  • Improved: FlowPlots2D replacing horizontal/vertical –> xy, xz, yz, more intuitive

• Wake models:

  • New: TurbOParkIX, integrates wake corrected TI along centreline for wake width sigma.

• Wake frames:

  • Improved: All yawed wake frames now also support centreline data integration

• Notebooks:

  • New: overview.ipynb, summarizes the setup steps

• Bug fixes:

  • Fix for bug in TurbOPark wake model: Brackets in Eq. (4) were wrong

  • Fix for bug with long streamlines

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.3.0

v0.3.1-alpha¶

• Input states:

  • Improved: FieldDataNC now supports states subset selection

• Wake frames:

  • Improved: Streamlines now with maximal length option

• Turbine models:

  • New: Calculator, simple variable calculation via user function

• Opt:

  • Now two abstract problems in core: FarmOptProblem and FarmVarsProblem, the latter invokes the SetFarmVars turbine model

  • New opt problem: RegularLayoutOptProblem, searches for a regular grid layout

  • New opt problem: GeomRegGrids, finds regular grid layout by purely geometrical means (no wind farm calculation)

  • New opt problem: GeomLayout, turbine positioning based on xy variables, also purely geometrical

  • New opt problem: GeomLayoutGridded, a purely geometrical optimization on a gridded background

• Examples:

  • New in foxes.opt: Example layout_regular_grid, demonstrates regular grid layout optimization

  • New in foxes.opt: Example geom_reggrids, purely geometrical regular layout optimization

• Utils:

  • New functions for shape file handling: read_shp, shp2csv, read_shp_polygons

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.3.1

v0.3.2-alpha¶

• Bug fixes:

  • Fix for bug in FarmResultsEval that affected time range calculations under Windows

  • Bug fixes for FarmResultsEval with time series data

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.3.2

v0.3.3-alpha¶

• Utils:

  • Now geopandas_helpers can handle interior areas

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.3.3

v0.3.4-alpha¶

• Installation:

  • foxes now supports Python 3.11

• Input:

  • New layout input option: add_from_df, adding turbines from a pandas DataFrame

  • New interface to windIO case yaml files

• Turbine types:

  • New turbine type CpCtFile, reads CP and CT data from file or pandas DataFrame

  • New turbine type CpCtFromTwo, reads CP and CT data from two files or pandas DataFrames

  • Improved: Turbine types now calculate P_nominal as maximum, if not explicitely given

• Constants:

  • Introducing P_UNITS, used in turbine types and output evaluation

• States:

  • Bug fixed in FieldDataNC with loading multiple files

• Core:

  • Improved DataCalcModel: Now cleaner treatment of runs with progress_bar=False. Also now slimmer for distributed scheduler

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.3.4

v0.3.5-alpha¶

• General:

  • Moving identifier-type entries from foxes.variables to foxes.constants

• Models:

  • New wake superposition model ProductSuperposition, computing factorized wake deltas

  • New turbine model: RotorCentreCalc, calculates data at rotor centre, irrespective of rotor model

• Bug fixes:

  • Bug fixed that caused problems when restarting the SectorManagement turbine model, e.g. for flow plots

• Documentation:

  • Completely new style, fixing issues with incomplete API entries

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.3.5

v0.4.0¶

• Models:

  • Improved: New option to specify wake growth variable name, such that multiple kTI models could be used, resulting in different k’s for different wake models

  • New turbine model: LookupTable, interpolates data based on a multi-dimensional lookup table

• Utils:

  • Improved DaskRunner: Now supports clusters that run the Slurm queueing system

• Examples:

  • New: timeseries_slurm, shows how to run foxes on a HPC with Slurm queueing system

• Optimization:

  • Improved: foxes.opt is now able to optimize for flow variables (at selected points in space) in addition to turbine variables

• Documentation:

  • Improved API docu, now based on python-apigen

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.4.0

v0.4.1¶

• Algorithms:

  • Improved: Iterative now iterates through outer loop instead of within chunk

• Models:

  • New wake frame: Timelines, propagates wakes for spatially uniform time series

• Tests:

  • New test: consistency/iterative, checks if iterative algorithm gives same result

• Examples:

  • New example: timelines, demonstrates the usage of the Timelines wake frame

  • Improved: All examples were update for the correct usage of the DaskRunner

• Notebooks:

  • New notebook: timelines.ipynb, showing how to use the Timelines wake frame in a notebook

• Data:

  • New states data timeseries_100.csv.gz, a short timeseries with timestep 1 min, varying wind direction only

• Output:

  • Improved: FlowPlots2D now has the optional argument runner, for computing plots with the chosen parallelization settings

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.4.1

v0.4.2¶

• Bug fixes:

  • Fix of bug that slowed down TurbOParkIX by considering all TI wakes instead of only source turbine wake during integration

  • Fix of bug that prevented plotly wind roses to be shown in the documentation

  • Fix in docu that excluded the algorithm models from the API

• Output:

  • New: Animator, creates animations based on generators that yield lists of artists

• Examples:

  • Improved: timelines now includes turbine REWS signal in animations

• Notebooks:

  • Improved: timelines.ipynb now includes turbine REWS signal in animations

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.4.2

v0.4.3¶

• Output:

  • Improved: RosePlotOutput and StatesRosePlotOutput now optionally return the binned data

• Models:

  • New vertical profile: DataProfile, data based profile from file or pandas DataFrame

  • Improved ti superposition: Now supporting n-th power superposition

  • Improved wake model TurbOParkIX: New option for consideration of all wakes in ti integral

• Bug fixes:

  • Fixed bug with windio input that resulted in wrong wind rose weights

  • Fixed bug with FlowPlots2D with value bounds in contour plots

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.4.3

v0.4.4¶

• Testing automated publishing to PyPi via workflow

v0.4.5¶

• Models:

  • New rotor model: LevelRotor, calculates the REWS from different height levels

  • New turbine type: WsTI2PCtFromTwo, reads turbulence-dependent ct- and power curves

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.4.5

v0.5¶

• Dependencies:

  • Dropping support for Python 3.7, due to netcdf4 dependency

• Core:

  • Simplified model initialization/finalization. Removing the keep_models idea, instead all models are now kept in the algorithm’s idata memory until finalization. Adding a new model now mostly requires that the sub_models and the load_data functions are overloaded, if applicable. The initialize and finalize only need to be addressed explicitly in non-standard cases.

• Algorithms:

  • New algorithm: Sequential, step wise evaluation of states for simulation environments that do not support chunking

  • Improved: Iterative now supports under-relaxation of parameters

• Output:

  • New sub package: output.flow_plots_2d is now a package instead of a module

• Models:

  • This version introduces induction models for modelling blockage effects.

  • New induction model: RHB, the classic Rankine-half-body model.

• Examples:

  • New example: sequential, demonstrating the usage of the sequential algorithm

  • New example: induction_RHB, showing the Rankine-half-body model for blockage

• Notebooks:

  • New notebook: blockage.ipynb, demonstrating how to apply the RHB induction model to a wind farm

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.5

v0.5.1¶

• Dependencies:

  • The windio package is now an optional dependency, to be installed by hand if needed. This is due to windio being not available at conda-forge

• Notebooks:

  • New notebook: sequential.ipynb, creating an animation showing state-by-state wake propagation

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.5.1

v0.5.2¶

• Algorithms:

  • Improved Downwind: Now optional arguments that allow calculation of subsets in calc_farm and calc_points, e.g. a selection of states

• Output:

  • New output: PointCalculator, a wrapper for the calc_points function of the algorithm

  • New output: SliceData, creating arrays, DataFrames or Datasets from 2D slices

  • Improved FlowPlots2D: Completely refactored, data based on SliceData. New option for creating figures for subsets of states only.

• Vertical profiles:

  • Now optional support for ustar input data instead of reference data

• Turbine types:

  • Fix for Cp-based models with interpolation within sparse input data

• Wake models:

  • Renaming: BastankhahWake class now called Bastankhah2014. Models in the model book are also renamed from Bastankhah to Bastankhah2014

  • Renaming: PorteAgelWake class now called Bastankhah2016. Models in the model book are also renamed from PorteAgel to Bastankhah2016

  • Renaming: RHB class now called RankineHalfBody

  • Fix: RankineHalfBody no longer shows a jump at the rotor disc, but a small region of constant deficit instead

  • New default values: Bastankhah2014 now has default value sbeta_factor=0.2 (previously 0.25). Models with the previous value are available in the model book as Bastankhah025 etc.

• Wake superpositions:

  • Restructured: Now simplified classes for WS or TI superposition only (less general but simpler), e.g. WSLinear or TIQuadratic, etc. Also in the model book the models are now called ws_linear or ti_quadratic, etc.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.5.2

v0.5.2.1¶

• Bug fixes:

  • Bugs fixed that messed up the colorbar and the title in animations

• Notebooks:

  • Improved animations in timelines.ipynb and sequential.ipynb

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.5.2.1

v0.6¶

• Dependencies:

  • Replacing dependency on plotly by dependency on windrose, since the latter is lighter

• Core:

  • This version introduces AxialInductionModel classes, computing the axial induction factor a(ct)

• Wake models:

  • Reformulating beta in terms of induction in Bastankhah2014 and CrespoHernandez

  • New default: Bastankhah2014, Bastankhah2016, TurbOPark and TurbOParkIX now with default axial induction model Madsen

  • New induction wake models: Rathmann, SelfSimilar and SelfSimilar2020, for blockage modelling

  • Introducing WakeMirror wrapper around wake models, modelling wake reflection from ground or horizontal plane via virtual mirrored turbines

• Wake frames:

  • Renaming Streamlines to Streamlines2D, no changes in model book names

• Axial induction models:

  • New induction model: BetzAxialInduction, the classic a = 0.5(1 - sqrt(1-ct)) relation. In the model book this is called Betz.

  • New induction model: MadsenAxialInduction, a third-order polynomial approximation of a(ct). In the model book this is called Madsen.

• Output:

  • Improved: FlowPlots2D now includes an option for indicating the rotor disk by a colored line

  • Improved: RosePlotOutput no longer depends on plotly, but on the new utility TabWindroseAxes

• Utils:

  • New: TabWindroseAxes, a derivative of windrose.WindroseAxes for input data that is based on bins with weights (and not timeseries)

• Notebooks:

  • New: blockage_comparison.ipynb, comparing four turbine induction models

• Bug fixes:

  • Fix for bug in Streamlines2D when used in combination with WakeMirror

• Tests:

  • Fresh flappy v0.6.2 test data for all Bastankhah and CrespoHernandez wakes, also without the sbeta limitation

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.6

v0.6.1¶

• Input:

  • New ambient states: TabStates, single tab-file input

• Data

  • New static data: winds100.tab, an example tab file

• Examples:

  • New example: tab_file, demonstrating the usage of the TabStates

• Bug fixes:

  • Bug fixed for RankineHalfBody turbine induction model that produced wrong results for wind directions unequal 270 degrees

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.6.1

v0.6.2¶

• Models:

  • New partial wakes model: PartialCentre, evaluating wake deltas at the rotor centre point only

• Inputs:

  • New farm layout input: add_random, adds turbines at random positions, respecting a minimal distance

  • New states creation: random_timseries_data, creates uniform random timeseries data

• Utils:

  • New function random_xy_square, generates random xy positions with minimal distance

• Examples:

  • New example: random_timeseries, computes a random farm in a random timeseries. Both sizes are defined by user input

• Bug fixes:

  • Fix for bug in gen_states_fig_xz and gen_states_fig_xz with parameter x_direction, which had no effect on the image

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.6.2

v0.7¶

• Python versions:

  • Now supporting Python 3.12

• Documentation:

  • New page Models, summarizing and explaining the foxes model choices.

• Core:

  • Speed-up in comparison with version 0.6.x, by internally handling all turbine data in downwind order, and then translating it back to farm order once computations are complete.

  • Internally, all point evaluation data is now translated into so called “target” data, where each target is understood as being composed of a certain number of target points. During wake computations, these are the points per rotor (as defined by partial wakes models). For computations at user given points, the points are interpreted as targets with a single target point each. Final point output data is then given to the user again with point index coodinates, i.e., in the same format as in previous versions.

  • New data classes: MData, FData, TData, all derived from the foxes Data class. These specialize model, farm and target data, respectively, during model calculations.

• Algorithms:

  • All algorithm constructors now take farm, states, wake_models as the first three arguments. If no model book is given, the default ModelBook() will be used.

  • Partial wakes are now chosen either

    • by a dictionary, which maps wake model names to model choices (or default choices, if not found),

    • or by a list, where the mapping to the wake models is in order of appearance,

    • or by a string, in which case all models are either mapped to the given model, or, if that fails with TypeError, to their defaults,

    • or by None, which means all models are mapped to the default choice.

• Partial wakes:

  • New PartialSegregated abstract model, from which the PartialGrid model is derived. Segregated models now average background results and wake deltas separatly, and then add the results. Notice that with the choice of RotorPoints partial wakes, the mathematically correct average over a discretized rotor is calculated. This is more accurate, but it may be slower than some models (e.g. for PartialAxiWake models) or not applicable for some rotor choices (e.g. the LevelRotor, where a wake average makes no sense).

• Outputs:

  • New output RotorPointPlot, creating rotor point plots.

• Notebooks:

  • New notebook rotor_models.ipynb, visualizing rotor points.

  • New notebook partial_wakes.ipynb, verifying partial wakes models.

• Bug fixes:

  • Various fixes here and there.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.7

v0.7.1¶

• Models:

  • Improved ModelBook, adding some smartness thanks to model factories. Parameters are parsed from model name strings that follow given patterns. E.g., Jensen_<superposition>_k<k> represents all JensenWake models with any superposition model choice and and choice of k, and Jensen_linear_k0075 would be an accepted name choice by the user.

  • New wind speed superpositions WSLinearLocal, WSQuadraticLocal, WSPowLocal, WSMaxLocal: Adding dimensionless wind deficits, and then evaluating the overall effect for the ambient results at evaluation points (no scaling with rotor effective data)

• Utils:

  • New utility Factory, creating class instances from selections of allowed parameter choises

  • New utility FDict, a dictionary that supports factories

• Bug fixes:

  • Bug fixed with TurbOParkIX, that prevented it from running

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.7.1

v0.7.2¶

• Bug fixes:

  • Fix for bug with Factory that confused templates A_B<..> type with B<..> type templates

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.7.2

v0.7.3¶

• Dependencies:

  • Restriction to numpy<2 due to an incompatibility of dependecy autograd

• Core:

  • New model type GroundModel, which manages the call of the partial wakes models in case of farm calculations and wake models in case of point calculations

• Inputs:

  • New height dependent states: MultiHeightNCStates and MultiHeightNCTimeseries, same as MultiHeightStates and MultiHeightTimeseries, but starting from a netCDF file or xarray.Dataset input

• Models:

  • New wake model helper class WakeK, handling the k, ka, kb choices for all wake models

  • New ground model NoGround, plain call of the partial wakes and wakes models

  • New ground models WakeMirror and GroundMirror, replacing the equivalent former wake models. Realizing wake reflection at horizontal planes.

  • New induction model VortexSheet, which is a radial implementation of the centreline deficit model in Medici 2012 https://doi.org/10.1002/we.451

• Utils:

  • New utility WakeKFactory class, enabling the choice of either k or ka, kb directly from the wake model names in the model book.

• Inputs:

  • Work on windio, but unfinished on their side when it comes to analysis requests

• Examples:

  • Example multi_height: Now based on MultiHeightNCTimeseries

• Bug fixes:

  • Fox for bug with wake mirrors and partial wakes

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.7.3

v0.7.4¶

• Input:

  • Adding output options to windio

• Output:

  • New output StateTurbineTable, exporting state-turbine data to NetCDF

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.7.4

v0.8¶

Starting with this version, the previous foxes.opt sub-package is is now its own package at github, pypi and conda-forge: foxes-opt, with its own documentation.

If you are planning to run wind farm optimizations, install it via

pip install foxes[opt]

or

pip install foxes-opt

or

conda install foxes-opt -c conda-forge

Compared to older versions, replace foxes.opt by foxes_opt in all your scripts - then everything should just run as before.

If you are not running any optimizations, just don’t do any of the above and enjoy the lighter version with less dependencies.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.8

v0.8.1¶

• Updated requirements.txt

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.8.1

v0.8.2¶

• Removing plotly_helpers.py from utils

• Updating dependencies

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.8.2

v0.8.3¶

• Outputs:

  • Improved SliceData output: Now either specify resolution or n_img_points, e.g. n_img_points=(100, 100) for an image with 100 x 100 points

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v0.8.3

v1.0¶

This major version introduces the concept of Engines which handle the chunking and parallelization of all foxes calculations. The default choice now prefers the concurrent.futures package and provides a significant speedup compared to previous versions. See the documentation for more details and all engine choices. The Engines replace the Runners of previous versions.

• Engines

  • New engine ThreadsEngine (short threads): Sends chunks to threads, based on concurrent.futures

  • New engine ProcessEngine (short process): Sends chunks to processes, based on concurrent.futures

  • New engine MultiprocessEngine (short multiprocess): Sends chunks to a multiprocessing pool

  • New engine XArrayEngine (short xarray): Runs parallelization via xarray.apply_ufunc

  • New engine DaskEngine (short dask): Submits chunk calculation functions to dask

  • New engine LocalClusterEngine (short local_cluster): Creates a virtual cluster on the local machine

  • New engine MPIEngine (short mpi): Parallel computation via mpi4py, also multi-node

  • New engine SlurmClusterEngine (short slurm_cluster): Submits jobs to a SLURM system

  • New engine NumpyEngine (short numpy): Runs a loop over chunks

  • New engine SingleChunkEngine (short single): Runs single-chunk calculations

  • New engine DefaultEngine (short default): Switches between single and process, depending on the case size

• Inputs:

  • New states OnePointFlowStates, OnePointFlowTimeseries, OnePointFlowMultiHeightTimeseries, OnePointFlowMultiHeightNCTimeseries: Generating horizontally inhomogeneous inflow from horizontally homogeneous input data

  • New farm layout option: add_ring, adding a ring of turbines

• Models:

  • Wake frame Timelines now also accept spatially uniform multi-height states

  • New wake frame DynamicWakes: Dynamic wakes for any kind of timeseries states, compatible with chunking

  • New turbine type FromLookupTable, computes power and thrust coefficient from a lookup table

• Outputs:

  • New sub package seq_plugins, in case more of these will be added in the future

  • New sequential plugin SeqWakeDebugPlugin, adding wake centres and velocity vectors to flow animations, for debugging

• Examples:

  • New example: dyn_wakes, similar to timelines but with dynamic wakes and OnePointFlowTimeseries inflow

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.0

v1.1¶

• Python versions:

  • Adding support for Python3.13

• Engines:

  • Default number of processes is now os.cpu_count(), or os.process_cpu_count() for python3.13

  • New engine: RayEngine (short name ray), runs parallel computations based on the Ray package

• Inputs:

  • New states SliceDataNC, much like FieldDataNC but without height dependency

• Models:

  • New turbine type TBLFile: Reads power, ct, D, H, P_rated from a *.tbl file

  • Turbine induction models SelfSimilar, SelfSimilar2020, Rathmann, VortexSheet now optionally accept any wind speed superposition model, i.e., they are no longer based on hard-coded linear superposition

• Bug fixes:

  • Bug fixed in WSProduct, causing zero wind speed at regions not touched by wakes

  • Bugs fixes in FarmLayoutOutput, concerning the writing of the layout csv file

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.1

v1.1.1¶

• Dependencies:

  • Removing optional dependencies io. The installation advice will be printed when trying to use the foxes.input.windio sub package. The reason is that for now this depends on a personal fork by kilojoules, and the inclusion of such dependencies is not supported by PyPi.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.1.1

v1.2¶

• Command line applications:

  • New command line application: foxes_yaml, runs foxes by interpreting a yaml input parameter file (no script neccessary)

  • New command line application: foxes_windio, runs foxes by interpreting a windio yaml input parameter file (no script neccessary)

• Config:

  • Introducing a dictionary-like configuration object foxes.config, setting default data types, the directory choices, etc.

• Inputs:

  • New input package foxes.input.yaml, realizing the run via yaml parameter files. This is not following windio, but a foxes specific structure.

  • Reworked windio input package, now at foxes.input.yaml.windio and based on foxes.input.yaml

• Outputs:

  • Introducing the idea of a default output directory, accessible via foxes.config.out_dir. The file writing output classes now refer to this, or a output specific output directory that is given as an argument to the constructor of the output class.

• Examples:

  • New example yaml_input, demonstrating the command line application foxes_yaml and the new yaml parameter file

• Bug fixes:

  • Fix for TurbOParkIX wake model: Now k is truly k, such that only ambka or ka choices from WakeK correspond to multiplication with AMB_TI or TI, respectively.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.2

v1.2.1¶

• Dependencies:

  • Removing dependency on windrose package

• Outputs:

  • Output FarmResultsEval: Renaming contraction rule mean into weights. Introducing new rule mean_no_weights.

  • New Output WindRoseBinPlot, for example showing variable mean values for wind rose bins

  • Re-implementation of WindRosePlot, no longer needs the windrose package

• Inputs:

  • Generalizing ScanWS states to ScanStates, for arbitrary tensor products of variable lists

  • States FieldDataNC now also accepts fields without heights, signature (state, x, y) and (state, y, x)

• Bug fixes:

  • Fixing some examples that were computing the turbine efficiency incorrectly

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.2.1

v1.2.2¶

• Developments required by foxes_opt v0.3, related to running also wind farm optimizations by yaml parameter files

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.2.2

v1.2.3¶

• Dependencies:

  • Switching windIO support to fork at https://github.com/EUFLOW/windIO

  • Removing dependency netcdf4, since this or h5netcdf are now a dynamic choice, called when NetCDF files are being used

• Config:

  • Now three directories are defined: work_dir, input_dir, output_dir. If not defined, falling back to work_dir, defaulting to cwd.

• Inputs:

  • Removing SliceDataNC, since now FieldDataNC can handle data without height dependency

  • Enhancing FieldDataNC for more efficient parallel loading of large data, replacing pre_load flag with load mode options preload, lazy, fly

  • Enhancing windio inputs for output generation

• Outputs:

  • New output SlicesData which combines slices from SliceData into a single NetCDF file, resulting in full 3d field output (or 4d, with time/states dimension)

• Models:

  • New point model Ustar2TI, estimates TI from friction velocity

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.2.3

v1.2.4¶

• Config:

  • Fix for bug with nc_engine for Python 3.8

• Utils:

  • New util ReaderWRG, reading WRG files. Thanks @HassanKassem

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.2.4

v1.2.5¶

• Models:

  • Fix for bug in VortexSheet turbine induction wake model. Thanks @MellBeli

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.2.5

v1.3¶

• Core:

  • The statistical weights of the states are now no longer required to be known a-priori, but they are computed at target points during the states calculation. So in principle, weights can now depend on the evaluation point, as it is required by the new WRGStates, see below.

• Input:

  • Introducing WRGStates, reading and running Wind Resource Grid (WRG) data files which describe Weibull factors on a regular horizontal grid.

• Examples:

  • New example quickstart, demonstrating foxes in a minimal script

• Bug fixes:

  • Bug fixed with xarray.Dataset input instead of file for FieldDataNC states

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.3

v1.4¶

• Core:

  • This version introduces WakeDeflection models. This replaces and generalizes the YawedWakes wake frame, such that now any choice of wake frame can be combined with yawed wakes, for example streamlines or dynamic wakes. The wake deflection model is selected through the parameter wake_deflection or the algorithm and then applies to all wakes.

  • Wake models, wake deflections or wake superpositions can now induce a horizontal rotation of the waked wind vector. This is realized by inhternally switching from a (WS, WD) description to a UV vector, and vice-versa at the end of calculations.

  • A new abstract class WindVectorWakeSuperposition was instroduced for the purpose of handling UV wind wake effects.

• Inputs:

  • New states WeibullSectors, reading and interpreting spatially homogeneous Weibull scale and shape parameters for wind direction sectors, from csv or nc files

• Models:

  • New wake deflection model Bastankhah2016Deflection, replacing the previous YawedWakes wake frame. Notice that this model does not modify the wind speed deficit or the wind direction in the wake, it only changes the wake path. Meant to be used together with the Bastankhah2016 wake model.

  • New wake deflection model JimenezDeflection, which bends the wake and optionally also rotates the waked wind vector along the path. If that option is chosen, this model requires a vector type superposition of the wind deficit.

  • New wake deflection model NoDeflection, which switches off wake bending and should be used if no yaw misalignment occurs in the simulation (default).

  • New wake superposition model WindVectorLinear, realizing linear vector type wind vector superposition.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.4

v1.5¶

• Dependencies:

  • Introducing minimal package versions of dependencies to pyproject.toml

• Inputs:

  • New abstract states class DatasetStates, serving as a common base for NetCDF type input data

  • Renamed FieldDataNC as FieldData, now with improved data handling

  • New states class PointCloudData, interpolating data with irregular point cloud support

  • New states class WeibullField, representing Weibull sector data on a regular grid

  • New states class WeibullPointCloud, representing Weibull sector data with irregular point cloud support

  • New farm layout input function add_from_arrays, for directly adding turbines through one dimensional data lists or arrays

  • Advancing windio to version 2.0

• Models:

  • New turbine type CalculatorType, for direct infusion of farm data at a turbine through a user function

  • New rotor model DirectMDataInfusion, piping stored model data directly into turbines

  • Generalizing the Calculator turbine model, now additionally passing algo, mdata, fdata to the user function

• Examples:

  • New example point_cloud, demonstrating the useage of the PointCloud ambient states class

  • New example weibull_grid, demonstrating the useage of the WeibullField ambient states class

  • New example weibull_cloud, demonstrating the useage of the WeibullPointCloud ambient states class

  • New example direct_mdata_infusion, demonstrating how to infuse states data dirctly into rotor data without interpolation

• Bug fixes:

  • Bug fixed with turbine model SetFarmVars that mixed up the turbine ordering

  • Bug fixed with turbine models that are marked as pre_rotor that mixed up the turbine ordering

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.5

v1.5.1¶

• Examples:

  • Now all examples are making use of an engine via the with context, as it should be done in scripts

• Bug fixes:

• Bug fixed that caused a reset in the dyn_wakes wake frame when calling calc_points

• Bug fixed with write_nc utility, concerning string type coordinate arrays

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.5.1

v1.5.2¶

• Models:

  • New farm controller OpFlagController, switching off turbines based on state-time dependent boolean input data

  • Wake blockage model SelfSimilar: Promoting alpha and beta to constructor arguments

• Tests:

  • New test test_set_farm_vars, checking if correct turbine ordering is maintained

• Notebooks:

  • New notebook operating_flag.ipynb, demonstrating the usage of the OpFlagController class

• Bug fixes:

  • Bug fixed with negative beta in sqrt for some wake models, Madsen axial induction and high ct values

  • Bug fixed with wake model bastankhah16 to set default value beta to 0.077 (was 0.07)

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.5.2

v1.6¶

• Python version:

  • Adding support for Python 3.14

• Core:

  • Wind farms now support turbine locations in terms of longitude, lattitude. They will automatically be translated into a uniquely fixed UTM zone, stored in the global config object.

  • Turbines now carry an optional property wind_farm_name, intended for cases which model multiple wind farms. The WindFarm class offers a property wind_farm_names and a function get_wind_farm_mapping(), returning a mapping from wind farm names to turbine indices. This is for example used by layout plots with the choice annotate=3, which results in printing wind farm names instead of turbine indices.

• Algorithms:

  • New optional dict-type parameter write_nc for algo.calc_farm() and algo.calc_points(), allowing the parallel writing of results into multiple files without ever constructing the complete dataset in memory. If writing results to file is the only intention of the run, this is the recommended approach.

  • Introducing new parameter population_params to Downwind algorithm, for setting up a vectorized parameter study. This is based on a set of values for selected variables, which are then evaluated for each state. This is realized via the algorithm specific states class PopulationStates and turbine model class PopulationModel.

  • Reworking the Iterative algorithm, such that it now computes only non-converged states in each iteration. In general, this leads to a substantial speed-up.

  • Introducing new boolean parameter clear_mem to algo.calc_farm(), clearing memory after initiating the calculation. This is intended to be used when there is only one single call of that function, or any of the used models.

• Engines:

  • Now all engines have submit and await_result functions (which are trivial for serial engines), also usable by the user from the engine object, e.g. for parallel plotting during post-processing.

  • Adding support for parallel results writing via write_nc parameter, optionally without ever creating the full dataset in memory. Independent of writing, the futures of individual chunks are now computed directly when needed, no longer in advance.

  • Engines have a new boolean parameter progress_bar, and if set to False, they print the progress to terminal instead. This is recommended for runs on clusters.

  • Dropping XArrayEngine, since no longer compatible

• Inputs:

  • New farm layout input from_eww, reading wind farm input in EuroWindWakes format

  • New farm layout input from_wrf, reading a folder that contains wind farm input data in a format readable by WRF

  • New states NEWAStates, directly reading NEWA-WRF data files into foxes

• Models:

  • Addition of IEA37Gaussian wake model to the model book, requiring slight generalization of the Bastankhah2014 wake model class.

  • Adding default parameter wake models to the model book, following the papers: Bastankhah2014, TurbOPark.

• Utils:

  • Improved write_nc utility, now better data compression through the usage of appropriate encoding

• Examples:

  • New example parameter_study, demonstrating the setup of a parameter study

• Bug fixes:

  • Fix for max x selection in FieldDataNC, thanks @qzheng75

  • Fix for Sequential algorithm with non-updated chunk store data

  • Fix for TurbOPark wake model in the context of windio input files

  • Fix for bug with writing time stamps to NetCDF in write_nc

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.6

v1.6.1¶

• Input:

  • All NetCDF based states now have improved performance in speed an memory, thanks to reworking the treatment of states during spatial interpolation

• Engines:

  • Engines can now optionally write flow animations of individual chunks to file during calculations. This is triggered by the new write_chunk_ani parameter of the calc_farm and calc_points functions.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.6.1

v1.6.2¶

• Engines:

  • Reworked engine implementations, in particular pool engines. This concerns progress display and a smoother balance between submissions and post-processing chunk results.

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.6.2

v1.7.0¶

• Documentation:

  • The docu has moved to a new location: https://fraunhoferiwes.github.io/foxes/index.html

• Python version:

  • Removing support for Python 3.14, will be re-added at a later stage

• Engines:

  • All engines are now required to be used though a with block. If algo.calc_farm() or algo.calc_points() are called outside such a block, the DefaultEngine will be created temporarily and the with block will be added in the background.

• Models:

  • Improving the Streamlines2D wake frame, now substantially faster and lighter on memory

• Input:

  • Aligning the windio input with WIFA conventions for turbine_results.nc output writing, now (turbine, time)

• Output:

  • Image generating outputs like FlowPlots2D now separate plot data computation from figure generation. This reasoning is that matplotlib.pyplot is not thread-safe, and now all data can first be computed in parallel before using it for graphics after exiting the engine context.

• Utils:

  • New function eval_dict_values, which recursively parses string dictionary values like e.g. np.arange(100) to data, applied in Dict.eval()

• Bug fixes:

  • Fixed a bug with the Sequential algorithm in combination with OnePointFlowStates

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.7.0

v1.7.1¶

• Bug fixes

  • Fixes for bugs with foxes-opt

Full Changelog: https://github.com/FraunhoferIWES/foxes/commits/v1.7.1