Ard integration for wind farm combined performance and cost model #312

jaredthomas68 · 2025-10-22T17:41:25Z

Ard integration for wind farm combined performance and cost model

This PR adds Ard as an available wind farm performance and cost model. Ard wraps many wind farm modeling capabilities including aerodynamics (FLORIS), cable array design (OptiWindNet), and cost/finance (WISDEM). Ard is built specifically for optimization, and as such includes common constraints, like arbitrary polygonal boundaries, the ability to use multiple discrete boundaries, and multiple parameterization approaches, like ordered grids and continuous location definitions.

The current (first iteration) wrapper only allows for ordered grid layouts.

Type of Contribution

General PR Checklist

CHANGELOG.md has been updated to describe the changes made in this PR
Documentation
- Docstrings are up-to-date
- Related docs/ files are up-to-date, or added when necessary
- Documentation has been rebuilt successfully
- Examples have been updated (if applicable)
Tests pass (If not, and this is expected, please elaborate in the tests section)
Added tests for new functionality or bug fixes
PR description thoroughly describes the new feature, bug fix, etc.

New Technology Checklist

Related issues

This PR would be enhanced by allowing more resource type integrations, such as detailed in #237

Impacted areas of the software

CHANGELOG.md: note inclusion of Ard
docs/_toc.yml: include Ard docs in TOC
docs/technology_models/technology_overview.md: include Ard in overview
docs/technology_models/wind_plant_ard.md: high-level documentation/introduction to Ard
examples/xx_wind_ard/ard_inputs/NREL_Reference_5MW_126.csv: wind turbine definition
examples/xx_wind_ard/ard_inputs/open-meteo-56.20N8.54E86m-short.yaml: abbreviated wind resource input file for Ard
examples/xx_wind_ard/ard_inputs/open-meteo-56.20N8.54E86m.yaml: wind resource input file for Ard
examples/xx_wind_ard/ard_inputs/power_thrust_table_ccblade_NREL-5p0-126-RWT.csv: wind turbine performance input for Ard
examples/xx_wind_ard/ard_inputs/windIO-plant_turbine_NREL-5.0MW-126m-RWT.yaml: wind turbine input definition for Ard
examples/xx_wind_ard/ard_inputs/windio.yaml: WindIO input file for Ard
examples/xx_wind_ard/h2i_inputs/driver_config.yaml: H2I driver input yaml for example
examples/xx_wind_ard/h2i_inputs/plant_config.yaml: H2I plant input yaml for example
examples/xx_wind_ard/h2i_inputs/tech_config.yaml: H2I tech input yaml for example
examples/xx_wind_ard/h2i_inputs/wind_pv_battery.yaml: top-level yaml input to run script for example
examples/xx_wind_ard/run_windard_pv_battery.py: main example run script
h2integrate/converters/wind/wind_plant_ard.py
- WindArdCostComponent: mostly empty child of CostModelBaseClass to allow H2I access to cost_year parameter
- WindPlantArdModelConfig: defines inputs to Ard (ard_system, and ard_data_path)
- ArdWindPlantModel: OpenMDAO Group that includes and Ard OpenMDAO problem object as an OpenMDAO SubmodelComp.
h2integrate/core/h2integrate_model.py: include wind_plant_ard in combined_performance_and_cost_models
h2integrate/core/supported_models.py: include ArdWindPlantModel as "wind_plant_ard"
pyproject.toml: include Ard as a dependency
tests/h2integrate/test_all_examples.py
- test_windard_pv_battery_dispatch_example: test Ard integration with H2I
path/to/file.extension
- method1: What and why something was changed in one sentence or less.

Additional supporting information

The location used in the example is specific to IEA Task 50 WP 2. The current solar resource is not from the correct location.

Test results, if applicable

kbrunik

This is super exciting--our first international example and some really cool functionality with the full-stack integration! Thanks for the great PR @jaredthomas68! Only a few small things from my perspective.

kbrunik · 2025-10-26T18:47:53Z

docs/technology_models/wind_plant_ard.md

+The intent of [Ard](https://github.com/WISDEM/Ard) is to be a modular, full-stack multi-disciplinary optimization tool for wind farms. By incorporating Ard in H2Integrate, we are able to draw on many wind technology models developed at NREL and other institutions without managing them or their connections in Ard. Models connected in Ard include many parts of [WISDEM](https://github.com/WISDEM/WISDEM), [FLORIS](https://github.com/NREL/floris), and [OptiWindNet](https://github.com/DTUWindEnergy/OptiWindNet). Ard also provides constraint functions and wind farm layout generation capabilities among other things. Because Ard has been developed in a modular way, you may extend Ard fairly easily to include other wind models of interest.
+
+Ard is included in H2Integrate as an [OpenMDAO sub-model](https://openmdao.org/newdocs/versions/latest/features/building_blocks/components/submodel_comp.html), which means that Ard is treated as an OpenMDAO system within an OpenMDAO system. In this way, the user can run an independent wind farm optimization within Ard, or allow H2Integrate to manage the wind farm design variables directly. The drawback of including Ard as a sub-model is that N2 diagrams made from the H2Integrate problems will show Ard only as a single model, rather than showing all the subsystems within Ard. IF you wish to view an N2 diagram of Ard, you will need to use the Ard problem instead.
+


Would it be possible to show how a user would create an Ard problem instead of just the H2Integrate problem?

kbrunik · 2025-10-26T18:52:06Z

docs/technology_models/wind_plant_ard.md

+        - FinanceSE
+
+## Wind Resource
+The wind resource capabilities of H2Integrate are not yet connected with Ard, so the user must provide a wind resource file directly to the Ard model inputs.


Is this because of how the wind resource needs to be formatted? If yes, could you explain the necessary format of the file and what the attributes--or column--requirements are)?

it looks like you're using openmeteo, is this something that you're integrating @elenya-grant?

yes - PR #332

kbrunik · 2025-10-26T19:05:53Z

docs/technology_models/wind_plant_ard.md

+The wind resource capabilities of H2Integrate are not yet connected with Ard, so the user must provide a wind resource file directly to the Ard model inputs.
+
+## Examples
+For an example of using Ard in an H2Integrate model, see `examples/xx_wind_ard`. Note that Ard uses a combination of input files, including a [wind IO](https://github.com/IEAWindSystems/windIO) file.


Could you add a list of the necessary input files for ARD, with maybe a one sentence description about them?

Agreed, and pointing to the Ard docs where possible here would be quite helpful!

kbrunik · 2025-10-26T19:10:18Z

examples/xx_wind_ard/ard_inputs/power_thrust_table_ccblade_NREL-5p0-126-RWT.csv

is this file different from the NREL_Reference_5MW_126.csv? Could one of them be removed?

I am also confused on why either of these CSV files are included because the turbine Cp curve information appears to be in examples/xx_wind_ard/ard_inputs/windIO-plant_turbine_NREL-5.0MW-126m-RWT.yaml

kbrunik · 2025-10-26T19:15:48Z

examples/xx_wind_ard/ard_inputs/windIO-plant_turbine_NREL-5.0MW-126m-RWT.yaml

it seems like this has the same info as the NREL_Reference_5MW_126.csv and the power_thrust_table_ccblade_NREL-5p0-126-RWT.csv, is it necessary to have the information duplicated?

kbrunik · 2025-10-26T19:24:11Z

examples/xx_wind_ard/h2i_inputs/tech_config.yaml

+          modeling_options:
+            windIO_plant: !include ../ard_inputs/windio.yaml
+            layout:
+              N_turbines: 9 #65 # should be 65 turbines


Should this be changed to 65?

kbrunik · 2025-10-26T19:29:44Z

h2integrate/converters/wind/wind_plant_ard.py

+
+
+@define
+class WindPlantArdModelConfig(BaseConfig):


I think you could have WindPlantArdModelConfig() inherit CostModelBaseClass because the cost model base class already inherits BaseConfig and then you'd be able to get rid of the lines 9-22 and still have access to the cost_year, but @elenya-grant or @johnjasa correct me if I'm wrong. Or maybe this is specific to the sub_model setup?

I looked this over, I believe this WindArdCostComponent is necessary because the setup() method of the CostModelBaseClass assumes that you have created self.config. If we don't call that here in this inheriting class' setup() and just tried to use CostModelBaseClass directly, it would fail on the lack of self.config.

I don't think we need to change anything in the framework for the cost model baseclasses to behave differently as part of this PR. It did take me a sec to understand why this is here so I've added a note in the docstring.

Thanks @johnjasa

kbrunik · 2025-10-26T19:30:19Z

h2integrate/converters/wind/wind_plant_ard.py

+class WindPlantArdModelConfig(BaseConfig):
+    """Configuration container for Ard wind plant model inputs.
+
+    Attributes


Could we reformat this docstring to Google style?

kbrunik · 2025-10-26T19:33:10Z

h2integrate/converters/wind/wind_plant_ard.py

+
+class ArdWindPlantModel(om.Group):
+    """
+    OpenMDAO Group integrating the Ard wind plant as a sub-problem..


Please reformat to Google style please and I see a double period on line 42.

Done and removed

kbrunik · 2025-10-26T19:37:13Z

tests/h2integrate/test_all_examples.py

+    # Post-process the results
+    model.post_process()
+
+    with subtests.test("Check wind generation"):


Is there a reason to not set the generation to a value? It seems like it might be easier to check in the future if this test breaks if we have a better idea of the value.

I agree with Kaitlin - I think that having subtests on the value of the total wind generation profile and total solar generation profile would be a good so if other subtests fail (like LCOE or total electricity dispatched, etc), it will be easier to identify what causes other subtest failures.

johnjasa

Cool stuff! It's really fun to see different projects and tools get connected. I think this strengthens our group and its capabilities well.

I've pushed some very minor changes directly. I'm requesting changes, mostly for things that @kbrunik noted that I'd like to see addressed or discussed. Thank you!

johnjasa · 2025-11-04T03:57:29Z

docs/technology_models/wind_plant_ard.md

+The wind resource capabilities of H2Integrate are not yet connected with Ard, so the user must provide a wind resource file directly to the Ard model inputs.
+
+## Examples
+For an example of using Ard in an H2Integrate model, see `examples/xx_wind_ard`. Note that Ard uses a combination of input files, including a [wind IO](https://github.com/IEAWindSystems/windIO) file.


Agreed, and pointing to the Ard docs where possible here would be quite helpful!

johnjasa · 2025-11-04T04:00:13Z

examples/xx_wind_ard/h2i_inputs/plant_config.yaml

+        resource_dir: "../11_hybrid_energy_plant/tech_inputs/weather/solar"
+        resource_filename: "30.6617_-101.7096_psmv3_60_2013.csv"


I like the re-use of the existing solar data, though it might be confusing for folks looking at this to see a different lat/lon location used. I'd suggest either adding a comment here noting that it's a notional location and being used generically, or adding a different weather file.

johnjasa · 2025-11-04T04:11:07Z

h2integrate/converters/wind/wind_plant_ard.py

+
+
+@define
+class WindPlantArdModelConfig(BaseConfig):


I looked this over, I believe this WindArdCostComponent is necessary because the setup() method of the CostModelBaseClass assumes that you have created self.config. If we don't call that here in this inheriting class' setup() and just tried to use CostModelBaseClass directly, it would fail on the lack of self.config.

I don't think we need to change anything in the framework for the cost model baseclasses to behave differently as part of this PR. It did take me a sec to understand why this is here so I've added a note in the docstring.

johnjasa · 2025-11-04T04:16:52Z

h2integrate/converters/wind/wind_plant_ard.py

+class WindPlantArdModelConfig(BaseConfig):
+    """Configuration container for Ard wind plant model inputs.
+
+    Attributes


johnjasa · 2025-11-04T04:16:58Z

h2integrate/converters/wind/wind_plant_ard.py

+
+class ArdWindPlantModel(om.Group):
+    """
+    OpenMDAO Group integrating the Ard wind plant as a sub-problem..


Done and removed

johnjasa · 2025-11-04T04:20:22Z

h2integrate/converters/wind/wind_plant_ard.py

+        subprob_ard = om.SubmodelComp(
+            problem=ard_prob,
+            inputs=[
+                "spacing_primary",
+                "spacing_secondary",
+                "angle_orientation",
+                "angle_skew",
+                "x_substations",
+                "y_substations",
+            ],
+            outputs=[
+                ("aepFLORIS.power_farm", "electricity_out"),
+                ("tcc.tcc", "CapEx"),
+                ("opex.opex", "OpEx"),
+                "boundary_distances",
+                "turbine_spacing",
+            ],
+        )


I appreciate the use of the submodel here! This is mostly for my own curiosity, but why did you choose this route instead of making a group of actual Ard components? Was it mostly to mimic the existing H2I converter style, or due to some of the problem formulation in Ard, or something else?

Ard sets up a full model in openmdao, including driver etc. I took the submodel approach here to allow H2I to take full advantage of Ard (can do its own optimization, DOE, etc) and to not have to rewrite Ard (emphasis on the latter). This is particularly important because Ard is already bringing together many external codes that need special handling, so I thought it would be easier to have a clear line between the H2I and Ard systems.

johnjasa · 2025-11-04T04:21:39Z

h2integrate/converters/wind/wind_plant_ard.py

+        # add ard sub-problem
+        ard_input_dict = self.options["tech_config"]["model_inputs"]["performance_parameters"][
+            "ard_system"
+        ]
+        ard_data_path = self.options["tech_config"]["model_inputs"]["performance_parameters"][
+            "ard_data_path"
+        ]


I'll change this and push up directly, but I'm commenting here so you see this design paradigm. Instead of using:

ard_data_path = self.options["tech_config"]["model_inputs"]["performance_parameters"][ "ard_data_path" ]

you can just do

ard_data_path = self.config.ard_data_path

Got it, thanks!

elenya-grant

Thanks for doing this Jared! I agree with Kaitlin's and John's feedback and added a few comments. My biggest request is to add a unit test or two for the ArdWindPlantModel into h2integrate/converters/wind/test/ to compliment the integration test you added with the example.

One other minor point is that, if possible and easy, it'd be nice if some of the inputs/outputs that exist in PYSAMWindPlantPerformanceModel were replicated in this wind plant model (I am not sure if this is possible given the subgroup setup you introduced).

I will also likely have questions in the future about connecting it with resource data, but that doesnt have to be addressed in this PR.

elenya-grant · 2025-11-05T18:16:23Z

tests/h2integrate/test_all_examples.py

+    # Post-process the results
+    model.post_process()
+
+    with subtests.test("Check wind generation"):


I agree with Kaitlin - I think that having subtests on the value of the total wind generation profile and total solar generation profile would be a good so if other subtests fail (like LCOE or total electricity dispatched, etc), it will be easier to identify what causes other subtest failures.

elenya-grant · 2025-11-05T18:17:43Z

examples/xx_wind_ard/h2i_inputs/plant_config.yaml

+]
+
+resource_to_tech_connections: [
+  # connect the wind resource to the wind technology


nitpick - can you change the comment in resource to tech connections to be about solar instead of wind?

elenya-grant · 2025-11-05T18:19:38Z

examples/xx_wind_ard/h2i_inputs/driver_config.yaml

I love this driver config - I was curious about how this model could be used in a DOE or optimization given its different type of setup but happy to see this included!

elenya-grant · 2025-11-05T18:21:04Z

examples/xx_wind_ard/h2i_inputs/driver_config.yaml

+  flag: True
+  file: "wind_h2_opt.sql"
+  includes: ["*"]
+  excludes: ["wind_resource.wind_resource_data"]


you could do:

excludes: ["*_resource_data"]

to exclude wind and solar resource data from the output file.

elenya-grant · 2025-11-05T18:25:08Z

examples/xx_wind_ard/run_windard_pv_battery.py

I think in the example it'd be nice if the optimization ran (set the optimization flag to True in the driver config file) but I know this optimization wouldn't be ideal for the test corresponding to this example. I'm making a note that after PR #313 is merged in, this would be good example to showcase loading the config files, modifying them, then initializing H2I with a dictionary rather than a filepath.

elenya-grant · 2025-11-05T18:27:17Z

examples/xx_wind_ard/ard_inputs/power_thrust_table_ccblade_NREL-5p0-126-RWT.csv

I am also confused on why either of these CSV files are included because the turbine Cp curve information appears to be in examples/xx_wind_ard/ard_inputs/windIO-plant_turbine_NREL-5.0MW-126m-RWT.yaml

elenya-grant · 2025-11-05T18:29:28Z

h2integrate/converters/wind/wind_plant_ard.py

+        ard_input_dict = self.config.ard_system
+        ard_data_path = self.config.ard_data_path
+
+        ard_prob = set_up_ard_model(input_dict=ard_input_dict, root_data_path=ard_data_path)


This may be out of scope for this PR - but in this Ard wrapper, could we update the site polygon boundaries input to Ard to be set as the site boundaries defined in the h2integrate plant config?

This isn't necessary in my opinion because the site boundaries for H2I are largely meaningless (for now). I think it's fine for Ard to have its own site boundary definition

elenya-grant · 2025-11-05T18:33:07Z

docs/technology_models/wind_plant_ard.md

+        - ORBIT
+        - FinanceSE
+
+## Wind Resource


Could you specify what wind resource data is needed by ard and what the units should be? like that and needs wind speeds in m/s or whatever? Cause if folks are downloading resource data from the openmeteo web API - they may have wind speed in km/h (that's the default unit)

jaredthomas68 added 28 commits August 5, 2025 13:03

working on ard integration

5d1fd35

wip ard integration

bb38a07

complete merge

7a50791

fix syntax

cf6f99c

wip: ard integration needs full year wind data

70b9cbc

add wind resource with api capability

7e807b9

Merge branch 'develop' into ard

3383d99

update type for resource_config option

c499c7f

move ard tests into a new subfolder

1af4bdf

wip ard integration stuck at needing time-series

1402a47

add test for ard

95b311c

wip debuggin ard integration

92744e7

update ard example for some refactor

284291c

add resource request

a0834cf

inputs for example

69fb21c

complete merge

9c0c415

resolve merge conflicts

e31e296

remove jared's attempt at wind resource integration

4f01420

wip: add ard example

4bf6bb2

get ard example working

9bf8ff2

wip: debugging tests

2f3dd3e

merge from develop

80cf490

update docs and changelog

6def61c

include ard in pyproject.toml dependencies

3be97e1

include ard in tech overview

f8f0ae7

backmerge from develop

f503605

install ard from PyPI

b7d4934

remove duplicate test. update docs

0a122f1

jaredthomas68 requested review from johnjasa and kbrunik October 22, 2025 17:41

jaredthomas68 requested review from bayc and elenya-grant October 22, 2025 17:41

Merge branch 'develop' into ard

9f1fb67

kbrunik requested changes Oct 26, 2025

View reviewed changes

jaredthomas68 and others added 7 commits October 27, 2025 08:59

finish merge

67d1842

update naming in ard example

0254d0a

resolve merge conflicts

bc5dfac

Merge branch 'develop' into ard

cb27a1d

Merge branch 'develop' into ard

fd6247c

resolve merge conflicts and renaming corrections

46696b7

Minor docstring and Ard edits

f1d5b7d

johnjasa requested changes Nov 4, 2025

View reviewed changes

elenya-grant requested changes Nov 5, 2025

View reviewed changes

elenya-grant mentioned this pull request Nov 5, 2025

Open-Meteo Wind Resource #332

Merged

31 tasks

kbrunik and others added 3 commits November 5, 2025 15:00

Merge branch 'develop' into ard

ae42667

Merge branch 'develop' into ard

c42c48a

backmerge develop

ea68c23

johnjasa added the needs modifications This PR has been reviewed, at least partially, and is ready for PR author response label Nov 12, 2025

jaredthomas68 and others added 7 commits November 13, 2025 10:40

wip: converting to use h2i resource management for wind

572b99e

finish merge

3382e0e

Merge branch 'develop' into ard

5c299f8

working through pre-commit

9b6ce3b

Merge branch 'develop' into ard

7dd8472

apple silicon install instructions wip

8ef1120

resolve conflicts

d88502a

jaredthomas68 mentioned this pull request Dec 4, 2025

Removing more of old GreenHEART #384

Merged

10 tasks

johnjasa mentioned this pull request Dec 5, 2025

Offshore wind costs #398

Open

jaredthomas68 mentioned this pull request Dec 9, 2025

Added FLORIS wind plant model #372

Merged

29 tasks

		The intent of [Ard](https://github.com/WISDEM/Ard) is to be a modular, full-stack multi-disciplinary optimization tool for wind farms. By incorporating Ard in H2Integrate, we are able to draw on many wind technology models developed at NREL and other institutions without managing them or their connections in Ard. Models connected in Ard include many parts of [WISDEM](https://github.com/WISDEM/WISDEM), [FLORIS](https://github.com/NREL/floris), and [OptiWindNet](https://github.com/DTUWindEnergy/OptiWindNet). Ard also provides constraint functions and wind farm layout generation capabilities among other things. Because Ard has been developed in a modular way, you may extend Ard fairly easily to include other wind models of interest.

		Ard is included in H2Integrate as an [OpenMDAO sub-model](https://openmdao.org/newdocs/versions/latest/features/building_blocks/components/submodel_comp.html), which means that Ard is treated as an OpenMDAO system within an OpenMDAO system. In this way, the user can run an independent wind farm optimization within Ard, or allow H2Integrate to manage the wind farm design variables directly. The drawback of including Ard as a sub-model is that N2 diagrams made from the H2Integrate problems will show Ard only as a single model, rather than showing all the subsystems within Ard. IF you wish to view an N2 diagram of Ard, you will need to use the Ard problem instead.

		resource_dir: "../11_hybrid_energy_plant/tech_inputs/weather/solar"
		resource_filename: "30.6617_-101.7096_psmv3_60_2013.csv"

Ard integration for wind farm combined performance and cost model #312

Are you sure you want to change the base?

Ard integration for wind farm combined performance and cost model #312

Uh oh!

Conversation

jaredthomas68 commented Oct 22, 2025 • edited by johnjasa Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Ard integration for wind farm combined performance and cost model

Type of Contribution

General PR Checklist

New Technology Checklist

Related issues

Impacted areas of the software

Additional supporting information

Test results, if applicable

Uh oh!

kbrunik left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

johnjasa left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

elenya-grant left a comment

Choose a reason for hiding this comment

jaredthomas68 commented Oct 22, 2025 •

edited by johnjasa

Loading