Iron: Adding electrowinning capabilities by jmartin4u · Pull Request #432 · NatLabRockies/H2Integrate

jmartin4u · 2026-01-02T20:15:43Z

Iron: Adding electrowinning capabilities

This PR adds three iron electrowinning technologies:

Aqueous hydroxide electrolysis (AHE)
Molten salt electrolysis (MSE)
Molten oxide electrolysis (MOE)

For the time being, the performance and cost are modeled fairly simply. Performance is modeled based on upper and lower bounds of energy required for each type as defined by Humbert. Cost is modeled based on correlations of capital cost developed for electrowinning of many different metals developed by Stinn and Allanore, and an opex model developed by Humbert in the spreadsheet available as SI to the linked paper, which also provides input values for the Stinn/Allanore capex model.

Section 1: Type of Contribution

Section 2: Draft PR Checklist

N/A

Type of Reviewer Feedback Requested (on Draft PR)

Structural feedback:

Implementation feedback:

Other feedback:

Section 3: General PR Checklist

PR description thoroughly describes the new feature, bug fix, etc.
Added tests for new functionality or bug fixes
Tests pass (If not, and this is expected, please elaborate in the Section 6: Test Results)
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)
CHANGELOG.md has been updated to describe the changes made in this PR

Section 3: Related Issues

Section 4: Impacted Areas of the Software

Section 4.1: New Files

h2integrate\converters\iron\:
- humbert_ewin_perf.py: New iron electrowinning performance model humbert_electrowinning_performance based on Humbert
- humbert_stinn_ewin_cost.py: New iron electrowinning cost model humbert_stinn_electrowinning_cost based on Humbert and Stinn and Allanore
- stinn\table1.csv: Input data table mostly from Stinn and Allanore but iron costs from Humbert
h2integrate\converters\iron\stinn\cost_model.py: Modified the transcription of the Humbert SI equations to interface with the humbert_stinn_electrowinning_cost model
examples\27_iron_electrowinning\: New example that produces an LCOI for all 3 types of iron electrowinning

Section 4.2: Modified Files

h2integrate\converters\iron\stinn\:
- cost_model.py: Modified the transcription of the Stinn and Allanore equations to interface with the humbert_stinn_electrowinning_cost model
- cost_coeffs.csv: Added in some coefficients that were previously hard-coded into cost_model.py
h2integrate\core\supported_models.py: Added humbert_electrowinning_performance and humbert_stinn_electrowinning_cost as supported models
h2integrate\finances\profast_base.py: Added sponge_iron as a mass commodity

Section 5: Additional Supporting Information

Section 6: Test Results, if applicable

Section 7 (Optional): New Model Checklist

kbrunik

Thanks for working on this! Overall super cool and exciting to have some more functionality added to H2I. I know this isn't quite finished with regards to documentation and testing but thought I could share some initial thoughts.

examples/27_iron_electrowinning/pysam_options_6000MW.yaml

examples/21_iron_mn_to_il/plant_config_old.yaml

examples/27_iron_electrowinning/27_iron_electrowinning.yaml

examples/27_iron_electrowinning/driver_config.yaml

h2integrate/converters/iron/humbert_ewin_perf.py

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

h2integrate/converters/iron/stinn/cost_model.py

h2integrate/converters/iron/stinn/cost_coeffs.csv

h2integrate/converters/iron/stinn/table1.txt

into HEAD

elenya-grant

Hi Jonathan! I think this is super cool and the example is very well done! I didn't take a super deep look into the new models but provided some initial feedback (some of it may be irrelevant depending on other changes you plan on making). Would be happy to do a deeper review once it's ready! Thanks for the work on this! Very exciting!

h2integrate/converters/iron/rosner/cost_coeffs.csv

h2integrate/converters/iron/stinn/cost_model.py

h2integrate/converters/iron/test/test_humbert_ewin_perf.py

h2integrate/converters/iron/test/test_humbert_stinn_ewin_cost.py

h2integrate/converters/iron/stinn/cost_model.py

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

elenya-grant · 2026-01-20T18:20:14Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+        self.add_output("labor_opex", val=0.0, units="USD/year")
+        self.add_output("NaOH_opex", val=0.0, units="USD/year")
+        self.add_output("CaCl2_opex", val=0.0, units="USD/year")
+        self.add_output("limestone_opex", val=0.0, units="USD/year")


shouldn't feedstock costs (lime, electricity, ore, etc) be taken care of with feedstock component(s)? In my mind, this should only output costs associated with this tech, not its feedstocks.

elenya-grant · 2026-01-20T18:21:08Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+        elec_price = inputs["price_electricity"]
+
+        # Add ore transport cost TODO: turn iron_transport into proper transporter
+        ore_price += ore_transport_cost


ore transport is now a separate model, should not be included here.

elenya-grant · 2026-01-20T18:22:35Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+
+        # Humbert Opex model - from SI spreadsheet (doi.org/10.1007/s40831-024-00878-3)
+        # Default costs - adjusted to 2018 to match Stinn via CPI
+        labor_rate = 55.90  # USD/person-hour


Should labor_rate be an attribute of the configuration class?

elenya-grant · 2026-01-20T18:22:58Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+        labor_rate = 55.90  # USD/person-hour
+        NaOH_cost = 415.179  # USD/tonne
+        CaCl2_cost = 207.59  # USD/tonne
+        limestone_cost = 0


I think these feedstock costs should be taken care of with the feedstock component.

elenya-grant

This is coming along nicely! Thanks for the work on this! I think the performance model is looking awesome, I have some remaining questions and comments on the cost model. Another small nit-pick (non-blocking) is that links to papers can be embedded in docstrings as hyper-links if they're formatted like this:

 All of these values come from the SI spreadsheet for the Humbert paper that can be downloaded
    at `doi.org/10.1007/s40831-024-00878-3 <https://link.springer.com/article/10.1007/s40831-024-00878-3>`_

Thanks for making a lot of the previously suggested changes. I think it's nearly ready! Let me know if you want to discuss any of the comments I left!

elenya-grant · 2026-01-29T18:08:38Z

h2integrate/converters/iron/humbert_ewin_perf.py

+        ewin_type = self.config.electrolysis_type
+        # Look up performance parameters for each electrolysis type from Humbert Table 10
+        if ewin_type == "ahe":
+            E_all_lo = 2.781


could you add in-line comments about what each of these variables are and the units

E_all_lo = 2.781 # [kWh/kg] the something something of something

Added as a comment block at the top so I didn't have to repeat 3 times, fine if you want to change

elenya-grant · 2026-01-29T18:09:56Z

h2integrate/converters/iron/humbert_ewin_perf.py

+        n_timesteps = self.options["plant_config"]["plant"]["simulation"]["n_timesteps"]
+        self.config = HumbertEwinConfig.from_dict(
+            merge_shared_inputs(self.options["tech_config"]["model_inputs"], "performance"),
+            strict=False,


why strict=False - this is normally only used (as far I I know) for dispatch/storage models because theres more than 2 models connected. But in this case, it appears that its only a performance and cost model, so strict=True would be appropriate

Not sure if I need to change to strict=True or remove the whole statement - will leave in your hands!

elenya-grant · 2026-01-29T18:10:42Z

h2integrate/converters/iron/humbert_ewin_perf.py

+        self.add_input("spec_energy_electrolysis", val=E_electrolysis, units="kW*h/kg")
+        self.add_input("capacity", val=self.config.capacity_mw, units="MW")
+
+        self.add_output(


could you add iron_ore_consumed as an output here and then calculate it in the compute() method?

elenya-grant · 2026-01-29T18:16:35Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+
+            # MSE - Opex
+            positions = 499.2 / 2e6  # Labor rate (position-years/tonne)
+            NaOH_ratio = 0  # Ratio of NaOH consumption to annual iron production


should NaOH and CaCl2 be inputs to the performance model that are consumed and may limit the output production similar to electricity and iron ore? Then the costs associated with these could be represented with a feedstock component and not included in this cost model. I could see it going either way, if it makes sense to include within this cost model, then its fine as-is

Yes, it makes sense to make them into feedstocks and limit production. If you can make that change really quick great, I don't see it as a requirement - I see a wave of converting lots of these little VarOpExes into feedstocks at once across multiple models, because I bet there's more of them buried.

elenya-grant · 2026-01-29T18:17:06Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+            anode_replace_int = 3  # Replacement interval of anodes (years)
+
+        # Set up connected inputs
+        self.add_input("output_capacity", val=0.0, units="Mg/year")  # Mg = tonnes


why not use t/yr instead of Mg/yr?

elenya-grant · 2026-01-29T18:19:27Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+        elec_in = inputs["electricity_in"]
+        elec_price = inputs["price_electricity"]
+
+        # Add ore transport cost TODO: turn iron_transport into proper transporter


Reminder to remove the transport cost from this model since its a separate component now.

Already removed, also removing reference in docstring

elenya-grant · 2026-01-29T18:20:20Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+
+        # Humbert Opex model - from SI spreadsheet (doi.org/10.1007/s40831-024-00878-3)
+        # Default costs - adjusted to 2018 to match Stinn via CPI
+        labor_rate = 55.90  # USD/person-hour


could these costs be in the configuration class? They could default to the values you have here but could then be modified by the user.

Yes, those could easily be config parameters. Lazy on my part. If you have time to fix, please do

elenya-grant · 2026-01-29T18:21:35Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+        outputs["VarOpEx"] = (
+            NaOH_opex + CaCl2_opex + limestone_opex + anode_opex + ore_opex + elec_opex
+        )
+        outputs["labor_opex"] = labor_opex


Are you outputting the individual costs for testing purposes to make sure it's aligned with the results from the paper? (Just checking since only OpEx, CapEx, and VarOpEx will be used in the finance models)

I want to have the ability for capex and opex to be broken out in more detail - I don't want the only outputs of the cost model to be the entire sum of capex and opex. I want to be able to make plots and tables (straight from the output files) that show which parts of the capex (the electrolyzer, the processing) are sensitive to which inputs (the temperature, the iron content of the ore. Even if the finance model isn't using them I'd like to keep these outputs, and I'll put a note into the comments so there's no confusion on why they're there.

elenya-grant · 2026-01-29T18:22:25Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+        self.add_input("positions", val=positions, units="year/Mg")
+        self.add_input("NaOH_ratio", val=NaOH_ratio, units=None)
+        self.add_input("CaCl2_ratio", val=CaCl2_ratio, units=None)
+        self.add_input("limestone_ratio", val=limestone_ratio, units=None)


instead of units of None could you make these unitless?

elenya-grant · 2026-01-29T18:24:18Z

h2integrate/converters/iron/humbert_stinn_ewin_cost.py

+        self.add_input("iron_ore_in", val=0.0, shape=n_timesteps, units="kg/h")
+        self.add_input("iron_transport_cost", val=0.0, units="USD/t")
+        self.add_input("price_iron_ore", val=0.0, units="USD/Mg")
+        self.add_input("electricity_in", val=0.0, shape=n_timesteps, units="kW")


I think electricity and iron ore costs should be calculated in feedstock models

Indeed! Looks like you already removed, also deleting references from docstrings

jmartin4u and others added 20 commits November 6, 2025 15:19

Reading in Stinn input tables

26a6ec6

Sucessfully testing Stinn model

3a803fd

Merge remote-tracking branch 'upstream/develop' into ewin-dev

6fa32fd

Making humbert perf model

9d4ff83

Adding humbert/stinn cost model inputs only

84aad1c

Progress on humbert cost

674cd47

Humbert Stinn complete, not debugged

0728b2d

Debugging electrowinning

9601cce

Merge remote-tracking branch 'upstream/develop' into ewin-dev

3351d51

Debugging ewinning

6310a12

Merge remote-tracking branch 'upstream/develop' into ewin-dev

6455234

Electrowinning cases running, need to add DRI case

4f978b8

Merge remote-tracking branch 'upstream/develop' into ewin-dev

bd5cdb4

Cleanup of old trailing underscores

9388822

Fix input csv names

db6864b

Merge remote-tracking branch 'upstream/develop' into ewin-merge

c130111

Integrate ewin example with feedstocks

0b41cbf

Moving functions out of h2i module and renumbering example

ab63af7

Merge branch 'develop' into ewin-merge

47b6535

Small fixes to PR

149ee57

kbrunik requested changes Jan 12, 2026

View reviewed changes

kbrunik and others added 9 commits January 12, 2026 17:36

start of unit test

efa47cc

Merge branch 'ewin-merge' of https://github.com/jmartin4nrel/H2Integrate

5b44977

into HEAD

Testing example

502aa8c

Expanding unit tests

8c79a30

Merge branch 'develop' into ewin-merge

95c5e70

fix import

d30a44a

Docstrings begun

3992372

Docstrings plus a little reorg

6845b13

Everything but the docs page

9229ed7

elenya-grant requested changes Jan 20, 2026

View reviewed changes

johnjasa added the needs modifications This PR has been reviewed, at least partially, and is ready for PR author response label Jan 20, 2026

Addressing reviews

c34f4e5

jmartin4u requested review from elenya-grant, johnjasa and kbrunik January 22, 2026 18:25

jmartin4u added ready for review This PR is ready for input from folks and removed needs modifications This PR has been reviewed, at least partially, and is ready for PR author response labels Jan 23, 2026

johnjasa and others added 8 commits January 23, 2026 18:57

Merge branch 'develop' into ewin-merge

71ea58a

site -> sites

a7e62c7

Updating test values

f4b24f5

Merge remote-tracking branch 'upstream/develop' into ewin-merge

c4cc271

Merge branch 'develop' into ewin-merge

09fa9a2

Added autodoc to the electrowinning docs

98bb8c5

Merge branch 'develop' into ewin-merge

fe7d704

Merge remote-tracking branch 'upstream/develop' into ewin-merge

2a0d6d5

elenya-grant requested changes Jan 29, 2026

View reviewed changes

Merge branch 'develop' into ewin-merge

6464330

johnjasa added the needs modifications This PR has been reviewed, at least partially, and is ready for PR author response label Feb 2, 2026

kbrunik and others added 11 commits February 3, 2026 10:19

Merge branch 'develop' into ewin-merge

85a11a7

merged in develop

39b9ba7

updated iron electrowinning tech config

89b1c5f

updated plant config for electrowinning example

4b43951

minor udpates to cost model

71967dc

minor updates to cost and performance models

f8ef558

fixed initialization error in cost model

1544cbd

removed duplicate costs and updated to use PerformanceModelBaseClass

1c5a82a

updated example test values

9038c6e

Merge branch 'develop' into ewin-merge

1ea3e07

Addressing elenya comments

61dced2

Conversation

jmartin4u commented Jan 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Iron: Adding electrowinning capabilities

Section 1: Type of Contribution

Section 2: Draft PR Checklist

Type of Reviewer Feedback Requested (on Draft PR)

Section 3: General PR Checklist

Section 3: Related Issues

Section 4: Impacted Areas of the Software

Section 4.1: New Files

Section 4.2: Modified Files

Section 5: Additional Supporting Information

Section 6: Test Results, if applicable

Section 7 (Optional): New Model Checklist

Uh oh!

kbrunik left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

elenya-grant left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

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

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

jmartin4u commented Jan 2, 2026 •

edited

Loading