Starting structure for OM refactor by elenya-grant · Pull Request #92 · NatLabRockies/electrolyzer

elenya-grant · 2026-07-13T22:48:03Z

Starting structure for OM refactor

This PR requires PR #90 to be merged in!

This PR introduces a different structure of the functionality within the electrolyzer repo to accommodate an Open-MDAO structure. This is in a somewhat intermediate state so that future development has smaller PRs and can have more rigorous testing and a focus on individual component development. This PR really focuses on the structure for the "cluster" and the connection from a cluster to a controller. The "controller" right now is just an IndepVarComp because the structure is only set-up to handle one cluster at the moment. The cluster is comprised of 3 main subgroups:

converter: perhaps this should be renamed to "pre_translator". This defines the current operational bounds and also is used to get curve coefficients for the power-to-current conversion.
translator: this is used to take a cluster input power signal and convert it to the cell-level current input
simulation: this takes the cell-level current input and simulates cell degradation, includes dynamic operating constraints, and simulates the cell performance.

Plans for follow-on PRs:

update framework to handle hydrogen as the control variable (Issue OM Structure: Hydrogen control variable #93)
update framework so that all cell design variables are promoted to the top-level (Issue OM Structure: Update framework so cell design variables are promoted to higher-level #94)
update framework to use multiple clusters with a controller (Issue OM Structure: Multiple Clusters Connected to a Controller #95)
add component that tracks how close to EOL the cluster is and when its replacement schedule is (Issue OM Refactor: Component to Track Remaining Life #96)
add a component to "classify" the cluster and system (rated capacity, etc)
add a "scale-up" component after the "simulation" component to scale outputs (Issue OM Refactor: Scale-up components and System Classification #97)
add a system-type component (or something?) that can handle cases if multiple feedstocks are limiting (Issue OM Refactor: Feedstock limited cases #98)

Files to Remove (were helper/drafts):

electrolyzer/translators/power_translator_baseclass.py
electrolyzer/core/simulate_structure.py
electrolyzer/core/pre_processing_structure_v2.py

Related issue, if one exists

Impacted areas of the software

Additional supporting information

Test results, if applicable

…olyzer into om_refactor/structure

…aseclasses

…or components

…commit

…cture

…rolyzer into om_refactor/structure

johnjasa

Some initial thoughts and comments from me! I'm liking what I'm seeing, especially with the care you've put to introducing base classes and then having smart inheritance. The similarities between the structure you've chosen here and H2I is fun to see, as well as the differences since we're tackling slightly different problems.

Nothing blocking, though I am treating this PR as part of a larger refactor effort so I'm not being picky about docstrings, etc. Glad you're able to push through some great progress here!

johnjasa · 2026-07-28T02:32:30Z

+    def initialize(self):
+        super().initialize()
+        # self.options.declare("mode", types=str, default="normal")
+
    def setup(self):
        # self.n = 2  # number of electrons transferred in reaction

        self.config = PEMCellConfig.from_dict(self.options["tech_config"]["cell_parameters"])

        super().setup()

+        # self.add_input("H2_demand", val=0.0, shape_by_conn=True, units="g/s")
+        # self.add_output("I_demand", val=0.0, copy_shape="H2_demand", units="A")
+        # self.add_output("J_demand", val=0.0, copy_shape="H2_demand", units="A/(cm**2)")


remove commented code? or is useful for later?

this is removed in the follow-on PR I think?

johnjasa · 2026-07-28T02:33:13Z

maybe rename to cell_baseclass.py? not necessary

johnjasa · 2026-07-28T02:35:16Z

I'm guessing you're going to be fleshing this out more later on, but I still think it'd be worthwhile to have a docstring or two in here explaining the very basics of the simple dynamics class, what it does, potential applications for it

johnjasa · 2026-07-28T02:37:59Z

+                outputs[o_name] = inputs[in_name] / inputs["n_stacks"]
+
+
+class ScaleUp(om.ExplicitComponent):


These two classes are named very broadly; is this on purpose? Will these be used in multiple different contexts, or could we name them more clearly to say at what level they'd be used, e.g. "stack to cluster" or whatever it is?

removed the building blocks file

johnjasa · 2026-07-28T02:38:10Z

+
+
+class ScaleUp(om.ExplicitComponent):
+    """Scale things down"""


docstring flipped

johnjasa · 2026-07-28T02:38:24Z

+class Draft(om.ExplicitComponent):
+    def initialize(self):
+        self.options.declare("tech_config", types=dict, default={})
+
+    def setup(self):
+        pass
+
+    def compute(self, inputs, outputs):
+        pass


sorry I meant to delete the entire building_blocks.py file!

johnjasa · 2026-07-28T02:39:04Z

+    A_cell: float = field(validator=(validators.gt(0.0)))
+
+
+class CurrentBoundsBase(om.ExplicitComponent):


Tell me about this component; its name/docstring/filename doesn't make it immediately clear to me what it's doing or where it's used

It supplies the min/max current bounds to the dynamics component (and would likely send these to the controller). It also calculates a bunch of current reference points from min to max for the curve-fitting.

johnjasa · 2026-07-28T02:40:23Z

+        # Step 1: Create controller cluster connector components
+        pre_translator = self.create_controller_cluster_connector()
+        # Translator has scale down + power to current conversion
+        translator = self.create_controller_translator()
+        # Step 2: Create the simulate block of a cluster
+        simulator = self.create_cluster_simulation_block()
+
+        cluster_group.add_subsystem(
+            "converter", pre_translator, promotes=["A_cell", "I_min", "I_max"]
+        )
+        cluster_group.add_subsystem("translator", translator, promotes=["n_stacks", "n_cells"])
+        cluster_group.add_subsystem("simulation", simulator, promotes=["I_min", "I_max", "A_cell"])


I'm not necessarily sold on the translator and converter terms. Is there something where we could call it static_design or pre_simulation or something else? It could be more clear what those names mean

I'm not sold on them either. I did have it the "converter" called the "pre-translator" at one point. I don't think I'm sold on the static_design name. I did make an issue about the naming in general and started brainstorming some new names for the classes here #100

johnjasa · 2026-07-28T02:41:50Z

+        cell_scale_down = self.create_scale_down_component(
+            scale_comp="cells", n_comps=self.config["stack"]["n_cells"]
+        )
+        stack_scale_down = self.create_scale_down_component(
+            scale_comp="stacks", n_comps=self.config["cluster"]["n_stacks"]
+        )
+
+        translator.add_subsystem(
+            "cluster_to_stack",
+            stack_scale_down,
+            promotes_inputs=["n_stacks"],
+        )
+        translator.add_subsystem(
+            "stack_to_cell",
+            cell_scale_down,
+            promotes_inputs=["n_cells"],
+        )


Right on, I see how the scale components are used now, please mostly disregard my prior comment. I do still think when you get to documenting the components it'd be useful to include a few examples of where they'd be used in the docstring so folks understand where you'd be scaling up or down

elenya-grant added 30 commits July 6, 2026 15:36

fleshed out more of stack

ed2b738

changes to existing files as temporary save point

58d0801

added play scripts with structure drafts

e84028f

added nonworking draft of possible config input file

bd08712

re-imagined some of input config

6023e2f

merged in pem cell branch

36214e1

Merge branch 'om_refactor/pem_cell' of github.com:elenya-grant/electr…

d30a492

…olyzer into om_refactor/structure

added cluster level grouping and connections using building blocks

9c104ff

translated dynamics, degradation, and PI curve building blocks into b…

d944e8c

…aseclasses

added Bounds baseclass and power scale-up/down baseclass

ef08966

updated cell output names

e306de2

updated stack to be named degradation and inherit new baseclass

72c2c22

updated cluster.py to use dynamics baseclass

491416b

renamed cluster.py and stack.py files

b17e505

added power to current translator class

c506a95

added very basic controller

3c35439

working on syncing models for workflow

8aed93c

redid openloop controller

6a7d142

tried to finish up simple implementation of clusters in bert.py

3acb7d4

got bert structure working with building blocks

cb2949e

removed some old draft code

f67a38f

replaced cell with PEMCell model

c5540c0

added in actual degradation component

f400b29

added in dynamics component

f13a930

added power to current translator component

98e5b99

added bounds component

f99c1fe

added use of scale down component

7903872

added in logic to handle with or without controller

1d1394b

starting playing with example 00 to test functionality

f2135ca

updated to have better differentiation between converter and translat…

2002335

…or components

added tiny test for example 00

375d04a

elenya-grant mentioned this pull request Jul 20, 2026

OM Structure: Multiple Clusters Connected to a Controller #95

Open

elenya-grant requested a review from johnjasa July 20, 2026 22:05

elenya-grant and others added 7 commits July 20, 2026 16:09

added some comments to bert.py and maybe other cleanups

d445bb1

removed draft or unused files

acb6747

bugfix in A_cell units and updated test for example 00 from previous …

6319064

…commit

updated pem cell test for updated V_cell_out naming

89c0220

Merge branch 'develop' into om_refactor/structure

724d4d3

Merge remote-tracking branch 'upstream/develop' into om_refactor/stru…

5f50d2a

…cture

Merge branch 'om_refactor/structure' of github.com:elenya-grant/elect…

9d4cf2b

…rolyzer into om_refactor/structure

elenya-grant marked this pull request as ready for review July 27, 2026 17:35

johnjasa approved these changes Jul 28, 2026

View reviewed changes

elenya-grant added 3 commits July 28, 2026 11:15

removed building blocks file

3b2f4d4

added docs/comments to dynamics baseclass and simple_dynamics

0aa1bc6

added docstring to CurrentBoundsBase

5d06031

		outputs[o_name] = inputs[in_name] / inputs["n_stacks"]


		class ScaleUp(om.ExplicitComponent):

		A_cell: float = field(validator=(validators.gt(0.0)))


		class CurrentBoundsBase(om.ExplicitComponent):

Uh oh!

Conversation

elenya-grant commented Jul 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Starting structure for OM refactor

Related issue, if one exists

Impacted areas of the software

Additional supporting information

Test results, if applicable

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!

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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

elenya-grant commented Jul 13, 2026 •

edited

Loading