Summary

Adds h2-fuel-cell (hydrogen-powered fuel cell) as a new technology option in ReEDS, complementing the existing ng-fuel-cell (natural gas fuel cell) already in the model. The H2 fuel cell consumes hydrogen (h2fuel, the same fuel as H2-CT/H2-CC), is gated by a new dedicated switch GSw_H2FuelCell, and offers three cost scenarios (conservative/moderate/advanced) based on Reznicek et al. (2026).

Technical details

Implementation notes

• New tech subset: h2-fuel-cell is added to a new H2_FUEL_CELL subset in i_subtech.csv and flagged under both H2_COMBUSTION and H2_FUEL_CELL in tech-subset-table.csv. It is not merged into ng-fuel-cell's FUEL_CELL subset — the two fuel cells are toggled separately.
• GAMS gating (b_inputs.gms): adds a new h2_fuel_cell(i) set; the ng-fuel-cell ban was refactored from a subset ban to an explicit ban('ng-fuel-cell') (under the existing Sw_FuelCell=0 block, behavior preserved), and a new block bans h2-fuel-cell when Sw_H2FuelCell=0. The h2_combustion set comment was updated to include fuel cells.
• Fuel & classification: h2-fuel-cell uses h2fuel (fuel2tech.csv), is added to i.csv and techs_default.csv (region type p).
• Cost data loaded via plantcostprep.py by adding h2fuelcell to the existing conv_techs loader loop, using switch plantchar_h2fuelcell. Dollar year 2022$ (dollaryear.csv), converted to 2004$ via the standard deflator.
• Key parameters: unit size 100 MW; lifetime (max age) 40 years; financing/eval period 30 years (same as Gas-CT); construction time 3 years; min load 1% ; heat rate 8.461 MMBtu/MWh ; VOM 1.7054 $/MWh; FOM ~27 $/kW-yr (moderate, 2030); ramp 100%/min; 6% min CF; 5% forced & scheduled outage; 0 start cost; zero direct CO₂/SO₂.
• Per-technology h2-fuel-cell rows were added across the plant-characteristic and financial input files (emitrate, cost_opres_default/_market, minCF, mttr, outage_forced_static, outage_scheduled_static, ramprate, startcost, unitsize_atb, heat_rate_adj, heat_rate_penalty_spin, degradation_annual_default, years_until_endogenous, construction_times_default, financials_tech_ATB2023/_CRP20/ATB2024, reg_cap_cost_diff_default, nat_gen_tech_frac, ivt_default/_small/_step, futurefiles, runfiles).

Additional changes

• postprocessing/tech_aggregation.csv, tech_map.csv, tech_style.csv updated for h2-fuel-cell display: mapped to H2-Fuel-Cell, aggregated as "H2 Fuel Cell", color #87A21D. (A matching H2 Fuel Cell style key was added so the aggregated bar renders in compare_cases.py PPTs.)
• cases_h2-fuelcell.csv added with 5 test scenarios: USA_defaults_fc_pr (runs without fc switch turning on), USA_defaults_fc_mid (moderate cost/deployment cost case), USA_defaults_fc_low (advanced/high deployment cost case), USA_defaults_fc_high (conservative/low deployment cost case), USA_defaults_fc_mid_w_ng_fc (moderate level + NG fuel cell switch turned on).
• Documentation: new "Hydrogen fuel cells" subsection in docs/source/model_documentation.md, file entries in sources.md, and a Reznicek 2026 entry in references.bib.

Switches added/removed/changed

(GSw_FuelCell is unchanged by this PR and continues to govern only ng-fuel-cell.)

Relevant sources or documentation

• Reznicek, E., Berres, A., Li, Y., et al., Cost Analysis of Heavy-Duty Vehicle Proton Exchange Membrane Fuel Cell Stationary Power Plants (2026), https://doi.org/10.2172/3018256 — unit size and performance parameters

Validation, testing, and comparison report(s)

Ran 1/6 cases under main ReEDS branch, also ran the 5/6 cases under cases_h2-fuelcell.csv to test scenarios (fc_pr, fc_low, fc_mid, fc_high, fc_mid_w_ng_fc) and compared capacity, location, and dispatch with compare_cases.py. The new h2-fuel-cell technology builds, is selected over H2-CT as costs decline, and dispatches as expected.

Location & technology selection (2030)

In 2030 the H2 capacity lands in exactly two BAs : p10 (692 MW) and p31 (210 MW) = 902 MW total in every scenario. The location does not change; only the technology filling those two slots swaps between scenarios as the fuel-cell cost case improves:

This 902 MW is prescribed build from inputs/capacity_exogenous/interconnection_queues.csv . It appears as a step in 2030 in p10 + p31, holds flat through 2035, then declines. The model fills that prescribed requirement with the cheapest available H2 technology, which is why fuel cell displaces H2-CT in the cheaper cost cases (low/mid), partially in high (p10 only), and not at all in pr (H2-CT baseline).

H2 fuel cell capacity build (MW)

Dispatch / capacity factor

Annual CF is a flat 6.0% across every scenario and year. The 6% is the annual minimum CF requirement imposed in inputs/plant_characteristics/minCF.csv; peak hourly utilization reaches ~16% max when looking at the gen_h.csv file.

Checklist for author

Details to double-check

• References for all H2 fuel cell operational data (Reznicek et al. 2026, DOI 10.2172/3018256)
• Charge code provided to reviewers
• Included comparison reports for appropriate test cases
• Documentation updated if necessary
• If input data added/modified:
  • Dollar year recorded and converted to 2004$ for GAMS (dollaryear.csv updated for h2fuelcell_* scenarios, 2022$)
  • Units specified (MW capacity, $/kW capex, $/kW-yr FOM, $/MWh VOM, MMBtu/MWh heat rate)
  • No large data files — all inputs are small CSVs
• No change to process flow (runreeds.py, solve.py)
• No change to spatial/temporal resolution

General information to guide review

• Zero impact on results of default case (GSw_H2FuelCell=0 by default — no change when switch is off)
• No large data file(s) added/modified
• No substantive impact on runtime
• No change to process flow
• No change to package requirements

Did you use LLM tools (chatbot or copilot) in the preparation of this PR? If so, describe how

Yes — Claude Code (Anthropic) assisted with documentation modification and writing this PR language.

Tag points of contact

@yijinli22