Data from: Grid-connected PEM electrolysis: Experimental characterization and dispatch simulation across a decarbonizing grid

Veatch, Evan 1 ; Windom, Bret1; Zdanowicz, Andrew1

Published Sep 03, 2026 on Dryad. https://doi.org/10.5061/dryad.8931zcs64

Data files

Sep 03, 2026 version files 72.10 MB

Fixed_ERCOT_Electrolyzer_Sim_Engine.zip

72.10 MB
README.md

5.69 KB

Abstract

This dataset contains the code, inputs, and outputs supporting a grid-connected PEM electrolyzer dispatch simulation. It includes experimental characterization data from a 60 kW NEL C-Series PEM electrolyzer collected across its full operating range (AC power, DC power, hydrogen production rate, and system efficiency), six years (2020–2025) of ERCOT locational marginal price and generation-mix data used as simulation inputs, and dispatch simulation outputs from a parametric sweep of approximately 9,468 supervisory control configurations. MATLAB code performs the core dispatch simulation; Python scripts handle data pre- and post-processing. All source files are plain text and viewable in standard code editors such as Visual Studio Code, including the MATLAB scripts. Users without access to MATLAB will need to reimplement the simulation logic in an open-source environment such as Python or Octave to reproduce results directly. The dataset supports reuse for evaluating electrolyzer performance under alternative grid conditions, testing supervisory control strategies against historical or synthetic price and emissions signals, and benchmarking hydrogen production cost and carbon intensity against other pathways such as steam methane reforming. The underlying experimental data reflects a commercial system originally configured for hydrogen refueling applications and may not generalize to electrolyzer stacks of different scale or design without adjustment. No personally identifiable or sensitive information is contained in this dataset; ERCOT market data used here is publicly available.

Dataset DOI: 10.5061/dryad.8931zcs64

Description of the data and file structure

This dataset supports characterization-informed simulation of a grid-connected proton-exchange membrane (PEM) electrolyzer. Experimental power and hydrogen-production traces provide the basis for the modeled system behavior. The MATLAB simulation engine applies supervisory-control cases to time-aligned ERCOT-like grid, generation, and electricity-price signals from 2020 through 2025.

The included outputs quantify electrolyzer electrical consumption, hydrogen production, electricity cost, grid-associated carbon dioxide emissions, operating states, and aggregate performance metrics. The dataset is organized so that the supplied inputs and case definitions can be used to reproduce the included machine-readable and tabular results.

Files and variables

configs/paper_cases/supervisory_control_cases.csv: supervisory-control settings for the simulated cases. Variables are:
- ID: case identifier.
- min_up_steps, min_down_steps: minimum on and off durations, in simulation steps.
- price_start, price_turndown, price_stop: electricity-price thresholds in USD/MWh.
- clean_ratio_start, clean_ratio_turndown, clean_ratio_stop: clean-energy-ratio thresholds.
- turndown_delay, recover_delay, price_delay: controller delay settings in simulation steps.
data/inputs/: time-aligned grid input signals used by the simulation. Each text file contains a timestamp and a numeric signal value. load.txt is grid load; wind.txt, solar.txt, nuclear.txt, hydro.txt, biomass.txt, bess.txt, and wsl.txt are non-fossil or project-specific generation/storage channels; gas_cc.txt, gas_other.txt, and coal.txt are fossil-generation channels; and price.txt is the electricity-price signal in USD/MWh. Power channels are expressed on the MW basis used by the model.
data/experiment/: synchronized experimental PEM electrolyzer traces. timestamp.txt supplies the common time axis. power_command.txt, stack_power.txt, smps_power.txt, chiller_power.txt, and subsystem_power.txt record commanded or measured power signals. hydrogen_flow.txt and hydrogen_flow_cs.txt record hydrogen-flow signals.
src/matlab/: MATLAB implementation of the simulation engine. The engine/ directory contains clean-energy, cost, and emissions calculations; models/ contains system and component models; sim/ contains the simulation and post-processing scripts; and io/ contains input and HDF5/CSV export functions.
outputs/results/: HDF5 result files. Included result files such as default_case.h5 and clean_case.h5 contain time, case identifiers, simulated grid power (P_grid_kW), hydrogen production rate (h2_kgph), cost, carbon dioxide emissions, aggregate energy/cost/hydrogen/emissions totals, and startup counts.
outputs/tables/: CSV result exports. *_timeseries.csv files contain time-resolved variables including clean_ratio, energy_price, system power, energy, cost, hydrogen production, and emissions. *_case.csv files provide aggregate and annual metrics alongside time-series data.
src/python/analysis/: Python scripts for analysis, statistical summaries, and visualization of exported results.
docs/: model-design notes, control-logic documentation, and project workflow documentation.
tools/: optional utilities for inspecting HDF5 files, exporting result data, formatting materials, and visualizing time series.

Data conventions

P_grid_kW and System Power (kW): modeled electrolyzer system electrical draw, in kW.
h2_kgph: hydrogen production rate, in kg/h.
h2_kg: hydrogen produced during a simulation time step, in kg.
e_MWh: electrical energy consumed during a simulation time step, in MWh.
cost: electricity cost during a simulation time step, in USD.
co2_kg: grid-associated carbon dioxide emissions during a simulation time step, in kg CO₂.
co2_per_h2: carbon dioxide intensity, in kg CO₂ per kg H₂.
clean_ratio: fraction of the modeled grid supply represented by the project's clean-energy calculation.
Missing numeric values are represented by empty fields or NaN where applicable. Ratio values may be undefined when hydrogen production is zero.

Code/software

The simulation engine is written in MATLAB. MATLAB must be available to reproduce simulation outputs. From the repository root, initialize the project path and run the simulation scripts:

run('src/matlab/startup.m')
run('src/matlab/sim/run_sim.m')
run('src/matlab/sim/post_sim.m')

run_sim.m loads the input signals and supervisory-control cases, simulates each case, and writes a chunked HDF5 file to outputs/results/results_timeseries.h5. post_sim.m reads that file and writes outputs/tables/results_timeseries.csv.

Python 3 is required for the scripts in src/python/analysis/ and the utilities in tools/. Packages used across these scripts include numpy, pandas, h5py, matplotlib, scipy, and statsmodels.

Access information

The dataset is available through Dryad at 10.5061/dryad.8931zcs64. The accompanying code repository is available at HydrogenDude/ERCOT_Electrolyzer_Sim_Engine.

The dataset contains grid and engineering simulation data and experimental system traces. It contains no human-subject or personally identifiable data.