Update 2025 ATB

[Yunzhi-Chen]

Summary

Updates the ATB workflow from ATB 2024 v3 to ATB 2025 v1, in 2023 dollars.

ATB 2025 renamed its cases, dropped the market case for most technologies, moved its base year to 2023, and extended its horizon to 2060. This updates the configuration and the code that depended on ATB's 2024 vocabulary, and revises the default historical source for most technologies.

Technical details

• Pinned to the ATB 2025 v1.0.0 flat file and the 2025 v1 workbook; all costs are converted to 2023 dollars, the base year the workbook states.
• ATB renamed Market to Exp (a pilot "Expanded Cost Drivers" case) and split tax credits into a separate + TC suffix. Only 9 of 20 technologies have the pilot case, so Exp is selected for utility PV, land-based wind and utility-scale battery, and R&D for the rest, which have nothing else published.
• The + TC variants are not used. Tax credits move LCOE and financing, not overnight cost, so every metric ReEDS reads is identical with and without the suffix.
• The battery workbook sheet was renamed by ATB; the extractor points at the Expanded case sheet.

Historical source

• The actual historical source was used for available technologies for historical years before 2025.

Output range

• New reeds_end_year, set to 2060 to match the horizon ATB 2025 publishes.

Smoothing changes

• slope_change_threshold and max_kink_years was turned to facilitate new data.

Relevant sources or documentation

• ATB 2025 v1 flat file
• ATB 2025 v1 workbook

Additional details

Comparison against the current 2024 ReEDS inputs

Comparison against ReEDS ATB 2024:

compare_to_reeds_2025_1.zip

Output verification

The formatter completed successfully and generated 32 ReEDS input files. Every series backed by ATB data spans 2010 to 2060; the five series ATB does not publish span 2010 to 2050. Each plot below shows one technology's final curves against the same run with smoothing disabled, with historical years coloured by their configured source; years filled from observed data inside the 2023-2025 window use the same colour as observed history.

The generated CSVs are not attached; they are reproducible by running the workflow with the committed atb/config.yaml.

Before/after smoothing plots:

smoothing_comparison_2025_1.zip

Known limitations

• The sources of the manual historical inputs are undocumented. When concatenating the manual historical inputs from the ReEDS 2024 ATB inputs with the new 2025 ATB flat files, we may see some weird curves, as shown below. This is because the new 2025 ATB data has relatively large differences from the 2024 projection in the start year. The near term future fix would be to clean up the manual historical files using historical ATB project data, for example, using ATB 2016 to fill in the 2015 historical values. Long term future fix would be fill up all historical data with the real historical data.

LLM usage

LLM tools assisted with implementation, documentation, debugging, and validation. Outputs were reviewed and tested against the current ReEDS inputs.