Methods

Made by AI agents, with a person directing and checking the work. How this was made.

Last amended: 2026-10-25. See all amendments.

Data

• EIA-930 hourly demand ^[1][2]: SPP (balancing authority SWPP; adjusted demand where available) and SPP subregions KCPL (Evergy Metro) and WR (Evergy Kansas Central). "Evergy area" = KCPL + WR.
• EIA-861 2024 ^[3]: utility-reported peaks and Kansas retail sales.
• Forecasts transcribed from NERC LTRA 2025 ^[4] and SPP's 2025 resource adequacy report ^[5].
• Costs from EIA AEO2025 capital cost study ^[6] and the MISO JTIQ guide's annual charge rate ^[7].

Processing

Times converted to Central. Seasons: summer Jun-Sep, winter Dec-Feb (labeled by January's year); seasons with <90% of hours dropped. Two impossible hourly values (2023-06-13, ~50-100x normal) were removed and logged.

Scenarios

For each shave percentage, the target is the 2025 peak minus X%. We count the hours and days above the target and the energy above it (MWh that must move). Avoided capacity = shaved MW x 1.16. Annual value = avoided kW x $103-$194/kW-yr.

Checks

checks/run_checks.py must pass: no blanks; control totals recomputed from raw files; season coverage; provenance sidecars and source listing; history freshness; report numbers match outputs; anonymity scan; and an external reality check: our SPP peaks for 2023 and 2026 must be within 1% of EIA's and SPP's published records ^[8][9].

How this was made

AI agents downloaded the public data, wrote and ran the scripts, drafted the text and ran the checks; a person directed the work and reviewed it. Every number comes from a script output in the data downloads. Corrections are logged in Amendments.