Sequence changes ideal storage accounting.

Nominal source-profile model, not a physical French storage estimate. Each reported half-hour label is assigned 0.5 nominal hours. The fixed daily label grid does not establish a physically complete UTC timeline across daylight-saving changes. The initial physical-time design was stopped before capacity results; its failed check and the pre-outcome revision remain available.

At the same peaks and variance, the order of supply and demand can change the energy capacity required by a fixed balancing model.

10.7%

Median extra nominal capacity in the reported order relative to the median of within-month day reorders, across ten year-labelled profiles in the primary 70% wind / 30% solar scenario.

10 / 10

Primary year-labelled profiles above their reordered-reference median. This is a descriptive result, not an independent-years significance test.

The check

We used 175,344 complete source-labelled actual-value rows from RTE’s 2015–2024 profiles. Wind and solar shapes were scaled to a hypothetical 70/30 annual-energy mix equal to model demand. For each year-labelled profile, 999 reorders changed the sequence of whole days within each month while preserving all values, matched demand and generation, within-day patterns, monthly totals, annual variance and peak.

A useful result with a narrow scope

The primary original/reference ratios ranged from 1.0269 to 1.2781. Removing the two daylight-saving transition dates gave a median ratio of 1.0972; zero-clipping the signed generation inputs gave 1.1089. Other wind/solar mixes changed the magnitude; one 90/10 profile lay below its reference median. All five scenarios are reported.

capacity primary — Nominal-profile model; reference bars describe the central 95% of within-month day reorders, not uncertainty in a real-grid estimate. Full methods and every scenario appear in the report.

What follows

Peaks and variance alone do not determine the ideal cyclic energy requirement of these constructed sequences. This is compatible with the original proposal, which already includes duration; it does not prove its causal explanation, establish a new theorem, or estimate actual French storage needs.

Read the complete methods, results and limits →

Download the full reproducibility package

Download the report · Inspect source and file hashes