# Routine supply and prolonged-outage cover

Reviewed 18 September 2026. This interpretation uses the unchanged 32 hourly cases in the archive-to-model review. It adds no dispatch runs or outage probabilities.

## Regular winter supply

The thirteen routine cases retain 30 days of scheduled maintenance for each reactor and a 98% output allowance outside maintenance. They exclude the additional prolonged Leibstadt outage. Their largest October–March supply gap is 1.5009337340832514 TWh, under 2016/17 weather and hydro conditions. This is the largest tested gap, not expected annual purchases or a final contract size. The separate data-centre demand allowance has not been run with routine operation.

## Immediate backup

Sudden plant loss requires a separate assessment of response speed, available power, grid constraints and restoration. The hourly energy model does not establish the required fast-response reserve. Its hourly power gaps must not be labelled frequency-response requirements.

## Prolonged-outage cover

The adverse cases remove Leibstadt for the first 90 calendar days of the ending year, in addition to scheduled maintenance and the normal output allowance. Ninety days is a chosen stress duration inherited from the earlier hourly model. It has no calibrated probability, recurrence interval or regulatory-standard status. Winter placement deliberately tests a demanding period.

The largest base winter gap is about 3.71 TWh, also under 2016/17 weather. The public chart uses the complete archived values. Three added-demand cases include about 0.9973 TWh/year beyond the data-centre growth already represented in the reference demand path. Across those three cases and the thirteen base outage cases, rounding the maximum winter gap upward to 0.1 TWh gives 4.2 TWh. Rounding the maximum annual hourly gap upward to 0.1 GW gives 2.5 GW. The energy and power maxima occur in different weather years.

Those figures describe a severe-contingency supply envelope including the routine deficit. They are neither expected annual production nor an economically optimized procurement recommendation. Adding the routine gap to this total would double-count energy. The difference between separate maxima also does not establish a final reserve contract.

## Shared boundaries and proposed action

All these comparisons allow up to 2 GW imports, retain 26 GWp solar, Gösgen and Leibstadt, finite pooled hydro and retained thermal/wind generation, and use fixed vehicle charging. Beznau and new nuclear receive no 2035 output credit. Perfect foresight, optimized cyclic reservoir starting stocks, smoothed monthly inflows and simplified grid operation remain optimistic assumptions. Added demand, lower imports and operating limits can raise supply needs.

The editorial recommendation is to seek separately priced regular winter energy, immediate backup and prolonged-outage cover. Reserve offers should distinguish payment for readiness from payment for activation, fuel and delivered energy. Owner contracts and operational access remain necessary. Compare outage durations and starting dates, credible import access, response costs and an explicit reliability objective before fixing reserve quantities. The archived cases provide useful bounds under their assumptions; they do not price or select that reserve.

Underlying records: `data/model-evidence-2026-09-18/case-index.json`, `portfolio_model/parameters.json`, `portfolio_model/run.py`, and the archived model-evidence method and reproduction bundle.