# Storage sizing, flexible demand and stress results

This extension adds 36 deterministic hourly cases (315,936 modeled hours) using the existing independent demand and weather inputs. It improves the hydro representation, sizes battery power and energy separately, and tests flexibility and adverse conditions. Every result remains conditional; the search does not optimize generation construction costs or certify a 2050 plan.

**Main finding:** storage need depends strongly on accessible hydro and demand timing. In the 2011/12 weather case, eight assumed 1.1 GW nuclear units plus 26 GWp solar require approximately 5.26 GW/29.25 GWh of batteries under the base decomposition. Road flexibility reduces this to 2.25 GW/11.19 GWh; adding the thermal-shift proxy gives 1.25 GW/5.19 GWh. These reductions rely on unverified participation and thermal assumptions and exclude their costs.

## Selected sensitivities

All rows below use 8.8 GW hypothetical nuclear, 26 GWp solar and the hydro year ending September 2012. Zero imports are allowed except in the explicitly connected case.

| Case | Battery GW | AC energy GWh | Unserved GWh |
|---|---:|---:|---:|
|Base finite-pump decomposition|5.262|29.25|0.000|
|No battery|0.000|0.00|216.685|
|Battery constrained to four hours|7.313|29.25|0.000|
|50% road energy movable within each day|2.252|11.19|0.000|
|Road flexibility plus 1 GW/6 GWh thermal proxy|1.252|5.19|0.000|
|75% residual natural energy capacity|7.092|354.34|0.000|
|50% residual natural energy capacity|7.092|2481.97|0.000|
|75% hydro turbine/pump power|8.240|393.43|0.000|
|Combined weather/outage/stock stress|8.641|4705.43|0.000|
|Cold-weather heating uplift|11.309|998.28|0.000|
|+365 GWh conditional pooled hydro expansion|5.262|29.25|0.000|
|Extra 1 GW/40 GWh unmapped pump partition|6.570|44.39|0.000|
|Up to 2 GW imports permitted|3.262|17.25|0.000|

A four-hour solution is 7.31 GW/29.25 GWh, approximately CHF 7.31–14.63 bn initially at the assumed 250–500 CHF/kWh band. The unconstrained-duration capital proxy gives about CHF 9.74 bn centrally. These are scenario allowances, not Swiss bids; financing, replacement, exceptional connections and owner revenue must be evaluated separately.

Restricting power to 1,2 or 4 GW leaves 141.49,78.44 or 11.92 GWh unserved respectively even after energy sizing. Increasing an energy tank cannot cure an insufficient discharge-power bottleneck. The conditional 365 GWh hydro expansion does not reduce base-case battery size in this particular profile, although it changes flows. That is not evidence that the projects have no value; this simplified expansion adds pooled energy without new annual inflow, power or mapped pump routes.

With 75% usable residual natural energy, required battery energy rises to 354 GWh; with 50%, 2,482 GWh. The compound case reaches 4,705 GWh. These flag a seasonal adequacy problem. Applying a linear lithium-ion cost formula to multi-TWh stores is not credible procurement evidence and should not be interpreted as an executable investment proposal.

## Weather and portfolio comparison

The 8.8 GW/26 GWp portfolio needs 29.25,10.35,0.45 and approximately zero GWh in the 2012,2018,2021 and 2022/2023/2024 base cases respectively. This is six historical weather years, not a statistical guarantee. The larger solar/smaller nuclear mixes often require seasonal rather than overnight storage; some hit the 20 GW/10,000 GWh search caps and still shed load. All mixes omit fossil backup in these tests.

Four additional cases add an assumed 2 GW of wind and both flexibility options to the 3.3 GW/65 GWp and 0 GW/90 GWp mixes. They reuse actual Swiss fleet hourly wind shapes for 2022 and 2024, normalize toward 2,500 full-load hours and clip at nameplate. This is a small-fleet proxy, not a future siting/yield assessment. It improves those portfolios, but leaves substantial modeled storage needs. The actual input series have 8,760 and 8,784 hours, no missing wind values, and hydro-year totals 145.00 and 190.00 GWh respectively. [energy-charts-api](<https://api.energy-charts.info/openapi.json>); [ch-public-power-2022](<https://api.energy-charts.info/public_power?country=ch&start=2021-10-01&end=2022-09-30>); [ch-public-power-2024](<https://api.energy-charts.info/public_power?country=ch&start=2023-10-01&end=2024-09-30>).

| Case ID: n = number of 1.1 GW units, pv =GWp; year ends September | Battery GW | GWh | Unserved GWh | Net exports TWh |
|---|---:|---:|---:|---:|
|n8-pv26-2012-split|5.26|29.25|0.000|21.09|
|n8-pv26-2018-split|2.39|10.35|0.000|20.95|
|n8-pv26-2022-split|0.00|0.00|0.000|20.85|
|n6-pv40-2012-split|14.08|2413.09|0.000|16.86|
|n6-pv40-2018-split|19.51|2854.02|0.000|16.47|
|n6-pv40-2022-split|16.10|1046.58|0.000|15.83|
|n3-pv65-2012-split|20.00|8687.45|0.000|15.56|
|n3-pv65-2018-split|20.00|10000.00|67.080|15.99|
|n3-pv65-2022-split|20.00|7204.00|0.000|15.40|
|n0-pv90-2012-split|20.00|10000.00|4199.569|16.03|
|n0-pv90-2018-split|20.00|10000.00|5698.763|16.26|
|n0-pv90-2022-split|20.00|10000.00|2697.961|16.12|
|n8-pv26-2012-battery_none|0.00|0.00|216.685|21.08|
|n8-pv26-2012-battery_4h|7.31|29.25|0.000|21.13|
|n8-pv26-2012-road_flex|2.25|11.19|0.000|19.58|
|n8-pv26-2012-thermal_flex|1.25|5.19|0.000|19.58|
|n8-pv26-2012-hydro_75pct|7.09|354.34|0.000|19.14|
|n8-pv26-2012-hydro_50pct|7.09|2481.97|0.000|19.14|
|n8-pv26-2012-compound|8.64|4705.43|0.000|13.44|
|n8-pv26-2012-hydro_expansion|5.26|29.25|0.000|21.41|
|n8-pv26-2012-pump_residual|6.57|44.39|0.000|21.18|
|n8-pv26-2012-cold_heat|11.31|998.28|0.000|20.48|
|n8-pv26-2012-connected_2gw|3.26|17.25|0.000|20.90|
|n8-pv26-2012-hydro_power_75pct|8.24|393.43|0.000|20.98|
|n8-pv26-2021-split|0.29|0.45|0.000|21.49|
|n8-pv26-2023-split|0.00|0.00|0.000|21.70|
|n8-pv26-2024-split|0.00|0.00|0.000|22.28|
|n8-pv26-2012-power1|1.00|6.00|141.488|20.82|
|n8-pv26-2012-power2|2.00|12.00|78.436|20.82|
|n8-pv26-2012-power4|4.00|24.00|11.921|21.00|
|n8-pv26-2012-power8|8.00|29.25|0.000|21.13|
|n8-pv26-2012-power10|10.00|29.25|0.000|21.13|
|n3-pv65-2022-wind2-flex|20.00|4287.03|0.000|16.76|
|n3-pv65-2024-wind2-flex|19.13|887.93|0.000|16.58|
|n0-pv90-2022-wind2-flex|20.00|9022.20|0.000|16.87|
|n0-pv90-2024-wind2-flex|20.00|5963.98|0.000|16.76|

**Net exports can coexist with unserved hours.** The table makes that failure visible: annual energy balance is not hourly adequacy. Wind/flex cases change several assumptions together and therefore do not isolate wind alone. There is no gas, new thermal generation or imported offshore wind credited.

## Method and limits

The LP first minimizes unserved energy with a very large penalty and an assumed initial battery capital proxy. It then holds the selected capacity fixed and dispatches to minimize shortages, imports and storage throughput. Power and energy have separate costs and bounds. This is a fixed-generation storage-sizing screen, not a total-system least-cost optimization.

Hourly bus balance, cyclic annual stocks, finite power/energy, 85% battery and 80% pumped round-trip efficiency are enforced. State is AC-deliverable energy: charging is multiplied by efficiency and discharge is not reduced a second time. Initial stock is optimized subject to annual closure, except for the compound fixed-initial-stock case. Perfect year-ahead foresight is optimistic. Independent checks audit simultaneous charging/discharging, not just the solver status.

The model retains the prior monthly hydro inflow/river profiles, solar weather proxy, 6% final-to-bus loss conversion, scheduled nuclear maintenance and one 90-day unit outage. Residual natural hydro is pooled nationally; there is no internal transmission network. The separate 75% hydro-power case is a uniform derating sensitivity, not an N-1 network simulation.

Road flexibility preserves each local calendar day’s road energy, including daylight-saving days. Half of all modeled road energy is flexible; connected charging power is capped at three times its daily mean and limited to chosen home/day/evening windows. This includes freight and should not be mistaken for observed car-owner participation. Thermal flexibility is a 1 GW/6 GWh electrical-equivalent buffer at 95% efficiency; discharge cannot exceed space-heating load. It does not model building temperatures, emitter constraints or COP explicitly.

A 2,003-driver BFE survey reports 36% stated willingness at CHF 2/month and 79% at CHF 20/month; EKZ’s pilot involves roughly 500 participants. Neither establishes a national quantity of flexible GW. [bfe-ev-survey-2025](<https://pubdb.bfe.admin.ch/de/publication/download/12299>); [ekz-smart-charging-2025](<https://www.ekz.ch/ekz/de/ueber-ekz/medien/medien/medienmitteilungen/Elektromobilitaet--Pilotprojekt--Smart-Charging-.html>).

Cold heat adds up to 50% more space-heating electricity as weighted outside temperature falls from 0 to−10°C. The compound case combines 20% lower natural inflow/river output, 10% less solar, 75% natural energy capacity, 70% initial natural stock and a second January–March nuclear outage. These are stress assumptions without assigned occurrence probabilities.

Reproduce with `python -m feasibility_model.run` from the project root. Existing completed case directories are reused; change case IDs or use a new versioned output directory for changed inputs. Code: [run.py](</home/niko/Documents/swiss-energy/feasibility_model/run.py>), [dispatch.py](</home/niko/Documents/swiss-energy/feasibility_model/dispatch.py>). Inputs/results: [case index](</home/niko/Documents/swiss-energy/data/system-feasibility-2026-09-10/case-index.json>); each case folder includes its complete compressed hourly ledger.