Swiss electrification-demand calculator — first results, 10 September 2026

The annual calculator reconstructs the detailed 2024 national final-electricity balance of **57.512 TWh** and produces editable sector scenarios through 2050. The central illustrative case is **93.4 TWh/year in 2050**. These are conditional calculations, not forecasts, confidence intervals or a generation plan.

| Year | Lower demand, TWh | Central, TWh | Higher demand, TWh |
| --- | ---: | ---: | ---: |
| 2024 | 57.5 | 57.5 | 57.5 |
| 2035 | 60.6 | 73.5 | 92.6 |
| 2040 | 62.3 | 82.2 | 113.2 |
| 2050 | 60.8 | 93.4 | 147.2 |

![Annual demand cases](/home/niko/Documents/swiss-energy/outputs/demand-model-v1/annual-demand.png)

The cases share the same broad technology-replacement targets. They differ in activity, population, efficiency, heat-pump performance, cooling and domestic charging allocation. Their spread therefore exposes assumptions rather than establishing statistical likelihood. All three default cases leave explicit additional data-centre, hydrogen/e-fuel and district-heat-production electricity at zero. Existing electricity for these uses remains in the sector baseline where represented. The zero increments do not establish zero future growth; the calculator supplies separate controls and sensitivity cases.

The central 2050 composition is:

| Component | TWh/year |
| --- | ---: |
| households | 22.30 |
| services | 22.94 |
| industry | 25.33 |
| transport | 21.55 |
| unallocated | 1.24 |
| explicit additions | 0.00 |

![Sector changes](/home/niko/Documents/swiss-energy/outputs/demand-model-v1/sector-demand.png)

The central population assumption is 10.5 million, road activity rises 15%, useful space heat per unit of activity falls 25%, and new space-/water-heating seasonal performance ratios are 3.3/2.7. New car consumption is assumed at 0.18 battery kWh/km, charging efficiency 90%, and Swiss charging energy is 95% of that associated with territorial vehicle-km. These are visible analyst assumptions. None is presented as a measured 2050 Swiss fact. Published heat-pump measurements give context, but no single house or unweighted sample supplies a national coefficient.

All identified household resistance heating and the chosen oil/gas/coal building-heat pool are replaced by the end of the default adoption path. Industrial conversion covers 90% of oil/gas/coal process heat, 20% of wood and 25% of mixed other process energy. The assumed industrial heat-pump share is checked against published carrier/temperature marginal bounds. Passing those bounds establishes arithmetic feasibility, not site-level feasibility. Wood, district heat and some industrial fuels remain; this is not a claim of complete fossil elimination.

**What the baseline reconciliation establishes.** The archived [BFE electricity statistics](https://pubdb.bfe.admin.ch/de/publication/download/12228), table 6, give 81.054 TWh generation plus 25.955 imports minus 40.354 exports = 66.655 TWh. Subtracting 4.817 pumping and 4.326 published losses gives 57.512 final electricity. The end-use workbook reports 207.04 PJ, or 57.511111 TWh; its 0.000889 TWh difference is consistent with published rounding. The end-use sector model sums to 56.270214 TWh. The calculator therefore carries a separate 1.241786 TWh baseline residual. Its numerical cause is reconciled; its physical allocation remains unresolved. The central case explicitly holds it constant and the other cases vary it.

There is also a concrete sector-classification bridge. Electricity Statistics table 21 places 0.272 TWh of public lighting and 1.148 TWh of other infrastructure in transport. The [end-use methods report](https://pubdb.bfe.admin.ch/de/publication/download/12358), sections 3.1.2 and the service-sector methods, places these activities in services. Moving those published categories reduces the transport difference to about 0.000015 TWh. Remaining household, service and industry differences are retained; sectors are never silently rescaled.

**What the observed load comparison establishes.** We extracted all 35,136 quarter hours and 8,784 hours from the archived Swissgrid 2024 workbook. Every source timestamp matches the Europe/Zurich calendar, including both daylight-saving transitions, and summed energy agrees with the workbook summary. Metered end-user withdrawals total 53.369089 TWh, which is 4.142911 TWh below the BFE national final balance. The metered control-block curve has a different coverage and must not be relabelled as complete national consumption. Its observed hourly maximum is 9.401 GW. January–March plus October–December account for 29.421 TWh; these are six calendar-year months, not one contiguous hydrological winter.

**Sensitivity and practical limits.** The [sensitivity dataset](/home/niko/Documents/swiss-energy/data/demand-model-v1/one-at-a-time-sensitivity.json) changes each parameter separately. Optional hydrogen/e-fuel electricity adds one-for-one to final demand. Service and industrial activity, general electrical efficiency, vehicle energy use and heating performance all materially affect the result. The exact ranking depends on the chosen parameter ranges.

![Sensitivity to selected assumptions](/home/niko/Documents/swiss-energy/outputs/demand-model-v1/sensitivity.png)

Annual outputs are on the 2024 weather basis, with an explicit annual weather-stress multiplier. They do not calculate hourly heat-pump capacity, future cold peaks, future winter imports, storage dispatch, generation adequacy or net exports. Domestic non-road/navigation conversion and international aviation are not represented as explicit conversion modules. Foreign van activity and services/industry heating-auxiliary allocation remain material boundary uncertainties. Physical causes of the remaining electricity residual and the complete Swissgrid-to-national load bridge are also unresolved. These are listed in the result files rather than filled with invented measurements.

Implementation checks cover reconstruction, energy accounting, replacement without double counting, unit conversion, input rejection, industrial feasibility bounds and scenario totals. They establish accounting and software correctness. Predictive validation against withheld years has not been performed. The next model stage is a geographically weighted, hourly demand model validated across historical weather years, followed by generation/storage comparisons using the same demand cases.

The [methods and usage guide](/home/niko/Documents/swiss-energy/aidocs/022_demand_model_methods_and_usage.md) explains the editable inputs and equations. [Annual results](/home/niko/Documents/swiss-energy/data/demand-model-v1/annual-demand.csv), [component results](/home/niko/Documents/swiss-energy/data/demand-model-v1/annual-components.csv), [central case](/home/niko/Documents/swiss-energy/data/demand-model-v1/central-2050.json), [baseline evidence](/home/niko/Documents/swiss-energy/data/demand-model-v1/baseline.json) and [validation](/home/niko/Documents/swiss-energy/data/demand-model-v1/validation.json) are local and reproducible. Earlier archives are preserved; nothing was published or sent to third parties.