Solar Yield Method 2026: Data, Simulation and Known Limits

Every figure on this site comes from one of five data files that scripts in scripts/data build from official sources and date. Calculators and texts read only from these files. When a feed-in rate or electricity price changes, it changes on every page at once; the worked examples in the texts are recalculated on every build.

1. Radiation and yield: PVGIS at the grid point

For 838 grid points (every 0.2 degrees of latitude and 0.3 degrees of longitude, about 22 by 20 kilometres, wherever a postcode lies) we queried PVGIS 5.3 of the European Commission's Joint Research Centre on 15 October 2026 for the annual yield of a 1 kWp system: crystalline silicon, 14% system losses, 35 degrees tilt facing south, PVGIS-SARAH3 radiation data for 2005 to 2023, horizon from the digital terrain model. Values range from 843 to 1,185 kWh per kWp; the population-weighted mean is 1,057 kWh.

In addition we queried PVGIS for each of the 10,815 postcodes individually, at the centre of all places in the postcode according to GeoNames. The calculator uses that value; the grid now only assigns the region (orientation matrix, monthly profile) and the place name, the largest municipality in the grid cell according to the Destatis register. In the Alps the grid value would otherwise be off by up to a sixth, because valley and slope see a different horizon.

2. Tilt and direction

Germany is split into 28 regions of 1.5 by 2 degrees. For one reference point per region there are 81 further PVGIS queries: tilt from 0 to 90 degrees, eight directions. Your yield is the grid value at 35° south times the ratio of your orientation to 35° south in your region, interpolated linearly between the support points. East-west roofs are taken as half east, half west.

3. Self-consumption: 74 households, quarter-hour by quarter-hour

The basis is the representative load profile set of HTW Berlin: 74 households metered minute by minute in 2010 (IZES). We average them to 15 minutes and scale each profile to the annual consumption you enter. Generation comes from PVGIS hourly values for the same weather year at a reference location in central Germany (Kassel), for 20 orientation shapes. For each combination of yield, consumption and battery, the full year is simulated; the calculator shows the mean of the 74 households and interpolates between support points.

Plug-in solar: 800 VA inverter; output above that is clipped. A battery sits on the DC side, stores the surplus including the peak above 800 VA and releases it in the evening up to the inverter limit.
Rooftop system: home first, then battery, then grid. Without a smart meter, feed-in from systems commissioned from 25 February 2025 is capped at 60 % of rated output (section 9(2) EEG 2023); what exceeds the cap and cannot be stored is lost.
Battery: efficiency 94.9% each way (about 90% round trip), power 0.5 kW per kWh, at least 0.8 kW. The usable capacity is what you enter.

4. Money: saving, feed-in payment, payback

Self-used power saves the electricity price you enter; the default is 36.2 ct/kWh, the Eurostat average for German households using 2,500 to 4,999 kWh (2026-S1). Power fed in from a rooftop system earns the feed-in tariff for its commissioning period from the Federal Network Agency table (7 periods since 1 January 2023, currently 1 August 2026 to 31 December 2026), blended by capacity above 10 kWp. New systems with a smart meter receive nothing at negative prices; we estimate the share affected from the share of German solar power produced in hours with negative prices (2025: 573 hours, 24.1 % of solar output, SMARD). Power from plug-in devices is not paid for. Payback = purchase price divided by the first-year benefit; the 20-year total assumes a constant electricity price unless you enter a change.

5. Tests that must pass before publishing

Feed-in rates: all 7 periods against the Federal Network Agency spreadsheet; blended rates above 10 and 40 kWp recalculated by hand.
Locations: 16 postcodes from Hamburg to Oberstdorf against a second, independent PVGIS query; largest difference 0.1 %.
Orientation: values for upright, east, west and flat against our own PVGIS queries at the same place.
Energy balance: self-used + fed in + clipped = generated, in every combination; self-consumption rises with the battery and falls with system size.
Law: 2,000 Wp and 800 VA, the 60% limit by cut-off date and smart meter, negative prices under section 51(2), payments under section 52 EEG.

What the calculators cannot do

Measure your own shading: trees, neighbouring buildings and dormers are not in the terrain model; the shading option is a flat reduction.
Know your load profile: a household with a home office or heat pump differs from the mean. Heat pumps and electric cars hardly appear in the 2010 profiles. We checked the daily pattern against BDEW’s new H25 standard load profile (2025, built from the settlement data of 62 grid operators for 2018/2019): on June weekdays, 25.3 % of the 74 households’ daily use falls between 10:00 and 16:00, against 25.6 % in H25; from April to August the gap is at most 0.7 percentage points (script scripts/data/compare-h25.py). H25 does not replace the individual households: as an aggregate it smooths the peaks that self-consumption depends on, and it already contains some PV self-supply.
Capture short gaps in the clouds: generation is hourly, so clipping above 800 VA and the 60% limit are, if anything, slightly underestimated.
Future law: the draft EEG 2027 (printed paper 21/7867) is described but never applied as law in any calculator.
Individual tax cases, tenant electricity schemes, businesses and systems above 100 kWp.

Updates

A scheduled job reloads Federal Network Agency rates, Eurostat prices, SMARD market prices and the register counts every week and rebuilds the site when something changes. All texts are fully reviewed at least every 12 months and whenever the law changes. Please report errors to kontakt@solarrechnen.de.

Sources

City register figures: Market Master Data Register, retrieved on 15 October 2026, Data licence Germany attribution 2.0. Postcodes: GeoNames, CC BY 4.0.