What Germany's Solar Peak Act costs a new rooftop solar system, and which response actually pays off. Simulated hour by hour with real market prices and real weather.
Since 25 February 2025, new systems above 2 kW get no feed-in payment when the day-ahead price is negative, and may feed in at most 60% of their power until a smart meter is installed.
| Option | Gain / year | Cost | Payback | Unpaid kWh |
|---|
Gain is compared with the new rules, no battery and no smart meter. It includes lower electricity bills (self-consumption), feed-in income and the delayed payment for unpaid hours. Yearly smart-meter fees are already subtracted.
With 2025 prices, 19–23% of a 10 kWp system's production falls in negative-price hours in the five cities tested, yet the law costs only €78–85 a year. That energy was worth just 7.7 ct/kWh, and part of it is paid back after the 20 years.
For a south-facing system the 60% limit cuts only 30–260 kWh a year, worth far less than the meter's yearly fee of around €100.
Every stored kWh replaces grid power at 37 ct. Charging it in the negative-price hours instead of first thing in the morning cuts unpaid energy by 13–16%. A 10 kWh battery pays back in 11–14 years.
It lowers the midday peak, but produces 16–17% less energy over the year: it loses €90–130 a year against a south roof.
Hour by hour, one real year. Solar output per kWp for each city and roof layout comes from PVGIS (European Commission). Day-ahead prices for the DE-LU zone come from SMARD (Bundesnetzagentur). The negative hours match the official counts: 457 in 2024 and 573 in 2025. Household demand follows the BDEW H0 standard load profile.
2025 weather. PVGIS ends in 2023, so the 2025 profile keeps each month's measured energy and takes the timing of sunny days and hours from Germany's actual national solar generation in 2025. Tested out of sample (2022 → 2023), it reaches an hourly correlation of 0.83–0.93.
Peak-aware charging uses the day-ahead prices, which are published the day before, and assumes the day's solar output is known. It is an upper bound for a real controller.
Limits. A standard profile is the average of many homes: it is smoother than a single house, so self-consumption and battery gains are likely overestimated by around 10 percentage points. Heat pumps, electric cars and dynamic tariffs are not modelled yet. This is an engineering estimate, not financial advice.