“For us this was not a cosmetic update, it was a question of accuracy. Our customers put a six-figure investment case in front of their own clients, and they have to be able to stand behind it after the installation too.”
Green Energy Tools builds software that installers, developers and manufacturers use to calculate the economics of commercial battery storage. The workflow is always the same: a company's real metered load profile is uploaded in 15-minute values, the PV system is simulated for that exact location, and then we simulate every relevant use case. Self-consumption optimisation, peak shaving, atypical grid usage and procurement optimisation with dynamic electricity tariffs. The result is the storage size with the best payback, traceable down to the individual quarter hour.
The quality of that calculation depends, among other things, on the quality of the generation profile. That is why we switched data source.
Problem 1: hourly values smooth away reality
PVGIS provides irradiance data at hourly resolution. A commercial load profile, on the other hand, comes in quarter-hourly values, exactly as the grid operator measures it. To combine the two, the PV side has to be split into four identical quarter-hour values. That removes the short generation peaks and dips the battery could potentially have absorbed.
The effect always points in the same direction: a smoothed profile overstates the self-consumption of the PV system on its own and understates the increase in self-consumption the battery delivers, and therefore its economics. The difference between hourly and quarter-hourly values is not dramatic, but it is visible in the payback period. And it systematically works against the battery.
Problem 2: PV generation and power prices belong to the same year
PV generation and spot power prices are negatively correlated: when there is a lot of solar in the grid, prices are low. Anyone calculating with a typical meteorological year or a random historical year ignores that relationship, and gets several use cases wrong as a result.
On the procurement side: whether switching to a dynamic tariff pays off, and how much extra a battery extracts there, can only be assessed if the PV profile, the load profile and the spot prices all come from the same year. Otherwise the cheap midday hour in the price profile lands on a day when the simulated PV system produced nothing at all, and the battery is evaluated against hours that never existed.
The same applies on the feed-in side, three times over:
- For systems commissioned since February 2025, remuneration is dropped for every quarter hour with a negative spot price. What that costs depends on how much the system produces in exactly those quarter hours. Without generation and price being time-aligned, the effect cannot be traced.
- Systems from 100 kWp upwards sell through direct marketing and therefore the market premium model. Revenue there can be optimised by choosing when to feed in (green power storage). That, too, can only be assessed correctly with the right correlation between production and prices.
- From 2027 this becomes even more relevant. Under the draft EEG 2027, the fixed feed-in tariff disappears and the direct marketing obligation already applies from 25 kWp.
Current PVGIS data only goes up to 2023. But usually you want to run the analysis on data that is as recent as possible, typically the load profile and the spot prices from last year. And then you need the solar data from that same year.
Why pvnode
pvnode gives us both: time series at 15-minute resolution and historical years we can merge with real load profiles and real spot prices from the same period. Instead of a synthetic average year, our customers calculate on real years in which weather, consumption and prices match up.
For us this was not a cosmetic update, it was a question of accuracy. Our customers put a six-figure investment case in front of their own clients, and they have to be able to stand behind it after the installation too.
Working with pvnode has stayed straightforward. The integration was done in a reasonable amount of time, the API runs reliably, and when we have questions we get fast, technically sound answers. Exactly what you expect from a data provider that is deeply integrated into your simulation.
