A lifetime production estimate from a single reading

The first time you connect via a token, there's no history yet — only the current state. And yet something can already be said, on day one, about how a system has done over its entire operating life. This explains how, and why the result needs more caution than a monthly verdict does.

Where it comes from

The inverter keeps a cumulative counter of total production since installation (yieldTotal). Dividing it by the years of operation gives a rough annual average — without waiting to collect monthly data:

annual average = total production since install / years of operation

That's compared against what the system should deliver at its location according to PVGIS, giving a ratio — the same principle as the annual rule, just computed from a single number instead of a measured series.

The ageing correction

Panels produce a little less year after year — that's normal wear, not a fault. We apply a conservative degradation of 0.5 % per year, and not against the system's current age but its average age over the whole operating life (half the years since installation). For a one-year-old system the correction is practically zero; for a ten-year-old one it's around 2.5 %.

Without this correction, older installations would look worse than they actually are — compared against a new system's expectation rather than their real, slightly lower potential.

Three bands of result

Both the top and bottom outer bands are reported only as info, never a confirmed finding — above 120 % almost always means a mis-entered power or install date, not real overproduction (more in your solar is producing far more than expected).

Why this estimate never rises to a serious problem

This is the most important property of the whole rule. Even if the ratio falls deep below 75 %, the finding never exceeds warning — it's never marked a serious problem, however large the shortfall looks.

The reason: a single cumulative number mixes together years of different weather, gradual panel ageing, and often an incomplete first or last year of operation. From such an average you can't reliably tell real long-term underperformance apart from a mis-entered power rating, long-term shading, or simply unlucky weather in particular years. Claiming "serious problem" would assert more than this one number supports.

For the same reason the finding always carries low confidence — it's the one kind of diagnostic conclusion here where that's fixed by design, not derived from data quality. It's the opposite of a fault the inverter reports itself, which is the most certain finding we can give — there the device reports its own state, here we're inferring from one rough average.

What has to be true before an estimate is computed at all

• At least a year of operation. A shorter period isn't worth averaging — instead you see "collecting data".
• Installed power and install date entered. Without them there's nothing to compute from.
• Location (GPS). Without it expectation can't be computed via PVGIS, and no estimate is made at all — it isn't replaced with a worse one.

How this relates to the monthly verdict

The lifetime estimate is a temporary bridge, not the end state. It exists precisely so the first day after connecting isn't empty while enough monthly readings accumulate for a more accurate comparison. Once you have a full measured year, the annual rule takes over with real, not inferred, history — and it outranks this one in both confidence and strictness.

In practice: on day one you see a rough estimate at low confidence. After a year of watching, you see a verdict built on what we actually measured.

Where Slunora helps

Right after connecting an inverter you'll see this estimate as the first orientation figure. It's treated with exactly the caution it deserves — and once we've gathered monthly data, the verdict sharpens itself, with nothing for you to do.