P50, P90, and uncertainty
Probability figures and the uncertainty table.
Reviewed
What this is for
An uncertainty statement, not a second simulation. A typical meteorological year is one synthesised year with no inter-annual spread in it. P90 is the modelled P50 read against a combined standard uncertainty, not a second 8,760. Treating it as one is how a bank gets a number nobody can reproduce from the weather file.
Steps
- On a finished result, Production probability lists exceedance levels against annual production, and each level against P50. The run's own choice, set as Production probability on the condition set's Weather tab, is the marked row. The headline tile is that same adjusted energy at the combined uncertainty.
- Under the table, six contributors:
- Weather variability (a property of the site's record).
- PV module modelling parameters.
- Inverter efficiency.
- Soiling and mismatch.
- Degradation estimation.
- Custom variability.
- A dash is a contributor nobody has stated, not one measured at nothing. The five modelling terms start unstated. Until a firm enters its own budget on the condition set, the combination is the weather term alone.
- Combined in quadrature (the root of the sum of squares), because the six are independent. Adding them overstates the combination as soon as more than one is stated. Five separate one-percent modelling terms come to
2.24%, not5%. - The printable report carries the same table on its last page, in the same words.
Until a pairing has been run, there is no table. The matrix cell says Not ready.
Tips and common mistakes
- Do not invent a P90 by knocking a percentage off annual energy. The table is the definition.
- Filling an unstated term with zero "to be conservative" is the opposite: it shrinks the combination and looks measured.
- Weather variability is not soiling. One is the year-to-year spread of the record; the other is dirt on this installation.