How to Read a PVsyst Report: P50, P90, Performance Ratio and Losses
Updated 23 October 2026 · SEO Dons Editorial
What does a PVsyst report contain?
PVsyst describes itself as a PC software package for the study, sizing and data analysis of PV systems. Its documentation says simulation results are summarised in a printable report with an exhaustive table of the parameters used and a short description of the main results. The report is assembled from forms chosen at printing, including general parameters, the horizon drawing and points table, the near shadings, the main results and the detailed loss diagram. A short report may simply leave the detailed forms out, so ask for all of them.
Check the kind of study too. PVsyst's preliminary design mode gives monthly values from a few general characteristics, without specifying actual system components, while project design uses detailed simulations with hourly or sub-hourly time steps. A proposal for a real roof should rest on the second. A project can also hold several simulation runs, which PVsyst calls variants, so confirm that the report is the variant behind the quoted figures.
What are specific production and performance ratio?
Specific production is the system's annual production per unit of installed capacity in kWh/kWp/year, and PVsyst places it on the main results page. The performance ratio (PR) is the energy effectively produced divided by the energy that would be produced if the system always ran at its nominal STC efficiency, as defined in IEC EN 61724 for grid-connected systems without storage.
They answer different questions. Specific production depends on place and orientation, so it is the number to check against PVGIS. PVsyst says PR is not directly dependent on weather data or plane orientation, so it compares the quality of designs in different places. PR takes in optical losses (shading, IAM, soiling), array losses and system losses such as the inverter.
PR is not a score to maximise. PVsyst notes that raising the albedo setting raises the sunlight counted on the panels and lowers the PR, and that east-west vertical bifacial layouts can pass 100%.
How do I read the loss diagram?
PVsyst says the loss diagram gives a quick look at the quality of a design by identifying the main sources of losses, and that it is always present in the report for the entire year. Each loss is a percentage of the previous energy quantity, so the percentages are not additive. A string of small losses cannot be summed by eye.
| Line in the diagram | Question to ask |
|---|---|
| Shading (horizon and near) | Is it built from a scene that matches the actual roof? |
| Soiling | What value, and why for this site? |
| Module quality and mismatch | Taken from the datasheet tolerance or chosen by the designer? |
| Wiring (DC and AC) | Default, or worked out from the actual cable runs? |
| Unavailability | Is it zero? |
For where the kWh go in more detail, see the guide to solar PV system losses.
Which weather and shading inputs does the report use?
Weather data
PVsyst's documentation calls weather data the starting point of a project and the main source of uncertainty. Its own comparison of sources finds big discrepancies between databases, although different sources typically agree within 10% of the mean. PVsyst recommends running the simulation with data from several providers to gauge the bias in one source.
The report should name the source, the period and the file type. A multi-year average file generally gives the P50, while a single specific year cannot be taken as representative of the P50.
Shading
PVsyst separates far shadings, described by a horizon line for objects typically more than about ten times the field size away, from near shadings, which it says require a detailed 3D description of the system and its surroundings. On a commercial roof, parapets, plant, stair cores, roof lights and the rows themselves are near objects that a horizon line cannot capture.
A missing horizon can be legitimate, because PVsyst ignores a horizon below 2 degrees and does not show it on the report. A missing near-shading scene on a cluttered roof is harder to explain. Ask which electrical treatment was used: PVsyst says linear shading is a lower limit of the full losses and the strings option a higher limit. The shading guide covers what shading costs.
How are P50 and P90 worked out?
PVsyst assumes annual yields follow a normal distribution centred on the simulated result, with a spread (sigma) found by combining the uncertainties in quadrature. P50 is exceeded with 50% probability and P90 with 90%, and for a normal distribution P90 lies 1.28 sigma below the mean. Weather variability dominates the spread.
As an illustration, with a P50 of 1,000 kWh/kWp and a sigma of 5%, the P90 is 1,000 x (1 - 1.28 x 0.05), which is 936 kWh/kWp. PVsyst puts typical year-to-year variability of annual irradiation at 3-4%, and PVGIS's own standard deviation of annual output at the five locations below is 2.6% to 4.1%.
Three limits matter. PVsyst says its P90 tool has no general method for the bias between weather sources, so a P90 from one source does not capture it. P50 and P90 are only meaningful for annual yields, not months. And long-term degradation is not compatible with the P90 concept, so ask how ageing is shown. If a funder or landlord wants a cautious figure, ask for the P90 with the uncertainty inputs behind it.
What are the five red flags in a PVsyst report?
- Optimistic or unnamed weather data. No source or period, or a single source with no comparison.
- Missing shading. No near-shading scene for a roof with obstructions, or only linear shading.
- Light losses. PVsyst says soiling is typically under 1% in moderately rainy climates and residential areas but can be significant in rural areas and several per cent in industrial zones. It also leaves module quality loss to the user, and its loss parameters start at defaults that it advises defining carefully.
- The wrong module file. PVsyst advises checking database parameters against recent datasheets, and modified modules are stored as PAN files. The module in the report should match the quoted one.
- Unrealistic availability. PVsyst lets the designer set unavailability as a time fraction, with the system treated as off in those periods. Zero means the forecast assumes the plant never stops.
The quote comparison checklist lists what else to put side by side, and the design software comparison covers reports from other tools.
How do I check the headline yield against PVGIS?
Divide the report's annual kWh by the kWp, then compare it with PVGIS for the nearest place and layout. The table uses PVGIS 5.3 with 1 kWp and 14% losses, the tool's default. The last column is the PVGIS yield divided by the sunlight on the panels, which is PR by PVsyst's definition. PVsyst measures PR before horizon loss, so treat this as a rough check.
| Location | South, 35 degrees (kWh/kWp/year) | East-west, 10 degrees (kWh/kWp/year) | Sunlight on the south panels (kWh/m2/year) | South yield divided by sunlight |
|---|---|---|---|---|
| Plymouth | 1,122 | 932 | 1,368 | 0.82 |
| London | 1,019 | 840 | 1,261 | 0.81 |
| Manchester | 886 | 741 | 1,096 | 0.81 |
| Belfast | 919 | 756 | 1,125 | 0.82 |
| Glasgow | 853 | 707 | 1,052 | 0.81 |
The ratio sits between 0.81 and 0.82. A PR well above that has assumed fewer losses than PVGIS's default, and one well below has assumed more or modelled heavy shading. The report should show why. PVsyst's own comparison finds weather sources typically within 10% of their mean, so a gap of that order between a report and PVGIS can come from the data alone, while a gap well beyond it needs another explanation, such as lighter losses. Thirteen locations are on the regional yield page, and the commercial PV systems guide explains the components behind the report. Readers who want a roof looked at against these checks can send an enquiry through the quote form.
Sources
- PVsyst SA, PVsyst 8.1 documentation: Overview; Results; Loss diagram; Performance ratio; PR warranty; Normalised performance index; P50 - P90 evaluations; Shadings; Far shadings; Near shadings; Weather data; Provider comparison; Array and system losses; Soiling; Module quality; Unavailability; Modules database, checked 22 October 2026.
- European Commission Joint Research Centre: PVGIS 5 user manual, and PVGIS 5.3 PVcalc runs for 1 kWp at 14% losses (south 35 degrees and east-west 10 degrees, the latter as the mean of an east and a west run), checked 22 October 2026.
Percentages, ratios and the P90 example are illustrative arithmetic.
Common questions
Does a higher performance ratio mean a better system?
Not always. PVsyst notes that more albedo gain raises the sunlight counted on the panels and lowers the PR, and that vertical east-west bifacial layouts can pass 100%. Compare PR between similar designs, and use specific production for yield.
Can a monthly P90 be quoted?
PVsyst says a P90 for monthly, daily or hourly totals is not statistically meaningful, because monthly variability is much larger than annual and a reliable P90 needs a long weather record. Quote P90 for the year.
What if the proposal uses different software?
The same questions apply: weather source and period, shading scene, loss assumptions, module and inverter files, and availability. Ask the designer to state each in the document.