What PR measures
PR answers one question: of the energy the sun made available to this plant, how much reached the meter? A plant with a PR of 78% delivered 78% of its theoretical output at standard test conditions, after accounting for how much irradiation it actually received.
Because it is normalised for irradiation, PR lets you compare a sunny week with a cloudy one, and one site with another. It does not, on its own, tell you why the remaining energy was lost. That is the job of loss analysis.
The formula
The common definition, consistent with IEC 61724-1, is measured AC energy divided by a reference energy based on plane-of-array irradiation and installed DC capacity:
Always state three things alongside a PR figure: which meter the energy comes from, which sensor provides irradiation, and which DC capacity value you used. Two teams using different conventions can report PRs several points apart for the same plant on the same day.
Worked example
Plane-of-array irradiation: 6.2 kWh/m². Installed DC capacity: 50,000 kWp. AC energy at the plant meter: 236,000 kWh.
Reference energy = 6.2 ÷ 1 × 50,000 = 310,000 kWh.
PR = 236,000 ÷ 310,000 = 76.1%.
If the design expectation for this month is 78.6%, the plant is about 2.5 points short. That gap is what loss analysis should explain.
What moves PR
| Factor | Effect on PR | What to check |
|---|---|---|
| Module temperature | Crystalline silicon loses roughly 0.3–0.4% of power per °C above 25 °C, so PR dips on hot days | Module temperature sensors; compare with temperature-corrected PR |
| Soiling | Dust blocks irradiance before it reaches the cells; losses build between cleaning cycles | PR trend between cleanings; soiling sensors or reference strings |
| Plant downtime | Inverter trips, string and SCB faults remove generating capacity | Inverter availability, string currents, alarm history |
| Clipping | When DC power exceeds inverter AC rating, the excess is lost | Inverter output flat at rated AC power around noon |
| Grid outages and curtailment | Energy that could not be exported lowers PR unless excluded | Grid availability records; agreed exclusion rules |
| Sensor problems | A dirty or misaligned pyranometer inflates or deflates PR | Compare pyranometers; cleaning log; irradiance vs output shape |
| Wrong DC capacity | Using a nameplate that doesn't match the as-built plant skews every PR | Asset register vs commissioning records |
PR vs CUF vs specific yield
| Metric | Question it answers | Weather-normalised? |
|---|---|---|
| Performance ratio (PR) | How efficiently did the plant convert available sunlight? | Yes |
| Capacity utilisation factor (CUF) | How much energy did the plant produce relative to running at full capacity all the time? | No |
| Specific yield (kWh/kWp) | How much energy per installed kWp? | No |
Use PR to judge operational performance, and CUF or specific yield to judge commercial output. A plant can have a healthy PR and a low CUF in a cloudy month, or the reverse.
Common mistakes
- Averaging daily PRs. To get a monthly PR, sum energies and irradiation for the month, then divide. Averaging daily percentages over-weights cloudy days.
- Using GHI instead of POA irradiation. Global horizontal irradiance ignores panel tilt and tracking, and makes PR look wrong by a large margin.
- Ignoring sensor cleaning. An uncleaned pyranometer under-reads irradiance and makes PR look better than it is.
- Silent exclusions. Excluding grid outages or curtailment is fine if the rule is written down and applied every time.
- Comparing across conventions. PRs from different meters or capacity values are not comparable.
How operations teams should use PR
- Check portfolio and plant PR every morning against the expected value for the day's conditions.
- When a plant is below expectation, drill down to blocks and inverters to find where the gap sits.
- Break the gap into loss categories: downtime, soiling, temperature, clipping, grid.
- Turn credible, material losses into work orders with an owner and due date.
- After the fix, check that PR recovers on comparable days before closing the issue.
How nxtGrid helps
- Plant technical dashboard with PR, CUF, availability and specific yield alongside live power against irradiance
- Loss bucketing and a loss waterfall from target to actual generation
- Per-inverter power heatmaps and top and bottom inverter rankings to find where PR is lost
- Daily generation reports and loss reports generated automatically
- nxtGrid AI: ask “why is PR low at this site today?” in plain English