Solar Performance Ratio: How to Calculate and Interpret PR

Most articles on performance ratio give you the formula and stop there. PR = E_AC ÷ (H_POA × P_nom). Done. But knowing the formula and knowing what to do with a PR of 71.4% on a Bologna C&I system in July are entirely different problems. This post closes that gap.

What follows is a step-by-step calculation tutorial — three fully worked examples at different scales — plus an interpretation framework that maps any PR result to a specific diagnosis and action. If you want the formula derivation and benchmark context before the numbers, start with Solar Performance Ratio: Formula and Benchmarks. If you are ready to calculate and interpret, stay here.

In this guide:

• What performance ratio actually measures — and where the formula has hard limits
• The four time intervals and why monthly is the right default
• Step-by-step input gathering with common data mistakes
• Worked Example 1: 6 kWp residential system in Munich — monthly table and annual PR
• Worked Example 2: 250 kWp C&I system in Bologna — summer dip explained
• Worked Example 3: 5 MWp utility-scale plant in Rajasthan — MWh units and high-PR winter months
• A six-band decision framework to diagnose any PR result

What Performance Ratio Actually Measures (and What It Doesn’t)

Performance ratio is a dimensionless efficiency index defined in IEC 61724-1:2021 as the ratio of actual AC energy yield to the energy the system would have produced if it operated at its rated STC efficiency for every hour of incident irradiance. The formula:

PR = E_AC / (H_POA × P_nom)

Where E_AC is measured AC output in kWh, H_POA is POA irradiance in kWh/m², and P_nom is the DC nameplate rating in kWp.

The key insight is what PR removes from the equation. By dividing AC output by H_POA × P_nom, you strip out the effect of how much sun the site receives. A system in Munich and a system in Seville running identically well will produce very different absolute kWh values — but they will show nearly identical PR values. This site-independence is what makes PR the standard metric for comparing system quality across geographies, benchmarking against contract guarantees, and flagging underperformance that is not explained by weather.

PR does not measure everything. Three things it deliberately excludes:

1. Irradiance level. High irradiance causes higher module temperatures, which suppress output. PR will be lower on hot sunny days than on cool bright days — this is physically correct behavior, not a flaw.
2. Curtailment. If an inverter clips at high irradiance or a grid operator curtails export, that lost energy appears as lower PR even though the hardware is functioning normally.
3. System size. A 6 kWp residential system and a 5 MWp utility plant can both show PR of 78%. PR says nothing about absolute output.

PR can exceed 100% in cold, high-irradiance conditions — typically winter days with fresh snow reflection. This is not a measurement error. Modules operating well below 25°C produce more than their STC-rated output. PR of 103–108% for a single day in January in northern Europe is documented. Annual PR above 100% is physically impossible for a correctly sized system over a full year.

For systems where shading is a variable — adjacent structures, trees, seasonal shadow angles — the solar shadow analysis software used during design should produce a POA-corrected irradiance estimate that accounts for shading losses before they appear in the PR calculation. If you are back-calculating PR from metered data, shading that was not modeled at design stage will silently compress PR without any obvious fault.

For a deeper treatment of temperature-corrected PR methodology, see Solar Performance Ratio: Formula and Benchmarks.

The Four Time Intervals: Instantaneous, Daily, Monthly, Annual

PR is not a single number — it changes depending on the time window you use to calculate it. The four standard intervals each serve a different diagnostic purpose.

NREL research (Dierauf et al., 2013) shows seasonal PR variation of ±10% around the annual mean is normal for crystalline silicon systems in continental climates. A German system showing PR of 82% in February and 72% in August is performing normally. The same 10% spread seen month-on-month within a single season is not normal and warrants investigation.

Monthly is the recommended standard for operational monitoring. It is granular enough to catch a failing string or a soiling event, but stable enough to avoid false alarms from single-day weather anomalies. IEC 61724-1 uses monthly PR as the default reporting interval for Class B and Class C monitoring systems.

Note on PR above 100%: on cold, clear winter days — particularly in alpine or northern European locations — daily PR of 103–110% is physically possible. Modules operating at -5°C to 0°C produce output above their 25°C STC rating. This does not indicate a sensor error unless it persists for full months.

Step 1: Gather the Inputs You Actually Need

Three inputs are required for any PR calculation:

1. AC energy output (kWh) Read from the revenue-grade meter or the inverter’s data logger. Use the AC side — never DC. The common mistake is pulling DC string data from a monitoring portal and calling it E_AC. DC-side data misses inverter losses (modern string inverters typically run at 96 to 98% European-weighted efficiency per PVsyst documentation) and will inflate PR by 2 to 4%.

2. POA irradiance (kWh/m²) Measured by a pyranometer or reference cell mounted in the plane of the array — same tilt and azimuth as the modules. This is the most frequently corrupted input. The mistake: substituting GHI (global horizontal irradiance) from a nearby weather station. For a 30° south-facing array, GHI understates POA by 10–20% depending on latitude and season. The resulting PR appears 10–20% higher than the system actually deserves.

If you have GHI and need to convert, the process is called irradiance transposition and requires tilt angle, azimuth, latitude, and a transposition model (Perez or Hay-Davies).

3. System nameplate (kWp) Use the DC STC nameplate from the datasheet — the sum of all installed module Wp ratings. The common mistake is using the inverter AC rating or the as-built MWp rounded figure from the EPC contract. A 250 kWp array connected to a 220 kVA inverter is still a 250 kWp system for PR purposes. Using 220 kWp in the denominator inflates PR artificially.

IEC 61724-1 Monitoring Class Reference

IEC 61724-1:2021 defines three monitoring classes that specify the minimum sensor requirements for reliable PR calculation:

Low-irradiance filter: IEC 61724-1 specifies that data points with POA irradiance below 50 W/m² should be excluded from PR calculations. At very low irradiance, inverters may be at the edge of their operating range, and the PR formula amplifies small measurement errors into large swings. Filter these out before computing monthly totals.

Worked Example 1: 6 kWp Residential System, Munich

Munich sits at 48°N with a continental climate — moderate irradiance, cold winters, warm summers. A well-installed 6 kWp south-facing rooftop system with a single-phase string inverter and no shading should show PR in the 72–78% range annually.

Here is the 12-month dataset for this system:

Walking through June:

PR = E_AC ÷ (H_POA × P_nom) = 710 kWh ÷ (160 kWh/m² × 6 kWp) = 710 ÷ 960 = 0.740 = 74.0%

June is the lowest-PR month of the year, yet June also produces the highest absolute AC output (tied with July). This is the temperature effect in action. Module temperatures in Munich in June regularly reach 55 to 65°C, and crystalline silicon modules lose 0.3 to 0.5%/°C above 25°C (NREL, 2013). At 60°C module temperature, that is a 10.5–17.5% power reduction from STC rating — which maps directly to compressed PR.

Why winter PR is higher:

January PR is 0.722, which is only marginally below June despite much lower absolute output. The denominator is much smaller in January (45 kWh/m² vs. 160 kWh/m²) and so is the numerator (195 kWh vs. 710 kWh). But the ratio is preserved because modules operating at 0–10°C are significantly more efficient than at 60°C. The temperature coefficient effect runs in your favor in winter.

Annual:

PR = 5,595 ÷ (1,255 × 6) = 5,595 ÷ 7,530 = 0.743 = 74.3%

This is solid performance for Munich. The annual result is shaped primarily by summer months when both output and irradiance are high — the high-irradiance summer months dominate the weighted average.

Worked Example 2: 250 kWp C&I System, Bologna

Bologna sits at 44°N in the Po Valley — hotter summers than Munich, higher annual irradiance, and significant soiling risk from agricultural dust in the region. A well-maintained flat-rooftop commercial system at 15° tilt facing south should achieve 75–80% PR annually.

Here is the 12-month dataset:

Sanity check: 270,425 ÷ (1,430 × 250) = 270,425 ÷ 357,500 = 0.7565 ✓

Walking through July:

PR = 33,763 ÷ (185 × 250) = 33,763 ÷ 46,250 = 0.730 = 73.0%

July is the lowest-PR month in Bologna — and for the same reason as Munich: module temperatures. In Bologna’s summer, ambient temperatures regularly exceed 35°C. At that ambient, module temperatures under full irradiance can reach 70–75°C, compressing output by 13–20% relative to STC. The inverters also run hotter, slightly reducing efficiency.

Interpreting July vs. April:

April PR is 0.790 — the highest month. Module temperatures in April in Bologna are close to 25°C, so the system operates near STC conditions. The 6 percentage point difference between April and July is almost entirely explained by temperature. Soiling loss in the Po Valley agricultural environment typically adds 2 to 4% additional loss in summer (industry-observed range; see NREL PV Soiling Losses, 2022), which is visible in the July figure.

What would trigger an alert here?

If July in Year 2 showed PR of 0.695 instead of 0.730, that 3.5 percentage point drop year-on-year would cross the threshold for investigation. Common causes at this scale: partial soiling on several strings, one inverter MPPT operating with a fault, or module mismatch from a replacement batch with different voltage characteristics.

Worked Example 3: 5 MWp Utility-Scale Plant, Rajasthan

Rajasthan sits at 27°N with one of the highest irradiance resources in the world — annual GHI of 2,000–2,400 kWh/m²/year. The trade-off is extreme heat: ambient temperatures above 45°C are common from April through June, with module temperatures regularly exceeding 80°C. Annual PR expectations for a well-designed fixed-tilt bifacial system in Rajasthan are 75–80%.

At utility scale, energy is reported in MWh and system size in MWp. The PR formula is identical — units must match:

PR = E_AC (MWh) ÷ (H_POA (kWh/m²) × P_nom (MWp))

Note that kWh/m² × MWp = MWh, so the units cancel correctly.

Here is the 12-month dataset:

Sanity check: 8,490 MWh ÷ (2,190 kWh/m² × 5 MWp) = 8,490 ÷ 10,950 = 0.7753 ✓

Walking through January:

PR = 668 MWh ÷ (165 kWh/m² × 5 MWp) = 668 ÷ 825 = 0.810 = 81.0%

January achieves the highest PR of the year at 81.0%. This is the mirror image of the Munich pattern: Rajasthan in January has abundant irradiance (165 kWh/m²) and cool ambient temperatures (15–22°C daytime). Module temperatures at 35–45°C are well within the range where the temperature coefficient barely suppresses output. The system runs close to STC efficiency.

The May heat problem:

May PR drops to 73.0% — the lowest of the year. Ambient temperatures of 42–46°C push module temperatures to 75–85°C. At 80°C with a -0.4%/°C coefficient, output is depressed by 22% relative to STC. This is normal for Rajasthan. An annual PR of 77.5% here is excellent performance in this climate.

Interpreting Rajasthan PR against European benchmarks:

A fund manager comparing this plant against a German portfolio using a flat “PR should be above 78%” standard would flag the annual 77.5% as borderline. That interpretation is wrong. Rajasthan’s thermal environment structurally depresses PR by 5 to 7% versus a northern European climate (field-observed range for hot, dry climates per NREL Weather-Corrected PR study, 2013). The correct benchmark for Rajasthan is 73–79%. This plant at 77.5% is in the upper-middle band for the region.

In India, capacity utilization factor (CUF) is the primary regulatory and PPA metric, not PR. CUF = E_AC (MWh) ÷ (P_nom × 8,760 hours). The Rajasthan plant above: 8,490 MWh ÷ (5,000 kW × 8,760 hr) = 8,490,000 kWh ÷ 43,800,000 kWh = 19.4% CUF. This is a standard result for fixed-tilt in Rajasthan.

How to Interpret Any PR Result: A Six-Band Decision Framework

A PR number without context is not actionable. The six-band framework below maps any PR result to a specific diagnosis and next step. The thresholds assume temperate or Mediterranean climates (central Europe, coastal US, Japan). For hot climates, subtract 5–7 percentage points from every band boundary.

Six-step decision tree for any PR result:

1. Apply climate correction. If the site is in a region with mean annual temperature above 25°C (India, Middle East, southern US, North Africa), subtract 5–7% from band thresholds. A Rajasthan system at 74% falls in the “Acceptable” band after correction.
2. Check the time interval. If calculating daily or instantaneous PR, expect high volatility. Apply the six-band framework only to monthly or annual figures.
3. Compare to baseline, not benchmark. If you have 12 months of historical data for the same system, a 6% drop from baseline is more meaningful than the absolute band it falls into.
4. Isolate the season. Compare summer months to prior-year summer, not to spring or autumn. Seasonal PR variation of 8 to 12% within a year is normal for crystalline silicon (NREL, 2013).
5. Check the denominator first. Before diagnosing hardware, verify the POA sensor reading is plausible. A dirty or poorly calibrated pyranometer is the most common source of unexplained PR shifts. If POA reads low, PR reads high — and vice versa.
6. String-level drill-down. If monthly PR falls 4% below same-month baseline, check string-level current data. A single failed string on a 250 kWp system reduces PR by roughly 0.5–2% depending on string count — visible at monthly resolution but easily missed in annual averaging.

Climate-band caveat:

The thresholds above are calibrated to European and temperate North American conditions. DOE/FEMP (2021) reports an average PR of 78.6% across 75 US federal solar installations. Reich et al. (2012), cited in the same DOE report, found a median PR of 84% across 100 German residential systems. These benchmarks should not be applied directly to hot-climate installations.

When consistent underperformance signals degradation:

If a system shows PR 6% below its 12-month baseline consistently for 3 or more consecutive months, and soiling and inverter faults have been ruled out, the remaining candidates are:

• Module degradation at a rate above 0.5%/year (the advanced-scenario floor in the NREL ATB 2024) due to potential-induced degradation (PID) or UV-accelerated encapsulant browning
• Sustained shading from new obstructions not present at commissioning
• DC wiring deterioration (connector corrosion, damaged insulation) increasing resistive losses

A 3% year-on-year PR decline — sustained over 2+ years — is the signal to commission a full site inspection with EL (electroluminescence) imaging.

PR vs. Specific Yield vs. Capacity Factor: Which Metric to Report

Performance ratio is not the only output metric used in the solar industry. The right metric depends on who you are reporting to and what decision they need to make.

Guidance by audience:

Use PR when comparing two systems at different sites. PR removes location from the equation, so a Munich system and a Bologna system can be fairly compared. Use specific yield when a developer is deciding whether a site justifies the capital expenditure. A site producing 900 kWh/kWp/year cannot compete with one producing 1,600 kWh/kWp/year regardless of PR. Use capacity factor (or CUF) when reporting to grid operators or structuring PPA commitments. It directly measures the fraction of theoretical maximum output, which is the number that matters for grid dispatch and financial modeling.

The Indian CUF flag:

Indian solar PPAs and CERC performance standards are denominated in CUF, not PR. If you are modeling an Indian utility project for a DISCOM offtake agreement, CUF is the contractual metric. The target is typically 17–21% CUF for ground-mount fixed-tilt in most Indian states, with tracker-equipped plants reaching 22–24%. Converting: CUF = (Specific Yield) ÷ 8,760. For the Rajasthan example above, specific yield = 8,490,000 kWh ÷ 5,000 kWp = 1,698 kWh/kWp, and CUF = 1,698 ÷ 8,760 = 19.4%.

Solar software that reports all four metrics simultaneously — PR, specific yield, capacity factor, and CUF — allows you to satisfy every stakeholder in a single output without manual unit conversion.

The Three Most Common PR Calculation Mistakes

Getting the formula right is the easy part. Getting the inputs right is where most PR calculations fail.

1. Using GHI instead of POA in the denominator. Global horizontal irradiance measures sunlight on a flat surface. Plane-of-array irradiance measures sunlight on the tilted, oriented module surface. For a 30° south-facing array in central Europe, POA is typically 10–15% higher than GHI. If you substitute GHI for POA, the denominator shrinks and PR appears inflated — sometimes by 15%. You cannot compare a PR calculated with GHI to one calculated with POA. They are different numbers that look identical. Always use POA irradiance in the denominator, and always verify which irradiance channel your monitoring platform is reporting.

2. Using DC nameplate or inverter AC rating instead of DC STC nameplate. The correct denominator is P_nom = sum of all module Wp ratings at STC, in kWp. Using the inverter’s AC rating understates the denominator if the array is DC-oversized (DC/AC ratio above 1.0), which inflates PR. Using a rounded EPC contract figure (e.g., “5 MW” for a plant that is actually 5,240 kWp as-built) understates the denominator and inflates PR. Pull the number from the as-built module schedule.

3. Misaligned time periods for AC energy and irradiance. If your inverter logs energy in calendar months but your pyranometer data arrives in 30-day rolling windows, you will mix periods. A single day of mismatch at the start or end of a high-irradiance month can shift monthly PR by 2–3%. Align your AC meter and irradiance logger to the same timestamp reference — UTC is standard under IEC 61724-1. Apply the low-irradiance filter (exclude all intervals with POA less than 50 W/m²) before summing monthly POA totals.

Frequently Asked Questions

What is a good performance ratio for a solar system?

A PR of 75–85% is the accepted range for well-functioning systems in temperate climates. Systems above 80% are well-optimized, with low soiling, good inverter efficiency, and minimal shading losses. German residential systems studied by Reich et al. (2012, cited in DOE/FEMP 2021) showed a median PR of 84%, which reflects high-quality installations with proper monitoring. For hot-climate systems — India, Middle East, North Africa, and southern US states — the structural temperature penalty reduces expected PR to 68–78%. A Rajasthan utility plant at 77% is in the top quartile for its climate zone. Below 70% warrants investigation in any climate.

What is the difference between PR and specific yield?

Performance ratio strips out location by dividing AC output by irradiance × nameplate, making it a pure efficiency measure. Specific yield, measured in kWh/kWp per year, tells you how much absolute energy the system produced per kilowatt of installed capacity, which depends directly on how much sun the site receives. A Munich system can achieve PR of 82% while producing 1,050 kWh/kWp/year. A Rajasthan system might show PR of 76% while producing 1,698 kWh/kWp/year. The Munich system is technically more efficient; the Rajasthan system generates far more revenue. Both metrics matter, and they answer different questions.

Why does PR drop in summer?

PR drops in summer because module temperature rises. Crystalline silicon modules lose 0.3 to 0.5%/°C of output above 25°C (industry-typical range, per NREL, 2013). On a hot summer afternoon with module temperatures at 65°C, that is a 16 to 20% output reduction relative to STC. The irradiance in the denominator peaks in summer as well, but it peaks proportionally less than temperature suppresses the numerator — so the ratio compresses. This behavior is physically expected and should not trigger an alarm. The correct test is comparing this July to last July, not to this April.

Should I use GHI or POA irradiance in the PR formula?

Always use plane-of-array (POA) irradiance — never GHI. GHI measures radiation on a horizontal surface. POA irradiance measures radiation on the tilted, oriented module plane. For any non-horizontal array, these values differ. For a 30° south-facing system in central Europe, POA typically exceeds GHI by 10–15%. Using GHI inflates PR by the same proportion, making the system appear to perform better than it does. More importantly, a PR calculated with GHI cannot be compared to a PR calculated with POA — they are not the same metric.

What causes a performance ratio below 65%?

Below 65% indicates losses beyond normal temperature and optical effects. The most common causes are heavy soiling (dust, bird fouling, or pollution film), module degradation from long-term UV and thermal cycling, shading from trees or structures that were not present at commissioning, inverter faults or sustained clipping, and wiring losses from undersized or corroded DC cables. Before diagnosing hardware, verify AC metering accuracy — a revenue meter with a wrong CT ratio will understate E_AC and depress PR without any system fault. In climates with ambient temperatures consistently above 35°C, apply the 5–7% climate correction before concluding that 65% is a problem.

How often should performance ratio be calculated?

Monthly is the operational standard. Monthly PR catches seasonal patterns, flags a single bad month before it becomes a multi-month trend, and provides enough data volume to smooth out single-day weather anomalies. Calculate annual PR for bankability reports, investor returns, and year-on-year degradation tracking — a 3% year-on-year decline in annual PR, sustained over 2 or more years, is the signal to commission a full site inspection. Daily and instantaneous PR are useful for commissioning checks and fault diagnosis immediately after an alert, but they are too volatile for routine performance management.

You now have everything needed to calculate and interpret PR at any scale. The formula is simple; the judgment is in the inputs and the context.

Three actions to take from here:

• Calculate monthly PR using POA in the denominator and AC output in the numerator — not GHI, not DC output, not rounded nameplate figures
• Map the result to the six-band framework, adjusting -5 to -7% for hot climates before drawing any conclusions
• Track PR as a year-on-year trend; a 3% year-on-year decline sustained over two years is the signal to act