Performance ratio, or PR, compares a PV system's delivered energy with a reference derived from its installed module capacity and received irradiation. It helps separate the effect of available sunlight from the overall system result. It is not identical to capacity factor, and it is not a direct measure of forecast accuracy or module conversion efficiency.
The formula and its inputs
PR = final yield / reference yield. Final yield Yf = E / P0, where E is the system's net energy output over the period and P0 is its stated DC nameplate capacity. Reference yield Yr = H / Gref, where H is irradiation on the plane of the array over the same period and Gref is the reference irradiance, conventionally 1 kW/m². Consistent units give both yields in hours or kWh/kW.
Consider an illustrative system with 100 kW of DC module capacity, producing 12,000 kWh during a period with 150 kWh/m² of array-plane irradiation. Final yield is 120 kWh/kW. Reference yield is 150 hours using a 1 kW/m² reference. PR is 120 / 150 = 0.80, or 80%. This example demonstrates the arithmetic, not a recommended benchmark or an enercast performance claim.
Use the right radiation plane
The denominator concerns sunlight received on the array plane. Horizontal irradiation is not automatically an appropriate substitute for a tilted or tracking array. If a model converts horizontal radiation into array-plane radiation, that transformation introduces assumptions and uncertainty. Document whether irradiation is measured or modelled, and how orientation and tracking are represented.
For systems with several orientations, explain how the reference is constructed. A single sensor may not represent every sub-array equally. Bifacial systems require particular care in defining the reference because rear-side contributions and their measurement or modelling affect interpretation. Comparisons need consistent methodology rather than a percentage alone.
Measurement boundaries shape the result
Energy measured at inverter output may differ from energy at the export meter because of downstream losses and auxiliary consumption. DC capacity must describe the modules relevant to that energy measurement. Capacity changes during the period need a corresponding change in the calculation, not simply today's rating applied to all historical energy.
Check timestamps, missing intervals and sensor quality. Soiling or shading on an irradiance sensor can bias the denominator while leaving plant energy unchanged. A misaligned energy interval can distort daily comparisons. Before diagnosing a plant issue from PR, confirm that numerator and denominator cover the same valid period.
Interpret changes before applying a benchmark
Temperature, conversion losses, availability, soiling, shading, clipping and restrictions can influence the result. Some effects vary seasonally or with operating conditions. A change in uncorrected PR therefore does not automatically identify a new fault, and a stable value does not prove every component is healthy.
There is no universal target appropriate for every design and climate. Compare with the project's documented model, warranty or operating reference under its stated conditions. Where temperature-corrected or weather-normalized metrics are used, name the method rather than comparing them directly with an uncorrected value. Matching the metric definition is the first step in a meaningful benchmark.
PR and capacity factor answer different questions
Capacity factor divides generated energy by rated capacity times the hours in the period. PR instead normalizes energy using received irradiation. A site with abundant sunlight can have a higher capacity factor while having a lower PR than another site. The metrics describe different aspects of performance and should be presented together only with their definitions.
Forecast accuracy asks whether predicted values match subsequent observations at particular horizons. A forecast can reproduce total energy while mistiming production, so neither forecast PR nor a good period total replaces interval-level validation. Use MAE, RMSE, bias and event analysis where appropriate to the forecasting target.
Connect performance analysis to forecasting
An unexpected PR change can prompt investigation of asset metadata, status and loss assumptions used in a forecast. Retain evidence rather than immediately recalibrating the model to lower production. If the plant is temporarily constrained, learning that period as normal conversion behaviour can make future unconstrained forecasts too low.
A useful review records the formula, capacity boundary, energy meter, irradiation source, period and exclusions. Then it compares observations with a documented expectation and investigates the residuals. This makes PR a reproducible diagnostic tool. It avoids turning one percentage into an unsupported claim about the health, profitability or forecast quality of an entire PV project.