Interval Data Gaps in Performance Ratio Calculations

A missing record is not a measured loss

Interval data gaps can turn an apparently precise performance ratio into a misleading result. Missing AC energy, missing irradiance and a verified period of zero generation are different conditions. Replacing all three with zero, or deleting them without explanation, changes the question the calculation answers. A commercial operator should first establish what information is absent and whether the plant's physical operation is known.

The immediate decision is whether the available records support the intended report: a diagnostic ratio over valid intervals, a complete-period operational result, or a contractual assessment under agreed rules. These outputs may legitimately differ. Name the intended output before cleaning the data. Otherwise, a contractor can produce a mathematically consistent ratio that excludes the very outages the owner expected the report to reveal.

Define the numerator, denominator and common period

For a conventional simplified PR, divide measured AC energy by the energy represented by DC nameplate capacity multiplied by plane-of-array irradiation and divided by reference irradiance. With capacity in kWp, irradiation in kWh per square metre and reference irradiance of 1 kW per square metre, the denominator is expressed in kWh. The electrical boundary and installed capacity must correspond to the system being evaluated.

Energy and irradiation must describe the same included period. The national-laboratory report on weather-corrected PR also explains that conventional PR depends on weather and can vary seasonally; matching records does not remove that separate issue. If a temperature- or weather-corrected metric is required, its additional inputs and method need to be specified. A basic PR should not quietly acquire a corrected label because the analyst applied a generic adjustment. Weather-Corrected Performance Ratio.

Build a quality ledger before filling gaps

Retain the original exports and create a separate quality ledger for each expected interval. Useful states include valid measurement, missing record, duplicate, invalid sensor value, estimated value and confirmed operational event. Record the reason for every exclusion or substitution. This lets another analyst reconstruct how the reporting dataset differs from the raw files.

The IEA PVPS technical and economic KPI report identifies completeness, accuracy, consistency, timeliness and reliability as data-quality considerations. It explains that missing measurements and their imputation affect calculated indicators and that suitable completeness criteria depend on the task. Those findings support a declared project method rather than one universal allowable gap percentage. Examine the timing and concentration of gaps as well as their count: lost midday records and absent night records do not remove the same solar resource. IEA PVPS KPI report.

Use matched intervals without hiding excluded operation

For a valid-interval diagnostic ratio, apply a common inclusion mask to both energy and irradiation. If one input is unavailable for an interval, the other cannot remain in that ratio as though the pair were complete. Report the included period and coverage alongside the result. This is a disciplined calculation over the retained observations, not automatically an estimate of the whole month.

Keep operational exclusions distinct from measurement exclusions. A confirmed inverter outage with valid energy and irradiance measurements is evidence about plant performance, not bad data merely because output is low. Removing it may be appropriate for a separately defined equipment-performance diagnostic, but then report the outage and explain the metric's scope. The same principle applies to curtailment. A calculation intended to isolate conversion behaviour and a report of actual delivered performance require different treatments and labels.

A hypothetical gap that changes PR in both directions

Consider a hypothetical 100 kWp system with 100 kWh per square metre of plane-of-array irradiation during a reporting period. Its simplified reference energy is 10,000 kWh. Suppose complete measurements would show 8,000 kWh of AC energy, giving PR of 80%. One missing group of energy records actually corresponds to 2,000 kWh, while irradiation for those same intervals corresponds to 25 kWh per square metre.

Treating missing energy as zero while retaining all irradiation produces 6,000 divided by 10,000, or 60%. Removing the affected irradiation as well produces 6,000 divided by 7,500, or 80% for the retained intervals. In this deliberately simplified example, that equals the complete-period result because the missing intervals happen to have the same ratio. It would not generally do so. Retaining all 8,000 kWh while using only 75 kWh per square metre would instead produce approximately 106.7%, exposing the opposite mismatch.

Recover measurements before estimating them

Check whether missing portal records can be retrieved from the local logger, meter or another retained source. A continuous cumulative energy register may establish total energy across a communications gap when its endpoints and integrity are verified. It does not necessarily reveal how that energy was distributed among the missing intervals. Preserve that distinction when the analysis needs interval-level alignment rather than only a period total.

If estimation remains necessary, state which variable is estimated and why the chosen source is suitable. A nearby irradiance sensor, a model or satellite information may differ from the actual array conditions. Interpolating through a long cloudy period can create a smooth series with little evidential value. Test a proposed recovery method on comparable periods where measurements exist, and retain an estimation flag. The aim is to quantify what can be recovered, not make every row appear measured.

Quantify coverage and sensitivity around the decision

Report expected and valid daylight intervals, the longest material gap and the affected operating states. Where trustworthy irradiation remains available, also describe the share of irradiation associated with excluded intervals. If irradiation itself is missing, label any resource-coverage estimate according to its source. A statement that 98% of all timestamps are present can conceal the loss of a much larger share of useful daytime information.

Show how reasonable alternative treatments affect the conclusion. Compare the valid-interval result with a separately labelled reconstructed-period estimate where justified. Identify whether the decision remains stable or crosses a contractual threshold under different assumptions. Do not manufacture confidence bounds without a supporting method. If the missing period is too important to reconstruct defensibly, report that the intended full-period PR cannot be established from the available evidence. That is more useful than an unsupported pass or failure.

Deliver an auditable result and repair the data process

The final report should include the metric definition, electrical boundary, capacity, timestamp convention, input sources, inclusion rules, gap ledger, recovery methods, results and limitations. Provide the calculation version and a reproducible link between the reported totals and underlying records. Assign the monitoring owner a corrective action where repeated gaps suggest a logger, communications or sensor-maintenance problem.

The existing interval-load guide covers timing and data interpretation for sizing; PR adds the requirement to pair measured output with the solar resource over the same accepted intervals. The commercial solar hub discusses PR in broader project assessment, so agree the reporting definition before using it in a guarantee discussion. Through Solar Panels Cyprus, request a specialist review of the monitoring and calculation scope, with measured, recovered and estimated evidence priced and described separately.

Frequently asked questions

Should missing generation records be replaced with zero?

Only if evidence establishes zero generation for those intervals. An absent record does not establish a physical outage.

Can a cumulative meter recover a communications gap?

Verified register endpoints may recover total energy across the gap. They do not automatically recover its interval distribution, which may still matter for alignment and corrected metrics.

Does matching valid intervals produce the true monthly PR?

It produces a ratio for the retained intervals. Whether that represents the whole month depends on the excluded periods and the purpose of the assessment.

Is there one acceptable missing-data percentage?

No universal percentage is supplied here. The assessment must consider the metric, gap duration and timing, available recovery evidence, and any agreed contractual rules.

Discuss the monitoring evidence and PR calculation scope with a qualified solar specialist.