Calculating defensible PV energy loss from interval data

A defensible energy-loss estimate begins at the telemetry interval, where expected and measured power can be aligned and qualified. Aggregating first and explaining later can hide missing data, nighttime periods, and offsetting behavior. Interval calculations make every included and excluded contribution traceable.

1. Define eligible intervals

An interval is eligible only when actual AC power is measured, expected AC power is valid, the expected-production model quality is usable, and the interval duration is known. Exclude nighttime intervals where expected output is effectively zero. Missing actual telemetry creates unknown impact; it must not be treated as zero production.

2. Apply transparent interval equations

• Power deficit (kW) = max(expected AC power − actual AC power, 0).
• Estimated lost energy (kWh) = power deficit × interval duration in hours.
• Estimated lost revenue = estimated lost energy × configured site energy value per kWh.
• Negative deficits are clipped to zero rather than counted as recovered historical loss.

3. Aggregate with coverage attached

Sum estimated loss only across eligible intervals and report the corresponding expected energy, actual energy, data completeness, and unknown-impact interval count. A site with incomplete data should not be ranked as having zero loss. Use an explicit unknown state whenever the available comparison cannot support an estimate.

4. State the financial assumptions

Use a site-specific energy value when available and identify its units and currency. The value may represent an avoided purchase price, contract rate, or another organization-approved assumption. Do not combine unlike assumptions silently across a portfolio, and do not describe estimated revenue exposure as guaranteed recovery.

5. Validate the result before prioritizing

1. Inspect the expected and actual curves for timestamp or aggregation errors.
2. Confirm clipping, curtailment, and planned downtime are interpreted consistently.
3. Review model confidence and calibration history.
4. Compare loss magnitude with incident persistence and data completeness.
5. Retain the interval evidence behind every site and portfolio total.

For forward-looking cost-of-waiting estimates, use a recent rolling daily loss estimate across representative daylight conditions rather than multiplying one extreme interval. Future weather and operations will differ, so projections should include confidence, method, and limitations.