Key takeaway
Equal AC output at a cap can come from different pre-cap states. An AC-only plateau therefore cannot, by itself, establish unchanged potential production.
Test the proposed label before analyzing a whole plant
A solar archive owner proposes labeling two periods “no production degradation” because their peak AC readings are equal. Before fitting a trend, ask whether the retained output can distinguish the underlying states at those peaks. A capped response can erase changes even when every timestamp is present.
SAM’s inverter documentation describes power-limiting clipping in which modeled AC output is limited to inverter nameplate capacity. It separately discusses losses associated with the maximum-power-point-tracking voltage limits. The mechanisms should not be merged into one unexplained “clipping” flag.
The immediate owner decision is about the proposed archive label: does unchanged AC establish unchanged pre-cap production? The following counterexample answers that limited question. It does not diagnose any actual inverter or plant.
Compute three states through the same fictional cap
Define a deliberately simplified hypothetical fixture: Pout = min(Ppotential, 100 kW). Here Ppotential already incorporates the illustration’s conversion assumption and represents potential power before the selected output cap. It is not a measured DC input and is not a complete PVWatts calculation.
The first two rows differ before the cap but produce the same retained AC value. The third row lies below it. Because more than one input maps to 100 kW, that retained output alone cannot recover which input occurred. The completed decision rejects “unchanged potential production” as a conclusion from these capped AC values alone.
This remains true without a missing sample or a timing defect. Increasing the sampling frequency of the same capped quantity does not reveal the hidden amplitude in this simplified fixture. Additional evidence must carry information about the pre-cap state or the relevant limiting mechanism.
| Fictional state | Ppotential after stated conversion assumption | Cap | Computed Pout |
|---|---|---|---|
| A | 120 kW | 100 kW | min(120,100) = 100 kW |
| B | 110 kW | 100 kW | min(110,100) = 100 kW |
| C | 90 kW | 100 kW | min(90,100) = 90 kW |
| A compared with B | Different potential values | Same cap | Equal AC does not identify equal potential |
Use the model description without pretending it was executed
The pvlib documentation read for version 0.16.1 describes its PVWatts inverter function with an AC output capped at the applicable limit. Its parameters and efficiency relationship are more detailed than this article’s three-row minimum function. The documentation also warns that the inverter function’s pdc0 has a different meaning from the similarly named module-model parameter.
An owner preparing model-derived records should preserve the exact function, version, parameter meanings and units. Do not replace an inverter input reference with array nameplate capacity because both fields look like DC power. That substitution can change the model basis while producing a plausible number.
Only the explicit minimum calculations above were evaluated for this article. No pvlib function, SAM simulation or physical plant model was run. The model sources support the cap mechanism; they do not turn the toy fixture into a validated lost-energy estimate.
A real plateau still needs a mechanism
A flat AC trace alone does not establish that real power-limiting clipping occurred. Inspect the relevant inverter configuration and nameplate basis, available DC and voltage telemetry, curtailment state and collection-system behavior. A limiting setpoint, a voltage constraint or a recording artifact can raise different interpretation questions.
Preserve the state and effective dates of those supporting records. A current configuration should not silently describe an earlier operating period. If the archive retains only AC values, identify that constraint before offering a degradation, soiling or available-power evaluation.
The purpose of the counterexample is not to infer that the panels deteriorated between the fictional states. It establishes only that unchanged capped output cannot rule out a pre-cap change. The actual cause, magnitude and period of a plant change need their own evidence and analysis.
Offer the evaluation the retained signal can support
The corrected owner description is: “AC output history with potential output-limit ambiguity; inverter configuration and supporting telemetry identified where retained.” That permits evaluation of the observed AC history without promising recovery of unobserved potential power.
For the fictional A–B pair, the completed result is simply “same retained output, different permitted pre-cap states.” A technical reviewer can now decide whether additional data make the intended question identifiable, instead of treating a flat line as a health certificate.