Curtailment vs Clipping: Two Different Losses, One Wrong Number

Two things cap your solar plant's output, they look almost identical in the data, and treating them as the same number is one of the most common — and most consequential — mistakes in loss accounting. The first is clipping: when your array can produce more DC power than the inverter can convert to AC, the inverter holds output at its ceiling and the excess is lost. The second is curtailment: when the grid, the market or an operator instructs the plant to produce less than it could, and you comply. Both show up as a flat top on the production curve. Both represent energy the plant could physically have made but did not. And there the resemblance ends — because one of them is a choice you made on purpose, and the other is money taken from you by someone else.

Definitions, and why the two get confused

Clipping — sometimes called inverter saturation — happens because most solar plants are built with more DC capacity than AC capacity. The ratio of DC array size to AC inverter size, the DC/AC ratio or 'overbuild', is chosen deliberately, because modules are relatively cheap and a bit of overbuild captures more energy across the low-light hours at the cost of clipping the few brightest ones. When irradiance is high enough that the array's potential DC output exceeds the inverter's AC rating, the inverter clips — holds AC output at its limit — and the surplus DC is simply not converted. This is expected. It was priced into the plant's design. Curtailment happens when an external party limits the plant's output below what it could produce and the grid could physically have taken from the inverters. It comes from grid congestion, negative prices, oversupply, or a direct dispatch instruction. Unlike clipping, curtailment is not a design consequence — it is an external constraint, and in many markets it is at least partly compensable or avoidable. They get confused because in the raw power data they can look the same: a ceiling on output. Distinguishing them requires looking at why the ceiling is where it is.

DC/AC ratio and expected clipping

Clipping is predictable because it follows directly from the plant's design and the weather. Given the DC/AC ratio, the module rating and the irradiance, you can calculate how much clipping to expect — it is the amount by which modelled DC output exceeds the inverter's AC ceiling, integrated across the bright hours. A plant with a higher overbuild clips more, by design, in exchange for more energy at the margins of the day and year. This is why clipping, correctly understood, is not really a 'loss' in the sense curtailment is. It is the expected cost side of a design trade that also delivers a benefit. A well-run plant clips exactly as much as its design predicts. The number worth watching is not the clipping itself but any deviation from expected clipping — more clipping than the design implies can signal an inverter capacity problem, and less can signal array underperformance.

Curtailment signatures in setpoint and meter data

Curtailment, unlike clipping, leaves fingerprints that reveal its external origin. The clearest is a setpoint or dispatch signal: when the plant is curtailed by instruction, there is usually a commanded power limit in the SCADA data, and the plant's output tracks that limit rather than the inverter's physical ceiling. A cap that sits below the inverter AC rating, and that moves in ways the weather does not explain, is curtailment, not clipping. Other signatures corroborate it. Curtailment often correlates with grid conditions — congestion, negative pricing intervals — rather than with irradiance. Clipping correlates tightly with irradiance and appears only at high output; curtailment can appear at any output level, whenever the external instruction arrives. And curtailment frequently shows a cap at a round or externally-set value, distinct from the physics-determined inverter ceiling.

Separating them in practice

The disciplined separation uses three questions applied to every capped interval:

1. 1.Where is the cap relative to the inverter AC rating? At the rating, in high irradiance — clipping. Below the rating — curtailment.
2. 2.Is there a commanded setpoint? A dispatch limit in the data that output is tracking is a direct curtailment signature.
3. 3.What does the cap correlate with? Irradiance and only the brightest hours — clipping. Grid conditions, prices, or arbitrary times — curtailment.

Applied consistently, these turn a single flat-topped curve into two distinct, correctly attributed losses. The energy above the cap during a clipping interval is expected design loss; the energy above the cap during a curtailment interval is externally imposed and potentially recoverable or compensable.

What each means commercially

The commercial consequences are where the distinction stops being academic. Clipping is a sunk design decision. There is no crew to send, no claim to file — it is the expected behaviour of a plant built with overbuild, and the only action it invites is validating that actual clipping matches design. Reporting clipping as a 'loss' to be recovered is a category error that wastes attention. Curtailment is the opposite. Depending on the market and the contract, curtailed energy may be compensable, may point to a congestion problem worth addressing, may inform a storage business case (shifting energy out of curtailed intervals), or may be a signal to revisit interconnection arrangements. Curtailment is a number that can drive action and, sometimes, recover revenue. Burying it inside a combined figure with clipping guarantees none of that happens.

How to report them separately

The fix is straightforward and belongs in every lost-production report: two lines, not one. An 'expected clipping' line, validated against design, that no one is expected to act on. And a 'curtailment' line, attributed to its external cause, that feeds curtailment analysis, compensation claims where applicable, and storage or congestion strategy. Each carries its own number, its own cause, and its own — very different — implication for what to do next. This is the same principle that governs any honest loss accounting: every megawatt-hour the plant did not deliver should be attributed to a named cause, and losses that behave differently should never be summed into a single figure that obscures both. Clipping and curtailment are the textbook case, because they look identical and mean opposite things.

Frequently asked questions

What is the difference between clipping and curtailment?

Clipping is an internal, designed-in consequence: when a solar array's potential DC output exceeds the inverter's AC rating, the inverter caps output and the surplus is lost — an expected result of building the plant with more DC than AC capacity. Curtailment is an external constraint: the grid, the market or an operator instructs the plant to produce below what it could, and it complies. Clipping was priced into the design and requires no action; curtailment is imposed from outside and is often recoverable, compensable or addressable.

Is clipping a loss?

Not in the way curtailment is. Clipping is the expected cost side of a deliberate design trade — building with a higher DC/AC ratio captures more energy across low-light hours in exchange for clipping the brightest ones. A plant clips exactly as much as its design predicts, and that clipping was accounted for when the plant was sized. Labelling it a recoverable 'loss' is a category error. The number actually worth watching is any deviation from expected clipping, which can signal an inverter or array problem.

How do you calculate curtailed energy?

Curtailed energy is the difference between what the plant could have produced and what it was permitted to produce during curtailment intervals. Identifying those intervals relies on curtailment signatures: a commanded setpoint the output is tracking, a cap sitting below the inverter AC rating, and correlation with grid conditions rather than irradiance. Once an interval is identified as curtailed, the recoverable quantity is the modelled achievable output — from irradiance and plant capability — minus the actual capped output, integrated across the curtailment period.

Should clipping appear in a lost-production report?

Yes, but as its own clearly-labelled line, validated against design expectation, and explicitly distinguished from curtailment. Clipping belongs in the report so that the full energy picture reconciles, but it should never be summed with curtailment into a single 'lost production' figure — because one is expected design behaviour requiring no action and the other is an external, often recoverable loss requiring analysis. Separating them is what lets the reader act on the number that can be acted on and ignore the one that cannot.

The takeaway