Lost output

Read the stored quantity, unit and quality, see what it was computed from, and never treat missing data as zero.

Lost output is the production lost during an event, as a quantity in an explicit unit, for example MWh. The unit is the one the installed vocabulary pack declares. Lost output is never a monetary figure.

The Events list shows the recorded Lost output / Quality. The event record shows the figure in Lost output, with what it was computed from.

Quality

Quality	Meaning
Measured	The figure was recorded as measured
Estimated	The figure compares an expected rate with the actual rate. It is a usable estimate, not a measurement
Not computable	The calculation recorded no usable value. No value is stored, and it is never a zero-loss result

The baseline methods available today record Estimated or Not computable, never Measured.

In the list, Unavailable means no value or quality was provided. On the record, Lost output is computed once the event closes for an open event, and No lost output has been computed for this event for a closed event that has none.

Estimated baselines

Expand About lost output on Events for the quality explanation. A good-only baseline remains Estimated. It compares the expected rate from a window before the event with the actual rate during the event; restricting the points to good quality does not turn that estimate into a fully measured result.

Point counts do not prove temporal coverage. Many points may still leave gaps in time. If either required window has no good points, the result is Not computable, never zero.

Historical v0 results keep their recorded policy. Reading an older result does not silently apply the newer good-only policy.

What the figure was computed from

Under the figure, the record lists:

Fact	What it shows
Method	The recorded method, and Good points only or All points (historical)
Computed	When the stored figure was computed
Pack	The vocabulary pack that declared the method, or Not settled among the packs that could have
Output KPI	The KPI the actual output was read from, with its unit
Reference window	The window before the event, its baseline hours, and its point counts: good, uncertain and unknown
Event window	The event's own window, its hours and its point counts
Conversion	How the KPI's rate unit converts to the pack's quantity unit
Read on	The asset the figure was read on: this event's own asset, or the unit it rolls up to

Not reached or Not read means the calculation stopped before that step. A not-computable figure states its reason in words, for example Several installed packs declare a method and nothing on the event picks one., followed by the recorded reason code.

Pending figures

When an event closes but the figure cannot be computed for a moment, the record shows Pending with Recomputation scheduled. The figure is retried automatically every few minutes for 24 hours, and the record says until when. No value is stored yet, and it is not zero. If the retries end without a figure, the record shows Not computed; compute it again to settle it.

Compute it again

Operators and admins can choose Compute again on a closed event. The dialog, Compute lost output again, asks for no reason, because the change is recorded under its own reason. The new figure replaces the stored one, and the previous figure stays in the journal as a Lost output computed entry.

Compute again after a trim, split or merge. Those changes leave the stored figure on the old range, and a contract's production availability treats such a stale figure as unusable until it is computed again.

Recorded reasons

Export CSV includes the recorded method and unavailable-data reason when present. A reason may explain missing good points, an open event, an unavailable input, an undeclared or ambiguous method, an unsupported unit conversion or a time window beyond the supported limit. Empty reason fields do not prove that every input was available.

The list shows the stored quality. It does not prove that the underlying points cover the whole interval.