Nuclear outages: rebuild lost energy from reactor status
A reactor’s daily reported power percentage is useful evidence about operating status. It is not an electricity meter covering every hour of that day. A company researcher can use the series to describe outage exposure, but must keep the resulting capacity-based approximation separate from actual gross generation, net generation and electricity sold. Otherwise a precise-looking energy total can conceal an unsupported conversion.
The completed worksheet should contain a unit-identity register, a dated status sequence, an explicitly assumed capacity conversion and a reconciliation to separately reported generation. It should also distinguish what the evidence establishes about a change in output from what remains unknown about its cause. A low-power day alone cannot prove either a planned refueling outage or an unexpected equipment failure.
Understand the source before multiplying
The EIA introduction to its Status of Nuclear Outages product explains that the underlying daily reactor information is compiled by the NRC and presented through EIA’s outage visualizations. Locate the paragraphs describing the source and historical coverage. The page provides a retrieval route, not a unit-specific assurance that every status value measures average net electric output.
For the actual investigation, obtain the relevant daily records and their data dictionary, the issuer’s unit ownership and capacity disclosures, and any available operating explanation. Record reactor identifier, date, reporting time, stated percentage basis, capacity basis and source version. Keep the publication date separate from the operating day represented. A correction should preserve the originally retrieved value and its replacement.
Do not quietly equate reactor thermal-power percentage with electric-power percentage. Even where they move together, conversion performance and station consumption can differ. If the available record does not support an exact electric interpretation, call the result a capacity-based status approximation. A label is part of the calculation, not a disclaimer to add after presenting it as measured generation.
Reproduce a ten-day approximation
Consider a fictional unit assigned an assumed constant 1,000 MW electric-capacity reference. For this teaching example only, treat each day’s status percentage as representative of the entire 24 hours and proportional to that reference. There is no daylight-saving change, missing record or within-day variation. These are analytical assumptions, not claims about the NRC measurement method.
| Status block | Days | Assumed factor | Approximate MWh |
|---|---|---|---|
| Full reference output | 4 | 100% | 96,000 |
| Zero status | 3 | 0% | 0 |
| Partial output | 2 | 50% | 24,000 |
| Restart block | 1 | 80% | 19,200 |
| Total | 10 | — | 139,200 |
For every block, multiply 1,000 MW × factor × days × 24 hours. Full-reference energy for ten days would be 240,000 MWh. The status-based difference is 100,800 MWh: 72,000 from the three zero-status days, 24,000 from the two half-output days, and 4,800 from the 80% day. Their sum provides an independent arithmetic check.
The weighted status factor is 139,200 / 240,000 = 58%. It is the result of the example’s mapping. Do not publish it as an actual net capacity factor without compatible generation and capacity evidence. Likewise, the difference is energy relative to the chosen reference; it is not established lost sales or economic damage.
If ownership is 60%, the same simplified attributable output is 83,520 MWh and the attributable reference difference is 60,480 MWh. Preserve the gross-unit series alongside that ownership allocation. An equity-method interest, an operating role and a power-purchase entitlement need not describe the same economic quantity.
Attribute causes only after building the status record
Join the dated changes with issuer outage notices, operating updates and other directly relevant first-party records. Use separate categories for confirmed planned activity, confirmed unplanned event, restart testing, external grid restriction and unresolved cause. The category should point to a specific source sentence, not a guess based on the month of the year.
Suppose the issuer confirms that two of the three zero days belong to a planned outage. It does not explain the remaining zero day or the half-power interval. Allocate only the supported two days to planned activity. Preserve the other observations as unresolved; assigning every surrounding day to the same event can exaggerate both its duration and its reported energy effect.
An expected restart date is a forecast. A later report of synchronization, testing or stable operation is a different milestone. Record each separately. A brief return to high status does not by itself establish a sustained return to commercial output or the end of all corrective work.
Test a boundary case with the same daily snapshot
Now consider a different fictional day whose published status is 100%. Scenario A operates at the reference all day, giving 24,000 MWh. Scenario B operates at zero for twelve hours and the reference for twelve, giving 12,000 MWh. A single snapshot near the end of the day could look identical in both scenarios.
This is a sampling limitation, not an arithmetic error. It shows why hourly operating or meter data can materially change an estimate around trips and restarts. Do not repair the gap by pretending the snapshot is an hourly series. Report the approximation, identify transition days and request finer evidence for those days first.
A second boundary is gross versus net generation. If the issuer reports gross generation of 140,000 MWh and station use of 7,000 MWh for the same scope, net generation is 133,000 MWh. Comparing the worksheet’s 139,200 directly with 133,000 creates a 6,200 MWh difference that combines modeling and measurement boundaries. Reconcile gross and net definitions before explaining the residual as operational underperformance.
Reconcile reported generation and commercial exposure
Obtain generation for exactly the same units and dates where available. If disclosure is monthly while the model covers ten days, do not subtract one from the other. Either extend the status exercise to the disclosed month or state that the direct comparison cannot yet be performed. Calendar alignment matters more than adding decimal places.
Keep replacement purchases, hedges and customer obligations in a separate commercial panel. A unit outage can lead to purchased power, reduced merchant sales, use of another asset or a different contract outcome. Multiplying the energy difference by a headline spot price establishes none of these. Actual settlement and contractual quantities are necessary before assigning a financial effect.
If a generation report and a status series disagree, check identities, ownership changes, capacity uprates, time zones, missing dates, thermal versus electric basis and station consumption. Do not choose whichever series supports the thesis. A preserved residual is a useful research result when its unresolved components are clearly listed.
A verification table that can be handed off
| Object | Required evidence | Passing check |
|---|---|---|
| Unit identity | Operator and ownership disclosures | Same reactor in every source |
| Status day | Raw record and data dictionary | Date and percentage meaning retained |
| Conversion | Stated capacity and assumptions | MW × hours produces MWh |
| Event cause | Dated first-party explanation | Cause applied only to supported interval |
| Generation | Compatible meter or issuer series | Gross/net and period reconciled |
| Economic effect | Purchase, sales and contract records | No invented spot-price loss |
Review the result when transition-day evidence arrives, the operator revises an outage explanation, a capacity basis changes or compatible generation becomes available. Preserve earlier estimates as dated approximations and explain which assumption the new evidence replaces. That produces a trackable operating judgment without turning a daily status observation into a false measurement of realized electricity.
Bound missing days instead of filling them with zero
Suppose only seven of ten status days are available and the three missing days belong to the assumed 1,000 MW reference. Under a deliberately broad zero-to-full-reference bound, missing-day energy can range from zero to 72,000 MWh. Add that interval to the observed-day subtotal. A single interpolated result would conceal uncertainty near an outage transition.
Fleet aggregation needs a second check. A 1,000 MW unit at full status and a 500 MW unit at zero produce an unweighted status average of 50%, but the capacity-weighted approximation is 66.67%. Use consistent capacities and dates rather than averaging percentages. If a unit changes ownership during the interval, apply the evidenced ownership schedule to its own days before aggregating. Neither missing observations nor acquisition timing can be repaired by a more precise group average.
If a supported capacity uprate changes the reference from 1,000 to 1,050 MW during the study, split the calendar at its effective date. Applying the new value retrospectively would create an extra reference-energy difference for earlier days. Keep the capacity revision and status observation as separate versioned inputs.
FAQ
Does zero status establish an unexpected outage?
No. Planned work, testing and other causes require their own dated evidence.
Can the modeled difference be called lost revenue?
Not without commercial entitlements, actual sales or purchases, and compatible settlement terms.