A generation report contains three measured intervals and one blank interval. The recorded energy is eighteen megawatt-hours. Should the blank be treated as zero, filled with the average, or left out of the performance calculation? Those choices answer different questions and can produce very different results.
The short answer is to preserve the gap and state what additional assumptions can bound or estimate it. Missing generation data are not automatically evidence of zero generation. An estimate can be useful, but it should remain distinguishable from a measured value.
This focused guide extends The Gold Standard of Wind Energy through an original fictional meter record. It teaches interpretation of supplied quantities, not plant operation, equipment selection, a contractual calculation or an investment assessment.
For power curves and the resource behind generation, use Wind Turbines, Blade Angles and Power Curves. Connect this account with learning that transfers into life and How X Works.
eduKateSG · Wind Energy
Find your next learning step
Choose the question closest to your task, or follow the expanded chapters in order.
Chapters 1–4 · Define the energy record
Name the quantity, interval and recorded subtotal.
Chapters 5–8 · Bound the missing quantity
Keep gap bounds distinct from filling assumptions.
Chapters 9–12 · Use additional evidence carefully
Inspect cumulative readings, boundaries and changing benchmarks.
Chapters 13–16 · Compare and transfer
Compare claims and solve an unequal-interval record.
Chapters 17–20 · Check and continue
Chapter contents
Define the energy record · Chapters 1–4
Bound the missing quantity · Chapters 5–8
Use additional evidence carefully · Chapters 9–12
Compare and transfer · Chapters 13–16
Check and continue · Chapters 17–20
CHAPTER 1 OF 20 · Define the energy record
1. Define the quantity before deciding how to fill a gap
A blank cell has no physical meaning until we know what should occupy it. It might hold an instantaneous power sample, an average power over an interval, energy accumulated during an interval or a cumulative meter reading. These are related quantities, but they are not interchangeable records.
The main exercise uses interval energy. Each row describes the energy generated during a stated six-hour period at one specified measurement boundary. The four rows cover one complete twenty-four-hour teaching day without overlaps or gaps in the schedule itself.
One row lacks a usable measurement. That is a gap in the evidence, rather than a missing part of the calendar. We know the interval’s duration even though its energy value remains unknown.
This distinction gives us a useful starting point. We can add the measured rows. We can identify the missing duration. We cannot obtain the missing energy merely by looking at an empty spreadsheet cell.
Before calculating, complete this sentence: the report is intended to describe energy at this boundary over this complete period. Then mark which parts of that intended quantity are measured and which are unresolved.
That small discipline prevents an ordinary editing choice from becoming an invisible physical assumption. A spreadsheet may make zero, blank and omitted row look convenient in different formulas. The explanation should decide what those states mean before the formula decides what to do with them.
CHAPTER 2 OF 20 · Define the energy record
2. Keep power and interval energy connected through time
Power is a rate of energy production. Energy over a period depends on the power through that period. In a supplied constant-output example, one megawatt maintained for six hours gives six megawatt-hours.
An interval energy of six megawatt-hours over six hours therefore implies an average power of one megawatt for that interval. It does not establish that output was exactly one megawatt at every moment.
This matters when the missing row is energy but another record contains one power reading. A two-megawatt reading at noon is not twelve megawatt-hours over a six-hour interval unless the relevant sustained-output assumption is justified.
Likewise, an interval average should remain attached to its duration. Equal numerical averages over two hours and six hours contribute different amounts of energy. Time belongs in the account rather than being a formatting label beside it.
The wider Energy guide and Wind Energy root develop these units. Here they allow us to interpret what a missing meter interval prevents us from knowing.
For a repair exercise, give learners three supplied quantities: one megawatt, six hours and six megawatt-hours. Ask which is a rate, which a duration and which an amount. Only then introduce a blank amount. The missing-data question becomes clearer when the student knows exactly which relationship could connect the available evidence to the unknown.
CHAPTER 3 OF 20 · Define the energy record
3. Read the complete fictional day before calculating a performance ratio
The fictional system has a stated two-megawatt benchmark throughout the day. For bounding purposes, the exercise additionally stipulates that generation at the named boundary is never negative and never exceeds two megawatts during the missing interval. This ceiling is a supplied model condition, not a universal claim about every nameplate rating.
| Fictional interval | Duration | Measured interval energy | Evidence state |
|---|---|---|---|
| 00:00–06:00 | 6 hours | 3 MWh | Measured |
| 06:00–12:00 | 6 hours | Unknown | Missing usable interval value |
| 12:00–18:00 | 6 hours | 9 MWh | Measured |
| 18:00–24:00 | 6 hours | 6 MWh | Measured |
The measured intervals total eighteen megawatt-hours across eighteen hours. The missing interval lasts six hours. Time coverage is therefore eighteen out of twenty-four hours, or 75%.
Do not write that 75% of the day’s energy was measured. That fraction concerns time, not energy. The missing interval could have a different average output from the observed intervals.
The distinction is already useful before any estimate is proposed. We have a complete schedule and partial energy evidence. A good account can describe both without disguising the missing amount or throwing away the measurements that remain available.
CHAPTER 4 OF 20 · Define the energy record
4. Calculate the recorded subtotal and label its scope
Add the three measured values: 3 + 9 + 6 = 18 megawatt-hours. That is the known subtotal in the fictional record. Its scope is the measured intervals, not automatically the full day.
The observed average power across those eighteen hours is eighteen divided by eighteen, or one megawatt. Because the observed intervals have equal durations here, averaging their individual average powers gives the same result. That coincidence should not be generalised to unequal intervals without checking weights.
A useful report might say that the available interval measurements sum to eighteen megawatt-hours and cover eighteen of the scheduled twenty-four hours. The remaining six-hour interval lacks a usable energy value.
This is a complete and informative statement of the supplied evidence. It does not need a falsely complete daily total to become useful. The next question is whether other information can establish, bound or estimate the missing amount.
For a learner check, ask whether eighteen megawatt-hours could be the actual daily total. Under the stated non-negative model, yes, if the missing interval produced zero. Ask whether the record establishes that condition. It does not.
Possible and established are different answers. The subtotal can equal the total in one compatible scenario while still failing to identify the total from the available evidence. Keeping those ideas separate prevents both overconfidence and the opposite mistake of treating the known subtotal as worthless.
CHAPTER 5 OF 20 · Bound the missing quantity
5. Derive a logical upper bound for the missing energy
The supplied ceiling is two megawatts throughout the six-hour missing interval. Sustaining that ceiling for all six hours would produce twelve megawatt-hours. Under the exercise’s conditions, the missing amount lies between zero and twelve.
Adding the known eighteen gives a complete-day energy range from eighteen to thirty megawatt-hours. These are logical bounds from the supplied ceiling and non-negative generation assumption. They are not a confidence interval, forecast distribution or statement that every value is equally likely.
The upper endpoint does not predict full output during the gap. It identifies the largest missing amount permitted by this simplified model. The lower endpoint likewise does not predict a shutdown.
Write the reasoning in words: known energy plus an unresolved non-negative amount that cannot exceed the stated ceiling multiplied by the missing duration. A learner who can explain that relationship can transfer it to a different gap length.
Now change the missing duration to three hours while keeping the same ceiling and known subtotal. The unresolved addition can be at most six megawatt-hours. The shorter gap produces a narrower range because the same maximum rate has less time to accumulate energy.
This is why a completeness percentage alone is insufficient. The location and duration of missing intervals, the applicable bounds and the quantity being reconstructed all affect what can be concluded from the remaining record.
CHAPTER 6 OF 20 · Bound the missing quantity
6. Keep full-period capacity factor separate from observed-period performance
The EIA capacity-factor definition compares generated energy with the energy implied by continuous full-power operation over the same period. The period in the numerator and denominator needs to match.
For the fictional full day, the benchmark is two megawatts times twenty-four hours, or 48 megawatt-hours. The bounded daily energy of eighteen to thirty therefore gives a bounded full-day ratio of 37.5% to 62.5% under the exercise’s benchmark.
For the observed eighteen hours, the benchmark is two times eighteen, or 36 megawatt-hours. The measured eighteen divided by 36 is 50%. That ratio describes the selected observed intervals.
It is not automatically the complete-day capacity factor. The missing interval may differ from the observed portion. Dividing partial energy by a partial benchmark can produce a valid conditional description while leaving the complete-period question unresolved.
Conversely, dividing the partial eighteen by the full 48 gives 37.5%. In this non-negative model that is the lower endpoint of the daily range, not an identified daily result.
The arithmetic is simple; the language carries the important distinction. State observed-period ratio, full-day lower bound or full-day estimate as appropriate. A bare percentage can conceal which of those questions was answered, especially when a reporting system presents every result in the same colourful performance box.
CHAPTER 7 OF 20 · Bound the missing quantity
7. Recognise what average-based filling adds to the evidence
Suppose an analyst fills the six-hour gap using the observed average power of one megawatt. The estimated missing energy is six megawatt-hours, giving an estimated daily total of twenty-four and an estimated full-day ratio of 50%.
The estimate lies inside the logical range. That is necessary for compatibility with the stated ceiling, but it does not prove that the estimate is accurate. Many other gap values remain compatible with the original record.
Its key assumption is that the observed average provides a suitable basis for the missing interval. If the gap coincides with unusually different wind or operating conditions, that assumption can fail. The record supplied so far does not settle the issue.
Notice a numerical coincidence. The filled-day ratio is 50%, exactly the observed-period ratio. That happens because the filling method extends the same observed average over the remaining time. The agreement is built into the method rather than independent confirmation of it.
A useful teaching comparison asks students to calculate both values and explain why they match. Then choose a compatible missing amount of twelve instead of six. The daily ratio becomes 62.5%, while the observed-period ratio remains 50%.
The purpose is not to ban estimation. It is to make the estimate’s added assumption visible. A model can support a decision or further analysis while remaining clearly labelled as a model rather than silently acquiring the status of a recovered measurement.
CHAPTER 8 OF 20 · Bound the missing quantity
8. Ask whether the gap mechanism supports the filling assumption
A missing interval may reflect several different processes: loss of communications, an unusable reading, an incomplete export or another record problem. Missingness alone does not identify what the generating equipment was doing.
In one fictional scenario, data transmission fails while generation continues. In another, a confirmed period of zero generation overlaps a lost meter value. Both can leave a blank cell in a summary, while their implications for missing energy differ.
The supplied evidence should distinguish those cases. A note saying no portal data is not the same as a verified statement that generation at the relevant boundary was zero throughout the interval.
Likewise, a record of availability is not automatically an energy measurement. Equipment can be available under a definition while output depends on the resource and other conditions. Keep the relevant state and the intended quantity separate.
For a learner task, provide two fictional incident notes with the same six-hour blank. Ask which note supports assigning zero and which leaves generation unresolved. Require the reason to refer to the actual evidence rather than the appearance of the spreadsheet.
This is the practical heart of missing-data interpretation. The best reconstruction method depends partly on why information is absent and what independent records survive. A convenient mathematical rule should not decide the physical story before those questions are examined.
CHAPTER 9 OF 20 · Use additional evidence carefully
9. Tighten the bound when a supported interval limit is supplied
Now add a new fictional condition: an independent, applicable record establishes that output during the entire missing interval could not exceed half a megawatt. The original zero lower bound remains.
The maximum missing energy becomes 0.5 × 6 = 3 megawatt-hours. The full-day range narrows to eighteen through twenty-one. Against the same 48-megawatt-hour benchmark, the ratio lies between 37.5% and 43.75%.
This is a real improvement in what the supplied model identifies. The original average-based fill of six megawatt-hours is now incompatible with the added ceiling. A later piece of evidence can therefore invalidate an earlier convenient estimate.
Check applicability carefully. The half-megawatt limit must cover the missing duration and the same measurement boundary. A restriction recorded for a different asset or a shorter period cannot simply be extended to the whole gap.
This chapter illustrates how bounds can be updated without pretending to know an exact value. Stronger evidence may shrink a range while still leaving several outcomes possible.
Ask a learner to explain what changed: not the original measured subtotal, not the reporting period and not the full-day benchmark. The additional supported constraint changed the maximum permitted missing energy. That explanation is more transferable than memorising a new percentage because it identifies the relationship that the new evidence actually affects.
CHAPTER 10 OF 20 · Use additional evidence carefully
10. Split missing time when its applicable ceilings differ
A single missing duration need not have one uniform limit. Suppose a six-hour gap contains two hours with a supplied two-megawatt ceiling and four hours with a supplied half-megawatt ceiling.
The first part can contribute at most four megawatt-hours. The second can contribute at most two. Their combined maximum is six, rather than twelve from applying the larger ceiling to all six hours.
This calculation follows the stated time segments. It does not require guessing an average operating condition. Add each segment’s maximum rate times its duration, provided the segments cover the gap without overlap and the limits apply at the same boundary.
Do not average the two ceilings as though they lasted equally long. The arithmetic mean of two and one-half is 1.25 megawatts. Multiplying it by six would give 7.5 megawatt-hours, which does not match the unequal-duration account.
The correct duration-weighted ceiling average is six megawatt-hours divided by six hours, or one megawatt. That average summarises the segmented maximum calculation; it is not an observed power level.
For independent practice, reverse the durations. Four hours at two megawatts and two at one-half permit nine megawatt-hours. The same two ceiling values produce a different bound because their time weights changed. A student who notices that change has understood the role of duration rather than merely recognised familiar numbers.
CHAPTER 11 OF 20 · Use additional evidence carefully
11. Use cumulative meter evidence only when its conditions fit
Suppose a separate fictional cumulative generation register reads 1,000 megawatt-hours at the start and 1,026 at the end of the same day. The exercise explicitly states that it measures the same boundary, has no reset or rollover and provides usable endpoint readings.
The difference establishes twenty-six megawatt-hours for the complete day under those conditions. Subtracting the eighteen measured interval subtotal leaves eight megawatt-hours for the missing interval.
Eight lies inside the original zero-to-twelve range. Its implied interval average is eight divided by six, about 1.33 megawatts. The full-day ratio is twenty-six divided by 48, approximately 54.17%.
Here the gap is resolved through additional compatible measurement evidence rather than a mean-based estimate. The earlier range was useful because it provided a consistency check when the new evidence arrived.
Now compare this with the later half-megawatt ceiling scenario. Eight megawatt-hours would exceed that scenario’s three-megawatt-hour maximum. The accounts cannot all be accepted unchanged. The appropriate response is reconciliation, not silently selecting the most attractive result.
The cumulative-register method has conditions. A reset, changed channel, incompatible endpoints or different boundary can invalidate a simple subtraction. Those issues need the actual record. The educational rule is not always trust the counter; it is identify which independent evidence establishes the missing quantity and verify that its scope matches the question.
CHAPTER 12 OF 20 · Use additional evidence carefully
12. Distinguish a measured total from a reconstructed interval pattern
Even a complete daily total does not always reveal when energy was generated within the day. In the preceding case, the missing six-hour energy is identified as eight megawatt-hours, but its moment-by-moment power pattern remains unresolved.
Many profiles can integrate to eight over six hours while respecting the supplied two-megawatt ceiling. One could be constant at about 1.33 megawatts. Another could vary while producing the same area under the power-time curve.
The daily total therefore cannot establish a particular ramp, peak or match to a short demand event. Those questions need appropriate time-resolved evidence.
This distinction connects directly to the Wind Energy root’s equal-energy timing example. That overview shows why matching totals can hide different timing. The missing-data case adds a second issue: a recovered total does not necessarily recover the missing temporal detail.
A learner report can say the cumulative endpoints establish the daily amount and, together with the measured rows, the missing interval amount. It should separately state that the supplied record does not identify the power trajectory within that interval.
This is a useful stopping point. A reconstruction can solve one question completely while leaving another open. Treating every solved subtotal as a complete recovery of the original record would extend the evidence beyond what the arithmetic actually establishes.
CHAPTER 13 OF 20 · Compare and transfer
13. Preserve the measurement boundary when another meter disagrees
Imagine the fictional generation register establishes twenty-six megawatt-hours, while a separate export register reports twenty-three. The difference is three megawatt-hours, but the two meters describe different boundaries.
Do not add them to claim forty-nine megawatt-hours of production. Nor should the difference automatically be labelled a specific loss. Other consumption, conversion or accounting terms may be relevant, but their actual values and definitions need evidence.
The first useful step is a boundary map showing what each record measures. Only then can an appropriate account connect generation, export and any other relevant flows or changes.
The Renewable Energy Engineering guide provides the wider system route. A missing interval in one channel cannot automatically be filled from another merely because both columns use megawatt-hours.
For students, compare two questions. How much energy was generated at the specified generation boundary? How much crossed the export boundary? Both can be meaningful, and the same day’s values need not match.
This chapter protects against a tempting shortcut in data repair: replacing unavailable evidence with a nearby number of the right unit. Compatible units are necessary, but compatible scope, time and definition are also needed. Otherwise the repaired table becomes numerically complete by answering a different question.
CHAPTER 14 OF 20 · Compare and transfer
14. Revisit the benchmark when the represented asset changes
The main example keeps a two-megawatt benchmark fixed for twenty-four hours. A different fictional comparison might explicitly define one megawatt of represented capacity for the first twelve hours and two for the next twelve.
Under that exercise’s time-varying benchmark rule, the full-period benchmark is 1 × 12 + 2 × 12 = 36 megawatt-hours. Applying the final two-megawatt value to the entire day would instead give 48 and represent a different basis.
This is not a universal instruction for every official capacity-factor series. Actual reporting conventions need their own definitions. The teaching point is to preserve the agreed asset population and benchmark period rather than infer them from the latest equipment list.
If measured energy is eighteen megawatt-hours, the ratio against the stipulated 36 benchmark is 50%. Against 48 it is 37.5%. Before debating which result is correct, identify which benchmark the report claims to use and whether it fits the intended comparison.
Missing data and changing assets are separate problems. Solving one does not solve the other. A complete meter total can still be compared against an incompatible denominator.
For a transfer task, ask learners to mark every change of represented capacity on a supplied timeline. Then calculate the benchmark in pieces. This develops the same habit used for unequal missing-interval ceilings: rate-like quantities need the durations over which they actually apply.
CHAPTER 15 OF 20 · Compare and transfer
15. Compare reports by what each one establishes
Consider three fictional summaries of the original incomplete day. The first reports eighteen megawatt-hours as a complete-day total. The second reports twenty-four after average-based filling. The third reports eighteen measured, six hours missing and a model-bound full-day range of eighteen to thirty.
The first conceals the unresolved interval unless independent evidence supports zero. The second can be useful if its estimate and assumptions are made explicit. The third preserves the logical range but does not pretend to supply a best prediction.
Different purposes may call for different additional methods. A planning model may need an estimate; a measurement report may need a clear distinction between measured and reconstructed amounts. The method should follow the question and applicable arrangements.
Do not call the widest range the most accurate estimate. It is a bound under assumptions, not an estimate of the most likely value. Likewise, a precise filled number is not automatically better evidence merely because it occupies one cell neatly.
A good comparison asks what information was used, what was assumed, which quantity was produced and what uncertainty remains. Those questions allow a reader to appreciate the usefulness of an estimate while recognising its limits.
The practical outcome is an honest choice among descriptions. Rather than arguing over a bare number, the learner can explain why two reports differ and what further evidence would resolve the complete-day measurement question.
CHAPTER 16 OF 20 · Compare and transfer
16. Transfer the method to an unequal-interval record
Try this new fictional case. A system has a supplied three-megawatt benchmark across a ten-hour study window. Usable interval measurements total twelve megawatt-hours across seven hours. The remaining three hours are missing, with output stipulated between zero and three megawatts.
The missing energy lies between zero and nine megawatt-hours. The total lies between twelve and twenty-one. The full-window benchmark is thirty, so the corresponding ratio lies between 40% and 70%.
The observed-period ratio is twelve divided by twenty-one, approximately 57.14%, because its benchmark covers only seven hours. It should not be presented as the identified ten-hour ratio.
Average-based filling extends the observed average of twelve-sevenths megawatts over three hours. It adds 36/7 megawatt-hours, giving 120/7, or about 17.14, for the full window. This remains an estimate built from a representativeness assumption.
Now supply compatible cumulative endpoints showing an exact total of eighteen. The missing amount is six, and the complete-window ratio is 60%. The additional evidence resolves the period amount without proving a constant output during the missing hours.
The transfer check is whether the learner can keep measured subtotal, time coverage, logical bounds, filled estimate and recovered total separate when durations and capacity change. Correctly naming those quantities is as important as obtaining their numerical values.
CHAPTER 17 OF 20 · Check and continue
17. Build an evidence record that keeps the gap visible
A practical account records the asset, boundary, interval schedule, usable measurements, missing duration and applicable benchmark. Add the provenance of any ceiling, estimate or independent cumulative evidence used to reconstruct the period.
Keep the original missing-data flag alongside the reconstructed value where the system permits. A later reviewer needs to distinguish a recovered measurement account from an assumption-based fill. Overwriting every gap with a number can erase the very information needed to evaluate the reconstruction.
An appropriate record can also explain why some questions remain unresolved. A daily total might be known while interval ramps are not. A generation figure might be established while export is a separate account. A bound can be defensible while its physical endpoint values are not predictions.
For a learner, a short evidence table is enough. Use columns for quantity, period, boundary, evidence state and limitation. Ask another reader to explain the daily result from that table without receiving extra verbal hints.
This is a useful check of communication as well as arithmetic. If the reader cannot tell which values were measured, inferred or assumed, the account needs clearer labels.
The aim is a record that supports the next question. It should make a real gap easier to investigate, rather than conceal it behind a complete-looking dashboard or turn a classroom reconstruction into an unsupported claim about an actual plant.
CHAPTER 18 OF 20 · Check and continue
18. Choose the learning route from the first uncertain relationship
The repair route begins with rates and amounts. Use one constant-power example and one interval-energy row. Check that the learner can explain megawatts versus megawatt-hours before asking about a complete reporting day.
The stabilisation route introduces the missing interval, derives bounds and compares observed-period with full-period ratios. Ask for a sentence naming the numerator and denominator each time.
For extension, add unequal interval ceilings, a cumulative counter and a different export boundary. The learner should decide which new evidence resolves which uncertainty and recognise when two supplied accounts cannot all hold simultaneously.
Original data exercises can be completed from supplied records without approaching turbines or changing equipment. The relevant practical action is interpretation: identify the missing information, state the assumptions and ask for appropriate evidence.
The strongest final task changes one condition without announcing which formula to use. Shorten the gap, change its ceiling, introduce a reset or revise the represented capacity. Ask the learner to explain which earlier result must change and why.
Progress appears when the student can preserve useful conclusions while narrowing unsupported ones. They need not choose between pretending to know everything and abandoning the record entirely. A measured subtotal, a justified bound and a clearly labelled estimate can each be valuable when their distinct roles remain visible.
CHAPTER 19 OF 20 · Check and continue
19. Questions about incomplete generation records
Does missing data mean zero generation?
No. It means the intended measurement is unavailable or unusable under the record’s definition. Assigning zero requires supporting evidence about generation at the relevant boundary and interval.
Can I simply remove the missing hours?
You can describe the observed period if clearly labelled. That does not automatically answer the complete-period question. The denominator and the claimed period must remain compatible.
Is a logical range a confidence interval?
No. The worked bounds follow from supplied constraints. They do not assign probabilities or identify a most likely missing value. Statistical intervals require an appropriate model and evidence.
Does agreement with an observed-period average verify a fill?
Not when the fill was constructed from that same average. The agreement can be built into the method. Independent compatible evidence is needed to test the reconstruction.
Can a cumulative counter resolve the gap?
Yes, under suitable conditions such as matching boundaries and usable endpoints without an unresolved reset or rollover. A complete-period total may still leave the within-interval power pattern unknown.
Does a recovered energy total establish project quality?
No. Resource, operations, delivery, environmental responsibilities and the intended service need their own evidence. A solved missing-data account answers a defined measurement question, not every project decision.
CHAPTER 20 OF 20 · Check and continue
20. Continue through the energy and measurement routes
For the complete conversion and service chain, return to Wind Energy, Energy and Renewable Energy. For a deeper explanation of power curves, wind distributions and uncertainty in model inputs, use Wind Turbines, Blade Angles and Power Curves.
For the connection between generation and its receiving system, follow Renewable Energy Engineering. For the distinct solar-system handover context where monitoring gaps also matter, read Solar PV System Commissioning. That guide does not turn a solar acceptance method into a universal wind-reporting rule.
The primary definition used here is EIA's capacity-factor explanation, checked on 21 October 2026. It supports matching generated energy with the same period’s full-power benchmark. The missing-interval records, ceilings, counters and transfer problems are original fictional teaching cases.
For your next independent task, take a supplied incomplete record and write three separate statements: what is measured, what can be bounded and what needs an additional assumption. That is a concrete example of learning that transfers into life, where a useful answer remains honest about the evidence it actually has.