Technology / Engineering guide
Wind Energy P50, P90 & Exceedance Probability
P50 and P90 are probability statements about a defined energy quantity. The labels are incomplete without a time horizon and modelling basis. They describe an assessment’s uncertainty distribution, rather than guaranteed turbine output or the percentage of its nameplate rating.
State what is exceeded, over which period.
A P90 estimate uses an energy threshold at or below P50 for the same distribution; it is not 90% of P50 by definition.
01 / Exceedance meaning
Read the probability in the correct direction.
A P90 energy threshold is intended to be met or exceeded with 90% probability under the stated model.
A P90 energy threshold is intended to be met or exceeded with 90% probability under the stated model. For the same energy distribution, P50 is the median threshold and P90 is at or below it. Higher exceedance probability corresponds to a more conservative energy threshold.
The SAM probability reference explains this exceedance convention. It is different from saying that P90 is the upper 90th percentile of an ordinary increasing cumulative distribution. That reversal would change the meaning of the result.
02 / Energy boundary
Define the quantity before reading its percentile.
Identify whether the result concerns gross generation, net energy at a defined output or energy delivered at a connection point.
Identify whether the result concerns gross generation, net energy at a defined output or energy delivered at a connection point. Retain losses, auxiliary use and operating assumptions included in that quantity.
A gross P90 estimate cannot be compared directly with a net P50 estimate and attributed entirely to probability. First align their physical and accounting boundaries. The annual-energy guide addresses that broader calculation scope.
03 / Time horizon
Separate a single year from a multi-year average.
A distribution of one-year energy and a distribution of an average over several years describe different quantities.
A distribution of one-year energy and a distribution of an average over several years describe different quantities. State the represented horizon explicitly, including the meaning of any operating-year or long-term-average result.
A P90 statement about a multi-year average does not establish that each individual year exceeds that value. Likewise, a single-year estimate does not describe the longest low-resource episode. Use time-sequence assessment for continuity questions.
04 / Uncertainty model
Keep variability separate from estimation uncertainty.
Interannual resource variability concerns differences between years.
Interannual resource variability concerns differences between years. Uncertainty in resource assessment, turbine performance or losses concerns imperfect knowledge of the estimate. A model may include both, but they are not interchangeable contributions.
The SAM wind-uncertainty documentation describes probabilistic annual-output modelling with stated uncertainty inputs. Record which contributions are included, their dependence and their basis. Adding percentages without considering definitions or correlation does not establish a valid uncertainty model.
05 / Distribution shape
Avoid an automatic percentage discount.
There is no universal rule that P90 equals 90% of P50.
There is no universal rule that P90 equals 90% of P50. Their difference depends on the assessed distribution and uncertainty assumptions. Different sites and boundaries can have different spreads even where their P50 values are similar.
A normal-distribution approximation, where used, is a modelling choice with assumptions. Do not apply it solely because two percentile labels are requested. Keep the method, tail behaviour and limitations with the reported values.
06 / Result review
Keep probability statements qualified.
An assessment handover should identify the quantity, horizon, reference resource, performance evidence, losses, uncertainty model and exceedance convention.
An assessment handover should identify the quantity, horizon, reference resource, performance evidence, losses, uncertainty model and exceedance convention. These allow a reviewer to understand what could change the stated thresholds.
P50 and P90 are not product certifications or production guarantees. No numerical MARSTON energy quantiles are established here. Connect a project-specific assessment to the long-term resource review before comparing candidate outcomes.
Continue the review with Long-Term Wind Resource & MCP Assessment Review.
Continue the review with Seasonal Wind Energy & Load Matching.
Continue the review with Wind Data Quality Flags & Aggregation Rules.
General background: NLR SAM: energy exceedance probability convention · NLR SAM: wind-resource and output uncertainty inputs. These references explain industry concepts; they are separate from the applicable MARSTON product and project documents.
Prepare the information for this review.
Use the checklist to organise the project information. Mark unavailable items and retain document references.
| Review item | Information to provide | Purpose |
|---|---|---|
| Quantity boundary | Gross or net energy, units and included losses | Identify what is exceeded. |
| Horizon | Single year or defined multi-year average | Interpret the probability period. |
| Contributions | Variability and estimation uncertainties included | Understand the distribution inputs. |
| Model | Distribution, dependence and calculation basis | Avoid automatic discounts. |
| Reported meaning | Exceedance convention and limitations | Separate estimates from guarantees. |
Engineering questions.
Is P90 90% of P50?
No. The ratio depends on the assessed distribution; it is not defined by the labels.
Is P90 normally higher than P50?
For the same energy distribution, a 90% exceedance threshold is at or below the 50% threshold.
Does P90 mean the upper 90th percentile?
No. In the exceedance convention it corresponds to a lower-tail energy threshold.
Does a multi-year P90 protect every single year?
No. A multi-year average and an individual year are different assessment quantities.
Can a P90 estimate guarantee continuous supply?
No. Energy probability and time-sequence continuity describe different requirements.
Are numerical MARSTON P50 or P90 values provided here?
No. Such values require a project-specific resource, performance and uncertainty assessment.
Technical enquiry
Bring your project information to the discussion.
P50 and P90 are probability statements about a defined energy quantity. Share the available information and the equipment scope you want reviewed in a technical quotation.