The landing calculation clears, but only just, and someone eventually asks the one question that matters: this number out of the manual — who obtained it, how, and with what built into it? The published distance is not a prediction of what the aircraft will do that day. It is the endpoint of a chain — flight test campaign, data reduction, modelling, expansion, approval — in which every link shifts the meaning of the number. Knowing where the margins were placed, and above all where they were not, changes how a performance page is read.
What the campaign measures, and what it never measures
A performance campaign does not produce tables, it produces points. On a heavily instrumented development aircraft — trailing air-data boom, external trajectography, wheel-by-wheel brake pressure sensors, thrust derived from engine parameters, weighing before every flight — the test crew repeats the same manoeuvre at chosen weights and pressure altitudes, holding a technique written down to the exact gesture.
The choice of points is not arbitrary at all: it covers the extremes of what the model will later have to reproduce. The two weight limits, a few pressure-altitude levels, the configurations retained in the manual, the published thrust ratings, and, for the stop, the brake conditions that penalise most. Between these points, nothing is flown.
Several reference conditions are imposed by the regulation, and none of them describes an ordinary day. Accelerate-stop is demonstrated with brakes worn to their allowable wear limit, not with new brakes that absorb energy better. The thrust retained is that of an engine at the minimum of its specification, not that of the test engine, which almost always produces more. On a dry runway, no credit is given for reversers. These are not margins in the usual sense: they are definitions of the case to be demonstrated, and they are paid for in published distance.
The numbers in the manual come from test flights, but they’re not just raw data. They’re adjusted to match standard conditions, and the worst-case scenarios are used to make sure the plane will always stop safely, even if things go wrong. The actual distance you’ll need in real life might be longer because of how pilots fly and how the runway really is.
An Airbus A320’s landing distance in the manual is based on tests with worn brakes, no reverser credit, and a strict flare technique. But in real operations, a pilot might flare a bit later, use reversers, or land slightly faster, adding up to 20% more distance than the manual suggests.
Scatter is not a flaw in the test
Repeat the same landing ten times, at the same weight, with the same crew, within the same hour: you will get ten different distances. Exact height at the threshold, the rhythm of the flare, the instant of touchdown, the delay before full brake pressure, a gust during derotation, a temperature deviation — each of these terms translates into metres, and they do not cancel each other out.
This scatter is information, not noise to be erased. Its width first tells you whether the technique is repeatable: a manoeuvre that does not repeat in testing, with a crew trained on a new machine, will never repeat in line operation and cannot be the basis for a published figure. When the cloud widens, it is not the measurement that needs fixing, it is the procedure.
Finally, it forces one idea to be kept in view: the final value is not a single test, it is a property of a population of tests.
Bringing each point back to reference conditions
No test is flown in the conditions of the table. Each point has to be shifted before it can be compared with the others: this is data reduction, and a good part of the result is built right there.
The measured wind component is removed, the runway slope is corrected for, the actual weight is brought back to the series’ nominal weight, temperature and pressure deviations are corrected, and the observed thrust is brought back to the specified minimum thrust. Every correction relies on a model: wind sensitivity is not measured on the day, it is calculated. Data reduction is already modelling, and the uncertainty of the corrections adds to that of the measurement.
Then comes the most underestimated step. The corrected points are used to identify the coefficients of a physical model: braking law as a function of speed, drag polar by configuration, thrust as a function of pressure altitude, the reaction times retained. It is this model, not the tests themselves, that is run across the whole envelope to generate every cell of the manual. The cell you are consulting was in all likelihood never flown: it was calculated by a model whose constants the tests fixed.
The package submitted to the authority therefore contains three things: the raw measurements, the reduction method and the model, and the tables that come out of it. Approval covers the entire chain, not the isolated numbers — which is why a manual figure cannot be touched up locally.
From the cloud to the single value: what the manual retains
That leaves the question that decides everything: within this corrected population, which value gets published? The criterion is not primarily statistical, it is behavioural. The figure must be obtainable by a crew of average proficiency applying the manual’s procedure, without exceptional skill or luck, and without any manoeuvre that would not be asked for in line operation.
Translated into practice, this criterion leads to retaining a central estimate of the corrected cloud — the fit that passes through the middle of the points — rather than an upper bound that would envelope all of them. The consequence is the one practitioners internalise worst: the published distance is not a distance that will never be exceeded. It is, at best, the distance obtained on average when flying exactly the certified technique, on a new aircraft.
The word ‘guaranteed’ nonetheless remains accurate, though in a different sense from the one usually assumed. What is guaranteed is the method: a value established under an approved protocol, on real tests, with conservative reference conditions. The guarantee covers how the number was built, not the fact that it will never be exceeded.
The trap. Setting the manual’s landing distance in flight against the announced runway length, seeing that there is margin left, and stopping there — typically during a diversion, a late runway change, or a runway shortened by works, where the departure calculation no longer applies. The manual value is a central estimate obtained with the test technique on a new aircraft: about half of the corrected tests that produced it stopped further along, and an ordinary line landing — a comfortable flare, progressive braking, unhurried reversers — departs from it considerably more still. Reasoning that ‘the manual is already conservative’ is wrong twice over: it is conservative in how the case is defined, it is not conservative in the value itself.
Where the margins are really placed
Once it is accepted that the measured value is central, it becomes clear that the margins have been placed elsewhere, and differently depending on the phase. Three mechanisms coexist; confusing them leads either to double-counting one or forgetting another.

The first is the definition of the case to be demonstrated, already described. The second is the definition of the required distance, which is not the measured distance: at takeoff, the length to retain is the greater of the critical-engine-failure case and one hundred fifteen percent of the all-engines case; for the stop, a distance equivalent to two seconds travelled at decision speed is added to what was measured. The third is the line-operation factor, which exists only for landing.
| Phase | What is measured, and how | Margin built into the definition | Margin added at line operation |
|---|---|---|---|
| Takeoff | Up to the 35 ft screen, critical engine failed, minimum specified thrust | The greater of the two cases: critical engine failure, or 115% of the all-engines case | No factor: the required distance must fit within the usable length |
| Accelerate-stop | Brakes at their wear limit, no reverser credit on dry | Addition of a distance equivalent to two seconds at decision speed | No factor, same rule as takeoff |
| Landing | From 50 ft to full stop, maximum manual braking, no reverser credit on dry | None: the published value is the central estimate itself | Must fit within 60% of the usable length on dry; required length increased by 15% on wet |
Reading down the columns is instructive. At takeoff and for the stop, the margin sits in the calculation rule and operations add nothing on top: the required distance must fit, full stop. For landing, it is the opposite: the certified value is bare, and the entire cushion is deferred to the dispatch factor. A ‘manual’ distance and a ‘required at dispatch’ length are not comparable.
Two clarifications avoid miscounting. Not everything in the case definition points toward severity: on a wet runway, the screen height to be cleared at takeoff is lowered, which shortens the distance to be demonstrated. This is a deliberate trade-off against the higher accelerate-stop distance required on wet, not an extra precaution; filing it among the margins amounts to crediting the aircraft twice. And this distribution holds for a turbojet transport aircraft. In Class B performance — light twins, low-weight turboprops — the logic does not carry over: correction factors also apply at takeoff, the nature and condition of the surface enter directly into the required-distance calculation, and the fraction of runway to be respected at landing does not read the same as in Class A. A reflex carried over from one class to another is not off by a few metres, it is off in method.
Since the generalisation of arrival assessment, a separate family of data must be added: the landing distance of the day, computed in flight from the reported runway condition, the actual configuration, the braking mode selected and any reverser credit, increased by a margin on the order of fifteen percent. These are not the certified data: they assume a technique representative of operations and do not always carry the same approval status.
What the margins do not cover
The margins described above cover one precise thing: the gap between a new aircraft flown with the certified technique under measured conditions, and a production aircraft flown normally under reported conditions. Everything else is not covered.
- Airframe drift. Engine degradation is bounded by the retained minimum specification, but nothing bounds the drag of an aged airframe: fatigued seals, ill-fitting doors, repairs, control-surface misalignment.
- Technique drift. A flare stretched by a few seconds, braking resumed later for comfort, reversers selected late: each of these actions is defensible on its own, and their combination consumes a considerable share of the landing factor.
- Wind measurement. The reported value is an average, at a single point, at a conventional height, with no certain relation to what the aircraft encounters during the flare. Hence the weighting that credits at most half the headwind component and counts at least one and a half times the tailwind component.
- Runway condition. Beyond dry, the data rest on modelled friction coefficients and a rough correspondence between the reported code and the braking actually available. The certifier’s ‘wet’ runway is a thin, continuous film of water; a runway with puddles and ruts is not the same family.
- Failure combinations. Configuration or system supplements are established separately and presented as additive. Additivity is a convenient assumption, not a test result.
The right reflex. Faced with a performance page, establish three things before reading the number: where it comes from (certified manual data or manufacturer operational-support data), what technique it assumes (maximum manual or autobrake, reversers counted or not), and what margin it already contains. A margin added without this check often duplicates one already present, and is missing where it would actually have been needed.
What to go and check in your own documentation
Three checks are done once and for all, by type, and are then passed on better than any general reminder.
First, the assumptions behind the published landing distance for your aircraft: threshold height and speed, flare technique, demonstrated braking mode, the treatment of reversers on each runway condition. Two types within the same category can differ on this last point, and the gap is far from marginal when the runway is short.
Next, the status of the arrival data used on board: approved or advisory, margin included or to be added, and how the operator has settled the question. A policy that imposes an additional margin on data that already contains one is not safer: it is less practicable, and ends up being bypassed on the day the runway is limiting.
Finally, the grey area that no text fills: the correspondence between the runway condition reported by the aerodrome operator and the braking the aircraft will actually find. This is the one link whose data has no measured lineage on the aircraft in question, and it is where runway excursions take root. For those preparing a type rating or a supervisory role — the FL330 training programmes devote part of their performance modules to it — knowing how to answer ‘where does this number come from’ is worth more than reciting the factor: the former lets you reason through an unforeseen case, the latter stops at the case foreseen.
The applicable certification specifications, the type’s approved flight manual, and the operational documentation validated by the authority prevail in all circumstances over the content of this article, whose purpose is explanatory.
Questions fréquentes
Is the flight manual's landing distance a maximum distance?
No. It is a central estimate of a population of corrected tests, obtained by flying the certified technique on a new aircraft. A significant share of the tests used to establish it lies beyond the published value. What is guaranteed is the method by which it was obtained and compliance with conservative reference conditions, not the fact that the distance will never be exceeded: safety comes from the factor subsequently applied at dispatch.
Why is the landing margin applied at dispatch rather than built into the certified data?
Because the two numbers do not serve the same purpose. The certified data describe a physical capability of the aircraft, measured under defined conditions, and must remain comparable from one type to another. The dispatch factor covers the gap between this capability and actual operation: line technique, the condition of the aircraft, and uncertainty in the reported conditions. Keeping them separate allows the operating rule to evolve without touching the type certificate.
Why is the accelerate-stop distance demonstrated with worn brakes?
Because the figure must remain valid throughout the brakes’ entire service life, up to the allowable wear limit. A new brake has more material and absorbs energy better; using its performance would publish a distance that a fleet aircraft could no longer achieve near the end of its brake life. The same reasoning leads to retaining an engine at the minimum of its specification rather than the test engine, which almost always produces more.
Why is only a fraction of the reported headwind retained in performance calculations?
Because the reported value is an average measured at one point, at a conventional height, which does not describe the wind actually encountered during rotation or the flare. The weighting credits at most half the headwind component and counts at least one and a half times the tailwind component. This is not a comfort margin: it compensates for a known measurement uncertainty whose consequences are asymmetric.
Do contaminated-runway data carry the same status as the flight manual tables?
Not always, and that is the first thing to establish. Part of this data is provided by the manufacturer as operational support, established on modelled friction coefficients rather than on tests covering every runway condition. Its use depends on what the operator has written into its manual and on what the authority has accepted. The correspondence between a reported runway condition code and the braking actually available remains the least substantiated link in the chain.
Why have most cells in a performance table never actually been flown?
Because it is physically impossible to measure every combination of weight, pressure altitude, temperature, configuration and thrust setting. The campaign measures a limited number of well-chosen points, at the edges of the envelope, and these points are used to identify the coefficients of a physical model. It is this model that is then run across the whole envelope to generate the tables. Approval covers the whole — measurements, reduction method and model — which is why a single cell cannot be touched up on its own.

Leave a Reply
You must be logged in to post a comment.