Low Visibility Operations Approval: The Demonstration That Conditions the Grant

Piste dans un brouillard dense de nuit, barres de feux d'approche visibles en enfilade sur l'asphalte mouillé

An operator files its Category III application. The aircraft is certified for it, the crews are trained, the procedure is written. Yet the authority asks for something that appears in no manufacturer’s manual: by what means will the operator know, a year or two later, that its automatic landing system has degraded. That is the question that conditions the grant, and it is the one that most applications address last.

The Approval Validates a Chain of Evidence, Not an Intention

Low visibility operations approval belongs to the family of special approvals: it is not simply declared, it is examined. It is issued to a named operator, for a named aircraft type, for a named class of operations — reduced-visibility take-off, Category II approach, Category III approach with or without decision height — and its exercise then plays out runway by runway. Two airlines operating the same type may hold different privileges, and the same aircraft may be qualified for an operation on one runway and not on the one next to it.

The application rests on four pillars. The aircraft: what the approved flight manual has actually demonstrated. The aerodrome: what the runway, its aids and their protection allow at the moment of approach. The crew: what training and checking guarantee, individually and as a constituted crew. The organisation: procedures, minima calculation, the minimum equipment list, maintenance of the systems concerned. A fifth element links the four, and it is this element that separates an accepted application from one sent back: the data that will prove, after the grant, that the four still hold.

This is the shift in perspective to understand. The authority is not trying to establish whether the operator can land an aircraft in reduced visual range: the manufacturer has demonstrated it, the simulator shows it. It is trying to establish how the operator will notice when that is no longer true. An automatic landing system does not fail outright; it drifts, one card at a time, one runway at a time, one crew at a time. The monitoring arrangement is therefore not an appendix to the application, it is its central argument.

The key is not just proving you can do it, but showing how you’ll catch any future problems. The system won’t just break – it will slowly get worse, and the regulator wants to see you have a plan to spot that before it becomes dangerous.

Inspector's desk covered with stacked files, a stamp and pen resting near the edge, low raking window light
The examination does not judge an intention: every privilege requested must be backed by a verifiable piece of evidence.

A nuance recurs in almost every examination: the approval and its use do not sit at the same level. The approval is granted to the operator for a capability; its use is decided flight by flight, by weighing that capability against the actual state of the aircraft and of the aerodrome. An airline may hold Category III without decision height and yet exercise, on a given morning, only Category I, because a system has been deferred, because the runway lacks the required lighting, or because the aerodrome has not implemented its procedures. Confusing the two levels produces the most common deviation in the field: a crew that believes it holds a privilege because its airline holds it.

When the type is new to the fleet, the examination is spread out over time. The authority requires a demonstration period during which the operator flies, collects data and reports, before the lowest minima are opened up. This period is not an administrative formality: it is the first sample produced by the monitoring arrangement, and its quality foreshadows the entire relationship that follows.

The Aircraft’s Share: What the Flight Manual Demonstrates, and What It Bounds

The starting point is not regulatory, it is certified. The approved flight manual states the aircraft’s demonstrated capability and, above all, the configuration in which it was demonstrated: number of autopilot channels engaged, autothrottle status, engine-out case, runway slope and elevation, crosswind and tailwind components retained for automatic landing. These limits often bite before the regulatory minima do. A forecast crosswind beyond the demonstrated value takes automatic landing off the table before visibility even comes into play, and this is a far more frequent cause of diversion than a system failure.

Fail-Passive, Fail-Operational

The structuring distinction is not about the number of computers but about behaviour after a single failure. A fail-passive system stops cleanly: no significant imbalance, no abrupt deviation in flight path or attitude, but no one left at the automatic controls. It therefore requires a decision height, because a height is needed at which a crew can take over. A fail-operational system carries on: after a single failure, the remaining redundancy flies the flare, the landing and, depending on the architecture, the rollout guidance. It is this system, and this one alone, that opens up operations without a decision height.

From this architecture follows a notion that crews handle without always seeing where it comes from: the alert height. It is not an operational minimum, it is a design boundary. Above it, a failure that causes loss of redundancy causes loss of the privilege, and a go-around is flown. Below it, the same failure leaves a system sufficient to finish the job, and the landing continues — because going around at that height would expose the aircraft more than continuing would. An application that presents the alert height as merely a higher decision height has misunderstood what it is asking for, and this shows immediately.

On a Boeing 787 with a fail-operational system, the alert height might be set at 150 feet. If a system failure occurs above this height, the aircraft must go around. But if the same failure happens at 100 feet, the remaining systems can complete the landing safely. This is why the crew must understand this boundary – it’s not about the weather, it’s about the system’s design limits.

Alert Height and Decision Height on an Approach ProfileSide view of a precision approach: the flight path descends from the left down to the runway threshold. Two horizontal dashed lines mark the alert height, higher, and the decision height, lower. Above the alert height, a failure that causes loss of redundancy requires a go-around; below it, a fail-operational system continues the landing.heightdistancethresholdalert heightdecision heightfailure above:go-aroundfailure below:landing continues
The alert height is not a minimum, it is an architectural boundary. Above it, loss of redundancy removes the privilege and requires a go-around. Below it, the remaining system finishes the landing, because going around at that height would be more exposed than continuing.
Two Automatic Landing System Architectures, Two Operating Regimes
Criterion Fail-passive Fail-operational
Behaviour after a single failure Clean disconnection, without significant deviation in flight path or attitude Approach, flare and landing continue with the remaining redundancy
Who finishes the landing The crew, manually, or a go-around The remaining system, with the crew monitoring
Decision height Required: a height is needed at which to take over May be absent, depending on what has been demonstrated
Alert height Not applicable Defined: above it, go-around; below it, continuation
Operations generally accessible Category II, Category III with decision height Category III up to the lowest demonstrated minima
What monitoring primarily looks at Height and quality of manual take-overs Loss of redundancy on approach, track-keeping during rollout

There remains orderly degradation. The minimum equipment list does not only state what authorises departure; it defines the fallback path, from Category III without decision height to Category III with decision height, then to Category II, then to an ordinary precision approach. This path must be executable in flight, with the charts carried on board and the fuel loaded, not merely readable on the ground. A solid application shows where the fallback is decided, by whom, and on the basis of what information is actually displayed on the flight deck.

The Aerodrome’s Share: The Approval Is Decided on the Ground

An approval does not land; it is the runway that receives the aircraft. The aerodrome share of the application consists of demonstrating that the operator knows, before each approach, whether the infrastructure is in a state that makes the operation legitimate. This covers the existence of low visibility procedures implemented by the aerodrome, the measurement and dissemination of runway visual range at the required locations, the lighting — approach lighting, runway centreline, touchdown zone, taxiway centreline, stop bars — and a standby power supply whose switch-over time remains compatible with the intended operation.

The most poorly understood point is signal protection. Around a localizer, a distinction is made between a critical area, where nothing parks or moves while the system is in use, and a sensitive area, larger, whose occupation is controlled only when operations require it. The mechanism is physical: a large metal surface placed within the field re-radiates the signal, and the aircraft no longer receives the expected composite signal but a distorted one. The error appears as a displacement of the course line. A coupled autopilot has no way of distinguishing a displaced course line from a true one: it follows. This is why the implementation of low visibility procedures pushes back holding points and spaces out movements, at the cost of a reduced aerodrome throughput.

Localizer Sensitive Area and Holding Point Set-BackTop view of a runway and its localizer, positioned beyond the end opposite the arrival direction. A dashed rectangle, the sensitive area, encloses the runway and the antenna. A taxiway joins the runway: the ordinary holding point, shown as a solid line, lies inside the sensitive area, while the holding point used in low visibility, shown dashed, is set back outside that area.arrivalrunwaylocalizer sensitive arealocalizerordinary holding pointlow visibility holding pointtaxiway
A metallic mass within the sensitive area displaces the course line that the autopilot follows. Hence the set-back of holding points and the spacing-out of movements as soon as low visibility procedures are in force: aerodrome throughput drops so that the signal remains true.

Runway visual range, for its part, is measured at several points — touchdown zone, midpoint, stop end — and not all carry the same weight. The touchdown zone value always governs; the others become decisive as the minima get lower, because the crew must still be able to taxi and clear the runway after touchdown. This hierarchy explains a common scene while holding: a value that rises at the midpoint, while the touchdown value stagnates, changes nothing about the decision.

The trap. « Low visibility procedures in force » is not synonymous with « everything is operational. » The procedures describe a mode of operation for the aerodrome; they do not guarantee that touchdown zone lighting, a stop bar or a visual range sensor is available that day. The information travels by notice to airmen or over the frequency, often late, and it is the operator who bears the responsibility of recalculating the applicable minimum. An approach begun on yesterday’s value is an approach begun without a valid minimum.

The right approach. Set the ceiling for the day before the descent, not during it. The approach briefing benefits from stating three things in this order: the best operation the state of the aircraft allows, the best the state of the aerodrome allows, and the more limiting of the two, which becomes the minimum for the day. Any degradation announced afterwards is then compared against a decision already made, rather than against an intention.

The Crew’s Share: What Is Demonstrated in the Simulator

Training for low visibility operations is demonstrated through a programme and through records, never through a certificate. Ground training deals with what crews often believe they already know: the logic of displayed capability annunciations, the hierarchy of degradations, the calculation of minima, what low visibility procedures change during taxi and in reading stop bars. The simulator deals with what cannot be improvised.

What the examination cares about is not the series of successful automatic landings — they almost always are — but the failure cases and their placement in time. Loss of a system just above the alert height, then the same just below it, so that the crew experiences the difference in the response required. Engine failure during a coupled approach. Degradation of the displayed capability at a few hundred feet, when the briefing is already done and the workload is high. Visual range dropping below the minimum between the final approach fix and the decision height. Lighting failure. Go-around from a very low height, and on some fleets from touchdown. Manual take-over after a low disconnection, which is the most demanding manoeuvre in the repertoire and the least practised.

To this are added the recent experience rules — a number of approaches and landings of this type within a given period, restored in the simulator when the period has lapsed — and pairing restrictions: two pilots freshly checked out on the type do not constitute the same crew as two experienced ones, and the operator’s policy must state this in black and white. Finally comes the detail that proves costly in audit: on a single type, several automatic flight system standards often coexist within the fleet. The crew must know, before the briefing, which one they have in front of them, and the operations manual must make this reading immediate rather than something to be deduced.

The real difficulty is not the flying. It lies in the decision architecture: who calls what, what exactly « continue » means when the capability annunciation changes at a few hundred feet, and at what point the decision ceases to belong to the pilot flying and belongs instead to the procedure. This is what an inspector observes during a session, and it is what a poorly written programme leaves to each individual’s judgement.

After the Grant: The Monitoring That Keeps It Alive

The privilege wears down. An approved operator whose crews fly only a few real low-minima approaches a year holds a paper capability, and the authority knows it. This is why the approval comes with a collection and analysis programme that, unlike the approval itself, never stops.

What is collected is simple to state and hard to sustain: every approach and every automatic landing conducted under the approval, classified by type and by registration, with their outcome — completed, degraded, interrupted, track deviation, system failure, ground-based cause — and the context that allows that outcome to be attributed. Without the context, the table says nothing. With it, a registration that fails twice in a row at the same point in the sequence becomes a signal, whereas three failures scattered across three different runways are not one. The safety objective assigned to an automatic landing concerns a hazardous event probability so low that no operator sample, however long, could ever measure it. What the series detects is not the residual probability, it is the drift.

Monitoring the Rate of Successful Automatic LandingsCurve of the successful automatic landing rate over time, without numerical values. A horizontal dashed line represents the action threshold set by the operator, and a vertical line separates the initial demonstration from ongoing monitoring. The curve stays above the threshold during the demonstration, dips below it during monitoring at a cluster of manual take-overs, then rises again after the corrective action.success ratetimeaction thresholdinitial demonstrationongoing monitoringcluster of manual take-oversrecovery after action
It is not the rate that is informative, it is the cluster. Three failures scattered across three runways do not carry the same weight as two consecutive failures of the same registration at the same point in the sequence. The action threshold serves to trigger an investigation, not to pass judgement.
Classifying an Automatic Landing: What Is Recorded, What Is Done With It
Outcome observed Context enabling attribution Action taken
Landing completed Runway, registration, systems engaged, wind Feeds the series, no action
Degradation of the displayed capability on approach Height of occurrence, system standard, associated message Technical analysis of the system concerned
Low disconnection and manual take-over Height, protection areas safeguarded or not, wind compared with demonstrated limits Excluded from the rate if the cause is ground-based, investigated if it is on board
Course or alignment deviation at touchdown Ground movements, published unavailability, visual range readings Reported to the aerodrome if the cause is the infrastructure
Go-around, automatic or decided Trigger, height, visual range at the time Counted as an event, never by default as an aircraft failure

The trap. Currency-maintenance automatic landings, flown in good weather, poison the log both ways. Conducted on a runway whose protection areas are not safeguarded — nothing requires it when visibility is good — they produce deviations and disconnections that are not aircraft defects and that drag a rate down. Excluded from the log on principle, they instead deprive the fleet of the only abundant data series it has. The sorting is done on the attribution of the cause, never on the weather of the day.

The right approach. Have the context recorded at the time of the event, in the same report: low visibility procedures in force or not, runway classification, wind compared with the flight manual’s demonstrated limits, systems engaged, system standard. A note written the following day is a reconstruction. And setting the action threshold in advance, along with what it triggers — technical investigation, restriction of an aircraft, notification of the authority — avoids having to negotiate the threshold on the day it is crossed.

This material does not serve the operator alone. It is what the authority looks at first when adding a registration, changing a system standard, extending to a new aerodrome, or following a notable event. An operator that presents a continuous series, attributed and annotated, obtains its changes quickly. An operator that presents a global rate without context is asked for the detailed sample, and the examination starts over from zero.

What to Check in Your Own Documentation

Three checks are worth the time they take. The first: are the automatic landing limits from the flight manual carried over exactly into the operations manual, or has a more permissive operator value crept in between the two over successive revisions? The second: is the degradation path written as a set of actions, with the information on which each step is decided, or merely as a correspondence table? The third: who receives the automatic landing log, how often, and what actually happens when the threshold is crossed — the name of a function, not that of a department.

One grey area remains, and it is acknowledged as such. The status of an aerodrome’s aids is published by the aerodrome, but the applicable minimum is calculated by the operator: between the two, the propagation delay of an unavailability is not zero, and no text makes it zero. Enhanced vision systems, moreover, shift the boundary of the field — they open up operational credits that fall under neither the same demonstration nor the same training, and an operator who mentally adds them to its Category III approval is confusing two different matters.

For anyone preparing a type rating or taking up a role in this field, the most effective reading order is not that of the regulation: start with the automatic landing chapter of one’s own type’s flight manual, continue with the part of the operations manual that calculates minima, and finish with the regulatory text. Understood this way, what each requirement protects becomes clear, and the FL330 training courses approach the subject from the same end.

The applicable regulation, the type certification specifications and the operator’s approved documentation take precedence over this article, whose purpose is to shed light on the mechanism, not to substitute for it.

Questions fréquentes

Is the airline's Category III approval enough to land the aircraft without a decision height?

No. The approval assigns a capability to the operator; its exercise is decided flight by flight. The aircraft must have the required architecture and state, the runway must have the corresponding aids and protection, the aerodrome must have implemented its low visibility procedures, and the crew must be qualified and current in recent experience. The most limiting of these conditions sets the minimum for the day.

What is the practical difference between a fail-passive system and a fail-operational system?

It shows in the behaviour after a single failure. A fail-passive system disconnects cleanly, without an abrupt deviation, but the crew must take over: a decision height is therefore required. A fail-operational system continues the approach, flare and landing with the remaining redundancy, which permits operations without a decision height when the demonstration covers it.

What exactly is the alert height for?

It is an architectural boundary, not an operational minimum. Above it, a failure that causes loss of redundancy removes the privilege and requires a go-around. Below it, the remaining system is sufficient to finish the landing, and continuing exposes the aircraft less than going around at that height would. It exists only on fail-operational architectures.

Should currency-maintenance automatic landings flown in good weather be logged?

Yes, but with their context. On a runway whose protection areas are not safeguarded, a disconnection or a course deviation reflects a signal disturbance, not an aircraft defect: it is logged and classified outside the rate once attributed. Excluding them outright would deprive the fleet of its only abundant series; counting them indiscriminately would distort the trend.

What happens to the minimum if touchdown zone lighting or a visual range sensor is unavailable?

It is recalculated. The operator’s documentation sets out, element by element, the effect of an unavailability on the applicable minima, which most often leads to a fallback to a less demanding operation. The aerodrome’s implementation of low visibility procedures does not exempt from this check: it describes a mode of operation, not the state of each piece of equipment.

Can an operator lose its low visibility approval?

Yes, and it rarely begins with a withdrawal. Monitoring of automatic landings reveals a drift, the operator first restricts an aircraft or an operation, and the authority is informed. If the conditions that underpinned the grant are no longer met — a training programme not maintained, monitoring without usable data, failing maintenance of the systems concerned — the privilege may be limited or suspended.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

FL330, c’est toute l’aviation professionnelle — explorez nos autres espaces.