Who Gets the Delay Minute: Why the Code Names the Last Link, Not the Cause

A turnaround coordinator enters delay data on a handheld device beside a parked aircraft on a dimly lit airport ramp at dusk.

A delay code looks like an answer. It is a subtraction with a label stapled to it, written by somebody with three other aircraft to push and a deadline measured in minutes. The code names whoever was holding the aeroplane when the clock crossed the threshold, which is almost never whoever created the constraint. Read that way, the fact that flow management sits near the top of most punctuality reports says less about controllers than about how the last link in a chain gets the invoice.

The minute is a subtraction, not a diagnosis

Strip the vocabulary away and a delay is one arithmetic operation: actual off-block time minus scheduled time of departure. That subtraction contains no information about causes. It does not know that the aeroplane was fuelled late, that the inbound arrived from a diversion, or that a sector two countries away was regulated. It knows a number of minutes, and the number is identical whatever produced it.

Everything else on the delay line is added afterwards by a human being. That is the first thing to hold on to when a monthly report shows a category climbing: the categories are annotations, not measurements. The measurement is the clock.

Two conventions shape what even counts. Most operators publish on-time performance against a fifteen-minute tolerance, so a departure at plus fourteen counts as on time while a departure at plus sixteen is recorded in full. Whether the smaller event is coded at all depends on a separate threshold set by the operator, and the two are not always the same number. And the reference event is usually off-block, not airborne. Everything that happens after the tug disconnects — the taxi queue, the de-icing pad, holding short waiting for a slot — either lands in a different reporting stream or vanishes. A flight can be perfectly punctual off-block and forty minutes late in the air, and the departure file will show nothing.

The trap. Comparing punctuality between two operators, or two bases, without first checking the threshold and the reference event compares two different quantities. A carrier measuring at three minutes off-block and a carrier measuring at fifteen minutes at the gate are not in the same conversation.

A delay code is just a number showing how late a flight was, not why. The person who writes the code often doesn’t know the real cause, and the system doesn’t record it. The code just says who was holding the plane when the delay happened, not who caused it.

An Airbus A320 at Paris Charles de Gaulle is delayed 50 minutes because it missed its slot after an 8-minute late catering truck. The code says “flow management,” but the real cause was the catering delay. The system doesn’t show the small issue that started the chain of delays.

Who writes the code, and against what deadline

In most organisations the code is entered by the turnaround coordinator or the station duty officer, into the movement or departure control system, within minutes of pushback. At that instant the person coding often knows what stopped them last, and only that. The fuel truck that arrived late may have been late because the previous stand ran over, which happened because a tow was held for a crossing aircraft. None of that reaches the field.

Amendments are usually possible for a defined window afterwards, and are usually not made. By the time anybody has a clearer picture, the figure has already fed the daily operations report, the handling agent’s performance dashboard and, in some structures, an internal cost transfer. Correcting a code means explaining why the first one was wrong, in front of the department it would move the minutes to.

That is the second thing to hold on to. Delay coding is not a neutral act of record-keeping. It sits inside a set of incentives: ground handling contracts with punctuality penalties, engineering budgets charged for technical delay, crew planning defending its rosters, and a legal function that will later have to argue whether a disruption was within the carrier’s control for passenger compensation purposes. The code does not decide any of those questions, but it is the first document anyone reaches for.

Systems differ on how much nuance they will accept. Some allow the minutes to be split across several codes with one designated as primary; others force a single code for the whole block. Where only one code is allowed, the largest single contiguous block of lost time wins by construction — and the largest block is very often the one created downstream, after a small upstream event has already knocked the flight out of its window.

What the code bands actually prove

The industry standard list — the numeric delay codes published in IATA’s airport handling material, extended locally by most operators with their own sub-codes — is organised in bands. The band is the useful part. The individual number is the part people argue about.

Delay code bands: what each one establishes, and what it leaves out
Band What it establishes What it hides
Passenger and baggage handling The aeroplane was held for people or bags at this station Whether those people were late because their inbound was late
Cargo and mail Load was still being processed at departure time Whether the load was accepted against an unrealistic cut-off
Ramp handling and aircraft servicing A ground process ran past the off-block time Whether the process started late or ran long, and why
Technical and aircraft equipment A defect or a maintenance action held the departure Whether the defect was known and deferred at a previous station
Aircraft damage and automated equipment failure Damage was found, or a handling or data system failed How long the inspection and release decision itself took
Flight operations and crewing Crew, documentation or dispatch held the departure Whether the crew were late because of a rotation, not a person
Weather Weather at this station affected the turnaround Whether the station’s de-icing or contingency arrangements made a bad hand worse
Flow management, airport and authorities An external constraint sat between the flight and the runway The reason for the constraint, and what pushed the flight into it
Reactionary The delay came from a previous leg or another flight Everything about the origin of that previous delay

Read the right-hand column in order and a pattern appears: the further down the list, the less the code says about causation and the more it says about position in a queue. The two bands that dominate most European punctuality files — flow management and reactionary — are precisely the two that carry the least causal information.

The slot arrives after the delay has already happened

Air traffic flow and capacity management works by protecting a constrained element: a sector, an approach sequence, an aerodrome’s arrival rate. When demand is forecast to exceed what that element can absorb, a regulation is applied and departing flights are given a calculated take-off time so that the queue forms on the ground rather than in the air. The delay is served at the departure airport, but it was created somewhere else entirely.

A man in uniform operates a laptop at an airport gate with an airplane in the background.
Digital systems often obscure the real reasons for delays.

In the European network the calculated take-off time comes with a tolerance window running from five minutes before to ten minutes after. Airborne inside that window, the flight has used its slot. Outside it, the slot is gone and a new one must be requested — and the new one reflects the queue as it stands now, not as it stood an hour ago.

A flight caught by more than one regulation is allocated against the most penalising of them, and each regulation carries its own reason, drawn from a short structured list that includes capacity, staffing, equipment, weather, aerodrome capacity, airspace management, industrial action and special events. The operator’s delay code is far coarser. It records that flow management was involved and, at best, whether the constraint sat en route or at the destination. The network’s own reason for the regulation does not survive the transfer into the punctuality file, so nothing in that file separates a weather regulation from a capacity one.

Then causality inverts. A turnaround loses eight minutes to a late catering uplift — not enough to breach the reporting threshold, not enough to be worth a phone call. Those eight minutes push the aircraft past the end of its tolerance window. The new calculated time comes back forty-five minutes later because the sector is now full. The file records a fifty-three-minute departure delay, coded flow management. The eight minutes that actually caused it appear nowhere, and the ramp closes the day with a clean sheet.

The trap. A flow management code sitting on top of a small primary event is the single most common false attribution in punctuality reporting. Wherever the coded delay is much larger than any process that was observed on the ramp, assume a lost slot until the timestamps say otherwise.

Reactionary delay is honest about the mechanism and silent about the origin

Network reporting in Europe consistently shows reactionary delay — delay inherited from a previous leg or from another flight — as the largest single share of the total. That is not a statistical curiosity; it is what happens when aircraft fly four or five sectors a day against schedules built with thin buffers. One primary event in the morning propagates until a buffer somewhere is large enough to absorb it, or until the rotation ends.

The reactionary band splits by what was late: the aircraft itself, the flight crew, the cabin crew, connecting load or passengers, or an operations control decision such as a swap, a consolidation or a reroute. Every one of those codes is accurate. Every one of them points backwards one step and then stops. It says the previous leg was late. It does not say why the previous leg was late, and it does not say whether the buffer that failed to absorb it was undersized.

Which means root-cause work on reactionary delay is not a coding exercise at all. It is a query: take the aircraft registration, walk the rotation back through the day until you find the first leg with a primary code, and look at what the schedule gave you between each pair of legs. Two carriers with identical delay-code profiles can have completely different problems — one with a maintenance issue, one with a schedule that never had a chance.

Three cascades, coded three ways

The same file read as a code and as a timeline
What happened What the code says What the timeline shows
Catering uplift eight minutes late; slot tolerance window missed; new calculated time forty-five minutes later Flow management, en route An eight-minute ramp event and a lost slot. The constraint existed either way; the flight would have made the first slot
Inbound arrives twenty-five minutes late after a technical stop; operating crew would exceed their duty limit; replacement crew called out Crew rotation, reactionary A technical event two legs earlier, plus a roster with no standby at that base at that hour
Fog at destination cuts the arrival rate; a regulation is applied; departure held on stand Flow management, destination Weather, at the other end, in nobody’s control — and correctly outside the carrier’s exposure
Loading equipment unserviceable; hold loading finishes eighteen minutes late Ramp handling A ground service level agreement issue, correctly attributed for once

Only the bottom two rows put the code and the cause in the same place. That is the practical test for any coded delay: not whether the code is defensible on the night, but whether somebody reading the timeline a week later would still choose it.

The timestamps that do carry evidence

Where collaborative decision making is in place at an aerodrome, the milestone timestamps are the objective record the codes are not. They are stamped by the systems that handle the flight rather than written up afterwards from memory, they are shared between the operator, the handler, the aerodrome and the network, and — the operator’s own target off-block time aside — nobody edits them to defend a budget.

Departure milestones and what each one can prove
Timestamp What it proves What it cannot prove
Estimated off-block time, as filed The intention declared in the flight plan That the intention was ever realistic
Target off-block time The operator’s own commitment, updatable as the turnaround progresses That it was updated honestly rather than optimistically
Calculated take-off time That an external constraint applied, and how much of it Why the constraint existed, or which flight lost which slot
Target start-up approval time When the ground movement plan expected this aircraft to move Whether the aircraft was actually ready then
Actual start-up request time When the crew declared themselves ready — the cleanest single indicator of turnaround performance What made them ready late
Actual off-block time The number the whole punctuality report is built on Anything at all about causation
Actual take-off time Whether the slot was used or lost Whether the taxi time was normal for that configuration

The gap between the actual start-up request and the actual off-block is the interesting one. Large and repeated, it points at the ground movement plan or at a slot. Small, with a late request, it points squarely at the turnaround. No delay code distinguishes those two situations; the pair of timestamps does it in one subtraction.

The right reflex. When you code a delay that ends in a lost slot, record the primary event separately — its own code if the system allows split minutes, free text if it does not. Two lines cost nothing and are the only trace the eight minutes will ever leave. A month later, that free-text field is what tells the difference between a capacity problem and a catering contract.

What a delay code is honestly good for

None of this makes the code worthless. It is a good instrument for the job it was built for: aggregating a network’s lost time into categories that can be compared between stations and between seasons, driving handling contracts, and giving operations control a fast common vocabulary during a disruption. As a queue-position record it works.

What it cannot do is carry the weight people put on it. It is not a fault finding. It is not, by itself, an argument that a disruption was outside the carrier’s control for passenger compensation purposes — the facts behind the code are what matter there, and a flow management code sitting on a self-inflicted lost slot will not survive contact with a timeline. And it is not the starting point of a root cause analysis, because it was written before the cause was known.

The operational rule that follows is simple enough to apply at a handover. Never present a delay code as a finding. Present the timeline, and let the code be what it is: a note about who was holding the aeroplane when the clock ran out. The arguments turn out to be remarkably consistent from one network to the next, which usually means the problem is structural rather than local — worth comparing with people who code delays every day on the FL330 forums.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

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