Where the operator ends and the workshop begins: the risk that falls between two management systems

A technician and supervisor in a hangar examine a deferred defect on an aircraft wing at dusk, highlighting the gap between maintenance and operations

An operator can run a mature safety management system. A continuing airworthiness organisation can hold a clean audit record. An approved maintenance organisation can have exemplary human-factors training and a compliance department that finds its own faults before the authority does. All three can be true at once and the aeroplane can still be dispatched with a deferred item nobody is tracking, a limitation the crew has never been told about, or a finding that was reported into a system the people who needed it do not read. The reason is not negligence. It is that each management system is built to see hazards inside its own boundary, and the hazard here lives in the transfer — a place that has procedures on both sides, an owner on neither, and no category in either reporting scheme.

Three organisations, one aeroplane

In the European structure the responsibilities are deliberately separated. The operator holds the air operator certificate and remains responsible for the airworthiness of every aeroplane it flies. The management of continuing airworthiness — the airworthiness programme, the modification and airworthiness directive status, the planning, the airworthiness review — sits with a continuing airworthiness management organisation, which may be a department of the airline or a separate company. The physical work and the release to service belong to an approved maintenance organisation under Part-145, which may be a subsidiary, a competitor, or a contractor on another continent.

Each of the three has its own accountable manager, its own compliance monitoring function, its own safety reporting channel, its own record system and its own authority oversight. Each is audited against its own approval. That architecture is sound: it puts a technically competent organisation in charge of each part of the problem and it makes accountability explicit. What it does not do — what no approval scheme does on its own — is give anybody the job of watching the seams.

The separation survives being brought inside one company. An airline holding all three approvals still runs three departments with three budgets and three sets of indicators. The boundary moves from a contract to an organisation chart, and goes on being a boundary.

The main boundaries, what crosses them, and how the crossing typically fails
Boundary What crosses it Usual artefact Classic failure
Flight crew to maintenance A defect described as a symptom Technical log entry, free text The symptom and the rectification do not describe the same thing
Maintenance to flight crew A deferred item with an operational procedure Deferred defect list, minimum equipment list The item is visible, the operational procedure behind it is not
Continuing airworthiness to maintenance A work package and a due date Work order, maintenance programme Findings outside the package have no defined route back
Maintenance to continuing airworthiness Work performed, findings, deviations Certificate of release to service, work pack return Records arrive complete and late, or fast and partial
Continuing airworthiness to operations Technical decisions with operational effects Engineering order, service bulletin, directive The decision never becomes an instruction anyone flies with
Operator to contracted line station Authority to act, and to spend Maintenance contract, station procedures The escalation path is discovered at two in the morning

Why the gap is structural, not a matter of goodwill

Safety management works by identifying hazards, assessing them and tracking the mitigations. Every step of that method assumes an owner. A hazard that lives entirely inside one organisation has one: somebody’s process produced it, somebody’s indicator will show it, somebody’s action plan will close it. A hazard that exists only in the handover has none. It is nobody’s process failing; it is two processes that were each correct and that do not quite meet.

Two mechanisms do most of the damage. The first is symmetric ignorance. The technician who defers an item does not see the flight the crew will fly with it, and rarely learns whether the operational procedure was workable. The crew that flies with the deferral does not see the constraint that produced it — the part on back order, the tooling at the wrong station, the interval that was already extended once. Each side reasons correctly from an incomplete picture, and neither has any cue that the picture is incomplete.

The second is the illusion of double checking. When two competent organisations both touch an item, each is entitled to assume the other looked. That assumption is usually right, which is what makes it dangerous: it is reinforced hundreds of times before the one occasion when both assumed and neither looked. The reliability of the interface is not the product of two reliabilities; it is whatever is left after both sides have quietly delegated the same check to the other.

The problem isn’t that people are careless. It’s that when two different teams handle the same issue, they might both think the other team is taking care of it. This creates gaps where mistakes can slip through, even if everyone is doing their job right.

An A320 with a deferred item (e.g., a faulty flap actuator, with flap extension limited to a reduced speed) is dispatched with a 7-day rectification interval. The maintenance team schedules the fix, but the flight crew isn’t properly briefed on operational restrictions. On day 6, the crew extends the flaps at the normal speed, above the reduced limit set by the deferral, risking structural damage. The maintenance team assumed the crew knew, and the crew assumed the deferral had no operational impact.

International guidance on safety management is explicit that interfaces between service providers must be identified and assessed as part of safety risk management, and that contracting out an activity does not contract out responsibility for it. Turning that principle into something an organisation can act on means naming the interfaces one at a time and giving each of them evidence.

The deferred defect: the whole problem in miniature

Nothing exposes the seams like a deferred item, because its life crosses every boundary at least twice. A crew reports a symptom. Certifying staff assess it and, if the minimum equipment list allows, defer it under a category with a rectification interval — the familiar A, B, C and D bands, whose durations come from the master document and which an operator may shorten but not lengthen at will. Two things then have to happen in parallel: an operational procedure has to reach the crews who will fly the aeroplane, and a maintenance procedure has to be performed or scheduled. Meanwhile the continuing airworthiness organisation counts the interval, plans the rectification, and needs the aeroplane, the part, the tooling and the licensed staff to be in the same place before the count runs out.

Where a deferred item can quietly go wrong, step by step
Step Who owns it How it drops
Symptom recorded Flight crew Described in terms that do not let the engineer reproduce it
Deferral decided and categorised Certifying staff Interval counted from the wrong day, or the wrong item selected
Operational procedure published Operator, operations documentation Exists in the manual, never appears in what the crew reads before flight
Item entered in the tracking system Continuing airworthiness Transcribed into a second system, updated in only one
Combination with other open items Nobody, by default Two deferrals the document never intended to coexist
Rectification planned Continuing airworthiness and network planning The aeroplane never routes to a station with the capability
Extension, if any Operator, with defined approval Granted informally because the alternative is a cancellation
Closure and record Maintenance organisation Closed technically, still open in the crew-facing list

None of those failures requires anyone to be incompetent. They require only that two systems hold the same item and that nobody is responsible for the fact that the two copies say different things.

The trap. A deferred item that is technically correct and operationally invisible is worse than one that is neither. The correct entry generates confidence in the record, the missing operational procedure means the crew is unaware of the constraint, and the two together produce an aeroplane flown outside the conditions under which the deferral was valid — with a paper trail that looks perfect.

Technical decisions that never become operational instructions

The second recurring pattern is a decision taken correctly on the engineering side that never crosses into the operational world. An airworthiness directive is embodied with an associated flight-manual limitation. A structural repair is approved with a restriction. A component is substituted with one of a different standard. A deactivation removes a capability the aeroplane is approved to use — a low-visibility category, an extended-range approval, a height-keeping capability, a particular autoflight function.

On the technical side the loop is closed: the directive is embodied, the record is updated, the review will find it compliant. On the operational side nothing has happened, because nothing in the technical process obliges anyone to ask the question. The crew plans an approach the aeroplane is no longer approved to fly, or a route it is no longer approved to take, and the discovery is made either by an alert crew or by an event report.

The remedy is procedural and cheap: every technical decision record carries a field that must be completed before closure, naming the operational consequence or stating explicitly that there is none, with a named recipient on the operations side and a closure that only that recipient can sign. It is not sophisticated. It works because it makes the absence of an answer visible, which is the whole trick with interfaces.

The contract scope is a safety document

Where maintenance is contracted, the maintenance contract is normally negotiated for scope, turnaround and price, reviewed by procurement and by the technical department, and read by the safety department never. Yet the risk sits precisely in the negative space of that document: what is not in scope, who supplies and certifies parts, who owns tooling calibration, who has authority to stop work when a finding falls outside the package, and — the question that decides how fast the answer comes at two in the morning — who pays for the extra work.

An escalation path that exists only in a contract clause is not an escalation path. It has to be tested, at least once, in conditions that resemble the ones in which it will be used: a night, an outstation, a finding outside the package, a departure slot in ninety minutes, and a duty engineer who has to reach somebody with the authority to spend money. Organisations that run that drill usually find that the named contact left the company, that the out-of-hours number rings in an empty office, or that the authority threshold is so low that every real decision escalates two levels.

The good habit. Before signature, have the safety function read the scope annex of every maintenance contract with one question in mind: for each activity, who reports what, to whom, and within what time? An interface with no reporting obligation written into it is an interface that will be discovered after an event, not before one.

Occurrence reporting that has to cross a boundary

Both the operator and the maintenance organisation are reporting entities in their own right, each with a mandatory reporting obligation and each with its own internal scheme. That produces two symmetrical failures. In the first, an event is reported by neither, because each assumed the other would: the crew saw a maintenance matter, the technician saw an operational one. In the second, it is reported twice into two systems that never link the reports, so two organisations each analyse half an event and both conclude it was isolated.

The cure is unglamorous — a shared reference for events that touch both organisations, and a written rule saying who reports and who is copied — but the harder half is feedback. A technician who reports a finding and never hears what came of it will report less next time, and there is no regulation that fixes that. The operator’s investigation conclusion has to travel back to the workshop that raised the report, in writing, with a closure date, and somebody has to measure whether it did.

Repetitive defects deserve their own agreement, because the definitions genuinely differ between organisations: how many occurrences, in how many days, on what level of the system, and whether a change of base or of registration resets the count. Two organisations using two definitions will each conclude that the threshold was not reached.

Control points that make the grey zone visible

The useful test of any interface control is not whether it sounds sensible but whether it produces evidence that can be examined later by somebody who was not there. The following list is deliberately short, because a long one is not implemented.

  • An interface register naming each boundary, its flow and an owner on both sides
  • One record of truth for deferred items, read by both sides
  • Mandatory operational-consequence field on technical decisions
  • Tested escalation and spending authority for contracted stations
  • Two-way occurrence feedback with a measured closure time
  • A single agreed definition of a repetitive defect
  • Joint safety review with both accountable managers
Control points and the evidence each one leaves behind
Control point Evidence it produces
An interface register naming each boundary, its flow and an owner on both sides A document in which every flow has two names and a review date
One record of truth for deferred items, read by both sides No second copy that can disagree with the first
Mandatory operational-consequence field on technical decisions A signed statement, including the explicit statement that there is none
Tested escalation and spending authority for contracted stations A dated drill report with the names that answered
Two-way occurrence feedback with a measured closure time A distribution of closure times, per organisation
A single agreed definition of a repetitive defect One counter both organisations accept
Joint safety review with both accountable managers Minutes that discuss hazards rather than supplier performance

Audit the item, not the organisation

Separate audits of three compliant organisations will produce three satisfactory reports and will not touch the subject of this article. The audit that does is the one that follows an object across the boundaries: take five deferred items closed in the last quarter and reconstruct each one end to end — the crew entry, the assessment, both systems’ records, the operational procedure exactly as it appeared to the crew before the next departure, the planning record, the rectification, the closure. Then compare the timestamps and the wording. Transcription errors and timing gaps are invisible from inside either organisation and obvious from across the boundary.

Two men in a hangar, one inspecting a component while the other uses a tablet, with an aircraft in the background.
Operational and maintenance teams must align to mitigate risks.

Do the same with one airworthiness directive that carried an operational limitation, and follow it forward until you find it, or fail to find it, in a document a crew actually reads. Three or four indicators sustain the exercise between audits: the proportion of deferred items closed before half their interval has run, the number of items whose operational procedure was not visible to the crew, the time from a crew report to an acknowledgement by the continuing airworthiness organisation, and the proportion of workshop reports that received a written answer. None of them blames anyone, and all of them go wrong before an event rather than after it. Practitioners on both sides of these boundaries compare notes in the maintenance and safety sections of the FL330 forums, which is often where the first symptom of a broken interface is described out loud.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

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