Where an Aircraft Programme Actually Loses Its Months

Engineers inspect a partially assembled aircraft in a dimly lit hangar at dusk, highlighting the behind-the-scenes work of an aviation programme.

When a manufacturer announces that entry into service moves by a year, the reason offered is almost always the thing that is visibly late at that moment: a test campaign running long, a supplier that missed a delivery, a certification finding. That explanation is usually true and almost always beside the point. The months were lost earlier, in a place with no press release attached to it, and they became visible only when the schedule buffer that had been absorbing them ran out. Reading a programme means knowing where those months actually accumulate, and which of them can ever be recovered.

A slip is a buffer running out, not a task going wrong

Every industrial programme is built with float. Some of it is explicit and sits in the schedule as margin; much of it is implicit, hidden in optimistic durations, in the assumption that two activities will overlap cleanly, in a supplier’s quoted lead time that already contains its own cushion. Work that runs long consumes this float silently. Nothing is announced, because nothing at the top level has moved.

The announcement comes when the last of the float on the critical path is gone. By then the originating problem may be a year old and may already be solved. This is why the publicly named cause of a delay so often looks disproportionate: a two-month issue is blamed for a twelve-month slip because it was the one that arrived after the cushion was spent. The task that broke the schedule is rarely the task that emptied it.

There is a second reason the announced cause misleads. A programme schedule is also a commercial document — it underpins delivery contracts, penalty clauses, customer fleet planning and financing. There is a strong incentive to hold the published date until it is unarguable. The interval between internal knowledge and public revision is not deception so much as an artefact of how the commitment is structured, and it is why suppliers and airlines frequently know before the market does.

Delays often come from small problems adding up over time, not from one big disaster. The schedule has hidden buffers that absorb these small delays until they run out. When the buffers are gone, that’s when we hear about the delay – but the real problems might have started months earlier.

The Airbus A380 programme famously slipped by two years. The public explanation focused on wiring issues during final assembly, but the real problems started much earlier with design changes and coordination between different design teams. The wiring problems were just the final visible symptom of months of accumulated delays.

The visible milestones — rollout, first flight, certification, first delivery — are ceremonies attached to states. Each has a technical predecessor that decides whether it can happen, and that predecessor is where the schedule is really made or lost.

Visible milestone, real gate, and the quiet predecessor
Milestone What actually gates it Where the time is lost before it
Design freeze Configuration stable enough to release drawings and order long-lead items Requirements churn driven by launch customers and by weight or performance shortfalls
First metal cut Released engineering and available tooling Engineering release running behind plan while assembly dates stay fixed
Rollout A structurally complete airframe, whatever is fitted inside it Unfinished work moved out of station and into final assembly
First flight Systems integrated and functioning, a flight-worthy configuration, permit to fly Integration bench and iron bird slipping, software builds arriving late
Type certificate Compliance demonstrated and documented for the whole certification basis Ground test articles, and findings that require a design change and a re-test
First delivery Production aircraft conforming, and the operational package complete Training devices, maintenance and crew data, spares, customer acceptance

Notice how few of these gates are about flying. The aeroplane in the air is the most photographed part of a development programme and among the least likely to be its constraint.

Configuration churn is the most expensive month there is

A month lost while the configuration is moving is not one month. Every change after release propagates: the drawing, the tool, the supplier’s part, the qualification test that used the old standard, the analysis that supported a compliance argument, the manual that described it. The industry measures the health of this phase through the engineering release curve, and a curve that lags its plan early is the single most reliable predictor of trouble later — more reliable than anything visible from outside.

Concurrency makes it worse in a specific way. Compressing a schedule means starting production engineering, tooling and supplier work before the design is fully stable, on the reasonable assumption that most of it will not change. What concurrency actually does is convert schedule risk into rework risk: when a change does land, it lands on work already done. The programmes that suffer most are not those that were slow to freeze, but those that froze on paper while continuing to change in practice.

The trap. Reading a rollout as evidence of maturity. An airframe can be rolled out structurally complete and systems-hollow, with thousands of open jobs travelling with it. The meaningful number is not whether the aircraft left the hall, but how much work left the hall unfinished — and that number is internal. Two widely documented wide-body programmes of the past two decades each slipped by years for exactly this reason: one because digital definitions between design sites were not fully compatible, turning wiring into a physical fit problem discovered at assembly; the other because a distributed build with risk-sharing partners moved unfinished structure into final assembly, where it is far more expensive to complete.

Supplier qualification runs in series and cannot be parallelised

A modern airframer integrates more than it builds. That is a deliberate risk transfer, and it works in production. In development it introduces a specific fragility: the qualification of an item of equipment is a serial chain — design, qualification testing to the environmental standards, software and hardware development assurance evidence for anything airborne and critical, first article inspection, then integration on the aircraft. None of those steps can be shortened by adding people, and all of them must be repeated if the requirement changes.

This is why a single-source item on the critical path is so dangerous. The instinctive remedy, qualifying a second source, takes about as long as the first qualification did, so it protects the series but not the programme in front of it. Long-lead items behave the same way: large forgings, castings, landing gear structure and engines have order-to-delivery times measured in many months, and they are ordered against a configuration that must therefore be frozen long before anyone would like to freeze it.

Tooling and rate: a month here is paid across the whole series

The distinction that matters most when judging a delay is between a month lost on something that happens once and a month lost on something that constrains every aircraft afterwards. Test time is largely the first category. Tooling is the second.

Assembly jigs, automated drilling and fastening systems and the layout of the final assembly line set the achievable rate. If they arrive late, or arrive and require rework, the effect is not a one-off delay to the first aircraft: it caps deliveries for as long as the constraint exists, and the backlog it creates has to be worked off at a rate the line does not yet have. Add the learning curve — the first units take a multiple of the mature hours — and a tooling problem discovered during early production quietly reprices dozens of aircraft.

Travelled work is the mechanism through which this becomes visible. When a station cannot finish its scope, the job moves downstream to a position where access is worse and the labour cost per job is higher. A line running with heavy travelled work looks superficially on schedule, because aircraft keep moving; it is in fact accumulating a debt that shows up as an inability to raise rate.

Certification is limited by evidence, not by flying hours

Flight testing is the most visible certification activity and the easiest to misread. Hours accumulate on a schedule that is largely plannable; what is not plannable is what those hours find. A finding that requires only a documentation change costs days. A finding that requires a design change costs the change, the re-qualification of anything affected, a modification to the test aircraft, and the repetition of the tests already flown against the old standard. Programme schedules are broken by the second kind, and the number of them is not knowable in advance.

Two ground activities gate the type certificate at least as firmly as flying does. The static test article must complete its structural programme, including ultimate load, before the structure can be shown compliant; and the full-scale fatigue test has to stay ahead of the fleet-leading aircraft, which means it must be running and demonstrably ahead well before service entry. Both compete for the same manufactured airframes as the flight test fleet, and both are affected by exactly the same production problems as delivery aircraft.

Behind all of it sits the compliance record itself. The certification basis is agreed early, with special conditions where the design outruns the existing requirements; from there each requirement needs an agreed means of compliance and a document that closes it. Authorities work through that record at a rate the applicant does not control, and a programme can be finished flying while months of closure work remain.

The right reflex. When judging how far a programme has to go, ask about compliance closure rather than flight test hours. The percentage of the certification basis actually closed, and the number of open findings requiring design change, describe the remaining distance far better than hours flown — which is why hours flown is the figure most often published.

Entry into service has a critical chain of its own

A type certificate approves the design. It does not by itself let the manufacturer build the aircraft, which needs a production approval, nor an airline operate it, which needs a certificate of airworthiness and an operational package of its own. The gap between the type certificate and the first revenue flight contains its own dependencies, most of them derived from data that only exists once flight testing is well advanced.

Crews have to be trained, which requires qualified full flight simulators, which require a data package generated by flight test and a device qualification process of its own. Maintenance programmes derive from the manufacturer’s planning data through the industry analysis process and its review board, and the first operators need that output to build approved programmes. Operational suitability data — flight crew data, the master minimum equipment list, maintenance certifying staff data — is approved alongside the type certificate and is what the training organisations and operators build on. Extended operations approval, where it applies, has its own evidence requirements. Spares provisioning and field support have to be in place before the first revenue flight, not after it.

These items are frequently underestimated because they are not engineering, and because they run in parallel with everything else until they do not. A flight test slip that does not move certification can still move service entry, simply because the simulator data package arrived too late for the launch operator’s training footprint.

What compounds, and what can still be recovered

Not every lost month behaves the same way, and the difference decides whether an announced recovery plan is credible. Time lost on work that happens once can often be bought back with resource; time lost on work that every subsequent aircraft depends on cannot.

Where a lost month stays a month, and where it multiplies
Where the month is lost Behaviour Practical consequence
Flight test campaign Often recoverable Add aircraft, crews and shifts; the constraint is aircraft availability and findings, not calendar
Compliance documentation Partly recoverable Can be resourced, but authority review rate is outside the applicant’s control
Design change after release Compounds Propagates into tooling, suppliers, qualification and previously closed compliance
Supplier qualification Serial, poorly recoverable Cannot be shortened by adding resource; a second source protects the series, not the first aircraft
Assembly tooling and rate capability Compounds across the series Caps deliveries for years and makes the backlog harder to clear
Service entry package Recoverable only if started early Simulators, maintenance and crew data and spares have their own lead times

Reading a programme from the outside

Airlines, lessors, suppliers and candidates all have reasons to form a view on whether a published date will hold, and none of them will see the engineering release curve. A few observable signals carry real information, provided they are read for what they actually indicate.

A group of engineers in high-visibility vests examine a digital tablet displaying an aircraft schematic, with a partially assembled fuselage in the background.
Engineers review digital schematics to optimize aircraft assembly timelines.
Outside signals and what they are worth
What can be observed What it suggests What it does not prove
Number of test aircraft actually flying, and how often Production of conforming airframes is working Nothing about compliance closure or findings
Long-lead orders placed with suppliers Configuration considered stable enough to commit That it will not change afterwards
Recruitment for final assembly and field support The industrial ramp is being resourced That tooling and rate capability exist
Simulator orders and training centre build-out Service entry package is being taken seriously That the data package is available
Launch operator adjusting its fleet plan Someone with contractual visibility has re-planned The size of the slip
A milestone met exactly on the announced date Schedule discipline, or a milestone redefined to be achievable That the work behind it was complete

The last row is the one that repays attention. Milestones are defined by the programme, and a definition can be adjusted more easily than an aircraft can be finished. A rollout with a hollow interior, a first flight in a configuration far from the certified one, a delivery accepted with a long list of open items: all can be reported as achieved on time. The questions that separate a healthy programme from a late one are unglamorous — how much work is travelling, how much of the certification basis is closed, how many changes are still landing on released engineering — and they are almost never the questions asked at the ceremony. Engineering and operations staff tracking these programmes for their own planning exchange that kind of reading on the FL330 forums.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

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