Spares under strain: what the supplier queue costs the operator

Comptoir de magasin de rechanges de nuit, rayonnages vides et un seul colis scellé posé dessus

The supplier quotes twenty weeks on a bleed valve. Stores have none left, neither does the regional pool, and the aircraft leaves tomorrow morning. The question is not whether it will fly: it will fly. The question is under which tolerance, for how many days, and what that delay will have consumed by the time the part finally arrives. A replenishment lead time is settled less in euros than in operational margin, and that currency appears on no invoice.

A lead time is not a duration, it is a place in a queue

When a manufacturer quotes twenty weeks, it almost never describes the time needed to make the part. Machining a valve body is counted in hours, a surface treatment in days. What is counted in months is the wait in front of shared workstations: the foundry that also pours for other programmes, the heat-treatment furnace, the single test bench that qualifies a whole family of equipment, the electronic component line whose production window opens only a few times a year.

Hence a property worth taking on board once and for all: waiting time does not grow in proportion to load. As long as a workstation runs at half its capacity, an incoming order finds the slot free. Beyond a certain utilisation rate, every additional point of load costs not a few days but weeks, because the queue no longer has time to empty between two arrivals. It is the mechanism that makes a single runway absorb moderate traffic effortlessly, then slide into a holding sequence for just a few more movements.

Wait in front of a production workstation according to its utilisation rateThe curve stays low and almost flat as long as the workstation runs well below its capacity, then it rises almost vertically as saturation approaches. A vertical marker indicates a high load level, already located on the rising part.high loadwait before processingworkstation utilisation ratecomfort zone
A shop close to saturation does not wait twice as long as a shop loaded to half, it waits weeks longer. The curve is flat for a long time then vertical: the lead times of a whole supply chain therefore degrade at the same moment, and a quoted lead time slips with every follow-up, recalculated from a longer queue.

When a factory is very busy, new orders don’t just take a little longer – they can take weeks longer. It’s like a runway: a few extra planes don’t cause delays, but just a few more can create long holding patterns.

An A320 bleed valve with a 20-week quoted lead time might actually take 30 weeks when production is at 90% capacity. If an airline needs 5 valves, the last one might arrive 6 months after the first, not 10 weeks later.

The same mechanism governs repair, the true way of life of a rotable part. A removed unit does not go away “for repair”: it goes to join a queue, in a shop that also receives the inductions of ten other operators. Breaking down a complete turnaround — shipping, waiting, work, testing, documentary release — shows that the technical part occupies only a fraction of the calendar.

Breakdown of the turnaround time of a repairable unitA horizontal bar cut into four segments: transport, queue at the shop, the repair itself, then testing and documentary release. The waiting segment, hatched, takes up more than half of the bar on its own.transportshop queuerepairtest and releaseduration of an equipment turnaround
What takes longest in a repair is not the repair. The proportions depict a common case, not a measurement: they are a reminder that negotiating the work time without negotiating the position in the queue moves almost nothing.

Three currencies: the aircraft, the money, the tolerance

A delay is always paid for, but rarely where it is being watched. The first currency is the aircraft: grounding, AOG, with the rotation cancelled, passengers reaccommodated and the crew up against its duty limits. The most visible, the most expensive, and the one the operator almost always avoids.

The second is money, and it is acceptable: a short lease, a standard exchange with an urgency surcharge, dedicated transport, labour outside the planned slots. These costs are visible, can be budgeted and can be compared.

The third appears nowhere: dispatch tolerance. Deferring a defect means consuming a reserve defined by the design and validated at certification. It is finite, it is not renewable at will, and consuming it triggers no accounting alert. That is precisely why it serves as the adjustment variable as soon as the supply chain tightens.

When parts are delayed, airlines have three options: cancel the flight (most expensive), pay for urgent solutions (visible cost), or fly with temporary fixes (hidden risk). The last option is often chosen because it’s invisible in the budget.

The minimum equipment list is not a buffer stock

Deferring a defect under the minimum equipment list is not a waiver: it is an approved degraded regime, bounded in time, almost always subject to conditions. The rectification intervals taken from the manufacturer master list are counted in consecutive calendar days, the day of discovery not being counted. The durations recalled below are those of the usual reference convention: the operator approved list may be more restrictive than the manufacturer list, and it alone applies to the flight of the day.

What each deferral category actually grants, at the usual reference durations
Category Rectification interval What it assumes in practice
A Set case by case in the remarks of the item Often a number of flights or cycles; to be read item by item, never out of habit
B Three consecutive days The pace of a round trip to a maintenance base; incompatible with any order for a part not held in stock
C Ten consecutive days The time of a short logistics turnaround; assumes an existing part, already released
D One hundred and twenty consecutive days The only category compatible with a manufacturing lead time; reserved for items whose absence weighs little on safety

The design reasoning behind these limits governs the way they are used. The master list assesses each item in isolation, assuming the other functions are available; it does not claim to cover the accumulation of open items or their interactions. Three innocuous deferrals taken separately can, together, remove the redundancy of a single function or degrade a take-off performance on a short and wet runway.

Rectification intervals compared with a supply lead timeOn the same scale of days, category A is depicted by an empty dashed frame, because it has no fixed length. Categories B and C are two very short marks, category D a long bar. An arrow depicting the supplier lead time extends beyond the category D bar and leaves the frame on the right.flight time authorised with the defectcat. A — case by casecat. B — 3 dayscat. C — 10 dayscat. D — 120 dayssupplier lead time0120 days
On the same time scale, categories B and C are marks, not margins. Only category D has the order of magnitude of a manufacturing run: on a part not held in stock, a B or C item serves only to bring the aircraft home.

The trap. Treating the extension of an interval as an acquired reserve. A single extension, when the approved procedure of the operator provides for it and the authority has accepted it, remains a reasoned management act: it assumes a documented cause of delay, associated conditions still satisfied and a reassessed accumulation. A technical department that builds its supply plan by counting in advance on the extension has turned a safety valve into nominal capacity, and will have nothing left to open the day the delay really slips.

Buy, exchange, lease, repair, cannibalise: the real scale

Faced with a shortage, five routes exist, and they are distinguished first not by their price but by what they demand of the organisation. The choice is generally made within a few hours, often without anyone having the complete picture in front of them.

Five routes to an unavailable part, and what each one demands
Route What you obtain What you pay What it requires
Outright purchase A new part, owned by the operator The list price and the full lead time of the production queue Cash tied up, a credible consumption forecast
Standard exchange A serviceable unit immediately, against the removed part A flat fee, surcharged if the returned unit falls outside the repairable scope Returning the core within the deadline, otherwise billing switches to a purchase
Lease A temporary right of use A daily rental for as long as the order does not arrive Daily follow-up: a lease forgotten after the order has been received ends up exceeding the purchase price
Repair of the removed part The same unit, restored and released The full turnaround time, queue included A replacement during its absence, hence one of the routes above
Cannibalisation from another aircraft Immediate availability with no financial flow Two removals, two refittings, a second aircraft made dependent An approved procedure, complete traceability, a dated return plan

Cannibalising a grounded aircraft is the most tempting reflex and the most poorly assessed. It creates no part: it moves the problem, doubles the work and adds the risk of collateral damage inherent in any unplanned removal. Its real limit lies elsewhere: it masks the shortage instead of escalating it, and the supply department never sees the signal.

The right reflex. Open the repair order in the same movement as the part request, and ask not for the contractual turnaround time but for the one observed over the last twelve months for that part number, at that shop. The gap between the two is the only useful measure of the queue. The first figure appears in the contract, the second has to be asked for — and a supplier who refuses to disclose it has just told you something.

A stock is not sized on an average

Demand for parts is a rare and irregular phenomenon, not a steady flow. A part number removed on average three times a year is not removed every four months: it is removed twice in the same week, then nothing for more than a year. Sizing the stock on the average guarantees the stockout; what protects is covering the variability throughout the whole replenishment lead time.

Two technicians inspect and repair an aircraft engine in a hangar.
Delays in spare parts can halt critical maintenance work.
Stock level, reorder point and the effect of accelerated consumptionThe stock decreases in a sawtooth pattern; it triggers an order by crossing the reorder point, keeps falling during the delivery lead time then rises again on receipt. A second trajectory, dashed, corresponds to faster consumption: it reaches zero before the order arrives.delivery lead timereorder pointsafety stockstockoutstock on handtime
Safety stock does not cover average consumption: it covers the deviation during the lead time. Lengthening the supplier lead time raises the reorder point; with unchanged stock, a variability of consumption that until then had no consequence produces a stockout.

Two effects make the calculation worse. The first is the gap between mean time between failures and mean time between unscheduled removals: it is removals that consume the stock, and they are more frequent than actual failures, because a poorly isolated intermittent defect goes through two or three units before being understood. No-fault-found returns are not a shop problem: they are healthy units spending weeks in a queue instead of being on the shelf. The second effect is the price of the service level: bringing it close to perfection costs far more, for the same geometric reason as the queue.

Decision matrix: criticality of the item against replenishment lead timeA four-cell matrix. Top left, critical item and short lead time: local stock at the outstation. Top right, critical item and long lead time: pool or exchange contract. Bottom left, item of low criticality and short lead time: replenishment on consumption. Bottom right, item of low criticality and long lead time: anticipated order on a fixed date.local stockat the outstationpool or standardexchange contractreplenishmenton consumptionadvance orderon a fixed datecriticalityshortreplenishment lead timelong
It is not the price of the item that governs the stock policy, it is the criticality-lead time pair. A cheap part that is nonetheless blocking and long to obtain belongs in the upper right cell: it is a matter for pooling, not for the shelf.

Airlines need to keep spare parts based on how critical they are and how long they take to get, not just their price. A cheap but essential part with a long delivery time should be shared between airlines, not kept on every shelf.

A regional airline with 20 aircraft might share 10 critical parts with other airlines, reducing their total stock from 20 to 12. If one airline needs a part, another might have it available, saving everyone from keeping extra spares.

This is where pooling takes on its real meaning. A pool shortens no manufacturing lead time and does not reduce average consumption: it mutualises variability. Ten operators sharing a common batch need fewer units than ten separate stocks for the same service level, because their peaks do not fall on the same day. What is bought is therefore not a part but a probability of availability — and the clauses that count are those that set the location of the forward stocks, the commitment lead time and the behaviour of the contract when the supplier itself is out of stock.

The shape of the network governs the answer more than the size of the fleet. An operator whose aircraft come home every evening to a single base concentrates its stock at one point and can absorb internal delays: the part stays a few hours from the aircraft, and the store sees every removal go past, and therefore learns. A network that lets its aircraft sleep at outstations must arbitrate between forward-positioning stock at rarely served sites, relying on a mutual assistance agreement with another operator of the same type, and accepting that a failure far from base ends in an empty ferry flight. Business aviation shifts the constraint once more: the mission is often born a few hours before departure, there is no planned rotation on which to line up a repair, and the only variable left is the depth of the parts access contract subscribed to long beforehand.

The paperwork is part of the part

A part without its authorised release certificate is not a part: it is an object. This obvious fact regularly turns hours gained into days lost. The certificate must be consistent with the intended field of use, traceability must go back to an acceptable origin, and dual release is in no way automatic when the aircraft is registered in a third country.

Two situations recur. The first is the part obtained in a hurry from an intermediary, whose documentation arrives incomplete or carries a chain of ownership that cannot be reconstituted: it will not be fitted, and the aircraft will stay on the ground with the part on the counter. The second is the part manufactured under a design approval distinct from that of the manufacturer: legitimate within its own framework, it assumes that acceptability for the type and the registration concerned has been verified beforehand. That verification takes weeks and is carried out before the shortage, never during.

Component obsolescence is the grey area on the rise. A computer stays in service for decades, its electronic components far less. When the production line of the component closes, the manufacturer must requalify an alternative, which reopens a design file — a lead time no longer measured in weeks but in years. The operator has no grip on that lead time; it does have, on the other hand, a means of being warned of it, by following the discontinuance notices of the manufacturer, published well before the part number becomes impossible to find.

What to go and check in your own documentation

This subject is not handled by one global decision but by a handful of specific checks, each of which takes half a day. Four are worth carrying out before the next period of strain.

  1. The list of long lead time items specific to the type operated, and not that of the manufacturer: it rarely exists in that form and is reconstituted from the orders of the last two years.
  2. The turnaround time observed by part number and by shop, compared with the contractual time, together with the exclusion cases that allow the commitment to be suspended.
  3. The consistency between the minimum list of the operator and the master list in its current revision: an operator list that has not followed a restrictive revision leaves open deferrals that are no longer open.
  4. The actual distribution of the associated operational and maintenance procedures. A deferral is only valid if its procedure is in the hands of the crew and of maintenance, not only in the master document.

For anyone preparing a type rating or a first job in a strained sector, the stake is simpler: understanding why the same item has been open in the technical log for three weeks, knowing how to read its interval and its associated condition, and asking the question that is most often missing in the crew room — what is the expiry date of this deferral, and what has been planned if the part does not arrive. The professional preparation programmes deal with this mechanism from the operational side; the supply side remains outside the syllabus, even though it governs half of the decisions of the morning.

The applicable regulation, the minimum list approved for the operator and the maintenance documentation in force prevail over this article, whose only purpose is to explain a mechanism.

Questions fréquentes

Can a rectification interval be extended when the part does not arrive?

A single extension of equivalent duration is possible for certain categories, provided the operator procedure provides for it and it has been accepted by the authority. It is not a formality: the cause of the delay must be documented, the conditions associated with the deferral must still be satisfied and the accumulation with the other open items reassessed. An extension used by default, for want of a supply plan, turns a safety valve into nominal capacity.

Why does the lead time quoted by the supplier get longer with every follow-up?

Because the figure quoted is not a delivery date but an estimate of waiting time in front of a shared workstation. That waiting time grows in a highly non-linear way with the load of the shop: beyond a certain utilisation rate, a few additional orders add weeks. With each follow-up, the estimate is recalculated from a queue that has become longer.

Should standard exchange or a lease be preferred in the event of a shortage?

Standard exchange is suitable when the removed part is repairable and can be returned within the contractual deadline: beyond that, billing often switches to an outright purchase. A lease is suitable for filling a wait whose end is known, provided the rental is monitored daily. A lease forgotten after the ordered part has arrived regularly ends up costing more than a purchase.

Is cannibalising a part from another aircraft a good solution?

It is an immediate but costly and masking solution. It imposes two removals and two refittings, complete traceability of both movements, and exposes to the collateral damage inherent in any unplanned work. Above all, it makes the shortage signal disappear: the supply department never sees the strain, and the missing part is not ordered any earlier.

What exactly is a parts pooling contract for?

It shortens no manufacturing lead time and does not reduce average consumption: it mutualises the variability of demand between several operators, whose peaks do not fall on the same day. What is bought is a probability of availability. The decisive clauses are the location of the forward stocks, the commitment lead time and the behaviour of the contract when the supplier itself is out of stock.

Why can an available part remain unusable?

Because the authorised release certificate is part of the part. A certificate inconsistent with the intended field of use, a chain of traceability that cannot be reconstituted, or a design approval whose acceptability has not been verified for the type and the registration concerned are enough to block installation. That verification takes weeks and is carried out before the shortage, never during.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

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