How to build a network so the aircraft flies more

What utilisation is made of

Let us start with a day. An aircraft can fly while the airports it lands at and departs from are open. Many large European airports have night bans or quotas, and there a base's operating window runs roughly from six in the morning to midnight, about 18 hours. But this is not a law of nature: many airports operate around the clock — Istanbul, Doha ¹⁰ and other big hubs in the Gulf and Asia — and flying goes on all night there: a night flight arrives for the morning connecting bank, late arrivals and early departures connect overnight. The head of Qatar Airways has explicitly called a 24-hour airport an advantage over European competitors¹⁰. Even there, the whole day almost never goes into flying; it is just that the limit is no longer the airport but three other things. Demand: there are few passengers at night, and a night flight sells for less, except on long-haul and leisure routes. Crews: a night duty is shorter than a day one — under EASA rules, a duty starting late in the evening or at night is limited to 11 hours, against 13 by day¹¹. And maintenance: daily checks and the A-check are done on the ground, usually overnight at base — more on this below. So the 18-hour window in the calculations below is the typical case for a base with night restrictions; a base without them has a wider window, 20–22 hours, but pays for it with night duties and with having to find other time for maintenance. Within the window the aircraft does not only fly: after each landing it has to be deplaned, cleaned, refuelled and boarded again — the turnaround, from half an hour to an hour for narrowbodies. On top of that, a based aircraft has to be home by night, so only a whole number of out-and-back pairs fits into the window, and the part of the window that a pair does not fit into is lost. That is how 24 hours become a peak day — what the aircraft flies on a good summer day without disruptions.

Average annual utilisation is always below the peak day. In winter demand falls, and some aircraft are parked or sent for heavy maintenance; a few are kept in reserve for breakdowns; delays eat into schedule buffers. How much is lost depends on the airline; the chart below uses a factor of 0.8 — our calibration for illustration, not a quote from a source.

Schematic: window 06:00–24:00, 2.5-block-hour sector, 25-minute turnaround — six flights by the formula from the network section; the annual factor of 0.8 is an assumption.

When talking about utilisation it is easy to compare apples and oranges, so three clarifications. First, block hours and flight hours. Block time runs from chocks off before departure to chocks on after landing, including taxi; flight time counts only the time in the air. Taxi out and in takes about 20 minutes per flight¹², so on short sectors there are 10–15 % fewer flight hours than block hours. Ryanair reports in flight hours; most carriers and the calculator use block hours.

Second, flying aircraft versus the whole fleet. In financial year 2025, a Wizz Air aircraft that was flying logged 12 hours 28 minutes a day, while the average over the whole fleet came to 9 hours 51 minutes: on average 44 aircraft were grounded by engine defects¹³. What matters for the cost of a flight is utilisation across the whole fleet: the lease is paid for a grounded aircraft just the same.

Third, peak month versus the year. In summer an airline gets the most out of an aircraft, in winter it flies less: easyJet in the UK flew 11.9 hours a day in September 2024 and 8.9 in January^14,15. The annual lease is divided by the annual average, not by the best month.

How much aircraft really fly

The chart shows average annual utilisation from airline reports for 2023–2025, in block hours per aircraft per day (flight hours for Ryanair). The dot is the annual average; the line, where present, runs from the worst month to the best or from the whole fleet to flying aircraft. Orange marks figures where the average was estimated from monthly data, taken from a presentation, or where the reading of the report is not yet confirmed.

The most telling comparison is within one group. Aeroflot publishes detailed reporting for its three airlines: flight hours for the year and the year-end fleet¹. Divide the hours by the average fleet and by the number of days, and you get utilisation per aircraft per day:

Flight hours from Aeroflot's Databook ÷ average fleet (mean of start and end of year) ÷ days in the year ¹. The fleet is every aircraft on the balance sheet, including grounded ones; so this is utilisation across the whole fleet, not per flying aircraft.

Pobeda — a low-cost carrier with a single type, the Boeing 737-800, and a dense shuttle from its bases — flies 12.7 hours a day, on a par with the world's best low-cost carriers. Aeroflot, with its Sheremetyevo hub, connecting banks and long-haul flights, flies 9.8, and this figure grew quickly after 2022, when the fleet shrank and traffic recovered. Rossiya flew less than five hours per aircraft per day in 2025. The report does not say why: the fleet on the balance sheet includes every aircraft, flying and grounded, and utilisation by type is not published. But a two-and-a-half-fold gap between airlines of one group with shared procurement, engineering and rules shows the main point well: utilisation is set not by the brand or the country but by how the network, the fleet and maintenance are organised.

The picture is the same abroad. Low-cost does not mean more flying: Allegiant flies to resorts a few times a week and logs 6.3 hours a day, Spirit with its dense network 11.1^16,22. Network carrier does not mean less: Copa with its Panama hub flies about 11 hours²¹, on a par with easyJet and Norwegian ^19,20. Widebodies on long-haul routes fly the most — 14–15 hours a day at British Airways and Virgin Atlantic in 2024¹⁸: a long flight wastes almost no time on turnarounds.

And a separate word on talk of “16 hours a day or more”. Such utilisation does happen — for a single aircraft on a summer day, in a peak month, in a plan for a new type. No public report shows it as an annual average across a whole fleet: even the best low-cost carriers hold 12–13 hours, and the summer month of a single subsidiary 13 ^3,18. A peak day and an average year are different quantities, and for costs it is the second that matters.

How to build a network that flies more

A base is the first morning departure

A low-cost network is built around bases: the aircraft spends the night at base, leaves with the first wave in the morning, flies several sectors during the day and returns home in the evening. In Dublin Ryanair's first wave departs between 06:00 and 07:00⁶. The value of a base lies in that morning slot: if the aircraft cannot leave early, it loses a whole flight, and the economics of a based aircraft do not add up³⁰. So the first thing checked when choosing a base is the airport's night regime — not only at the base but at the destinations too: the last flight of the day must land at home before the curfew.

The wider a base's window, the more pairs fit into it. Airports with a ban from 23:00 to 05:00–06:00 cut the window at both ends, while privileges for based carriers — as in Düsseldorf, where delayed based aircraft may land until midnight⁷ — lengthen it. A 24-hour base allows a late arrival and an early departure — the window then stretches to 20–22 hours, and the limits become crew night duties and the time the aircraft needs on the ground for maintenance.

Sector length — for a whole number of pairs

A based aircraft flies a whole number of out-and-back pairs. How many fit into the window is calculated as: flights = ⌊(window + turnaround) ÷ (block time + turnaround)⌋, and for a shuttle rounded down to an even number, so that the last flight brings the aircraft home. Because of the rounding the relationship is a sawtooth^4,5,6,7,8,9: as the sector grows, utilisation grows with it, but as soon as the last pair no longer fits into the window, utilisation drops by a step. An extra 15 minutes on the sector can cost a whole pair of flights.

Horizontal axis — sector block time in minutes; vertical — aircraft block hours on a peak day with an 18-hour window (06:00–24:00), 19 and 21 hours (a base without a night ban). Block time already includes taxi, so only the turnaround is added between flights. This is a ceiling without delays, maintenance or crew limits, not a forecast.

Hence the planning rule: sector length and turnaround are chosen so that the pairs fill the window with nothing left over. With an 18-hour window and 25-minute turnarounds, sectors of about 1.5, 2–2.5 and 4 hours fit well; a 3-hour sector is worse than a 2-hour one per day. If a destination has an awkward length, the rest of the window is filled with other patterns — see below.

A short turnaround is the main lever on short sectors

The shorter the sector, the more turnarounds per day and the more utilisation depends on their length. According to EUROCONTROL data for 2015–2019, the average scheduled turnaround of a narrowbody in Europe is 48–49 minutes, the actual 53–55; for widebodies 80–92 and 86–97 minutes⁴. We found no more recent public summary. Low-cost carriers shorten the turnaround by boarding without jet bridges, a single class, no catering and cleaning by the cabin crew; the chart shows that on a one-hour sector the difference between 25 and 45 minutes is a pair of flights a day.

The crew decides how many duties the aircraft needs

The diagram below shows an aircraft's peak day for two sector lengths and the crew duties covering it. With one-and-a-half-hour sectors the aircraft flies eight flights, covered by two crews — morning and evening — of four sectors each. With 2.5-hour sectors there are six flights and more utilisation, but four sectors no longer fit into an allowed duty: under EASA rules, a duty starting in the morning is limited to 12 hours with four sectors ¹¹, while four 2.5-hour flights with three turnarounds and reporting an hour before departure take more than 12. So three crews a day are needed, each flying one pair. Aircraft utilisation goes up while hours per crew go down: this matters for the calculation, and we come back to it in the crew section.

Schematic: window from 06:00, 25-minute turnaround, crew reports an hour before the first departure (assumption). The duty runs from report to the landing of the last flight.

Rotations: shuttle, W and hub banks

The simplest pattern is the base shuttle: out-and-back pairs to the same or different points, night at home. The crew returns to base every evening, no hotels are needed. Its drawback is that the rest of the window that a pair does not fit into is lost. Charter carriers flying tourists to resorts build their economics on two full rotations a day, and losing an early or late slot breaks the pattern by a step³⁰.

The rest of the window is filled with W rotations: the aircraft flies from base to A, comes back, flies to B, from there to C and home — or spends the night away from base. This makes use of awkwardly long sectors, but costs a crew hotel and an overnight stand away from home.

A network carrier works differently: all flights go through a hub — a connecting airport where passengers change planes — and aircraft fly back and forth between the hub and the cities of the network (the hub-and-spoke model). To keep connections short, flights are grouped into banks: aircraft from different cities arrive at the hub within one window, passengers transfer, and the aircraft leave almost at once. SWISS in Zurich has six banks between 06:10 and 22:45 with a minimum connecting time of 40 minutes³¹. Convenient connections are paid for with utilisation: between banks the aircraft waits. American Airlines, abandoning rigid banks at some hubs in the early 2000s, raised utilisation at the cost of longer connections — six minutes on average³².

Long-haul flights are scheduled around time zones: a night flight arrives in the morning for the hub's first bank, and the aircraft flies back during the day. Narrowbodies also use the night window where airports are open: Southwest has more than 50 overnight flights on a peak day³³.

Crews: what really limits

A crew's daily duty rarely limits aircraft utilisation: the aircraft is covered by several duties, as in the diagram above. EASA rules allow up to 13 hours for a duty starting between 06:00 and 13:29 with one or two sectors, and half an hour less for each additional one — 12 hours with four¹¹. The low-cost “two out-and-backs” pattern on sectors of up to two hours fits within this with room to spare; morning and evening crews together give the aircraft 14–16 hours^11,34.

How many crews an aircraft needs is decided by monthly and annual limits. EASA caps a pilot's flight time at 100 hours in 28 days and 900 hours in a calendar year¹¹; in Russia, from 1 March 2026 under Ministry of Transport Order No. 381, it is 80 hours a month and 800 a year, and with the pilot's consent up to 90 and 900 ^35,36. Rest between duties is at least twice the duty, at base at least 12 hours, weekly rest 42 hours ³⁶. So crews per aircraft = aircraft annual block hours ÷ a pilot's annual flight hours.

Without reserve for sickness, leave and training. 900 hours is the EASA ceiling; in practice a pilot flies less.

Real figures are well below the ceiling. At IndiGo in financial year 2025 — about 567 block hours per pilot and 6.3 crews per aircraft^37,38. In the Aeroflot group in 2025 there were 4,319 pilots for 352 aircraft — about 12 per aircraft; with two pilots on each flight, that is about 490 flight hours per pilot a year¹. This is an upper-bound estimate from pilot numbers: long-haul cockpits carry more than two people, and the flight crew headcount includes reserves, trainees and people on leave. The gap between 490 and 900 hours is a buffer that also has to be paid for.

Hence the main conclusion for the calculation: higher aircraft utilisation makes an hour cheaper only on ownership. The monthly lease is the same at 8 and at 13 hours a day, but at 13 hours more crews are needed, and the crew cost per flight hour does not fall. And if the sector length means a crew covers only one pair per duty, as in the diagram with the 2.5-hour sector, crew cost per hour actually rises.

How not to sit on the ground: maintenance, contracts, stations and spare parts

An aircraft earns money only in the air, and maintenance is the main legitimate reason to keep it on the ground. That is why airlines flying 12–13 hours a day do not choose between utilisation and maintenance: they fit maintenance into the gaps the network leaves anyway — nights at base and the winter low season — and use contracts, reserves and spare parts to turn unscheduled failures into predictable costs. Here is how it works, step by step.

The maintenance programme: checks, intervals and whichever comes first

The aircraft manufacturer issues a maintenance planning document (MPD); the airline turns it into its own maintenance programme, approved by the aviation authority. The programme's tasks are grouped into checks — work packages with thresholds in flight hours, cycles (one take-off and landing) and calendar time; a check is due when the first threshold is reached. Light checks are frequent and short, heavy checks rare and long:

Two things follow from this table for the network. Daily and weekly checks and the A-check are done overnight — so the aircraft must regularly spend the night where there are mechanics, a hangar and spare parts, and this is one of the main requirements when choosing bases. Outstations do the minimum; the rest is saved up for the weekly check at base ^39,40,43. Second: at high utilisation, hours and cycles run out before the calendar does. In the specification of one 737-8 MAX operated by SmartLynx (2024), the base-check limit is 7,500 hours: at 10 block hours a day it is reached after about 2.4 years, at 16 after a year and a half; at six flights a day, 5,000 cycles are reached in 2.3 years^41,44. Every extra thousand flight hours brings the next downtime closer.

Schematic: monthly utilisation is illustrative, 2-hour sectors, 20 minutes of taxi per flight; A-check every 750 flight hours (up to 1,000 in optimised programmes). The C-check is placed in January.

Long downtimes are shortened in two ways. The first is an equalized programme: C-check tasks are spread over overnight visits, so there is no multi-week downtime at all. easyJet has worked this way since 2004; JetBlue split its A-check package into 32 segmented overnight checks^39,42,57. The price is a hangar and a crew at base every night, and as the fleet ages there are more findings and the scheme works less well: easyJet reviewed it in 2019⁵⁷. The second is to move heavy checks to winter, when fewer aircraft are needed anyway and third-party hangar slots are cheaper⁴⁵. Ryanair buys heavy-maintenance “lines” at third-party MROs for the winter seasons — in 2024, for instance, ten lines in Amman for ten years^54,58. So the winter dip in utilisation is partly filled with maintenance, and in summer every aircraft flies.

How downtime is avoided

Unscheduled downtime — the aircraft cannot depart because of a defect, AOG (aircraft on ground) in the jargon — is rare but expensive. Normal technical dispatch reliability for a narrowbody fleet is 99.4–99.8 % of departures without a technical delay of more than 15 minutes or a cancellation: the world average for the A320neo is 99.59 %, easyJet's is 99.57 %^59,60,61,62. That is 2–6 technical disruptions per thousand departures; for a fleet of a hundred aircraft with six flights a day, one to four a day.

The first tool against downtime is the minimum equipment list (MEL): it allows departure with certain defects provided they are fixed in time — 3 days for category B, 10 for C, 120 for D⁶³. For the network this is key: a defect found at an outstation is deferred until the night at base, where both the part and the mechanic are.

The second is reliability and failure prediction. These are no longer marketing claims but confirmed figures: Delta cut maintenance cancellations from 5,647 in 2010 to 55 in 2018^64,65; between January 2019 and September 2025 easyJet's predictive models prevented 1,343 cancellations — about two hundred a year for a fleet of roughly 350 aircraft — and 78 % of on-wing failures of such components would have meant more than three hours of downtime^60,66. By replacing starter valves before failure, flagged by slow engine starts, British Airways cut A320neo fleet downtime by about 4 %⁶⁷. A small airline usually has no data science of its own, but it gets part of the effect through its contracts: the engine maker and the component provider monitor the units themselves^46,50.

The third is reserves. In summer 2024 easyJet held 14 spare aircraft, about 4 % of the fleet, Ryanair about eight, 1.3 %^68,69,70,71. Spare engines are 14–15 % of installed engines for the CFM56 and V2500 and about 21 % at Wizz Air with its troubled GTFs^72,73,74. If there is no reserve of one's own or a repair drags on, airlines wet-lease an aircraft with crew (ACMI) — from a few flights to months⁷⁵. A reserve means lease payments without flying, so it directly lowers the fleet's average utilisation.

Contracts: who carries the risk

Across the IATA sample, maintenance averages 1,522 dollars per flight hour, about 11.5 % of operating costs; half of that money goes on engines, 18 % on components⁷⁶. Two thirds of engine costs are shop visits⁴⁹: rare, huge and hard-to-predict payments. So the key question of any maintenance contract is who carries the risk of failure.

Under a “by the hour” engine contract (power by the hour, PBH) the manufacturer takes on the failure risk and the cost of all repairs, and the airline pays a rate for every engine flight hour^46,47,48. These contracts run for 10–12 years and are not only for giants: China's Colorful Guizhou, with a fleet of about fifteen A320neo, signed an agreement with CFM for a twelve-year contract worth about 500 million dollars at list price ^47,49. PBH is particularly attractive on a new engine type — the manufacturer shares the reliability risk of new technology⁷⁷. The price of predictability is a risk premium in the rate and annual escalation in the high single to low double digits, plus dependence on a single supplier who knows more about the engine than the airline^49,78.

For components — pumps, valves, avionics boxes — the PBH equivalent is total component support: the provider gives access to its pool, repairs removed units, ships parts and places a kit of critical components right at the customer's main base, for 5 to 11 years^50,51. Everything else is bought case by case — by exchange, consignment or one-off order:

Small and large airlines benefit from different things, and the reason is scale. Component stock grows more slowly than the fleet: in a shared pool one spare unit covers many aircraft at once, and the larger the fleet in the pool, the fewer spares are needed per hundred aircraft^79,80. A small airline cannot get this effect on its own, so PBH for engines and total component support pay off for it: it buys someone else's pool scale. Large airlines build their own, but even Ryanair, with six hundred aircraft, mixes in-house and bought-in maintenance ⁸¹.

Line stations: where to keep your own mechanics

Line maintenance — checks before departure, during turnarounds and overnight — is always done where the aircraft is, whereas heavy maintenance can be “flown in”: third-party MROs serve customers within roughly three hours' flying ⁸². So for each airport in the network the airline decides whether to keep its own line maintenance station there or to buy the service from a contractor. We found no public numerical threshold; the deciding factor is the workload, above all the number of aircraft staying overnight. In 2019 American Airlines opened its own station in Houston, where it had about 38 departures a day and up to 10 aircraft overnight, with a staff of about 46 — to have more serviceable aircraft for the morning bank⁸³. Where flights are few, the station is handed to a contractor⁸⁴; third-party providers run on-call stations and mobile teams⁵⁵. As a price benchmark, S7 Technics' public 2025 price list for line maintenance for foreign carriers: 88 dollars per man-hour, a technician call-out with defect rectification of up to one hour — 450 dollars⁵⁶.

In Russia, requirements for maintenance organisations have been set since 1 March 2024 by FAP-109 (Russian aviation rules), which replaced FAP-285⁸⁵. A line station is a unit of a certified maintenance organisation, not a separate certificate⁸⁶. To open one you need qualified staff entitled to sign the aircraft's release to service (no more than half of them contracted from outside), calibrated tools and equipment, and a store keeping serviceable and unserviceable parts separately^85,86. At the end of 2024 Aeroflot Technics had 20 line maintenance stations in Russia, five of them in the Far East; they carry out line maintenance on the A320 and Boeing 737 ^87,88.

Where the spare parts are

Keeping every part at every station is too expensive⁷⁹, so stock is laid out in layers: the closer to the aircraft, the faster, but the more money is tied up on the shelf.

Schematic based on the KLM Component Services stock study and the AFI KLM E&M, IATP and fly-away kit providers' contracts^80,89,90,91.

Stock is sized for a target service level — the share of cases in which the part is there when needed: KLM Component Services' contractual level is 95 %, its calculation targets 99 %, yet for the Boeing 737 it actually achieved 82 % ⁸⁰. Today the main risk is engines: shop repair times have grown by 35 % or more for older models and by more than 150 % for new ones⁹². In 2025 this costs airlines 2.6 billion dollars in extra engine leases and 1.4 billion in excess inventory⁹³: the longer the repair, the more parts must be in circulation.

Russia after 2022

Russian airlines lost access to foreign shops and to CFM and Pratt & Whitney engine parts⁹⁴. Aeroflot Technics maintains more than 300 group aircraft and since 2023 has performed more than a thousand A- and C-checks a year ⁹⁵; the first in-house repair of three CFM56-5B engines has been completed, and an engine repair plant is planned for the end of 2027⁹⁴. The group's maintenance costs under IFRS rose from 23.4 billion roubles in 2023 to 57.1 billion in 2024 — 2.4 times — while capacity grew by 18 %⁹⁶; the company explains the growth of operating costs as a whole by the expanded programme, and public sources do not break down the maintenance increase separately. According to media reports, at the end of 2022 the government allowed components to be taken from some aircraft to repair others, but the text of the decision has not been published⁹⁷.

What this means for the network

High utilisation rests on four pillars. Every aircraft regularly spends the night at a base with its own station, hangar and stock — that is where daily and weekly checks and the A-check are done and deferred defects are fixed. Heavy checks are scheduled for winter or spread across nights. A reserve of 2–4 % of the fleet or an ACMI arrangement covers breakdowns. “By the hour” contracts and pools turn rare, expensive failures into a rate per flight hour — which is exactly how the calculator counts maintenance.

What a tight schedule costs

The more flights an aircraft has in a day, the further the last one drifts from the plan. In Europe in 2024 the average delay was 17.5 minutes per flight, and 46 % of those minutes were reactionary: the delay came from the same aircraft's previous flight⁹⁸. The first morning wave leaves almost on time — 6.6 minutes of delay on average in 2025 — and delays build up towards the evening^98,99. Summer is worse than winter: 21 minutes per departure in July 2025, 11.6 in November^100,101.

To counter this, buffers are built into the schedule — and paid for with utilisation. A minute added to block time removes between 0.24 and 0.57 minutes of arrival delay; a minute added to the turnaround, about 0.17 ^102,103. Ten minutes of buffer on each of six turnarounds is an hour of flying a day ^98,102,103. The sawtooth chart shows this literally: extra minutes on the turnaround can knock utilisation down by a whole pair of flights.

Low-cost does not necessarily mean later: Iberia Express was Europe's most punctual airline in 2025 with 88.9 % of flights on time, while in the UK in 2023 the largest average delay was Wizz Air's — 31.6 minutes — against 19.9 for Ryanair and 18.5 for Lufthansa^104,105. Punctuality is a matter of buffers and reserves, not of the business model.

Finally, the season. In winter demand drops, and an airline has three options: park some aircraft, send them for heavy maintenance, or lease them out with crews to where it is high season. European ACMI operators (wet lease with crew) take aircraft to the southern hemisphere in winter: in winter 2024/25 six such aircraft flew for Argentina's Flybondi ^106,107. Wizz Air shows the seasonal swing well: 11 hours 35 minutes a day across the whole fleet from October to December 2023¹⁰⁸, and 10 hours 9 minutes from April to September 2025, when some aircraft were grounded by engine problems⁷³.

Table for the calculation

Benchmarks by network type — for entering your own utilisation into the calculator. The annual average is what to enter; the summer peak is for checking that the schedule is realistic at all.

What utilisation does to ownership cost per flight

Current-generation A320, the calculator's default lease of 200,000 EUR a month, Frankfurt — Barcelona flight of 2.12 block hours.

What we are changing in the calculator

1. ⚑ Mind when entering dataEnter utilisation across the whole fleet, averaged over the year, not the summer peak and not the utilisation of flying aircraft: the lease is paid for winter too, and for aircraft in reserve or in maintenance. The table above is the benchmark.
2. ✓ Built into the calculatorUtilisation by network type instead of a single 300-hour figure (9 Nov 2026). In the “Network” mode the default is the peak day from the window formula on the network's sectors, multiplied by the annual factor of the network type. The factor turned out not to be 0.8 for all: from Form 41 T-2 2025 (block hours ÷ aircraft days of US carriers against the peak day at their average sector length), a base shuttle is 0.81, a hub 0.74, low-cost and leisure carriers with strong seasonality 0.52, long-haul 0.73 (one type, weak). Frontier and Southwest were left out of the fit; on them the model is off by 4–10 %. US regional carriers are not used: their aircraft days are counted differently.
3. ✓ Built into the calculatorCrews from annual hours. Crews per aircraft = aircraft hours ÷ pilot flight hours (750 hours by default, range 500–900); if the aircraft's peak day does not fit one EASA duty, a warning says how many duties a day are needed. The crew cost per block hour does not change.
4. ⚑ Mind when entering dataBase window — a field in the “Network” mode, 18 hours by default. ✓ Built into the calculatorA night ban at the base, from the airport reference data, narrows the window (wider than 18 hours only by hand: the annual factor is calibrated at 18); quotas and bans at other airports of the network come as a warning. An agent fills the reference data, eight airports a week.
5. ✓ Built into the calculatorReserve and scheduled maintenance as explicit shares within the annual factor: reserve 3 %, heavy maintenance 3 % (assumptions: there is no open per-carrier data), the rest is winter and delays. In the fleet report the reserve is a separate line of aircraft, and utilisation there is per aircraft in service, with no double counting.

How it was checked

Four researchers kept search logs — 898 steps including extracts from reports (trace/n1–n5): airline utilisation, network mechanics and design, crews and delays, maintenance, contracts and spare parts. Other agents opened every source and checked the quotes (verify/): utilisation — 65 of 65 found; mechanics — 68, of the inferences 5 correct, two corrected; crews and delays — 74 verbatim and 1 partially; maintenance, contracts and spare parts — 179 verbatim and 3 partially, all 10 calculations recomputed.

From the check of the maintenance section: the pool-effect example in the KLM study turned out to be illustrative — we give the principle without the numbers; the “3 years” interval for the 737-8 MAX comes from one aircraft's specification, not from a maintenance programme; the Colorful Guizhou contract with CFM is a memorandum of understanding; Aeroflot explains the rise in maintenance costs only as part of operating costs as a whole.

Corrected after checking: the utilisation formula is in block hours with turnaround, without double-counting taxi; the comparison “low-cost carriers fly more than network carriers” was removed — the data do not show it; the Düsseldorf window is scheduled landings until 23:00; the long-haul rotation example was removed (not proven to be one aircraft); crew numbers are based on limits and reporting, not on an old pilot-hours figure; the duty table by number of landings from Order No. 381 is not given until checked against the official text. Aeroflot group utilisation was calculated from the Databook (operating and fleet sheets); utilisation series older than 2023 were removed.