Today the calculator assumes that an aircraft can take off fully loaded on any day. In heat and at altitude this is not so: the allowable (performance-limited) take-off weight drops, and on a long sector with a full cabin some passengers or cargo have to be offloaded. The same flight costs are then split over fewer passengers. Below: when this happens and by how much.
Density decides everything. Wing lift and drag are directly proportional to air density, and density depends on pressure, temperature and humidity6. That is why pilots work not from the airfield elevation but from “density altitude” — the altitude at which the standard atmosphere would have the same air7. To an aircraft, a hot day at sea level is the same as a cool one in the mountains: it behaves as if it were taking off from its density altitude7.
Two blows to the take-off. Thin air has two effects at once: a higher lift-off speed is needed and thrust is lower, so acceleration is weaker7. The aircraft lifts off at the same indicated airspeed, but its true airspeed is higher — the take-off run is longer7. The opposite lever is weight: +10 % weight lengthens the take-off by at least 21 %7. So when the runway is not long enough, weight comes off.
Flat rating. A jet engine is set up to hold take-off thrust constant up to a certain temperature — the “kink” (corner) point: usually 30 °C at sea level1, up to 45 °C for A320 engines, up to 30 °C for the A3218, up to 37.4 °C for the Dash 8-4009. Below the kink, heat hardly matters; above it, thrust falls together with the allowable weight. Hence the threshold in the calculation.
The limit is not just the runway. Certification rules require take-off performance to be determined for every weight, altitude and temperature10. The take-off must fit on the runway even with an engine failure10, and after lift-off a twin-engine aircraft on one engine must still climb at a gradient of at least 2.4 %10. The allowable weight is the lowest of all limits: runway, climb, obstacles, tyre speed, brakes11. Denver's runway is 4.9 km long, yet the 737-800 runs into its tyre speed limit2.
Which temperature to use. US airports are designed for the mean daily maximum of the hottest month11 — we take the same temperature for the flight calculation.
An A320 at maximum take-off weight: how many passengers (100 kg each with baggage) would have to be offloaded. Dot — estimated correction, line — spread of the slope; orange — where the basis is weaker (high altitude, short runway).
Denver — with the ISA+12 threshold for high airfields; in reality the weight is limited there even on a cool day (tyre speed), so the estimate is, if anything, too low.
Reduction of the 737-800 allowable take-off weight, % of maximum (79 t). Dots — Coffel & Horton (2015), table 22, from Boeing charts; straight lines between the dots.
Each line starts at the threshold and then rises almost linearly. The shorter the runway, the steeper: at LaGuardia (2.1 km) the limit grows from 4.5 to 6.8 t between 31 and 33 °C. The slopes and thresholds were derived by the researcher from the paper's three points — the paper itself does not give them. The authors convert the same 4.5 t into 52 passengers — they assume 86 kg per passenger with carry-on2; the calculator uses 100 kg with checked baggage, so we get fewer passengers. In practice cargo is offloaded first2.
For airport planning, manufacturers publish charts showing what weight the aircraft can lift from a runway of a given length at a given elevation, at standard temperature and in the heat. We read the figures off them for the 737-800 (standard, +15, +25 and +35 °C above standard)4 and for the A320ceo and A320neo (standard and +15 °C)5. The chart below is the 737-800 on a 2.9 km runway: at sea level the weight starts to fall only beyond +25 °C above standard, i.e. around 40 °C; at 4,000 ft already beyond +15 °C, and past +25 °C the drop speeds up; at 8,000 ft the weight is limited even on a standard day.
Horizontal axis — degrees above standard; one airfield elevation — 6,000 ft. For the A320, Airbus charts exist only for standard and +15 °C.
Horizontal axis — airfield elevation, thousand ft; the temperature relative to standard is constant. 737-800 — 2.9 km runway; at +25 °C and low elevation the weight exceeds the maximum take-off weight, so the line starts at 4,000 ft.
Horizontal axis — degrees above the standard atmosphere. 737-800, CFM56-7B27, 9,500 ft (2.9 km) dry runway; weight capped at the maximum take-off weight (79 t).
These charts give the slopes for the calculation. For the 737-800 the loss of allowable weight is 0.26 % per degree from standard to +25 °C and 0.53 % (0.32–0.67) from +25 to +35 °C; every 1,000 ft of elevation takes about 2.1 %4. The second figure matches the slope the researcher derived from Coffel and Horton for Phoenix — 0.48–0.57 %: two independent bases agree. For the A320 the weight barely changes up to +15 °C above standard — 0.05–0.15 % per degree, because its engines hold thrust up to 45 °C; elevation takes 1.2–1.4 % per 1,000 ft5. Airbus has no charts above +15 °C, so for heat beyond the kink point we use the 737-800 slope for the A320 — an assumption.
| Case | Threshold | Weight loss per °C | A320, pax per °C | Basis and spread |
|---|---|---|---|---|
| long runway (≥3.5 km), airfield <2,000 ft (Delhi, Gulf) | ISA+25 | 0.5 % | 4.0 | 0.48–0.57 |
| high altitude ~5,000 ft, long runway (Denver, Nairobi, Johannesburg — extrapolation) | ISA+12 | 0.4 % | 3.2 | 0.41; plus ≥0.6 % at any temperature (tyres) |
| short runway (~2.1 km) at sea level | ISA+12 | 1.4 % | 11.1 | 1.43 |
| 7,000–8,400 ft (Addis Ababa, Bogotá, Mexico City), La Paz, Cusco | — | — | — | no data; extrapolating from DEN understates |
Threshold — how many degrees the hot month is warmer than the standard atmosphere (ISA: 15 °C at sea level, 2 °C less for every 300 m of elevation) before the limit starts.
| Type | What is known | Figure | Conditions |
|---|---|---|---|
| A320 (ceo) | take-off distance, not weight | TODR +4.2 m/K (min. load), −1.2 m/hPa; at max. load ≈7.4 m/K and 2.2 m/hPa (paraphrase)12 | force-balance model on OpenAP data, ISA |
| A320 (ceo) | MTOM on a 1,511 m runway (Chios) | −133 kg/year (climate trend in temperature and wind); 3,900 kg ≈ 38 passengers since 198813 | runway-limited airfield |
| A320/A321neo | TOW vs runway length charts (ISA and ISA+15) are in the Airbus AC; figures not extracted | —14 | AC A320 06/2024 §3-3 |
| 737 MAX 8 | BFL at 92 t | 8,300 ft (JFK) → 11,500 ft (Denver), standard day (+39 %)15 | Leeham model example |
| ATR 72-600 | take-off distance at the same weight (300 nm, max. passengers) | 1,156 m (SL, ISA) → 1,383 m (3,000 ft, ISA+10), +20 %16 | ATR factsheet |
| Dash 8-400 | flat rating 37.4 °C at SL | —9 | DHC spec sheet 2025 |
| E190/E195 | not found | —17 | only the E195-E2 TOFL of 1,840 m is available (MTOW, ISA, SL) |
| Case | Figures |
|---|---|
| Aeroméxico: weight restrictions on some flights out of MEX due to heat, April 2024 (34.2 °C — the highest since 1998)18 | seats/kg not published |
| IndiGo: Delhi–Bagdogra delays >3 h, 'high ground temperatures', June 202419 | none |
| Leh: IndiGo and SpiceJet cancelled ~12 daytime flights over ~3 days, July 2024, 36 °C20 | ~12 flights |
| Delhi: airport operations disrupted for ~3 h by heat (Palam 44.9 °C); 'load carrying capacity falls'21 | 3 h |
| Aeroméxico MEX–NRT: from 2024-07-10, 5 of 7 flights via MTY, non-stop on the way back (787)22 | 5/7 a week |
| Air France restricted A340 loads on take-off from Bogotá (2,548 m)23 | none |
| Coffel et al. 2017 projection: 10–30 % of flights in the hottest hours will need restrictions, on average 0.5–4 % of payload+fuel; 4 % ≈ 12–13 passengers out of 16024,25 | see text |
Airlines describe the restrictions in words — nobody publishes how many seats were blocked.
ΔISA = Thot month − (15 − 1.98 · h)
Mallow = min( MTOW ; Mrwy(L) · [1 − ah·h − k1·min(ΔISA, D) − k2·max(0, ΔISA − D)] )
offload = max(0, Mplan − Mallow); passengers = offload ÷ 100 kg
cost per passenger = flight cost ÷ (N − passengers)
| Term | What it is | 737-800 | A320ceo/neo |
|---|---|---|---|
| h | airfield elevation, thousand ft | — | — |
| Mrwy(L) | weight a runway of length L allows at sea level on a standard day (manufacturer chart) | ≈ 90 t at 2.9 km (extrapolated: the curves are cut off above MTOW) | 84 t at 3.0 km |
| ah | loss per 1,000 ft of elevation | 2,1 % | 1,2–1,4 % |
| k1 | loss per degree up to the thrust kink | 0,26 % | 0,05–0,15 % |
| D | kink: how many degrees above standard thrust holds | +25 °C | ≈ +30 °C (45 °C at sea level) |
| k2 | loss per degree beyond the kink | 0,53 % (0,32–0,67) | as 737-800 — assumption |
| Mplan, N | planned take-off weight of the flight and number of seats — already in the calculator | — | — |
Coefficients from the Airbus and Boeing charts4,5; if the runway length is unknown, the simplified table above applies: one slope k from the threshold.
The log has 176 steps: the researcher's search (trace/h1.jsonl) and the theory from the owner's documents (trace/h2.jsonl). Quotes were checked (verify/): found verbatim 74, partly 2, not found 0. A separate agent reproduced the calculation of the thresholds, the slopes and the Delhi example; table 2 of the journal version of Coffel & Horton matched the conference version in every figure.
trace/h3.jsonl; digitised values in out/*_digitized.json. There are no charts yet for the A321, E-Jets and turboprops.trace/h2.jsonl).