Most disasters on this site are errors of translation — someone says metres and someone hears feet, a thruster fires in pound-force-seconds and a satellite reads newton-seconds. Air France Flight 4590 is different. Nobody confused one unit with another. The math in the cockpit and on the loading sheets was, on paper, correct. What went wrong was subtler, and in its own way more instructive: an aircraft whose entire existence depended on flying at the extreme edge of its numbers — weight, balance, temperature, speed — had those numbers pushed, one by one, to their very limits. When a single 43-centimetre strip of the wrong metal appeared on the runway, there was no margin left to absorb it.
To understand why the world's only supersonic airliner died on a clear July afternoon, you have to understand two things. First, what Mach actually is — a unit that is not a fixed speed at all, but a ratio that changes with the temperature of the air. Second, what "no margin" means when the margin is measured in kilograms, in knots, and in a few centimetres of metal.
Mach: The Speed Unit That Isn't a Speed
Concorde was built around a single number — Mach 2 — and Mach is the strangest unit anyone ever bought a ticket for. A Mach number is not a speed. It is a ratio: the aircraft's speed divided by the speed of sound in the air around it. Mach 1 means you are travelling exactly at the speed of sound. Mach 2 means twice that. And the denominator — the speed of sound — is not a constant. It rises and falls with the temperature of the air, which is why the same "Mach 2" is a different physical speed at different altitudes.
At sea level, in air at 15°C, sound travels at about 1,225 km/h (761 mph). Climb to Concorde's cruising altitude of 60,000 feet, where the air sits near −57°C, and sound slows to roughly 1,062 km/h (660 mph). The physics is exact: for an ideal gas, the speed of sound depends on the absolute temperature — not on pressure or density on their own — through
where γ ≈ 1.4 is the ratio of specific heats of air, R ≈ 287 J/(kg·K) is the gas constant, and T is the temperature in kelvin. Halve the absolute temperature and the speed of sound does not halve — it falls by the square root of that change. That is why a 100-degree drop in temperature costs an aircraft a meaningful chunk of its Mach speed.
So when Concorde's pilots pushed the nose to Mach 2.02 over the Atlantic, they were not flying at 2.02 × 1,225 km/h. They were flying at 2.02 × 1,062 — roughly 2,150 km/h, about 1,330 mph. The commonly quoted top speed of 2,179 km/h is Mach 2.04 at a slightly lower altitude. The point stands either way: a Mach number is meaningless until you know the temperature of the air it was measured in. Read it without that, and you are reading a number without its unit. If you need to move between the speed units in a hurry, convert km/h to mph or work in any of the common speed units.
Concorde's whole design was an argument with this number. Its aluminium skin, thinned to save weight, heated up from Mach-2 friction until the fuselage stretched several centimetres in flight — enough that a gap opened between the flight engineer's console and the cockpit bulkhead. Its long, drooping nose existed purely so the pilots could see the runway over the steep angle of approach the delta wing demanded. Every kilogram mattered, because at Mach 2, drag and heat punish every extra gram. But on July 25, 2000, none of that mattered. Concorde never reached Mach 2. It never got above 200 feet. The numbers that killed it were not Mach numbers; they were the ground-level numbers of a takeoff.
The Edge: 186 Tonnes, a Tail Wind, and Three Speeds
For this takeoff, the critical reference speeds were expressed in knots, and they were fixed in advance. V1 — the decision speed, the point after which a rejected takeoff is no longer possible — was 150 knots. VR, the rotation speed, was 198 knots. V2, the takeoff safety speed, was 220 knots. In kilometres per hour, that is 278, 367, and 407. Below V1 the crew could still slam on the brakes and stop; above it, they were committed to fly no matter what went wrong. Knots convert to km/h at 1.852, and it is the knot that every one of these decisions hangs on.
The aircraft was heavy. It weighed about 186 tonnes — roughly a tonne over its certified maximum takeoff weight of 185 tonnes, and 810 kg (about 1,790 lb) over the weight the day's conditions actually allowed. Nineteen items of luggage, about half a tonne, had been loaded at the last minute without appearing on the manifest. And the load was placed tail-heavy: the centre of gravity sat aft of the takeoff limit.
Then the wind turned. An 8-knot tailwind sprang up — nothing much on its own, but a tailwind eats runway, and it quietly cut the regulated takeoff weight by a further six tonnes. The crew did not turn the aircraft around to take off into the wind.
None of this, the investigators later concluded, was what brought the aircraft down — the excess weight was judged to have a negligible effect on takeoff performance. But it is the context. These conditions had trimmed the operational margin a healthy aircraft keeps for emergencies, in kilograms and in knots, before the wheels even began to roll. Each trim was too small to matter on its own; together they meant the aircraft began its takeoff with less slack than its paperwork assumed.
435 Millimetres of the Wrong Metal
Five minutes earlier, a Continental Airlines DC-10 had taken off from the same runway. As it climbed away, a small metal strip fell from the thrust-reverser cowl door of its number 3 engine and landed on the tarmac. The strip was 435 mm long, 29 to 34 mm wide, and about 1.4 mm thick — roughly 17 inches long and a little over an inch wide, barely thicker than a coin.
The size was not the problem. The material was. It was a wear strip, and the manufacturer's specification called for it to be stainless steel. The one on the runway was a TA6V titanium alloy — roughly 90 per cent titanium, with about 7 per cent aluminium, 2 per cent vanadium, and a trace of iron — fitted two weeks earlier at a maintenance facility in Houston in the wrong material. Titanium alloy is strong and light, which is exactly why a mechanic might reach for it. But it is also harder than steel, and when a Concorde tyre rolled over it at 300 km/h it behaved very differently from steel. It did not bend or flatten. It held its edge — and sliced.
At about 300 km/h — 186 mph, 162 knots, just past V1 — the number 2 tyre of the left main landing gear ran over the strip and burst.
The Hydrodynamic Surge Nobody Had Seen Before
A tyre bursting at speed is not a puncture; it is an explosion. A chunk of the failed tyre weighing 4.5 kilograms — about 10 pounds — was hurled upward into the underside of the left wing at an estimated 140 metres per second. That is 504 km/h, over 300 mph, and it carried about 44 kilojoules of energy — the kinetic energy ½ m v², or ½ × 4.5 kg × (140 m/s)², working out to roughly 44,100 joules.
For years the accepted story was that this chunk of rubber simply punched a hole in the fuel tank. The official investigation found something stranger. The debris did not pierce the tank. It slammed into the wing skin directly over the number 5 fuel tank — which, because of how fuel had been transferred during taxi, was 94 per cent full. Investigators concluded that two physical processes, acting nearly simultaneously, produced the rupture. The first was a continuity effect: the impact deformed the tank skin inward, straining and rotating the surrounding structure outward. The second was a hydrodynamic pressure surge — the impact sent a pressure wave racing through the nearly full, nearly incompressible fuel, and that wave pushed back against the tank from the inside. Together they blew a section of the tank's bottom skin outward. A panel about 32 cm square tore clean away and was later found lying on the runway.
The BEA called it "a type of failure never experienced before in the history of civil aviation." Fuel began pouring from the breach at roughly 60 litres per second — about 3,600 litres a minute. It ignited almost immediately; the most likely ignition source, investigators concluded, was electrical arcing from damaged wiring in the landing-gear bay. Litres per second convert to litres per minute with a simple factor of 60, and at that rate the leak did not need long to become a fire.
By now the aircraft was past V1, and stopping was no longer an option; the pilots had no choice but to rotate and fly. Engine 2, on the burning left side, surged and lost thrust, and the flight engineer shut it down on a fire warning. Engine 1 began to fail behind it. The landing gear, damaged, would not retract, and hung down as drag. The Concorde lifted off, climbed to about 200 feet at 200 knots, and could not accelerate or climb any further. The first officer called out for Le Bourget, the airfield a few kilometres away. The captain answered, "Too late." About 90 seconds after the takeoff roll began, the aircraft stalled, rolled, and fell onto the Hôtelissimo Les Relais Bleus hotel in Gonesse, roughly ten kilometres from the airport. All 109 people on board and four on the ground were killed.
What the Crash Teaches Engineers
Air France Flight 4590 was not a unit conversion error, and it would be dishonest to dress it up as one. The thing that failed was not a unit of measurement but a unit of engineering judgment: the margin. Concorde was an aircraft that could only exist at the edge of its envelope, and on July 25 every edge had been reached at once — the weight, the balance, the wind, and then a piece of foreign metal that should never have been there. Three lessons survive the crash.
The first is about material properties, which engineers learn to treat as a number in themselves. The strip that should have been steel was a titanium alloy, and the two metals answer a 300 km/h impact in opposite ways. A part substitution is a unit conversion in disguise: you cannot swap one material for another and assume the part is otherwise identical, because the property that mattered here was not the strip's size but its hardness. The same dimension, made of a different substance, is a different part.
The second is about margins at the edge of the envelope. The overweight, the tail-heavy balance, and the tailwind were each judged, on their own, to be negligible — and on their own, they were. What they did was consume the slack. A margin that is negligible in isolation is still a margin you no longer have when everything else goes wrong at once.
The third is about the unit Concorde was defined by. A Mach number is not a speed; it is a speed measured against the local speed of sound, and the speed of sound changes with temperature. Read a Mach number without its temperature, and you are reading a number without its unit. The habit of mind that refuses to trust a number until it knows what the number was measured against — that is what separates a unit that informs from a unit that lies. Concorde spent its whole life at Mach 2, and died, in the end, on a number nobody had to convert at all.
Related Converters
The crash turns on speed, weight, length, and energy — the four unit families that defined Concorde's envelope:
- Speed — all speed converters: km/h, mph, m/s, knots, ft/s
- Knots to km/h — V1 = 150 kt, VR = 198 kt, V2 = 220 kt (278 / 367 / 407 km/h)
- km/h to mph — the 300 km/h tyre impact = 186 mph
- m/s to km/h — the 140 m/s debris strike = 504 km/h
- kg to lb — the 810 kg overweight = 1,790 lb
- mm to inches — the 435 mm titanium-alloy strip = 17.1 in
Related Guides
- Speed Conversion Guide — m/s, km/h, mph, knots, ft/s, and why the knot is defined by the Earth's circumference.
- The 8 Most Expensive Unit Conversion Mistakes in Engineering History — the Mars Climate Orbiter, Gimli Glider, and the other disasters where a number went wrong.
- Korean Air Cargo 6316: When 1,500 Meters Became 1,500 Feet — the other aviation disaster where a number, not a machine, failed.
- Deepwater Horizon: The 0.2 ppg That Broke the Gulf — the same "margin consumed" failure chain, in a well instead of a wing.
Frequently Asked Questions
Was the Concorde crash caused by a unit conversion error?
No. Unlike the Mars Climate Orbiter or Korean Air 6316, nobody confused one unit with another. Air France Flight 4590 was caused by a 43 cm strip of titanium alloy — the wrong material — that burst a tyre at 300 km/h. The tyre debris triggered a hydrodynamic pressure surge that ruptured a fuel tank, and the resulting fire brought the aircraft down. The overweight and tail-heavy loading were judged to have a negligible effect on their own.
What does Mach 2 mean, and why isn't it a fixed speed?
A Mach number is a ratio: the aircraft's speed divided by the local speed of sound. Because the speed of sound depends on air temperature — about 1,225 km/h at sea level (15°C) but only 1,062 km/h at 60,000 ft (−57°C) — the same "Mach 2" is a different physical speed at different altitudes. Concorde's Mach 2.02 cruise worked out to roughly 2,150 km/h (1,330 mph).
Why did the Concorde tyre fail?
The number 2 tyre on the left main gear ran over a strip of metal lying on the runway at about 300 km/h, just past V1. The strip should have been stainless steel but was a harder TA6V titanium alloy, which held its sharp edge instead of flattening — and sliced through the tyre, bursting it instantly.
What was the titanium strip made of?
It was a TA6V titanium alloy — roughly 90 per cent titanium, with about 7 per cent aluminium, 2 per cent vanadium, and a trace of iron. The manufacturer's specification called for stainless steel, but the strip was fitted in titanium alloy during a repair in Houston, in the wrong material.
How did a 43 cm strip of metal bring down Concorde?
The strip, which should have been stainless steel but was a titanium alloy, held its sharp edge under a 300 km/h impact and burst the number 2 tyre. A 4.5 kg chunk of tyre hit the wing at 140 m/s and — instead of piercing the tank — set off a continuity effect and a hydrodynamic pressure surge inside the 94%-full number 5 tank, blowing a 32 cm panel outward. Fuel leaked at 60 L/s and ignited, and the aircraft could not climb away.
What should engineers learn from this accident?
Three things. Material properties are a number in themselves — swapping steel for a titanium alloy changes the part even if the dimensions are identical. Margins consumed in isolation are margins you no longer have when everything fails at once. And a unit like Mach is only meaningful against what it is measured against — the speed of sound, which itself changes with temperature.
Sources
This account is based on the following sources, grouped by what they establish:
Accident investigation (primary):
- Bureau d'Enquêtes et d'Analyses (BEA), Accident on 25 July 2000 at La Patte d'Oie in Gonesse (95) to the Concorde registered F-BTSC operated by Air France — final report, 16 January 2002
- FAA, Lessons Learned from Civil Aviation Accidents: Concorde (F-BTSC)
- NTSB safety recommendations to the BEA regarding Concorde tyre failures (November 1981), and the aircraft's ~70 tyre/wheel incidents across 27 years
Physics and unit definitions (secondary):
- ICAO International Standard Atmosphere — air temperature and the speed of sound against altitude
- Speed of sound in an ideal gas, a = √(γRT), with γ ≈ 1.4 and R ≈ 287 J/(kg·K)
- Unit definitions used throughout: 1 kt = 1.852 km/h, 1 in = 25.4 mm, 1 kg = 2.20462 lb