The real story of Flight 4590
A freak ‘single cause’ accident was blamed when an Air France Concorde crashed in flames last year. But a number of avoidable errors may have played a part in the disaster, writes David Rose
It is an indelible image, heavy with symbolism: the photograph taken on July 25 last year at the moment Concorde became a technological Icarus.
The great white bird rears up over runway 26 at Charles de Gaulle, immediately after takeoff.
Already mortally wounded, flames bleed uncontrollably from beneath the left-hand wing. Less than two minutes later, the world’s only supersonic airliner will fling itself into the Paris suburb of Gonesse, killing all 109 on board and another five on the ground.
In the 10 months since the crash, the official investigation into its causes has focused almost entirely on the fire. According to the French accident investigation bureau, the BEA, it broke out when the plane passed over a strip of metal on the runway. A tyre burst; a chunk of rubber thudded into a fuel tank inside the wing; jet fuel poured out of a hole and ignited. The hot gases caused two of the engines to falter, and despite a valiant struggle by Captain Christian Marty, the loss of thrust made the crash inevitable.
This “single cause” explanation is why the Civil Aviation Authority (CAA) grounded BA’s seven Concordes three weeks after the disaster. Its chair, Sir Malcolm Field, said: “It is clear to all of us in the CAA that a tyre-burst alone should never cause the loss of a public transport aircraft ... tyre debris alone is thought to have led to this catastrophic accident.”
An investigation by the Mail & Guardian’s sister newspaper, The Observer, suggests the truth is much more complicated and as far as Air France is concerned, far more worrying. In the words of John Hutchinson, a BA Concorde captain for 15 years, the fire on its own should have been “eminently survivable; the pilot should have been able to fly his way out of trouble”.
The reason he failed to do so, Hutchinson believes, was a lethal combination of operational error and “negligence” by the maintenance department of Air France. This appears to have been a crash with more than one contributing factor, most of which were avoidable.
Go back to that photo. An amazing picture: but where was it taken? The answer is: inside an Air France Boeing 747 that had just landed from Japan and was waiting to cross Concorde’s runway on its way back to the terminal. Its passengers included French President Jacques Chirac and his wife.
Concorde seems to be nearby because it had been close to hitting the 747, which would have turned both aircraft into a giant fireball. Veering wildly to the left, like a recalcitrant supermarket trolley with a jammed wheel, Concorde’s undercarriage had locked askew.
When Marty pulled back on the control column to raise the nose and take to the air the process pilots call “rotation” the plane’s airspeed was only 188 knots, 11 knots below the minimum recommended velocity required for this manoeuvre.
But he had no choice: the plane was about to leave the tarmac altogether and plough into the soft and bumpy grass at its side. That might have ripped off the landing gear, leaving Concorde to overturn and blow up on its own. If not, the 747 lay straight ahead. So he took to the air, although he knew he was travelling too slowly, which would impair the damaged plane’s chances of survival.
The BEA’s interim report notes that according to the cockpit voice record, the instant before Marty rotated, co-pilot Jean Marcot screamed: “Watch out!” The report states: “At this stage it is not possible to explain this exclamation.” Marcot was at the front of a Concorde thundering off a runway at 300kph towards a 747 carrying the president: his exclamation seems reasonably explicable.
But why was the plane in this disastrous position? Shocking evidence now emerging suggests that the underlying reason may have been that Air France Concorde F-BTSC had not been properly maintained. The airline’s ground staff had failed to replace a “spacer”, a vital component of the landing gear that keeps the wheels in proper alignment, when they serviced and reassembled the plane’s undercarriage four days before the disaster. Although the BEA disputes it, there is compelling evidence that it was the missing spacer that may have caused the plane to skew to the left, so forcing Marty to leave the ground too early.
At the same time, the plane was operating outside its legally certified limits. When it stood at the end of the runway, ready to roll, it was more than six tonnes over its approved maximum takeoff weight for the given conditions, with its centre of gravity pushed dangerously far to the rear. According to Hutchinson, even before the blowout, Marty was already “pushing the envelope” of safe flying explored by the test pilots when the plane was being developed in the early 1970s.
Had the plane not hit the metal strip, Marty would almost certainly have got away with it. Faced with an emergency, with his plane in the air flying below a sustainable speed, his options were severely compromised.
The Observer’s investigation suggests Concorde need not have been grounded. Now undergoing a 30-million refit to equip the fuel tanks with new kevlar linings, it was already basically safe. As for Air France, the sense of sorrow over what took place remains palpable: the crash was its worst nightmare. Its lawyers are close to reaching a settlement with the families of the 109 German victims, tourists from Mnchengladbach about to start the holiday of a lifetime. Each family has been offered about 1-million the largest settlement in German legal history.
Disclosure that this may not have been a “single cause” catastrophe may place this in jeopardy. Because the plane was bound for New York, the relatives would be entitled to sue in the United States. If the courts there found Air France culpable, the damages they might award could make 1-million seem trivial.
Most of the raw data about the crash can be found on the Internet, in the two reports already published by the BEA. Its final report is due in a few weeks’ time. Meanwhile, a separate inquiry is being conducted by an investigating judge. The missing undercarriage spacer, and the effects this had on Flight 4590, are now emerging as the judicial investigation’s central issues. If the French courts decide there is sufficient evidence, they have the power to charge Air France with “homicide involuntaire” what British law would term corporate manslaughter.
The stresses on Concorde’s landing gear are unusually severe. Unlike ordinary aircraft, its delta wings generate hardly any “lift” until the captain pulls up the nose and pitches the plane upwards at an angle of 18 degrees at the point of rotation. Until then, the wheels and bogeys will bear all of Concorde’s weight in the case of a fully laden plane at takeoff, about 185 tonnes.
The procedures both Air France and BA impose on their ground crews reflect the obvious danger of getting anything wrong. At regular intervals of a few hundred flying hours, the various load-bearing components become “lifed”, and must be replaced. When the undercarriage bogeys are taken apart and reassembled, the work must be done according to a rigid formula, and rigorously inspected and assessed.
Concorde F-BTSC went into the hangar at Charles de Gaulle on July 18, a week before the crash. The part that was “lifed” was the left undercarriage “beam” the horizontal tube through which the two wheel axles pass at each end. In the middle is a low-friction pivot that connects the beam to the vertical “leg” extending down from inside the wing. The bits of the pivot which bear the load are two steel “shear bushes”.
To keep them in position, they are separated by the spacer: a piece of grey, anodised aluminium about 12cm in diameter and 30cm long. When the plane left the hangar on July 21, the spacer was missing. After the crash, it was found in the Air France workshop, still attached to the old beam that had been replaced.
In the days before the accident, the aircraft flew to New York and back twice. At first, the load-bearing shear bushes remained in the right positions. But each time the plane took off, the landing gear was retracted into the wing. On the ground the two shear bushes are positioned horizontally on either side of the beam. With the gear retracted the right-hand bush lies vertically above the left.
On F-BTSC it began to slip down into the gap where there should have been a spacer. By the day of the crash it had moved about 18cm until the two washers were almost touching. Instead of being held firmly in a snug-fitting pivot the beam and the wheels were wobbling, with about three degrees of movement possible in any direction. As the plane taxied to the start of the runway, there was nothing to keep the front wheels of the undercarriage in line with the back. The supermarket trolley was ready to jam. Exactly when it started to do so is uncertain. Jean-Marie Chauve, who flew Concordes with Air France until his retirement six years ago, and Michel Suaud, for many years a Concorde flight engineer, believe the undercarriage was already out of alignment when the plane began to move down the runway.
They have spent the past six months preparing a 60-page report on the crash, which they have submitted to the investigating judge. Chauve said: “The acceleration was abnormally slow from the start. There were something retarding the aircraft, holding it back.” In his view, it must have been friction from the undercarriage. Chauve and Suaud’s report contains detailed calculations that conclude that without this retardation, the plane would have taken off 1,694m from the start of the runway before reaching the fateful metal strip.
The BEA contests these findings, saying that the acceleration was normal until the tyre burst. It also maintains that even after the blowout the missing spacer was insignificant.
The BEA’s critics say that once the tyre burst the load on the remaining tyres became uneven, and even if the wheels had been more or less straight before, they now twisted to the side. The “smoking gun” is a remarkable series of photographs in the BEA’s own preliminary report. They show unmistakably the skid marks of four tyres heading off the runway on to its concrete shoulder, almost reaching the rough grass beyond.
In one picture the foreground depicts a smashed yellow steel landing light on the very edge of the made-up surface, which was clipped by the aircraft as Marty tried to wrest it into the air. Industry sources have confirmed this probably had further, damaging results. Until then the number one engine had been functioning almost normally, but when the plane hit the landing light it ingested hard material that caused it to surge and fail. This hard material, the sources say, was probably parts of the broken light.
And as one industry insider put it: “You would not see four marks if the wheels had been straight, with the back wheels behind the front. And you should not see such marks at all after a normal takeoff. This plane was skidding sideways. It was out of control.”
Hutchinson said: “The blowout alone would not cause these marks. You’d get intermittent blobs from flapping rubber, but these are very clearly skids.” Overall, the effect on the plane was like trying to take off in an exceptional crosswind a situation Concorde pilots are trained to avoid. The captain tried to overcome the leftward drift by turning the rudder to the right. It made negligible difference.
In its interim report, and in a statement issued last month after Chauve submitted his dossier to the judge, the BEA said that the leftwards “yaw” was caused not by the faulty landing gear but by “the loss of thrust from engines one and two”.
There are several problems with this analysis. First, as the BEA’s own published data reveals, the thrust from engine one was almost normal until the end of the skid, when it took in the parts of the landing light. It is simply not true that the yaw began when both engines failed.
Second, those who fly the plane say that a loss of engine power will not cause an uncontrollable yaw. Concorde’s engines, unlike, say, a 747, are not mounted out near the wingtips but close to the tail and fuselage. The Observer has spoken to five former and serving Concorde captains and flying officers. All have experienced the loss of an engine shortly before takeoff in the computerised Concorde training simulator; one of them, twice, has done so for real. All agree, in Hutchinson’s words, that “it’s no big deal at all. You’re not using anything like the full amount of rudder to keep the plane straight; the yaw is totally containable.”
Finally, there are the skid marks. Last weekend the BEA claimed the plane had not skidded at all. Its chief representative, Helen Bastianelli, confirmed that the wheels were “not in a symmetric trajectory” in other words out of alignment when the Concorde took off, the first time the BEA has made this crucial disclosure. But despite the photographic evidence, she still insisted the yaw was caused by engine failure on the grounds that photos or no photos, there was no sign of skid marks.
Later I was telephoned by Philip Swan, an Englishman who works for the BEA in Le Bourget the airstrip near Gonesse where Marty was hoping to try to land. He accepted that the pictures did show the marks of four tyres, and that they were clearly out of alignment. But he concluded: “The photograph does depict that the tyres of the aircraft made sideways marks under stress. But I prefer not to use the term skid.” Think of your car, resisting your attempts to drive it straight, lurching off to the left. The BEA would say those black smears it leaves behind on the road are not evidence of a skid, merely the marks of the wheels moving sideways under stress.
The fact that Marty had to rotate his plane 11 knots below its stipulated rotation velocity was always going to make it difficult to save. In the event, he never got close to “V2”, the 220-knot airspeed that would have represented stable flight. For a few seconds in the agonising minute between takeoff and catastrophe he got up to about 210 knots, only for the number one engine which had begun to recover to fail for a second time.
But despite everything already against him the skewed bogey; the fire other avoidable factors were making it still more difficult to rescue the plane. When Marty paused at the start of the runway, his instruments told him that his Concorde had 1,2 tonnes of extra fuel which should have been burnt during the taxi.
In addition, it contained 19 bags of luggage that were not included on the manifest, and had been loaded at the last minute, weighing a further 500kg. These took the total mass to about 186 tonnes a tonne above the aircraft’s certified “maximum structural weight” the weight its physical components were designed and tested to carry in safety.
Meanwhile, in the interval between Concorde’s leaving the terminal and reaching the start of the runway, something very important had changed: the wind. It had been still. Now, as the control tower told Marty, he had an eight knot tailwind. The first thing pilots learn is that one takes off against the wind. Yet as the voice record makes clear, Marty and his crew seemed not to react to this information at all.
Had they paused for a moment, they might have recomputed the data on which they had planned their takeoff. If they had, they would have learned a very worrying fact. Flying a tonne over maximum structural weight was theoretically unlawful, but was not an outrageous risk.
However, more important than this measure is what pilots call the RTOW, the regulated takeoff weight: a limit set according to detailed tables for a given plane in the conditions obtaining at a particular time and place. As Marty released the brake with the eight-knot wind behind him, the tailwind meant that Concorde’s RTOW was just 180 tonnes six tonnes less than the weight of Flight 4590.
Hutchinson said: “The change in the wind was an incredible revelation, and no one says anything. Marty should have done the sums and told the tower, ‘Hang on, we’ve got to redo our calculations.’” Once he realised how far he was above the RTOW, he should have insisted on taxiing back to the other end of the runway and taken off against the wind. “If I’m honest, I’ve probably taken off 30, 40kg overweight after all, you can never be sure because you don’t weigh the passengers or the hand baggage. But not six tonnes! They were already at the limits of the envelope. Once the wind changed, they were beyond it.”
The extra weight had a further consequence beyond simply making it harder to get into the air. It shifted the centre of gravity backwards: the extra bags almost certainly went into the rear hold, and all the extra fuel was in the rearmost tank. A plane’s centre of gravity is expressed as a percentage: so many per cent “fore” or “aft”.
Brian Trubshaw and John Cochrane, Concorde’s two test pilots when the aircraft was being developed in the 1970s, set the aft operating limit at 54% beyond that, they found, it risked becoming uncontrollable, likely to rear up backwards and crash, exactly as Flight 4590 did in its final moments over Gonesse.
The doomed plane’s centre of gravity went beyond 54%. The BEA states a figure of 54,2%. A senior industry source says the true figure may have been worse: with the extra fuel and bags, it may have been up to 54,6%.
He said: “This is very significant. Even in a takeoff with all four engines working normally, you are well beyond the point where the test pilots would have been prepared to tread.” And as the fuel gushed from the hole in the forward number five tank, the centre of gravity moved still further back. Once again, Air France and Marty had closed off their options.
Marty’s crew was to do so one final time. When the plane was just 7,5m off the ground, the flight engineer shut down the ailing number two engine. Both French and British pilots say it was another disastrous mistake, which breached all set procedures. The engine itself was not on fire, and as the tank emptied and the fire burned itself out, it would probably have recovered. The fixed drill for shutting down an engine requires the crew to wait until the flight is stable at 1 200m, and to do so then only on a set of commands from the captain.
In a comment that might be applied to the whole tragedy, Hutchinson said: “Discipline had broken down. The captain doesn’t know what’s happening; the co-pilot doesn’t know; it’s a shambles. Once you deviate from rules and procedures, it’s chaos.”
Previous reports of the tragedy have described the crash as an “act of God”, a freak occurrence which exposed a fatal structural weakness in the aircraft that could have appeared at any time. Following the lead from the BEA, the international media have looked at previous tyre blowouts, suggesting that any one of them could have set in motion an ineluctable chain reaction; they have taken as a given the proposition that once the fire began, and hot gases interfered with the running of the left-hand engines, then the disaster that followed was inevitable.
The investigation by The Observer suggests the truth may not only be more complicated, but also sadder, more sordid. Men, not God, caused Concorde to crash, and their omissions and errors may have turned an escapable mishap to catastrophe.