J. Brit. Astron. Assoc., 106, 4, 1996, pp.179–181

The fiery death of Ariane 5

Ten years of work and the hopes of hundreds of scientists lay scattered across the South American rainforest after the explosion of Ariane 501. Paul Murdin was there.

A grey dawn was breaking on Tuesday 1996 June 4 as I approached the French Guiana space centre. It revealed a damp jungle of dark mangrove swamps beneath ragged, broken clouds, and a higher ceiling of more solid-looking cloud above. Within the high security confines of the European Spaceport near Kourou were the flags of the 14 member states of the European Space Agency (including the UK which contributes 8% of the cost of the spaceport), scattered tall buildings, and the floodlit Ariane 501, a fat central rocket with the main engine and payload, flanked by two thinner solid fuel boosters to provide extra thrust for the first two minutes. The payload consisted of four spacecraft named Cluster. These were of special concern to the British scientists at the launch, because it was Cluster to which they had contributed their expertise and from which they expected scientific data.

Cluster was half of the Solar-Terrestrial Physics Cornerstone of the ESA Science Programme. The other half, SOHO (Solar and Heliospheric Observatory) had been successfully launched six months before and is already on station observing the Sun. SOHO is looking at the whole structure of the Sun, from its deepest interior to its surface and into the atmosphere above. In particular SOHO was to look for solar storms gathering on the surface of the Sun, and study how they were ejected into space.

Solar storms – clouds of gaseous plasma – traverse space across the solar system, sometimes towards our Earth. They wrap around us, mostly deflected from the Earth's surface by the Earth's magnetic field. The clouds of plasma drag back the magnetic field like a comet's tail. Some plasma leaks down to the magnetic poles where its impact on Earth's atmosphere makes the aurora. The four Cluster spacecraft were to fly in formation through all this activity, determining the three dimensional structure of the magnetic field and how it changes in the fourth dimension of time. Four dimensions and four spacecraft – Cluster was an exactly defined mission which was intended definitively to solve the problems of the Earth's magnetosphere. Together with SOHO, Cluster would be a major attack on fundamental problems in space science.

The launch of Cluster on the first flight of Ariane 5 was a calculated risk: no one starts off wanting to be on the maiden flight of a new rocket. Although it has a similar name to the previous four generations of Ariane, Ariane 5 is a very new rocket, with new technology. But none of the previous Arianes were powerful enough to deliver the four Cluster spacecraft where they needed to go. There were Russian and American alternatives, of course, but ESA has a 'European launcher policy' – to give the European rocket industry the best chance against the international competition. The Science Programme in ESA is supposed to stretch the European space industry and one of the ways it can do so is to challenge European launcher manufacturers. Also, the launch was free – that is, paid for by ESA's Ariane Development Programme, to prove Ariane 5 and publicise it. The less you pay for a rocket launch, the more science payload you can buy for a given amount of money, so the free launch was tempting for the scientists. Without a free launch, Cluster would have been too expensive. Of course, there is no such thing as a free launch.

However, there was good reason to be confident about the first launch of Ariane 5. Although a new design, it was a design originally intended for manned space flight. This was why Britain had declined at the outset to join the Ariane 5 programme. When the human factor was postponed, Britain joined in last year. The safety features built in from the start of the project were still based on human-qualified space flight; the test programme was rigorous, and conditions for testing were as realistic as possible. The success factor is 98.5%. Those who made the decision to put Cluster on Ariane 501 still think they were right: it is only with hindsight that a calculated risk might look like a gamble.

The final minutes of an Ariane 5 launch are automatically controlled. Of course, a manual override is possible, but there was no indication of any problem in the countdown to engine ignition. As T=0 approached, the audience in the control room moved rapidly out to balconies, with mounting excitement. At a distance of 8km, we saw the nose-cone of the rocket above the trees, and the first glow of the engines. There were squeals, oohs, scattered applause and cheers as Ariane 501 lifted off the launch pad. Its three engines, the main Vulcain engine in the central stage, two solid propellant boosters either side, shone so brightly in the sky, far brighter than I had imagined, glorious and promising. The lift off seemed so measured, so deliberate. It was hard to reconcile the motion which we saw with a speed of many times the speed of sound. Smoke and fumes trailed back to the ground as the rocket passed up through the broken lower cloud layer. The rocket slowly rolled a quarter turn around its axis, presenting its correct aspect to the telemetry receivers.

But 36 seconds into the flight, the rocket was at an altitude of 4km when disaster struck, and the bright glory turned to a brighter catastrophe. According to a later analysis, the onboard control computer decided, for reasons which are not known as I write this account the day afterwards, that the rocket was off course. The possibility that the rocket will depart from its nominated course, for example due to an unexpected gust of wind, or an imbalance in the solid fuel boosters, is programmed in to the computer. It responds by deflecting the nozzles on the boosters, and in extreme circumstances on the main Vulcain engine. This action, like the action of an outboard motor on a boat, pushes the back end of the rocket sideways, to nudge it back on course. The computer senses the course of the rocket from data fed from two, duplicated inertial guidance systems.

According to the telemetry direct from the rocket, the records from which are the rocketry equivalent of a crashed aircraft's 'black box', the sensors showed the course of the rocket was spot on. But, nevertheless, both inertial guidance systems declared themselves malfunctioning, and the control computer decided that the rocket was very badly off course. It thought the top of the rocket was leaning to the left. It sent commands to push both nozzles of the boosters to maximum deflection, shoving the back of the rocket to the left to position it under where it thought the top was. The Vulcain engine independently decided to give a helping push, because this manoeuvre with the boosters seemed not to work – according to the computer the rocket was still leaning to the left. The Vulcain engine swivelled too. In fact, the course of the rocket was true, and the well-intentioned correction efforts of the three engines shoved the back of the rocket well to the left, out from under the top.

The rocket was now crabbing upwards and to the right, its engines twisting it sideways. The computer displays in the control room showed it veering off track. 'We are going to lose this one,' remarked one ESA official in a matter of fact tone. Even we unknowledgeable spectators could see it veer sideways. Rockets are just not made to take this kind of manoeuvre. Rockets are cylindrical, like drinking straws, and have their greatest strength against a force along their length. Apply a sideways force and they snap. This is what happened to Ariane 501. Its upper stages broke into a few large pieces. Moments later, the on-board safety mechanism destroyed the big pieces with explosions. The range safety officer, sitting isolated in his lonely glass booth, also manually sent a destruction command, so close to the explosion that he thought he had been the one who had had to do it, as ESA at first announced. A shower of smaller glowing fragments sprayed into the sky, trailing smoke and burning fuel. All around me most spectators fell silent. Two French rocket engineers swore – 'Merde, merde, merde!' A woman sobbed.

Most fragments were lobbed forwards from the explosion point, and fell to Earth across 15km of mangrove swamp between the launch pad and the sea. Some pieces fell within 500 metres of the launch pad and the launch control personnel. They were in a concrete bunker, protected by a roof three metres thick, able to survive practically anything. Three kilometres away a group of spectators, including British scientists from the University of Sheffield, Imperial College and the Rutherford Appleton Laboratory, were hurried onto their buses, donning gas masks and rapidly driven out of the area to safety. Amongst them was the widow of one of the British scientists responsible for the manufacture of some of the instrumentation in Cluster. Her husband died some weeks before the launch, and she had come to Kourou to see the culmination of his work. Sadly, she had to endure its destruction.

A TV transmission of the launch had been in progress in the room behind the balconies, and as the rocket fired into the sky we heard the roar of the takeoff on the loudspeakers, curtailed by the explosion. There was a silence, made the more intense by the sobs. To my utter astonishment, the launch replayed itself in direct sound. The sound of the launch had taken the time of the short flight to cross to us from the launch pad. As the fiery pieces of the destroyed rocket drifted down from the sky, we heard the takeoff again, a ghostly invisible rocket rising in a crescendo into the sky towards its already manifest doom, which in direct sound was signalled by a sharp crack.

We were ushered inside the control room and the doors were closed against the possibility of drifting fumes of hydrochloric acid. I sat stunned, thinking about what I had seen. The energy of the rocket had been made visible to us in an uncontrolled way, and that brought home the power and complexity of the task that is a space mission. The science loss is of course unique. Cluster died on June 4 in the explosion of Ariane 501. It is very doubtful if it will be attempted again, at least in the same form. There will be many more Ariane launches, which will build on this experience, the results of the ESA Board of Enquiry (which Britain will aid by supplying expertise on robust software) and the investigations being carried out by the prime contractors for the rocket control system, Aerospatiale and Matra Marconi. Indeed the success of Ariane 5 is essential to the European science programme, ESA's plans to participate in the Space Station, and the dream, particularly strong in France, of manned European space flight.

So I feel sorry for the rocket engineers who saw their creation explode, but they will get other chances. I feel sorrier for the scientists who had given ten, twenty years to Cluster, now with little to show for it. I feel sorriest for the young people who were to have worked on Cluster data, whose careers have suffered a serious blow. Already they are thinking about how to capitalise on the preparatory work that has been done, how to use the supplementary equipment without Cluster, and the definition of a replacement project, already optimistically named Phoenix.

I am confident that the setbacks which are the consequence of the ill-fated Ariane 501 will be temporary. The scientists of Europe are beginning metaphorically to pick up the pieces from the Ariane 501 disaster and its forlorn science mission, just as the spaceport technical staff and Foreign Legionnaires in French Guiana were literally picking up the pieces, as I left the rocket range on Wednesday morning.


Paul Murdin, who joined the BAA in 1957, is head of astronomy at the Particle Physics and Astronomy Research Council (PPARC). Parts of this article appeared in the Guardian and New Scientist.


Back to top of page

Return to Journal 1996 August contents page