IN THE aftermath of the accident that destroyed Ariane 5, the question on
everybody鈥檚 lips is: what went wrong? Rumours from the launch site in Kourou,
French Guiana, suggest that a simple bug in the rocket鈥檚 onboard software was to
blame.
The bald facts of the rocket鈥檚 demise are already clear. All was well until
37 seconds into the flight. With the rocket 3500 metres above the ground and
travelling at 70 per cent of the speed of sound, it flipped on its side, snapped
in two and began to disintegrate. Within two seconds, the automatic onboard
self-destruct system had blown up the remains. Apparently unaware of this,
flight controllers sent two self-destruct signals at 66 and 72 seconds after
liftoff.
The disaster followed 6 months of delays to a programme in which ESA and the
French space agency CNES had invested $7 billion鈥�20 per cent more
than the allotted budget. The first launch was originally planned for October
last year, but problems with software used by ground-control computers stalled
the programme for several months. Mechanical problems followed and as late as
May engineers were adding extra thermal protection to parts of the rocket and
changing faulty valves. 鈥淭hese are run-of-the-mill problems that happen the
first time you launch a rocket,鈥� says a spokesman for Aerospatiale, the French
company that coordinated much of the work.
Advertisement
More worrying was the discovery during a routine test three weeks before the
launch that the electrical wires controlling the ignition and jettison of the
solid fuel rocket boosters had been mislabelled, which could have led to the
boosters being jettisoned, rather then ignited, on the launchpad. ESA officials
are playing down the significance of this incident. 鈥淭he problem was solved in
no time,鈥� says one.
The focus of ESA鈥檚 investigation, which will report by 15 July, is why Ariane
5 suddenly veered off course. The rocket is steered using its exhaust nozzles,
which are designed to swivel. Small changes in the angle of these nozzles alter
the position of the rocket鈥檚 rear end to compensate for sudden gusts of
wind.
Preliminary analysis of the flight data has shown that the hardware鈥攖he
main engine and the solid fuel rocket boosters鈥攑erformed perfectly. This
conclusion comes as a relief for ESA, since a major problem with the hardware
would be hugely expensive to fix. The inquiry is now focusing on the launcher鈥檚
electrical and software systems.
The first suspect was the inertial guidance system which senses the rocket鈥檚
position and attitude. But the raw data from these units, which are broadcast
directly to ground control, suggest that they were functioning correctly.
However, the rocket鈥檚 onboard computer evidently had other ideas. At the instant
before the accident, the computer believed that the forces on the rocket had
risen to 240 g. 鈥淭hey should not be more than 10 g,鈥� says Paul
Murdin, head of astronomy at the Particle Physics and Astronomy Research
Council, who witnessed the accident. This points to two further possible causes:
faults in the rocket鈥檚 electrical wiring or its onboard software.
Last week, staff at Kourou reconstructed the entire launch using the flight
data. Sources at the space centre say that this rerun suggests that the accident
was caused by a bug in the software run by the onboard computer. A similar
problem is thought to have caused the failure of a Chinese Long March rocket
launch in February.
鈥淚鈥檝e seen no data to confirm this, but it would be very sad if a simple
error caused the disaster,鈥� says Ivan 脰fverholm, president of Saab Ericsson
Space in Gothenburg, Sweden, the company that supplied the onboard computer and
some of its software. He says that, prior to the launch, the computer had been
tested in isolation and linked up to the rest of the launcher.