EVEN rocket science isn鈥檛 rocket science any more. The number of launch
vehicles goes up and up, but somehow the technology isn鈥檛 keeping pace. A
devastating analysis from French insurance firm AGF says that new rocket designs
are less reliable than they were 20 years ago, when commercial satellite
launchers first became big business.
Satellite television, satellite telephones and wireless Internet all add up
to a burgeoning demand for space launches. The number of different types of
rocket available on the multibillion-dollar commercial launch market has
quadrupled over the past two decades. But despite advances in propulsion
technology and the increasing use of computerised flight control and guidance
systems, rocket builders don鈥檛 appear to be any better at getting their new
craft safely aloft.
In fact, it鈥檚 these computer systems that many industry insiders say should
take much of the blame for the problem, along with the need to carry heavier
payloads. The rigorous drive to cut costs is another factor鈥攖hough that
could prove counterproductive if the failures continue.
Advertisement
As if to underline AGF鈥檚 findings, woeful news emerged last month from NASDA,
one of Japan鈥檚 two space agencies. The American satellite maker Hughes has just
cancelled an order with NASDA for 10 launches, following repeated failures of
its heavy-lifting rocket.
AGF鈥檚 survey set out to check how successful new rockets were in their first
few launches鈥攖he benchmark measure for new designs. It found that during
the first five launches of a new type of rocket between 1980 and 1984, 60 per
cent experienced at least one failure. And AGF鈥檚 Guy Lallour points out that in
the past five years almost two-thirds of new launches experienced at least one
failure鈥攁 worse record than 20 years ago.
鈥淣ew launchers are not getting any more reliable because the nature of
satellite missions is changing,鈥� says Lallour. 鈥淭hey are launching much heavier
satellites that make more demands on the launcher.鈥� The practice of
鈥渃o-locating鈥� a number of satellites in the same orbits鈥攕o that their
signals can all be picked up by a single TV dish, for example鈥攊s also
technically challenging. 鈥淎nd these come at a time when the market is
increasingly commercial, with big budget and time constraints.鈥�
AGF鈥檚 analysis is centred on new rocket designs. You might think this means
that its figures are of academic interest only: after all, even launchers that
get off to a shaky start generally get it right in the end. But the commercial
world doesn鈥檛 work like that. Currently, satellite makers demand flat rates
across all launcher types, says Lallour. But pressure from insurers is set to
change this. Individual rocket manufacturers will have to ensure their products
compete keenly in terms of reliability. Failure to improve the reliability of
early flights will send future insurance premiums鈥攁nd therefore the total
cost of the satellites they insure鈥攊nto orbit, and their commercial
ambitions crashing to Earth. 鈥淭he reliability of launchers has to be reflected
in the insurance rates in future,鈥� says Lallour. 鈥淭he price has to be
commensurate with the risk.鈥�
The new rocket designs now coming into commercial service include Lockheed
Martin鈥檚 Atlas 3 and Boeing鈥檚 Delta 3 from the US, Ariane 5 from 鈥淓urope鈥檚
Arianespace, and the Proton M from Khrunichev of Russia. The last three all
failed on their maiden flights. People are starting to ask why rocket makers
haven鈥檛 smoothed out technological difficulties and driven failure rates down
for new designs, says Space News. Lallour puts the blame for this on
cost cutting. 鈥淭he budget constraints mean there is a lot less actual testing of
components being done, with more being done in simulation,鈥� he says.
Pierre-Eric Lys, AGF鈥檚 manager of space operations insurance, says ideas like
NASA鈥檚 鈥淔aster, Better, Cheaper鈥� programme鈥攔ecently criticised in an
independent report on the failures that have dogged its Mars missions鈥攁re
part of the problem. 鈥淒esign-to-cost programmes like that mean people have to
stick to budgets no matter what technical problems come up,鈥� Lys points out.
Not surprisingly, the launchers鈥� manufacturers see things differently. 鈥淚
don鈥檛 agree with AGF鈥檚 analysis. They are looking probably at our competitors,鈥�
says Mario Delpine of Arianespace. 鈥淎riane 4 has had 54 successful launches in a
row鈥攚e stand at a success rate of 97.5 per cent, which is the best record
on the commercial launch market today.鈥�
But while Ariane 4 is undoubtedly a solid performer, the history of
Arianespace鈥檚 heavy-lifting rocket Ariane 5 illustrates the problem highlighted
by AGF. Ariane 5 blew up on its maiden flight in 1996, destroying the European
Space Agency鈥檚 four-satellite Cluster mission. Its second flight, with a dummy
payload, failed to reach its intended orbit. The three flights since then have
been fine, however.
鈥淲ith a rocket, all your eggs are in that one flight and there鈥檚 always a
chance of a failure,鈥� says David Todd, who edits the SpaceTrak database.
SpaceTrak monitors launchers鈥� success rate, and is used by the companies that
insure satellite payloads. It is published by Airclaims, a London company that
advises insurers when they face claims from the aviation industry. 鈥淎 maiden
flight is a brand-new vehicle,鈥� says Todd. 鈥淚t鈥檚 not like an aircraft that鈥檚
been test flown in small altitude and speed steps.鈥� He says the industry should
be wary of using maiden flights for launches.
In an analysis of rockets鈥� maiden flights, Airclaims found that 12 out of 23
new designs that flew in the past decade鈥攏ot including more reliable
converted ballistic missiles鈥攕uffered failures first time round. And of
the nine failures that made a second attempt, six failed then too.
Computerisation of rocket guidance and control may be a key factor in these
failures, says Todd.
Poor design
The recent failure of a Zenit 3 rocket launched from Boeing鈥檚 Sea Launch
platform in the Pacific was due to what Todd dubs a 鈥減rogrammer medley鈥�. When
software engineers tried to 鈥渢idy up鈥� a computer program after two successful
flights, they accidentally introduced an error that left a flight-critical valve
open. Ariane 5鈥檚 maiden failure is another famous computer foul-up
(New 杏吧原创, 27 July 1996, p 10).
More than four decades after the Soviet Sputnik probe blazed a trail into
space, rocket makers have worked out how to make propulsion reliable. So is
software now scuppering launches? 鈥淕uidance systems do seem to go wrong more
than anything,鈥� says Todd.
But other types of fault may be to blame, too. Nowhere has the failure of
rocket technology been felt more keenly than in Japan. NASDA鈥檚 lost contract
with Hughes was worth $836 million. Yet in 1996, when the contract was
signed, it was seen as a vindication of Japan鈥檚 H 2 programme. Its cancellation
now is a body blow for Japan鈥檚 ambitions as a commercial satellite launcher.
Poor design and manufacture led to the H 2鈥檚 downfall, says Todd. Observers have
blamed things as varied as stray insects causing short circuits and bubbles in
liquid hydrogen fuel that fractured a turbine blade. Production of the main H 2
rocket was suspended last December after two failed launches in a row.
Development of its successor, the cheaper H 2A version, is also on hold.
Is this the end of the road for Japan? The European Space Agency had already
booked an experimental communications satellite called Artemis onto the maiden H
2A flight. It opted for the Japanese launcher instead of the twice-failed Boeing
Delta 3. Todd wonders why. 鈥淚鈥檇 choose the Delta 3,鈥� he says, 鈥渙n the basis of
third time lucky.鈥�
