The Centaur upper stage, the first hydrolox rocket stage ever flown and (in its highly-evolved forms) still one of the most-used, as well as one of the highest-performance (if not the highest-performance) upper stages available, was used from the 1960s onward, as part of the Atlas-Centaur, to launch heavy probes to the Moon and to other planets, and to launch very heavy satellites into Earth orbit; in the 1970s, it was also used, as part of the Titan-Centaur, to launch six even heavier probes1 away from the Earth.

In the early 1980s, with the advent of the space shuttle, a new version of the Centaur was developed to launch heavy spacecraft from a shuttle orbiter’s payload bay; known as the Shuttle-Centaur, or, technically, as the Centaur G (for the Air Force) and Centaur G-prime (for NASA), its first missions were to be to launch the Galileo and Ulysses probes (to study Jupiter and the polar regions of the sun, respectively) in May 1986, with the Magellan Venus orbiter following at a later date. The shuttle’s first G-series Centaur was rolled out in August 1985, followed by a second shortly afterwards, and, by late January 1986, the two Centaurs and the two probes were at KSC and already mated to their payload adapters in preparation for being mounted in their orbiters’ payload bays.

However, when the shuttle orbiter OV-099 Challenger was destroyed in a fatal launch failure on the 28th of that month, it sounded the death knell for the Shuttle-Centaur programme, as the high-profile deaths of the seven crewmembers onboard caused NASA and its astronauts to suddenly become far more risk-averse, eliminating any realistic chance of a Centaur flying on board a shuttle (notably, following the loss of Challenger, a large fraction of the corps of shuttle astronauts flat-out refused to fly with a Centaur under any circumstances whatsoever) even once the shuttle’s grounding was lifted;2 although the Shuttle-Centaur was initially able to linger on, work towards the Galileo and Ulysses launches was ordered stopped in February, and the entire Shuttle-Centaur programme was cancelled in June.

At the time of the Challenger launch failure, there was still a ready supply of Atlas-Centaurs3 which could have been used to launch Galileo and Ulysses, potentially even still being able to make the May launch windows had the two probes immediately been transferred from Shuttle-Centaur to Atlas-Centaur. Instead, however, Galileo and Ulysses, which remained slated for shuttle launches, had to wait until the shuttles’ return to flight, and, without the Centaur, when they finally did fly (in October 1989 and October 1990, respectively), they had to make do with the anemic Inertial Upper Stage (IUS; actually two stages), cobbled together from two solid-fuel kickmotors originally designed to propel satellites no further than GTO.4

This was especially bad for Galileo; as even the two-stage IUS had less than two-fifths the oomph of a Centaur, the spacecraft had to get three gravity assists from Venus and Earth before it could get to Jupiter, taking a total of six years to arrive at its destination, whereas the Centaur would have been easily capable of boosting it directly to Jupiter in only two years.5 Moreover, in the three years Galileo sat warehoused to await its ride (and exacerbated by its two coast-to-coast journeys as it was shipped back to JPL in 1986 and then once again to KSC in 1989, both times by truck), the lubricant for its high-gain antenna deployment mechanism broke down, with the result that, when ground control tried to extend the antenna in 1991, it didn’t; switching it to a Centaur on an expendable rocket when the shuttles were grounded and launching it during the original May 1986 launch window would have eliminated the wait and cross-country travel that together allowed the lubricant to deteriorate, in addition to getting Galileo to Jupiter seven years earlier (in 1988 rather than 1995).

Why were Galileo and Ulysses made to wait for the shuttles to return to flight and then be launched from the orbiters using a mediocre propulsion system, rather than simply transferring them from a Shuttle-Centaur to an Atlas-Centaur?

1: Two each of the Helios, Viking, and Voyager probes, to be specific.

2: Which, as it turned out, took the better part of two years.

3: For that matter, there still is a large ready supply, albeit of a much-evolved and much-more-capable version from which the Centaur is practically the only part still recognisable.

4: Magellan also used an IUS to throw it at Venus in May 1989; ironically, what had originally been planned to be the third Shuttle-deployed interplanetary probe ended up actually being the first to launch!

5: Ulysses, which also had to go to Jupiter (for gravity-assist purposes), instead stuck yet another kickmotor on top of the IUS.

  • 9
    $\begingroup$ Um, so... why did astronauts refuse to fly with a Centaur on board after the Challenger accident, when they were presumably okay with it before the accident, and the Centaur had nothing to do with the accident? $\endgroup$
    – Greg
    Jun 1, 2019 at 23:36
  • 11
    $\begingroup$ @Greg The post-Challenger investigation shed a lot of light on some uncomfortable truths about how the shuttle program was being managed. In addition, I imagine that a lot of the astronauts had concerns about Shuttle-Centaur prior to Challenger, and simply got more openly vocal about it afterward. $\endgroup$ Jun 1, 2019 at 23:39
  • 8
    $\begingroup$ @Greg they weren't OK with it before the Challenger failure. It got the name Death Star before then. $\endgroup$ Jun 2, 2019 at 1:14
  • 4
    $\begingroup$ @OrganicMarble Were there specific safety concerns beyond the generic "we have lots of extra boom-stuff on board"? $\endgroup$ Jun 2, 2019 at 6:29
  • 6
    $\begingroup$ @chrylis Normally, If a spacecraft has a tankage leak, it’s venting to space and dispersing. If a shuttle Centaur leaked during ascent (a rough ride, keep in mind), you could get gaseous hydrogen + oxygen accumulating in the payload bay. On any given flight the odds of springing a leak in both the LH and LOX tanks would be extremely small, of course... $\endgroup$ Jun 2, 2019 at 6:50

2 Answers 2


The performance of Shuttle-Centaur would have greatly exceeded that of either the Atlas-Centaur or Titan-Centaur combination.

Neither the Atlas nor Titan were able to put a fully fueled Centaur into Earth orbit using only their lower stages. The Centaur would have burned part of its propellant completing the orbital insertion.

In contrast, the shuttle stack would reach orbit before deploying a completely fresh Centaur.

In addition, the Centaur-G designed for the Shuttle-Centaur had a larger hydrogen tank than the Centaur-D being used on the contemporary expendable launchers. It would have carried 21.1 tons of propellant versus 13.6 tons (according to Astronautix) with only 144kg increase in dry mass -- a big difference to mass ratio, yielding a difference of about 1400 m/s ∆v for a Galileo-sized payload.

According to the payload graphs, Titan IIIE-Centaur can get about 1100kg to Jupiter, maybe 1400kg with a Star-37 kick motor, well short of Galileo's 2560kg.

Atlas-Centaur can do about 750kg to Jupiter with the addition of a Star-37 kick motor.

  • $\begingroup$ Ahh, that makes sense. :-/ $\endgroup$
    – Vikki
    Jun 1, 2019 at 23:17
  • 1
    $\begingroup$ Too bad we couldn't borrow an Energia booster to put a Centaur G in orbit (or nearly so). $\endgroup$
    – Joshua
    May 10, 2020 at 22:42

According to the Galileo Wikipedia article under Reconsideration-Paragraph 6 (https://en.wikipedia.org/wiki/Galileo_(spacecraft)#Reconsideration), there was consideration that Galileo could of been launched by a Air Force Titan IV-Centaur G in May 1991, but the AirForce could not provide NASA a Titan rocket due to the backlog of Department of Defense Payload.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.