The Space Shuttle Main Engines operated for around 8.5 minutes in every Shuttle launch, yet were removed from the orbiter after each flight and went through an overhaul, parts often being traded from engine to engine. This is appalling compared to even the most primitive turbojets (later-model Junkers Jumo 004s made of aluminium-coated iron had a MTBO of 5 hours at absolute minimum) and also to many other rocket engines, even reusable ones—for examples, the engine of the Breeze-M burns for 50 minutes, the reusable XLR99 had a MTBO of >1 hour, and (though not really comparable due to its pressure-fed radiatively-cooled nature) the Shuttle's own OMS was rated for 1,000 restarts and 15 hours of run time.

Was this removal and overhaul always truly necessary? If not, why wasn't there any changes to ground processing to eliminate overhauls on some flights, and if so, why wasn't there any serious work done on actually making the engines more reusable given their purpose?

Supposedly, improvements to service life were made in Phase I and Block IIA, but as they didn't result in the un-overhauled reflight of an engine, they were objectively negligible.

  • 3
    $\begingroup$ You have to consider that some people here have worked on the Shuttle, so the more objective your wording is and the less indignant it is, the better your chances are of getting a well-written and authoritative answer. Welcome to Space! $\endgroup$
    – uhoh
    Dec 20, 2019 at 2:11

2 Answers 2


No Space Shuttle questions are simple, since it was a long program with many different drives beyond engineering perfection*, so the short answer is 'yes, given money, time and perfect management/direction a reflyable engine could have been built'. The longer answer is more complex.

The space shuttle payload was around 22 tonnes, each engine weighed 3.5, if your changes to allow the engine to fly 2-3 times double the weight you add 16.5 tonnes leaving a payload of 5.5 tonnes and needs to fly four times as often for same capavity, so service numbers per year are similar or higher and you are paying for many more launches (and placing crew at more risk).

Obviously real world getting two launches out of an engine would not double the weight, since many parts are structural and do not 'wear out' just by holding loads but the overall math of 'we need to get x tonnes to orbit per year, how do we do that most effectively' holds true where each launch/fly/recover cycle has a fixed cost from people and equipment.

A highly relevant question is, which suggests that among other things the seals on the pumps were a key item for service, which makes sense since they operate in a chemically hostile environment and suffer mechanical wear. They also tie directly to performance, making them bigger/thicker/tougher may directly impact pump throughput and therefore thrust, rather than just weight.

Since the pump internals needed to be accessed anyway there is little gain in making other parts of the engine longer lifed, and if you find during service that a part can survive multiple flights the question becomes 'hey maybe make this lighter and less robust, we can get another kg to orbit'. At >$10k per kg to orbit the naive math is that lighter (possibly less exotic/cheaper) parts are a good deal.

In our current timeline the engines impacted a number of flights and the loss of Challenger and Colombia placed risk management under close scrutiny. A decision to spend more money to change something that already worked in a way that reduced safety checks and reduced payload would have been a hard political sell even if the pure math made it perfectly sensible.

In a different timeline where the original program goals of being able to turn around an entire orbiter for reflight succeeded trading a percentage of payload for longer engine service intervals makes perfect sense, and is why for example airliners do not have afterburners or similar. As soon as 'something' needs intrusive work every flight then being able to refly all related systems becomes less useful and trading ground service for payload becomes a valid (inherently expensive) choice .


*looking around in 2019 it appears the answer for 'how to make better shuttle engines' is to not send them all the way to orbit, and delete the crew.


The SSME is an exceptionally complex engine, producing a very large amount of thrust from a compact package. It's working much closer to the margins of what is possible than any of your points of comparison. This was largely a result of a very ambitious specification; in order to reach the required thrust-to-weight ratio and high specific impulse, it had to run very close to its mechanical limits, whereas less ambitious engines can run under less stress. Imagine two identical car engines, one running at 3000 rpm, the other at 6000 rpm. One's definitely going to win the race, but you're going to want to inspect the engine afterward.

According to Wikipedia:

since NASA was interested in pushing the state of the art in every way they decided to select a much more advanced design [for the SSME] in order to "force an advancement of rocket engine technology".

In retrospect, NASA may have gone too far with this strategy; SSME development was challenging and expensive, and as you note, the engine required more maintenance than was ideal for a reusable launcher. Unfortunately, the size, mass, and thrust of the engines was fixed in the specification before it was known how hard it would be to build and operate them. A less "bleeding-edge" engine design would have cut severely into the space shuttle's payload.

In some alternate universe, NASA somehow got a budget to develop the space shuttle without having to cave to the Air Force's requirements; this one has a smaller payload bay and stubby wings like the X-37, and maybe it would have used a simpler engine that didn't need as much maintenance between flights.

This is appalling compared to even the most primitive turbojets

The most powerful modern high-bypass turbofans in the GE90 family produce around 500kN of thrust from an 8.7 ton engine; SSME produces 1890 kN at sea level from a 3.5 ton engine. The GE90s rotors run at 2355 rpm and 9332 rpm; the SSME has four turbopumps; the slowest turns at 5150 rpm and the fastest at over 35000 rpm -- almost 600 revolutions per second.

the engine of the Breeze-M burns for 50 minutes

SSME produces 100 times the thrust of the Briz-M's engine. It has a lot more heating to contend with. Over the course of its 8.5-minute burn it moves about 10 times as much mass through its pumps as Briz would on a 50-minute burn. Briz-M's engine has a thrust-to-weight ratio of around 27:1; SSME around 73:1.

(though not really comparable due to its pressure-fed radiatively-cooled nature) the Shuttle's own OMS was rated for 1,000 restarts and 15 hours of run time.

OMS uses a remarkably simple engine. No regenerative cooling, no pumps, as you note; as long as the chamber remains intact, the only points of concern are the valves, which operate under relatively low pressure. I don't know how much work was done on these between flights; it would probably be possible to swap out the valves for brand new ones every time at a tiny fraction of the cost of the SSME inspections/overhauls.

the reusable XLR99 had a MTBO of >1 hour

Again, a fairly modest engine that didn't have to achieve anywhere near the performance of the SSME.

Was this removal and overhaul always truly necessary?

This, I don't actually know.

  • $\begingroup$ This makes me wonder actually how the SSME compares to the engines being designed for the Starship upper stage. $\endgroup$
    – ikrase
    Dec 22, 2019 at 2:05
  • 1
    $\begingroup$ @ikrase I'm really not sure. Raptor uses a full flow staged cycle design, which is complex; it's going to be running at a very high chamber pressure, and it's supposed to outperform SSME in thrust and thrust:weight. On the other hand, no hydrogen plumbing to deal with, and the state of the art in computational fluid dynamics simulation has vastly changed since the SSME was developed, so SpaceX may be able to meet those challenges more efficiently than was possible in the 1970s. $\endgroup$ Dec 22, 2019 at 2:31
  • $\begingroup$ huh, didn't know it outperformed the SSME. Massive number of engines also probably means failures are more tolerable. $\endgroup$
    – ikrase
    Jan 2, 2020 at 5:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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