I understand the philosophy of using relatively many smaller engines so that a single engine failure has minimal impact on the mission (provided you can keep the shrapnel from shredding the others) but 31 engines for the BFR seems an awful lot.

Nine engines for F9 seems reasonable enough, and they had an engine that size from the F1. 27 for FH is because it's really three rockets tied together, but there seems no obvious reason for 31 rather than (say) 9 on the BFR, given that it's both a new rocket and a new engine.

Anyone know what the story is? are bigger engines disproprtionately harder to design or maintain? Had they got too far through the engine design before the BFR design stabilised?


4 Answers 4


Anyone know what the story is?

The main driver for the large number of engines on the BFR first stage is the desire to use a common engine design (albeit with different optimized nozzles) for both the booster stage and the interplanetary spacecraft stage. Building and maintaining only one type of engine makes things more efficient, and is a strategy that has worked well for them in Falcon 9/Heavy. The upper stage uses 6 of them (in the September 2017 design), and that number is driven by redundancy and thrust-range requirements. If the upper stage is using 6, the much larger booster stage needs many more.

are bigger engines disproprtionately harder to design or maintain?

Large engines are generally harder to build, although they can be more efficient by mass. Combustion instability problems plagued development the F-1 engine used on the Saturn V rocket.

SpaceX seems to be good at building lightweight, powerful engines (the Merlin series is said to have the highest thrust-to-weight ratio of any liquid fueled rocket engine), so they don't need large engines to be efficient.

More engines means more possible failures, but as long as the failure modes are contained to a single engine, losing one of 31 engines is less of a problem than losing one of 9. SpaceX's engines appear to be generally reliable, so that doesn't seem to be a concern either.

  • 1
    $\begingroup$ chuckles But if one of the failure modes escapes the single engine, traveling horizontally at great speeds, the result can still be colorful =) $\endgroup$
    – Cort Ammon
    Commented Feb 9, 2018 at 20:10
  • 3
    $\begingroup$ @CortAmmon Yup. One Merlin went bang in a 2012 Falcon 9 1.0 flight, and didn't seem to hurt the neighbors despite being a very visible explosion, so I imagine they've given a lot of thought to failure containment. $\endgroup$ Commented Feb 9, 2018 at 23:24
  • 2
    $\begingroup$ I can’t find it now but I read somewhere that Falcon Heavy could lose 6 out of 27 engines and be fine, so we could estimate maybe 7 engines could fail on the BFR without jeopardizing the mission. $\endgroup$ Commented Feb 10, 2018 at 6:54
  • $\begingroup$ @ToddWilcox IIRC Musk said that in his post F9H press conference. $\endgroup$ Commented Feb 10, 2018 at 19:58
  • $\begingroup$ "maybe 7 engines could fail on the BFR" ASSUMING none of those 7 failures caused neighboring engines to fall. The BFR rocket engines have (at least) two high speed turbo pumps, an exploding fan disk could send high energy shrapnel in 360 degrees. $\endgroup$
    – user8269
    Commented Feb 11, 2018 at 7:55

Russell Borogove's answer gets to the root of the issue. They wanted minimum 6 engines on the BFS, which meant that the BFR had to have several times more, and they picked the number 31.

31 nozzles, however, is not a record. The Soyuz has 5 cores. The centre core has 4 main and 4 vernier nozzles, and the side cores have 4 main and 2 vernier nozzles each, making a total of 20 main nozzles + 12 vernier nozzles = 32 nozzles. That's a total of 32 nozzles and combustion chambers, but each core has only one set of turbopumps, so they are usually counted as one engine per core for a total of 5.

The Russians / Soviets have a history of making engines comprising a single turbopump feeding 4 main combustion chambers (and sometimes some vernier combustion chambers.) The reason they do this is to avoid the problems of combustion instability that come with very large combustion chambers. It probably also saves on tooling costs and on development costs.

The largest of these 4 combustion chamber designs is the RD170 / RD171M. Smaller 2 chambered and 1 chambered derivatives have been made. The Energiya Uragan was a proposed rocket comprising a central core surrounded by eight boosters each with one RD170 engine. Including the 4 nozzles on the central core, it would have had a total of 36 nozzles. Unfortunately this design coincided with the collapse of the Soviet union, and in the political and economic turmoil the space program was shelved.

The Saturn V's F1 engine (6770kN thrust at sea level) had by far the largest combustion chamber of any liquid fuel rocket engine, and was completely unthrottleable. The next largest combustion chamber is found on the RS-68 (single chamber, 3137kN thrust at sea level, used on the Delta IV.) That's about 4 and 2 times larger than the raptor (1700kN thrust at sea level) respectively

Therefore, attempting to develop a throttleable combustion chamber much larger than Raptor would have been a completely new engineering challenge, and not one Spacex was interested in - they are interested in optimising thrust to weight ratio, specific impulse, and cost.

SpaceX could have taken a similar route to the Russians, and equipped the BFR with eight sets of turbopumps supplying 4 combustion chambers each for a total of 32. This would have enabled them to reduce their nominal "engine count" and retain some commonality of parts with the BFS. It would not, however have reduced the amount of piping required compared to their chosen design.

SpaceX chose instead to have 31 completely independent engines for greater failure tolerance.

People often point to the failure of the 30-engined Soviet N-1 moon rocket as a reason not to use large numbers of engines. But this was a rushed design (ultimately cancelled by the Americans getting to the moon first) and while it did have some plumbing issues, the biggest problem was trying to control all those engines with a 1960's computer, which was as dumb as a modern microwave oven, and vastly more laborious to program!

Musk believes that with modern computers, a large number of independent engines is an advantage.

  • 4
    $\begingroup$ Yeah, the F1 was a beast and only had two modes, on and off. But dang, the on mode was impressive. (I lived near the Rocketdyne test area in Chatsworth and when they ran the whole world shook.) $\endgroup$
    – zeta-band
    Commented Feb 9, 2018 at 21:30

SpaceX has been developing a methane/LOX engine called Raptor which is about 2 times more powerful than the Merlin engine. (This is the first Raptor iteration at around 400Klbs thrust. The first Merlin (1A) was 75Klbs, current version is closer to 200Klbs. So expect performance growth).

31 engines at about 400Klbs thrust gives about 12.4 million lbs of thrust.

What are the existing alternatives? Some of the largest engines ever built (by thrust) were the F-1 (US) and RD-170 (Russia). Both are close to 1.5-1.8 million lbs thrust. (Actual value changes over time. There was 1.8mlbs F-1 version that never flew for example).

You would still need a cluster of such engines, 7 or more to achieve the same performance. Once you get to that point, the question becomes, how to optimize everything. There are likely many possible solutions, SpaceX has bet on one that fits their needs.

SpaceX also needs to balance landing, where the thrust (when throttled down all the way possible) should allow landing the booster/ship. Of course you could use larger engines for launching and smaller ones for landing, but now you need to develop two engines. More cost.

As it turns out SpaceX is developing two versions, sea level and vacuum, and they are suggesting the upper stage would use mostly vacuum models, with 2 or 3 (iterations are changing all the time) as sea level versions for landing, thus they leverage existing needs, to solve existing problems.

The F-1 and RD-170 were very hard to develop. Engines that large, really are quite hard.

SpaceX after the experience of developing multiple Merlin iterations obviously spent a lot of time thinking about the proper size for a Raptor engine. Some of their considerations would have been, ease of development, landing ability, sea level and vacuum performance.


It's simple really. First of all, large numbers of engines were desired to allow for failures to still allow one to reach orbit. Secondly, the size of the Raptor engines are about the same size as a Merlin engine. Being about the same allows for the same machines to make them.

That being said, they want to put the most of these engines on a single rocket as they can. 31 will fit, and thus that is what they are planning to do.

  • $\begingroup$ Are they actually using common tooling for Merlin and Raptor production? $\endgroup$ Commented Feb 9, 2018 at 15:07
  • 1
    $\begingroup$ I can't find if they are using the exact same tooling, but I remember hearing that making them the same size was a deliberate decision to reduce cost, as they know the equipment that can manufacture and manage the Merlin engines can also manage the Raptor engine production. $\endgroup$
    – PearsonArtPhoto
    Commented Feb 9, 2018 at 15:10

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.