Why does SpaceX use RP-1 in the first and second stages of their rockets? In my view at least the second stage could use hydrogen as it currently is not being reused. I am aware that you have higher initial technical requirements than with RP-1 but wouldn't it pay off because of the higher impulse?

Is it because the storage of RP-1 is more simple?

Or would a rocket with LH2/LOx tanks be too fragile in order to be reused?


4 Answers 4


SpaceX looked at the design of a booster with fresh eyes. Their concern was cost and reusability.

Different fuels, while more efficient are much more expensive to manage (two fuel systems), and develop (two different engines).

They took a more pragmatic approach of can we build an effective booster that is cheap, vs super performant.

An RL-10 cluster as the second stage of the Falcon 9 could potentially boost performance quite a bit. But thing is, performance is pretty great with LOX/RP1. But the RL-10 is super expensive compared to the Merlin engine. Thus SpaceX kept in mind cost, at what cost performance? The results do seem to speak for themselves.

The main consequence is that Falcon is great for LEO, pretty good for GTO, but not very good for interplanetary missions, compared to RL-10 based upper stages (Atlas V, Delta 4). But that is ok, the vast majority of missions are in that set of parameters.

Now SpaceX is also developing a Raptor engine running on Methane (CH4) and LOX. A lot of speculation has gone into performance of a Raptor powered second stage. But then the cost is looked at. New infrastructure at the pad (potentially 3 pads, LC-39A, LC-40, and SLC-4W) and a basically all new upper stage. With the lifespan of Falcon 9 limited by the coming monster of BFR, it is unlikely they will spend the money.

Also, SpaceX developed the Merlin engine 1A and 1B (for Falcon 1) and the Kestrel (upper stage of F1). Then the turned to the Falcon 9 and clustered the engines (and continued updating Merlin to the current 1D+++++ they are flying now. 75Klbs thrust to 190Klbs thrust earns the ++++ moniker).

So the F9 development needed engine improvements but not a new engine, new fuel.

As you look at the move from F9 1.0 to 1.1 builds with the move from Merlin 1C to 1D hindsight suggests this was a good call.

Also, if you are going to LH2/LOX you want it in your upper stage, not your lower stage. Lower stages do better with higher thrust and increased density.

Get out of the atmosphere and turn, then let the second/third stage do the rest. Performance is for the upper stage.

Some interesting related questions:

  • 1
    $\begingroup$ An RL10-based upper stage for Falcon would be much higher performance, but RL10 is a notoriously expensive engine to build -- according to some sources, over US\$25 million per (and list-priced at more like \$38M?), compared to under a million for a Merlin producing 9 times as much thrust. Merlin is a little overpowered for the stage, so you could get by with fewer than 9 RL10s in a hydrolox upper stage for Falcon, but it would still dramatically increase costs. Cost estimates from some internet rando here: forum.nasaspaceflight.com/index.php?topic=43053.0 $\endgroup$ Commented Jun 13, 2018 at 14:45

To make a long story short, liquid hydrogen has a very low density of just 70 kg/m3. RP1, on the other hand, has a density very close to that of water - about 1000 kg/m3. This means that for the same mass of hydrogen fuel, you'd need a tank 14 times as large. Couple that with the need to keep LH2 cryogenic or lose it to boiloff, and it becomes a very complex tradeoff between efficiency and complexity.

I can't tell you exactly what the engineers at SpaceX were thinking, but at the end of the day they chose to go with a less efficient but easier to handle RP-1/LOX solution for their first stage. This isn't new though - the engineers working on the Saturn V faced the exact same problems and came to the exact same solution. The Space Shuttle (and other rockets with external boosters) also followed a similar paradigm. At launch, a whopping 82% of the Shuttles thrust came from the low-ISP but high density (And thrust) SRBs.

enter image description here

Case in point, the Space shuttles boosters almost perfectly encapsulate the tradeoff between low and high density fuels. 4/5ths of the thrust comes from engines that are a fraction the size of a massive LH2/LOX tank.

  • 1
    $\begingroup$ At a certain point, the SRBs were much closer to 0% trust. (You've got a typo, but it's not enough for me to fix..) $\endgroup$
    – T.J.L.
    Commented Jun 13, 2018 at 17:40
  • $\begingroup$ #T.J.L. Typo location? $\endgroup$
    – UIDAlexD
    Commented Jun 13, 2018 at 17:51
  • 3
    $\begingroup$ @UIDAlexD: Last sentence: "4/5 of the trust..." $\endgroup$
    – cHao
    Commented Jun 13, 2018 at 17:55

SpaceX optimizes for cost over performance, and everything SpaceX does with respect to the Falcon architecture works towards that goal. SpaceX will gladly trade a few m/s of ΔV if a) it saves significant amounts of money and b) there's enough left over to complete the missions most of their customers need.

One significant cost optimization is to use the same materials, engines, and tooling for both the booster and upper stage. The only difference between the Merlin on the US vs the Merlins on the booster is the larger nozzle. This not only simplifies the construction of the vehicle, it also simplifies pad construction and operations (only one type of fuel line to mess with, only need one type of storage tank, etc.), further reducing cost. And while LH2 itself is relatively cheap (it's liquid air, c'mon), it's expensive to store and doesn't produce a lot of thrust by itself (there's a reason the SLS will continue to use SRBs). And its low density requires much bigger (and correspondingly heavier) tanks.

Yes, there is a definite performance tradeoff. While the F9 can send heavier payloads to LEO than an Atlas, an Atlas with the Centaur hydrolox upper stage can boost a heavier payload to higher orbits than the F9. However, the F9 is powerful enough to satisfy the needs of most comsat operators, at a much lower cost than the Atlas.

SpaceX is trying to wring as much performance as possible out of the F9, they're just coming at it from a different direction (lighter weight materials, 3D-printed engine components, densified propellants, etc.).

  • 1
    $\begingroup$ I believe the Merlin 1D Vacuum engine has a few more differences from the sea-level Merlin 1D than just a vacuum-optimized nozzle. Additional redundancy (no 8 other engines to pick up for you if something goes wrong), better thermal tolerances (long orbital coasts risk freezing of anything not in sunlight; I believe they had to add insulation to the igniter system, for example, to support restart after long coasts), and possibly some changes to the gimbal capability all come to mind as things I think I've heard mentioned. With that said, yes, the similar design saves lots of money. $\endgroup$
    – CBHacking
    Commented Jun 13, 2018 at 21:16
  • 1
    $\begingroup$ Another optimization to consider is design cost. Others have mentioned that LH2 imposes higher technical requirements. That means more complexity, more design work, more testing. Given that they are trying to do technically crazy things like land used rockets on a boat, it makes sense that they would cut back on complexity elsewhere. This is sort of like using Python instead of C -- 90% of the time, faster development is way more valuable than faster code, and I bet the analogy to rockets is a good one. $\endgroup$
    – senderle
    Commented Jun 14, 2018 at 13:35

I think geoffc is correct. It's primarily about cost and if that is a major concern then Hydrogen is almost immediately off the table. It is good for high efficiency upper stages but as both other answers mentioned it's bad for a lower stage as the thrust is on the lower end and the tanks are heavier. You need a pretty advanced engine just to make a Hydrogen first stage possible without boosters to help out.

You could still do a Kerolox first stage and a hydrogen second stage but that has added costs too. As Geoffc noted you have added cost from the ground systems at the launch sites and extra costs for development of the two different types of engines but for SpaceX it goes beyond even that as they have significant similarities between the first stage and the second stage. On top of using the same engine which the are producing in relatively significant quantities for rocket engines (~150 last year for the launches they performed) the two stages are also the same size and use the same tank designs so they can share tooling and their construction is largely similar. This helps drive some economies of scale and improve efficiencies of the assembly process instead of needing two sets of tooling and processes for assembling two completely different types of tanks.

Additionally you have to look at how the rocket as a whole works. The F9 first stage shuts off relatively early compared to other rockets so it is lower and going slower than say an Atlas or Delta first stage when its done its job. This ends up working well for recovery but means that the second stage has to do a lot more. It's quite a bit bigger than most other second stages and has a lot more thrust. A hydrogen powered F9 second stage would require much more powerful engine than the typical RL10 or possibly multiples to get the thrust needed.

The second stage would also need much larger tanks than a delta or atlas second stage or even the current F9 stage. The first stage is already at the limit of what is road transportable, any bigger round and it won't fit under overpasses and any longer and corners will be a problem so you can't really push it any further even you can still handle recovery. Making the second stage bigger to handle the hydrogen tank needed might start bumping into the same limitations on size that the first stage hits. The entire rocket as a whole is already really tall and skinny and going even taller and skinnier would also be difficult.

Road transportability is important for cost too and critical to the way they are currently setup. The rocket is built in California. Sent to Texas to be tested. Then to Florida or back to California for launch. That gets much harder and more expensive if it isn't road transportable. Larger rockets are typically built near the launch site (BO is building the New Glenn in Florida right near the launch site) or near the water like the BFR is being build at the Port of LA.

You also have to keep in mind the history of SpaceX and the F9 itself. SpaceX isn't the first rocket company startup but they are the first to really be successful. Many others failed and went bankrupt and SpaceX was very close to following that same path. Go back to September 2008. They've failed at their first three attempts at launching the Falcon 1. Musk has scrapped together enough money for one more attempt at launching the Falcon 1 before they go bankrupt. Everything is on this launch and if it fails SpaceX will go the way of countless other rocket company startups but hard work and perseverance pays off. Their launch is successful. Their business plan at this point is to launch smaller payloads on the Falcon 1 to hopefully fund the company while working on improving the Falcon 1 and eventually building a Falcon 5 and Falcon 9 to launch progressively larger payloads. Jump ahead a couple months to the end of 2008. NASA has awarded you a contract for Commercial Resupply Services to deliver cargo to the ISS. Your old plan is out the window you need a much bigger rocket. You need to get the F9 going and doing delivery runs to the ISS ASAP in rocket development timelines. That means you reuse as much as you can and you certainly don't have time to design a whole new hydrogen engine for the upper stage.


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.