In a previous post I wrote about the technological argument against using orbital refuelling. That post was about a thought experiment, using 1970s technology to perform the mission profile proposed for Artemis 3. The thing about Starship, however, is that it’s such a massive leap forward that many are hesitant to believe it’ll perform as promised.
The thing is, orbital refuelling can be a massive enabling technology even when you limit yourself completely to already existing rockets. Any launch vehicle launching to Low Earth Orbit will bring with it a large empty tank in the form of its upper stage. That’s usually discarded, but if we had a depot floating around in LEO, it’s possible to top it up.
It doesn’t need some massive orbital facility either. The upper stage you want to refuel can be used as a depot as well. If you want to top up a Falcon 9 upper stage, you could just use another Falcon 9 upper stage. Some modifications are necessary, to allow orbital refuelling as well as long term on-orbit storage of propellant. The point isn’t that we could do this today, but instead, that something like Starship isn’t necessary to unlock the potential of orbital refuelling.
In the graph below I mapped out the performance of number of orbital launch vehicle upper stages if they could be fully refuelled in orbit. The change in velocity (or ∆V) is mapped out on the y-axis, the payload in metric tons is on the x-axis. The straight black lines show the ∆V required for certain destinations. The vehicle masses are taken from a number of sources, mainly Wikipedia, user’s guides or manufacturers websites. This whole exercise is purely hypothetical, and these figures are not intended to be very exact. They function mainly as rough estimates for illustrative purposes.
Most upper stages, when fully fuelled, have a potential ∆V of ~8-10 km/s, which is a little surprising, because they all use very different propellants. While hydrolox beats kerolox which beats hypergolics in pure performance, the higher density of kerolox and hypergolics mostly negates this performance gap. The performance curves aren’t the exact same shape, which is mostly a function of stage size: smaller stages tend to see their performance decline faster as payload increases, which makes sense.
Despite the efficient hydrolox engines, the Centaur upper stage on the Atlas V has the lowest payload for everything but the most demanding missions, because it’s small. Even so the tiny Centaur can lob ~18 tons to Trans Lunar Injection, could insert ~12 tons into a Low Lunar Orbit, and could even land ~5 tons on the lunar surface or insert a similar mass straight into a Low Mars Orbit . The third stage of the Proton rocket has similar if slightly better performance yet it relies solely on storable propellant.
A Falcon 9 upper stage if fully fuelled could land ~14 tons on the lunar surface, insert ~35 tons into lunar orbit, or send ~60 tons through TLI. Finally, the Exploration Upper Stage on SLS Block 1B could land 25 tons on the lunar surface or send a 100 tons through TLI. None of these stages could land on the surface of the moon or last months in orbit without significant modifications, of course. But that’s not what I’m trying to argue. I instead want you to think about the potential here. A single Falcon 9 launch, a normal, medium lift commercial vehicle capable of launching a modest 22 tons to LEO, could with a single refuelling outmatch the performance of the Saturn V.
Don’t forget, these aren’t total masses sent to a destination, these are useful payloads. Depending on the mission, some of that would be eaten up by necessary changes (a lander would obviously need landing legs, for example). The Falcon 9 US is almost capable enough to send the Constellation stack of Orion and Altair through Trans Lunar Injection. Nor does that performance require massively expensive on-orbit facilities or advanced long term cryogenic storage. These sorts of performances are possible with kerolox and hypergolic upper stages.
Falcon 9 is not unique here among commercial vehicles. It just has the benefit of having quite a big, LEO optimised upper stage. We could go even bigger when we look at sustainer stages. Launch vehicles like the European Ariane 5 use a center core surrounded by a pair of large solid boosters to propel both payload and small upper stage almost all the way to orbit. That core stage could be inserted into LEO and be refuelled just like an upper stage could.
If you did that for an Ariane 5 core, you would get even better performance than the EUS from SLS Block 1B. If you do it with the SLS core stage, which also makes it to orbit, then the sky is the limit.
This isn’t free performance, obviously. To top up a Falcon 9 upper stage, you would need five Falcon 9 launches to fill up a depot stage. Initial mass in LEO would be on the order of ~120 tons, similar to the ~140 ton IMLEO on Saturn V. Prop depots aren’t magic, but they do provide a way around the biggest problem of multi-launch architectures, which is that mission risk massively increases the more mission critical launches there are. This graph from ESAS (page 392 from the final report) shows a knee in the curve for mission succes around 3 launches, which is part of the reason NASA tends to favour large rockets and low launch rates for human exploration.
When there’s a depot involved, however, all those propellant tanker flights become non-critical. Propellant tankers can be much more disposable than, say, a lunar lander, and the depot can loiter for months. That means only launches carrying complex mission critical hardware - an Orion spacecraft, a lunar lander, a hab, etc - are mission critical. A hypothetical mission in which Falcon 9 launches a crew capsule, gets topped up, and then inserts the crew capsule into Low Lunar Orbit would only have two mission critical launches and one critical docking, even though the total number of launches and dockings required is six. Depots take complex, multi-launch architectures and make them simpler and lower risk. They can do so without massive in-space infrastructure, without new in-space stages and without advanced zero boil off hydrolox technology. There’s really no reason not to develop that capability.