Falcon Heavy Moon Mission Update--Now With Raptor Upper Stage!

I got a comment (I got a comment!!!) from "redneck" on my previous post, which looked at a 2-launch Falcon Heavy architecture for getting crews to low lunar orbit.  He suggested that going for on-orbit refueling of the F9 S2 solved a lot of problems, and I've come to believe that he's right.

While refueling one S2 from another one presents fairly nightmarish plumbing difficulties, reusing one of the second stages from the launch of one of the Falcon Heavies certainly reduces the need to engineer a chopped-down TLI stage and figure out how to launch it within the current FH length constraints. With refueling, you'd launch the Dragon 2 crew module with the lunar maneuvering stage, using the attached FH Stage 2 to do the TLI burn, even though it doesn't yet have enough propellant to meet the delta-v requirement.  Then you'd launch the support D2 on its own, leaving plenty of propellant in its S2 "tanker" stage once it was on-orbit.  After rendezvous and docking the two D2's nose-to-nose, you'd somehow magically transfer the remaining prop from the tanker S2 to the TLI S2, and off you go.

This has many nice properties.  First, you can make the D2's somewhat heavier.  There's still some customization that has to be done for the D2 support module, but it's nowhere near the previous strip-it-to-the-studs requirement we had when we were launching a fully-fueled TLI stage on top of the FH.  I still think that the support D2 probably doesn't have a heat shield and may not have SuperDracos, but my guess is that you can wind up with a lot more common equipment between the two versions.

Also note that our problem with sending the crew D2 ass-backwards into the unknown (and therefore pulling negative gees) has gone away.  That removes the need for couches and control panels in the support D2.

So let's talk about transferring fuel.  The way it's planned for the BFS to do this is to dock tail-to-tail.  That makes a lot of sense if you've designed your stages to do it, but it would require a completely new interstage to do it with an F9 S2, because you have to protect the Merlin vacuum nozzle(s) from damage.

Furthermore, we know we want our D2's docked nose-to-nose.  So we either have to do a whole bunch of undock-turn around-dock to fuel-undock-turn around-redock the payload stuff (which sounds nuts to me), or we have to find a way to transfer fuel from the FH #2 S2, on one extreme end of the docked stack, to FH #1 S2, on the other extreme end.

I can only think of two ways to do this, and both of them will cause the NASA crew safety people to be wearing their brown trousers on rendezvous day:

1. Extend some kind of double refueling boom (one for the LOX and one for the RP-1) from the the tanker S2, around the LMS and both D2's, to mate with fueling collars on the TLI S2.  You'd have to stow the booms on the outside of the stage, then either swing or extend them, then mate them securely to the collars.  The fact that the two D2's and the LMS exceed the length of the current F9 S2 by almost 7 meters isn't going to help things be any safer or more reliable.
2. Build refueling lines into the LMS and both D2's to transfer the propellant.  This has actually been envisioned in the IDSS docking standard, but never implemented.  And of course you'd have to figure out how to run the lines through the D2's and their trunks.

This frankly sounds butt-ugly, which is why I didn't consider it in the first go.

However, this isn't our only problem.  After thinking about it some more, I think our stage lifetime problems were more acute even before we decided to try refueling, and now they're even worse.  Here's an arm-wave at just how long it takes before you can do the TLI burn and be rid of all of your S2's:

1. Launch first FH. (This starts the clock running on the TLI stage life.)
2. Launch second FH.  Even if you have the ability to launch two FHs into the same launch window (which you probably don't), you have to plan for a glitch on the second one once the first is gone.  That's at least a 24 hour wait for a launch window if you're going to do everything into an inclined LEO (and you are).
3. Rendezvous and dock the S2+LMS+D2Crew with the S2+D2Support.  The shortest ISS rendezvous has taken about 6 hours.  Let's budget 8.
4. Do the fuel transfer.  How long this takes depends on the mechanism used, but I wouldn't budget less than 6 hours with checkout, settling burns, transfer, and closeout.
5. Coast to your TLI insertion point.  Could be as much as 90 minutes.
6. Throw in 8 hours for things to go wrong, but not quite wrong enough to abort.

Total:  48 hours.

I'm pretty sure that the longest coast from launch to stage restart by the F9 S2 is about 45 minutes, so we're just a tiny bit short.  I am far from an expert in the various design factors that go into increasing stage life, but here are the ones I know about:

1. Boil-off.  Not a problem for RP-1, but even subcooled LOX will warm up and start to boil, and the boiling will cause it to be vented to avoid pressure buildup.  Boil-off can be minimized with insulation, or by really big tanks.  Since we're pretty much stuck with something close to the F9 S2 form factor, the latter doesn't work.
   
   Since we only need some of our propellant in each stage reserved for TLI, I wonder if the "header tank" scheme proposed for BFS would work.  Not only does it prevent sloshing during landing (only interesting for the BFS), it also has the effect of putting some prop in a kind of thermos bottle.  If, after achieving orbit, the outside main tank is vented to vacuum, it will insulate the inner tank quite effectively, since it won't be subject to much radiant heating.
2. Helium.  Everything on the F9 S2 is pressurized with helium, which has its own problems with heating and venting.  Once the pressurant is gone, the stage is dead.  If you could use autogenous pressurization (pressurizing each tank with boil-off from that tank), you could drastically reduce helium requirements and lengthen lifetime.  But you can't do that with RP-1, because its vapor pressure is so low.  (For hopefully obvious reasons, it's a really bad idea to pressurize your RP-1 tank with gaseous oxygen...)
3. Power.  It's one thing to run all your avionics and actuators off a fairly small battery for less than an hour.  It's quite another to use battery power for two days.  The stage is either going to need its own solar panels, or it's going to have to draw power from the D2's solar arrays through a bus that probably doesn't exist right now.
4. Attitude control.  If you're using hypergolic or cold gas RCS thrusters, at some point you run out of hypergolics or nitrogen.

I think it might be time to give up on the F9/FH S2.  Is there an alternative?

Oddly, there might be.  We know the US Air Force is interested in developing a Raptor upper stage for the FH.  There are two good reasons for USAF to be interested in this, and there might be a couple that would interest SpaceX:

1. The second round of Evolved Expendable Launch Vehicle requirements include the ability to put satellite into GEO without attaching payload assist systems.  That requires a longer coast period to get to GTO apogee (about 5.25 hours) before doing the insertion burn into GEO.
2. In addition to EELV2, there have been rumors that USAF wants to make the X-37C operational.  This is a bigger version of the X-37B, which SpaceX has already launched once.  Rumors are that it's big enough that USAF needs an upper stage with a bit more energy to deploy it for some of its missions.
3. Beyond just the USAF requirements, SpaceX would have an interest in getting a testbed for some of technologies needed for the BFS, which turn out to be almost exactly the technologies we need up above:  autogenous pressurization (which requires switching from RP-1 to liquid methane), low boil-off of some amount of propellant, and refueling.
4. In addition, switching the primary second stage for both FH and F9 to the Raptor means that the number of Merlins they have to produce drops to a trickle after they finish building out their fleet of medium-reusable block 5 cores, freeing up line space and personnel to ramp Raptor production.

So, without further ado, let's unveil the new Raptor Upper Stage!  The RUS:

1. Uses a single Raptor vacuum engine for propulsion.
2. Uses liquid methane (LCH4) instead of RP-1 for fuel.
3. Has about 25% more energy than the current F9 S2.
4. Retains its interstage as a skirt around the engine nozzle, enabling tail-to-tail docking and secure refueling attachments.  Note that the current F9/FH land the interstage as part of the core.  Using it as a skirt for the S2 means that it's no longer reusable.  (This might be a good reason to keep the current F9 S2 in the inventory for lighter launches, although then you're back in the Merlin-making business.)
5. Has a dry mass of about 7.6 t vs. the ~4 t of the current F9 S2, and a propellant mass of 115.9 t vs. 107.5 t.  (Dry mass is heavier because of  the docking skirt/interstage, and because LCH4 is less dense than RP-1, requiring a bigger tank.)
6. Is 5 m in diameter, vs. the current 3.66 m F9 S2.  (The fairing is 5 m in diameter, so I don't think this messes up the aerodynamics.  It does significantly complicate the attachment of the D2; My guess is that it'll require a tapered interstage to attach to the existing D2 trunk.
7. Is slightly shorter than the F9 S2 (due to the wider stage).
8. Has header tanks inside the main tanks, capable of holding roughly half of the propellant in a low-boil-off state.
9. Keeps its nitrogen for its cold-gas thrusters in the thermos bottle, and makes the LN2 tanks bigger.
10. Takes power from an umbilical to whatever is mounted above it.

When I run the numbers on this, everything gets significantly better.  We can now comfortably carry a crew of 5 with a mission duration of 3 weeks.  We can carry at least 1 t of extra (non-crew) payload to LLO.  And we now have one FH launch with two-booster reuse, and one with full reuse.  Here's the mass budget:

Module Name	Dry Mass	Payload Mass	Prop Mass	Wet Mass	Engine	Specific Impulse	Delta-v Needed	Delta-v Available
Dragon 2 Crew	6,400	1,323	500	8,223	Draco	300	-	184
Dragon 2 Support	6,400	1,323	500	8,223	Draco	300	-	184
Lunar Maneuvering Module	1,000	-	15,300	16,300	OME (AJ10)	316	1,950	1,950
TLI Stage 2	7,570	-	55,723	63,293	Raptor	375	3,190	3,190
Launch Name	Description		Payload	Length	Launch Type	Prop at LEO Burnout
FH #1	Tanker + D2Supt		8,223	8.1	2 Booster Reuse	41,008
FH #2	TLI + LMS + D2Crew		24,522	10.7	Full Reuse	16,665
Excess Above TLI Delta-v:						1,950

The updated mission profile is:

1. Launch FH #1 as the S2 (RUS) tanker, with the support D2 as payload.  This is a 2-booster-reusable launch.
2. Launch FH #2 as the S2 (RUS) TLI stage, with the LMS and the D2 crew on top of it.  I'm going to assume that it's better to adapt a 3.66 m LMS to the 5 m RUS and get rid of the interstage than it is to adapt the D2 to a 5 m LMS and keep the interstage through most of the mission.  This flight is fully reusable.
3. The two launches rendezvous.
4. D2 support undocks from tanker S2.
5. Tanker RUS docks tail-to-tail with the TLI stage + LMS + D2 Crew. Propellant is transferred.
6. Tanker RUS is jettisoned.
7. D2 support now docks nose-to-nose with the D2 crew.
8. Now we can do TLI.

Still some unanswered questions:

1. Obviously the RUS is a lot of work.  If it weren't for the USAF being interested, I'd say it was silly.  But I'm now convinced that you really can't do this mission with the existing RP-1 S2, because you'll never get the stage life you need.  So it's the RUS or nothing.  Given that, taking the lunar requirements into account seems like a pretty good idea. (I'm sure Elon is hanging on my every word...)
2. Going to a 5 m diameter RUS is a big pain, and interstage weirdness will ensue.  In theory, you could do this with a 3.66 m RUS, but it's a really long, thin rocket when you do.  Seems like SpaceX must be mighty close to the aspect ratio limit already.
3. Crew safety issues abound here, but you pretty much have to transfer prop at some point.  I suppose that the NASA folks might decide that it would be safer to launch the crew on a third F9, dock with the D2 crew after refueling, transfer crew, undock, and then dock the D2 support, but that requires even longer stage life, and it's actually more docking operations with the crew exposed to the point of contact than the way I have it above.

All-in-all, though, this seems quite a bit cleaner.

If I were really ambitious, I'd spec out a nice lunar lander for this architecture.  I'd guess you'd mount it on top of the LMS, and fly the tanker up with no payload (other than its own prop, of course).

Maybe next time.