sevenperforce

The whole thing should scale down pretty cleanly...if the reusable hab only has 50% mass growth over the Apollo LM, you can do the lander in 16-18 tonnes, which Falcon Heavy can send to TLI expendable.

Just for gits and shiggles, I decided to rerun the numbers on hypergols instead. With the same expected specific impulse of 316 seconds and a landing stage commensurate in size to a Titan II GLV upper stage (descent stage dry mass est. 2.8 tonnes), you need to throw 31 tonnes to TLI. SLS Block 1B can do it if they ever get the EUS working. There's a dry mass advantage with using hypergols because of the density issue; the lander has the same dry mass even though it carries way more props. Another option would be to have a reusable lander hab with about 800 m/s of onboard props and no main engine: basically a Dragon 2 with no heat shield, aeroshell, or SuperDracos. For each lunar sortie, you only send a landing stage vehicle with no hab to TLI. It would brake past the moon and rendezvous with LOP-G, then hang out until Orion arrived. Once Orion was there, the astronauts would transfer into the lander hab vehicle and then you'd use a Canadarm to berth the landing stage onto the bottom of the lander hab. The landing stage vehicle would refuel the hab, then would take it into LLO and down to the surface, then bring it back up to LLO, at which point the lander hab vehicle would unberth and return to LOP-G using RCS only. I estimate the Apollo APS dry mass at 0.7 tonnes, meaning the hab-associated mass of the LM ascent module would have been about 1.36 tonnes. Let's do 150% mass growth on that to allow for more crew, more capability, and down/upmass, bringing it to 3.4 tonnes. This is something Congress would actually like. Getting 3.4 tonnes from LLO to LOP-G costs 770 m/s (to give margin) and requires about 1.1 tonnes of props with about 300 kg of tankage dry mass. Let's say that the fully-loaded "taxi" rounds up to 4.8 tonnes. You'll note this is roughly the same mass as the Apollo ascent module. However, instead of only delivering the module from LLO to the surface (1.87 km/s), we need to deliver it from LOP-G to the surface and back to LLO (4.47 km/s). A pair of OMS engines will do the job, at 100 kg each. Now, as a single stage on hypergols, getting to 4.47 km/s would be rough; you'd need a whopping 40 tonnes of props and 7.6 tonnes of tankage and structure, not counting what's needed to brake to LOP-G. But we don't have to do that. Getting from the surface to LLO only takes 1.870 km/s, which can be done with the two engines, 665 kg of tankage, and 4.7 tonnes of props. If this is what breaks away from the landing stage, we aren't lugging a lot of dead structure up on that final burn. The Apollo descent stage had an engine massing 180 kg, 8.2 tonnes of props, and 1.92 tonnes of structure and tankage. Using that as a benchmark, you end up with an engineless landing "stage" massing 2.53 tonnes and carrying 10.8 tonnes of props, for a gross mass (with cargo) in LLO of 23.7 tonnes. If we add 1.7-tonne drop tanks at this point, carrying about 12 tonnes of props, we will burn 6.75 tonnes of those props getting from LOP-G to LLO. The tanks also need to refuel the reusable hab vehicle, leaving 4.1 tonnes of props for getting from TLI to the LOP-G. The total stack ends up being 32.58 tonnes...just slightly more than before, but now with a reusable hab that's twice as big as before. That's clever staging for you. EDIT: Note that the reusable "hab" is only 4.8 tonnes fully loaded and has ample dV to get from TLI to LOP-G on its own. That's well within the capability of an expendable Falcon 9 or a recovered Falcon Heavy. The modular design also allows the sortie stage to be customized for the mission. The base model could just have an airlock; a slightly larger one could also deliver a rover, additional experiments, etc.

Pushing against itself will not happen. Too bad there is no such thing as a "gravity brake" or some other way to interact with gravitational field lines and not constantly expend energy. Well i'm only thinking of using Antimatter for propulsion; fusion would be providing power as it scales up pretty well. But yes; it's all cumulative. And heat transport would be an issue; but graphene radiators could be pretty epic. Antimatter-catalyzed nuclear pulse propulsion is a good intermediate step and could scale Orion down to pulsejet size, which would make for one hell of a spectacular upgrade to Project Pluto. In the nearer term, z-pinch fusion pulse propulsion is an even easier (and clean-burning) alternative. It can be scaled quite small. Ordinarily it needs a magnetic nozzle but I wonder if it could be used in-atmosphere with a turboladunglufteinlassraketenmotor. At liftoff you'd be using lithium fusion to heat an airflow as working mass; as you accelerated you'd switch from airflow-only to a hydrogen or even methane working fluid, and then once orbital you could turn on the magnetic nozzle and do your Earth escape burn at 19,400 s.

No need for a prop depot. Just launch the cryo lander from a stack on the SLS. Even if Orion is stuck at Gateway and runs out of time, it's not a LOCV, just a benign LOM, which is not really a problem for the program. I agree that using hypergols for the lander is prohibitive. I agree. 45 tonnes. Almost enough to throw a barebones cryo lander plus Orion for a repeat of Apollo. Even expendable FH can't throw 23+ tonnes to TLI, and that's cryo, which is incompatible with FH.

I did allow for boiloff. Launch cadence isn't as big of an issue if you have a little extra propellant margin. You can send Orion to the Gateway and if they need to extend their mission by an extra two weeks because the second SLS slips, nbd. I would say no. Though the architecture does allow for the same lander to drop 10 tonnes of downmass on the moon as a precursor. So that's nice.

One of the nice things about NSWRs is that if something DOES go horribly wrong, it becomes an impromptu single-pulse Orion. So structure your vehicle accordingly, and you're fine.

Yes, I am aware. You need not inform me of the difference between pressurized reactors, molten salt reactors, and a NSWR. No, not necessarily. You go on to list many of the very real challenges, which are readily acknowledged. They are by no means solved. But they are not significantly more challenging than the early materials and engineering hurdles for pressurized nuclear reactors. They are different challenges, obviously, but they are not dramatically more different. It is nothing whatsoever like the challenges in creating a Halo drive, for example.

I didn't misunderstand, I threw in the Apollo baseline for what a minimal lunar surface sortie would take mass wise. The bare minimum is 45 tons to TLI, regardless of the number of launches (where the TLI is to LLO). Orion CSM is actually lighter than Apollo CSM, but the capsule itself is much heavier, so it has less propellant. A minimal lander would be the Apollo LM (16.4 tons). That of course requires a lander that only ever operates to and from LLO, however. Since Orion cannot do LLO, the lander then has to go from Gateway to the surface, and the ascent stage comes back to Gateway. This is a substantial mass change. In addition, with 2 launches the lander ALSO has to do the LOI burn, which at least at Gateway is not so bad. The end result is that the lander doesn't need 4400 m/s, it needs maybe 1000 m/s more (a few hundred for LOI at Gateway, and 7-800 m/s to LLO) to get to the surface, and the ascent stage needs an extra 7-800m/s to get back to Gateway (less prop because only the ascent stage). So we need a lander with like 6.2km/s of dv (again, this is slightly confusing because the descent stage needs maybe 3200 (LOI+transfer to LLO), and the ascent stage (smaller) needs maybe 3000 m/s). Hmm. Let's allow 50% mass growth over the Apollo ascent module, since we obviously wouldn't be satisfied with just replicating Apollo. The ascent vehicle had a wet mass of 4.7 tonnes, of which 2.55 tonnes were props, and had 2.22 km/s of dV. Trying to pull off cryos for the ascent module is just ridiculous, so we'll stick with hypergols. To be generous, we'll boost up to 316 s of isp on the ascent propulsion system, which will get us to 2.42 km/s. Still not enough for the gateway. Let's pile on pure props to get to 3 km/s. That puts us at a wet mass of 8.49 tonnes for the ascent module; we'll round up to 8.6 tonnes to allow for tankage and propulsion system mass growth. So our cryogenic lander, launched on SLS, needs to deliver 8.6 tonnes from TLI to gateway and then from gateway to the lunar surface. You need 420 m/s from TLI to gateway, 730 m/s from gateway to polar LLO, and 1,870 m/s from LLO to the surface: 3020 m/s. It's going to need margin for correction and boiloff so let's give it 3.1 km/s total to be fair. Just for a baseline, let's develop our descent module based on the Delta Cryogenic Second Stage formerly used on Delta III, because I suspect it will come out to the right ballpark. Dry mass is 2.5 tonnes; let's grow that to 2.8 tonnes to allow for landing legs and other lander-y stuff. This brings our lunar touchdown mass to 11.4 tonnes, which with a single RL-10 is about one gee. With the amazing 462 s isp of the RL-10, you only need 11.2 tonnes of props to get 3.1 km/s, bringing the total mass SLS must throw to TLI to only 23 tonnes, well within the performance of even SLS Block 1. So it's doable, but only with cryos.

The challenges are not substantially different than those involved in building molten-salt reactors or pressurized heavy water reactors. There are some critical elements that need work, but the engineering problems are really not that bad.

I tried translating back into English and got "Enriched with liquid oxygen fortified ether-supercharged nuclear shred" which isn't quite right but sounds cool enough. I played around with it a bit. German actually has a word for an afterburner fueled by liquid oxygen: Flüssigsauerstoffnachbrenner. And an air intake is a Lufteinlass. And a turbocharger is a Turbolader which just sounds freaking cool. A nuclear rocket engine is an Atomraketenmotor. So I suspect the correct term would be Turboladunglufteinlassatomraketenmotor mitflüssigsauerstoffnachbrenner, or "Turbocharged air-intake atomic rocket engine with liquid oxygen afterburner."

Mitflüssigemsauerstoffangereichertethermischaufgeladenekernramrakete.

Oh, it would work. Whether it would work at peak efficiency is another question, but it would definitely work. I wonder if you could build an open-cycle molten-salt reactor that could expend a working fluid like a NSWR during launch and then function more like a fission fragment rocket for high specific impulse during interplanetary flights. A vehicle based on this engine would need to use drop tanks for launch but would otherwise be pretty fully reusable and could fly single-stage to the surface of Mars easily enough.

That would get you into reactionless-thruster land. You gotta push against something (planetary magnetic field, atmosphere, etc.) or nothing happens. I was actually like 13 or 14 and designed a magnetic "flying saucer" that used gyrostabilizers and a superconducting magnetic repulsor field. The idea was to push against the Earth's magnetic field and hover. When I got a little older, took physics, and did the math, I learned that my idea would have worked...it just would have required exhorbitant amounts of electrical energy. Earth's magnetic field is immense, but very diffuse; levitating any sort of manned flying machine would require creating a magnetic field something like the size of Brazil. It brings a tear of joy to my eye. Such a thing of beauty. I wonder if I have those pixel drawings somewhere. I was using postimg to host them and evidently that was a bad idea. Of all the possible configurations, I believe that the LANTTCRR (LOX-augmented nuclear-thermal turbocharged ramrocket) running on an exotic carbide-alloy pebble-bed reactor burning either hydrazine or ammonia ends up resulting in one of the smallest vehicle cross-sections for an SSTO. Methane has a better mass fraction but I think ends up being larger. LH2 may squeeze out a slightly higher mass fraction than methane but is prohibitively larger.

You cannot reach a velocity in an airbreather which is substantially greater than the actual exhaust velocity of your engine. And even if you use antimatter to boost up the exhaust velocity so you can reach orbital velocity in-atmo, that doesn't get you into orbit. It's not about "extra fuel" for "more fine control" of orbital insertion; it's the fact that orbital insertion itself requires a separate rocket engine.

Ah, here it was. It's a shame the images died.

I did? Now I need to go hunt it down.....

I wonder what kind of margins you'd have if you strapped three Falcon Heavy side boosters to the appropriate points on that and lifted off asparagus-style.

No fully-airbreathing SSTO is possible, since you need to circularize once out of the atmosphere, and you cannot do that without reaction mass. Accordingly, you will always need some sort of propellant for the final orbital insertion. Building a vehicle becomes an optimization problem between retaining airbreathing capability for as long as possible and not wasting too much fuel fighting drag and heat rejection issues. I can build an SSTO that spams air-locked RAPIERs and can reach almost to Kerbin orbit -- such that I need only a Sepratron to circularize -- but I could do it more efficiently by using fewer RAPIERs and carrying a bit of regular liquid bipropellant. A simple liquid-core antimatter rocket using methane, ammonia, or even plain water can nearly make orbit twice in a day without refueling, so if you just built that with an air intake, you could do it every day and twice on the weekends, no problem. There's really no maximum possible size other than the constraints of your materials.

Depends on what happens to the dry mass associated with that extra thrust. With the Falcon 9, there was no addition to dry mass required to uprate the engines, and so the gains were essentially free. With the Atlas V, the added mass of the SRB casings is jettisoned at burnout, so there's no worry there either. The original Atlas that put John Glenn into orbit jettisoned its outboard engines once their thrust was no longer needed. However, if you try to replicate in KSP by simply adding more engines, you will run into dry mass losses before you run into drag losses. Real-life engines and tanks are much lighter than in KSP. In real life, I think piling on more thrust (more SRBs, etc.) would cause structural problems with your launch vehicle before the gravity drag advantage would be outweighed by the air drag disadvantage.

Great answers already from @Ultimate Steve and @tater, but an additional explanation would look at the uprating on the Falcon 9 Block 5 itself. There is no difference in propellant capacity between B3, B4, and B5, and in fact the B5 actually has slightly greater dry mass than its B3 and B4 predecessors, but the B5 has greater throw to LEO and beyond because the engines burn at higher thrust. Prior to the gravity turn, every second spent climbing costs you 9.8 m/s off your final velocity. Adding more propellant does not add linearly to your final stage because of the tyranny of the rocket equation, but adding more thrust does produce linear gains that propagate all the way up. If you can uprate your engines enough to shave 10 seconds off the pre-turn climb, then you stage at 98 m/s faster, which is 98 m/s less that your upper stage has to produce. That's why adding those inefficient SRBs to the bottom of the Atlas V and Delta IV is so very effective.

In order to be as cost-effective as possible, a launch company needs a "dial-a-rocket" model. Otherwise you end up with a far more expensive launch vehicle than needed for most missions. ULA accomplishes this because most of its launch vehicles are not 2-stage-to-orbit, but 2.5-stage-to-orbit, with parallel (usually solid) boosters. The solid boosters themselves don't contribute much in terms of absolute dV; their real advantage is to amp up TWR (not unlike F9B5 did with the final uprate on the Merlin engines) so gravity losses on the core stage are lower, giving the core greater speed at staging. Putting the bulk of the work on the core allows a low-thrust upper stage, which means a less heavy upper stage, which helps with mass ratio and obviates the famously horrible TWR of hydrolox engines. It's a great model, but it simply isn't amenable to amending into a fly-back-booster model. Look at the comparison. The original Falcon 9 could throw 9 tonnes to LEO, whereas the Atlas V 401 can throw 9.7 tonnes to LEO. Pretty comparable. Yet for the original Falcon 9, the upper stage packed 5.3 km/s and the lower stage packed 3.6 km/s, while for the 401, the Centaur packs 4.4 km/s and the Atlas CCB packs 5.6 km/s. The Falcon 9 upper stage was, in comparison to its lower stage, 87% beefier than the Centaur. Of course the Falcon 9 has evolved tremendously; for the same payload, it now packs 7.6 km/s into the upper stage while the lower stage has the same dV as before. SpaceX can "dial-a-rocket" by simply adjusting the recovery profile: RTLS has a lower operations cost than ASDS recovery, and near-shore ASDS recovery has a lower operations cost than distant recovery. Since the ULA model puts so much weight on the core, the only thing that really makes sense is downrange recovery of the engines alone. It's telling that they require an expandable heat shield just to try. At the same time, putting a fully-loaded Falcon 9 upper stage into LEO wildly outperforms putting a fully-loaded Centaur into LEO, because it's simply so much bigger. Not on Atlas V. Interestingly, Russian-derived launch vehicles (like CNSA and ISRO) may actually be more amenable to first-stage reuse than the ULA model. Rather than strapping on more core boosters, they typically add liquid kick stages to achieve their dial-a-rocket performance. Look at Fregat on the various Soyuz versions, for example.

I agree. I don't know whether Tory is factoring in hidden costs or development investments or anything else, though. Tory also seems to be making subtle digs about the fact that Boeing and Lockheed are publicly traded and SpaceX is not. Though really he has little ground to stand on in that arena. There is no evidence that SpaceX is getting significant private-investor bailouts. Of course this runs into Tory's point about how the fleet average must be ten flights. Flying expendable is a nasty datapoint. This would be more of an issue if not for the fact that even when flying with RTLS recovery, F9B5 has a ridiculous amount of throw to GTO. There really aren't any regular commercial payloads available that F9B5 with ASDS recovery cannot throw to GTO (or at least near-GTO). Also, I just realized that Tory said fly-back recovery requires a fleet average of ten flights. ASDS recovery reserves more performance, which means more expensive payloads, which presumably means a fleet average much lower than ten.

The genius of Falcon 9 was the Merlin engine. It has such a tremendously great thrust to weight ratio that they were able to uprate and uprate and uprate it and so kept gaining margin without adding significant dry mass, to the point that they can reserve props and still throw ANYTHING on commercial markets.

Tory's arguing that the total costs of a flyback, refurbishment, and reuse program, over time, offset 90% of the value of a booster. Even adjusting for opportunity cost, propellant margins, and a dozen other possible elements, I don't see how that's remotely possible.

With a magrail, the surface-based approach is far more successful, yes.