sevenperforce

Apparently they have dusted off their inanity and are now trying to launch a rocket out of a water tower?

After a little more poking around into this idea, I have what I think it is a pretty inventive way of doing a tripropellant power cycle. The trick is to use multiple turbines on the same shaft and to use a variable mixture ratio not in the chamber, but in the preburner. The attached image has the propane-LOX power cycle on the left, a blended power cycle in the center, and a hydrolox power cycle on the right. However, for the sake of explanation, we'll start at the right and move to the left. One of the limitations on an expander cycle is a lack of heat. If you can get a little more heat, you can have more power to operate the expander turbopumps and thus improve your pump power and total thrust. So I have an oxygen-rich preburner that burns a little hydrogen with all of the oxygen, not to operate a turbine, but to provide extra thermal energy to an oxygen-loop heat exchanger that operates the oxygen turbopump. On the right, you can see that LOX comes out of the pump, through a heat exchanger inside the preburner, through the turbine to operate the pump, and then into the preburner and subsequently into the engine. The hydrogen side operates like an ordinary expander cycle, pulling heat off the nozzle and chamber and splitting off just a little bit of hydrogen to operate the preburner while pushing the rest into the chamber. That's going to have really high efficiency (you really only need to put a tiny bit of hydrogen into the preburner, just enough to give you the heat you need to operate the oxygen turbopump), but it's not going to have great thrust. So let's switch over to the left-hand side, where you have the liftoff configuration: a propane-LOX engine running on a hydrolox gas generator cycle. Let's unpack this. On the LOX side, you've used the three-way valve to run significantly less LOX through the heat exchanger and into the preburner; instead, you're pushing most of the LOX directly into the combustion chamber. On the fuel side, you're no longer pushing any hydrogen into the combustion chamber; you're sending it all into the preburner. You're pumping propane into the combustion chamber in its place. Because the preburner is no longer getting as much LOX and is getting significantly more hydrogen, it's now mildly fuel-rich. But more importantly, it's not exhausting into the combustion chamber anymore; instead, it's going into a turbine that exhausts into the nozzle extension. Because this exhaust is extremely low-pressure, the available power from that turbine goes through the roof, and so there's substantially more LOX flowing through the entire system at higher pressure. Similarly, because all of the liquid hydrogen is going into the preburner which exhausts to the nozzle extension, you've gone from a closed expander to an open expander, and so the power on the fuel side also skyrockets, allowing you to pump the same amount of hydrogen plus all of the propane at significantly higher pressures. The result is a high-pressure propane-LOX reaction in the combustion chamber with a hydrolox gas generator. The center version demonstrates the transition from high-thrust, low-isp to low-thrust, high-isp: the preburner is now slightly oxygen-rich and is partially exhausting into the chamber and partially exhausting through the turbine, while the hydrogen is partly going into the chamber along with a small amount of propane. With three different valves in operation, you get a smooth transition between the two power cycles. And since you're pushing a little more fuel into the chamber than you have oxidizer, this reheats in an afterburn with the slightly oxidizer-rich preburner exhaust in the nozzle. I feel like this design simultaneously maximizes the advantages of gas generator cycles, closed expander cycles, open expander cycles, oxidizer-rich staged combustion, and nozzle-injection afterburning. Seems promising. Also, because the propane is never in a fuel-rich precombustion state and isn't used for cooling, you can utilize any fuel here: methane, RP-1, even more hydrogen if you want.

Kerbal sim isn't really necessary; we can do the math easily enough. Elon is talking about a velocity kick from an eccentric orbit...let's say GTO for the sake of simplicity (much higher and phasing gets froggy). Starlinks are on the order of 260 kg, but let's assume modifications bring them up to 300 kg (they're going to need some sort of attitude control and increased solar capacity since they won't be in LEO). Interpreting "a few dozen" as 40, that's 12 tonnes of payload. Total dV on the expendable Starship would be 11.7 km/s. Add the 2.3 km/s you get from your GTO staging orbit and that's a total of 14 km/s to play with (relative to LEO). That's a lot. But what would you do with it? This feels like a throwaway tweet by Musk, given that Starlink satellites don't have imaging or other science systems or anything else to contribute, and their solar power makes them pretty useless beyond 2 AU. The New Horizons spacecraft had a launch mass of 478 kg, not much more than a modified Starlink. With the same mass budget, you could launch no less than 25 New Horizons clones all over the solar system, if you wanted.

Interestingly, it looks like Aerojet already had a similar idea: This is a hydrolox expander-cycle engine that uses a fuel-rich preburner as a "first stage" of the combustion chamber, not to operate a turbine, but simply to provide an additional source of heat for the hydrogen to use in the expander: Very cool concept. Provides some of the advantages of staged combustion without the moving turbine in superheated gases, which is the hardest part. Ordinarily, you can't use the heat of a staged combustion preburner directly, because it is the heat which drives the expansion and powers the preburner. However, it might be possible to extract some heat energy from a standard ORSC preburner-turbine combo to lower the temperature and thus allow a lighter, simpler turbine while still operating at reasonably high pressures, and use that extra heat energy to make the expander cycle more powerful and thus able to match the high pressures of the preburner-turbine combo.

Stage 2. Well yes, I assumed THAT. Okay, yeah, that tracks. That would make the flat thing over on the right the PAF.

Any clue which part of the rocket is which? What are we looking at? Presumably this is the second stage, but is this just one tank or both tanks? And is the aft end on the left or the right?

Just for some added fun... Someone on NSF saw this idea and remarked that tripropellant designs can also be good for a Thrust-Augmented Nozzle, where additional propellant and oxidizer are injected into the nozzle extension at launch in order to fully fill the nozzle (preventing overexpansion) and add thrust. Once you get high enough and specific impulse becomes more important than thrust, you shut off the additional injection; at altitude you can fill up the whole nozzle without overexpansion. I proposed this configuration: Here, the ORSC preburner+turbine pump only the LOX, while the closed hydrogen expander pumps the hydrogen along with the small amount of propane needed to operate the preburner. Using propane for the preburner is better than using hydrogen because propane is more dense and so the preburner will have more power this way. In the low-thrust, high-efficiency mode, both of the valves are closed, and so all of the preburner gases flow through the first turbine stage and directly into the combustion chamber. However, in high-thrust, augmented injection mode, the valves are open and so a portion of the preburner gases flow through a second turbine stage. This increases the power output of the LOX turbopump, and so you still have the same amount of gases flowing into the combustion chamber, but you also have extra gases that are coming out at lower pressure. Those lower-pressure gases can be injected into the nozzle. The boost pumps (not shown) will give enough pressure to inject the corresponding amount of propane as well, since the pressure down there is pretty low.

I could only speculate, but I would think there are a few reasons why the milestone would ask for a demonstration of LOX transfer: This is still baby steps, so you only need to show one thing at a time Starship (like all launch vehicles) carries proportionally much more LOX than fuel, so there would be more LOX left over at the end to work with All launch vehicles use LOX, so demonstrating LOX transfer is more broadly applicable than demonstrating transfer of fuel, which may differ from vehicle to vehicle As a medium-deep cryogen, LOX occupies a temperature situated between liquid hydrogen on the one hand and liquid methane or kerosene on the other hand, and so a demonstration of cryogenic LOX transfer is more likely to be broadly applicable to other propellants than an outlier

The contract only requires a demonstration of a LOX transfer: SpaceX of Hawthorne, California, $53.2 million Large-scale flight demonstration to transfer 10 metric tons of cryogenic propellant, specifically liquid oxygen, between tanks on a Starship vehicle. SpaceX will collaborate with Glenn and Marshall. From here. So they could have a methane tank inside the fairing to demonstrate transferring both, but it's not necessary.

Cams are now showing that Ship 26 in fact has three tanks: So clearly a prop transfer demonstrator. Although that doesn't mean it couldn't also be set up as a depot. The ability to transfer propellant between vehicles is not primarily an issue of docking (which is comparatively quite well-understood and well-demonstrated) but an issue of controlling fluid flow in microgravity. If they can transfer between two tanks in the same vehicle, they can hook up docking connections and transfer between two tanks in two different vehicles.

Example: This is almost a staged-expander staged combustion cycle, because the expander cycles run in series.

I calculated payload without any nose cone at all. I only saw these calculation for the two stage. Right, because I was analyzing a hypothetical expendable second stage (for depot purposes, etc.) with fewer engines and no nosecone. None of this changes the fact that (a) Ship 26 does have a fairing, and (b) with six engines, a fully-filled Starship can barely get off the ground and would have extraordinarily high gravity drag losses.

Thank you for taking my half-baked musing and (hah) expanding on it. Here you go, a power cycle diagram. I ended up putting the LOX and propane pumps on a single shaft attached to the ORSC preburner turbine, rather than shoving two separate turbines into the mix. You can eliminate a tricky fuel-ox shaft seal on the boost pumps by removing the LOX boost pump entirely and running a separate LOX boost pump off a simple tap-off downstream of the LOX turbopump. You could go all the way and move the propane turbopump off the preburner turbine and onto the hydrogen expander turbine, but that would tend to limit chamber pressure because the hydrogen is having to do considerably more work than before, and it will have a tough time competing with the high pressure capabilities of the ORSC preburner. However, if you do a low-performing version of the ORSC preburner and a high-performing version of the much simpler expander cycle, this could work.

This is not a modern reference; it's from 1983. But it's by Aerojet under a NASA contract so it should be pretty reliable for that era. In the results summary on page 4, it says that coking started at under 500°F and that higher propane purity reduced the coking rate but did not reduce the coking threshold temperature, suggesting that it is an issue with propane itself. I haven't had time to read the rest of the paper but I'm assuming it talks a lot more about propane as a fuel type.

That's been done upthread. I calculated payload without any nose cone at all. Ship 27 has a dispenser so it is expected to be an expendable "get something to orbit" demonstrator. Ship 26 has no dispenser and no payload bay door so it won't be used to launch any payloads. It is likely to be a prop transfer demonstrator, but its role may also be expanded to a prototype prop depot demonstrator if the prop transfer works. If they start adding more external stuff like extra thrusters, etc., then it could suggest the latter.

I see what you did there. There's something just delightful about the problem being a lack of sufficiently-aggressive metallic chemistry and the solution being "staple it!" Hmm, I don't think I've heard of this. Link? Did a little digging on this. You are correct that propane does have the significant advantage of having kerosene-level density (when subcooled) with a theoretical specific impulse only about 2% lower than methane. A further advantage is that it can share an uninsulated common bulkhead with LOX because its subcooled temperature is comparable to ordinary LOX. However, there's a problem: it cannot readily be used for cryogenic cooling. Properly understood, regenerative cooling in rocket engines is a supercritical open-loop organic rankine cycle. In an ordinary rankine cycle, an incompressible liquid coolant is pressurized by a pump (requiring very little work because it starts in its liquid phase), forced into a boiler that turns it into a "saturated vapor" (a fluid at equilibrium below the critical point), allowed to expand almost isobarically through a turbine to perform work, and then condensed at constant pressure to return to its liquid state and make the loop again: The total work done by the system is the work extracted from the turbine minus the work done by the pump. The rankine cycle has advantages over other thermodynamic cycles because the turbine operates entirely in the "dry vapor" phase while the pump operates entirely in the "incompressible liquid" phase, limiting the overall change in pressure. The phase change maximizes the amount of heat that the fluid is able to accept. An open-loop rankine cycle uses a working fluid that is exhausted after the expansion and thus is never recondensed. In a supercritical rankine cycle, however, the fluid is allowed to be heated beyond its critical point to become a supercritical fluid that is neither liquid nor vapor. This dramatically increases the amount of work that the system can do: In a supercritical rankine cycle, there is no bubble formation: the transition from liquid to supercritical fluid is smooth. The main disadvantage, of course, is extremely high pressurizes and thus high material stresses. So why can't propane be used in a supercritical open-loop rankine cycle? Unfortunately, the three-carbon chain of propane will coke when forced into its supercritical phase, even at temperatures as low as 500°F. Trying to operate a regeneratively-cooled propane rocket would require keeping the fuel below 500°F which seriously limits the amount of work that can be extracted from an expander cycle. Due to the coking issues, propane is similarly ill-suited for fuel-rich staged combustion. However, the addition of hydrogen to the mix could prove interesting. Propane works beautifully for oxidizer-rich staged combustion, after all. Its ideal mixture ratio with LOX is a whopping 4.5:1, significantly higher than methane's 3.8:1, further increasing the overall bulk density of the propellant mix to offset the addition of (extremely fluffy) hydrogen. I'm imagining an engine with a single oxidizer-rich propalox preburner, two staged-combustion turbopumps, and two expander-cycle turbopumps: one high-temp, one low-temp. The propane-oxygen preburner exhausts into two separate turbines, one to pump the LOX for the entire engine and one to pump the propane. The combustion chamber and nozzle are cooled entirely by liquid hydrogen operating in a high-temperature expander cycle, while the preburner is cooled by a low-temperature propane split expander cycle that operates the boost pumps, allowing the propane to enter the combustion chamber in a supercritical state and thus have improved combustion. Because all of the hydrogen expander cycle's energy goes to pumping liquid hydrogen, it can operate closed at a power level comparable to a split bleed (open) expander cycle. You get the high chamber pressure of staged combustion plus the high specific impulse of hydrogen plus the maximum regenerative cooling power uptake. Because the propalox bulk density is so high, you could theoretically get an overall bulk density similar to methane but at significantly higher specific impulse due to the hydrogen you're adding to the system. Now to come up with a flow diagram...

The RD-701 was planned to be used on the MAKS spaceplane, which (being air-launched) would have had lower TWR requirements (around 1.2 instead of the 1.4-1.5 that is preferable for ground-launched kerolox rockets). It was supposed to be able to separate from the top of the double-tailed carrier aircraft with a separation mass of 275 tonnes, deliver an 8-tonne payload to LEO, drop its expendable fuel tank, and return for a winged landing. Using the Launch Vehicle Performance Calculator from Silverbird and imagining a ground-launched expendable SSTO, I get a modest 2.8 tonnes to LEO from the Cape. That's if you stack a shorter version of an Atlas V core on top of a shorter version of a Delta IV core and bolt a single RD-701 to the aft end with a TWR of 1.4 or thereabouts. Hydrogen is such a great working fluid. Using it for the power cycle is either a free lunch or a terrible idea, and I'm not sure which.

Here's a tweet showing how it was mounted and what the power cycle diagram looked like: Detail view of the power cycle: The engine used three separate-shaft boost pumps, two separate kerolox oxidizer-rich preburners, a separate-shaft high-pressure hydrogen pump operated off of one preburner, and common-shaft high-pressure oxygen and kerosene turbopumps operated off of the other preburner. The high-pressure hydrogen flow was sent through a cooling loop before being injected into the combustion chamber and also sent a little of the hot hydrogen back upstream to operate the hydrogen boost pump in an expander cycle (the expander turbine downstream still had a higher pressure than the boost pump downstream, so it could flow back into the boost pump downstream readily). The kerosene boost pump did the same, except that it didn't also pick up extra heat. Curiously, the LOX boost pump was operated not by high-pressure LOX tap-off, but from the actual preburner exhaust tapoff, meaning that hot oxygen-rich exhaust was being mixed back into the cryogenic liquid oxygen flow upstream of the turbopumps. Initially, both preburners operated at full throttle. However, as the kerosene tanks started to run low, one preburner (the one on the right in the diagram above) was throttled down significantly while the other stayed approximately constant. As a result, the flow of hydrogen to the engine was reduced by 7% but the LOX flow was reduced by 60% and the kerosene flow into the chamber was extinguished. However, since the entire power cycle was provided by the oxygen-rich kerolox preburners, both preburners stayed "on" for the full flight; the "throttled-down" turbopump needed to at least provide enough power to continue pumping the LOX for the "full thrust" turbopump as well as pump the kerosene it continued to use. This at least meant that the oxidizer flow coming into the engine was always an oxygen-rich gas, which helped maintain consistent combustion even though the chamber pressure was cut almost in half. Note that in the diagram above, it is difficult to see that the liquid hydrogen (Fuel2) is actually flowing into the combustion chamber, but it is. Officially, the dual-chamber engine had four preburners, seven turbines, and nine turbopumps (two of which were common-shaft), because while there was only one set of boost pumps downstream of the tank outlets, the two chambers had separate preburners, turbines, and pumps. Despite Due to this complexity, the whole engine assembly came in at just 1.9 tonnes a hefty 3.7 tonnes and boasted a whopping 3.2 MN of sea level thrust, for an incredible a reasonably serviceable TWR of 170:1 88.6:1 in Mode 1 with a sea level specific impulse the same as Raptor 1 and slightly better than Raptor 2. And in Mode 2 it had better vacuum specific impulse than the RS-25s despite requiring a trickle of kerosene the entire time. Truly a marvel of design and engineering. Too bad it never flew. Yes, the heat capacity of LH2 is really quite impressive. I wonder if there would be a way to do an expander-cycle tripropellant engine to take more advantage of hydrogen's heat capacity. EDIT: Fixed the dry weight and corresponding calculations, which Wikipedia had wildly, wildly wrong.

As long as the structure can be in a recessed position (like the current lift points) then presumably it will be safe enough from the re-entry plasma. The trick will be constructing a pop-out load structure that will give enough clearance between the tiles and the catching arms.

1.5 stage refers to a rocket that gets you most of the way there and a lightweight kick stage to finish, correct? No, that would still be a TSTO. As @mikegarrison explained above, a 1.5 stage rocket (also referred to as a "stage and a half design") has a single stage which fires from the launch pad all the way up to orbital insertion, but is also supported by one or more boosters. Examples here are the original R-7 that put Sputnik 1 in orbit (core that went to orbit, boosters that didn't), the Shuttle (operationally it used the OMS engines to circularize, but it COULD have burned the RS-25s all the way to orbit), the Atlas LV-3B that put the Mercury capsules in orbit (technically the "booster" here was just a pair of engines but close enough), and the Long March 5B (four boosters plus a core that goes all the way to orbit and then tends to drop back uncontrollably). A fuel-switching tripropellant engine is attractive because it can start with high thrust and lower specific impulse in a sea-level nozzle, then switch to lower thrust and higher specific impulse, using the same nozzle as a vacuum nozzle. Basically an altitude-compensating engine (as opposed to an altitude-compensating nozzle). If you wanted to build a reusable spaceplane, you could use this kind of engine and have it launch vertically with a couple of fairly modest strap-on boosters, and it would be able to have both high TWR at liftoff and high specific impulse at orbital insertion.

Here's what the current lift points look like: No way they can insert a catch rod into that upper load point using the grabber arms precisely enough while it's hovering. They've gotta have some plan for a pop-out catch pin of some kind.

One other note -- Ship 27 has no heat shield and has a functioning Starlink Pez dispenser, so it is generally expected to function as an expendable Starlink 2 demonstrator. It will not, however, be launched as a single stage. Also, no Starship has ballast tanks.

You will enjoy this short story very much. A relevant excerpt: A fuel-switching tripropellant engine has always been one of my favorite proposals for a 1.5-stage-to-orbit reusable launcher.

I had used 85 up thread. But using the claimed 250t payload with SH expended (which btw mirrors F9 numbers with a ~40% payload loss for RTLS) we'd have a 55t vehicle in LEO with 250t of residual props. That's 6349 m/s dv. Lunar landing from LEO is nominally ballparked at 6300 m/s. 85-90 tonnes is the generally-accepted figure for the bones-dry mass of the fully-reusable Starship. With reservation of deorbit and landing props the effective dry mass goes up to 115-120 tonnes. I was going with 55 tonnes in order to account for a chomper fairing (for an expendable version) or for insulation and extra tanks/systems (for a depot). I tend to think that Elon's aspirational 40 tonne version would basically be balloon tanks. Not realistic for most applications.

Even with six engines, the T/W ratio of a fully-fueled Starship is barely more than 1:1 at sea level. There are 1200 tonnes of propellant plus 40 tonnes dry mass plus 5 tonnes for the three extra engines is 1245 tonnes. Each Raptor 2 can lift 230 tonnes, with the vacuum-optimized Raptors being overexpanded (and thus under-thrusty, around 213 tonnes thrust) at sea level, so 1329 tonnes of thrust, or a T/W ratio of just 1.06:1. Gravity drag will be huge. And that's without a fairing (which Ship 26 has) or payload. Also, keep in mind that what you say about lower tank mass probably already applies; Elon's 40 tonne number was aspirational I'm sure it could get to orbit, although unclear whether it would get to orbit with more or less payload. A one-way or disposable Starship would have lower dry mass already because you don't need wings or a heat shield. Let's say 55 tonnes dry mass for the sake of this thought experiment. Let's say that a Starship with all six engines will fire all six together for the first half of the ascent and then fire only the vacuum engines for the second half of the ascent, leading to an average specific impulse of 372 seconds and a thrust at separation of 15.28 MN. Let's further say that your putative four-vacuum-engine Starship will fire all its engines for the whole ascent, leading to a constant specific impulse of 375 seconds and a constant thrust of 10.35 MN. In addition, the vacuum-engine Starship will have the same first stage characteristics but will have a second stage dry mass that is 3.2 tonnes lighter (52.8 tonnes). Let's say that the dry mass of Superheavy is 300 tonnes and needs to reserve 25% of its propellant in order to do a RTLS boostback and landing. Even if these numbers aren't quite right, it won't matter because we're doing a comparative analysis. Using these numbers, the Silverbird Astronautics calculator gives an estimated payload of 111 tonnes for the 4-engine Starship and a payload of 115 tonnes for the 6-engine Starship. So it looks like the gravity drag from not having all six engines would likely outweigh the advantage of lower dry mass and higher specific impulse. Slightly different booster numbers or dry mass numbers might reverse this, but in either case it's not going to be a very significant difference. For a fuel depot, it doesn't really matter, because the depot isn't going to be moving around enough for its dry mass or average specific impulse to make a difference. For an ultra-high-dV one-way mission (like a flagship payload to the outer planets), gravity drag won't really be an issue because you're going to be refilling in LEO anyway. So you'll probably want to only use a three-engine or even a two-engine version in that event.