They’re building one and have another one most-of-the-way-built, which is why I used the phrase “in operation”.
Liaoning - in operation
Type 001A - undergoing sea trials
Type 002 (rumored to be a larger, more-capable CATOBAR carrier) - under construction
I agree. Cranes on-hand for all of them. And ADCAPs in case the cranes fail to do the job.
I went through the USS Razorback, a WWII sub that underwent the guppy mod, was sold to Turkey and eventually bought by the city of North Little Rock as a museum ship. There was a mock nuclear torpedo in the aft torpedo room. If the balloon went up, they would fire it at whatever they were shadowing and run like hell.
They’re apparently trying to figure out how to recover the tipped-over booster. Current plan appears to be to tow it back into port as-is, and then use the crane there to fish it out of the water.
Looks surprisingly intact. The only obvious damage is a large hole through the interstage. I still don’t expect them to reuse much off the vehicle, but it looks to be in good enough shape to tow back.
The full landing looks like some of my better landings in Kerbal Space Program. Just thee right amount of wobbliness, only to straighten things out at the last second.
The “bathtub curve” of reliability applies specifically to non-serviceable component failures with potential latent defects where there are a few discrete failure modes that dominate the infant mortality and wear out failures. It is not a generic model of reliability with applicability to complex systems which have many potential failure modes along their lifespan and are capable of being inspected or serviced to detect latent defects or nascent failures and to bring them back to a fully serviceable condition indefinitely. Any car company that accepted the “bathtub curve” of reliability as part of their normal way of doing business would quickly go defunct (e.g. AMC) because nobody is going to buy or recommend a car from a manufacturer that dies within a few hundred hours of service.
Conventional (rocket) orbital space launch vehicles are not like any other terrestrial vehicle or transportation system; they experience environments and stresses that are well beyond anything seen by even the most high performance non-experimental aircraft; they operate for a few hundreds of seconds in ways that defy good sense or any ability to recover from all but a fraction of potential failure models; their performance is ultimately rated not on a cost basis of the flight itself but the potential for executing a successful deployment of multi-(often many hundreds) of million dollar payloads; and the economics of reuse are not as simple as just amortizing the initial build cost across however many missions the vehicle can fly before it is retired or plows into the ocean pointy-end first, because so much of the cost isn’t in materials, construction, or propellants (the fuel is an almost negligible cost of any launch) but in the labor that goes into integration and testing to make each launch successful. Even if SpaceX can manage tens of launches per each first stage booster without major rework (a capability that has yet to be demonstrated) it isn’t clear that it will dramatically reduce their launch costs because the actual costs are elsewhere in labor, facilities, et cetera. What it may allow them to do is maintain a higher launch tempo than the rate at which they could build all new first stages, which is an advantage upon itself, but that doesn’t mean it will translate into radical reductions in launch cost.
And the US Navy doesn’t have “accidents” or “failures”; only “anomalies”. Manipulating the language to mean something other than what it clearly means does not fool any intelligent person. SpaceX had a non-mission-critical failure in the return phase of their first stage. It’s not a big deal, and I’m sure they’ll learn something useful from it, as well as how to recover and salvage components for reuse, so it is at most a very minor setback with opportunity to learn. But it was not a nominally successful recovery or a controlled landing of the booster
Stranger
Here is a cool video of the recent … ummmm … water landing, from all four perspectives synchronized. Also, the booster is back on dry land. You can see some damage here and here
So good to see you Stranger!
Agreed. I missed your posts.
It’s been a while since the last BFR update. For one thing, they’ve changed the name: the upper stage is now called the Starship, and the booster is the Super Heavy.
An enormous change is that they’ve switched materials. Previously, they had been using carbon fiber, to the point of building giant 9-meter mandrels and putting in orders for large quantities of material. That is, apparently, discarded work.
The new material is stainless steel. SS has a venerable history in aerospace. The early Atlas rockets were made from stainless steel “balloon” tanks; they were highly mass-efficient though came with the downside that they required pressurization to simply remain upright; if lost, the entire stack would collapse.
One might think that, being an old material, SS is inferior to carbon fiber. But it carries two significant advantages. First is that it gains 50% strength without losing ductility (i.e., it doesn’t become brittle) at cryogenic temperatures. Since the launch config is cryogenic (liquid oxygen and methane), this is significant–SS becomes more mass efficient than CF.
Second, and perhaps more importantly, SS has amazing heat tolerance. Since the *Starship *is reusable, it has to survive orbital reentry–a rather warm process. SS maintains its strength up to 1500 F, which means you can just let it get super hot and lose most of the reentry heat via radiation. The remainder will be gotten rid of the same way as rocket engine nozzles–by flowing propellant in tubes (and then venting it via perforations). Musk goes over this stuff in a bit more detail in this article.
They do plan on avoid the Atlas problem and will not require pressurization. The increased strength from the double wall on the hot side will help; they’ll probably need additional structure elsewhere as well. Perhaps some of the reinforcements will use something other than SS since it won’t be exposed to high heat.
The plan for the Raptor engines has not changed–they are still full-flow staged combustion methane. And as they mentioned previously, the early iterations will have the same design for sea level and vacuum. Normally you put a large nozzle on the upper stage for better efficiency in a vacuum, but for a high-efficiency engine like this it matters a bit less and they can get away with a single design.
They are using one trick–normally, the reason you can’t put on a large nozzle at sea level is that the nozzle can’t expand the gas all the way to ambient pressure, and you then get separation between the combustion gases and the nozzle wall. This separation is unstable and can damage the engine. But the Raptor puts a small step in the nozzle where they want the separation to occur, and this makes the separation predicable, eliminating the instability.
There are now some pictures of the full-size engine. It’s quite a good-looking engine, IMO.
Oh, and they also put together a prototype for the initial “hopper” version of the Starship. It went together in weeks since they just hired a water tank company to weld together the stainless sheets (which don’t require any advanced handling). The final version will presumably be made with slightly more care, but that doesn’t matter much for initial hop testing.
All in all, they seem to have simplified their design massively with the switch to SS. It’s unclear that they ever would have solved the heat shield problems with CF, since it has such little heat tolerance. And steel has the side advantage that it’s easy to repair in the field–say, on Mars. While it’s slightly annoying to see them throw away work, it’s good to see that they aggressively avoid the sunk cost fallacy–they do not hesitate in switching to a better path even if it means losing past investment.
Some videos from their latest Raptor test firing. Some beautiful Mach diamonds in the second video.
One nice detail (that was known about, but it wasn’t clear if/when they would implement it) is that they’re using an electronic ignition system. Specifically, a “methalox torch igniter”; basically a spark-ignited flame, fueled by the same methane/oxygen propellant, which is then used to start the rest of the engine.
In contrast, the Merlin engines on the Falcon 9 use a chemical igniter called TEA/TEB, which spontaneously burst into flame when mixed. It’s reliable, but has the disadvantage that you have to carry a separate reservoir for the chemicals and can only start the engine as many times as you have decided in advance. The electronic system on the Raptor doesn’t require any special chemicals and can start an indefinite number of times.
Raptor testing has reached their initial design power levels.
This is extremely impressive. In particular, the 257 bar chamber pressure is on par with the best Russian engines ever built (their best series–the RD-170 and further variants such as the RD-180 are in the range of 250-262 bar; in contrast, the best American engine, the RS-25 Space Shuttle engine only reached 206 bar). Musk says that with subcooled propellants, another 10-20% is possible–which means they may reach the 300 bar they hoped for early on in BFR development (the target was later lowered to 250 bar).
And this is all with a sophisticated engine cycle, which is absolutely unique for their propellant combination: only two other full-flow staged combustion engines have been built, neither burning methane or ever making it off the test stand. If Raptor flies, as I have little doubt it will, it will be the first flight-worthy full-flow staged combustion rocket engine.
Thanks, Dr. S, firvthose helpful summaries and observations. Avoiding the sink cost fallacy, indeed — a lesson for us all.
Something to ponder: Raptor fuel and oxidizer turbopumps combined produce 3100000Hp.
The booster will have 31 Raptors.
Hoover Dam produces 3000000Hp
:eek:
Another thanks to Dr. S for his posts; you totally fucking rock, dude!
In case anyone else (besides me) got confused by this: the turbopumps on each Raptor produces about 100,000 hp. The turbopumps on all 31 Raptors on the booster add up to 3.1 million hp (2300 megawatt).
How many engines did the Russian moon rocket that blew up on the pad have?
The N1 had 30 engines on the first stage. However, note that the Falcon Heavy has 27 engines lit at once, and it worked perfectly. Even had a couple of engines failed, it would have succeeded. The N1 suffered from many things, among them: an inability to perform a static fire on individual engines; likewise an inability to do a static fire on the full stack; and generally a poor ability to model the behavior of the full setup. The N1 actually failed four times (out of four attempts), though only one of them obliterated the pad.
Glad people are getting value out of my posts. I plan on writing something up about the meaning and value of full-flow staged combustion in engine design, but am open to suggestions.
Please do! (If and when you can spare the time). I was wondering about just that. Of course we could look it up, but since you mentioned it…
It’s going to be tough to strike the balance between introduction and technical detail, but we’ll see how it goes. Also, while I’ll try to avoid too many “lies for children”, some amount is probably inevitable.
To start, there is an incredible variety of rocket engines, and to limit the scope of the discussion I’m going to only cover liquid-fueled cryogenic bipropellant engines. Fortunately, most orbital rockets use this style of engine, at least in part. Let’s unpack what the name even means:
liquid-fueled: The reaction chemicals are fully liquid and held in tanks in the body of rocket. They get pumped to the engine in some fashion.
cryogenic: At least one of the reaction chemicals is a gas condensed to a very cold liquid.
bipropellant: There are two different reaction chemicals. Almost always, one of them is an oxidizer. And for a cryogenic rocket, that oxidizer is almost always liquid oxygen. The other propellant is generally a fairly ordinary fuel: kerosene is one type, and liquid hydrogen is also common, but in the past gasoline and alcohol were also used. SpaceX’s new engine uses liquid methane, which is just purified natural gas.
For short, I’m just going to call these liquid engines. The fundamental thing that every liquid rocket engine has to solve is pumping the propellant into the combustion chamber. One might intuitively think that pumping liquids around is not a big deal–a minor implementation detail–but they are really almost everything. As The Vorlon noted above, the pumps on a single Raptor engine are 100,000 horsepower. On some engines, it’s over 250,000 horsepower.
Nevertheless, it is possible to start simple. The easiest engine designs are pressure-fed, which is exactly what it sounds like. There is no actual pump; instead, the propellant tanks are pressurized and valves are used to direct the flow to the engine. This style has a few downsides, as one might expect: it is heavy due to the extra tank reinforcement and bottles for pressurant gas, and the performance is not great. It is most widely used for small engines that need to be reliable, such as maneuvering thrusters. Nevertheless, pressure-fed engines are used on occasion for orbital rockets, such as the upper stage of SpaceX’s own Falcon 1. It has another interesting advantage that it scales almost indefinitely: the Sea Dragon was a design for an absolutely monstrous rocket (much larger than even the BFR) using pressure-fed engines. It made up for the low performance with sheer size.
But to really get high performance, one needs a true pump. And the most power-dense pumps around are turbopumps, which are not so different from the turbocharger on a car: they use a turbine driven by moving gases to drive a shaft, which drives another turbine that does the pumping action. Almost all liquid engines use a turbopump, though there are exceptions.
That raises some questions, though–turbopumps require a fluid stream to drive the input side of the pump. Where does that come from? And what, precisely, are we pumping?
The simplest turbopump design is what’s called a gas generator. The gas generator refers to a small combustion chamber separate from the main one. Its purpose is to combust a small proportion of the propellent into hot gas, then sending it through the turbopump to power it. The turbopump itself generally has two pumping sections: one for each of the propellants (liquid oxygen and, say, kerosene). The outputs mostly go to the main combustion chamber, with a small amount redirected back to the gas generator.
Gas generator designs are widely used–the F-1 engines on the Saturn V were that design, as are the Merlin engines on SpaceX’s Falcon series. They are reliable due to their relative simplicity, and have a high power-to-weight ratio. The downside is that while they are more efficient than pressure-fed engines, they are still quite wasteful. In particular, the waste gases from the gas generator are generally thrown overboard. Because the gas generator operates at a relatively low temperature (so as to keep things from melting), the exhaust gas isn’t very energetic, and can’t itself be turned into a reasonable amount of thrust. Rocket engines get their efficiency from throwing as much mass out the end as fast as they can, and for the engine to just dump waste gas like that means there is room for improvement.
I have to make a small aside. One key metric in engine design is chamber pressure. High chamber pressure has a number of benefits relating to efficiency. For one, it gives a compact design for a given power level. In the end, rockets are just squirting gas out a nozzle, so the higher the pressure the higher the force. High pressure also means we don’t have to use huge nozzles: another key factor in engine efficiency is the expansion ratio, which is just the ratio in area between the throat of the combustion chamber vs. the end of the nozzle. A higher pressure means we can have a smaller throat for the same propellant flow, which gives the nozzle a higher ratio. There are also some benefits related to combustion efficiency.
Getting back to gas generators–because they are wasteful, increasing the chamber pressure only has limited benefits. The pumps have to overcome the chamber pressure, so they need to become more powerful. But the pumps are themselves inefficient, and become a net burden at a certain point. Therefore, we find that gas generator engines tend to have lowish chamber pressures and lose efficiency from that as well.
Enter the staged combustion cycle. Staged combustion answers a question we might have about gas generators: why don’t we just take the gas generator exhaust and pump it into the combustion chamber? Then we aren’t just dumping it overboard. Furthermore, if the exhaust is going to the combustion chamber anyway, why don’t we push as much mass through the turbine as we can?
And so for a “standard” staged combustion engine, what you do is this: pick one of either fuel or oxidizer. Send all of that propellant’s flow into a preburner (similar to a gas generator). Divert a small amount of the other propellant to the preburner and combust it. Then send the entirety of the output to the combustion chamber.
The Russians mostly picked oxygen, and so their engines are called “oxygen-rich staged combustion”. All of the oxygen is sent to the preburner, where a little kerosene is mixed in and combusted. The result is a flow of hot, pressurized oxygen and some combustion waste (water, CO2, etc.). The remaining kerosene is pumped straight to the engine.
Because there is so much mass flow, and because we aren’t wasting the output, the turbopump can run at extremely high power levels (250,000 horsepower for the RD-170!). And because of this, the chamber pressure can be extremely high (250+ atmospheres, also called “bar”), leading to the efficiency improvements I mentioned above.
For the Russians, the difficulty was in the hot oxygen flowing around. With those pressures and temperatures, just about everything combusts (including any ordinary metal). However, they solved the problem with advanced metallurgy. American engineers, when they found out about the oxygen-rich engines, were almost incredulous. For kerosene-oxygen engines, the Russians have held a lead over American designs for decades.
Why not run a fuel-rich cycle, then? That avoids the advanced metallurgy while retaining the advantages of staged combustion. For kerosene engines, the problem is that unlike oxygen which vaporizes cleanly, kerosene is a mixture of a variety of hydrocarbons, and the hot exhaust ends up as a mess of products which clog up the works (often called “coking”). Fuel-rich staged combustion does work with hydrogen rockets, though, and so the RS-25 (SSME, “Space Shuttle Main Engine”) was a fuel-rich hydrogen staged combustion engine (a very efficient one at that).
So what is the next step? The normal staged combustion cycle captures all of the preburner exhaust, so surely there’s not much left to be had. But note that above, we picked either a fuel-rich or oxygen-rich system. Why not both? And what does that give us? Enter full-flow staged combustion (FFSC).
For one, we can use the entire mass flow of the propellant. The turbopumps are run by mass flowing through them. Although they are ultimately driven by the energy of the flow, the more mass we have, the slower it can go. We can use that extra margin to increase overall power levels, or make conditions milder (good for reusability), or some combination.
For another, we get better combustion efficiency when both propellants have turned to gas. Liquid+liquid mixing (as with a gas generator) is tough and requires sophisticated injector design. Liquid+gas mixing (as with partial staged combustion) is substantially better, but still difficult. Gas+gas mixing is very efficient and relatively easy to get right.
Yet another advantage relates to turbopump design. Turbopumps require seals to prevent gases from the driving side to make it to the pumping side. On an oxygen-rich SC engine, those seals have to be very good–we can’t have any of the oxygen-rich hot preburner exhaust making it to the fuel flow. The oxygen would immediately combust with the fuel, making the whole thing explode (turbopumps aren’t designed to support combustion themselves).
But with FFSC, the fuel pump is driven by the fuel-rich exhaust. This exhaust is hot but there’s no oxidizer present. So while we don’t want too much leakage, a small amount is fine and won’t harm anything. The seals are thus less critical. Likewise, the oxygen pump is driven by oxygen-rich exhaust. There’s no uncombusted fuel present, and some small leaks past the seals aren’t critical.
The obvious downside to FFSC is that in terms of design, it is the worst of both worlds: it requires the advanced metallurgy of oxygen-rich designs, but like fuel-rich designs is incompatible with fuels that don’t properly turn to gas. However, SpaceX is using methane for the Raptor, which isn’t quite as clean as hydrogen, but leagues better than kerosene, and thus shouldn’t pose any coking problems. As for the metallurgy, they seem to have solved the problem (the knowledge does seem to have slowly leaked out of Russia).
Because of the complexity, only two other FFSC designs have made it to test firings. One was the Russian RD-270, which was not cryogenic and used toxic propellants which are out of favor these days. The other was the Integrated Powerhead Demonstrator, which as you might guess was never more than a prototype. It was also hydrogen-only. SpaceX’s Raptor appears to be the only FFSC with a future to it.
Overall, Raptor’s FFSC design is critical to the success of the Starship and Super Heavy. The methane fuel allows FFSC to be possible at all, while also supporting Mars landing missions. SpaceX needs both the efficiency and performance that FFSC provides, alongside the high level of design margin they need for long-term reusability. Almost any part of the system could be changed except for Raptor, so it is very good to see how much progress they’ve made.
One minor point of clarity: The F1 used the pre-burner exhaust to cool the engine bells, it wasn’t dumped overboard like with the Merlin. If you look at the high speed Saturn liftoff films, that is the reason the exhaust of the F1’s was black coming out of the engines.
