sevenperforce

Not only that, they're gonna try full-flow staged-combustion. Which @Exoscientist thinks is a poor choice because of cost, complexity, and schedule. The issue is that for optimizing any two-stage launch vehicle, the first stage liftoff TWR is a more significant factor than the first stage vacuum specific impulse. Playing funky games with nozzles can help with vacuum specific impulse, but it doesn't improve liftoff TWR. Increasing chamber pressures will improve liftoff TWR while also giving better vacuum specific impulse as a bonus.

As I pointed out above, the SSME (the RS-25) achieves high vacuum specific impulse because it is a vacuum-optimized engine. It has an adaptive nozzle which allows it to fire safely at sea level, but at the expense of significant thrust losses. Anyhow, you're also comparing apples and oranges. First, you're talking about staged combustion specifically, not closed cycles generally (closed expander cycles, for example, tend to have very low chamber pressures). Second, "closed cycle" and "high chamber pressure" are not linearly related. The closed-cycle RS-25 boasts 206 bar, almost double the chamber pressure of the gas generator RS-68, but the closed-cycle BE-4 is only 17 bar higher than the gas generator Vulcain 2. Finally, cost does not directly correlate. Where are you getting that number? In 2006, the open-cycle RS-68s were $20 million each -- about $31 million in today's dollars. And it doesn't correlate. The closed-cycle BE-4 is $8 million, far cheaper today than the RS-68 was twenty years ago. I'm really lost as to what this has to do with vacuum specific impulse, but regardless, intentionally testing a design to failure is not how you come up with failure rates. Merlin's 108 bars of chamber pressure would like very much to know who you are calling low pressure, and politely directs your attention to the closed-cycle YF-75 and RL-10A's ~42 bar. They already appear to be using a cluster of engines on the lower stage, so I'm not sure about your point here. The engines on their second stage are far, far too small to be used on an appropriately-sized first stage. The XRS-2200 occupied a forward area of 7.8 square meters, giving it a sea level thrust/area ratio of 117 kN/m2. The RS-68A has a sea level thrust/area ratio of 676 kN/m2. So no, it doesn't work, not for this application. It doesn't work at all. And that's before we even start talking about the horrendous dry mass problems.

The XRS-2200 engine showed you can get quite high vacuum ISP of even a first stage engine by using an aero spike. The XRS-2200 engine was built from the J-2S powerplant, which is not a first stage engine at all. The J-2 was developed from the ground up as high-energy upper-stage engine to replace the cluster of eight RL-10A-3S engines on the Saturn I S-IV stage. Because the RL-10's closed expander cycle cannot be readily adapted to produce a reliable engine with greater than 155 kN, Rocketdyne designed the J-2 to use a gas generator design to achieve around 53 bar chamber pressure and produce over 1 MN of thrust. The expansion ratio wasn't great -- just 27.5:1 and no nozzle extension -- but the 53 bar chamber pressure was respectable. It was never, ever a first stage engine. Firing it at sea level would have achieved only 200 seconds of specific impulse. By using a more simplified cycle with the same expansion ratio, the J-2S reached 15 seconds greater vacuum specific impulse than the J-2 and eliminated about 84 pounds of unnecessary mass, increasing the vacuum TWR from the J-2's 73.2:1 to just over 85:1. It was still never a first stage engine. The J-2X, an updated design with a proper vacuum nozzle extension, would have achieved 448 seconds of specific impulse. The XRS-2200, on the other hand, went back to the gas generator cycle of the J-2 but split the propellant flow up between ten separate combustion chambers. Smaller combustion chambers and a bigger nozzle meant that the XRS-2200 could achieve similar 436.5 seconds of specific impulse, similar to the J-2S but still far short of what could be expected from the J-2X's true vacuum nozzle. The aerospike nozzle meant that it could be fired at sea level without serious losses or flow instability and still achieve 339 seconds of specific impulse. That's something, but it's still much lower than the RS-68A's 363 seconds of sea level specific impulse. The biggest problem with the XRS-2200? It packed on the pounds. At 3,450 kg, it was more than double the mass of the J-2 despite only producing 15% higher vacuum thrust. The added weight came both from the small clustered nozzles, which suffered from the square-cube law, and the extremely heavy aerospike ramp nozzle. The sea level thrust to weight ratio would have been just 21.2:1, barely half of a true sea level gas generator hydrogen engine like the RS-68A. The first stage of a two stage vehicle gains thrust due to underexpansion as it climbs into vacuum. Again, using the RS-68A as an example, the specific impulse at liftoff is 363 seconds while the specific impulse in a vacuum is 412 seconds. Assuming a similar launch profile to the Delta IV Heavy, a first stage powered by the XRS-2200 would take a full minute to catch up to the specific impulse of the RS-68A. By that time the booster has burned a full quarter of its propellant. And you'll need twice as much engine to achieve the same TWR when you're using the XRS-2200 because aerospikes are so heavy, which means you either accommodate increased gravity drag or you accommodate increased dry mass. You can't just say assume that it will increase the overall performance; you actually have to look at a real-world example. For the record, anything past about 20 km is essentially vacuum for the purposes of nozzle optimization. But again, an aerospike engine is NOT a vacuum-optimized engine. An aerospike gets lower vacuum specific impulse than a vacuum-optimized engine and lower sea level specific impulse than a sea-level optimized engine, and it will be significantly heavier than either. If you want to make a proper comparison, you'll need to integrate over the entire first-stage burn sequence. The first stage starts at 0m/s and ends somewhere much nearer to, but short of, orbital velocity, so the actual time spent at lower altitudes compared to higher altitudes is much greater than the diagram you posted suggests, as it is simply the path taken and the horizonal axis is not time Actually working up the numbers is challenging. You have to pick a launch TWR and you have to use numeric integration. The thing you want to minimize is total Δv loss, where total Δv loss is the sum of gravity drag, aerodynamic drag, and specific impulse shortfall (also known as pressure drag), added to the Tsiolkovsky Δv shortfall resulting from dry mass differences. You can make it easier of course if you pick a known-optimized trajectory and choose the same TWR. An adaptive nozzle on a first stage that would give it the same vacuum Isp as a vacuum optimized nozzle and the same sea level Isp as a sea level optimized nozzle would improve the overall rocket performance if the Tsiolkovsky Δv losses were low enough. That doesn't exist. There is no engine, aerospike or otherwise, with an adaptive nozzle that achieves the same sea level specific impulse as a sea level nozzle as well as the same vacuum specific impulse as a vacuum nozzle. And even if it DID exist, you would have to also factor in dry mass increase with the Tsiolkovsky rocket equation.

The RS-25 does have a vacuum-sized nozzle. That's why it is able to get over 450 seconds of specific impulse in vacuum. That's also why it has such a low thrust/area ratio. If you look at an RS-25 in person, you'll see that the powerhead itself is very small compared to the nozzle. The only reason that the RS-25 can fire at sea level without ripping itself apart is that the end of the nozzle angles back in toward the center, creating a high-pressure region at the outer lip which prevents flow separation. Unfortunately this results in a pretty severe loss of thrust and specific impulse at sea level. The RS-68 gets a nice respectable 365 seconds of specific impulse at sea level despite being a gas generator engine, while the RS-25 only gets 366 seconds -- just a single second of specific impulse more -- despite having a vastly more efficient turbopump cycle and much larger nozzle. Aerospikes (especially multi-chamber aerospikes) usually suffer from being overweight, so combustion chamber pressures are kept low to avoid making the chambers too heavy; the designs rely on the large nozzle to get respectable specific impulse despite the low chamber pressure. Thanks to staged combustion, the RS-25 has a pretty phenomenal chamber pressure at 206 bar, compared to the paltry 58 bar of a design like the XRS-2200.

From 2010 through 2019, Falcon 9 and Falcon Heavy launched 428.3 tonnes to orbit, not including the Zuma launch or the NROL-76 (although both of those were RTLS, so fairly low total mass). From 2019 through 2022, Falcon 9 and Falcon Heavy launched 1,251.1 tonnes to orbit, not including a number of classified NROL launches or Transporter-2 through Transporter-5. So far this year, SpaceX has sent 434.0 tonnes to orbit using forty launches and is on track for a total of 100 total launches. So the total mass SpaceX has put in orbit is somewhere slightly above 2,114 tonnes. Keep in mind that many of these launches went below LEO, so they had lower gross mass but much higher energy. With a current capacity of 100 Falcon 9 launches per year and a per-launch demonstrated max payload of 17.6 tonnes with RTLS, SpaceX can put 1,7600 tonnes into LEO annually. Yep. Also what the F-1 engine used, what Rocketlab uses for its Rutherford engine, and what Firefly Aerospace uses to ignite its Reaver engine. Its application is more common with rocket engines than jet engines, actually. The JP-7 fuel used for the SR-71 is notoriously difficult to ignite, so that's why they skipped a conventional ignition plug system and went with TEA-TEB. The SR-71 could refuel in-flight so its persistence was typically limited by how much TEA-TEB it could carry and how many times it fired its afterburners.

Does Mvac use film cooling? Sorta looks like it. The sea level Merlin 1D dumps the gas generator exhaust overboard, but the MVac delivers the gas generator exhaust into the downstream portion of the nozzle. The purpose here isn't really film cooling so much as it is squeezing a little extra Isp out of the engine by increasing mass flow through the end of the nozzle (just like the original F-1 engines did), but it does do a little film cooling, and that's why it looks the way it does. To @darthgently's question: I believe that SpaceX switched the nozzle extension material from a metal alloy to carbon-carbon some time around 2019 but I could be wrong. If it's RCC, any green tinge would not be from copper. The initial green tinge comes from the use of TEA-TEB for starts and restarts, but that usually fades quickly. It might just be a camera artifact. Amazing.

Same. It wouldn't make any sense at all to use an aerospike on the first stage. This doesn't really make sense. There can't be a "longer spike" beneath the first stage because there isn't a spike beneath the first stage. And for the record, the term "aerospike" really refers to a truncated plug engine, because the entrained air under the plug acts like a virtual spike. What you're referring to is a non-truncated plug nozzle, which looks like a lance. But anyway, there isn't any spike beneath the first stage right now, so converting from conventional engines to an aerospike is nontrivial. There are seven engines on the first stage. Do you propose converting all seven to individual annular aerospike engines, or are you envisioning a single plug nozzle with the engine combustion chambers wrapped around it? If the former, how do you suggest Stoke solve the combustion problems that have plagued all annular aerospike engines to date? If the latter, how do you suggest Stoke handle vertical landing? Well, that's not exactly how an aerospike works. An aerospike isn't vacuum-optimized in vacuum; it is simply better than a sea level nozzle. Similarly, while an aerospike is operable at sea level, it is not nearly as efficient at sea level as a sea level nozzle, so you end up with a thrust shortfall and specific impulse shortfall at liftoff. This couples destructively with the enormous weight penalty of an aerospike design to increase gravity drag. In most designs this seriously obviates any advantages from slightly increased specific impulse in vacuum.

With the now-infamous federal ChatGPT lawyer case, the attorney asked ChatGPT for caselaw and was given fake cases with fake quotes and fake citations. After finding out that the court and opposing party were questioning the veracity of the research, the attorney went back to ChatGPT and asked "are these cases real?" and ChatGPT assured him that they were and generated an entire fake first page of the case. I've seen it do this in legal research before -- I asked it to search for cases with a particular kind of fact pattern and it initially said it couldn't find any, but when I prompted it again, it confidently spat out case names, dates, fact patterns, and quotes. All utter fiction. While we might be able to create some sort of a "no fake cites" system to solve THIS problem, that seems like a work-around to the inherent problem, which is the lack of a persistent coherence model. I'd like to see a modification of ChatGPT which was trained to ask meaningful clarifying questions. Maybe then you could get the kind of "two-part" system that would work around AI hallucination.

WHOAAAAAA They're doing differential throttle control after all! Go Stoke!

I remember getting my kids one for Christmas 2021. We had a lot of fun with it. We figured out pretty quickly that more water meant a longer thrust duration while less water meant greater liftoff thrust AND greater liftoff specific impulse. Optimized at around 2/5ths full I think.

The jury is still out (imo) on whether LLMs are merely stochastic parrots whose propensity for AI hallucination is intrinsic and thus fundamentally hamstrings their utility for generative communication, or whether LLMs can defeat the AI hallucination problem with the right training data and safety systems. Human speech processing is VERY similar to a LLM in many ways. In ordinary conversation, we don't plan out an entire sentence word-for-word before we start speaking; we have an idea and start talking and the sentence just...flows. LLMs work the same way, by predicting the next word in the sentence. However, we do have a mental model of the world and the concept we are trying to communicate, and so they speech we use -- assuming we aren't just speaking gibberish -- is going to flow along the pathways in our mental model that lead to that particular concept. That's the part which LLMs seem to lack. By design, they will take any path as long as it ends with the target concept, which is what leads to nonsense like making up completely fake legal cases. My best guess is that the "stochastic parrot" element which leads to AI hallucination is fundamental and thus inescapable from within the LLM system. If it's fixable, it will require a completely different kind of AI as an intermediate filter, one which intelligently checks the LLM output for consistency with reality and bounces it back to the LLM to re-generate if it's inaccurate. And that AI can't simply operate as an extended or different LLM.

That's what I was assuming from the title of the thread TBH

It should be noted that the 30% increase in power production capacity will be the result of the installation and deployment of a total of six new IROSAs, not merely the two IROSAs carried on this launch.

The question is if this is the same vehicle. It's probably a safe assumption. If not, well, all we can do is wait for more information. I'm pretty sure it's not the same vehicle. The new design has a completely different OML so they would design the interior from scratch. To @tater's earlier point, it's possible that they could have the engine(s) protruding into a bulkhead inside the crew cabin, like the New Shepard crew capsule. Not sure how much space would be required, though. According to this source, the new HLS lander is 16 tonnes empty and will have a mass of over 45 metric tonnes with full props. But from the lunar surface, it will only need about 2.6 km/s of dV to get to Gateway. Let's say 2.7 km/s for margin. Assuming the BE-7(s) can get into the ballpark of 445 seconds specific impulse, the liftoff mass from the lunar surface will be about 29 tonnes. Liftoff mass for the Lunar Module was 4.7 tonnes and the engine delivered 16 kN of constant thrust, delivering a relatively anemic two lunar gees (0.35 geeEarth). I'm sure the HLS architectures will need more than that. If we say three lunar gees, that's the equivalent of an Earth T/W ratio of 1.4, which is considered pretty sporty. So we'd be looking at needing ~145 kN off the lunar surface, meaning at least three BE-7s are required. If they have a cluster of three BE-7s at the center, then you're looking at needing a 2.6-meter-wide circle space for clearance: Should be enough to put some storage, at least. That's with a notional 6.5-meter OML on the lander, assuming they go right up to the max volume capacity of New Glenn.

Not exactly. The difference in many words between Japanese and Mandarin just be like "turn a corner and you're right there". For example "Your point is correct", in Chinese can be "说得是 (shuō dé shì)", and in Japanese is "そうです (sodesu)". There're lots of other examples like that. Because, well, of course, is inextricably linked to the deep cultural exchanges of all kinds between China and Japan and others where around China, both peacefully and not. That started from very long ago. If must find a similar example like this, I would say is kind like Britain and France. On the other hand, doesn't Japanese have a lot of loan words from English as well? Maybe the more accurate statement is that Mandarin is almost as different from Japanese as Japanese is from English.

They are much farther removed apart than English and Russian. The Sino-Tibetan language family consists primarily of Chinese (in most of its various versions), Burmese, and Tibetan. It has no known ties to the Indo-European language family, the Austroasiatic language family, the Austronesian language family, or the Altaic language family, which themselves contain the following: Indo-European: German, English, Russian, Hindi, Punjabi, Greek, Spanish, French Austroasiatic: Vietnamese, Khmer Austronesian: Malay/Indonesian, Javanese, Tagalog Altaic: Japanese, Korean, Turkish, Mongolian So you could say that Mandarin is as different from Japanese as Japanese is from English. In comparing within the language families, however, there are closer similarities. Japanese and Korean (or Malay and Tagalog) have about as much in common as English and Russian. Since Chinese scripts go back so far in history, they were adopted for a number of other languages despite those languages having no recent common ancestry with each other. Just like you can use Roman characters to write transliterated Hebrew even though Hebrew and Latin aren't closely related.

The flow itself isn't laminar at the throat, but that IS the location of the choke point, with subsonic flow on one side and supersonic flow on the other side. And so it's the point of greatest temperature and pressure. They might place the gimbal location just downstream of the throat, so there is a fixed portion of the nozzle and a gimbaling portion of the nozzle. That's a good trick, as Anakin would say. Still not a straightforward problem though.

Not just this thread.

We know that the Blue Origin website labels something as a hydrogen pump which, if zoomed in on, clearly says oxygen pump. So the web/marketing team may be just wildly inaccurate in general. Can't remember if I said this before or not, but if they really are going with a nozzle-only gimbal (which is still a shocking choice for a regeneratively-cooled nozzle), then that also decreases the space required for gimbal authority. Assuming 10° pitch and yaw authority (not necessarily representative; just picking a round number), an engine that is 80" x 37" will need a circle 60.2" in diameter to accommodate gimbal from the thrust structure at the top of the engine, but will need a circle of only 47.4" in diameter to accommodate gimbal from the nozzle: When you think about a lander with where the engines are clustered with 2 or 3 in a line, that extra gimbal clearance starts to add up. So I can see why they would want to go with a throat gimbal for that reason. Throat gimbals also allow for slightly lower overall fixture weight since the engine load mount doesn't move (the mount still needs to be able to transfer the off-axis load; it just doesn't have to do so while also moving). What's unclear is whether these advantages are worth the added complexity of running both cryogenic cooling loops AND superheated high-pressure rocket exhaust through a flexible throat.

I don't know that this has been posted before, although it's fairly old news. The Blue Origin store is selling a t-shirt with engine dimensions: If you go to the sale page and look closely, you can read the engine dimensions (given in height x diameter): BE-3PM: 115" x 27" BE-3U: 175" x 99" BE-4: 150" x 76" BE-7: 80" x 37" Main takeaway: the BE-3U is a honking big engine. Bigger than the J-2, bigger than an RVac, and bigger than the largest RL-10s. Unless I miss my guess it will be the largest upper-stage engine ever flown, whenever it finally flies.

The trouble is with how inertial and gravitational mass are handled, I think. Everything in Earth's SOI is attracted toward the center of Earth, regardless of its mass. Even photons, with no rest mass, are deflected by Earth's gravitational well, albeit only slightly due to their great speed. And so reducing the effects of gravity on an object -- "screening" some portion of its gravitational mass -- wouldn't change the free-fall behavior of the object. If you "screen" 99% of the mass of a 1-tonne object, you will be able to lift that object 10 meters off the ground with relative ease, exerting just slightly more than 98 Newtons of force and expending only 981 Joules. However, if you then release that object, it will accelerate downward at 9.81 m/s2 just as if there was no gravity screen at all, and so all 1000 tonnes of its inertial mass will impact the ground at 44.3 meters per second, releasing 981000 joules of energy. This creates an infinite energy glitch and I think it also potentially violates the Copernican principle. The natural "fix" is to reduce the vehicle's felt perception of gravity. But this really mucks things up, because now you start to drift away from Earth relative to the sun, etc., because if the felt perception of the sun's gravity dropped by 99%, your 30 km/s orbital velocity around the sun would immediately become an escape trajectory.

Yeah, there's a problem here. I'm currently sitting on a chair being pulled toward the center of the Earth at 9.81m/s2, which is the equivalent of being accelerated at 9.81m/s2 for an object in free-fall. More importantly, this is the equivalent of being at the apogee of a highly eccentric 6378 km by 6.7 km orbit around the center of the Earth, with an apogee velocity of around 361 m/s (if you're clever, you can use that to determine my latitude). I rapidly come to notice this orbital path (and the change in orbital velocity relative to my fixed surroundings) if I jump off a ledge. A device or machine that could "screen" gravity and cancel my weight, leaving my inertial mass intact, would be the equivalent of stopping my orbit from being an orbit: transforming my apogee velocity of 361 m/s into a pure tangential velocity. Thus if gravity was "screened" and turned off for me, I would continue moving at 361 m/s in a straight line, while the surface of the Earth continued to move under me at 361 m/s along the 39th parallel, in a circle. After four minutes, both of us would have traversed 86.6 km, but Earth's surface would have done so along one decree of circular arc, causing it to drop away underneath me by 972 meters. After twenty minutes, both of us would have traversed 433 km, but Earth's surface would now be 24 km beneath me. After an hour, Earth's surface would be a whopping 225 km beneath me. A g-screen, then, would result in a perceived vertical levitation effect which increases quite rapidly with altitude for the first three hours, then increases more slowly, reaching a maximum of 361 m/s (or more, if you're at a lower latitude) after 6 hours. However, this becomes a problem when you start looking at other orbits. Earth is orbiting the sun. If I "screen" gravity for my spaceship, my 30 km/s of orbital velocity I share with Earth is transformed to tangential velocity. After ten minutes, my straight trajectory will have drifted 1068 meters away from Earth's near-circular trajectory. After an hour, it will be 38 km. So my rate of drift relative to the surface of the Earth will be different if I am on the daylight side or the night side. (Fortunately, the rate of the sun's rotation around the center of the galaxy is slow enough (in rad/sec) that drift between my vehicle and the sun is not a significant issue.) I suppose the easiest way to fix this is to posit that the g-screen, while very efficient, is not QUITE 100% efficient, and so portions of the ship still experience SOME gravitational attraction, which anchors the ship to the planetary surface at least a little. That would fix the drift problem, but I'm not sure what it does to orbits themselves, since the inertial mass of the vehicle is still intact. Agreed. It was a good balance. Not Star Wars handwavium, where everything floats for no visible reason, but also not so detailed that it introduces obvious contradictions or otherwise adversely impacts the plot. Another reasonably good system (which I suppose approximates the Star Wars thing to some degree) is the "gravity rail" approach. There's some sort of superconducting loop which "locks" a vehicle to whatever spot in the local gravitational gradient where it finds itself, without using up any energy. That allows an object to experience gravity (and have the objects in or on it also experience gravity) while floating, but also while following the local curvature of space so that it doesn't drift. However, such a system doesn't allow you to follow the terrain; if you run off a cliff you'll stay at the same elevation, and if you try to drive up a slight hill you'll plow straight into it. Perhaps the superconducting hoverloop thingy stores energy relative to your own gravitational potential, so when you drop off a cliff, it absorbs more energy in cushioning your fall, and when you need to ascend a hill, you lose potential energy in your hoverloop while gaining gravitational potential energy. But this suggests some maximum altitude to which a speederbike can ascend...

I believe that in the Firefly universe the ships like Serenity have two systems that work like this: a g-screen and a g-field. The g-screen insulates the bulk of the ship from the effects of gravity, so that the engines don't have to fight against gravity during the climb to space. The g-field, on the other hand, produces a directional gravitational field for discrete objects within the ship to allow the use of decks, chairs, bunks, and so forth. It's reasonably clear that although these use similar technology, they are two separate systems. Both the g-field and the g-screen, while requiring a lot of energy to set up, require very little energy to keep going and decay slowly after a power loss, slow enough that issues like oxygen and temperature are more pressing. When the ship's main pulse drive is activated, the surge of power allows for the g-field and the g-screen to be aligned together, which effectively cancels the inertia of the entire vehicle and allows for very high acceleration. Feels very much like the sort of of stuff that @Spacescifi using goes on about. He should really just borrow the Firefly mechanic and be done with it. One unanswered question (I think) is how far the effects of any g-screen system actually extend. Ordinary objects are in free-fall, but free-fall itself is actually an extremely eccentric orbit around Earth's center of mass. If you cancel that, are you floating relative to Earth, or are you also floating relative to the sun? Relative to the galaxy?

It looks like they've switched over to a regeneratively cooled nozzle for BE-7:

Ah, I see where you're coming from. Do you know of any studies reviewing the effects of atmospheric Orion? With the radiation problem, I suspect that high enough standoff would make it work. The plasma envelope is opaque to x-rays so as long as the standoff distance is greater than the initial fireball radius there shouldn't be problems there. The flash will certainly transfer quite a bit of energy to the surrounding air, but air is famously quite transparent and so the amount of energy that is absorbed and re-radiated before the ship is out of range of the blast should be low. The shockwave propagation around the pusher plate might be a bigger problem.