> This is basically how SpaceX seems to operate: rather than optimizing a single part of the system (like trying to get the most efficient engine), they try to optimize from a systems level.

They're optimizing for "real" reusability, as in eventually aiming for airline-like turnarounds. None of the earlier engines you noted were reusable except the RS-25, and given how long it took to turn the Shuttle around one could argue it doesn't really count as "reusable" in sense SpaceX is aiming for.

Aiming for reusability and ease of maintenance leads to different optimization paths, in particular we should expect that any one component is going to be less powerful, it'll need to account for wear & tear, not use once and crash it into the ocean.

But raw performance on a single trip doesn't matter much if you can just deliver the same payload in two trips to orbit instead.

> and given how long it took to turn the Shuttle around one could argue it doesn't really count as "reusable" in sense SpaceX is aiming for.

Not just the time but the cost, if you take the figure of how much the space shuttle program cost, and divide it by total number of flights (including the two lethal ones), it cost somewhere between $900 million to $1.1bn every time the thing launched. It was very far from the early 1970s dream of an economically reusable spacecraft.

Indeed, the space shuttle was so expensive it burned up much of the rest of NASAs budget-- keeping that disastrous, failed program running is why NASA doesn't have manned spaceflight capability at the moment.

Well, from an outsider perspective, the SLS/Orion program is pure pork. Nobody holding the purse strings cares whether the thing ever flies. Heck, probably they even prefer if it never flies (always just a few years and tens of billions in the future); a failed launch is potentially politically catastrophic.

> But raw performance on a single trip doesn't matter much if you can just deliver the same payload in two trips to orbit instead.

Except that you're doubling the cost if you have to do two trips.

If you throw away the rocket, sure, you're doubling the cost if you have to do trips.

If the rocket is reusable, then the vehicle cost gets amortized over the total number of trips, and the per-trip cost gets closer to just the cost of fuel (hopefully at least an order of magnitude less than a new vehicle+fuel on each flight).

Fuel and oxidizer is 2-3 orders of magnitude less than the cost of an expendable rocket.

Entirely valid point, but doubling 1% of the cost of a single use system is still pretty good, comparatively speaking.

You just need twice as much methane. So just have PF Chang’s catered and you’re back in orbit in a couple hours.

> The issue with liquid hydrogen is twofold: it is not very dense, and to keep it from boiling away in liquid form, it needs to be really cold

It’s also a bitch to work with. When it’s not trying to immolate or suffocate you it’s busy turning every metal part it touches to garbage [1].

[1] https://en.m.wikipedia.org/wiki/Hydrogen_embrittlement

You missed one of the most important parts about methane as well: The Sabatier Reaction [1]

Hydrogen + CO2 => Methane + Water

This is an extremely important reaction. Mars, for instance, has effectively unlimited Hydrogen and CO2 which also means effectively unlimited Methane and Water. How just absurdly convenient is that?

But it also plays a major role in things like life support. We breathe out CO2. On Earth where we have an enormous atmosphere this doesn't matter. But in closed systems, such as the ISS or what will be the habitation regions on Mars, this is a major issue that needs to be dealt with or we'd kill ourselves with our own exhaust.

NASA, on the ISS, used to deal with life support by producing oxygen from imported water, and discarding the produced hydrogen. They'd then collect exhaled CO2 from the air using CO2 scrubbers and also discard that. That was a pretty inefficient system that required large amounts of imported water to sustain. Now they use the Sabatier Reaction to convert the exhaled CO2 into water and methane. Only the methane is discarded. And as a result of this, instead of needing to import large amounts of water, now all they need to import is a small amount of hydrogen to keep the reaction running.

On Mars this will result in a really cool system. What will be an increasingly large scale life support system will be literally creating rocket fuel as a byproduct. Of course methane is also a controlled and efficient gas, so it can also be safely used for things like cooking dinner.

[1] - https://en.wikipedia.org/wiki/Sabatier_reaction

If you invest more energy you can also turn Methane into longer chain hydrocarbons and produce things like plastics or lubricants.

Here's a Wired article about the raptor engine - it's "a full-flow staged combustion engine, only the third engine in history to employ this technique ... The previous two attempts at such an engine, one in the Soviet Union in the 1960s and another in the US in the early 2000s, never made it beyond testing."

https://www.wired.co.uk/article/spacex-raptor-engine-starshi...

And if you want to get a lot more detail on what makes a "full flow staged combustion engine" so special, this article from Everyday Astronaut goes into some amazing detail on this and many other engines.

https://everydayastronaut.com/raptor-engine/

This was AMaZING. Exactly the kind of content I want to consume on HN

This guy is amazing. Look out for his interview with Elon Musk that should be released in the next 24 hours on his YT channel [1].

[1] https://www.youtube.com/channel/UC6uKrU_WqJ1R2HMTY3LIx5Q/vid...

I'm sorry but did anyone do a rudimentary spelling and grammar check on that?

"Raptor is designed to power the new reusable vehicles SpaceX’s is building, the Starship spacecraft and the Super Heavy rocket."

Also they call Tim Dodd an "industry expert"...

https://youtu.be/Sdwy9fzQzl4 Scott Manley does a great job explaining such things!

I'm surprised the guy being a youtuber. He is the codecs guy at Apple Computers

Eh, it's just another form of blog at this point. Not the greatest for searching, but a valid medium for expression.

I think they might be suggesting that Scott may have specific exemption from Apple's PR policy (summary: "say nothing to anyone") to be allowed an independent public voice - particularly on a technical/engineering topic. Apple is not, so far as we know, currently developing orbital-class rocketry, but there's genuine potential for product topic overlap in Project Titan vs Tesla and perhaps more besides.

I would not be at all surprised if there's a List Of Things Apple PR Hereby Politely Requests That Scott Doesn't Mention.

When I was at AWS I was similarly forbidden from mentioning anything to do with video games to just about anybody, despite not having anything to do with video games myself, and many other topics besides the obvious (e.g. "don't disclose EC2/S3 capacity numbers"[1], "don't reveal how Glacier works"[2]), and in fact there was a whole programme of PR training and tiered permission to speak at conferences or to the press/analysts that boiled down, mostly, to knowing what you could and couldn't say, and how diplomatically you parried questions on the latter.

Either that or Terry Pratchett's law of rewritten rules applies.

[1] no-one would believe me anyway

[2] it's a massively redundant array of vinyl records

It's also great self-marketing.

He started out (or at least found fame) doing Kerbal Space Program tutorials and play throughs, that's how I first came across him. He's also a pro DJ with an awe inspiring music collection and knowledge of music, though I'm not sure how much of that he still does. Oh, and he's also a fully qualified Astrophysicist and has worked in the field.

I'm shocked that I'm the last person to know this. I didn't know I could be more impressed with him.

Isn't he an astronomer / astrophysicist? You must be talking about another Scott Manley.

Nope, same guy. Scott Manley has a degree in astrophysics (I think) and worked as an astronomer for a while, then got a job coding at Apple. He's a smart guy

Technically he got a job at a startup, which was then bought by Apple.

Yeah, I looked him up on Wikipedia, looks like I got the wrong impression from his videos. Smart guy indeed!

I really like this one:

https://www.youtube.com/watch?v=LbH1ZDImaI8

I'm no rocket scientist, so bear that in mind, but something has been bugging me about the flight paths for this vehicle [0]. It has to make some very large, specific orientation changes several times during descent, using the heat shield on one side of the vehicle, and then only rotating to upright orientation in the lower atmosphere, after quite a lot of falling sideways. The orientation maneuvers seem much more complicated than their current booster re-entry flight paths [1]. and even more complex than the space shuttle [2].

[0] https://www.youtube.com/watch?v=LQTnWEHl5qU

[1] https://www.youtube.com/watch?v=6YyV-otP3pI

[2] https://www.youtube.com/watch?&v=TOieURpnbm0

This appears to me to make each flight much more complicated to pull off, introducing more points for it to fail without contingencies, and it makes me nervous. But, again, I am an just an unenlightened observer from an unrelated field. I don't even play Kerbal Space Program! I'm sure they've weighed a million different options and their risks, and have arrived at this one after a lot of thinking, but it has been bugging me. The stainless steel construction certainly looks beautiful out there in Texas.

It’s actually similar to the shuttle, In a couple of ways simpler, and a couple of ways more complex. The shuttles body was also designed to both generate lift and be a blunt body - with a transition of mode between them. The blunt body allowed it to survive reentry while the lifting body allowed it to glide. However that glide was way more complex then NASA generally talked about. The shuttle pulled some aggressive s-turns to bleed off energy during the re-entry.

This approach is more or less the same as what the starship is doing here, except instead of using a glider trajectory for the controlled landing they instead use retropropulsive landing. This is needed because what SpaceX is landing here is a second stage, not a first stage. Way more heat, way more power.

Also they plan to land where there isn't any atmosphere, or an atmosphere too thin to brake significantly.

So they need propulsive landing.

A lot of the complexity is hidden in the second video. It's the exact same deal, it just doesn't look like it. In particular those innocuous 'waffle irons' on the side play an incredibly important role. You miss the scale in the video thanks to how big everything is. Those fins are each 4' x 5' large. And each of them can independently yaw/pitch/roll. And of course the exact angle that the rocket is at when it fires its boostback (as well as when it's coming down), exactly how long the boostback is fired, and a large number of other variables all need to be lined up exactly perfectly. Otherwise you end up literally thousands of miles from your destination, run out of fuel, or otherwise condemn your mission in a surprisingly large number of innovative ways.

The big difference you're seeing compensated for is that Mars atmosphere is less than 1% dense as Earth. So to slow yourself down you need a much larger exposed surface area. Do that exact maneuver on Earth and you'd get an 'unplanned rapid disassembly' during reentry thanks to the thicker atmosphere. That danger and complexity is also something that's hidden.

I cannot recommend Kerbal enough for anybody even vaguely interested in space. Just absurdly fun and does really help you get a feel for all of these issues.

> Do that exact maneuver on Earth and you'd get an 'unplanned rapid disassembly' during reentry thanks to the thicker atmosphere

That is the planned Earth re-entry maneuver. Belly first almost all the way down, using the heat shield on one side, then flipping and completely re-orienting near the ground so that the vehicle is pointed upwards, and then the final burn to land upright.

But it does make sense that this path has arisen from a secondary capability: to land on planets or moons with less atmosphere. It still doesn't make me less nervous about their ability to pulli it off reliably. Here's hoping.

Learning how to reliably control the damn thing on descent is certainly their greatest hurdle. Refueling in orbit being the second biggest. I expect things to go boom at least a few times before they learn to land properly. Saying people Wil go to orbit within a year is most likely not to happen.

Elon said they learned a lot of what they need for orbital refueling just from docking with the ISS.

Methane was chosen to a large extent because you could make it fairly easily on Mars which is necessary if you want to fly people there and then get them back again.

Musk talks about making it on Mars here https://www.youtube.com/watch?v=sOpMrVnjYeY&feature=youtu.be...

> This is basically how SpaceX seems to operate: rather than optimizing a single part of the system (like trying to get the most efficient engine), they try to optimize from a systems level.

A system of local optimums is an inefficient system. Sort of the self-identified takeaway of the book _The Goal_, and I haven't stopped thinking about this statement since I read it.

Musk thinks like a software engineer. Create simple, reusable, components that can scale efficiently. I’m taking some liberties in saying this, but the gist remains the same. Musk keeps it simple and focused.

Agreed! Only a software engineer would risk a product launch delay because of an uber-bikeshedded pet feature (Model X doors) or wait until the last minute to ship and then construct a makeshift “factory” in a parking lot and make staff go through a death march, QA be damned.

The Model X doors are terrible. It's an SUV that you can't put a roof rack on. Does seem like a case of geek myopia over usability.

Or… optimized for a different case (ease of access for small kids).

I haven't used or wanted a roof rack in 30 years (extra mass and drag at the highest location? No thanks), but the Model X has a lot of storage space, including the frunk so it's a non-issue.

The falcon doors also provide a surprising benefit: cover in the rain. And they are very practical with small children.

Sliding minivan doors get you both.

Optimized for showy and unique features, when a young Tesla really needed its products to stand out.

He actually thinks like a systems engineer and a highly skilled and mature one at that.

https://www.youtube.com/watch?v=sOpMrVnjYeY&feature=youtu.be...

Musk does not design the rockets.

Musk was chief designer for Falcon 1, their first rocket. Musk is very much involved in detailed technical decisions.

Musk is very much responsible for why the team works the way it does. He specifically focused on hiring people who experiment and try out things a lot. He did not want just pure theoretical guys.

He doesn't do most of the work but he's very involved. Here's an interview Mueller gave that describes some things Elon did with the Merlin engine, search for "face-shutoff."

https://zlsadesign.com/post/tom-mueller-interview-2017-05-02...

> And, uh, I’ve seen that hurt us before, I’ve seen that fail, but I’ve also seen— where nobody thought it would work— it was the right decision. It was the harder way to do it, but in the end, it was the right thing. One of the things that we did with the Merlin 1D was; he kept complaining— I talked earlier about how expensive the engine was. [inaudible] [I said,] “[the] only way is to get rid of all these valves. Because that’s what’s really driving the complexity and cost.” And how can you do that? And I said, “Well, on smaller engines, we’d go face-shutoff, but nobody’s done it on a really large engine. It’ll be really difficult.” And he said, “We need to do face-shutoff. Explain how that works?” So I drew it up, did some, you know, sketches, and said “here’s what we’d do,” and he said “That’s what we need to do.” And I advised him against it; I said it’s going to be too hard to do, and it’s not going to save that much. But he made the decision that we were going to do face-shutoff.

So, he knew about a problem, he heard someone mention a possible solution, and he insisted they use that solution.

That is not what designing is.

Mueller begs to differ "And now we have the lowest-cost, most reliable engines in the world. And it was basically because of that decision, to go to do that. So that’s one of the examples of Elon just really pushing— he always says we need to push to the limits of physics."

His title is literally CEO and Chief Designer.

ISRU on Mars makes Methane fuel and excellent choice. Zubrin’s favorite in ‘A case for Mars’.

There won't be much literature because Raptor is essentially the first production methane engine. Everyone in the past seems to have decided to optimize strongly for either Isp, bulk density, or storability, and methane is the best at nothing.

Right, because in the past we were building first stages which had to be optimised for operations in the lower atmosphere, and upper stages which had to be optimised for vacuum. Hence the Saturn V had an RP-1 powered first stage and Hydrolox second stage.

But the raptors on the second stage have to operate efficiently in a vacuuum, but also work well and sea level for the landing. So even taking Mars out of the equation (which we really shouldn't, but just for sake of argument), it's requirements are quite different to anything we've built before.

SpaceX has a good design philosophy. They use cheap, reliable, and sturdy components, take the hit on payload fraction, and just make really big rockets to make sure _total_ payload is adequate. I'm reminded of the old quip apocryphally attributed to Stalin: "quantity has a quality all its own"

RD-58MF - https://en.wikipedia.org/wiki/RD-58 - uses kerosene (or syntin?) and has Isp of 372 s.

Methane, which is cryogenic and less dense, ought to deliver more? Otherwise bigger tanks and operational issues will eat out the benefits.

You can raise the Isp of any vacuum engine by using a bigger nozzle. The RD-58MF is a small engine and there's room for a relatively huge nozzle. That's not true for big engines.

For example, Merlin vacuum has a 1:165 expansion ratio. The RD-58MF has a 1:500 ratio. The Wikipedia article for the RD-58 family shows the increase in Isp from various expansion ratios.

https://en.wikipedia.org/wiki/RD-58

Right. There is however an argument that the smaller total impulse a stage does (RD-58 is used for upper stages, where the stage mass is closer to the stage payload), the smaller the relative mass of fuel+engine of that stage is, and so the fixed mass of the bigger nozzle added to the stage costs more, in Tsiolkovsky equation, that the mass for the bigger stage. Going further, we can argue that for larger stages it's beneficial to spend some mass on nozzle, because that mass is used for larger impulse provided by the stage.

Interesting that using http://rocketworkbench.sourceforge.net/equil.phtml I couldn't get declared Isp (380) for methane-LOX for reasonable nozzle expansion ratios. Maybe the model is too approximate...