Ars does seem to have far more "fluffy-ish fan articles" on a fair number of subjects in recent years. Maybe I'm old. Or maybe Condé Nast (their owner) is more interested in profits than in hard core tech nerd cred.
They did put this to their audience in a poll back in 2017[1] and the results overwhelmingly said to accept the car manufacturers invitation. The fact they asked publicly and run with full disclosure makes them far better than their peers imo.
They should indicate the nature of the deal in front of every car article, so that people aren't mislead into thinking they are reading a neutral review. Approximately no current reader knows about a six year old article. The way they are currently doing it is deceptive.
Isn't that exactly what they have done? Peeking at yesterday's Ars article about a Mercedes SUV - right at the top of it I see a really, really obvious box with this text:
"Mercedes-Benz provided a flight from DC to San Diego and a return from Los Angeles, plus four nights in hotels so we could drive the AMG EQE SUV as well as see the Vision One-Eleven and drive the AMG S-Class. Ars does not accept paid editorial content."
And their bottom-line conclusion about that car is probably not what Mercedes was hoping for, either:
"I'm not sure this EV makes a particularly compelling case for itself, except for the fact that it will appeal to the kind of person who wants to walk into a Mercedes dealership to buy the most expensive one."
You really think this is comparable to bring it up ?
Worst case a few sats get lost or the booster doesn't land correctly. As long as they test this on their own sats first (which are rather cheap) I do not see the issue.
They will want to test to breaking point after all.
There is a pretty good reason why they launch rockets over the ocean or large expanses of empty land. If an RSO terminates a flight with a perfectly functional system, the debris will still spread over the surface for some distance, depending on the altitude that the flight was terminated at, of course. Launch trajectories are designed in such a way that the portion where the altitude would allow substantial debris to return is usually placed over uninhabited regions or the ocean.
Are you forgetting that little something at the back of rocket that is so hot it can melt itself?
Even managing propellant tanks is complicated dilema. Take a big jug of water that is half full and slosh it around. Its one of thousands small things that needs solving. And solutions need to be implemented with as few grams as possible.
You're right, rocket engineering is not as simple as some people would have it sound.
I think the SpaceX organization has proven to be prudent risk takers. They have had many spectacular failures, but they have been good at communicating the risks ahead of time.
James Cameron spent 10M to built his submarine. It is impossible at present time to build a orbital manned rocket for that amount. Therefore the submarine is easier to design and build.
Hmm. The pressure inside a Merlin engine is about 1410 psi (about 88 atmospheres) and the temperatures inside the engine reach 2200 C. The turbopumps in the engine spin at about 3600 rpm and produce about 10,000 horsepower, which is just used to pump fuel into the main combustion chamber. Each Merlin engine produces about 200,000 pounds of thrust and is still reusable.
Just starting a rocket engine is a complicated dance that can easily turn out badly if not handled with extreme precision. Sorry, but I don't think there is any way one can equate the complexity of a diving bell with that of a rocket.
A spacecraft has to deal with zero to one atmospheres.
A rocket has to deal with rocket thrust/structural cross-section units of pressure. I'm neither a engineer nor informed about the structure of a Falcon 9, but that number is > 1 atm even if the entire cross-section is structural.
The reusability of the Falcon 9 is something still something without a lot of public information. I'd love to know how much they're spending to refurbish each of these – is it 80% of build cost or 20%? Are they decommissioning other rockets earlier than 16 launches because they're finding faults, or are most just waiting for more launches? What are the decisions to re-fly based on, are there common failure modes that aren't causing mission or recovery failures but are causing SpaceX to rule out the booster for future missions.
We do know that they're turning around boosters for re-use in less than 30 days which suggests they aren't doing extensive work, although I don't know how long building a new booster takes so maybe this is most of the time anyway.
Maybe none of these details really matter with Starship on the way, in the long term that will probably address all of these much more obviously, but it's still a way off and for now the F9 is the core of their business.
> I'd love to know how much they're spending to refurbish each of these – is it 80% of build cost or 20%?
"In August 2020, Elon Musk tweeted that refurbishment and reuse of a booster is done for less than 10% the price of a new booster while the payload reduction is below 40%. According to his tweet, SpaceX breaks even with a second flight per booster and saves money from the third flight on.[81] At that time, Falcon 9 Block 5 had made 35 flights with 11 boosters."
F9 is a combination of simplicity, reusability, and high cadence. Simplicity reduces the cost of the expended components and the cost of refurbishment. Reusability amortizes the cost of the components over more launches. And high cadence means both that fixed costs are amortized over more launches, and also that the expendable parts are produced at higher volume, reducing their cost. They must be really good at making a second stage now.
The highest stress points of the rocket will be the engines. If I were to guess (which is all anyone can do), the booster can be re-used a practically indefinite number of times (>100 launches) - while the engines need to be replaced every K launches on average. They might be able to stretch engine lifetime if they are good enough at shutting the engine down on failure. If they can tolerate one flaky engine on a launch, then they can blend new engines with old engines to improve re-usability by running engines to failure. Question is how good they are at shutting down the engine as catastrophic failure could take out the whole vehicle.
Space Shuttle main engines could be re-used 3x on hydrogen/oxygen fuel, and those maintenance schedules were set in the 80s.
Refurb is a million or so, a small fraction of the total cost of new. The new second stages is about ten mil, the fairing, often reused but not always is a few mil for each half. The ops and recovery are probably a million. Listen to the Musk Aviation Week interview for his public statements on all of that. Best case, he says, is about fifteen million for a reused launch and people paying attention think it was typically about $28 million at that time. See Chris Colouris (sp) comments off the record for that number. I suspect in the last several years, the gap between those has closed and it's probably around $20 million for a typical LEO RTLS launch that SpaceX charges about $67 million for. (Note, the full build cost of a new F9 is also a matter of speculation, and I think it's between $40 and $50 million today.)
What I think is interesting here is how in the larger rocket industry, specially small launch all the competitors talk about how they need to make rocket production like car production. Lowering the price or production by mass production.
SpaceX the most successful company doing by far the most launches, is producing about 1 new Falcon 9 Upper Stage about 2x a week at most.
A successful car is produced 5000x a week, maybe even 10000x. That gives you real cost savings, but 1-2 is really just serial production. That does give some savings but nothing like automotive.
So any small rocket company that talks about improving launch cost by mass production and talking about 'using methods from the automotive industry' is just delusional. Even if the were spectacularly successful and the was actual demand (there isn't) and launching as often as SpaceX.
The cost saving from mass production would still be really quite small. Going from the price of your first rocket to say 100 every year saves money, but nowhere near as much as reuse does.
RocketLab the only company that actually launches small sat regularly was absolutely convinced mass production was the way to go and then the CEO literally eat a hat and pivoted to making the rocket reusable instead. But somehow the we gone make small launch cheap with mass production story still convinces some investors.
Edit: Prat of why I mention this is just recently I was listening to interview with the CEO of Augsburg Rocket Factory from Germany and they made that argument again.
I'm not sure I agree. Mass production scales exponentially, sure, but that doesn't mean one producing two Falcon 9 Upper Stages per week isn't experiencing exponentially more efficiencies than one producing a competing rocket at a rate of one per month. No matter how you cut it, more production is typically better for cost efficiencies.
Yes its better, but my point is your not gone have cheap rocket launches because you produce 2 rockets a week. Your not gone beat a company that has rocket reuse. And you sure as shit wont compete with a larger reusable rocket.
Reusability isn't a panacea. Refurbishment is hand bespoke labor and its costs can add up quickly. The obvious example is the space shuttle, which cost $1B to refurbish.
Elon has said that it costs 10% of new cost to refurbish a first stage. It's believed that a first stage costs $20M to make, so that puts refurb at about $2M.
It's plausible that the first stage of a medium-class rocket can be built for less than $2M. Just barely plausible, but plausible. You're not going to do it if you pay $16M for a pair of BE-4 engines, but if you can get two Raptors for $250M apiece...
The elephant in the room is Starship. Designed for reusability from the beginning, it's quite possible it pushes that $2M figure down besides reusing the second stage as well as the first.
You can potentially build a small vehicle for the cost of a reused Falcon 9, but you don't compete on the rocket market, you compete on the payload to orbit market.
> but if you can get two Raptors for $250M apiece...
But if your competitor builds the same rocket with 3 Raptors that is reusable your still gone lose.
It’s rarely discussed but a disproportionate amount of aerospace cost is quality control. High level QC doesn’t tend to scale as well as mass production.
I think as you do more pieces you can invest in more automated ways of doing QC as well. As the increased fixed cost for the QC infrastructure is amortized across the larger number of produced units.
It all adds up. Scaling up production requires simplifying both product design and the entire process, minimizing the amount of complex manual work and bespoke components. That reduces QA/QC workload as well.
A lot of the QA cost is with pedigree and traceability and that doesn’t scale nearly as well. A $1k bolt isn’t expensive because it’s a bespoke design; it’s expensive because you have to maintain tight control from everything from material coupons to storage. This is one area that makes a lot of manufacturers balk at the selling to aerospace companies; the amount of QC work just doesn't make sense for the amount of product they're making.
Another area is supplier control, which SpaceX has struggled with in the past. That doesn't scale well either, but maybe they can bring that cost down by vertical integration of manufacturing (although that doesn't really help as much for the pedigree cost).
Yes, but I believe there are still synergies here. A $1k bolt may be expensive because of QA, but if you're making/buying enough of them, it could make sense to change the QA process to one that has higher false positive rate, but is cheaper to perform. Even if the change makes you throw every other bolt out, most of them perfectly good by the old process, if the new process means dropping per-item-cost from $1k to $250, it's very much worth it.
In a similar way, increased scale also makes it reasonable to invest up front to redesign the final product so it doesn't need any of those $1k bolts in the first place.
I think this is reasonable, but there’s a couple important points:
Aerospace quality is largely driven by reducing false negatives. You’re generally more ok throwing out something good than accepting something bad. Reducing false negative is what becomes costly.
Hiding quality issues by increasing throughout is a classic fallacy in the process control world because it increases waste. Sorry if I misinterpreted your point here.
What are you thinking on trends of revamping QA processes? Maybe I’m not imaginative enough, but I’m not sure what kings of different prices controls would bring down costs dramatically. Is there some specific new NDT you’re thinking?
> Aerospace quality is largely driven by reducing false negatives. You’re generally more ok throwing out something good than accepting something bad. Reducing false negative is what becomes costly.
Right. But the special case here is that if you can eliminate part of the process entirely, you no longer have to pay for reducing false negatives in it.
> Hiding quality issues by increasing throughout is a classic fallacy in the process control world because it increases waste. Sorry if I misinterpreted your point here.
All I'm saying is that trying to increase throughput will force you to simplify and streamline, which often will increase waste, but still significantly lower the total cost. Random example: CNC milling is incredibly wasteful compared to some more complex (and more manual) processes, but it's also much simpler and faster, and this saves you way more money across the board than you lose over wasted material. Same principle applies to QA: if you have two tests with equal false negative rate, with the second one having 2x the false positive rate but being 4x cheaper to run, it's worth it to switch to the second one - while this doubles the waste at this stage, it also lets you double the production (or just save you half the costs).
Now, to give specific SpaceX / aerospace-relevant examples, I'd have to dig through some of the things I've read over the years that gave me this overall impression. Having not done that, all I'm saying is that there are strong synergies involved in scaling production up. You don't just get bulk rates and pump stuff out faster - you get better at it faster, which lets you optimize it further and do it quicker. Simplifying design to simultaneously cut costs of parts, assembly and QA, is just one specific case.
> Is there some specific new NDT you’re thinking?
Not really, not right now. I could look over my notes and some references, and perhaps point something out - but that would really be pretending I know more about manufacturing processes than I do.
>But the special case here is that if you can eliminate part of the process entirely, you no longer have to pay for reducing false negatives in it.
I don’t disagree, I was just asking for a concrete example. In the abstract it’s a bit of hand-wavey marketing speak, but I’m not clear on what part of the process would be cut out. You’d still need traceability and pedigree regardless of how the part was made. That is what I pointed out drives so much of the cost.
I think we have a fundamental difference of opinion on reducing costs through quality control. Having worked at different times in process design and quality control in both automotive manufacturing and aerospace, using excess inventory to try to drive down costs is generally a bad practice. But I may an old codger stuck in my ways, so I’ll leave that discussion alone.
As a specific SpaceX example, they had a strut failure because of poor supplier quality control. Their remedy wasn’t to cut more out of the process but rather layer more quality control into it. I don’t know for certain, but I have to think that made it more expensive, not less.
> CNC milling is incredibly wasteful compared to some more complex (and more manual) processes, but it's also much simpler and faster, and this saves you way more money across the board than you lose over wasted material
Interestingly SpaceX opted for the exact opposite. Instead of having huge CNC milling machines like CNC has. SpaceX just manually welds in stringers into the tank.
And they were able to build the rockets far cheaper and arguably simpler. So it really depends what the manual process is and why you are doing it.
This and the parent comment has to constricted view of QA in aerospace. Yes, manufacturing quality is a big part. And yes, it’s possible that new manufacturing processes like additive manufacturing can streamline the QA. But the part being missed is all the downstream QA.
Those welds still need to be inspected. Any soldering needs to be inspected to the J-STD. The materials still need to be kept in an environmentally controlled environment. The materials still need samples kept, for possibly decades, in a similar manner in case something is found wrong later. All this requires administrative overhead like a chain of custody etc. That isn’t cheap.
Maybe, but I think the tech in that is still some times away. For example, there’s a lot of borescope inspection on a reused engine. Maybe we can eventually train image recognition software to help, but most likely this is still going to be done by humans in the short and medium terms. A lot of QC boils down to validated trust, and I think we’re some ways away from outsourcing that to machines in safety critical spaces like spaceflight.
You actually hear a lot of that in the industry. “We don’t make widgets” is a common refrain to illustrate that mass production principles don’t always completely map to the space sector.
There is also a point where you simply run out of either payloads or launchpads and processing facilities. While it's arguable that running out of payloads isn't at all bad, as it opens the market for more competitive pricing, running out of launchpads and facilities to use requires sizeable investment that, in turn, requires higher cadence to be profitable.
Mass production techniques can still apply to low production volumes. Controlling production, vertical integration, driving down cost of manufacturing.
Sure if you look at the whole rocket you can say "only 2 per week" but zoom in a little bit, Falcon 9s have 9 engines, Starship has what 33? It's still not car volume levels of course, but SpaceX are dreaming of a day where there are 1000 starships making a mars trip at once. 33,000+ raptor engines for a single transfer window. I know this is pie in the sky at this point, but it kind of makes sense.
I think you misunderstand your point. I'm not saying it bad to have mass production. Its clearly a benefit and increasingly so as you scale up.
I'm saying that if you are a small rocket maker who argues to investors 'our launches are gone be cheap because we will have mass production' is not really credible.
And many current small launch vehicle startups still argue this regularly.
The other issue is these companies aren’t really beating spacex with mass production- rocket engines being a good example - spacex has manufactured well over 700 Merlin engines. So when these companies claim they are going to undercut spacex dramatically with ‘mass production’ (10x spacex) they never get close to that and many struggle with basic reliability
As a controls engineer who does a lot of work for the automotive industry, I think you might be missing the long tail of production here.
Sure, a few current flagships have production rates of 5,000 per week, with takt times of 45 seconds or less, and really impressive rolling production lines that won't stop for years, which feature a screaming-fast robot at each corner of the car that get featured in documentaries. But while those lines turn out an impressive amount of product, they're a small part of the whole industry.
For every billion-dollar line at Ford building one black Model T (or F-150) a minute, there's a few dozen machines at tier 2 and tier 3 suppliers that might be building, say, a brown leather center console trim panel for the special color option for a less popular car. They might run 3 shifts per week while it's in production, and one a month for the next 10 years building service parts for mechanics. But they're still built with mass-production manufacturing techniques: The parts go in a nest sized just for them cut by CNCs, the right amount of glue is applied evenly by automated sprayers, the adhesive is wet out by an automated press that records the force and position to a production database, the completed assembly undergoes 100% inspection in an automated poka yoke station.
In contrast, that brown leather trim panel assembly machine is a one-off, hand-built by craftsmen. The nest might be cut by a CNC, but it goes on a rectangular sub-base with a hole pattern cut on a Bridgeport and held in a generic machinist's vise. The weldment that forms the table was cut, welded, and powder-coated by hand. The dowels that locate the nest were pressed using a manual hydraulic press, or maybe using a hammer, into holes reamed by hand. It underwent weeks of debug and run-off, with a handful of last-minute changes, until eventually the manufacturer told the integrator they were happy.
Small launch vehicle startups that argue they'll save money by using mass production are not suggesting they'll have TAKT times of 20 seconds per rocket, 3 shifts per day 7 days a week. They're merely saying they won't have individual master craftsmen fitting parts together by hand. There's a lot of room between the two extremes.
I agree with you there, but they are touting this as a competitive advantage over their competitors. We are gone be cheaper because of mass production.
Its not like their competition has a old guy with fill hand grind an aluminum block into a rocket.
This is a logical approach when many of the launches are delivering your own payloads to orbit. Best case, no issues and you get more data and more launches out of a booster. Worst case, you loose a few dozen Starlinks and you learn a lot about how your vehicle will eventually fail and maybe learn where to design in improvements to avoid that specific failure case.
That said, I'm sure they are still well within the safety margins, but pushing a bit (with little risk) to be able to see if your margins are correctly placed is just smart.
They've said repeatedly that they will push the limits to failure on their own dime (Starlink payloads) and in the mean time, they have a fleet that can meet almost any emerging demand by simply delaying a Starlink launch here and there. This allowed them to sweep up all the Russian business and now increasingly the ULA and EU business as those rockets continue to be delayed.
1.5 million subscribers, $150 per month, 4400 satellites, 5 year lifetime, that makes pretty much 3 million dollar for building, launching and operating one of the satellites. That seems not much.
Buying a launch used to cost around 60 million or so, that would still leave you with only 2 million dollar to build and operate one satellite. The costs to SpaceX are certainly lower, especially for reused ones, but still makes me wonder how cheap the satellites are.
I mean its still growing, not limited to 1.5 million people, and some of them also pay more then $150 per month. Ships, planes and the military.
If we assume cost of launch being 20-30million $ per Falcon 9 and an avg of 30-60 sats per launch. Launch is maybe 1 million per sat. Earlier Starlink 1 sats were likely less, the Starlink 2 might be higher.
Production 1 million per sat doesn't seem impossible given the vertical integration and large production numbers. Life time operation cost are likely below 1million given how few people SpaceX actually has to manage the whole fleet.
The economics will look much better once Starship is operational. If you can cut the cost of launch by 10x.
SpaceX is also doing a lot on the sat front. For example, switching from Kyropton to Argon. Next generation processors and so on.
SpaceX has set it up in way where their cost will continue to come down and their revenue will continue to grow as they sign up more people. And they have many more years before the competition Amazon will come online.
So I think this is a pretty good business for SpaceX. Building up a sales, distribution, regulation around the globe is a challenge but worth doing. Unlike Amazon they wont have such a structure already in place.
But they also want to increase the number of satellites three times, maybe 10 times, that will lower the budget per satellite proportionally. And how many more subscribers can they realistically get? They are not only competing with other satellite services, they are also competing with copper, fiber and cellular networks in many cases.
I think there are way more subscribers they can get.
In terms of the largest industries, shipping and planes they are only at the very beginning of their role out.
There are also currently regions that are over capacity where they can sign up more people as soon as they have more sats.
In terms of geography they still have a huge amount of capacity in lots of places. These are currently hard to connect because of the requirement for ground stations. Once they have deployed a whole layer all with the laser capable sats they will have a much easier time to expand into truly remote regions.
There are also major use cases like using this for other government. Japan might very well use it for example. And other US allies might well adopt it too. That's a lot of money. It might even get NATO adoption.
So in my mind there is a huge amount of additional potential revenue for this system.
Scaling up production of sats, terminals, ground stations and rocket all at the same time is a pretty monumental task. The first version, Starlink 1, Ground Station and Falcon 9 was more of a proving ground that can break even in cash flow terms. Starlink 2+, Lasers, Starship is really where they will really start to make serious money.
SpaceX is basically creating global infrastructure and they are doing it mostly vertically integrated. It shouldn't really even be possible for one company to build something like this.
They can and likely will get many millions, perhaps tens of millions of subscribers. This is a build out. Once the build out is complete, it'll be maintenance at lower costs. Starship launching v2 and v3 sats with increased capacity and ever increasing value as a second internet backbone will assure they have more than enough business to keep this all going -- probably better and longer than Comcast can keep it's tens of millions of cable internet users.
Once the build out is complete, it'll be maintenance at lower costs.
The satellites have a lifetime of about 5 years, so with 12,000 satellites they will need to replace 200, if they go all the way to 42,000 even 700 of them each month, indefinitely. While in some sense this is maintenance it also means building and launching one complete constellation every 5 years. Or they have to increase the lifetime which is at least currently probably mainly limited by the available fuel for station keeping.
It's not, but SpaceX is very aggressive on cost control (it probably costs them less than $3M ea bird right now, which is amazing for space stuff, and the cost/capacity is only going to get better) and more importantly there's a lot of headroom for growth in the subscriber base.
Less than $500K per sat (well, at least the v1 sats). I suspect their goal is about half that and then maintaining that cost but increasing capacity as the tech marches forward on commodification and miniaturization.
The origin of the 747 is fascinating. It went from an agreement between the Pan Am CEO and Boeing CEO at a golf game to a prototype in 2.5 years at a time before computer-aided engineering (slide rules all the way). It was also meant to be a holdover until the SST came online (which never happened of course).
But the 747 was so succcessful and dominant for 40+ years that it was a massive profit-center for Boeing. The R&D had long since been paid for and the profit margin was huge. The dominance only really broke because engines became so efficient that 4 engine planes no longer made sense, especially with the lower speed modern passenger aircraft go (the 747 would go Mach ~0.9, a 787 or A350 will top out at Mach ~0.8 for efficiency reasons).
The Falcon 9 is in that same class of industry dominance. The profit margin on a reusable launch must be massive. It's also incredibly reliable. Even when competitors build a reusable rocket, it'll take time to have a proven track record. Until then, it's a license to print money.
According to SpaceX's original Mars timeline, they're going to start colonizing Mars next year. So color me unimpressed that people are reporting on SpaceX optimizing boosters.
Its funny to how people have called Musk unstable madman for years now. And even before the twitter stuff people often had that complaint.
And yet, he is the longest serving CEO in the car industry and the space industry (at least large companies) and he has been incredibly stable and predictable in working on EV and rockets for 25+ years.
And not only that, during that time both major companies experience incredibly success in that time period.
And yet somehow people think its reasonable to call him unstable, inconsistent, loss canon and so on.
I guess what you write on twitter matters more then what you do with 100B$+ cooperation's.
I mean... he had some incredible high-profile successes and some incredible high-profile failures, as well as the occasional public meltdown. "Unpredictable" seems as good a label as any.
What are these 'high-profile failures' and how much time and money investment was there in these? Some people claim 'hyperloop' is failure, ok that like a couple of weeks fun project with no investment behind it.
FSD might be considered a failure, but that's within the larger success of Tesla and it has not yet been fully decided. So far Tesla is likely making more money from self-driving research then anybody else.
Boring company isn't a huge success, but calling it a failure is also not accurate.
And what 'public meltdowns' are you referring to exactly? Like taking one hit of a joint? Or what?
This is probably something that deserves a thorough answer, but to cite a recent one: any event where a billionaire ends ups suing his own attorneys can probably be marked as a "high-profile failure".
So something is a 'high profile failure' when 99.99% (and I am talking about people who know Musk) of people don't know about and cost a few 10000$. That's not my definition of the term.
96 comments
[ 2.8 ms ] story [ 135 ms ] thread[1]https://arstechnica.com/cars/2017/03/ars-asks-how-do-you-fee...
Isn't that exactly what they have done? Peeking at yesterday's Ars article about a Mercedes SUV - right at the top of it I see a really, really obvious box with this text:
"Mercedes-Benz provided a flight from DC to San Diego and a return from Los Angeles, plus four nights in hotels so we could drive the AMG EQE SUV as well as see the Vision One-Eleven and drive the AMG S-Class. Ars does not accept paid editorial content."
And their bottom-line conclusion about that car is probably not what Mercedes was hoping for, either:
"I'm not sure this EV makes a particularly compelling case for itself, except for the fact that it will appeal to the kind of person who wants to walk into a Mercedes dealership to buy the most expensive one."
So - where's the deception?
https://arstechnica.com/cars/2023/06/driving-lotus-new-emira...
Overly positive, all car photos right from the marketing department, no disclaimer that the thing might be sponsored in any way.
Worst case a few sats get lost or the booster doesn't land correctly. As long as they test this on their own sats first (which are rather cheap) I do not see the issue.
They will want to test to breaking point after all.
So yeah, seems that they want to reduce their starlink cost
No people were hurt and no windows were broken.
Even managing propellant tanks is complicated dilema. Take a big jug of water that is half full and slosh it around. Its one of thousands small things that needs solving. And solutions need to be implemented with as few grams as possible.
I think the SpaceX organization has proven to be prudent risk takers. They have had many spectacular failures, but they have been good at communicating the risks ahead of time.
Just starting a rocket engine is a complicated dance that can easily turn out badly if not handled with extreme precision. Sorry, but I don't think there is any way one can equate the complexity of a diving bell with that of a rocket.
A rocket has to deal with rocket thrust/structural cross-section units of pressure. I'm neither a engineer nor informed about the structure of a Falcon 9, but that number is > 1 atm even if the entire cross-section is structural.
We do know that they're turning around boosters for re-use in less than 30 days which suggests they aren't doing extensive work, although I don't know how long building a new booster takes so maybe this is most of the time anyway.
Maybe none of these details really matter with Starship on the way, in the long term that will probably address all of these much more obviously, but it's still a way off and for now the F9 is the core of their business.
Another advantage of reuse is cadence. They can refurbish faster than they build
https://en.wikipedia.org/wiki/SpaceX_reusable_launch_system_...
Space Shuttle main engines could be re-used 3x on hydrogen/oxygen fuel, and those maintenance schedules were set in the 80s.
SpaceX the most successful company doing by far the most launches, is producing about 1 new Falcon 9 Upper Stage about 2x a week at most.
A successful car is produced 5000x a week, maybe even 10000x. That gives you real cost savings, but 1-2 is really just serial production. That does give some savings but nothing like automotive.
So any small rocket company that talks about improving launch cost by mass production and talking about 'using methods from the automotive industry' is just delusional. Even if the were spectacularly successful and the was actual demand (there isn't) and launching as often as SpaceX.
The cost saving from mass production would still be really quite small. Going from the price of your first rocket to say 100 every year saves money, but nowhere near as much as reuse does.
RocketLab the only company that actually launches small sat regularly was absolutely convinced mass production was the way to go and then the CEO literally eat a hat and pivoted to making the rocket reusable instead. But somehow the we gone make small launch cheap with mass production story still convinces some investors.
Edit: Prat of why I mention this is just recently I was listening to interview with the CEO of Augsburg Rocket Factory from Germany and they made that argument again.
Elon has said that it costs 10% of new cost to refurbish a first stage. It's believed that a first stage costs $20M to make, so that puts refurb at about $2M.
It's plausible that the first stage of a medium-class rocket can be built for less than $2M. Just barely plausible, but plausible. You're not going to do it if you pay $16M for a pair of BE-4 engines, but if you can get two Raptors for $250M apiece...
The elephant in the room is Starship. Designed for reusability from the beginning, it's quite possible it pushes that $2M figure down besides reusing the second stage as well as the first.
> but if you can get two Raptors for $250M apiece...
But if your competitor builds the same rocket with 3 Raptors that is reusable your still gone lose.
Sure if you have a $250K per Raptor that is mass produced then you can beat a reusable RS-25 that cost 140M.
Another area is supplier control, which SpaceX has struggled with in the past. That doesn't scale well either, but maybe they can bring that cost down by vertical integration of manufacturing (although that doesn't really help as much for the pedigree cost).
In a similar way, increased scale also makes it reasonable to invest up front to redesign the final product so it doesn't need any of those $1k bolts in the first place.
Aerospace quality is largely driven by reducing false negatives. You’re generally more ok throwing out something good than accepting something bad. Reducing false negative is what becomes costly.
Hiding quality issues by increasing throughout is a classic fallacy in the process control world because it increases waste. Sorry if I misinterpreted your point here.
What are you thinking on trends of revamping QA processes? Maybe I’m not imaginative enough, but I’m not sure what kings of different prices controls would bring down costs dramatically. Is there some specific new NDT you’re thinking?
Right. But the special case here is that if you can eliminate part of the process entirely, you no longer have to pay for reducing false negatives in it.
> Hiding quality issues by increasing throughout is a classic fallacy in the process control world because it increases waste. Sorry if I misinterpreted your point here.
All I'm saying is that trying to increase throughput will force you to simplify and streamline, which often will increase waste, but still significantly lower the total cost. Random example: CNC milling is incredibly wasteful compared to some more complex (and more manual) processes, but it's also much simpler and faster, and this saves you way more money across the board than you lose over wasted material. Same principle applies to QA: if you have two tests with equal false negative rate, with the second one having 2x the false positive rate but being 4x cheaper to run, it's worth it to switch to the second one - while this doubles the waste at this stage, it also lets you double the production (or just save you half the costs).
Now, to give specific SpaceX / aerospace-relevant examples, I'd have to dig through some of the things I've read over the years that gave me this overall impression. Having not done that, all I'm saying is that there are strong synergies involved in scaling production up. You don't just get bulk rates and pump stuff out faster - you get better at it faster, which lets you optimize it further and do it quicker. Simplifying design to simultaneously cut costs of parts, assembly and QA, is just one specific case.
> Is there some specific new NDT you’re thinking?
Not really, not right now. I could look over my notes and some references, and perhaps point something out - but that would really be pretending I know more about manufacturing processes than I do.
I don’t disagree, I was just asking for a concrete example. In the abstract it’s a bit of hand-wavey marketing speak, but I’m not clear on what part of the process would be cut out. You’d still need traceability and pedigree regardless of how the part was made. That is what I pointed out drives so much of the cost.
I think we have a fundamental difference of opinion on reducing costs through quality control. Having worked at different times in process design and quality control in both automotive manufacturing and aerospace, using excess inventory to try to drive down costs is generally a bad practice. But I may an old codger stuck in my ways, so I’ll leave that discussion alone.
As a specific SpaceX example, they had a strut failure because of poor supplier quality control. Their remedy wasn’t to cut more out of the process but rather layer more quality control into it. I don’t know for certain, but I have to think that made it more expensive, not less.
Interestingly SpaceX opted for the exact opposite. Instead of having huge CNC milling machines like CNC has. SpaceX just manually welds in stringers into the tank.
And they were able to build the rockets far cheaper and arguably simpler. So it really depends what the manual process is and why you are doing it.
Those welds still need to be inspected. Any soldering needs to be inspected to the J-STD. The materials still need to be kept in an environmentally controlled environment. The materials still need samples kept, for possibly decades, in a similar manner in case something is found wrong later. All this requires administrative overhead like a chain of custody etc. That isn’t cheap.
You actually hear a lot of that in the industry. “We don’t make widgets” is a common refrain to illustrate that mass production principles don’t always completely map to the space sector.
Sure if you look at the whole rocket you can say "only 2 per week" but zoom in a little bit, Falcon 9s have 9 engines, Starship has what 33? It's still not car volume levels of course, but SpaceX are dreaming of a day where there are 1000 starships making a mars trip at once. 33,000+ raptor engines for a single transfer window. I know this is pie in the sky at this point, but it kind of makes sense.
I'm saying that if you are a small rocket maker who argues to investors 'our launches are gone be cheap because we will have mass production' is not really credible.
And many current small launch vehicle startups still argue this regularly.
Sure, a few current flagships have production rates of 5,000 per week, with takt times of 45 seconds or less, and really impressive rolling production lines that won't stop for years, which feature a screaming-fast robot at each corner of the car that get featured in documentaries. But while those lines turn out an impressive amount of product, they're a small part of the whole industry.
For every billion-dollar line at Ford building one black Model T (or F-150) a minute, there's a few dozen machines at tier 2 and tier 3 suppliers that might be building, say, a brown leather center console trim panel for the special color option for a less popular car. They might run 3 shifts per week while it's in production, and one a month for the next 10 years building service parts for mechanics. But they're still built with mass-production manufacturing techniques: The parts go in a nest sized just for them cut by CNCs, the right amount of glue is applied evenly by automated sprayers, the adhesive is wet out by an automated press that records the force and position to a production database, the completed assembly undergoes 100% inspection in an automated poka yoke station.
In contrast, that brown leather trim panel assembly machine is a one-off, hand-built by craftsmen. The nest might be cut by a CNC, but it goes on a rectangular sub-base with a hole pattern cut on a Bridgeport and held in a generic machinist's vise. The weldment that forms the table was cut, welded, and powder-coated by hand. The dowels that locate the nest were pressed using a manual hydraulic press, or maybe using a hammer, into holes reamed by hand. It underwent weeks of debug and run-off, with a handful of last-minute changes, until eventually the manufacturer told the integrator they were happy.
Small launch vehicle startups that argue they'll save money by using mass production are not suggesting they'll have TAKT times of 20 seconds per rocket, 3 shifts per day 7 days a week. They're merely saying they won't have individual master craftsmen fitting parts together by hand. There's a lot of room between the two extremes.
Its not like their competition has a old guy with fill hand grind an aluminum block into a rocket.
That said, I'm sure they are still well within the safety margins, but pushing a bit (with little risk) to be able to see if your margins are correctly placed is just smart.
[1] https://www.reddit.com/r/Starlink/comments/qurjn7/cost_to_bu...
If we assume cost of launch being 20-30million $ per Falcon 9 and an avg of 30-60 sats per launch. Launch is maybe 1 million per sat. Earlier Starlink 1 sats were likely less, the Starlink 2 might be higher.
Production 1 million per sat doesn't seem impossible given the vertical integration and large production numbers. Life time operation cost are likely below 1million given how few people SpaceX actually has to manage the whole fleet.
The economics will look much better once Starship is operational. If you can cut the cost of launch by 10x.
SpaceX is also doing a lot on the sat front. For example, switching from Kyropton to Argon. Next generation processors and so on.
SpaceX has set it up in way where their cost will continue to come down and their revenue will continue to grow as they sign up more people. And they have many more years before the competition Amazon will come online.
So I think this is a pretty good business for SpaceX. Building up a sales, distribution, regulation around the globe is a challenge but worth doing. Unlike Amazon they wont have such a structure already in place.
In terms of the largest industries, shipping and planes they are only at the very beginning of their role out.
There are also currently regions that are over capacity where they can sign up more people as soon as they have more sats.
In terms of geography they still have a huge amount of capacity in lots of places. These are currently hard to connect because of the requirement for ground stations. Once they have deployed a whole layer all with the laser capable sats they will have a much easier time to expand into truly remote regions.
There are also major use cases like using this for other government. Japan might very well use it for example. And other US allies might well adopt it too. That's a lot of money. It might even get NATO adoption.
So in my mind there is a huge amount of additional potential revenue for this system.
Scaling up production of sats, terminals, ground stations and rocket all at the same time is a pretty monumental task. The first version, Starlink 1, Ground Station and Falcon 9 was more of a proving ground that can break even in cash flow terms. Starlink 2+, Lasers, Starship is really where they will really start to make serious money.
SpaceX is basically creating global infrastructure and they are doing it mostly vertically integrated. It shouldn't really even be possible for one company to build something like this.
The satellites have a lifetime of about 5 years, so with 12,000 satellites they will need to replace 200, if they go all the way to 42,000 even 700 of them each month, indefinitely. While in some sense this is maintenance it also means building and launching one complete constellation every 5 years. Or they have to increase the lifetime which is at least currently probably mainly limited by the available fuel for station keeping.
The origin of the 747 is fascinating. It went from an agreement between the Pan Am CEO and Boeing CEO at a golf game to a prototype in 2.5 years at a time before computer-aided engineering (slide rules all the way). It was also meant to be a holdover until the SST came online (which never happened of course).
But the 747 was so succcessful and dominant for 40+ years that it was a massive profit-center for Boeing. The R&D had long since been paid for and the profit margin was huge. The dominance only really broke because engines became so efficient that 4 engine planes no longer made sense, especially with the lower speed modern passenger aircraft go (the 747 would go Mach ~0.9, a 787 or A350 will top out at Mach ~0.8 for efficiency reasons).
The Falcon 9 is in that same class of industry dominance. The profit margin on a reusable launch must be massive. It's also incredibly reliable. Even when competitors build a reusable rocket, it'll take time to have a proven track record. Until then, it's a license to print money.
'Yeah your car can go 1000 miles per gallon, but you promised be 10000 next year so color me unimpressed'. Doesn't that sound crazy?
And, by judging from Twitter, that's a good thing.
And yet, he is the longest serving CEO in the car industry and the space industry (at least large companies) and he has been incredibly stable and predictable in working on EV and rockets for 25+ years.
And not only that, during that time both major companies experience incredibly success in that time period.
And yet somehow people think its reasonable to call him unstable, inconsistent, loss canon and so on.
I guess what you write on twitter matters more then what you do with 100B$+ cooperation's.
Once you're rich enough, you're insulated from the consequences of your actions.
And likely he believes (based on good evidence) that he can make the company successful. He might be wrong or it will turn out he will be wrong.
> Once you're rich enough, you're insulated from the consequences of your actions.
That's a totally different argument then the one you made before.
FSD might be considered a failure, but that's within the larger success of Tesla and it has not yet been fully decided. So far Tesla is likely making more money from self-driving research then anybody else.
Boring company isn't a huge success, but calling it a failure is also not accurate.
And what 'public meltdowns' are you referring to exactly? Like taking one hit of a joint? Or what?