Not really true. Consider this article from 2008: Larry Williams, SpaceX vice president of strategic relations, told the Space Studies Board that the company is offering Falcon 9 launches for $36.75 million to $57.75 million depending on the size of the payload and desired destination.
In 2017 dollars, that’s $41M to $64.5M. Current published price on the SpaceX site for a baseline launch is $62M–on the high end but within the published range. Since the current iteration of the F9 (“Full Thrust”) is a much more capable craft than the original F9 1.0, it’s fair that they would be able to price things at that level.
Clearly, what’s also happened is that they no longer have to offer huge discounts to gain customers due to the difference in credibility between June 2008 SpaceX (which hadn’t even launched their F1 into orbit) and 2017 SpaceX (which has a pretty decent track record).
Obviously, those are the baseline prices and almost no one pays that little (though Iridium got close). But so were the original prices, so it wouldn’t be fair to compare a no-frills 2008 price to a fully-loaded 2017 one.
And in fact, by the end of 2009, the minimum published F9 price was $44M, or $49.8M in today’s dollars. Current prices are only 25% higher than that, but they can launch 80% more than they could originally ($44M price to GTO is 3000 kg, while their current price is for 5500 kg). Price per kg has gone down substantially almost no matter now you look at it.
Again, the prices that SpaceX advertizes on their website are bare manifesting prices with minimal services and support. It is unclear whether SpaceX is actually making a profit at this point; their costs can’t just be evaluated on a per launch basis becuase many of the costs, including ground support equipment and facilities, non-direct support labor, engineering labor, et cetera are more or less fixed costs that have to be amortized or spread across multiple launches, and they are well short of their expected launch tempo, although they have been gradually ramping up. There is this tendency among self-described “New Space” advocates to uncritically accept advertized capabilities and costs while they continually critique established launch providers for their empirically demonstrated costs (albeit often with some justification, especially in the case of ULA). But despite claims of significant cost savings, SpaceX has neither provided a publically available accounting of costs nor rendered the savings in the form of dramatic discounts for accepting previously flown hardware. I’ll believe that the SpaceX plan for first stage reuse is viable when they have a significant number of successful launches and offer more than a few percent of the total flight cost in discounts.
You’ll probably be waiting a while. There’s no reason for SpaceX to make deep price cuts right now. The small discounts currently given are already low enough to entice satellite operators; it would be silly for SpaceX to charge any less than the difference in “risk premium”, which will also go lower over time assuming SpaceX doesn’t have any reusability-related mishaps. So they can, should, and will make as much money as they can while competitors figure out how they can match them, or at least figure out which niches they can stay viable in.
I’m quite aware that the listed prices are not the full cost; my point is that neither were the 2009 prices. I compared list prices from both years. Iridium did get pretty close with a $492M contract for 7 launches (~$70M each).
What could happen is a big contract (with a publicly traded company) for lots of launches and “bulk discount”. It’s not impossible that someone might need 50 launches, but can only close the business case if the launches are (say) $35M each. I could just about see SpaceX biting on that.
Weisshund is suggesting that Apollo technology, presumably adapted and upgraded, is good enough. That we already know how to fly people in space, so we just need to put something together and go there. Something like ram it through and tough it out.
The Boeing 777 was not a huge advance in airplane technology/design, but that small-ish change cost seven billion dollars and nearly five years to get it into the air. Adapting Apollo/Shuttle/ISS technology to a Mars flight, by contrast, would be a huge and difficult undertaking that would cost far more just to get going.
More importantly, with Mars, you basically get one try. Apollo landed men on the moon six times in 11 flights plus one really fucked up ground test. With a Mars flight, you have to get it right the first time, because one scrub-classl mistake and you are done for good. Fail and you probably do not get a second chance, especially if people do not come home.
Part of a Mars mission absolutely has to be an onboard workshop. There can be no jury-rigging. If something has to be fixed or modified, it will have to be done millions of miles away from Earth, and possibly without much guidance from mission control. Full or triple redundancy is expensive and heavy, but failure is far more expensive.
And as much as we want all six crew to land back on Earth, I believe there is at least about an even probability that there will be at least one loss. That is a huge problem that must be faced up to and prepared for, especially for the crew, because not being ready to deal with that is unthinkably shortsighted.
Yeah, maybe we could adapt traditional, existing designs for a Mars trip, but doing it even a tiny bit wrong would be beyond disastrous for the space program. Any practical parallels to Apollo are vanishingly small.
I can only assume that you have not worked in any material product engineering or production capacity, because this is just not how manufacturing and operation of complex systems works. Products do not get arbitrarily cheaper just because you build more of them, and while every product has a so-called “economy of scale” where you can amoritize the fixed costs of tooling and support effort over a larger number of units, there is ultimately a floor defined by the production rate and cost of individual production.
In the case of rocket launch systems, as I have previously elucidated, the majority of actual costs of a launch are not in the physical hardware (generally around 10% for liquid propulsion systems, and not exceeding 20%), or propellant (<1% for common hydrocarbon/LOx systems) but all of the effort and costs involved in transporting, handling, processing, testing, integrating, and preparing a system for launch. The part of those costs that are fixed, such as equipment, can be amoritized over a volume of missions up to maximum throughput, but there is a substantial amount of “hand labor” that has to be done by trained people who can only operate to a certain degree of efficacy, and even though SpaceX has made efforts to streamline their processes in this regard, there is no evidence whatsoever that they could perform a complete Falcon 9 launch for US$35M even in volume, even assuming full reuse of the first stage for multiple flights (which despite their claims, it is not plausible that it is saving $15M per flight; a more reasonable figure is probably $3M to $5M at best). And the Falcon 9 Stage 1 does not have the capacity for flyback when used for GSO or other highly energetic trajectories, or for maximum payload weight, so even a theoretical great savings in reuse only applies to flights that can tolerate giving up some payload margin (which is a fair number of potential satellites, and especially multiple satellite deployments).
SpaceX has made some impressive accomplishments so far, both technically and in challenging the conventional launch providers in terms of their costs, but this doesn’t mean that we should uncritically accept every claim they make about costs or future capability, especially without any evidence and in the light that many prior claims made about costs and launch rates have so far failed to come to pass. Musk and Shotwell are in the role of entrepreneurial promoters, and as such will make the most optimistic claims conceivable to generate interest and bring in new business. Musk is also a self-described “visionary”, which means he advocates some future capability even if there is no currently viable path, e.g. sending thousands of people to colonize (as he describes it “retiring on”) Mars, even though the actual challenges of that effort are vastly beyond what SpaceX is currently capable of both in terms of cost and technical capability.
My family is long-lived, so I’ve got a decent shot at 2050. But nope, not happening.
Anyone who thinks getting to Mars is 2x or 3x or even 10x as difficult as getting to the moon is fooling themselves. Probably somewhere between 100x and 1000x. And getting to the moon is hard enough that nobody’s been back in >44 years, even though we got there with 1960s technology.
Mars’ mean distance from the sun is 141 million miles; Earth’s mean distance from the sun is 93 million miles. So even when Earth and Mars are in the same direction from the sun, Mars is 200 times further from the Earth than the moon is. And the moon stays roughly the same distance from the Earth, while the Earth and Mars travel around the sun at different speeds, meaning that most of the time, Mars is much further away even than that.
It’s easy to say we’re going to Mars, but there’s no pressing reason to spend a shitload of money on a Mars mission. Getting to the moon was prompted by one of the comparatively peaceful aspects of the Cold War, the space race. The Russians put up Sputnik and Gagarin, and we couldn’t just let them have ‘outer space’ (as we called it back then) to themselves.
There’s nothing like that to prompt another nation to even replicate our moon landings - something that is comparatively do-able. Mars just ain’t happening between now and 2050.
And rockets are a case where that floor is so far below the real-world production costs that it’s basically not worth considering.
I do in fact have some small experience with manufacturing aerospace hardware, though I wouldn’t claim that it has more than a passing relevance here. What’s clear though is that economies of scale are most easily had for very low rates. Making 10 of something might only cost double that of making one of something.
As a simple but relatively obscure example, any machined part of reasonable complexity needs some form of custom workholding. For lowish production rates, this hardware is simply a fixed cost, and the same whether making one or 100. Switching production is also a cost; it may even dominate the total cost if you’re only making a single unit.
Your figures of hardware costs of 10-20% are off-base unless you think that Musk/Shotwell/etc. are outright lying. For example: As Elon Musk explains, “The boost stage [of a Falcon 9 rocket] is about 70% of the cost of the rocket … it’s sort of on the order of $30 to $35 million dollars.”
That puts the full cost in the ballpark of $50M. Of that, the fairings are $6M:*
Musk said at last week’s briefing that each payload fairing costs about $6 million. “At one point we were debating if we should try to recover it or not,” he said. “Imagine if you had $6 million in cash in a pallet flying through the air, and it was going to smash into the ocean. Would you try to recover that? Yes, yes you would.”
*
50% of the aforementioned $30M is the $15M savings I mentioned. This is what they achieved on their very first attempt, without any of their refurbishment improvements (the upcoming Block 5 will have improved landing legs, for instance). Save another $5M with fairing recovery. With these two improvements, they will likely hit a total rocket cost of <$20M.
It is not plausible to me that it really only costs SpaceX $7-14M for a rocket. The engines alone are probably $1-2M each, and a F9 has 10 of them.
There are, as you’ll note, still many other costs involved. But they seem to be more on the order of 30% of launch cost as opposed to 80-90%. Some of these amortize as well.
Heck, forget about SpaceX. The Atlas V has a somewhat variable cost, but a fairly recent launch cost NASA $132M (and this appears to be the “all-in” cost). The RD-180 engine alone costs $25M, though, so they’re spending 20% just on that. Even bloated ULA manages to spend much less than 80% on non-HW costs.
Sure. GSO was always going to be hard; in fact, their latest insertion (of a 6 t bird) wasn’t recovered at all. But no one makes 50-satellite GSO constellations, so that’s hardly what I had in mind. A contract like I mentioned would have to be an Iridium-like LEO constellation (but bigger), or perhaps some long-term government thing (an extended version of the current ISS resupply contract).
Constellations of all kinds also tend to be more amenable to making tradeoffs between payload and reusability efforts; there’s no need to pack (say) 12 satellites in a launch when you can fly 9 instead and get RTLS as payback. There is more design flexibility when optimizing for cost.
To bring this back to the OP, this same flexibility also applies to a large portion of a potential Mars mission: in particular, bulk cargo (food, water, etc.) destined for Mars and fuel for on-orbit refueling around Earth. These cargoes can be subdivided easily to whatever is optimal for a given rocket.
Since there is absolutely no detail in what that costs represents and how it is broken down, there is no way to evaluate the merit of that claim. But I have worked with costing models for rocket launch vehicles, including metrics to assess where cost reductions and efficiencies of volume or commonality can be had, and I’ll reiterate the point that the bulk of the costs is in the handling, processing, integration, documentation, and operations of building up and flying the launch vehicle. The so-called economies of scale that come from maximizing production to amoritize the costs of tooling, fixturing, and other capital expenses are real and certainly applicable to structural components such as weldments and machined components, but what is often overlooked with more complex assemblies is that the tooling can only support a certain throughput rate that governs the rate of return of value. If an assembly fixture is required for a two month process, you can only amortize the cost per year across six units, and if you require a higher build volume you habe to buy another fixture.
I’ll conceed that SpaceX, by virtue of their somewhat vertically integrated manufacturing infrastructure, almost completely digital inventory management and build ‘paper’, and some streamlining of their integration processes may have a somewhat different balance of costs than traditional launch vehicle,manufacturers such as McDonell Douglas (now Boeing) and Martin Marietta (now Lockheed-Martin), but if there is “$30 to $35 million dollars” of materials, purchased parts, and manifacturing costs in the Falcon 9 first stage, or even in the entire vehicle, I’ll eat my fucking hat. The only way to get to that cost is to include integration and testing costs, and unless SpaceX is literally going to refuel and refly the stage without any servicing there is some significant recurring portion of those costs. To claim otherwise is bombast until unserviced reflight is demonstated (and no, the reflown F9 Stage 1 was not “unserviced”; it underwent significant refurbishment and testing).
The RD-180 cost is well understood to be profiteering by NGO Energomash, owing to the lack of domestic engine production capability at the time. In fact, Pratt & Whitney tried to develop a license-built version of the engine in the mid-2000, citing that they even with the standup of manufacturing capability and the license fees they would be able to produce the engine at a fraction of the cost. That they were not able to successfully manufacture and qualify the engine is more of a commentary on the sorry state of propulsion engineering in the US than the inherent cost of the engine. Rocket engines are complex assemblies operating near material capabilities, but so are jet turbine engines (with better margins but operating lifetimes measured in thousands of hours rather than hundreds of seconds) and we build those quite economically.
I’ve studied reuse and read many analyses by others on the economics and physical requirements for reuse. I won’t say that there is no case for it, but I don’t believe that the largely conventional Falcon 9 or SpaceX claimed hardware costs actually over the kind of savings that you are uncritically accepting, or that SpaceX is necessarily making a net profit at this point. To get sufficient fiscal margins to save money would require dozens of flights per unit, reductions in processing and integration labor to the point that not much more than an airliner ground crew can integrate and operate the vehicle, and throughput to maintain a steady steam of flights. I highly suspect that any advantage of F9 Stage 1 reflight is to reduce bottle-necking manufacturing throughput and maintain a higher volume of,total flights rather than the cost savings from not manufacturing new hardware.
That’s fair enough. I think there’s probably an interesting discussion to be had on what manufacturing costs even mean in this context, and probably part of the disagreement circles around this. Clearly (and SpaceX has said as much), material costs are basically a rounding error. Amortized cost of the machining time and operator salary is probably also in the single-digit-percent range.
So to get stage costs that high would certainly have to include some kinds of testing, but that covers a wide range of things. Basic acceptance testing of dimensional requirements? To me, that’s a basic requirement and I’d lump that in with manufacturing. It only has to be done once, even with reuse. But similar types of acceptance testing, like that there are no cracks in your turbopump, will probably have to be done every time. Maybe we don’t lump that in with manufacturing since it’s more like a regular inspection requirement.
No doubt, but ultimately they still had to pay it. Before SpaceX really embraced vertical integration, they got dicked around by a lot of suppliers who inflated their costs as soon as they thought they had a captive target. Probably not all of this was malicious, in that the low volumes in aerospace HW tend to lead to high overhead and “risk-adjusted” prices. Regardless, if you can’t do it yourself, then you have to pay someone, and that someone might have you over a barrel.
That’s also fair. SpaceX is still growing and even if their reuse is beneficial “on paper”, that doesn’t necessarily mean their launches are profitable as a whole. If nothing else, they’ve invested quite a bit in reuse and it’s not yet clear that the investment will ever really be paid back.
As for reducing bottlenecks, they’ve basically said as much regarding fairing reuse when Musk isn’t making imaginative analogies about airdropping pallets of money. Depending on how the numbers worked out, it could be advantageous to spend more on refurbishment if it meant you could avoid a big capital expenditure. A $100M autoclave that sits idle in lean times is a waste. If you can avoid that at the expense of some labor-intensive refurbishment, then at least you can reassign/fire those people if you don’t need the throughput right now.
There appears to be a major roadblock to a manned exploration of Mars. Admittedly, I think Stranger may have spoken to this in an earlier post, but NASA has extremely stringent policies related to contamination, which would not be easy for a manned mission to adhere to. And they seem to be in control of any missions that are US-based, including private launches and unmanned probes, so no US company is sending anything to Mars any time soon. There is a tad more world than the US here, though.
