EPTPast Entries
Part 1: (The Basics of) Rocketry and Spaceflight
Part 2: Disasters and Anomalies
Introduction
It almost seems like common sense - rockets are an expensive piece of hardware, and the structure of the rocket costs many millions while the fuel is merely a couple hundred thousand dollars. So why do we shoot off these rockets just once, and then leave the expensive hardware we spent millions of dollars building to either just float away or disintegrate as it burns up on reentry? After all, we just don't throw away cars or airplanes after one use, we make repairs and then fire them up again. So why not rockets?
To answer that question, perhaps we should start by looking at technologies that we actually have in both the reusable and expendable variety, to try to understand why you might or might not reuse them. Two of the most common would be water bottles and cameras. You can either pay for a nicer, multi-use water bottle and use it many times, or just get one of those one-use, low-hassle disposable ones. Same with cameras - a disposable one is going to be far less of a hassle, far less troublesome to lose, far less complicated than a higher-end, reusable one.
Though rockets and cameras are far from similar, the rationale for forgoing reusability is pretty much the same: logistics and economics. If you don't launch that many rockets, you don't gain much from recovering them. You have to make a more complex, less powerful device that is a lot more hassle. Of course, you can save a lot of money over the course of many uses, but if you don't have all that many uses on your rocket then there is no reason to put in that extra effort.
The technical feasibility of reusing a rocket has long ago been demonstrated; the Space Shuttle has been reused, the Falcon 9 has been reused, and any number of spacecraft in between has been partially or fully reused after launch. Some repairs always have to be done to make sure that your rocket is up to the task of undergoing another launch, but hopefully that's cheaper than just building a new one completely from scratch. Sometimes it is, sometimes it isn't. But in the end, reused rocket or not, it shouldn't be launched until you have the proper level of certainty that your rocket is good for launch. So for now, it would not be unreasonable to set aside the topic of which variety - reusable or expendable - is safer, because at the end of the day they both should be safe.
Ultimately, reusability is a question of economics - is it a better business decision to make your rocket from scratch every time, or to recover and reuse components? We're going to start with the Space Shuttle, to understand what can be learned about reusability from the poster boy of "reusable, but not worth it" rockets. The main focus is going to be on SpaceX's Falcon 9, which has impressed the world with its booster landing and made some fantastic claims about how effective reusability is. We will compare that to the fully expendable Atlas V, the overall best rocket of SpaceX's most important US competitor, United Launch Alliance. For the sake of brevity, we will set aside all international attempts at making reusable rockets - but know that this debate extends well beyond the borders of the United States.
We will be making the case for - or against - reusability as a business case. I'm sure you all have heard some variation of the adage that just because something works on paper, doesn't mean it works in the real world. This is true - but an important corollary to that saying is that if something doesn't work on paper, then we can be pretty sure that it won't work in the real world. So before you go out and make your own rocket with full reusability, hold your horses - we have to sit down and see if the numbers look good on paper. You want to make money, not lose it, don't you?
Lessons from the Past: The Space Shuttle
Before we get into the economics, first we should sit down and discuss the vehicles that we're shooting up into the sky and why the decisions of their design were made as they were.
We will start, of course, with the Space Shuttle, the first serious attempt at a reusable rocket. I talked a lot about it in my previous entry in this series, and overall it's probably the most studied rocket in history. For reference, let's re-post a diagram of its design.
![[image loading]](http://i.imgur.com/DvbYyhS.jpg)
The Space Shuttle has, in recent times, earned quite a unfortunate reputation for being too expensive and too complicated. It was certainly kept running well beyond its time, running for 30 years from 1981 to 2011. It was a rocket that tried to do too much - an attempt to create an all-purpose spacecraft that could do everything and that would be cheap because you could reuse it dozens of times. Unfortunately, the finances never really worked - the Shuttle cost $450 million a launch, or if you take the approach of dividing the program cost by the number of launches, you get closer to $1.5 billion a pop. As much of a beauty as it was, even its fans can't deny it was severely overpriced.
Ultimately, the economic problems of the Shuttle were Shuttle problems, not reusability problems. Its design was poorly conceived from an economic perspective, a result of a "lowest bidder" scenario that turned into massive cost overruns that should have been considered when the Shuttle was still in the planning stages. But as a science project, its successes and failures are important in order to understand the technical aspects of reusability.
As the Shuttle launched, the Solid Rocket Boosters (SRBs) provided a massive burst of thrust all in one go - provide enough thrust to get into space, then detach and fall into the ocean, where it's picked up by a boat. Then the main Orbiter, with a fuel refill from the External Tank, provided the rest of the thrust to make it back into space. After it runs out of fuel, the External Tank drops off and is not reused. The Orbiter lands like a plane after it finishes its mission.
![[image loading]](http://i.imgur.com/OP3xcWS.jpg)
Picking up the SRBs from the ocean
![[image loading]](http://i.imgur.com/uDvvJre.jpg)
Landing of the Orbiter
Both the SRBs and the Orbiter were reused between flights of the Shuttle. The Orbiter was actually quite well-positioned for reuse; parts had to be replaced and repairs had to be made, but it was not all that bad overall. Much of the original structure of the first flights of the Shuttle lasted up to its last flights. And as the Shuttle was reused more and more, it started to be easier to repair and re-certify the Orbiter for flight - once they knew how the Shuttle tended to fail, they had a good idea of what they needed to repair, so it became easier and easier each time.
The SRBs were a different story. The way that a solid rocket burns, the entire inside was pretty badly charred and needed refurbishing. And even worse, it landed in the ocean, so it filled full of salt water. That's the kind of corrosion that means you should probably just throw the whole thing out. A car or plane that find itself underwater is headed for the junkyard, so it's no surprise that a rocket booster that was found under water was not cost-efficient to repair, at all. They saved little to no money on reusing the SRBs. But the benefit of being able to analyze the wreckage was certainly useful. Having to send out boats to fetch the SRBs? Not so much.
As the pioneer of reusability, the Shuttle certainly made many mistakes, some due to short-sighted design decisions, but many simply because they were the first to try it and they had to figure everything out themselves. So what can we learn from them? A few lessons, I think are worth taking to heart.
The easiest to talk about is the Solid Rocket Boosters - a decision that, from the perspective of rocket design, was clearly the wrong one. Solid rockets are less efficient than liquid ones and the design decision was made simply because it was less money up-front to develop a solid booster than a liquid one. Sure, it cost more money in the long run and was harder to reuse (because the entire cylinder was charred), but it was a necessity to calm a short-sighted Congress with a "lowest bidder" mentality. More important, however, is simply to realize that the better the condition of your recovered pieces, the easier it will be to economically reuse them - so a saltwater-filled booster is not a great thing to reuse, but a spaceplane that lands safely on a runway might be.
From a more nuanced perspective, though, there are a lot of other important lessons to learn that were not borne just of a Congressional "lowest bidder" mentality. The first is simply that reusing a craft multiple times means that it gets easier the more times you do it - you get a handle on what needs to be repaired and what doesn't, you can modify your craft accordingly, and have better luck next time. In and of itself, certainly a good reason to reuse rockets - it will make your next design a fair bit easier to manufacture. But not nearly enough.
However, the Shuttle did also reveal the logistical obstacles that exist that make reuse more troublesome. When you are reusing rockets, you have to both develop an infrastructure for building new rocket components and for recovering and refurbishing old ones. This could be tricky if you're not launching very often as a result of the issue of economies of scale.
For those less familiar with economics, an economy of scale is the phenomenon where a combination of a large fixed cost (such as the establishment of a factory, a cost which is paid only once) and a low variable cost (the cost for the construction of each individual rocket) comes together in such a way that the average cost is lower when you make a lot of them than when you make just a few. See Wikipedia if you want more details.
Although there were 135 missions on Space Shuttles, only six Shuttles were ever built - and one never flew. So five Shuttles took care of all the missions. This comes with a few issues. The first is that if you ever need to build a new Shuttle - say, because one exploded, it's going to be a lot more expensive as an average per-unit cost to make it. You aren't making many but you have to keep that supply line running - at a hefty price. And you still have to have all the infrastructure involved in picking up the pieces you reused. Drone ships if you're performing sea landings of boosters, heavy lift helicopters if you're jettisoning your engine for pickup, boats to pick up gigantic SRBs out of the ocean, and so on. Plus the cost of repairing the rockets and making them ready for liftoff again. That is a whole separate capacity you have to have ready, and once again one subject to economies of scale.
How do you get around this "economy of scale" issue? Simply by having enough customers so that both the production end and the reuse end are well-provisioned. Cars and airplanes have this; enough people need both new aircraft and reused aircraft that both production lines are economical. But rockets? Less than 100 of those suckers get popped off per year worldwide. Unless that changes, it's not going to be economical to have two large infrastructure investments instead of just one.
I want to just take a moment to draw attention to one generally underappreciated fact: even though it's been more than 100 years since the concept of a rocket has developed, the spacebound variety of rocket cannot be called a mature technology; it is more appropriately termed "experimental." A single aircraft or a single car failing will not cause you to take your entire line of products out of service for a few months, because there is no need - we are confident enough in the function of either of them that one fault does not have to be assumed to be a technical defect of the entire product line. And even when you do find a product-wide defect and issue a recall, you don't pull your entire line out of service; you just replace the broken parts one by one. None of this would be acceptable with rockets. We have yet to figure out how to make them have an acceptable level of reliability for car/aircraft treatment to be acceptable - which is one important way in which the analogy breaks down. More on that later.
Another problem, demonstrated by the Shuttle, from having only a few craft by virtue of reuse - difficulty of upgrades. When you have only a couple of rockets, taking any one of them out of service to upgrade it will significantly hamper your capacity to launch. If you lose one on launch, then you will also have reduced launch capabilities for a fairly long time. The Space Shuttle Columbia, when destroyed, was the only one equipped for the kind of advanced science missions it was working on - which meant that another Shuttle would have to be upgraded to have that capability. And if your rocket design isn't a mature design - most rockets are not - then you will essentially always have a prototype as your launch vehicle because upgrades are tough.
Furthermore, as the Shuttle learned through painful experience, the more advanced your equipment, the more brittle it tends to be - so rugged simplicity is a good approach to reusability. The more advanced your technology, generally the more frail it will be, a very bad thing for rockets. That doesn't mean that you should use all 60s-era technology, because that too will end badly - but keep in mind that the more complex your design, the more expensive it will be to reuse. You may have to sacrifice a fair bit of functionality to make it work.
One final, if somewhat incidental, lesson of the Shuttle: spaceplanes are a good device for reusability. That aerodynamic design turns out to be quite good for landing in one piece and takes advantage of the fact that unlike rockets, which have to put out as much thrust as possible to leave the Earth's gravitational influence ASAP, airplanes can spend a lot of time coasting, using not all that much fuel, and getting far just by virtue of the air helping it on its way. The government has made multiple experimental spaceplanes since the Shuttle - the first part of this talk considers a few of them (and the entire video is an interesting talk on the ins and outs of reusable rockets). Alternatively, if you prefer reading academic papers, this one talks in good depth about all the spaceplane projects in the context of rocket reusability. I myself know very little about the spaceplanes though, other than that they've generally tended to be much smaller than the Shuttle, they were all pretty expensive, and none of them have led to a mass-produced design. Fairly reusable, although the problem of lifting these vehicles up into space often makes them more trouble than they are worth on any appreciable scale.
So, to recap. The Shuttle may not have ultimately been the low-cost multi-purpose space taxi that it hoped to be, but it did provide a lot of lessons that future reusable schemes will need to overcome. If I were to summarize the lessons learned, it would be this: focus on simplicity, ensure you have a large enough market to scale properly, recover your rocket in a manner that reduces damage to the craft, and have a mature design when you start reusing. Any reuse scheme should be evaluated first by how well it addresses these issues. You will be able to reuse more effectively as you get more experience with your craft - so that is a windfall in favor of reuse.
The Expendables: Atlas V
Before we talk about Falcon 9, we should look at a craft that is fully expendable for comparison. I know that many might say that we could just compare it to the Falcon 9 without reuse - but that craft was designed with reuse specifically in mind, and its design choices will reflect that. Furthermore, given that SpaceX competes primarily for commercial business, it might seem to make more sense to look at Arianespace as a more direct competitor - but competition within the nation is a more meaningful metric since many opportunities for rocket launches are nation-locked. So we're going to look at Falcon 9's most important domestic competition: United Launch Alliance's Atlas V. A diagram of the rocket was given in the first entry in this series and is reposted here for reference.
![[image loading]](http://i.imgur.com/CgC1gmy.gif)
The Atlas V uses the kerosene/liquid oxygen RD-180 as its first stage engine, a half-sized version of the RD-170, the booster for the Russian monstrosity that was the Energia rocket. To date, the RD-170 is the most powerful liquid rocket engine ever invented, and both the RD-170 and its half-sized variant are true marvels of engineering. At just $10-15 million a pop for an engine that could power a large rocket, the RD-180 was a very good choice in its time - although the tensions between the US and Russia in the current day and age make the RD-180 a troublesome engine to keep around. Up to five small solid rocket boosters are used to give Atlas a little more power if need be.
The upper stage of the Atlas V is an all-American beauty - the RL-10, manufactured by Aerojet Rocketdyne and its predecessors. A liquid hydrogen/liquid oxygen engine, it has an integrated energy scheme that allows you to get some remarkable bang for your buck in terms of upper stage fuel. It has a very low thrust for a rocket engine at a mere 110 kN, but it can burn for an impressive 700 seconds. From an engineering perspective, an excellent design for operating in the vacuum of space once you've gotten far enough out of Earth's gravitational field to care more about efficiency than about generating enough acceleration not to be pulled back down to Earth.
Its major weakness? Price; the RL-10 is one expensive piece of hardware. At some points in the past it has cost as much as $40 million per engine, an insane price that could never be viable. I don't know what they are worth now, but United Launch Alliance says that they are roughly in line with the cost of the RD-180, slightly lower - so I'd put the RL-10 in the $10-12 million range right now. Although quite expensive, it is a very good engine from an engineering perspective, and it's no wonder that it has been a staple of American rocketry for half a century.
The Atlas V was developed for the US Air Force's Evolved Expendable Launch Vehicle (EELV) program, an Air Force program that sought to create high-quality rockets to ensure that the US military would not be locked out of being able to launch their valuable military payloads. After the Challenger disaster happened at the same time as the Titan rocket also was taken off the market for a time due to failures, the USAF was rightfully worried that they might be locked out of the business, and that they might lose a chance to be competitive in the commercial space industry that was poised for growth.
A staple of the past few decades has always been assuming that the commercial space industry was on the verge of taking off - and the Air Force, at the turn of the century, found that it wasn't actually going to happen. They were left with two high-quality, but highly expensive, launch vehicle families: Atlas and Delta, both of which eventually became managed by United Launch Alliance (Atlas coming from Lockheed Martin, Delta from Boeing). Of the two, Atlas was significantly cheaper, but the USAF wanted to make sure to have two separate rocket lines available, so they funded both of them. Although the price of the rockets was never all that great, they were both capable of performing all of the fancy maneuvers required to put Air Force satellites where and when the Air Force wanted them there, so both rocket lines survived despite being absurdly expensive. For reference, the Falcon 9 starts at $62 million, the Atlas V at $109 million, and the Delta IV at a whopping $200 million. And the Delta IV Heavy, a princely $400 million a pop.
In recent times, the focus has changed. While the Air Force is not a customer that is willing to lose rockets even if they have to pay more money for them, money is a finite resource and the Air Force is rightly interested in reducing their costs. The Falcon 9, which recently started to compete for Air Force missions, provided some degree of the competition that the Air Force was looking for. While it still isn't capable of performing many of the missions the Air Force requires (due to design limitations I will address shortly), it nevertheless managed to start competing for Air Force missions, forcing United Launch Alliance to start looking for ways to drive down costs.
The Atlas V, though a good rocket, does have two major issues that need addressing: its engines. The RD-180, though a fantastic and relatively cheap engine, has always made lawmakers quite uncomfortable; why should the US be dependent on Russia to be able to launch its rockets? In the wake of the US sanctions on Russia for the Ukraine crisis, SpaceX used a legal/political campaign to try to ban the RD-180 outright. Although unsuccessful. it did force ULA to finally start pushing to replace the engine for an American model. And the other problem they have is the RL-10, an engine that, while powerful and necessary to perform the kinds of missions ULA needs to perform, is just too expensive in a cost-sensitive world.
The answer to all of these issues that ULA has come up with was the Vulcan, an Atlas V derivative that hopes to replace both engines with cheaper engines from Blue Origin, a major New Space darling. They hope to reduce launch costs to just $100 million, to be able to have prices that are low enough to convince both commercial and military customers to launch on ULA rockets. While still not as cheap as the Falcon 9, that would allow Vulcan to perform its more complex missions while still being cheap enough to be worth it for commercial customers. Although it does plan to offer some degree of reusability (dropping its first stage engines inside a heat shield to be picked up by plane), it's clearly not Vulcan's main selling point.
![[image loading]](http://i.imgur.com/47NPIYg.jpg)
United Launch Alliance Advertisement for Vulcan
So what does all this history have to do with reusability? Simple: to understand why a rocket may or may not be designed to be reusable, you have to understand why it was designed in the first place. The Atlas was designed to be able to win launches from a customer desperately concerned with ensuring that it would never be denied the ability to launch its mission critical satellites into any orbit, at any time, so driving down launch costs to their absolute minimum was not really the biggest priority.
With mission success in the face of any kind of mission being the highest priority, the expendable design of Atlas made perfect sense. As the Shuttle showed, complexity and reusability don't go well together, so if complexity is required, it might make sense to toss aside reusability. The Air Force doesn't launch all that often either - so the scale of rocket use isn't sufficient to make reusability worth it. Since performance over the course of a small number of missions was what really mattered, reusability simply doesn't make sense for Atlas and we can conclude that without even looking in depth at the numbers. With Vulcan, which attempts to appeal to a much wider market, will that change? That's a slightly more difficult question to answer; it depends on the market. We'll be looking at that market later.
Incidentally, although it is seldom advertised, Atlas has performed many important missions for the Air Force and NASA that involve some rather complex and high-performance orbital maneuvers. Many important NASA missions, including Curiosity (Mars rover), New Horizons (Pluto satellite), and the Lunar Precursor Robotic Program (found water on the Moon), were launched on the back of an Atlas V rocket. A Falcon 9 rocket, for example, is not capable of performing missions like these.
Reduce, Reuse, Recycle: The Falcon 9
So now that we have set the stage, let's talk about the rocket that is the real focus of today: the Falcon 9. Designed by Space Exploration Technologies Corporation (SpaceX), it has promised to revolutionize space by offering rapid, cheap reusability to make it cheap to launch any payload into space. Recently, it's managed to reuse its first stage boosters, claiming some massive savings in the process. Have they managed the feat they promised? We shall see - but first we will talk about the design of the rocket itself.
![[image loading]](http://i.imgur.com/BYg7Bag.png)
Meet the Falconsons
In standard Silicon Valley fashion, the Falcon 9 was designed as a series of iterative improvements on the original design. Each successive version was a little more powerful, a little more advanced, and overall a little better. Two new designs are in the pipeline: a Falcon Heavy variant, a heavy-lift rocket capable of competing with ULA's Delta IV Heavy, and the Falcon 9 Block 5, an improved version of the current Falcon 9 Full Thrust that will make it even easier to reuse the rocket.
As far as I know, SpaceX doesn't make the kind of infographics that ULA does, so I can't just give you an official diagram of the Falcon 9 like for Atlas or the Shuttle. But I found a decent unofficial version - for an old Falcon 9. It will have to do for now.
![[image loading]](http://i.imgur.com/CyRiyqr.gif)
The Falcon 9 is a two-stage rocket, with both stages utilizing SpaceX's own Merlin engine line. Designed for ease of production, the Merlin engines are known for being remarkably light relative to the amount of thrust they produce (they have a very high "thrust to weight ratio") which, although generally not all that meaningful (the more meaningful metric is specific impulse, the "bang for your buck" for your fuel), would mean that it would not be all that hard to pick up and retrieve just the engines. The first stage consists of nine Merlin engines, with an upper stage that consists of just a single Merlin - for a grand total of ten per rocket.
The Merlin is a line of kerosene/liquid oxygen engines - which, if you read my first entry, you would know is great for first stages, but not all that great for second stages, where hydrogen is best. A hydrogen engine is more efficient on a per-mass basis, but kerosene is more efficient on a per-volume basis. This means that while hydrogen can give you better efficiency, kerosene does have one very important advantage: it's denser, so it's much easier to burn a lot of it and make a lot of thrust all at once. You need as much thrust as you can generate when you're trying to escape the Earth - every second you spend in the high-gravity environment is a lot of lost acceleration. Once you're clear of the Earth, though, you want and need some of that added efficiency to be able to get further on your small, but highly effective, last bit of fuel. And hydrogen can burn for far longer - as opposed to the 700 seconds of the RL-10, you can get 400 seconds of burn out of the most advanced Merlin. You could go into a lot more depth, but in short, the RL-10 is a far superior second stage.
Although it's easy to say that the RL-10 can run circles around the Merlin for an upper stage, it begs the question: why does the Falcon 9 use a clearly worse engine? Price. Although the exact cost of a Merlin is hard to determine, estimates realistically put it in somewhere in the $1-3 million per-unit range. My guess is that it would be somewhere around $2.5 million a piece; it's unlikely to be cheaper than $2 million a piece, and unless the Falcon has a significantly different cost structure than most rockets I've seen, that is a reasonable estimate. But by using a weaker engine, SpaceX saves a good $8-10 million on what the RL-10 can do. Is it worth it? Again, it depends on what you're going for. It's definitely an important staple of the low-cost reusability strategy that SpaceX is going for, though.
![[image loading]](http://i.imgur.com/iOai7HH.jpg)
Merlin 1D Test Fire
As with everything else about the Falcon 9, the Merlin engine has gone through many iterative improvements. With each iteration, it's become more powerful and allegedly cheaper and easier to produce. With nine of those little tykes in the first stage, each improvement gets you quite a bit more thrust, and lets you put larger and larger payloads into orbit. The current iteration is the Merlin 1D - so as you can see, the above diagram is fairly out of date.
An astute observer might notice another important piece missing from the Falcon 9 above: the landing legs. The factor most associated with SpaceX, of course, is their first stage booster landing on a sea-based drone platform. And that, of course, is at the very core of what the SpaceX plan for rocket reusability consists of. They cost some not-insignificant money to install, but if reusability works as they hope, it will be worth it.
From a technical perspective, this landing is certainly quite impressive - it requires some high-performance coordination between the rocket and the drone platform, and some well-timed rocket firings to stabilize the rocket to allow it to land like that. Nothing that looks like it might be impossible, but nevertheless highly impressive - and boy does it look cool. However, the hopeless cynic - or the rocket engineer - might ask if, for all the fanfare, this is a good way to reuse a rocket. The answer, in a traditional lawyerly fashion, is that it depends. It's certainly worth a look, that's for sure.
Let's start positive by looking at the advantages of this method. Unlike just allowing the booster to fall into the ocean and reach scrap-heap levels of damage, the rocket comes back quite intact, with little more than the standard wear-and-tear of use. While that wear-and-tear is significant because rockets are high-performance vehicles (the same thing happens to military jets, and rockets are basically what military jets wish they could be in terms of performance), it's still very manageable. Tick off minimizing wear; that definitely does the job well. It keeps the main structure of the rocket intact, which if well-executed would do a good job of reducing necessary maintenance work. And finally, though this may seem vain or a back-handed compliment, it's not: it looks cool. SpaceX has certainly shown that while finances matter, optics do too, and landing a rocket at sea certainly looks fantastic.
So what's wrong with this method? The most cited, and most significant, flaw of this method is that it uses up a lot of fuel. You have to save enough fuel to land your booster, and that hurts performance. The payload mass that can be carried while the Falcon 9 is operating in reusable mode is reduced by about 30%. Less cited, but also important issues, include the cost of the drone ship fleet (or the cost of clearing a landing area if you land on the ground), the added complexity to the mission and rocket hardware when you add in a landing maneuver, and of course all the logistics and infrastructure that have to be developed to support reuse of a rocket.
Playing the part of the "expendable-inclined skeptic" my first question would of course be, why not just use a smaller rocket? If you're not using all of your fuel, why bother recycling - you could just save by making an overall smaller - and cheaper - structure! And when you need a little more performance? Add a few strap-on solid rocket boosters like Atlas V does. If you reuse your rocket more often than not, then why not just save yourself the hassle of reuse - which, as I hope I've demonstrated, is significant - and just save on the production end?
This is probably the biggest question from the end of whether or not you should reuse at all. If we're going back to the "cars and planes" analogy, would you not rent a moving truck when you need to have that extra load and have something smaller like a sedan or minivan, rather than just drive a giant truck wherever you go? Would you pay for a fleet of high-end military-grade cargo planes when a cheaper cargo plane does the job just fine 95% of the time? But of course the other side of that analogy says that you wouldn't want to throw away your craft every time either, so the analogy goes both ways. So who's right? Without looking at the numbers, it's difficult to say, so this is a question we're going to leave aside for the moment.
Now let's play the "reusability-inclined skeptic" and ask whether this is really the best way to reuse the Falcon 9. There are two alternatives that quickly come to mind here: jettisoning and performing a mid-air recovery of just the engines using a heat shield and parachute, saving the most valuable parts at a fraction of the cost, or allowing the booster to float down most of the way with parachutes, saving yourself a lot of fuel by letting air resistance do most of the work for you. Why recover by landing a booster, rather than by reusing with some other method?
The "jettison the engines" technique, as advocated as a possibility by ULA among others, clearly is not right for Falcon. The Atlas has only a single, massive first stage engine, and the Vulcan that will succeed it will have two slightly smaller first stage engines. For that, it makes it not that hard to just bundle them up and send them on a skydiving adventure. But for Falcon? There are nine engines, and while there are certainly some important advantages to having smaller engines, it does come with an important drawback: increased complexity. You have to put down separate plumbing for each and every engine, you would have to detach each one separately, and bundling them together in a heat shield would pose all sorts of problems. Although having multiple, simpler engines might make the Merlin easier to mass produce, it does mean that this specific scheme loses a lot of its meaningful simplicity and is simply from a technical perspective clearly more trouble than it is worth.
What about parachuting the booster down? Well the idea of a midair recovery is essentially dead on arrival; the full booster is just too heavy for that to be worth it. But what about letting parachutes cushion most of the fall and then land it? It's hard to give a good, definitive answer without knowing what the process at SpaceX was for deciding how they would make their rocket. If I had to guess, though, I would say that it's a matter of predictability; it's much easier to track a rocket that is being guided by predictable engine firings than by the wind, and at the heights that the rocket travels at, the difference from the wind can be at least a few miles - which would make it quite hard for a drone ship to keep up with. Overall, I could easily buy that SpaceX decided that it was better to expend fuel than to deal with whatever way the wind blows.
Another question from the reusable-inclined skeptic: what about the second stage? While SpaceX says that ultimately, reusable second stages will be a necessity, right now they're not really an option. The Merlin is not a great upper stage, and it probably barely has enough fuel to finish the kinds of missions that SpaceX tends to take on, which are already among the less complex of space missions. As far as I know, the Merlin upper stage doesn't always even have enough juice to deorbit itself; it will often just sit in space and contribute to the cloud of space debris until gravity takes its toll on it. That's definitely not enough to make up a scheme in which the upper stage can be guided back home without compromising the mission itself. Instead, Falcon focuses on simply reducing the cost of the upper stage, by having a rather cheap engine on top. There will still be some rather valuable electronics (the avionics on the upper stage) that will be lost - probably most notably the flight computer - but it can't always be helped. You have to finish the mission, and you need your avionics to the end, so in this case it's probably best to just let them be expended until something better comes along.
So now I think we have enough to start analyzing the design of Falcon 9 as a whole. Adapting a concept from the Silicon Valley culture from which SpaceX was designed, we're going to focus on two main questions: did SpaceX build the rocket right, and did SpaceX build the right rocket? This same kind of question would be asked for software, and in this case, it seems appropriate to do the same. We have enough to understand what kind of rocket SpaceX was hoping to build - now we just have to ask ourselves if they did a good job, and whether or not they were barking up the wrong tree.
Did SpaceX build the rocket right? Well to answer that question, first we should ask what kind of rocket, in terms of high-level goals and ambitions, SpaceX was trying to build. Thankfully, that is something that SpaceX and its CEO Elon Musk talk a lot about. Musk has often said that the key to unlocking massive economies of scale in space is to reuse rockets, that since the cost of fuel is not even one percent of the full cost of a launch, that once we can reuse rockets rapidly and cheaply, the space economy will burst wide open to businesses that would otherwise not be profitable. If we take that as the firm, inflexible goal of SpaceX, are they designing their rocket right? Well I'm sure that even the most fanatical of SpaceX fans (well, maybe not the most fanatical, but you know what I mean) would agree that that level of cost reduction is absurdly optimistic. But if you look at how well they've done with that as a goal, it's hard to say that they made the wrong decision. They check off the major prerequisites of a reusable system quite well: cheap, low complexity, high-fidelity recovery, with a mature "final" design in the works: the improved Falcon 9 Block 5. So I can say they earn a passing score on this one for sure.
The next question we have to ask is, did they build the right rocket? Maybe they did a good job on reusability within the realities that were available to them, but was the scope and benefits of reusability just a pipe dream that was ill-conceived and not worth it under the current conditions of the market? That is the real meat-and-potatoes of this whole discussion, and what this already alarmingly long post is building up to. Atlas V is a good reference - it was built to give the Air Force what they wanted, and they did a good job at that. Falcon 9 was built to leverage reusability and low-cost manufacture to make space cheap; was that the right goal? As with many things in life, the answer to that question lies in the economics of it all.
A Brief Primer on Economic Analysis
Before we go off analyzing the economics of reusability, it would be a good idea to define the scope of our analysis so that we're all on the same page. Of course, for a more rudimentary analysis with only approximate numbers available, it makes sense to keep it relatively simple. Nevertheless, for those less familiar with economics, some definitions are in order. And as I've mentioned before, while it doesn't mean that if something works on paper it will work in the real world, it is true that if it doesn't work on paper then it almost certainly won't work in the real world. Furthermore, these analyses are almost by necessity optimistic - when you abstract away all the nitty-gritty, you usually find more and more things that could go wrong, but you will miss the forest for the trees if you look at all those numbers before you look at the bigger picture of whether or not the idea is worth pursuit at all. A corollary to that is that if your optimistic numbers look mediocre, if ultimately profitable, it's probably a bad investment.
The first idea we need to talk about is one already mentioned: economies of scale. If the above Wikipedia link wasn't to your liking, I can offer Investopedia as another explanation of the concept. Bottom line, when you have to pay a lot of up-front money but not so much for every individual item you produce, you want to be able to create more items, because that will shoot your average price of production way down. Conversely, if you're investing in something that will require a lot of up-front cost for just a few units, expect that it won't be particularly economical. Each of Elon Musk's businesses, SpaceX included, makes claims of economies of scale waiting to be unlocked - so it's understandable that this concept will see a lot of attention when discussing any of his ventures.
Tying into the idea of economies of sale is the idea of price elasticity of demand - briefly put, the percentage change in quantity of a good demanded per percentage change in the price of that good. Simply put, that ratio is a way to determine how sensitive your customers are to a change in price. Will a massive price hike cost you a lot of customers or just a few? Will a massive price drop net you a lot of customers or just a couple? The price elasticity of rocket launches is what's going to determine this.
A quick note on elasticity, though: no company is alone in the market. If, for example, the market is highly inelastic, high prices can be charged by all and all will make a reasonable profit - but if one producer wishes to charge less than their competitors anyways, they will just get a bigger slice of that pie, rather than attract a particularly large base of new customers. In general, not the best strategy in an inelastic market - you could charge more for less and make a bigger profit, even if you have to share your customers more.
The next concept of interest is the idea of present value of money. As a general rule, a dollar earned today is more valuable than a dollar earned tomorrow, which is more valuable than a dollar earned a year from now - colloquially, a nickel ain't worth a dime anymore. Of course inflation plays a role in that (and is incorporated into a present value analysis in sufficiently dedicated models), but the simple explanation for that is simply that a dollar today can be invested and utilized to make more than a dollar tomorrow. This number can be quantified into a percentage known as an interest rate, defined somewhat arbitrarily but not without meaning. However you choose to define your interest rate, if it's at, say, 8% per year, then a dollar today is worth $1.08 a year from now - or equivalently, $1.08 a year from now is worth $1 today (or $1 a year from now is worth 1/1.08 = ~$0.926, 93 cents rounded, today). Compound that number for every year - $1 today is worth (1.08)^2 = ~$1.17 in two years, and the same goes in reverse. But at the end of the day, you want all that money to be time-equivalent, and the best time to do that when looking into the future is to analyze it in terms of current worth - the present value of your future earnings.
A small extension to that concept - the net present value - is a useful tool for rudimentary economic analysis. All you do is you add the cost of your investments (a negative profit) to the same calculation as the present value of your profits, extend that calculation to however many years of operation as make sense for that calculation, and you find your net profit, in present value, at the interest rate of choice. Of course, as mentioned before, the choice of an interest rate is somewhat arbitrary - so wouldn't it make more sense not to have to choose an arbitrary number for our model? Indeed it does - so an important part of a net present value is an interest rate known as the internal rate of return (IRR) - the interest rate that makes your net present value equal to 0. The higher the better - an internal rate of return of, say, 10%, says that your investment is basically the equivalent of investing your money and making 10% a year on that investment. As a rule of thumb, a generally "good" IRR is around 20%, a number that has the makings of a pretty good investment. If your IRR is fairly low (<10%), then the realities of business will grind you down further and it's a pretty bad investment.
Finally, it's important to remember that economic models are just that: models. They come with approximations, estimates, and assumptions, and on their own they are little more than an educated guess. That's why it's important to cite some empirical evidence to show that your models are well-rooted in reality. And the best way to do that for SpaceX is to look at this now-famous Wall Street Journal piece that analyzes internal financial documents of the company. Unfortunately, it is paywalled - so unless you have a WSJ subscription you probably can't read it. I will summarize all the relevant parts as need be, though - and if you really want to read the original, then please PM me.
I will also note that much of my analysis is inspired by this paper and this paper and I will take a few ideas and possibly a graphic or two from them - but ultimately, the analysis is my own. Much of the general information about the satellite industry is taken from this report, and the WSJ piece will hold a lot of the information about the finances and the goals of SpaceX. Forgive me if I don't always cite which specific source I got any specific number from - if it isn't clear, just ask. But it should be fairly obvious; each of these sources covers a very different aspect of the topic.
If you've gotten this far, I'm fairly sure that you'll read to the end, so I have a favor to ask. I'm genuinely curious how many people read these through to the end, instead of just browsing, so I want to ask you, please vote in this poll: + Show Spoiler +
Poll: Vote below.
I read this far. (11)
100%
11 total votes
11 total votes
Your vote: Vote below.
(Vote): I read this far.
Rocket Economics
We can, roughly speaking, divide the costs of a rocket into five components:
1. R&D Costs. All rockets have these, but generally you will have to pay more for reusability. SpaceX claims that their R&D costs for reusability alone were around $1 billion. We will explore this number, but in general we will take those costs at their word - a range of $1-2 billion seems reasonable for Falcon 9 reuse development.
2. Rocket construction costs. For expendable rockets or components, you have to build one every time. For reusable rockets, this can be the average cost of building a rocket over a reusable rocket's lifespan. Keep in mind that there is an important economy of scale here - the more rockets you produce, the cheaper they are on a per-unit basis. When you are reusing, it's going to be a lot more costly to make new ones whenever that becomes a necessity. Incidentally, this could actually be a good advantage for SpaceX having a Merlin upper stage - it will make it easier to keep their Merlin supply line running if they have to make a new one each time. But in general, yes, reusable rockets will cost more to make unless your business size expands enough to make the point moot.
3. Flight Operations and Services. This is the biggest and most expensive part of your rocket, easily. This includes planning the mission, testing your rocket thoroughly leading up to the launch, transporting your vehicle to the pad, assembling your rocket on the pad, performing mission control operations upon launch, and everything in between. This is where the "rocket science" mythos really falls, and you have to do a whole lot of stuff each time. Every payload is special, every orbit is special, every launch condition is special, and every nook and cranny of possible vehicle performance is special - and rockets are too fickle to leave anything about this to chance. Paying specialists to do their job is fairly expensive, and you can expect to spend at least half the cost of any given mission on this. Incidentally, logistical improvements will do a lot to bring your costs down, whether you use a reusable or expendable rocket.
4. Reuse costs. This includes the cost of operating your drone ships, your refurbishing crew and equipment, and any other infrastructure you might need to have in order to actually reuse your craft. Ideally this is going to be small compared to the costs of building a new rocket, but in the current market it's not that easy for this to be true. You need an economy of scale for that.
5. Insurance. You have to pay to insure your products in case of trouble. Shouldn't really matter if reliability is set aside, but that has little to do with reusability.
In general, 3 and 5 are the same for expendable and reusable rockets, and their most important factor here is simply to show that fuel is far from the most expensive aspect of a rocket flight. With modern technology, I could not imagine a rocket that would not cost tens of millions of dollars to launch. Maybe once we have starships that are of the same caliber as those in Star Wars or Star Trek, a few hundred thousand for a launch might be feasible - but we're not even close to there yet. we're still working with experimental technologies here.
4 is only for reusable rockets; it goes without saying that if you just throw yours away you don't have to worry about reuse costs. 1 is common to both, but reusability will require a larger R&D cost. 2 is of course larger when you're producing a rocket each time, possibly significantly so. The real test of reusability is if you can make the difference in 1 so small that it doesn't really matter, and if your costs for 4 increase a lot less than your costs for 2 decrease. That's where the net profit is. It would be hopeful that your reuse of a rocket saves you at least some money, but keep in mind that if you're making a rocket to throw it away, your costs for 2 are probably lower than if you made the fully reusable rocket and never reused it, as that's something you're optimizing for.
A lot of this of course depends on how many rockets launches you can sell - and what your customers care most about. Roughly speaking, you can divide the current market into three groups:
1. The military. Satellites and certain other special payloads are important to the success of the military. These are expensive, often multi-billion dollar payloads, that really must get into space at any cost at the right time. Certainly, they will be price sensitive because budgets are a fact of life, but if the only option is to pay $400 million to get their cargo into space with a high level of certainty, they will suck it up and pay the bill every time. But you have to give them everything they want, when they want it, however painful or annoying you might find that to be. Also, this one is military secrets, so no foreign launchers allowed. Everything here is a matter of national security.
2. Civil and government launches. A few of these do fall under the military, but this essentially includes science missions and anything done by the government for the benefit of its citizens. GPS launches and ISS missions fall under this umbrella. These tend to be a bit more price-sensitive, but if the government is desperate it will foot the bill (the ISS has been in danger more times than NASA will ever admit - they will pay a premium for deliveries to the station if they must). Domestic launchers highly preferred, but not always possible - the Russian Soyuz-FG is as of now the only rocket that launches manned missions to the ISS. SpaceX makes at least half of their entire revenue here, between lucrative NASA contracts and some other launches often administered through the Air Force. These missions can range from simple (or not-so-simple) deliveries to low-Earth orbit to convoluted science missions that involve some rather impressive orbital maneuvers to complete.
3. Commercial launchers. These guys are out to make money - a well-planted satellite can generate a few billion dollars over its lifetime, costing only a few hundred million to both launch and build. Their goal is to make money, so they will pick the option that will save them the most money - and their payload is insured, so if they lose a satellite on launch they will get reimbursed; they will just lose some profit for a period of time. Nevertheless, price is not their only concern - waiting a long time for a launch will make them unhappy because it costs millions to not have their satellite operating, and the cost of the launch is significantly less expensive than the cost of making the satellite itself. For reference: in 2015, the launch industry had a revenue of $5.4 billion, the revenue from satellite construction was $16.6 billion, and the overall industry had a revenue of about $200 billion. Commercial payload owners are sensitive to price, but there are limits to that. SpaceX is a major player in this industry, competing largely with foreign rockets like Ariane and Proton.
The launch market is generally considered to be fairly inelastic, which should be no surprise. Two of the three major classes of customers, who together hold most of the money spent on launches, have concerns more important than price, and are willing to pay whatever it takes to get the mission done. Certainly, they will want discounts - but they will be willing to pay a premium to get what they want. That often, but not always, involves some rather specific or unusual requirements. Incidentally, this is why I think United Launch Alliance is more important as a competitor to SpaceX than Arianespace - Arianespace and SpaceX compete for commercial cargo, while ULA is able to compete with SpaceX on all fronts. And with Vulcan, they intend to try to push for commercial, which has tended to be too expensive to be worth it on Atlas or Delta.
If there is a hope for an economy of scale of launches to open up when the launch price goes down, it would have to be in the commercial sector. The military and the government pay a lot for the few launches they have, but when your cargo costs a few billion, you care less about an extra hundred million spent. And those billion-dollar payloads are few and far between as it is; cheaper prices won't change that. But for commercial launchers, who care about cargo? It's a possibility.
Unfortunately, it's not a good bet. If the numbers above are to be believed ($16.6 billion on sats, $5.4 billion on launches), even a launch provided for free would only net you about 25% savings on the production-plus-launch cost - and that would be a very small number compared to the revenue generated over the lifetime of the satellites. Massive growth in the satellite industry has been predicted in the past - it never really came to fruition; the growth has always been fairly small and incremental regardless of rocket cost. There is a notable slowdown in the commercial market within the past year and a half; this suggests that SpaceX is merely siphoning orders from their competitors rather than opening economies of scale. There is good reason to believe this market remains inelastic.
What about new industries? Well you could dream up an infinite number of them, but most of the commonly cited ones are not going to be all they are hyped to be. Space tourism is one often cited, unfortunately hindered by the lack of rich people willing to die on an experimental piece of hardware in the 5%+ chance that your vehicle explodes. The few exceptions can hitch a ride on a Soyuz when the Russians have a seat to spare, or on a Commercial Crew vehicle when they become available and NASA allows it. Space mining is a possibility, although probably further off than one would like - there are plenty of precious ores that would be worth mining as soon as the logistics of mining in a vacuum and large-scale space transportation figured out. It would require more than just cheap launches though; that technology would have to be developed significantly more than it is right now, so any profits from that possibility are a ways off. And the final oft-repeated idea is that of small satellites (cubesats), which are fairly cheap to produce individually - but most useful in large quantities that kind of make the entire point moot. Without a good engine for "economy of scale" growth, it's fairly reasonable to assume that the market will remain inelastic for the the foreseeable future - so it's hard to determine where all the extra growth will come from.
So that's the general market we're working with - from both the supply side and the demand side. Money is there to be made - but anyone hoping for massive growth has historical precedence to believe they will be horribly disappointed by the trajectory of the market.
SpaceXonomics: The Inside Scoop
SpaceX is a private company - and they tend to be quite tight-lipped about their technical specifications and their finances despite being quite media-friendly overall. The previously cited Wall Street Journal piece, reviewing SpaceX internal financial documents in the 2015-2016 period, spilled the beans with a deeper look into the company than we've ever had before. What was revealed was a company largely in line with standard Elon Musk fare: razor-thin margins, overall unprofitable, a stack of cash from investments into the company, and hopes for astronomical growth in the near future. That, along with some rumblings I've heard from the inside and outside, will make up this section - a sort of "gossip collection" that gives a look into the company beyond the scope of what is most generally talked about in the generally overwhelmingly bullish views on SpaceX. It will help inform our final analysis.
The operating margins on SpaceX operations are quite low - in a "good" year, with many launches, they get hundreds of millions in revenue which translates to only $10-20 million in profit, translating to a rather thin margin. In a bad year, SpaceX loses a rocket - and hundreds of millions are lost; in 2015 when they lost a NASA cargo mission, despite being paid 80 percent of the delivery cost they lost about $250 million (and they had to quietly reimburse NASA over later launches for that loss, which means they are probably not profiting off of NASA's Commercial Resupply Service program anymore). They lost another rocket in 2016, which along with those costs, destroyed one of their important launch pads, which they have to repair. They are buoyed by large cash investments - from Alphabet (Google) and Fidelity investing a billion, and up-front payments for multibillion-dollar NASA work such as Commercial Crew - and as of the first loss they were cushioned by a $1.3 billion cash reserve. They likely have been forced to raise more money since then, as the second launch would strain that supply very fast, because they have invested a lot in infrastructure since then, and that doesn't factor into operating income.
SpaceX also had goals for a rapid expansion of their flight rate - up to 20 in 2016 (ended at 8 after their 9th rocket blew up on the pad), 27 in 2017, 44 in 2018, 52 in 2019, then back down to 41 in 2020 - and presumably about 45 launches a year after that. They also had plans for a satellite internet network which would generate $30 billion a year by 2025. By 2020, their profits are suggested to go positive - and rise rapidly. All of these goals are wildly optimistic, to put it lightly - that would require cornering half of the entire satellite launch market (much of which SpaceX would not even be allowed to compete for due to operational/national requirements that they do not satisfy) and eclipsing the size of existing internet providers in just a couple years.
In truth, the company is in far more precarious a position than it would ever be willing to admit. In the 2014-2015, they were riding high on a tidal wave of hype - no rockets lost in years, many orders for their cheap rockets, completed Commercial Crew just on the horizon, Falcon Heavy about to launch, they got their "in" into the Air Force market, and they were able to put a lot of pressure on ULA to hamper their main competitor. When they lost their first rocket, that slowly started to change - NASA was unhappy, but forgiving, despite the fact that it was fairly embarrassing to be put in that situation (US failed to resupply the ISS, then Russia did, then SpaceX/US did, and then Russia finally managed to do the mission). With the 2016 loss, though, their fortunes started to look far more uncertain. The loss itself was pretty bad, and it led to delays galore - between the two, SpaceX delayed their launches for their customers by at least a year. Plus significant operating losses. They have hype again for their reuse of a rocket, well-planned as their return to flight media windfall - but behind the scenes are many unhappy customers with a media that won't be their friend forever.
By far SpaceX's most important customer is NASA - and I know many, many people in NASA that are displeased with SpaceX. For all the investment that NASA put into SpaceX, they feel that SpaceX has only marginally cared about doing what NASA wanted of them, despite some significant investment on NASA's part into their success. Not to mention that the Falcon 9, the rocket that is supposed to launch astronauts in just a year, had two accidents in the past two years. Similarly, the Air Force has its own set of gripes with SpaceX - they don't like losing rockets, and they don't like how the political ventures that SpaceX is pursuing, such trying to ban the RD-180 to make Atlas V not be able to fly, forcing the Air Force to either spend an extra $100 million a pop on Delta or wait until SpaceX develops the capabilities the Air Force needs, which is never on schedule. Commercial customers, too, are often unhappy - many of them have suffered multiple years of delays, which cost millions of dollars in lost revenue.
More than that, though, I think that SpaceX has a vulnerability to the possibility of rising costs on the personnel end. SpaceX is notorious for long hours and low pay, which many younger individuals are willing to endure because they are "working for the future." Once the hype subsides, those salaries will have to go up, or else. Operational costs are quite substantial for rocket launches.
The point of talking about all these issues is twofold. The first is that we need the launch numbers (cost, desired frequency, and so on) that SpaceX has to make good estimates. The second is that it's a good idea to get a handle on how optimistic any given prediction SpaceX makes is - so that we can adjust our expectations accordingly. And now, all the pieces are finally in place to crunch some numbers.
Falcon 9 Reusability: Return on Investment
As I mentioned before, we're going to do a net present value analysis on the SpaceX reusability scheme. For this analysis, we will pick a ten-year time period on which we will analyze return on investment; past ten years, the rest of the market will have had plenty of time to adjust and the advantage from that specific investment will have dissipated, if such an advantage exists. SpaceX will keep their launch costs at $60 million, which will be assumed to also be the cost of making the rocket new (which isn't far from the truth). Running multiple scenarios, we will ask a few questions:
1. How much money did the R&D and infrastructure to develop reusability cost?
2. How much money is saved on a reused launch of a Falcon 9 booster?
3. How many launches per year can SpaceX expect? What is the IRR at that condition?
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions?
We will do this calculation with a simple spreadsheet, given below:
![[image loading]](http://i.imgur.com/eIT9I4q.png)
These scenarios will be listed in order from most optimistic to least optimistic.
Scenario 1: Rainbows and Unicorns
Just for fun, let's start with the best possible conditions you could. Launches cost only the cost of fuel.
1. How much money did the R&D and infrastructure to develop reusability cost? $1 billion
2. How much money is saved on a reused launch of a Falcon 9 booster? Pretty much all of it, so $60 million.
3. How many launches per year can SpaceX expect? What is the IRR at that condition? Let's do 45 launches a year. Which gives us a princely IRR of 270%.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? Four.
Not realistic of course, but let's cover the extremes!
Scenario 2: Official Optimism
Gwynne Shotwell, president of SpaceX, says that they can save 50% per launch on their reuse. And I see about 20 missions scheduled per year - so we can use that number here as well. Let's assume no delays. And officially, R&D is still $1 billion for reuse.
1. How much money did the R&D and infrastructure to develop reusability cost? $1 billion.
2. How much money is saved on a reused launch of a Falcon 9 booster? Half, so $30 million.
3. How many launches per year can SpaceX expect? What is the IRR at that condition? 20 launches. Giving an IRR of 59.4 percent.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? Eight.
Scenario 3: Realistic Optimism
Estimates from those who are not trying to sell a reusability scheme to the public say that the best SpaceX could hope to save is around 25% of the cost. And, let's face it, SpaceX will slip on a few launches, so let's aim for 15 a year for 10 years. We can still look at an R&D cost of $1 billion.
1. How much money did the R&D and infrastructure to develop reusability cost? $1 billion
2. How much money is saved on a reused launch of a Falcon 9 booster? $15 million
3. How many launches per year can SpaceX expect? What is the IRR at that condition? 15 launches - which gives an IRR of 18.3%. Still a pretty good potential investment.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? 16. Just one more would do it.
Scenario 4: Realistic Optimism, Cost Overruns
Maybe R&D isn't quite $1 billion - you have to run drone ships and pay up-front for a lot of infrastructure, after all. Maybe it cost them $1.5 billion to develop it after all. What would that change?
1. How much money did the R&D and infrastructure to develop reusability cost? $1.5 billion.
2. How much money is saved on a reused launch of a Falcon 9 booster? $15 million.
3. How many launches per year can SpaceX expect? What is the IRR at that condition? 15 launches, and an IRR of 8.2%. Not really great. A good indication that the profit is very sensitive to up-front costs.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? Goes up quite a lot - to 24. Economies of scale start to kick in a fair bit over these cost expectations.
Scenario 5: Moderate Success on the Cheap
1. How much money did the R&D and infrastructure to develop reusability cost? $1 billion
2. How much money is saved on a reused launch of a Falcon 9 booster? $10 million
3. How many launches per year can SpaceX expect? What is the IRR at that condition? 15 launches, for an IRR of 8.2%.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? 24, as before.
Scenario 6: Moderate Success, Moderate Cost Overruns
1. How much money did the R&D and infrastructure to develop reusability cost? $1.5 billion
2. How much money is saved on a reused launch of a Falcon 9 booster? $10 million
3. How many launches per year can SpaceX expect? What is the IRR at that condition? 15 launches, IRR of zero.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? A whopping 36.
Scenario 7: Moderate Success, Severe Cost Overruns
1. How much money did the R&D and infrastructure to develop reusability cost? $2 billion
2. How much money is saved on a reused launch of a Falcon 9 booster? $10 million
3. How many launches per year can SpaceX expect? What is the IRR at that condition? At 15 launches you won't turn a profit on your investment at all.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? 48.
Scenario 8: Competitors' Predictions
Many competitors have put their predictions for launch savings at Falcon 9 reuse at 10%. What would be the results of this kind of profit margin at the specified $1 billion cost of R&D?
1. How much money did the R&D and infrastructure to develop reusability cost? $1 billion
2. How much money is saved on a reused launch of a Falcon 9 booster? $6 million
3. How many launches per year can SpaceX expect? What is the IRR at that condition? 15 launches won't turn a profit; 20 launches gives an IRR of 3.4%.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? 40.
Scenario 9: Flop
After $2 billion spent, the rockets are but $1 million cheaper a pop!
1. How much money did the R&D and infrastructure to develop reusability cost? $2 billion
2. How much money is saved on a reused launch of a Falcon 9 booster? $1 million
3. How many launches per year can SpaceX expect? What is the IRR at that condition? You won't turn a profit on any reasonable number of launches.
4. To achieve a 20% IRR, how many launches does SpaceX need at those cost conditions? 477.
Putting it All Together
So there we have it: a nice spectrum of possibilities for how well the investment in reusability will work out. If hype is to be believed, it is an amazing investment that will quickly and easily pay for itself. If the competition is to be believed, though, then it really was a losing prediction. So what is realistic for SpaceX? In truth, there is no easy answer, and we will have to do a bit of guesswork.
First question is, how much did the R&D actually cost? That is fairly difficult to accurately predict - but I do think it's fair to assume it's significantly over $1 billion. It took quite a lot of work to get it right, and a rather large amount of technological infrastructure is needed to make it work out. The $1.5-2 billion range is probably more reasonable, which will make it very difficult to make a return on investment here that was in any way, shape, or form, worth it.
The second question is of course how much reuse actually saves. Probably the most complex question to answer, because it's basically opening up a second supply chain that they have to manage. It's definitely not cheap to do that - and it will probably cut into a lot of the expected savings. I would be surprised if between increased per-unit production costs, reusability costs, and the necessity of creating a new second stage (with multimillion-dollar electronic components) each time, that the savings could be any more than $10 million at best.
How many launches can SpaceX pull off? I think it's fair to say that 15-20 is an optimistic projection, and any more than that is quite fantastic. The "economies of scale" that SpaceX predicts simply do not exist at present. At this point, for almost all launches, SpaceX is competing primarily on price, undercutting their competition by offering the launch at a cost that simply cannot be profitable. I'm simply not sure how much further they can go with that; even at a fairly large price advantage they are losing some important ground. If no more accidents happen, I see 10-12 as a more reasonable settling point for their launch number. Their hopes for launching 20 a year survive only as long as they have enough in their backlog from delays to have that many to launch.
And finally, was this all worth it compared to just making a cheaper launcher in expendable mode? If you're not worrying about recovering the rocket, then you don't have to worry about those reusability R&D costs, those landing legs, the larger rocket meant to accommodate a landing, and so on. How much could you save on that? Probably at least $10 million in this equivalent scenario - and that would be a much easier decision to justify.
Conclusion
The older folk who have worked with rockets often say that the market simply does not justify reusable rockets - I hope I've given some perspective as to why they take that position. Under current conditions, the factors that would have to come together to justify making a reusable launch vehicle are many, and the benefits are not great at the scale of rocket launches that are expected to exist in the present day. Of course, it is inevitable that one day, we will develop a market big enough that reusability simply becomes a necessity - but we are not even close to there yet. A market strategy for reusable rockets is not a good one right now. It should be clear from the model above that it's a decision that is vulnerable to any form of cost overrun or decrease in launch cadence - and it takes a pretty optimistic amount of savings per launch to get to a point where it would make sense to start reusing.
For as long as I can remember, low-cost reusability of rockets has been one of the major dreams of space-inclined individuals, and it is clear why SpaceX's promises of just the same draw so much attention. But unfortunately, we simply don't exist in a market where it makes sense to reuse rockets. It's a topic that sees light every decade or so - and unfortunately is always followed by disappointment.
It is, in fact, common sense that you should reuse your highly expensive technologies instead of just throwing them away - but strange as it may seem, we live in a world where throwing them away makes a whole lot of sense. These fire-breathing monstrosities are but an experimental high-performance craft, that we simply have never learned to use as well as we would hope. For the foreseeable future, it is not unlikely that expendability will reign supreme.
Anyone who read all of this is a champ; it is a monstrosity that is much longer than my previous entries. But that was simply because there is more to this topic than most ever realize - and the reasons why expendable rockets continue to dominate go well beyond some form of mass inability to think critically about the topic. I'm sure that even with this post, there are some who would not be satisfied - so if you have anything more to say on the topic, then by all means!
Happy 4th of July weekend - I can now say that I spent half of it writing about rocket reusability!
