|
Eager Space | Videos | All Video Text | Support | Community | About |
|---|
https://medium.com/@ToryBrunoULA/the-secrets-of-rocket-design-revealed-e2c7fc89694c
This is the third video in the series looking at ULA PR, looking at an article authored by ULA CEO Tory Bruno titled "The secrets of rocket design revealed".
We're going to be doing something a little different, which is why this talk is titled, "the real secrets of rocket design revealed", because I don't think the picture that he paints is accurate.
But don't worry, there are plenty of visual aids in this one and we'll be looking at those as well.
The article starts out with this paragraph: (read)
I mostly agree with it, though I would talk about markets rather than missions.
What I don't agree with is the assertion that the design of the rocket flows directly from the mission the rocket is intended to do and that efficiency is of high importance.
You really should go read the whole paper - it's linked in the description - but since you've been spending too much time on insta and not enough on homework, I'll summarize what I think the main argument is.
(read)
Before we dig into the paper, we need to do a little work.
We need to understand where rockets come from, and we're going to explore this by looking the birth of two rockets.
We'll start with SpaceX. They were working on the Falcon 1 when a great opportunity fell into their lap.
The shuttle was retiring and NASA needed a way to carry cargo to the international space station, so they were funding the commercial orbital transportation services program to help companies develop that capability, and after that was done they would hire companies to actually carry the cargo.
SpaceX's reaction was "we have to win a contract". The development contract would be for 100s of millions of dollars, and the ongoing cargo flights would be just the thing for a small rocket company with growth aspirations.
SpaceX would need a new rocket to get this contract, a much bigger rocket. Whatever the design, it has to be a rocket that not only can win the NASA contract, it needs to be a design that will win the contract. So the design comes down to what NASA wants.
NASA wants a high chance of success, and by success I mean "the rocket and capsule get designed, built, tested, and actually fly missions to the space station". This is not a technology development contract, this is to achieve a useful result.
NASA wants it to be timely - they need it ASAP
NASA wants it to be affordable. They aren't terribly price-sensitive because their only other option is to fly cargo with the Russians and that looks bad, but if it can be cheaper than resupply using the space shuttle, that would make them happy.
To sum it up, NASA wants boring. They are looking for the Ford Transit van of rockets, a utilitarian vehicle for carrying stuff around.
Now we get into what are known in engineering as "trades", or tradeoff studies. You look at a problem, come up with a set of options, and compare them to each other.
Rocket design always starts with engines, and typically with the second stage engine.
The Kestrel engine used on the Falcon 1 is an option. It's a pressure fed engine with no pumps, burns liquid oxygen and RP-1 kerosene, and has a specific impulse - or mass efficiency - of 317 and a thrust of 3 tons.
The RL-10 is made by aerojet rocketdyne and has been flying since the 1960s. It's a very efficient expander cycle engine that burns liquid oxygen and liquid hydrogen. It has a specific impulse of 450 and a thrust of 11 tons
And then there's the engine to be named later, an upper stage engine that SpaceX would develop themselves.
The RL-10 looks like an excellent choice - it's well understood, very reliable, and very efficient. It's also very expensive, at something like $10 million per engine, and SpaceX would be constrained by the capability of Aerojet rocketdyne to provide engines. SpaceX also has not flown a stage using liquid hydrogen as a fuel, and that would require more research and development. And the thrust is marginal - you need a big first stage if you choose the RL-10.
So the RL-10 isn't a good choice. The cost is too much, and buying engines isn't the SpaceX way.
Kestrel isn't very exciting technically; pressure fed designs mean thicker and heavier tanks and that's especially a problem for second stages. The specific impulse is low, and the thrust isn't close to sufficient. Kestrel doesn't work.
Which means SpaceX need a new engine for this rocket, and that's troubling because engine design takes a long time and is expensive and that's not going to make NASA excited. They need an engine that can be developed quickly
SpaceX decides to take the Merlin 1C first stage engine from the Falcon 1 and modify it to be the second stage engine for their new rocket. Merlin is a gas generator design that is less efficient than the other alternatives, burns RP-1 kerosene, and has a low specific impulse, though that is mostly because it doesn't have a large vacuum nozzle on it. It has a thrust of 49 tons
This is a really weird choice from a rocket design perspective. It loses efficiency both because it's a gas generator design and because it burns RP-1 kerosene, and it's really big for a second stage engine which means it's kindof heavy.
It's not a very *efficient* choice from that perspective.
But it's a great choice for SpaceX - they already have the Merlin 1C functioning and it will be *relatively* straightforward to modify it into a second stage engine, and it will also be quicker to do that than to try to develop a new engine from scratch. NASA will be happy and that is the main requirement.
It ends up looking like this without the large vacuum nozzle attached. The big nozzle gets the specific impulse up to 342, and the thrust drops a little to 42 tons, but it's still a beast.
For the first stage, there is a similar analysis.
There's the RS-68 from aeroject rocketdyne. It's kindof like the Merlin 1C except that it runs on liquid hydrogen rather than kerosene and that gives it a specific impulse of 365 at sea level. It's a big engine, putting out 306 tons of thrust.
A second option is a new SpaceX engine, much bigger than the Merlin.
And a third option is a cluster of the existing Merlin 1C.
The RS-68 isn't very exciting. Because it uses hydrogen, you need big tanks and those tanks are heavy and expensive to build. And it has the same issue as the RL-10 - it's expensive, probably at about $15 million an engine.
A new engine has the same issues we ran into for the second stage - you don't have the time and money to do it and you're already doing the vacuum version of the Merlin.
So a cluster looks interesting, and it was used quite successfully on both the first and second stages of the Saturn V moon rocket. The downside is that to get the thrust that is needed, SpaceX will need to use 9 of those engines. Nobody has every done 9 clustered engines and that means they will need to be able to build those engines quite cheaply.
Given the constraints of their situation, the cluster of 9 Merlin 1C engines is the only reasonable choice.
In reality, trades are more complicated than this, because trades will look at the whole vehicle together, not the stages by themselves, and changes to a detail in one stage will force changes in the other stage.
To oversimplify, SpaceX chose a Beefy second stage and a "stage early" philosophy because it made sense to build a high-thrust second stage engine. On a launch to geosynchronous transfer orbit, the second stage has a thrust/mass ratio of 0.88, which is quite high for a second stage, but it needs this high ratio because it stages early.
The other thing driving the trades for Falcon 9 was SpaceX's desire to reuse the first stage. They didn't know how to do it yet, but they did know that the slower and lower the rocket staged, the easier reuse would be, so that also argued towards a beefy second stage architecture.
All of their constraints resulted in the architecture of the Falcon 9. You can say that it is optimized towards low earth orbit missions, but it's really just a rocket designed to accomplish the NASA mission simply, easily, and in as boring of a manner as possible.
Or, to put it another way, you design your rocket to serve the customers that you care about within the business constraints that you have.
And that's why Falcon 9 is the way that it is.
Our second story is about United Launch Alliance's Vulcan, and for that we need a bit more context.
ULA had been gifted a virtual monopoly on US Government launches by the Department of defense when they brokered the creation of ULA from the rocket division of Lockheed Martin and Boeing. If you want that story, go watch spies, allies, and enterprise - the strange story of ULA.
With that monopoly, ULA launched payloads using three different rockets. The Atlas V, the Delta IV M+, and the delta IV heavy.
The Atlas V and Delta IV M+ were roughly the same price - around $160 million - but the delta IV Heavy was a mind-blowing $420 million per launch. The government paid those prices because they had no alternative.
That worked for a number of years and ULA made large profits. Then they ran into two problems.
The Atlas V first stage was powered by the highly efficient Russian RD-180 engine. This was initially viewed as a good thing, as it kept Russian rocket folks employed building engines rather than working for less friendly countries, but US relations with Russian deteriorated, and with the annexation of Crimea in 2014 it became clear that relying on the RD-180 was not a long-term solution - congress would simply not allow it.
So the Atlas V was out.
ULA could have switched all their launches over to the Delta IV, but the M+ variant was less capable than the Atlas V, which would have required more Delta IV Heavy missions and higher costs. Congress was already asking about the high cost of those missions, and when Falcon 9 showed up with much lower prices than the ULA launchers, just flying delta IV was not a winning strategy.
ULA needed a new rocket...
ULA had a specific set of goals...
First, it had to be able to fly all the government payloads that Atlas V, Delta IV Medium, and Delta IV Heavy could fly. Those missions are the bread and butter for ULA, and you have to be able to fly them all to get the contract.
Second, it needs to be done reasonable quickly.
Third, it needs to be reasonably price competitive with SpaceX for space force launches.
And finally, it needs to make the space force happy.
Like SpaceX and the Falcon 9, this is all about keeping a specific customer happy.
Once again, we'll start with the second stage.
One of the significant disadvantages that ULA has is that they do not build their own engines, so their engine choices are limited to what they can buy from other companies, companies that are in the business of making money selling expensive engines.
Aerojet rocketdyne bid the RL10, which we already talked about. A great engine but with low thrust and very expensive.
Traditionally, everybody flew the RL10 because there was no other choice. But Blue Origin decided to bid their BE-3U engine. It's also a hydrogen oxygen engine, but a much more powerful one, with a projected thrust of 70 tons. We don't know what the engine would cost.
That "projected" status is important; at the time, Blue had the BE-3 engine that they used on New Shepherd, but the BE-3U is really a very different engine, which made the timing iffy, and ULA needed an engine that could support the advanced scenarios required for government launches.
ULA decided to stick with the engine they knew and understood, the RL10. They would need two for the bigger and more powerful upper stage they needed to fly the difficult missions that Delta IV Heavy flew, and even then their second stage would be quite underpowered.
For the first stage, there were the same bidders, with Blue Origin pitching the BE-4 engine that would power their New Glenn rocket and aerojet rocketdyne pitching the AR1, an upscaled version of the RD-180 engine that flew on the Atlas V. In this case, the BE-4 was under development already and the AR1 was just a proposal. These are the only US engines that would work, and it would take two of each of them to power the first stage.
In this case, ULA chose the BE-4 as the first stage engine.
The design was settled, with two RL10 in the Centaur V second stage and two BE-4 in the first stage of the Vulcan rocket.
But that wasn't quite enough to handle the tough government missions they needed to handle, so the Vulcan centaur can add two, four, or six GEM 63XL solid rocket boosters to provide more liftoff thrust and higher performance.
The low thrust of the RL-10 engines meant that Vulcan would have a weak second stage and therefore it would need a beefy first stage and it would need to stage relatively late. The thrust to mass ratio launching a GTO payload with 6 solid boosters is only 0.29, much lower than what we would see on the Falcon 9 second stage.
Note that this wasn't a choice driven by rocket architectural concerns but one driven by engine availability, and also ULA's experience and comfort using this approach.
And that's two very different rockets with their differences driven very much by rocket engine availability and company history. Neither are what we would call "clean sheet" designs where you get to design the rocket you've always wanted.
Now we can return to the ULA content.
The first graphic we come to is this one, which is entitled "how hard is going to space?" It's a modification of...
This graphic, from the 2022 article entitled "how hard is going to space?
The arrow heading up to the right says "Increasing difficulty and complexity". I think this is a fairly good graphic, though it breaks rule 25: always start your bar charts at zero, and that's very annoying on this chart where you have lots of cases where it looks like you have zero values.
It's not the worst graph I've seen, however, and it expresses the concept pretty well - missions that require a higher energy trajectory or specific capabilities from the second stage add difficulty and complexity over missions that don't require those things.
This new version has the same information as the last one, but it's grouping the launches into two categories, low energy and high energy. The text says (read)
This is a particularly annoying thing to do.
The problem is that the term "energy" already has a use in rocketry, with high energy orbits ones being defined as ones that require more energy to get into, as you can see from this example from ULA's very own "cubesat launch to high energy orbits" document published in April of 2019.
This is about brand differentiation; ULA is redefining "high energy" to mean "difficult things" and - more importantly - things that ULA is uniquely qualified to do. And it's true that the launches on the right require a very capable launch system, but redefining a term that is already in use is not a helpful thing to do.
One of the main points of the paper is that the requirements of those high energy launches are best served by a specific rocket architecture, the one that Vulcan Centaur uses.
This is where the high energy redefinition gets messy. The slide is supposedly about rocket architecture but has a lot of items that are capabilities.
For example, payload fairings are a capability, so they don't belong on an architecture slide. We can get rid of them.
There are other things listed that are also capabilities - burns, thermal management, etc. They may be great to have in a rocket, but we are talking about architecture...
The rocket on the right side is obviously Vulcan Centaur, but it's not clear what the left rocket is. This is pretty obviously a strawman argument, where you argue about a hypothetical rather than your actual opponent, but we can work with it anyway.
This rocket stages early and can do fly home reuse. Note that when ULA says "fly home" it means either a drone ship landing or landing back at the launch site.
Here are the candidates I came up with - it could be the SpaceX Falcons, SpaceX starship, Blue Origin's New Glenn, or Rocket Lab's Neutron.
All of these are by companies that make their own engines and therefore they can choose whatever architecture makes sense for them. They are likely to be fairly close in architecture because propulsive landing means you want to stage fairly low and early.
I'm going to eliminate new glenn and neutron because they aren't yet flying and we therefore don't have much information about them.
And I'm going to eliminate starship because it is not in the same class as Vulcan and because - without refueling - it can't do the commercial GTO and interplanetary missions that the slide says the commercial optimized rocket can.
Which leaves us with Falcon 9 and Falcon Heavy, and wouldn't it have been easier to just draw one of those?
Which means we are really comparing the Falcon 9 and Falcon Heavy to Vulcan Centaur.
At which point the comparison falls apart.
As we saw at the beginning, Falcon 9 and Vulcan have different architectures because of the constraints that they were designed under; it's not at all about the missions that they can do.
Despite having a "commercial optimized" design, Falcon 9 and Falcon Heavy have the contract to fly all of the us government missions that Vulcan Centaur can and has in fact flown 12 National security space launch missions while Vulcan has yet to be certified and therefore has flown none of those government missions.
I'm not saying that Vulcan Centaur is a bad rocket - it appears to be a good rocket that is capable of doing everything ULA needs it to do at a cheaper price than Atlas V or Delta IV, and more good rockets is better.
It's just more than a bit disingenuous to do a comparison slide against a hypothetical rocket that is designed for a different market. You can build a rocket with a weak second stage and put a big hefty booster under it or you can build a stronger second stage and put a weaker booster under it. Both can get you to where you want to go, even for difficult "high energy" missions.
And now we come to my favorite graphic in the paper, the graph that launched 1000 tweets.
ULA wants to differentiate the Vulcan Centaur approach as better because of the high energy architecture, and they have decided to highlight the difference in how they stage.
The graph is pretty good; it's certainly true that Vulcan stages much higher than Falcon 9, and it also stages at a higher velocity, which is something I would have worked into the graph as well. We already know that this is mostly because the Falcon 9 has a big beefy second stage and wants to reuse the first stage, and the Vulcan has a wimpy second stage.
Then somebody comes along and slaps that title on it and you have a PR issue on your hands.
The immediate question is "are they actually asserting that the first stage of Vulcan Centaur will fly all the way to low earth orbit? That's not what the graph says, but it is what the title says.
If we look elsewhere in the text, we find this. I've added the highlights.
(read)
I think there's a good argument that neither of these are true - I'll show you why in a minute - but what primarily and nearly mean is open to interpretation.
And it's really obvious that the text of the paper does not say the first stages actually fly into low earth orbit.
We also have this graphic, and if you look at the parts I censored earlier, you see that the big label says "upper stage transits from LEO" but the detail says the second stage is carried nearly to orbit by the first stage.
So I guess there's some ambiguity there, but I'd still argue that "nearly" is the operational part.
BTW, does anybody think that it's weird that this graphic uses both "second stage" and "upper stage"? Yeah, me too.
This slide has come up a number of times and Tory Bruno has been asked about it.
The response is pretty simple to make; you say that the graphic is purely about the staging altitude and that the title was a mistake, and if you look at the rest of the document our position is more clear. And then, if people ask what "primarily" and "nearly" mean, you can have a nuanced discussion about that, but you've mostly defused the situation. That is what a good PR person would tell you to do.
Instead, Bruno tweets this.
(read)
My response on reading it was two-fold. The term LEO is well established to mean low earth orbit and nobody uses it to mean anything else. I looked at 6 space acronym sites and it means low earth orbit. Search for LEO on the nasa technical reports server and you get the same result. So it's pretty clear that Bruno is factually wrong, and very obviously factually wrong.
My second response was, "Oh, Tory, why didn't you do the really obvious thing and just say the title was wrong and it got missed during review?"
He later added this tweet, which only makes things worse.
(read)
You had an out but you decided to add fuel to the fire and very obviously didn't provide a reference to support your assertion.
To those of you who will assert that ULA said their boosters flew to low earth orbit, I'm sorry to say that you are wrong. But that "nearly" is something we can look at.
We can build ourselves a little model
We're going to build ourselves a very simple model using the rocket equation that can give us an estimate of how much work the second stage needs to do, and we can therefore estimate how much work the first stage does and compare that to what it takes to get into low earth orbit.
You can find more details in my video, the care and feeding of the rocket equation.
Based on the efficiency of the engine and the ratio of the mass of the stage full of propellant and the mass of the stage empty of propellant, we can figure out what the stage can do.
ULA has been nice enough to tell us how much propellant the centaur V carries and how much its empty weight is.
We're going to be trying to get a payload all the way to geostationary orbit, because that is the highest energy orbit we've been talking about. We'll be flying Vulcan with two boosters in the VC2S configuration, and ULA tells us that its payload to that orbit is 2.5 tons.
Plug some numbers in - the specific impulse of the RL10 is 450 - and we get a delta v of 9030 meters per second. Which is really great. We can subtract that number from the total required, and we find that the first stage needs to supply about 4130 meters per second of delta v.
This is of course an approximate number - I'm ignoring a number of factors - but it's close enough.
Using that, I've created a chart showing the relative delta v contribution of the first and second stages on a mission.
It takes about 9300 meters per second to get into low earth orbit, so I'll put a line on the graph at that point.
Once again, we're going all the way to geostationary orbit, requiring about 13,000 meters per second of delta v.
Here's the data for Vulcan, along with the payloads they carry. None of the first stages generate more than about 5500 meters per second of delta v, so they clearly don't take the second stage "nearly" to low earth orbit. Note that as we add more payload, the second stage can do less work and therefore the first stage adds solid rocket boosters so it can do more work.
Adding data for Falcon 9 and Falcon Heavy, along with some speculative payloads since SpaceX doesn't publish them.
Given what we know about the architectural choices that were driven by the engine choices, this isn't surprising at all. Falcon 9 has the beefier upper stage and therefore that stage does more work than the centaur upper stage in most cases, though the VC2S Vulcan configuration is pretty close in the split between the two stages as it has the least powerful first stage.
It's also interesting that the 6 booster Vulcan and falcon heavy are pretty close in terms of the ratio of work done between the two stages on this mission.
So does the Vulcan Centaur fly nearly to low earth orbit? Not in my book, no.
Pop quiz time.
Here are the time versus speed graphs for two different rockets launching into low earth orbit, from Declan Murphy's excellent FlightClub.io website.
Vulcan Centaur is absent today, so Atlas V has agreed to be a substitute, showing up with 5 solid rocket boosters in the 551 configuration.
Falcon 9 graciously agreed to be represented.
Here are the graphs. Which one is which?
The Atlas V is on the left; you can see the solid boosters separate at about 90 seconds and you can see centaur take over at about 270 seconds. The low thrust to mass of the centaur is very obvious here by how flat the green line is - the stage can only accelerate slowly even though the booster took it high and fast.
I guess you can make an argument that the staging point is nearly to LEO since it's about 80% of the way there, but it takes centaur 360 seconds to get into orbit.
The Falcon 9 launch is quite distinctive because of the booster path back to land on the drone ship. The staging is at a much slower speed, but the beefy Falcon 9 second stage gets to orbital velocity about 30 seconds quicker than centaur despite having to provide nearly 4 times the delta v.
Falcon 9 gets into orbit nearly 3 minutes earlier, and that's 3 minutes where the stage didn't have to waste energy on gravity losses, which you can argue makes it a more efficient rocket.
The important point is that both work.
We finally come to the last chart.
The text says the following:
(read)
It then lists three architectures:
A typical three core LEO optimized rocket
A high energy rocket
An extreme LEO optimized "super heavy leo lifter"
These are nicely specific descriptions - there is only one 3 core reusable rocket, and that's falcon heavy. High energy is obviously Vulcan, and I'm not sure, but the super heavy leo lifter might be starship.
Let's start looking at what it says for Falcon Heavy and Vulcan.
This dotted pink line shows the performance SpaceX gets out of Falcon 9 with drone ship landing, and this solid one is Falcon 9 in expendable mode. It's pretty clear that the gray line is not Falcon Heavy.
Falcon Heavy operates in a number of modes. In fully expendable mode, SpaceX numbers put it roughly here. I made up the MTO number as there are different MTO orbits and SpaceX doesn't talk about them, but it will be roughly the same as GTO.
There are no reusable numbers from SpaceX. If we assume that the LEO payload drops by 25% with drone ship landing - the same as F9 - then we end up with about 48 tons for landing the two boosters on the drone ships. GTO is likely about the same as Vulcan, and we know that GEO is at least 4 tons but is less than 6.7 tons, so I put it roughly where Vulcan is, but it might be a bit less.
I do want to make one clarification. The payload adapter for Falcon is designed to support up to 18.8 tons of payload, so it cannot carry the full payload for expendable Falcon 9 or any of the Falcon Heavy variants. To do so would require a new payload adapter and possibly a strengthened second stage. SpaceX would be happy to do this work if a customer wanted to pay for it, but there aren't any low earth orbit payloads of that size.
That's all for Falcon Heavy. You can draw your own conclusion about whether ULA correctly showed the data for the Falcons.
Let's move onto starship. I did a whole video on this topic titled "starship - what can we do with it?", so I'm going to keep this section short.
Starship is a very different architecture than Vulcan, and frankly to compare it to a rocket like Vulcan is to miss the point.
But it's on the chart.
The booster is designed to only fly back to the launch site, which means it needs to stage as low and close to the launch site as possible.
The second stage is not only huge but it's designed to be reused, so it has flaps and all the mechanisms to control them, heat shield tiles, and an enclosed payload section so it doesn't throw away a payload fairing every flight. All of those are heavy and they push the empty weight up quite a lot, which means it's not a very good second stage for higher energy orbits.
However, a number of people have done calculations and they agree that starship has a GTO payload of 20-some tons. That is probably a one-way trip. Would it have useful payload if it reserved enough propellant to get out of GTO and back to earth? I don't know the answer to that question.
Here's what ULA had to say about this scenario...
That is clearly not true and the calculations to estimate the GTO payload are straightforward.
Assuming starship refueling works - and you are willing to pay for the flights - the graph is quite a bit different; starship can take as much mass as it can get into LEO to any of these destinations.
And there are also options for non-reusable second stages that would be lighter and therefore would have better performance to higher energy destinations.
I talked about reuse at length in part 2 of this series, and frankly it feels like ULA is talking about it only because they have to...
The quick summary has no surprises; the low energy rocket stages lower and that makes flying the first stage home less challenging - I would say "practical" - where staging high means that it's too challenging to do that. And I think that's a fair summary of what's going on...
But this is missing one of the big drivers - all of the groups working towards propulsive landing build their own engines, and this is an enabler to do propulsive landing as you need a cluster of engines - 5, 7, or 9.
ULA buys their engines which both limits their choices and makes the economics very different, so they are stuck with their current architecture.
ULA makes one point in the reuse section that needs to be addressed.
They say that low energy reusable rockets must often fly expendable on the most challenging missions.
That is true, and it's a feature.
A communications satellite launched with a drone ship landing currently costs $70 million. Well, actually, like a car, it's priced at $69.75 million, but I'm going to call it 70.
If you need less performance than that, you might be able to get by with a return to launch site landing, and that will cost less.
If you need more performance, you will need to expend the first stage, and that will cost more.
You can tune the cost of your rocket based on what you need it to do...
In the same document where ULA lists that as a disadvantage, they talk about a Vulcan feature they call dial a rocket. The text describes it this way.
(read)
Four different first stages. That *kindof* reminds me of the way the Falcon 9 has three different first stages depending on how much propellant it has to save for landing.
What are the real secrets of rocket design?
(read)
And those are the real secrets of rocket design.
Please discuss in the comments.
If you enjoyed this video, please send me this KTEL album from 1979, with a list of hits that just don't stop.
I especially appreciate their audacity in putting Foreigner and Captain and Tennille on the same album.