I could probably Google this but how far away from the coast are the drone ships usually? When I see the rocket on the ship, I wonder about how this thing stays upright… obviously given a reasonable swell in the sea.
does the ship go back to port with the rocket upright? If so, what’s the worst weather it can land in?
It varies, but it’s in the ballpark of 150 to 400 miles. It takes a few days to get back.
Yes, it comes back upright. I’m not sure of the upper limit to weather conditions, but it can handle waves of a couple meters height.
Early landings did not have any special equipment to secure the rocket. When possible, crew would come aboard and weld devices to the deck to secure the legs. This wasn’t always possible, and there are examples of stages sliding around in rough conditions. See this video for an example (I’ve linked to the right place, but really watch the full video).
More recent landings, like the latest ISS mission, use a device called the OctoGrabber. Take a look at the white, plate-like device under the stage. It’s remote controlled and slides under the stage, grabbing on and holding it to the deck (probably by sheer mass).
As a Banks/Culture fan, I remain an odd mixture of delighted and appalled at this. Delighted that Culture ship names are being used, and appalled at who is is doing the naming.
It’d be like if Kim Jong-un started naming NK nuclear tests after Terry Pratchett characters…
Weren’t the Musks big apartheid supporters?
In the first part, what is the relevance to Elon himself?
In the second part, what is the relevance to the subject matter of this discussion?
In the third part, it does not appear to be true.
Not that I know of. There’s a lot of bullshit online about mines and yachts and shit, but no, as far as I’m aware there’s no real evidence for anything. Plus he left to avoid army service, and long before Apartheid actually ended.
Fantastic info here Dr. S. Can we convert this info into fuel % numbers, for instance, “in the LEO launch case, X% of a full fuel load is reserved for landings” etc.?
Every mission is different so it’s tough to make specific rules like that, but I’ll give two examples, the CRS-18 and CRS-19 missions. I’m using the fantastic site Flight Club, which uses telemetry data from launch videos to reverse engineer the trajectories, and from there things like propellant mass from known characteristics about the vehicle. They’ve proven themselves to make very good performance predictions.
CRS-18 Simulation (with RTLS)
CRS-19 Simulation (drone ship landing)
In the RTLS case, S1 (stage 1, the booster stage, and the one we’re concerned with here) starts with 405 tons of propellant. At stage separation, it has 63 tons remaining, or about 16%. Note though that due to the rocket equation, that 16% represents a tremendous amount of capability for an otherwise empty stage. It is much more than 16% of performance–in fact we can see that from the “total delta-V expended” chart, which shows S1 at 2984 m/s at stage separation, but then goes on to reach 5519 m/s total. In other words, the first 84% of the fuel produced 2984 m/s of delta-V, and the final 16% produced 2535 m/s of delta V. In fact it’s better than that since the rocket had 13 tons of fuel left at the end. Those last few drops of fuel have a dramatic effect since the stage weighs so little (no payload and almost no fuel).
Now consider CRS-19. At separation, there are only 31 tons of fuel remaining (same initial load, 405 t). That’s under 8%. Going back to the delta-V chart, we see stage separation at 3493 m/s, and a total of 5368 m/s. So a significantly higher speed at separation, but lower total delta-V. The landing only took 1875 m/s (as compared to 2535 m/s). And even this is high–for higher energy missions, they can do a “hot” landing as I mentioned and reduce the requirements even more. But I wanted to pick otherwise similar missions for this comparison.
I appreciate the long, dataful response. I’m curious, what’s the cost-benefit given all of that? What’s a reduced payload cost versus the savings from re-using the first stage? Obviously there’s a cost savings, but I’m wondering what the magnitude might be.
We only know rough numbers–SpaceX is private and so all we know is little fragments of info here and there.
Recently, Musk said the marginal cost of a launch is about $15M. That’s $10M for the upper stage (expended), $1M for 1st stage refurb, and the rest for incidentals.
Previously, it had been said that the first stage is ~75% of the cost of the rocket. That would mean the rocket is ~$40M total, with perhaps another $5M of incidentals, for $45M total.
The $15M didn’t seem to include the amortized 1st stage cost, which depends on the number of flights they get out of it. They’ve shown 5 uses so far, and claim 10, but their ability to land is still not at 100%, and losing a stage from a failed landing is as bad as losing one to wear. 5 seems reasonable, though, which would put the amortized cost at $30M/5=$6M.
So, $21M vs. $45M. They charge $62M for a baseline (reused) flight, which means their profits have gone from $17M to $41M. That’s a pretty good factor, and would be even higher if you subtract other fixed costs (R&D, etc.).
Again, these are very rough numbers since there isn’t a lot of data available, but they’re probably in the ballpark. Just looking at the construction, it’s easy to see that the 75% number for 1st stage fraction is close to the right answer. For the absolute costs, you have to trust Elon and Gwynne, but nothing they’ve said is too unreasonable.
Aside from Starship, SpaceX’s current reusability drive is the payload fairing. It’s a sophisticated carbon fiber structure and costs several million dollars. They parachute down to the ocean and are picked up by ship (they want to catch them with a ship, but that’s proven difficult). They’ve had moderate success so far but probably not quite to where it’s a net win yet. Once they get things dialed in, they’ll probably save another $5Mish per launch.
Another example from Flight Club:
Telstar 18V / APStar 5C
This is a heavy geostationary mission; their heaviest, I think. Right on the edge of being able to land.
Looking at the delta-V chart, they only used 1689 m/s for the landing burns. That’s decently less than the 1875 m/s for CRS-19. Furthermore, they only reserved 23.4 tons of fuel, as compared to 31.2 t for CRS-19. Finally, they landed out at 703 km instead of 343 km. In part this is due to the higher separation velocity, and part due to the timing of the burns. Definitely a “hot” landing.
A good article for those that don’t know too much of the history of the Falcon 9:
Forget Dragon, the Falcon 9 rocket is the secret sauce of SpaceX’s success
Even aside from all the technical achievements, it’s remarkable for any product to go from 0% to 70% market share in about 8 years with highly established players already there.
SpaceX alsolaunched and landed another Starlink mission today, bringing their satellite total to 482. It was the fifth landing of a booster, which almost certainly means it’ll be used on a sixth mission (the current record is five uses for a booster).
Doing some math since one question didn’t get answered: what is the payload hit, really? I’ll use the Telstar 18V launch as a starting point since it is at the limits of F9 reusability and we don’t have to worry too much about them wasting excessive margin.
To start, let’s do a consistency check. The Flight Club link above says that MECO (main engine cut-off) happened at 3642 m/s, the initial prop load was 412.7 t, and prop at MECO was 23.4 t. We know from elsewhere that the S1 dry mass is 22.2 t and we can compute the S2+payload+fairing mass as 116.4 t. That makes a total of 551.3 t.
If we run it through the rocket equation (online calculator here) to calculate the specific impulse (Isp, the efficiency of the rocket), we get 303 s. The Merlin engines are rated to run from 270 to 311 s depending on the ambient pressure, so this is a good match against average performance.
For the second stage, we know that it provided 7418 m/s of delta V. Since the first stage is gone and the fairing got deployed early on, the initial mass is 114.6 t. The Merlin Vacuum engine has an Isp of 348 s. Again going through the calculator, we get a final mass of 13 t. The payload and dry mass of the stage are 11.1 t, so it seems there are about 2 t of remaining propellant (not much considering the starting point).
Now that we can see the Flight Club numbers are sensible, let’s see what happens if there’s no landing. We’ll use 303 s for the first stage Isp and leave 3 t prop at MECO (not nearly enough for a landing, just a little bit of margin so that the engines don’t accidentally explode by running dry). We now get a MECO velocity of 4035 m/s. That’s a pretty nice improvement.
The second stage now only requires 7025 m/s to achieve the same orbit. But now we can’t use the calculator because the algebra is a tad harder. We have to solve:
dV = ve*ln(m0/m1)
7025 m/s = 348s * 9.81 m/s^2 * ln((107.5+p)/(6+p))
2.06 = ln((107.5+p)/(6+p))
(107.5+p)/(6+p) = 7.85
p = 8.8
If I’ve done my math right, then SpaceX could have sent 8.8 t to the same orbit (with the same margins) if they had not reserved an extra ~20 t of landing propellant. That means they lost about 20% of payload capacity against the actual payload of 7 t.
Different orbits, different payload masses, different landing sites, different margins, etc. will of course all change the result, sometimes significantly. And my numbers here are not exact because I don’t have access to SpaceX’s trajectory calculator, but they should be reasonably close.
Thank you! That’s the answer to the question I was asking. Given the margins you discuss further above, that 20% loss is nothing.
Sure thing. And please ask if you have any questions about the math.
One thing I didn’t do is account for the weight of the landing legs or other hardware. Musk has said they “weigh less than a Model S”, which would make them about 2 t. Doing the math again, I get a 9 t payload if the legs are removed (this makes intuitive sense, too: as a very rough rule of thumb, mass on the first stage is worth about 1/10 that on the second stage).
Here’s a prettty good video about the Falcon’s “hoverslam” landing strategy. Was surprised to learn that a single engine on an empty Falcon 9 has too much thrust to simply hover, even at minimum throttle. Not sure how large the discrepancy is - maybe they’re counting on the booster always having some non-zero fuel weight on board at touchdown?
A single Merlin 1D at 40% throttle (the minimum) produces about 35 t of thrust at sea level. The empty stage weighs 22 t. Remaining fuel may make up some of that, but it’s not guaranteed–the Telstar 18V landing only had 1.6 t of remaining fuel.
As the video notes, though, this doesn’t matter–it’s just a matter of timing. And in fact the greater the thrust, the greater the efficiency, essentially because the faster you decelerate, the less time gravity has to act on you, which means you need to expend less fuel cancelling that extra velocity. For some landings, SpaceX fires up three Merlins for a portion of the descent just to get that extra benefit (though for the very final part they go back down to a single engine for finer control).
An amusing bit of math: a Merlin 1D approximately fits into a cylinder 1 m in diameter and 2 m in height. A slug of solid gold of the same dimensions weighs about 30 tons. The Merlin at its minimum thrust would still easily lift the gold cylinder.
Is Falcon Heavy still a thing? I don’t recall hearing about any launched recently.
It’s still a thing. There are several upcoming launches, though only one in 2020. It has been about a year since the last flight.
The plain Falcon 9 is just too capable. The Falcon Heavy made sense in the early days of the F9 1.0, but it’s undergone significant payload upgrades since then, putting it past where the Falcon Heavy would have been. So it’s only appropriate for very heavy or high energy payloads.
