Launching Starlink with Starship

[TheRadicalModerate]

I've been fooling with how SpaceX would use the current Starlink on a Starship, and how it would have to change to be stacked as payload.  Shotwell claims that Starship can handle 400 birds per launch.  I can get about 340.  I figured that I'd share this stuff so you guys can tell me what I did wrong.

How Does F9 Launch 60 Starlinks?

The first attached image is just clipped from the Starlink web page, and gives you an idea of the layout.  Notice the three bronze-colored things, one centered in "front", and two in each corner in the "back".  I'm assuming that these are the hard points on which each Starlink rests, and they form four continuous spines that make up the rigid stack of Starlinks in the F9 fairing.

The second image is just an abstraction of the Starlink, so we can play with the stacking.  The hard points are not to scale, as this was way more measurement than I was willing to do.

The third image shows how Starlinks stack on F9.  Note that corner hard points extend below the bottom of the satellite (it appears that they're launched antenna-side up), so they stack kind of like upside-down cafeteria trays.

Note also that the corner hard points need to be half the height of the center hard point, because there are two stacks of birds in each F9 fairing, and the corners of both stacks form a common column.  That's why there are six hard points per layer of Starlinks, but only four columns.  Note also that this is why one stack is slightly higher than the other.

Each column of hard points is bracketed by pairs of tensioning bars, which are anchored in the platform just above the PAF, and ends with some kind of block at the top of the stack.  Each column is put into compression, so that when the tensioning bars are released, the whole stack springs open, beginning the deployment process.  The bars appear to have a hoop-like stabilizer every 15 birds or so.  You can see the entire assembly come off in some of the deployment videos.

This is what we know already works, because it's what on F9.

Adapting Starlink to the Starship Fairing

The fourth image is how I imagine you port this architecture over to Starship.  With the 8m-diameter fairing, you can put two stacks in the center and stack them about 15.3m high (this seems to be about where the fairing ogive becomes 4.6m in diameter, equal to the F9 cylinder diameter).  Then you can surround them with four additional stacks, which can rise up to about 10m, which is where 2.3m-wide Starlinks should start bumping into the ogive.

Since we know that a stack of 30 takes up the 6.7m height of the cylindrical part of the F9 fairing, we can determine the number of birds in each stack.  They're in the picture, and sum up to 338.

Do Starlinks port over to Starship as-is?  I think the answer is "almost, but not quite".  Some issues:

1) Unlike the F9, where the "front" hard points are all on the outside, this configuration has a center column between the two central stacks.  (Note that I made no attempt to get the spacing right in this picture, and the tensioning systems aren't shown.)  You obviously can't have that central tensioner just pop away like it did in the F9.  That may be OK or not.

Edit: 1a) That center column also now has two stacks hanging off of it, not one, so the full-size center hard points need to be half their height, while the center hard points on the outside stacks still need to be full height.  This may turn out to be a slightly bigger deal than changing the corner points (see #2 below), because the center points seem to be recessed deeper into the chassis, which won't work when you're stacking from two stacks. Any change from what's there currently will also ripple out to the corner columns, which will be spaced differently if you change how far the center hard point extends.

2) Each column created by corner hard points now is shared by three stacks, instead of two.  That means that they have to be 1/3 the height of the center hard points.  That's a real change, but it should be as simple as bolting on a different fitting.

3) I'm not sure how the tensioning of the corner columns works now.  I think it's just two tension bars like everything else, but there's no room for hoop stabilizers--at least not until the central stacks break out of the four surrounding stacks.  Also, like the central column, there's very little room for the tensioning bars to pop off and get out of the way.

4) The lengths of the tensioning rods are much longer than they are on the F9.  It's possible that you want to insert a platform at the "break out" point, where the surrounding stacks stop but the two central ones keep going.  That would allow for two independent tensioning systems, and possibly two different deployment phases, one for the top of the central stack, and one for the lower part with all six stacks.

5) The F9 gets the stacks to disperse by slightly rotating before deployment.  That's not going to be able to work with the chomper.  Instead, I suspect that the chomper will open, the tilt-table will angle the entire stack assembly out, the tilt-table will spin (which might be new functionality!), and then the whole assembly, tensioning platforms and all, will pop out into space.  Only then will the tension bars release and the whole stack start to disperse.

Can You Launch From Boca Chica?

It's entirely possible that I've been too conservative with my clearances, and you really can get 400 birds per launch here.  But that brings up another problem:  That's roughly 104t of payload, which will be fine for a Starship with RaptorVacs, but we now have a trajectory issue.

If Starship is launching from Florida, then there are no downrange population centers at all for a 53.xº inclination.  But there currently isn't a pad in Florida, and the time is fast approaching when you could be putting Starlinks on Starships.  So it seems that there's a plan for launching Starlinks from Boca Chica.

Getting to 53.xº from BC is... interesting.  A straight southern azimuth takes you over the base of the Yucatan Peninsula, which isn't super-populated in Mexico (anything away from the Gulf coast is jungle), but there are heavily settled areas a bit further downrange in Guatemala and Honduras.  I doubt that this trajectory will satisfy the criterion for less than 1:10,000 probability of harm to the public.

But there's a dogleg that might work.  You launch down the Yucatan Channel, then make a pretty serious right turn 400-600km downrange.  Given that we only care about where debris falls, not where the orbital track goes, I think (without having done a simulation), that this should have the effect of making the impact track stay out to sea as it traces an arc around the Yucatan.  By the time it makes landfall in Honduras, I think it's moving too fast to have a high enough probability of hitting anything.

But it's a huge dogleg.  I did a back-of-napkin and came up with about an extra 1500 m/s of delta-v.  That's likely not something that can be handled with 100t+ of payload, but it probably can be with 50-60t.  So maybe the way to go is to start with something like six 8-9m-high stacks that are all the same height.  That gives you somewhere in the neighborhood of 200-250 birds, which is still a huge improvement over F9.  But it's definitely not 400, as Gwynne claimed.

Does this seem like I'm barking up the right tree?

[Zed_Noir]

I've been fooling with how SpaceX would use the current Starlink on a Starship, and how it would have to change to be stacked as payload.  Shotwell claims that Starship can handle 400 birds per launch.
<snip>

IIRC each Starlink flatpack is about 250 kg. So the Starship nominally have a LEO payload of 100 tonnes. Ergo 400 Starlinks per Starship flight.

[su27k]

I've been fooling with how SpaceX would use the current Starlink on a Starship, and how it would have to change to be stacked as payload.

Interesting that you were able to fit 6 satellites per level for Starship, in contrast to 2 per level on Falcon 9, a 3x increase. And SpaceX said their Gen2 satellite will have 3 times the data capability of the current satellite, coincidence? Or could they be planning the Gen2 satellite that has 3 times the area and can fit 2 per level on Starship, similar to Falcon 9?

[born01930]

I wonder how far out you have to go to place a platform for sea-launch and avoid the dog leg? Maybe there are some mothballed oil platforms already there

[wannamoonbase]

I looked up a quote from Gwynne saying that Starship could do 400 Starlink satellite's at a time.  But maybe she was just going by mass and not the actual stacking.

The total volume of the SS cargo area is very impressive to be sure.  But what is going to be usable? 

The size and shape of the door is important, if the access is less than 180 degrees of the vehicle diameter then the usable volume of the cargo bay is reduced.

Also, how will the cargo be tilted and ejected, that device will take up some amount of space and weight.

Regarding Gen2, maybe they'd be larger and have more phased array antennas on board to increase the bandwidth. 

This is all very interesting to watch develop.

[RedLineTrain]

Regarding Gen2, the v-band antennas are also in play.  SpaceX is permitted to introduce those at any time they are ready.

[gongora]

I would not count on any of the numbers for Starship payload capacity, Starlink satellite throughput increases, number of Starlinks that would fit in Starship fairing, etc being all that accurate.  Considering them ballpark numbers that represent the thinking at SpaceX at the time they were said would probably be more useful.

[TheRadicalModerate]

Maybe it needs a new thread then.  Launching Starships from ocean platforms for polar trajectories is really not Starlink specific.

I doubt that you need a Starship-specific way to get access to the polar inclinations for Starlink; the number of satellites is small, so just use F9.  But  SpaceX won't meet their deployment deadlines without Starship being able to access 40º-55º inclinations, and right now there's no slam-dunk easy way to use Starship to do that.

That said, I agree that we'll go off into the launch facilities weeds really quickly if we're not careful, but I also agree with Phil that at least the strategy for how this gets done is on-topic.

My bet's on that new dance craze, the Boca Chica Dog-Leg, but that comes with a hefty reduction in birds to orbit.

[M.E.T.]

I thought Elon said Starship would launch from both Boca Chica and Florida. Problem solved.

[TheRadicalModerate]

I thought Elon said Starship would launch from both Boca Chica and Florida. Problem solved.

But all the active facilities work is in BC.  There's certainly going to be a lag before anything in Florida comes online.

It's going to be easy to launch Starships to orbit as soon as SuperHeavy works.  That makes full-up orbital EDL testing easy.  So why wouldn't you pack each Starship about to undergo an orbital EDL test with Starlinks?  It turns every EDL test into a revenue-generating opportunity.   The only thing holding you back is a trajectory that works out of BC.

[born01930]

Maybe it needs a new thread then.  Launching Starships from ocean platforms for polar trajectories is really not Starlink specific.

My thought was Starlink may be the driver for early sealaunch of Starship. The dog leg penalty seems too much to pay if BC is used for Starlink launches.

[TheRadicalModerate]

Maybe it needs a new thread then.  Launching Starships from ocean platforms for polar trajectories is really not Starlink specific.

My thought was Starlink may be the driver for early sealaunch of Starship. The dog leg penalty seems too much to pay if BC is used for Starlink launches.

I'm not ready to write off the dogleg just yet.  Just because you might only get half the payload capacity that a non-dogleg would provide doesn't mean that you don't have a system that handily reduces your launch cost per bird vs. the F9.  That's especially true when you're about to do a Starship EDL test where the alternative is launching a mass simulator instead of a big heap o' Starlinks.

However, a non-dogleg is certainly better, and that will require some new launch pad, somewhere.  At the risk of going to exactly the kind of off-topic we're trying to avoid, if the goal is to get to 53.8º inclination, then a launch out of shallow water in the Gulf would require being somewhere offshore of western Louisiana.  The alternative is somewhere southeast of BC, but that'll be in water that's 2-3km deep.  Both of those options seem unlikely.

That leaves Florida, where you could either launch from where SpaceX has done some half-hearted construction at LC-39A on a Starship launch mount that's supposed to go just off the ramp between the HIF and the pad, roughly 150m from both.  IMO, if this proceeds any further, both NASA and the DoD will lose their collective minds, because LC-39A will soon be home to not one but two vital national assets:  It's the only place where NASA can launch crews to ISS (at least until Starliner is up and running) and it's where the mobile service tower will be sited for vertical integration DoD missions requiring Falcon Heavy.  (Presumably, both Vulcan Heavy and and Atlas V 551 will also be available, but FHE is going to outperform both of them.)

That said, woods170, who usually has reliable SpaceX sources, says that they're going forward with the LC-39A add-on pad.  This still seems full-blown bat-guano crazy to me, but it has happened before that SpaceX has seen a diamond where I saw only bat guano. 

The other option is obviously an offshore platform in Florida, which makes a lot more sense to me.  It makes much more sense than an offshore pad in the Gulf.

[philw1776]

Fortunately it is shallow well offshore in FL making platform easier and less expensive. 
Maybe the ocean option consideration is why 39A efforts are a standstill.

[freddo411]

Maybe it needs a new thread then.  Launching Starships from ocean platforms for polar trajectories is really not Starlink specific.

My thought was Starlink may be the driver for early sealaunch of Starship. The dog leg penalty seems too much to pay if BC is used for Starlink launches.

I'm not ready to write off the dogleg just yet.  Just because you might only get half the payload capacity that a non-dogleg would provide doesn't mean that you don't have a system that handily reduces your launch cost per bird vs. the F9.  That's especially true when you're about to do a Starship EDL test where the alternative is launching a mass simulator instead of a big heap o' Starlinks.

However, a non-dogleg is certainly better, and that will require some new launch pad, somewhere.  At the risk of going to exactly the kind of off-topic we're trying to avoid, if the goal is to get to 53.8º inclination, then a launch out of shallow water in the Gulf would require being somewhere offshore of western Louisiana.  The alternative is somewhere southeast of BC, but that'll be in water that's 2-3km deep.  Both of those options seem unlikely.

That leaves Florida, where you could either launch from where SpaceX has done some half-hearted construction at LC-39A on a Starship launch mount that's supposed to go just off the ramp between the HIF and the pad, roughly 150m from both.  IMO, if this proceeds any further, both NASA and the DoD will lose their collective minds, because LC-39A will soon be home to not one but two vital national assets:  It's the only place where NASA can launch crews to ISS (at least until Starliner is up and running) and it's where the mobile service tower will be sited for vertical integration DoD missions requiring Falcon Heavy.  (Presumably, both Vulcan Heavy and and Atlas V 551 will also be available, but FHE is going to outperform both of them.)

That said, woods170, who usually has reliable SpaceX sources, says that they're going forward with the LC-39A add-on pad.  This still seems full-blown bat-guano crazy to me, but it has happened before that SpaceX has seen a diamond where I saw only bat guano. 

The other option is obviously an offshore platform in Florida, which makes a lot more sense to me.  It makes much more sense than an offshore pad in the Gulf.

Serious questions:

* Launch from BC, only the second stage will overfly parts of Mexico.   While this is not as good as only overflying ocean, are we sure that this can't be accommodated legally?    Note that the new polar launch corridor from the cape overflies Cuba so there is precedent.

* Why discount a platform "somewhere offshore of western Louisiana"?   Once you consider being offshore, one particular gulf location isn't very different than another one in many respects.   ( Yes, shipping lanes, etc, etc. matter ).

ADMIN:  feel free to relocate this to another thread on off shore platform launch, or alternate SX launch schemes.

[TheRadicalModerate]

Serious questions:

* Launch from BC, only the second stage will overfly parts of Mexico.   While this is not as good as only overflying ocean, are we sure that this can't be accommodated legally?    Note that the new polar launch corridor from the cape overflies Cuba so there is precedent.

The legal authority is the FAA, who has to license the launch.  (The country from which the spacecraft is launched is responsible for it.  The overflown countries have to complain to the host country.)  The criterion is that the risk of a human casualty on the ground, caused by any object with an energy greater than 15 joules, should be less than or equal to 1:10,000.  There are a lot of inputs to this computation:

1) What's the probability of an event that produces ballistic fragments at any point in the flight?  This will vary based on the risk associated with various flight phases.  Just post-launch is very risky.  Under first-stage boost is riskier than second-stage boost--usually.  Max q is riskier than lower-pressure regimes. Staging is much riskier than normal thrust.  This list goes on and on.

2) What's the distribution of sizes of fragments along the impact track?  They mostly get smaller, lighter, and less energetic as speed and altitude increase, but there are some notable exceptions.  Turbopumps from Columbia landed in East Texas as very large fragments, because they have high ballistic coefficients and are designed to operate at high temperatures.

3) Where does the impact track for the system go, and how quickly does it move?  At the start of launch, the impact track moves very slowly.  By the time you're in orbit, it's moving at 7800 m/s.

4) What's the population density through which the debris track moves?

As you can see, figuring this out without a detailed model with a huge amount of data is pretty much impossible, but the "Cuban Dogleg" proposed for putting stuff in polar orbits from Florida is instructive:  Cuba is only about 700 km downrange from the Cape, but the point at which the debris track crosses it is at a very narrow, mountainous part of the island with low population density.

Here's an image of the the possible area under the debris track for a direct 53.8º launch.  The northern line represents the launch azimuth, and the southern one represents the target inclination.  The actual track will be some curve between the two, as the Earth rotates during the launch:



If the track stays close to the northern line (and I haven't figured out how much the earth moves and how the trajectory approaches the inclination as the ascent progresses), but I'd guess that the real track is closer to the northern line than the southern one.  If that's true, then the overflight of populated but rural areas on the Mexican Gulf coast is pretty quick, and the interior of the Mexican Yucatan, the extreme northeastern part of Guatemala, and southern Belize is a giant jungle nature preserve--pretty ideal for minimizing risk of casualties to the public.  But if it's a little bit further south, you're going over heavily populated parts of Guatemala and Honduras.

The thing that it took me a while to understand (with the assistance of various people beating on me in other fora) is that the ground track of the flight path and the ground track of the debris impact path aren't the same thing.  At any given point in time, the debris track is out in front of the flight track, and it can be way out in front of the flight path in the latter parts of the ascent.  So just because Starship crosses the coast very high and fast doesn't mean that its debris track has a low risk at that point, because it's controlled by events that happen further up-range, at slow speeds and lower altitudes.

But this gets really interesting with doglegs, because now the two tracks don't even cover the same ground.  I have not figured this out, but the hope is that you could get the debris track to go down the Yucatan Channel, then wrap around the peninsula, only making landfall in Eastern Honduras.  I've attached an arm-wave of what I hope would be possible but, again, it's just an arm-wave.

I'm pretty sure that Starship can get 50t to 350km x 350km x 53.8º with about 1500 m/s to spare, so that's your dogleg budget.  I'm trying to figure out what that would buy you, but I'm still lost in roughing out Starship trajectories, to say nothing of the odd piece of spherical trigonometry and various laws of cosines.

* Why discount a platform "somewhere offshore of western Louisiana"?   Once you consider being offshore, one particular gulf location isn't very different than another one in many respects.   ( Yes, shipping lanes, etc, etc. matter ).

a) It's way far away from logistical support from Boca Chica.

b) Sea lanes and air routes going into Houston and New Orleans are all over the place, necessitating complicated range keep-out zones.

c) It's useless for any but launching to 53.xº.

Here's arm-wave on the dogleg (the pre-dogleg leg is way too long):

[envy887]

The thing that it took me a while to understand (with the assistance of various people beating on me in other fora) is that the ground track of the flight path and the ground track of the debris impact path aren't the same thing.

I resemble this comment. 

The relation between ground track and IIP trace is really non-intuitive. There are lots of ways to shape the IIP trace, and not all of them are really expensive in terms of performance. This isn't, unfortunately, something that I can help with running the numbers. But there are several users here that are adept in Flight Club and other sim software that should be able to crunch this problem.

[TheRadicalModerate]

OK, here's a somewhat better guess, but still a guess.  See attachment.

TL;DR version:

You can get about 70t of Starlinks (270 birds) to 300km x 300km x 53.8º, launching to azimuth 111º, as close to the Yucatan as you can get without actually letting the early impact points hit it, then doing a dogleg to azimuth 158.4 about 400 km downrange from launch.

Keeping the impact zone away from the Yucatan, the Caymans, and Panama City simultaneously is a little dicey, but likely doable.

Methodology and weasel words:

1) I've cribbed a lot of speed, altitude, and downrange numbers from the canned Flight Club Starship sim.

2) Per that sim, I get about 3090 m/s speed at 400 km downrange.  I used that, the earth rotational boost, and an inertial azimuth of 139º to reach the descending 53.8º inclination to compute the needed post-dogleg delta-v and azimuth.  I have about 70% confidence that I did this right, so if somebody who knows what they're doing has a little time to kill, let me know if I messed it up.

3) Per that computation, it looks like it costs about 1100 m/s extra to do the dogleg.  That's how I came up with the 70t payload.  (Note that the Flight Club sim assumes 380s RaptorVacs on Starship from the git-go, yielding a payload of 150t to a direct 300 x 300 x 26º orbit, so we're at a little less than 50% of max payload here.)

4) The post-dogleg trajectory is a guess, assuming an instantaneous heading change at the dogleg.

5) The impact zone computation is almost certainly wrong, but I'm guessing that I'm within 100 km.  When I tossed in drag with ballistic coefficients between 3 and 100 (which I cribbed from the Columbia accident docs), this is what I got, but I'm too lazy to write actual code for a clean numerical simulation, so this is some incredibly cheesy spreadsheet-smithing.  Also, as with the trajectory, the post-dogleg shape is just me guessing that it's a version of the (guessed) trajectory, suitably stretched downrange.  Guessing²!  If anybody has a real sim, that would be interesting.

Panama City is by far the biggest problem here, but I'm pretty confident that it's doable to move the impact zone east of it, and that looks like pretty high-quality jungle all the way down into Ecuador.  Also, the speed of the impact point will be really moving that far downrange.  Still, it's poor form dropping spacecraft on major cities.

Bottom line is that I'm now pretty sure that this is doable, and being able to launch 4.5x more Starlinks than an F9 out of Boca Chica is well worth it--especially if they're coupled with a Starship that's reliable to orbit but still needing EDL testing after the chomper has tossed the birds overboard.

[Jcc]

Depending on how quickly they get to orbit, and how far down range they are when they achieve orbit, the time period when the instantaneous debris path crosses over populated areas may be short enough to greatly reduce risk. The actual performance of Starship is evolving as they make improvements to Raptor, so I doubt that they have definitive numbers to be able to calculate trajectories with associated risk calculations. Could be wrong, maybe their projections are close enough.

[Bananas_on_Mars]

Do Starlinks port over to Starship as-is?  I think the answer is "almost, but not quite".

IMO what this thread entirely overlooked is that unlike Falcon 9, Starship has Vertical Integration by design.

The whole tension rod stuff isn’t necessarily needed on Starship. And unlike Falcon 9, i think they won’t want to leave anything on orbit other than the satellites.

Also, their current satellite design is centered completely around Falcon 9. There’s no need to carry this over to Starship other than possibly for a short period where Falcon 9 and Starship might deploy Starlink satellites in parallel.

My feeling is we might see automated on orbit assembly with dedicated Starships.

The following is an example how this could work.

For example solarpanels that don’t need to deploy(fold themselves open) could be paired with the rest of the satellite shortly before release.

Sun shades possibly the same.

This could be combined with a last health check, thruster check while in the largest vacuum chamber you can imagine, and then release.

Satellites that fail this test can be brought back and repaired.

[AC in NC]

My feeling is we might see automated on orbit assembly with dedicated Starships.

The following is an example how this could work.

For example solarpanels that don’t need to deploy(fold themselves open) could be paired with the rest of the satellite shortly before release.

Sun shades possibly the same.

This could be combined with a last health check, thruster check while in the largest vacuum chamber you can imagine, and then release.

Satellites that fail this test can be brought back and repaired.

I cannot imagine any scenario in which this makes any sense.  What does it solve that justifies this complexity?