Starlink Thread (split from SpaceX)

[mikegarrison]

12 minutes ago, Elthy said:

No, you are missunderstanding me (its hard to write that correctly). In this scenario, one sat had to move after 13 years to avoid a collision, the others can remain stationary. After about another 13 years another maneuver would be neccessary, propably by a different sat. If there is an imminent collision, only the affected sat has to move, not the whole constellation.

Um, you are saying there will only be one adjustment every 13 years? That's obviously not correct. They only have had 60 sats up for a few months and they have already had a conflict.

[Elthy]

Yeah, thats why i continued my calculations, read the next paragraph

[mikegarrison]

The problem with big numbers is hard to grasp sometimes.

In the US last year, there were about 0.65 fatalities per 100 million passenger miles driven by cars. That's great, right? Except if commercial airplanes had that fatality rate, there would be over 40,000 people killed worldwide on commercial flights every year.

12000 satellites is a LOT of satellites, and even if the per satellite rate of conflicts is quite low, that's not enough. It needs to be really, really really quite low, or else there will be problems.

[tater]

30 minutes ago, mikegarrison said:

'm not sure you are making the point you think you are making. Even if any individual Starlink satellite only has to adjust orbits every 13 years, that would still be about 1000 adjustments a year for the entire constellation. Three per day. One every 8 hours.

Good thing they have like 10 km/s of dv, and those maneuvers are what, 1 m/s?

The bigger issue is knowing to do the evasions that often, and not missing any

 

[mikegarrison]

Just now, tater said:

Good thing they have like 10 km/s of dv, and those maneuvers are what, 1 m/s?

The bigger issue is knowing to do the evasions that often, and not missing any

That was my point. Having 10 km/s for each sat isn't important if each sat has an avoidance maneuver every 10 years. But if each sat has an avoidance maneuver every 10 years, that means roughly 3 per day for whoever is watching the constellation. They have to watch 12000 of them 24/7 and make sure that the one maneuver that happens on each shift (on average) is done correctly. Without fail. Every time.

If those numbers are correct, it means they are going to have a pretty big job that either humans or maybe computers will have to do. And do it reliably.

[CatastrophicFailure]

1 hour ago, mikegarrison said:

If those numbers are correct, it means they are going to have a pretty big job that either humans or maybe computers will have to do. And do it reliably

Likely a big reason why they chose to go the AI route, with human oversight. It doesn’t have to be perfect, it only has to pare out a small enough number of potential collisions that the human overseers can reasonably verify and actively monitor. 

[Raven Industries]

While we have no info on how effective the Starlink satellites autonomous collision avoidance system is, I think it's not unreasonable to assume that they've at least considered the possibility of a large number of anti-collision maneuvers by the mere fact that such a feature was included. 

[tater]

1 hour ago, mikegarrison said:

That was my point. Having 10 km/s for each sat isn't important if each sat has an avoidance maneuver every 10 years. But if each sat has an avoidance maneuver every 10 years, that means roughly 3 per day for whoever is watching the constellation. They have to watch 12000 of them 24/7 and make sure that the one maneuver that happens on each shift (on average) is done correctly. Without fail. Every time.

If those numbers are correct, it means they are going to have a pretty big job that either humans or maybe computers will have to do. And do it reliably.

I think they said they plan on the constellation doing it itself---then of course we need to look at the failure rate for deciding to do THAT, and we're in the same boat, lol.

[Bosun]

I don't think the average person really thinks about how close we are to stranding ourselves on our planet with a cascading collision event.

[kerbiloid]

16 hours ago, tater said:

yeah, but nuclear weapons platforms...

Iirc, there should be 432 of them or so. Definitely not 12 000.

16 hours ago, Rakaydos said:

The circumference of the earth is over 40,000 km. We'll use this as an approximation of the hoop of our orbital path, which is even longer, because it's not at earth's surface.

Lets assume all 12,000 sats are in the same orbit. They arnt, but lets assume they are, just to show how ridiculus your claim is.

The sats all fit in a city bus-sized fairing. Lets be generous and call each satelite 3m by 1.5m. The solar panel folds out in the short direction about 5 times, so lets call the whole sat 3mx9m 

So we've got 12,000 satelites in 40,000 km. That's 3 satelites every 10 km, or about a sat every 2 miles. Each sat is nowhere near 2 miles long. To bring back the traffic analogy, there's plentry of room to merge with the freeway traffic.

Are you aware of the word inclination?
Please, at last have a look at the wiki table in my calculation post to realize the picture of reality.

14 hours ago, Elthy said:

Lets go with your numbers of one collision in every 80000 orbits in the lower layers. As an orbit takes about 90min thats over 13 years per necessary course correction of a single star link satellite (the one that would have collided).

No. You can't just multiply these values.
It's like you have a new probability test every 90 min.
As in a single test the probability of crash is p, so, the probability of not crash is (1-p).
The probability of not crash in n tests is (1-p)ⁿ.
The probability to crash in n tests is 1-(1-p)ⁿ
So, if take 80 000 orbits and probability of crash, say, =10^-6, you get 1-(1-10^-6)^{80 000} ~=0.077 probability of crash, 1:13.
So, if take 10 000 orbits, you get 1-(1-10^-6)^{10 000} ~=0.01 probability of crash, 1:100.
The dependence is not linear.

(Of course, this is a simplified calculation, as the orbiting is not discrete, so instead of 90 min long test we should integrate formulas ad so on, but it looks enough good for rough estimaton).

14 hours ago, Elthy said:

Actually its way to low since orbital paths arent measured precisely, so they will maneuver even with a low chance of a collision, ESA states over 1/10000 as the limit.

And the greater is their amount, the more often they need to do this. And the more often not one, but several of them, to avoid collisions with having just maneuvered neighbor satellite (whatever the "neighbor" means for satellites on crossing orbits).

13 hours ago, tater said:

Good thing they have like 10 km/s of dv, and those maneuvers are what, 1 m/s?

10⁴ reserved corrections is not that much if there is ~10³ potential objects to be evaded for 10⁴ orbit turns.

Also, at 300 km and having that sail on top they need some fuel to raise orbit from time to time.
Let's recall how long do debris live in similar orbits without engine burns.

[Snip]
"He is cheating in chess!"

All my calculations are available in the post. Maybe there is a logical or calculation error, why not, but I don't see it.
So, please, feel free to find an error in the calculations and untroll their result.

[Snip]

I'm optimistic in this sense, I think this project will be cancelled after 2-3 launches more.

Edited September 4, 2019 by James Kerman
redacted by a moderator

[Wjolcz]

4 minutes ago, kerbiloid said:

I'm optimistic in this sense, I think this project will be cancelled after 2-3 launches more.

Why?

[kerbiloid]

13 minutes ago, Wjolcz said:

Why?

Upd. to say more accurately.
Do you see a Falcon launch every two weeks, every second with Starlinks?
Do you see a Falcon launch with Starlink every two weeks, the every second launch of Falcons?
They need at least 50 Starlink launches next 4 years (see the table)

Edited September 4, 2019 by kerbiloid

[Wjolcz]

12 minutes ago, kerbiloid said:

Do you see a Falcon launch every two weeks, every second with Starlinks?

We've already seen Falcon being launched every two weeks or so, so not impossible.

13 minutes ago, kerbiloid said:

Do you see a Falcon launch with Starlink every two weeks, the every second launch of Falcons?

As I said previously: we've already seen Falcon being launched two times a month. If they can do that with customers then they can do that internally. In fact, it's probably easier to control the whole production, payload mating, etc. process internally than when you are a launch provider for someone else.

16 minutes ago, kerbiloid said:

They need at least 50 Starlink launches next 4 years (see the table)

Their current record of total rocket launches in a year is 19. Not impossible.

Now, I know it's a VERY BIG "if"at the moment, but if they use the Starship they won't need this many Falcon launches to complete the constellation.

[kerbiloid]

10 minutes ago, Wjolcz said:

but if they use the Starship

Currently a 1-engine proto-Starship hopper has hopped once or twice.

[Wjolcz]

22 minutes ago, kerbiloid said:

Currently a 1-engine proto-Starship hopper has hopped once or twice.

7 years ago I would've said the same thing about Grasshopper if only I had any interest in spaceflight.

[kerbiloid]

3 minutes ago, Wjolcz said:

7 years ago I would've said the same thing about Grasshopper if only I had any interest in spaceflight.

I don't say that BFR/Starship won't fly, I just have no idea about it.
But it's definitely not a Starlink lifter at least for a year or two.

Edited September 4, 2019 by kerbiloid

[Xd the great]

2 hours ago, kerbiloid said:

I don't say that BFR/Starship won't fly, I just have no idea about it.
But it's definitely not a Starlink lifter at least for a year or two.

Starlink would probably be one of the first payloads to launch abroad starship, because SpaceX have faith in SpaceX more than others would.

Edited September 4, 2019 by Xd the great

[Rakaydos]

7 hours ago, kerbiloid said:

 

Are you aware of the word inclination?
Please, at last have a look at the wiki table in my calculation post to realize the picture of reality

(...)

All my calculations are available in the post. Maybe there is a logical or calculation error, why not, but I don't see it.
So, please, feel free to find an error in the calculations and untroll their result.

 

[Snip] There's inclinations. They make it EASY to dodge other satelites.

Edited September 4, 2019 by James Kerman
redacted by a moderator

[kerbiloid]

2 hours ago, Rakaydos said:

[Snip] There's inclinations. They make it EASY to dodge other satelites.

Please, feel free to suggest your own calculations instead.

P.S.
Nice quoting, btw.

Edited September 4, 2019 by James Kerman
Redacted by a moderator

[Rakaydos]

9 minutes ago, kerbiloid said:

Please, feel free to suggest your own calculations instead.

P.S.
Nice quoting, btw.

Fine. Lets start with... what are you trying to calculate? (Because that's how nonsensical your "calculation" was, that the entire premise is obfuscated)

Edited September 4, 2019 by Rakaydos

[kerbiloid]

2 minutes ago, Rakaydos said:

Fine. Lets start with... what are you trying to calculate?

Didn't you read my post?
I was calculating chances (or probability if you wish) of a collision between a Starlink sat and either another one Starlink sat, or some space object 20x20 m (a big sat or an orbital ship with antennas, solar panels, possibly an upper stage) per reasonable amount of time (a day, a month, a year, a decade).

The given data from wiki describes the Starlink distribution in space.

[Rakaydos]

Just now, kerbiloid said:

Didn't you read my post?
I was calculating chances (or probability if you wish) of a collision between a Starlink sat and either another one Starlink sat, or some space object 20x20 m (a big sat or an orbital ship with antennas, solar panels, possibly an upper stage) per reasonable amount of time (a day, a month, a year, a decade).

The given data from wiki describes the Starlink distribution in space.

The  probability of one starlink colliding with another starlink satelite (after initial tumble deployment, that doesnt count) is zero. They are all operated by the same operator, are in constant communication with each other, and a synchronized to never occupy the same space at the same time.

The probability of something else colliding with starlink depends on it not being in the space debris database, so spaceX cant automatically evade it, unable to evade on it's own, AND hitting a 3m by 9m by 20cm target in an average of thousands of square KM, not even regarding cubic volume.

All this was already covered in other posts, if you bothered to read them.

[kerbiloid]

Just now, Rakaydos said:

The  probability of one starlink colliding with another starlink satelite (after initial tumble deployment, that doesnt count) is zero.

Believing a gentleman's word, still would prefer some calculatons.
Airplanes are operated by same traffic control but still collide.

2 minutes ago, Rakaydos said:

The probability of something else colliding with starlink depends on it not being in the space debris database, so spaceX cant automatically evade it, unable to evade on it's own, AND hitting a 3m by 9m by 20cm target in an average of thousands of square KM, not even regarding cubic volume.

In my calculations I was presuming Starlink sats having zero size when collide with a larger object.

P.S.
Still seeing no formulas and numbers except 0.

[Rakaydos]

19 minutes ago, kerbiloid said:

Believing a gentleman's word, still would prefer some calculatons.
Airplanes are operated by same traffic control but still collide.

In my calculations I was presuming Starlink sats having zero size when collide with a larger object.

P.S.
Still seeing no formulas and numbers except 0.

Zero is a singularity. Once you plug one in, most calculations return 0.

What's the probability of an object of zero size (your words) colliding with another object? Zero. Done.

As for starlink colliding with another starlink, have you heard the term Negative Feedback Loop?

Edited September 4, 2019 by Rakaydos

[Val]

24 minutes ago, Rakaydos said:

What's the probability of an object of zero size (your words) colliding with another object? Zero. Done.

That's not true. A point size object, can still hit something that has size, and be hit by it, if the path of point intersects the other object.

Mathematically, speaking.