SpaceX Starlink satellites had to make 25,000 collision-avoidance maneuvers | Hacker News Reader

SpaceX Starlink satellites had to make 25,000 collision-avoidance maneuvers

space.com

60 pointsAsymo50 comments

It's important to note that Starlink satellites will maneuver when they calculate a 1-in-100,000 probability of collision.[0] Most other satellite operators will only maneuver when they predict a 1-in-10,000 chance.

The scaling isn't linear either. By setting the threshold 10x lower, SpaceX performs more than 10x as many maneuvers.

Essentially they're complaining because SpaceX is too cautious.

[0] https://www.space.com/starlink-satellite-conjunction-increas...

Another thing is that Starlink is low enough that space junk spontaneously de-orbits relatively quickly. I think within ~10 years, all of the junk SpaceX is maneuvering around shouldn't be a concern to them any more.

As long as other satellites keep being launched into higher but not geosync orbits doesn't that provide a steady supply of junk to navigate around?

> As long as other satellites keep being launched into higher but not geosync orbits doesn't that provide a steady supply of junk to navigate around?

Sort of.

1. Modern satellites don't really shed junk - and almost all satellites today have to be launched with a plan to make sure that they deorbit in a timely manner. Doesn't always work out, but most are pretty safe.

2. For space junk, the time to deorbit increases super-linearly with orbital height. Which means that you are correct - stuff in higher orbits will slowly "fall" through the Starlink orbits on their way to Earth. But it's a very slow process which means there's really only a thin trickle of stuff falling from above.

What that means is that, Starlink will always have to avoid objects. But the bulk of the problem, which was caused by this[1] Russian anti-satellite missile strike will go away after a decade or so.

---

1. https://www.bbc.com/news/science-environment-59299101

Many of those de-orbit plans are things like "deploy all the solar panels in the maximal-drag configuration and wait for the thing to burn." That still poses risk to satellites in very low orbits, although they linger for weeks or months rather than years. I wouldn't expect the situation to get a lot better, since despite the reduced time to de-orbit, a lot more satellites are being launched.

If there is no collision, yes you're right, but if there is one.... Stuff gets thrown into various orbits. Most of the debris will deorbit quickly, but a small percentage can get a substantial kick and go higher.

> If there is no collision, yes you're right, but if there is one.... Stuff gets thrown into various orbits. Most of the debris will deorbit quickly, but a small percentage can get a substantial kick and go higher.

That's a misleading description.

While a collision may increase an object's apogee, it will be in an elliptical orbit: it will always intersect the orbit where the collision happened.

This isn't just a collision thing either: all spacecraft do this[1] as a first step (move to an elliptical orbit), and then typically relight their engines at the apogee in order to circularize their orbit.

---

1. https://en.wikipedia.org/wiki/Hohmann_transfer_orbit

Are you sure? My KSP intuition tells me this is wrong.

Orbits are circular and constant (accurate enough for my argument). An impact might cause some debris to go up, but when the debris that went up comes back around to the same spot in its orbit, it will again be going up. Well, extrapolate that backward a bit and you realize the debris that is going up was inside the atmosphere moments before.

In other words, if an instant impulse causes an orbit to have a higher apoapsis, it must also lower its periapsis. Which means the piece will go deeper into the atmosphere, and you don't come back from deep in the atmosphere.

> Stuff gets thrown into various orbits. Most of the debris will deorbit quickly, but a small percentage can get a substantial kick and go higher.

Only at apoapsis. But, any item thrown into an orbit with a significantly higher apoapsis will very likely have a significantly lower periapsis. So even for high-energy fragments, each orbit should have a period of passing through a higher-drag part of its orbit, and lose energy relatively quickly.

Significantly, the periapsis can never be lower than the altitude that the collision occurred. The worst case, IIRC, is for fragments ejected in the direction of orbital travel at the time of collision. But even then, they'll still spend some time in whatever drag environment they were in beforehand.

That's not to say collisions are fine, but nothing is going to get thrown into random high low-eccentricity orbits where they hang around and pose a threat for decades or more.

10 years for the satellites. Any debris coming off them will tend to be smaller, have a lower ballistic coefficient, and deorbit faster.

If you have 4000 satellites a 1-in-10,000 chance doesn't seem that unlikely anymore. The scale of their operation forces them to be more cautious

Not just cautious—responsible.

When you've got that many, you've got to expect to lose more than a few. Right?

Yes, but when it fails for whatever reason you want it to stay in one piece and just naturally deorbit within ~5 years due to drag. A collision is likely to create more fragments, that in turn have a chance of hitting your other satellites, which would create even more fragments. You don't want that to happen too frequently.

There's enough atmosphere in those low orbits that chain reactions aren't a huge concern for the rest of human spaceflight, but they are a concern for keeping the Starlink constellation alive.

Not really though as the risk is still per satellite. They have a greater chance of an individual loss but their individual risk per satellite is the same (or maybe higher because the collision could be with another starlink).

So it seems like the 1:10k vs 1:100k threshold would be a function of the cost of a loss from a particular incident rather than the risk of a single incident among the 4k satellites.

Thinking from an insurance standpoint, the cost to insure probably depends on the number of satellites overall and total potential loss, not just insuring for a single loss.

Okay so the data they offer is that the satellites are, on average, making one orbital correction per month to avoid any possible trajectory with a greater than 1 in 100,000 chance of a collision. Those satellites are around LEO, which corresponds to a mean velocity of ~17,000 mph. So they're claiming that making one course adjustment per month, traveling at 17,000 mph, to avoid a 1 in 100,000 chance of a collision, is "Like swerving on a highway every 10 meters" in big bold font? I remember there's a reason I removed space.com from my space bookmarks.

Thank you for putting this into perspective! It's a shame science journalism has become so sensationalist.

To compare: I find it interesting that, according to Britannica, the odds of someone dying in the United States from a lightning strike is about 1 in 15,300... over the course of one's entire lifetime.

Not exactly. The quote about highway collisions referred to a time in the future where there are substantially more satellites in orbit. If it’s true that these maneuvers are increasing exponentially, that could be a genuine problem.

It still won’t be accurate as every 10 meters is extremely common. The amount of junk can increase 100Mx and their analogy will still be stupid and make the author’s physics (or maybe stats) teacher feel like a little bit of a failure.

The language in this article seems to be making more of this than it is…

For example, “forced to swerve” is very different from “adjust course to avoid collision.”

There are 4,000 starlink satellites with an eventual plan for 12-40k [0] so it seems reasonable to need to maneuver to avoid crashes. Satellites weren’t “forced to swerve” 25,000 times, they just changed their course.

It would be like me describing my morning commute as “swerving 600 times” because I had to adjust around every object in my way during my 14 mile drive.

There’s traffic in space.

[0] https://en.wikipedia.org/wiki/Starlink

Yes I was going say, the framing seems weird. Isn’t it great the didn’t hit anything? Also that we know this accurately where all the debris is and can avoid this issues is incredible engineering.

Yes. The count also does not seem to account for the steadily increasing number of Starlink satellites (although that is increasing linearly but avoidance maneuvers are increasing exponentially).

I also wonder at what altitude are most of the maneuvers. Are they during the phase between achieving orbit and boosting to their 550km final orbit altitude, or at the final altitude? I have a vague recollection of reading that the 550km altitude was selected partly due to low traffic, and is SpaceX' traffic itself causing the maneuvers?

Also, the maneuvers are based on orbital data several days in advance. How close are we to the point where multiple satellites maneuver simultaneously, but without accounting for the other satellite's maneuver, and wind up causing a crash?

If the were no lanes and everyone was just driving in a huge field with no brakes, your analogy would make more sense. When there's only a few cars nobody would have to touch the steering wheel, but as the number of cars increases it gets more complicated, both per car and in aggregate.

Satellites do travel in lanes (orbits) and do have brakes (maneuvering thrusters), so the analogy is pretty spot on. Changing your speed (the magnitude of your velocity) to avoid a collision is also a collision avoidance maneuver, functionally equivalent to changing the direction of your velocity.

> There’s traffic in space.

Is there? I mean obviously there are other satellites, but typically the framing has been that space and LEO is so big that it doesn't matter. That even with all the other satellites you typically don't need any sort of course correction except in very rare circumstances.

Isn't the better analogy 'I used to never need to do any swerving at all on my morning commute, but now I need to do it 600 times'?

There's a space object and debris tracking visualization at http://astria.tacc.utexas.edu/AstriaGraph/

You can filter to "Low Earth Orbit" and play at 100x speed to get an idea of how things are moving around. There's a lot of intersecting paths, a lot of them Starlink satellite trains.

In those visualization aren't those orange spheres kilometers wide? I feel like we're still missing just how big space and the gap between the tiny objects is and that visualization blows it all out of proportion.

These visualizations are great to give you a basic sense of whats going on up there, but they do not represent any sort of realistic scale. If they did then the visualization would be boring because everything would be WAY too small.

Also it is stuck with a relatively 2D format. Are the satellites really intersecting or is there 10km of vertical distance between them? Its hard to tell with these forms of visualization.

>it [# of maneuvers] follows an exponential curve... it has gone up by a factor of 10 in just two years.

Ridiculous. Per this chart, the # of satellites has also gone up by a factor of 10 in that same 2 year time period from ~400 to ~4000. Maneuvers-per-satellite does not appear to be increasing, given available data.

https://en.wikipedia.org/wiki/List_of_Starlink_and_Starshiel...

Additionally, we have gotten better at tracking smaller objects in orbit as well, so we know of more things we need to maneuver around. Companies like ExoAnalytics have set up hundreds of sensors for finding and tracking objects in space. JSpOC has been increasing sensors has well. As we are able to track smaller things the number of things we are actually tracking can go up. If I remember correctly after the Space Fence came online in 2020 or 2021 the number of things being tracked by JSpOC went from like 25,000 to >100,000.

The article is hurting itself with it's slant.

But it is a dilemma. Having both so many more satellites than anyone else (SpaceX semi recently crossed the line to having more than half the satellites in orbit) and by having such a lower risk tolerance, I have to wonder if it is a phase shift for LEO.

It feels like orbital tracking models have, for a while, been fairly static. We know where everything is & where it's headed, & only have to make a couple of updates. Operators jobs are relatively simple in this world.

But as the number of satellites keeps increasing, and as they become more likely to manuver, it feels like we shift into a new phase, where we have to have much more dynamic planning. The article says we are up to 69 avoidances a day, 69 times where every other operators now need to see, am I still ok and for how long?

We're probably not there yet, but this idea of space going from mostly static to dynamic would be a huge shift. Also note, SpaceX has lightweight low cost craft with cheap argon thrusters & semi disposable craft; most operators have bigger heavier craft with more limited & expensive maneuvering capabilities: structurally SpaceX is set up to survive a chaotic system, and others will have more expense/difficulty if orbital planning becomes increasingly dynamic.

>by having such a lower risk tolerance

What do you mean by this? The article says that NASA makes an avoidance maneuver when the chance of crash is over 1 : 10 000, SpaceX does it when it's over 1 : 100 000.

They seem to mean exactly what you quoted - a lower risk tolerance means they tolerate less risk, meaning they are more cautious. Doing an avoidance maneuver if the risk exceeds 1/100K instead of 1/10K is being more cautious because 1/100K is a lower threshold than 1/10K.

Ah, you are right of course.

> There is a concern about the conjunctions that are occurring where no maneuvers are being made," said Lewis. "You could argue that the probability [of a collision in these cases] is very low, but given the very large number of them, they represent a quite substantial risk. It's like buying a ticket in a lottery. If you buy just one, you are unlikely to win, but if you buy a million tickets, you stand a pretty good chance."

I didn't see anything in the article that implied what sort of odds we're looking at, nor what the consequences of individual collisions would be. Without anything like that, it's hard to understand why this is a bad or meaningfully risky thing. I understand Kessler syndrome is a problem, but is there really nothing less than that that can cause serious issues? The article doesn't really touch on any specific dangers.

This article is sensationalizing pretty hard. SpaceX is doing a great job automating conjunction planning. If you run a collision model on their 1000s of satellites over a several day period, you'll find many potential collisions ("conjunctions"). But remember the probability on these is abysmally low - less than 1/100,000 or less than 10^-6. And these are conservative upper bounds. But thousands and thousands of those add up to an uncomfortably high probability over years of operation.

So SpaceX does a smart, responsible thing here. They make a course adjustment if the probability is above a small threshold (probably 10^-5). From my understanding this is all automated. It's not a "swerve" or an emergency. Their satellite simply needs to adjust its normal orbit-keeping thrust a small amount at some point in the next 12-24 hours. It's table-stakes for constellation management at this point. Planet does the same tracking with their 100-200 Dove satellites.

Additionally, SpaceX and other operators like Planet have higher fidelity tracking data than the USSF's radar data (available at https://www.space-track.org with a free account or through https://celestrak.org). They publish their tracking data for other operators (https://ephemerides.planet-labs.com). Having lower error-bounds reduces the number of potential collisions that need to be avoided and tracked.

Inflammatory language like this is frustrating because it feeds an emotional "have you heard about Kessler Syndrome!?!" gut response. Many people and VCs have been duped into spending time on exotic stuff like harpoon capture and trash collecting satellites. But the answer, to anyone who actually works on this problem and does any research, is already there. What SpaceX and Planet are doing is responsible. Debris mitigation agreements and deorbit timeline restrictions from the FCC/other bodies are enough. If someone truly does launch 10-100k satellites, there are great solutions available with defined orbital bands.

If SpaceX had done 0 of these maneuvers it's unlikely anything bad would have happened. But they're doing the right thing here. And Starlink satellites are very low and don't stick around as junk when they're not being used. Basically, the tone here should be "wow SpaceX does a great job with automating collision avoidance", not "satellites constantly swerving in space".

I recently took my first dive into Celestrak which provides coordinates (TLE's) for all sorts of satellites. Fascinating project I want to learn more about how changes in trajectory are executed by different operators and are communicated more broadly.

Here's some of their Starlink data: https://celestrak.org/NORAD/elements/table.php?GROUP=starlin...

More on the TLE format from wikipedia: https://en.wikipedia.org/wiki/Two-line_element_set

There's a nice blog post from Planet about their approach: https://www.planet.com/pulse/keeping-space-clean-responsible...

What's the story here? Is this an unusually-high number of maneuvers that's resulting in excess fuel usage and shortened lifespan? Orphan statistics are worthless. I make dozens of collision-avoidance maneuvers in my half-hour drive to work every morning.

More or less a repeat article https://news.ycombinator.com/item?id=36693879

How do these sats even know objects are coming? The relative velocity could be tens of kilometers per second. Meaning any optical method would direct it way too late.

And I don't think these are big enough to have serious radars on board capable of seeing hundreds of kilometers away. That just leaves ground control I guess?

Am I understanding correctly that they are mostly avoiding other Starlink satellites?

The article says there are 10,000 satellites in orbit, including dead ones. 4,000 of those are Starlink. I'm guessing that since they have similar altitudes, Starlink satellites are more likely to come near other Starlink satellites.

if they are basing avoidance on the known trajectory of other orbital objects, what happens when those object also manoeuvre?

Presumably if the other object is human-controlled, they coordinate (the counter-party also has interest in avoiding collision). If the other object is debris, they coordinate with physics which is highly predictable.

At least a portion of those maneuvers has to be autonomous, it averages to one every 10 minutes.

I imagine a human controlled object wouldn't be placed on a collision course with a Starlink sat, so it shouldn't come to them having to negotiate how to dodge each other.

What a nothing burger

Humans are like satellites in orbit ... inert.

It'll only take something like a starlink satellite punching a hole through the ISS for this to register or change anything ...

c.f. light pollution

That would be difficult since Starlink orbits more than 100km higher than the ISS.

There have been near collisions with Starlink satellites and China’s space station:

https://www.thespacereview.com/article/4338/1

No, that actually makes it possible. If they were lower then it would be difficult (or even impossible). But the chances of a collision are really small given how tiny the relative cross sections of satellite and ISS are compared to the total surface area of the shell the ISS orbits in.

Reply on news.ycombinator.com