Wooo it's up! New paper alert! I will write a summary thread about this paper tomorrow morning when I'm not quite as mentally exhausted!
"An Orbital House of Cards: Frequent Megaconstellation Close Conjunctions" by Thiele, Heiland, Boley, & Lawler https://arxiv.org/abs/2512.09643
Not recommended for reading right before bed. It's real bad up there in Low Earth Orbit, folks.
This paper started with a plot showing the density of satellites in orbit vs. altitude that Aaron Boley (professor at UBC) made. I knew this was probably bad, but what does 10^(-7) objects per cubic km really even mean when everything is flying around at 7km per second? It doesn't sound very scary.
I re-made the plot in a hand-wavy way assuming circular orbits, and looking at it in terms of 1km close-approaches instead, and it was a lot scarier. So scary, it was time to write a paper!
Two incredibly talented students led the project. We figured out a much less hand-wavy analytical way to calculate close approach rates using real data from public catalogues. And then we also ran n-body simulations to double check. They agree very well! And are really scary!!
In the densest part of LEO (Starlink), there are closer than 1km approaches every 15 minutes. 1km sounds like a lot, but remember everything in LEO is moving at 7km PER SECOND
Starlink themselves report an average of 1 collision avoidance maneuver every 2 minutes between Dec 2024-May 2025. So that's another (terrifying) double check on this calculation. https://www.scribd.com/document/883045105/SpaceX-Gen1-Gen2-Semi-Annual-Report-7-1-25?v=0.698
We also did this calculation for the catalogue of satellites, trackable debris, and rocket bodies from 2018, before megaconstellations, showing just how much less safe megaconstellations like Starlink have made orbit.
The densest part of orbit in 2018 had a closer than 1km approach a little more frequently than once a day. Now it's more frequently than once every 15 minutes.
How do you summarize how unsafe orbit is? This is where I get to tell you about my new favourite forced astronomy acronym, which I spent quite a while thinking about.
We needed a metric. I originally wanted to do something like "Kessler Countdown" or "Kessler Clock" but this isn't a countdown to Kessler Syndrome, it's just showing how bad things are in orbit, and how quickly they could get worse. So, our name for this metric is...
Collision Realization And Significant Harm: the CRASH Clock!
The CRASH Clock uses the current density in altitude bins (averaged over eccentric orbits) of satellites, rocket bodies, and tracked debris, assuming typical cross sections for each type and orbital speeds. This calculation tells us how long to a collision if all orbital maneuvers were to suddenly stop.
The CRASH Clock is currently* at 2.8 days.
In 2018 it was 121 days.
*This is actually for June 2025 because that's when we ran it. Will update soon!
@sundogplanets
One interesting number to estimate is how long before a crash in the less extreme case of worse orbital determinations (because a database is down or someone is dazzling telescopes), which then could lead to fewer maneuvers being ordered or maybe even to dangerous maneuvers being ordered.
@tobychev That's pretty much what the CRASH Clock is. If orbits are poorly known, collision avoidance maneuvers either won't happen or won't happen at the right place, so this is the random measure of how long until a collision.