What Happens When a Starlink Satellite Dies

Satellite disposal used to be an afterthought. For a constellation the size of Starlink it cannot be, because the rate matters: at steady state, a fleet of several thousand satellites on a five-year cycle retires roughly two every day, indefinitely.

The controlled case

A satellite that reaches end of life healthy is deorbited on purpose. It reserves propellant for this, so the decision to retire it comes before the tank is empty rather than after.

The satellite fires its thruster against its direction of travel, lowering the low point of its orbit. That drops it into thicker atmosphere, where drag does the rest of the work — each pass through the denser layers robs more energy, the orbit circularises lower and lower, and within weeks to months the satellite re-enters.

SpaceX has stated this process typically completes far faster than the twenty-five-year international guideline for post-mission disposal, which is one of the clearer regulatory advantages of operating at low altitude in the first place.

• Reserve propellant is set aside for disposal rather than used for station-keeping
• The thruster fires retrograde, lowering the orbit's perigee into thicker air
• Atmospheric drag takes over and removes the remaining energy over weeks to months
• The satellite re-enters and breaks up in the upper atmosphere

Design for demise

The design goal is that nothing survives. 'Design for demise' means choosing materials and structures that break up and vaporise during re-entry rather than reaching the ground as debris.

This is harder than it sounds. Dense, high-melting-point components — reaction wheels, certain optics, some structural fittings — are exactly the parts most likely to survive. Meeting a full-demise target constrains material choice throughout the satellite, not just at the surface.

SpaceX has redesigned components specifically for this, and states that current-generation Starlink satellites fully demise. It is worth noting that this is a design intent verified largely by analysis; direct observation of a re-entry is difficult, and independent confirmation is limited.

Full demise is a design requirement, not a happy accident. It shapes which materials can be used anywhere in the satellite.

The uncontrolled case

Not every satellite reaches end of life in working order. Some fail during checkout, some fail years in, some lose the ability to manoeuvre without warning.

This is where the deployment strategy earns its keep. New satellites are released into a low parking orbit a few hundred kilometres up — deliberately below the operational shells — and must climb to their assigned altitude under their own power. A satellite that fails its post-deployment checks never leaves that low orbit, where drag is strong enough to bring it down within months without any intervention.

A satellite that fails later, already at operational altitude, is a slower problem. At 550 km natural decay takes roughly five years. That is long enough to be a real collision-avoidance concern for other operators, and short enough that it is not a permanent addition to the debris population. Other operators are notified and manoeuvre around it.

Collision avoidance while the satellite still works

Disposal is the end of a longer process. Throughout its life, a Starlink satellite receives conjunction warnings — predictions that it will pass close to another catalogued object — and manoeuvres automatically to increase the miss distance.

SpaceX has said this happens autonomously, without a human approving each burn, which is a necessary consequence of the fleet size: at thousands of satellites and tens of thousands of tracked objects, the number of close approaches is far too large to handle manually.

The system depends on the quality of tracking data for everything else in orbit, which is not uniformly good. Small debris in particular is poorly catalogued, and the satellite cannot avoid what nobody has seen.

The open question: atmospheric effects

Here is the part that is genuinely unresolved. A satellite burning up in the upper atmosphere does not disappear — it becomes vapour and fine particulates, deposited in the stratosphere and mesosphere. The dominant products include aluminium oxides from the structure.

At historic re-entry rates this was negligible. At the rates implied by a large constellation with a five-year replacement cycle — plus every other constellation being built — several research groups have raised questions about accumulation of metal particulates at altitudes where nothing removes them quickly, and about possible interactions with stratospheric ozone chemistry.

This is an active area of research rather than a settled finding. The honest position is that the effect is real but poorly quantified, the modelling is immature, and it deserves attention rather than either dismissal or alarm. It is also a genuinely novel problem: nothing has ever put this much mass through controlled re-entry before.

Seeing the turnover in the data

Retirement shows up as a slow drift in orbital data. A satellite being deorbited loses altitude steadily rather than holding a stable shell, and the decay accelerates as it drops into thicker air.

The statistics page on this site charts the current constellation's altitude distribution. The satellites well below the main shells — the group at the low end of that chart — are a mix of newly launched satellites still climbing and old ones on their way down, which is the constellation's replacement cycle visible as a single snapshot.