Station-Keeping and End-of-Life Disposal
GEO station-keeping, longitude drift, Δv budgets, graveyard orbits and debris mitigation
A box 75 kilometres wide
A geostationary satellite does not own an orbit. It owns a box.
The box is defined by international agreement and enforced by the ITU: typically ±0.05° in longitude and ±0.05° in latitude. At 42,164 km that works out to about 75 kilometres on each side — which sounds enormous, until you remember the satellite is moving at 3.07 km/s and has neighbours in the adjacent slots.
It must stay in that box for fifteen years.
The box does not move. Everything else does. Lunisolar perturbations tilt its plane at 0.85° per year. The equator is not circular, so the satellite is pulled along it toward one of two gravitational valleys. Solar radiation pressure pumps its eccentricity annually. Left alone for a single year it is out of the box; left alone for five, it is a hazard.
So it fights. Small burns, every couple of weeks, for a decade and a half, consuming propellant that was loaded before launch and can never be replaced.
And on the day the tanks run dry, the satellite is still perfectly functional — every transponder working, power system healthy, ready for another decade of service — and it must be thrown away, on purpose, using the last few kilograms of propellant, because a dead satellite in the geostationary belt is a threat to everyone else there.
That last manoeuvre is the most important one it will ever perform. It is also, now, a legal obligation.
Watch first · presented by Shoaib Aziz
GEO station-keeping splits into two problems with wildly different costs, and the asymmetry is the whole story.
| Task | Cause | Annual Δv | Fraction of the budget |
|---|---|---|---|
| North–south (inclination) | Lunisolar pull tilting the plane at ~0.85°/yr | ~45–51 m/s | ~90% |
| East–west (longitude) | Equatorial ellipticity (J₂₂) pulling toward a stable well | ~2–6 m/s | ~10% |
| Eccentricity control | Solar radiation pressure | usually folded into east–west | small |
💡Cost of north–south versus east–west station-keeping
This is Lesson 6-4 collecting its debt.
North–south means correcting inclination, and inclination correction is a plane change: . With km/s and 0.85° of accumulated tilt per year, that is m/s annually. Plane changes are expensive because they scale with your orbital speed, and geostationary speed is not small.
East–west means correcting longitude drift, which is an energy correction: nudge the semi-major axis slightly and the satellite drifts back. Tiny changes in produce the required drift rates, and the whole year's worth costs a few m/s.
Over fifteen years: about 700 m/s for north–south and perhaps 60 for east–west. That 700 m/s is a very substantial fraction of a GEO satellite's entire propellant load, and it is spent doing nothing except staying still.
This is exactly why electric propulsion took over GEO station-keeping first. At s instead of 300, the same 700 m/s costs roughly a fifth the propellant mass — hundreds of kilograms saved on every satellite, and hundreds of kilograms is a smaller launch vehicle.
Now, here, you see, it takes all the running you can do, to keep in the same place.
Figure 7.12
The east–west problem has a geometry worth knowing about. Earth's equator is not a circle — the term describes an ellipticity of roughly 70 metres between its long and short axes. That tiny distortion creates a potential landscape along the geostationary belt, with two stable valleys and two unstable hills:
| Longitude | Type | Behaviour |
|---|---|---|
| 75.3° E | Stable | Uncontrolled satellites drift here and oscillate about it |
| 104.7° W | Stable | The other valley, over the Pacific |
| 11.5° W | Unstable | A satellite here drifts away in either direction |
| 161.9° E | Unstable | The other hill |
Figure 7.13
⚠The GEO libration points
A satellite abandoned at geostationary altitude does not stay put. It slides down the longitude landscape toward the nearest stable point and then oscillates about it with a period of two to three years and an amplitude of tens of degrees.
Which means uncontrolled objects accumulate at 75.3° E and 104.7° W. Those two longitudes have the highest density of derelict hardware in the geostationary belt, and every satellite that passes through — including working ones drifting between assigned slots — has to be screened against them.
It is a slow-motion traffic problem that nobody designed and nobody can undo, created by a 70-metre deviation in the shape of the equator.
Now the end of life. A satellite in LEO has an easy answer: burn retrograde, lower the perigee into the atmosphere, and let drag finish the job. From 400 km it costs about 100 m/s, and Lesson 7-2 showed the atmosphere will do the rest for free.
From GEO, deorbiting is not an option. Lowering perigee from 42,164 km to the atmosphere costs about 1.5 km/s — as much as the original circularisation burn, propellant that no operator carries.
So the rule is the opposite: go up. The IADC guideline sets the minimum boost:
Minimum graveyard altitude above the geostationary ring. The 235 km clears the operational belt and its station-keeping boxes; the second term accounts for radiation pressure pumping the eccentricity of an uncontrolled satellite back down.
Figure 7.14
🛰Kessler syndrome, 1978
Donald Kessler, a NASA scientist, published a paper in 1978 arguing that above a certain density of objects in orbit, collisions would generate debris faster than the atmosphere removed it. Each collision produces fragments; each fragment is a new projectile; the process runs away.
The endpoint is a debris belt dense enough to make some orbital regions unusable for generations. Not a sudden catastrophe — a slow, irreversible ratchet.
The evidence has been accumulating:
2007. China destroyed its own Fengyun-1C weather satellite in an anti-satellite test at 865 km, creating over 3,000 trackable fragments in one afternoon. At that altitude drag will not clear them for centuries. It remains the single worst debris event in history.
2009. Iridium 33, an operational communications satellite, collided with the derelict Cosmos 2251 at 789 km, closing at 11.7 km/s. Around 2,000 trackable fragments, plus far more too small to track and quite large enough to kill.
Today. Roughly 40,000 tracked objects; an estimated million fragments larger than a centimetre. The ISS performs collision avoidance manoeuvres several times a year. Starlink alone executes tens of thousands of automated avoidance manoeuvres per six-month reporting period.
The mitigation rules — the 25-year rule, now 5 years under FCC licensing for US operators; mandatory passivation; graveyard disposal — exist entirely because of that 1978 paper. Whether they are enough is genuinely unsettled.
Figure 7.15
📐Worked Example: the fifteen-year propellant ledger
A 3000 kg geostationary satellite, s, fifteen-year design life, .
North–south. 45.6 m/s per year × 15 = 684 m/s.
East–west. ~4 m/s per year × 15 = 60 m/s.
Disposal. The IADC minimum is km; operators fly to about 300 km for margin. Raising the orbit by 300 km costs 11 m/s, both burns included.
Total: 755 m/s.
Propellant, chemically. Mass ratio . Working backwards from a 3000 kg dry mass: 878 kg of propellant, nearly a quarter of the launch mass, spent entirely on staying still.
Propellant, electrically. At s: mass ratio , so 148 kg.
A saving of 730 kg — which either becomes 730 kg of extra revenue-generating payload, or drops the satellite into a cheaper launch class. This calculation, repeated across the industry through the 2010s, is why all-electric GEO platforms went from novelty to default in about a decade.
And note the last line of the ledger. Disposal costs 11 m/s out of 755 — 1.5% of the budget. There has never been a credible economic argument for skipping it. Operators who have skipped it did so through failure, not calculation.
↓But there is a problem
So this is where the Keplerian fiction ends.
Every orbit around Earth is a temporary arrangement. The bulge rotates it, the atmosphere consumes it, sunlight pushes it, the Moon tilts it. Staying anywhere is continuous work paid for in a resource that cannot be resupplied, and every satellite ever launched is on a countdown to the day it stops being able to hold its position.
Every orbit in this course has been a prisoner of Earth.
So let us stop trying to stay.
What does it take to leave — not to a higher orbit, not to a graveyard, but genuinely out, past the point where Earth is the thing you orbit at all?
And once you have left, what are you orbiting instead? Because there is no such thing as escaping to infinity in a solar system with a Sun in it. You do not become free. You change owner.
That is Unit 8.