Prev
Unit 7 · Lesson 7-2
Atmospheric Drag and Orbital Decay
38m75 XP
Next

Atmospheric Drag and Orbital Decay

The drag equation, ballistic coefficient, decay rates and the drag paradox

11 July 1979, Western Australia
Skylab was America's first space station: 77 tonnes, launched in 1973, abandoned in 1974 after the third crew departed. NASA's plan was to revisit it with the Space Shuttle around 1983 and boost it to a safe altitude. The models said it had until the mid-1980s at 435 km. Comfortable margin. Then solar cycle 21 arrived, and it was one of the most violent on record. Extreme ultraviolet output from the Sun surged, the thermosphere absorbed it, heated, and expanded. Atmospheric density at Skylab's altitude rose by a factor that no model of the era had predicted. Drag climbed. The orbit came down faster, into denser air, which increased the drag further — a runaway that only ends one way. By 1978 it was obvious Skylab would not survive to meet the Shuttle, which was itself running years late. NASA flew a desperate attitude-control campaign to minimise cross-section and bought a few months. It was not enough. On 11 July 1979 Skylab broke up over the Indian Ocean and scattered debris across Western Australia. The Shire of Esperance issued NASA a 400-dollar fine for littering. It went unpaid for thirty years, until a radio audience in California raised the money. The cause was an atmosphere so thin that at Skylab's altitude the mean free path between molecules was measured in kilometres. A vacuum better than anything achievable in a laboratory on Earth — and it brought down a building.
Watch first · presented by Shoaib Aziz
The numbers in this film are the labs' own: ballistic coefficients of about 50 kg/m² for a cubesat, 100 for the station, 200 for a spent rocket body and 1 for a drag sail; density falls tenfold every 115 km; at 400 km on a moderate solar day the station sinks about 126 m a day (46 km a year), costing 26 m/s and 3.7 t of propellant a year; from 400 km the drag sail lasts 8.5 days, the cubesat just over a year, the station-like body 2.2 years and the rocket body 4.5, while one satellite lasts 6.4 years at solar minimum and 7 months at solar maximum; dropping from 500 to 200 km loses 1.32 MJ/kg and gains 1.32 MJ/kg of kinetic energy, 7,613 → 7,784 m/s.
Drag is the only perturbation in this unit that is not gravitational, and it behaves completely differently as a result. Gravity accelerates all masses equally; drag does not. It depends on the shape and mass of the specific object:
a⃗drag=−12 ρ vrel2 CDAm v^rel\vec{a}_{\text{drag}} = -\frac{1}{2}\,\rho\,v_{\text{rel}}^2\,\frac{C_D A}{m}\,\hat{v}_{\text{rel}}

Drag acceleration. Note it opposes the velocity relative to the rotating atmosphere, not the inertial velocity — the air is going round with the planet.

💡The ballistic coefficient
Everything about how an object responds to drag collapses into a single quantity: B=mCDAB = \frac{m}{C_D A} High BB means dense and compact: a lot of mass behind a small cross-section, so it punches through and lasts. Low BB means light and spread out: it decays fast. A cubesat, B≈50B \approx 50 kg/m². Decays in months from 400 km. The ISS, B≈100B \approx 100 kg/m² — heavy but with enormous solar arrays. Requires reboosting several times a year. A spent rocket body, B≈200B \approx 200 kg/m². Dense, compact, and it persists for years. A deployed drag sail, B≈1B \approx 1 kg/m². Deliberately terrible, which is the point — it is a deorbit device. This is the first time in the entire course that the spacecraft's mass has mattered. Unit 2 made a point of the fact that mass cancels out of the two-body equation. Under drag it emphatically does not, and two objects released from the identical orbit at the identical moment will re-enter years apart.
The density itself is where the difficulty lives. To a first approximation the upper atmosphere falls off exponentially:
ρ(h)=ρ0exp⁡(−h−h0H)\rho(h) = \rho_0 \exp\left(-\frac{h - h_0}{H}\right)

Exponential atmosphere. The scale height H is 40–60 km in the thermosphere, so density falls by a factor of e roughly every 50 km — and by a factor of ten every 115 km or so.

Figure 7.5
10⁻¹⁵10⁻¹¹10⁻⁷10⁻³10⁰02004006008001000density (kg/m³)altitude (km)ISSIridium debrissea levelFifteen orders of magnitude,which is why the axis islogarithmic.At 400 km the density is4.0e-12 kg/m³ —a vacuum better than anylaboratory can produce.It brought down Skylab.
Atmospheric density from the surface to 1000 km, on a logarithmic scale spanning fifteen orders of magnitude. At 400 km the density is around 10⁻¹² kg/m³ — a hundred trillion times thinner than sea level. And it is still enough to bring down a space station.
⚠Why re-entry predictions are uncertain
The exponential model above is a teaching tool. Real thermospheric density is one of the worst-behaved quantities in aerospace engineering, and predictions of re-entry date routinely miss by 10–20% even days out. The solar cycle. Every 11 years, extreme ultraviolet output varies enormously. Density at 400 km can change by a factor of ten between solar minimum and solar maximum. This is what killed Skylab. Geomagnetic storms. A single large storm can double thermospheric density within hours. In February 2022, SpaceX launched 49 Starlink satellites into a 210 km insertion orbit the day after a modest geomagnetic storm. Density was roughly 50% above forecast, the satellites could not raise their orbits fast enough, and 38 of them re-entered within days. Day and night. The dayside thermosphere is hotter and puffier than the nightside; density at a fixed altitude can differ by a factor of several between local noon and local midnight. Attitude. CDAC_D A depends on which way the object is pointing, and a tumbling object presents a constantly changing cross-section. So when an agency says a re-entry will occur "within a window of ±20 hours", they are not being coy. Twenty hours at 7.7 km/s is half a million kilometres of ground track uncertainty.
For a near-circular orbit the decay rate has a clean closed form. Drag removes energy at a rate ε˙=−12ρv3/B\dot{\varepsilon} = -\tfrac{1}{2}\rho v^3/B, and with ε=−μ/2a\varepsilon = -\mu/2a this becomes:
dadt=−ρμaB=−ρ CDAm μa\frac{da}{dt} = -\frac{\rho\sqrt{\mu a}}{B} = -\rho\,\frac{C_D A}{m}\,\sqrt{\mu a}

Circular decay rate. Because ρ grows exponentially as a falls, this is a runaway: decay accelerates itself, gently for years and then catastrophically in the final days.

💡The drag paradox
Here is a result that sounds wrong every time. Drag is a retarding force. It opposes the motion, does negative work, removes energy. So the satellite should slow down. It speeds up. The mechanism is unavoidable once you write it down. Drag reduces ε\varepsilon, so a=−μ/2εa = -\mu/2\varepsilon falls. Circular speed is v=μ/av = \sqrt{\mu/a}, and a smaller aa means a larger vv. The satellite drops into a lower, faster orbit. Where did the energy go? Into the atmosphere as heat — but the satellite gave up twice as much potential energy as it lost in total, and half of that surplus went into kinetic energy. The bookkeeping: Δε=Δ(KE)+Δ(PE)withΔ(PE)=−2 Δ(KE)\Delta\varepsilon = \Delta(\text{KE}) + \Delta(\text{PE}) \quad\text{with}\quad \Delta(\text{PE}) = -2\,\Delta(\text{KE}) This is the virial theorem for a 1/r1/r potential, and it is not a curiosity of drag. Any energy loss from a bound orbit — atmospheric, tidal, gravitational-wave emission — makes the orbiting body speed up. It is why inspiralling binary neutron stars whirl faster and faster as they radiate energy away, right up to the moment they merge.
Figure 7.6
200300400500-30-20-100altitude (km) — decreasing to the leftspecific energy (km²/s²)7.607.80orbital speed (km/s)direction of decayenergy fallsspeed risesBoth at once, and bothcorrect.ΔPE = 2Δε — the orbitsheds twice as muchpotential energy as itloses overall, and thesurplus becomes speed.
Energy, altitude and speed through a decay. Total energy falls monotonically. Altitude falls with it. Speed *rises*. The satellite is losing energy and gaining velocity simultaneously, and both statements are correct.
Figure 7.7
012345100200300400yearsaltitude (km)B = 25 kg/m²a light cubesat — 0.6 yrB = 100 kg/m²ISS-like — 2.2 yrB = 200 kg/m²spent rocket body — 4.5 yrNearly flat for years, thenalmost vertical. Density risesexponentially as the orbitfalls, so the decay feedsitself.
Altitude against time for three ballistic coefficients from a 400 km start. Each curve is nearly flat for most of its life, then turns down almost vertically. That knee is why "it has years left" and "it re-enters next week" can be separated by a single bad solar season.
Figure 7.8
198020002020100150200yearF10.7 solar fluxSkylab re-entryF10.7 fluxthe EUV proxyρ at 400 kmlog scaleAn eleven-year cycle swingsthermospheric density by anorder of magnitude.No lifetime prediction can bebetter than the solar forecastit rests on — and that is poor.
F10.7 solar radio flux — the standard proxy for extreme ultraviolet output — over five solar cycles, with thermospheric density at 400 km plotted against it. The order-of-magnitude swing is the single largest uncertainty in any lifetime prediction.
📐Worked Example: the ISS reboost budget
ISS at 400 km: m=420,000m = 420{,}000 kg, CDA≈4200C_D A \approx 4200 m², so B=100B = 100 kg/m². Density at 400 km. Roughly 2.8×10−122.8\times10^{-12} kg/m³ at moderate solar activity — but anywhere from 5×10−135\times10^{-13} to 1×10−111\times10^{-11} across the cycle. Decay rate. a=6778a = 6778 km =6.778×106= 6.778\times10^6 m. dadt=−2.8×10−123.986×1014×6.778×106100=−2.8×10−12×5.198×1010100=−1.46×10−3\dfrac{da}{dt} = -\dfrac{2.8\times10^{-12}\sqrt{3.986\times10^{14}\times6.778\times10^6}}{100} = -\dfrac{2.8\times10^{-12}\times5.198\times10^{10}}{100} = -1.46\times10^{-3} m/s. That is 126 metres per day, or roughly 46 km per year. Reboost cost. Raising the orbit 126 m requires Δv≈12v Δaa=12(7669)(126/6.778×106)=0.071\Delta v \approx \tfrac{1}{2}v\,\tfrac{\Delta a}{a} = \tfrac{1}{2}(7669)(126/6.778\times10^6) = 0.071 m/s per day, so about 26 m/s per year. For a 420-tonne station with Isp=300I_{sp} = 300 s, that is roughly 3.7 tonnes of propellant per year, delivered by cargo vehicle, every year, forever. And the variability is the real problem. At solar maximum the decay rate can be four times higher. The ISS does not fly a fixed altitude; it drifts between roughly 400 and 420 km, reboosting opportunistically when a Progress or Cygnus is docked and has propellant to spare.
🛰Drag as a natural debris sink
Drag is the villain of this lesson and also the only reason low Earth orbit has not already been ruined. Every collision, every explosion, every discarded bolt in LEO produces debris. That debris is subject to drag, and drag eventually removes it. Orbital lifetimes are roughly: 200 km: days. 400 km: a few years. 600 km: a decade or two. 800 km: a century or more. 1000 km and above: effectively forever on human timescales. This is why the 25-year rule — the international guideline that satellites should deorbit within 25 years of end of mission, now tightened to 5 years by the FCC for US-licensed spacecraft — is written the way it is. Below about 600 km nature enforces the rule for you. Above it, nothing does. And it is why the 2009 Iridium–Cosmos collision at 789 km was so much worse than it would have been 300 km lower. Some of that debris will still be up there in the twenty-second century. The atmosphere is LEO's only janitor, and it does not work above 600 km.
↓But there is a problem
Two lies repaid. Earth is not a sphere, and space near Earth is not empty. Both of those are Earth's fault, and both fade out with altitude. J₂ weakens as a−7/2a^{-7/2}. Drag is gone entirely above about a thousand kilometres. So a satellite at geostationary altitude — 42,164 km, in a vacuum a billion times harder than the ISS enjoys, with the bulge reduced to a whisper — ought to be safe. Undisturbed. Keplerian at last. It is not. It is being pushed by sunlight, which has no mass and pushes anyway. It is being pulled sideways by the Moon, a quarter of a million kilometres away. It is being pulled by the Sun, at 150 million kilometres. And at that altitude those forces are no longer corrections. Left alone, a geostationary satellite's inclination grows by nearly a degree per year, and within a few decades it is no longer geostationary in any useful sense. The perturbations do not stop. They just change hands.