Aerobraking: Trimming an Orbit With an Atmosphere

Why a nudge at periapsis moves the far side of the orbit

An orbit is fixed by its energy and its angular momentum. The specific orbital energy of a Keplerian orbit depends only on the semi-major axis a: ε = −μ/2a, where μ = GM is the planet's gravitational parameter (μ = 4.283 × 10⁴ km³/s² for Mars). Speed anywhere on that orbit follows from the vis-viva equation, v² = μ(2/r − 1/a). Aerobraking is nothing more than a way of making ε less every time the spacecraft comes around.

The geometry does the clever part. Drag is proportional to ρv², and both ρ and v peak sharply at periapsis, so essentially all the braking happens in a few minutes at the low point. A retrograde impulse Δv delivered at periapsis changes the energy by v_pΔv, and since dε = (μ/2a²)da, the semi-major axis changes by Δa = (2a²/μ)v_pΔv. Periapsis radius is nearly untouched — you cannot lower the point you are standing on — so the whole change appears on the opposite side: Δr_apo = 2Δa.

Plug in MRO's early aerobraking orbit (a ≈ 25,800 km, with periapsis already walked down from its 426 km capture altitude to ~100 km, so r_p ≈ 3,500 km and v_p ≈ 4.8 km/s) and one metre per second of drag drops apoapsis by about 300 km. That is the whole trick: a bee-sting at the bottom of the orbit swings the top by hundreds of kilometres. The leverage collapses as the orbit shrinks — by the time a ≈ 5,000 km, the same 1 m/s buys only ~9 km — which is exactly why aerobraking campaigns start fast and end in a long, tedious tail of nearly identical passes.

Inside a drag pass: free-molecular flow, dynamic pressure, and heat

At the ~100 km periapsis altitudes used at Mars, density is a few times 10⁻⁸ kg/m³ — some 40 million times thinner than the 1.2 kg/m³ at Earth's surface. The mean free path between molecules is metres to kilometres, far larger than the spacecraft, so the Knudsen number is ≫ 1 and the flow is free-molecular: there is no shock layer, no bow wave, just individual CO₂ molecules striking the vehicle one at a time. Drag coefficients in this regime sit near C_D ≈ 2.0-2.2 for diffusely reflecting surfaces, roughly twice the continuum value.

Two numbers set the limits. Dynamic pressure q = ½ρv² with ρ ≈ 3 × 10⁻⁸ kg/m³ and v ≈ 4.8 km/s gives q ≈ 0.35 N/m² — a third of a pascal, under half the pressure a sheet of office paper exerts as it lies on a desk. Free-molecular heating goes as the cube of speed, q̇ ≈ ½αρv³ with accommodation coefficient α ≈ 0.8-1.0, giving ~1,600 W/m² = 0.16 W/cm². Radiating that away at emissivity 0.8 puts an unshaded solar array at εσT⁴ = q̇, or T ≈ 435 K (about 160 °C) — right up against typical array limits near 175 °C.

The acceleration itself is set by the ballistic coefficient β = m/(C_DA). MRO massed ~1,300 kg during aerobraking with an effective drag area of order 35 m², so β ≈ 18 kg/m² — a fluffy vehicle, which is precisely what you want. Integrating ½ρv²/β through a pass whose density e-folds in about 80 seconds yields Δv of order 1-3 m/s. Total time in atmosphere per orbit: roughly six minutes out of thirty-five hours — an early-campaign orbit is 99.7 % pure coasting.

Corridor control: flying inside a moving target

The upper atmosphere is not a fixed wall. Density at a given altitude on Mars swings by tens of percent orbit to orbit from thermal tides, by factors of several with season and solar activity, and by more than an order of magnitude when a regional or planet-encircling dust storm heats the lower atmosphere and inflates the whole column. With a thermospheric scale height H ≈ 8-12 km, ρ ∝ exp(−h/H) means a mere 1 km of periapsis altitude changes the heating by ~10 %. The spacecraft is flying a knife-edge in an exponential.

So navigators fly a corridor: an upper bound set by array temperature and dynamic pressure, a lower bound set by schedule and by how much orbit is left to remove. Periapsis is walked up or down with small aerobraking manoeuvres (ABMs) fired at apoapsis, where they are cheapest — typically 0.1-1 m/s each, a few dozen over a campaign. Crucially, the periapsis altitude also drifts on its own: Mars's oblateness (equatorial radius 3,396 km against polar 3,376 km) means the ground beneath periapsis rises and falls by ~20 km as the argument of periapsis precesses under J₂, and third-body effects add more.

Two feedback loops close the system. Onboard accelerometers measure the deceleration profile of every pass, from which the density structure is reconstructed within hours; and orbital cameras — Mars Orbiter Camera on MGS, MARCI and the Mars Climate Sounder on MRO — provide daily global dust-storm weather forecasts. The by-product is real science: Gerald Keating's MGS accelerometer team published the first sustained in-situ survey of the Martian thermosphere in Science in 1998 — the Viking landers had sampled it in 1976, but only along two entry tracks and revealed powerful non-migrating thermal tides that imprint the surface topography onto densities at 120-130 km, later confirmed in detail by MAVEN's NGIMS after its 2014 arrival.

The flight record, including the time it nearly went wrong

Walter Hohmann already discussed using a planet's air to brake in 1925, but the first spacecraft to actually do it was Japan's Hiten (MUSES-A), which on 19 March 1991 dipped to 125.5 km over Earth, lost 1.712 m/s, and dropped its apogee by 8,665 km — a clean demonstration of the leverage described above.

Magellan made the first operational use, at Venus. From 25 May to 3 August 1993, after its radar mapping was complete, it used ~730 passes near 140 km to convert a 3.3-hour ellipse into a 94-minute near-circular orbit for gravity-field mapping. Venus punishes mistakes: its nightside upper atmosphere is a cold cryosphere and its dayside is inflated, so density at fixed altitude varies enormously around the orbit.

Mars Global Surveyor is the cautionary tale. One of its solar arrays had failed to latch after deployment — a damper arm had broken — and in October 1997, three weeks into aerobraking, the panel flexed backwards past its design position under dynamic pressure. Controllers halted, then resumed at roughly a third of the planned pressure. A four-month campaign became a seventeen-month one, and the mapping orbit was not reached until March 1999. Mars Odyssey (Oct 2001 - 11 Jan 2002, ~330 passes) and MRO (30 March - 30 August 2006, ~445 passes) ran cleanly. ESA's ExoMars Trace Gas Orbiter flew the most drag passes of any campaign (MGS's stop-start campaign spanned more calendar time): from March 2017 to February 2018, interrupted by the July 2017 solar conjunction, roughly 950 passes took it from a 24-hour orbit to a 400 km circle. MAVEN ran a short campaign in early 2019, trimming apoapsis from ~6,200 to ~4,500 km purely to improve its relay geometry for surface rovers.

The propellant ledger — and the drag paradox

Doing MRO's job with rockets means changing periapsis speed from ~4.8 km/s to ~3.4 km/s: about 1.4 km/s of Δv. The Tsiolkovsky equation with a dry mass near 1,000 kg and monopropellant hydrazine at I_sp ≈ 230 s (exhaust velocity ~2.25 km/s) demands m_prop = m_dry(e^Δv/v_e − 1) ≈ 850 kg. NASA's published saving for MRO is ~600 kg against a 2,180 kg launch mass, and the same order — hundreds of kilograms — applies to TGO. That mass is not merely saved; it is converted, into instruments, into a smaller launch vehicle, or into a mission that would otherwise not close at all.

The payment is in time and risk. Five months of staffed, twice-a-day navigation is expensive, and every pass is an opportunity for a safe-mode entry at the worst moment. Aerobraking also works only where an atmosphere exists, ruling out the Moon, Mercury, and asteroids.

Finally, a genuine physics trap. Drag is a decelerating force, yet a satellite losing energy to drag ends up moving faster. For a near-circular orbit v = √(μ/r) while E = −μ/2r, so removing energy shrinks r and raises v: potential energy falls twice as fast as kinetic energy rises, and the difference is what the atmosphere carries away as heat. This is the orbital drag paradox, and it is why the endgame of an aerobraking campaign involves a spacecraft that is simultaneously being braked and speeding up.

Not aerocapture, and definitely not entry

Three atmospheric manoeuvres are routinely conflated. Aerobraking spreads the job over hundreds of shallow passes at ~0.2 W/cm², requires no heat shield, and takes months. Aerocapture would do it in a single deep pass, converting a hyperbolic arrival directly into a bound orbit, with heating rates one to two orders of magnitude higher, a real aeroshell or drag device, and closed-loop guidance that must hit an entry corridor typically only about a degree wide (a fraction of that at Neptune) — undershoot and you lithobrake, overshoot and you leave the system. Despite decades of study at NASA Langley and JPL for Mars, Titan, Venus, and a Neptune orbiter, aerocapture has never been flown. Atmospheric entry is the third case: you are not staying in orbit at all, and peak heat fluxes run from tens to a couple of hundred W/cm² — Curiosity's aeroshell in 2012 saw roughly a thousand times what an aerobraking array feels.

Two more distinctions matter. Aerobraking is not orbital decay, which is uncontrolled and ends in the ground; aerobraking is terminated deliberately with a periapsis-raise burn once apoapsis reaches target. Nor is it a gravity assist, which trades momentum with a planet and dissipates nothing — aerobraking genuinely destroys orbital energy, turning it into heat.

Open problems

The binding constraint on aerobraking today is not the physics but the forecast. Nobody can predict Martian thermospheric density at 100 km a week ahead to better than a factor of two, because global dust storms remain fundamentally unpredictable and because the tides that dominate the structure are driven by dust loading in the lower atmosphere. Campaigns therefore fly with large margins, which means they take longer than they physically need to.

Three lines of work are open. First, autonomous aerobraking: on-board drag-pass determination and ABM planning, studied at NASA Langley, which would remove the round-the-clock navigation staff that dominates the cost. Second, drag-modulated vehicles that deploy or retract a surface mid-pass to hold heating constant, letting a spacecraft fly deeper and finish faster. Third, and largest, aerocapture: at Neptune, where a chemical capture burn is prohibitive, it could turn a flyby-class launch into an orbiter, and it has been repeatedly flagged as high-value technology in the US planetary Decadal Surveys. Whether the first demonstration flies at Earth, Mars, or Venus is an open programmatic question — the aerodynamics are understood far better than the appetite for a single, unrepeatable pass.

The major operational aerobraking campaigns flown to date

Mission (body)	Campaign dates	Approx. drag passes	Orbit before → after
Magellan (Venus)	25 May - 3 Aug 1993	~730	8,470 × 294 km, 3.3 h → 541 × 197 km, 94 min
Mars Global Surveyor	Sep 1997 - Feb 1999 (two phases)	~890	54,000 × 263 km, 45 h → 378 km circular, 118 min
Mars Odyssey	Oct 2001 - 11 Jan 2002	~330	26,900 × 272 km, 18.6 h → ~400 km circular, 2.0 h
Mars Reconnaissance Orbiter	30 Mar - 30 Aug 2006	~445	45,000 × 426 km, 35 h → 320 × 255 km, 112 min
ExoMars Trace Gas Orbiter	Mar 2017 - Feb 2018	~950	33,000 × 200 km, 24 h → 400 km circular, 2 h