Orbital mechanics

Orbital mechanics, also called astrodynamics, is the study of how spacecraft move under gravity, and it sets the price of every mission: about 7.8 kilometers per second of speed to hold a low Earth orbit,^[9] 11.2 to escape Earth entirely.^[25] Its rules, Johannes Kepler's laws of planetary motion and Isaac Newton's law of gravitation, are four centuries old and still govern everything from station-keeping a communications satellite to threading a probe past four planets.^[1]

The subject has a reputation for being counterintuitive, and it earns it. In orbit, firing your engine forward eventually slows you down relative to a target ahead of you; the cheapest way to reach an outer planet can begin by flying toward an inner one; and coming home from space requires braking, not climbing down. The underlying reason is that a spacecraft with its engine off is not driving, it is falling, and everything follows from how falling works.

What an orbit is#

An orbit is a fall that keeps missing the ground. Newton explained it with a thought experiment: fire a cannonball horizontally from a high mountain and it curves down to Earth; fire it fast enough, about 7.8 kilometers per second near the surface, and the curve of its fall matches the curvature of the planet, so it circles forever.^[9] The International Space Station does precisely this at roughly 400 kilometers altitude, completing 16 laps a day, one about every 92 minutes.^[1] Nothing holds it up; it is falling around the Earth, which is also why its crew float.

Orbits need not be circles. Any orbit is an ellipse, characterized by its lowest point (perigee, for Earth orbits) and highest point (apogee). Higher orbits are slower: a satellite at 35,786 kilometers moves at just over 3 kilometers per second and takes one sidereal day, 23 hours 56 minutes 4 seconds, to go around, so it hovers over a single spot on the equator. That is the geostationary orbit used by weather and broadcast satellites.^[8][9]

Altitude, speed, and period are not three independent choices. Fix any one and the other two follow, because gravity at a given distance sets exactly how fast a circular orbit has to move. What a mission designer actually chooses is how big the orbit is, how eccentric, and how it is tilted.

Kepler's laws in plain language#

Kepler published his three laws between 1609 and 1619, from analysis of Mars observations, and they still describe every spacecraft trajectory.^[1][13]

First law: orbits are ellipses, with the central body at one focus rather than the center. A spacecraft's distance from Earth therefore rises and falls each revolution unless the orbit is perfectly circular.

Second law: an orbiting body sweeps out equal areas in equal times, which is a geometric way of saying it moves fastest at its lowest point and slowest at its highest. A comet crawls through aphelion and whips through perihelion.

Third law: the square of the orbital period is proportional to the cube of the orbit's average radius. Bigger orbits are not just longer paths, they are slower ones. The progression from the station's 92 minutes, to geostationary satellites' 24 hours, to the Moon's 27 days follows directly from it.

Kepler found the pattern but not its cause. Newton supplied that in 1687, showing that an attraction falling off as the square of distance, combined with his three laws of motion, produces exactly the paths Kepler had measured.^[7][13] The step from description to physics is what makes prediction possible: given a position and a velocity at one instant, the entire future trajectory follows. In Newton's hands the third law also stops being a proportion and becomes an equation. Orbital period equals two pi times the square root of the semi-major axis cubed divided by the central body's gravitational parameter, the product of the gravitational constant and that body's mass. An engineer can read a period straight off an altitude.

Describing an orbit: the six elements#

Six numbers pin an orbit down completely. They are usually called Keplerian elements, after the geometry Kepler established.^[8]

The first two describe the ellipse itself, the next three orient it in space, and the last places the spacecraft on it. In an idealized two-body problem the first five never change at all. Real orbits are not idealized, and every one of those five drifts.

Orbital speed, escape, and the vis-viva equation#

The speed of a circular orbit is the square root of the central body's gravitational parameter divided by the orbital radius, measured from the center of the planet rather than the surface. Earth's mean radius is about 6,371 kilometers, so a 400 kilometer orbit sits at a radius of roughly 6,771.^[25]

Escape velocity is the same calculation with a factor of two under the root, which makes it the circular speed at that radius multiplied by the square root of two. At Earth's surface that comes to 11.2 kilometers per second.^[25] The version that matters operationally is different: a spacecraft already circling at 400 kilometers is doing 7.7 kilometers per second and would need about 10.9 to break free, leaving it only some 3.2 kilometers per second short. That gap is why departing for the Moon or Mars costs far less than reaching orbit did in the first place.

One relation covers every case above. The vis-viva equation gives the speed at any point on any orbit: the square of the speed equals the gravitational parameter multiplied by the quantity two over the current radius minus one over the semi-major axis.^[5] Set the semi-major axis equal to the radius and it collapses to the circular result; let the semi-major axis run to infinity and it returns escape speed. Nearly every transfer calculation below is that one equation applied at two points with the difference taken.

Orbit types#

Spacecraft cluster into a handful of standard orbit types rather than scattering evenly through space. A few orbits solve recurring problems well enough that almost everything flies in one of them.

Sun-synchronous orbits are the clearest case of a perturbation being put to work. Earth's equatorial bulge drags the orbital plane around slowly, and at the right combination of altitude and slightly retrograde inclination that drag rotates the plane by almost exactly one degree a day, matching Earth's progress around the Sun.^[10] The result is a satellite that crosses the equator at the same local solar time on every pass, so images taken months apart share the same shadows. NASA's Aqua flew this way at 705 kilometers, crossing the equator northbound at about 1:30 in the afternoon local time.^[21]

The Molniya orbit solves the opposite problem. Geostationary satellites sit over the equator and appear very low or below the horizon from far northern latitudes, so Soviet engineers used a 12-hour ellipse with apogee high over the northern hemisphere, where the second law makes the satellite loiter for roughly two-thirds of each lap before dropping quickly through a low perigee in the south.^[10] The 63.4 degree inclination is not arbitrary.^[10] It is the critical inclination, the value at which the oblateness term leaves the perigee frozen rather than steadily rotating it around the orbit, so apogee stays parked over the same latitudes for the life of the mission instead of wandering.^[28]

Delta-v: the currency of spaceflight#

Mission designers measure maneuvers not in fuel but in delta-v, the change in velocity a burn produces. Every mission has a delta-v budget, and the rocket equation converts that budget into propellant mass (see how rockets work). Representative costs from Earth:^[5][9]

The launch figure, 9.3 to 9.5 kilometers per second just to reach low Earth orbit,^[5][9] is the one that looks anomalous, since orbital speed at that altitude is only about 7.8 kilometers per second.^[9] The difference is losses: propellant burned holding the vehicle up against gravity before the trajectory bends over, and a smaller amount spent pushing through the lower atmosphere. Everything after launch is cheap by comparison, which is why low Earth orbit has been called halfway to anywhere. The table also shows why landing on Mars costs little more departure energy than parking a television satellite, and why every kilometer per second saved by a clever trajectory translates into tonnes of propellant.

The conversion from delta-v to hardware is unforgiving because it is exponential. Tsiolkovsky's rocket equation sets delta-v equal to the exhaust velocity multiplied by the natural logarithm of the mass ratio, meaning the fueled mass divided by the dry mass.^[7] Chemical exhaust velocities top out near 4.5 kilometers per second for liquid propellants and around 2.5 for solids, so a stage that must supply the 3.9 kilometers per second of a LEO to geostationary transfer needs a mass ratio close to 2.4 even with the best chemistry available, which is to say roughly 58 percent of it is propellant before any payload is added.^[7]

Hohmann transfers and launch windows#

The workhorse maneuver between two circular orbits is the Hohmann transfer, described by Walter Hohmann in 1925: one burn stretches the orbit into an ellipse whose far end touches the destination orbit, and a second burn at arrival circularizes it. It is usually the cheapest two-burn route, at the price of patience.^[14]

Run the numbers for the most common commercial case, LEO to geostationary. A spacecraft in a 300 kilometer circular orbit is moving at about 7.7 kilometers per second; raising apogee to geostationary radius costs roughly 2.4 kilometers per second at perigee. It then coasts for about five and a quarter hours out to apogee, arriving at only 1.6 kilometers per second, well short of the 3.1 needed to stay there, so a second burn of about 1.5 kilometers per second circularizes the orbit. Total cost is close to 3.9 kilometers per second, and the whole ellipse is the geostationary transfer orbit that launch providers quote in their performance tables.^[5]

Between planets, the same geometry imposes schedules. A Hohmann-style transfer to Mars only works if Mars will be at the ellipse's far end when the spacecraft gets there, an alignment of Earth and Mars that recurs roughly every 26 months. Miss the window and the mission waits two years, which is why Mars launches cluster in bursts.^[22] A true Hohmann transfer to Mars coasts for about eight and a half months; NASA classifies the faster and slower variants missions actually fly as Type I, carrying the spacecraft less than 180 degrees around the Sun, and Type II, carrying it 180 degrees or more.^[1] Launch windows matter on smaller scales too: a rocket chasing the International Space Station must lift off within minutes of the moment its launch pad rotates under the station's orbital plane, because changing an orbit's tilt after launch is brutally expensive.

The Hohmann transfer is not always optimal. A bi-elliptic transfer uses three burns instead of two, flinging the spacecraft far beyond its destination on one enormous ellipse, changing the orbit cheaply out there where it is barely moving, then dropping back in. It wins on propellant when the ratio of final to initial orbital radius is above about 11.94, depending on how far out the intermediate apoapsis goes, and always wins above about 15.58.^[23] It also takes far longer, since the spacecraft must traverse two very large ellipses. LEO to geostationary is a ratio of only about 6.3, so the ordinary Hohmann transfer stays the right answer there.

Plane changes and why they cost so much#

Rotating an orbital plane without changing its size is the most expensive routine maneuver in spaceflight. The cost is twice the orbital speed multiplied by the sine of half the angle, which means it scales with how fast the spacecraft is already going.^[5] A 60 degree turn costs a full orbital velocity, about 7.7 kilometers per second in low Earth orbit, which is most of what the launch vehicle spent getting there.^[23]

The smaller numbers are just as instructive. At low Earth orbit speeds, a single degree of inclination change costs roughly 130 meters per second, and the 28.5 degree change needed to go from a due-east launch out of Florida to an equatorial orbit costs about 3.8 kilometers per second on its own, comparable to the entire LEO to geostationary transfer. Because the price tracks speed, the maneuver gets cheaper the slower the spacecraft is moving, so the same 28.5 degrees costs only about 1.5 kilometers per second once the spacecraft has already coasted out to geostationary distance.

In practice designers do better still by refusing to separate the two maneuvers. Instead of circularizing and then turning, the apogee burn does both at once as a single angled push, and the cost is the vector difference rather than the sum: roughly 1.8 kilometers per second instead of nearly 3.0.^[23] The practical consequence is that inclination is cheapest bought at liftoff, so launch azimuth and the latitude of the spaceport, not later maneuvering, are what set a satellite's inclination. Launching eastward near the equator also collects the largest share of Earth's rotation, worth over 1,650 kilometers per hour at the equator.^[24]

Gravity assists#

A gravity assist, or slingshot, lets a spacecraft trade momentum with a planet, and the trade happens in a way that depends entirely on which frame of reference you measure it from. Relative to the planet, the encounter is symmetric: the spacecraft accelerates falling in, decelerates climbing out, and leaves at the same speed it arrived, with only its direction changed. Relative to the Sun, that change of direction is everything, because the planet is itself moving at tens of kilometers per second and the turn adds some fraction of that motion to the spacecraft's own.^[11] The energy comes from the planet's orbit around the Sun, not from the planet's gravity as such, and the planet loses a corresponding but immeasurably small amount of orbital momentum.

NASA's own analogy is a tennis ball thrown at an oncoming train: the ball rebounds off the moving locomotive at roughly its approach speed relative to the train, but relative to the ground it has gained the train's speed twice over.^[11] The spacecraft is the ball, and Jupiter is a very large train.

The technique made the outer solar system reachable. Voyager 2 exploited a planetary alignment that recurs only about every 175 years to hop from Jupiter to Saturn to Uranus to Neptune between 1979 and 1989, with each flyby flinging it onward; gravity assists cut the trip to Neptune from roughly 30 years to 12.^[2] New Horizons used a Jupiter assist on February 28, 2007 that added about 14,000 kilometers per hour and shortened its Pluto flight by three years,^[16] and Cassini reached Saturn via two swings past Venus plus Earth and Jupiter flybys.

Assists work in both directions. Europa Clipper, launched on a Falcon Heavy in October 2024, skimmed 884 kilometers above Mars on March 1, 2025 in a maneuver that actually slowed it by about 2 kilometers per second relative to the Sun in order to reshape its solar orbit.^[3] It is due back past Earth in December 2026, passing about 3,200 kilometers above the surface to pick up speed for an April 2030 arrival at Jupiter and nearly 50 later flybys of Europa.^[15] Without the two assists the 6,000-kilogram spacecraft would have needed propellant it could not carry.^[3]

Lagrange points#

Lagrange points are the five locations in any system of two large bodies where gravity and orbital motion balance, so that a small object keeps station with the pair. The two used most, Sun-Earth L1 and L2, sit about 1.5 million kilometers from Earth, toward the Sun and directly away from it.^[12] All five are prime real estate for observatories, but they are not equally hospitable.

L1, L2, and L3 behave like a ball balanced on a hilltop: a small nudge grows, and anything parked there needs regular correction burns.^[12] L4 and L5 behave like a ball in a bowl, and they are genuinely stable as long as the mass ratio of the two large bodies exceeds about 24.96, a condition both the Sun-Earth and Earth-Moon systems satisfy comfortably.^[12] That is why Trojan asteroids accumulate at L4 and L5 across the solar system while nothing collects at the other three.

The James Webb Space Telescope does not sit at L2 but loops around it in a halo orbit once every 168 days, ranging between 250,000 and 832,000 kilometers from the point itself. Station-keeping maneuvers are scheduled every 21 days and skipped when the computed correction is negligible.^[4] The orbit keeps the observatory in constant sunlight for power, holds its instruments in stable shadow behind the sunshield, and never lets Earth eclipse the Sun.^[4] L2 previously hosted missions such as ESA's Herschel, Planck, and Gaia, while SOHO has watched the Sun from L1 since the 1990s.^[12]

Perturbations: why real orbits drift#

Perturbations are the small forces that push a real orbit away from the ideal two-body ellipse. The clean ellipse is a mathematical convenience; real orbits are shoved around constantly, and mission design is largely a matter of budgeting for that.

The bulge is the biggest single effect close to Earth, and it is the one designers exploit rather than fight, since it is what makes sun-synchronous and Molniya orbits possible at all.^[10] Drag rules the low end and is also the least predictable perturbation, because the density of the upper atmosphere swells and contracts with solar activity: near 400 kilometers a small dead satellite reenters within five years, while above roughly 500 kilometers natural decay can no longer be counted on to clear it on that timescale.^[18]

Higher up the ranking inverts. At geostationary altitude drag is negligible and the Sun and Moon take over, tugging the orbital plane away from the equator, so operators spend propellant on north-south station-keeping for the satellite's entire service life. NASA design studies take about 45 meters per second a year as representative for that correction, with east-west control an order of magnitude cheaper, and a decade or more of it has to be carried as propellant from the start.^[26] The same perturbations set disposal rules: the international guidelines note that a retired satellite boosted above the geostationary belt can still sink back by as much as 35 kilometers under lunisolar and geopotential effects, and they build that margin into the required disposal altitude.^[6]

Rendezvous and docking#

Rendezvous, meeting another spacecraft in orbit, is a chess game with Kepler's laws. Thrusting toward a target ahead of you raises your orbit, and by the third law a higher orbit is slower, so you fall behind: the harder you chase, the more you lose ground. Braking has the mirror effect, dropping the chaser into a lower, faster orbit that gains on the target. Real rendezvous therefore works vertically. The chaser flies a slightly lower, faster orbit to close the gap, then raises its orbit to arrive alongside the target with matched velocity before creeping in for docking. Close in, flight controllers work in a rotating frame centered on the target, where the Clohessy-Wiltshire equations describe this coupling between a burn's direction and where the chaser ends up.^[19]

The techniques were pioneered in 1965 and 1966. Gemini VI-A launched on December 15, 1965, spent roughly six hours phasing up to Gemini VII, halted at 130 feet with no relative velocity between the two, and then kept station for three orbits at ranges from 300 feet down to as close as one foot.^[17] Three months later Neil Armstrong and David Scott flew Gemini VIII to the first docking of two spacecraft in orbit, on March 16, 1966, before a short-circuited thruster stuck open and sent the joined vehicles tumbling, forcing an early return.^[27] The same phasing-and-approach logic, now largely automated, governs every cargo ship and crew capsule that visits the International Space Station.

Coming down: why deorbiting means slowing#

Leaving orbit reverses the problem of reaching it. A returning spacecraft fires its engine against its direction of travel; the retrograde burn, often only on the order of 100 meters per second from low Earth orbit, lowers the perigee into the upper atmosphere, and drag removes the rest of the enormous orbital energy as heat against a shield. Nothing needs to thrust downward: slow down, and gravity and air do the work.

The reason so small a burn suffices is that the atmosphere, not the engine, does the braking. Dropping perigee by a hundred kilometers or so is enough to guarantee that the next pass through the thermosphere bleeds off speed faster than the spacecraft can climb back out, at which point reentry is unavoidable. Everything after that is a thermal protection problem rather than an orbital mechanics one.

Orbital decay and disposal#

Atmospheric drag cleans house without any burn at all, pulling dead satellites in low orbits to fiery reentry over years or decades, a central factor in managing space debris. How long that takes depends steeply on altitude: a small spacecraft near 400 kilometers typically reenters within five years, while above about 500 kilometers natural decay alone cannot be relied on to clear an object within that time, and at 800 kilometers a drag device may need a full decade to bring a 180-kilogram satellite down.^[18] The United States once accepted a 25-year post-mission lifetime; the Federal Communications Commission cut that to five years for licensed satellites below 2,000 kilometers.^[18]

Operators have started designing around the same physics. In a January 2026 filing, SpaceX said it would move roughly 4,400 Starlink satellites from about 550 kilometers down to about 480 over the course of the year, arguing that the lower shell would cut the time a failed satellite takes to decay at solar minimum from more than four years to a few months, and that traffic and debris density are both lower below 500 kilometers.^[20]

Geostationary satellites orbit far above any usable drag, so at end of life they instead boost a few hundred kilometers higher into designated graveyard orbits. International guidelines set the target as a perigee at least 235 kilometers above the geostationary ring, plus an extra allowance that scales with how strongly solar radiation pressure acts on the particular spacecraft, and require the disposal orbit to be nearly circular.^[6]

References#

1. Basics of Space Flight - NASA.
2. Voyager: Planetary Voyage - NASA.
3. NASA's Europa Clipper Uses Mars to Go the Distance - NASA Jet Propulsion Laboratory.
4. JWST Orbit - JWST User Documentation, Space Telescope Science Institute.
5. Lecture L17: Orbit Transfers and Interplanetary Trajectories - MIT 16.07 Dynamics, MIT OpenCourseWare.
6. IADC Space Debris Mitigation Guidelines, IADC-02-01 Rev. 4, 16 January 2025 - Inter-Agency Space Debris Coordination Committee, via UNOOSA.
7. Chapter 3: Gravity and Mechanics - NASA.
8. Chapter 5: Planetary Orbits - NASA.
9. Types of Orbits - European Space Agency.
10. Catalog of Earth Satellite Orbits - NASA Earth Observatory.
11. Basics of Spaceflight: A Gravity Assist Primer - NASA.
12. What Is a Lagrange Point? - NASA.
13. Orbits and Kepler's Laws - NASA.
14. Hohmann Orbit - Encyclopaedia Britannica.
15. Europa Clipper: Mission Timeline - NASA.
16. Exploring the Unexplored: New Horizons' Mission to Pluto - NASA.
17. 55 Years Ago: The Spirit of 76, the First Rendezvous in Space - NASA.
18. State of the Art of Small Spacecraft Technology, Chapter 13: Deorbit Systems - NASA Ames Research Center.
19. Multi-Maneuver Clohessy-Wiltshire Targeting - NASA Technical Reports Server.
20. SpaceX to lower orbits of some Starlink satellites - SpaceNews, January 2026.
21. Aqua Orbit - NASA Aqua Project Science.
22. Launch Windows - NASA Mars Exploration.
23. Bi-elliptic Hohmann Transfer and Plane Change Maneuvers - Orbital Mechanics and Astrodynamics, Bryan Weber.
24. Chapter 14: Launch - NASA.
25. Planetary Physical Parameters and Planetary Fact Sheet - NASA Jet Propulsion Laboratory Solar System Dynamics and NASA Space Science Data Coordinated Archive.
26. Advanced Propulsion for Geostationary Orbit Insertion and North-South Station Keeping - NASA Technical Reports Server.
27. 55 Years Ago: Gemini VIII, the First Docking in Space - NASA.
28. Martin Lara, On Inclination Resonances in Artificial Satellite Theory - Acta Astronautica, vol. 110, 2015, pp. 239-246.