The Patched Conic Approximation

The trick: tape together solved problems

Planning a flight to Mars looks hopeless if you write it out honestly. The spacecraft feels the Sun, Earth, Mars, the Moon, Jupiter, and every other body in the solar system simultaneously, and the gravitational N-body problem has no closed-form solution for more than two bodies — Henri Poincaré proved in the 1890s that even three bodies admit no general analytic integral. You cannot write down the trajectory as a formula.

The patched conic approximation sidesteps this entirely with a beautifully lazy idea: at any given moment, pretend only one body matters. Near Earth, model only Earth's gravity. Out in interplanetary space, model only the Sun. Near Mars, model only Mars. Each of those is a two-body problem, and the two-body problem is solved exactly — the orbit is a conic section, an ellipse, parabola, or hyperbola, courtesy of Kepler and Newton. You compute three separate conic arcs and then "patch" them together at the boundaries where one body hands the spacecraft off to the next. The unsolvable problem becomes a chain of solved ones.

A typical Earth-to-Mars flight is three conics in a row: a departure hyperbola as the craft escapes Earth, a heliocentric transfer ellipse coasting between the two planetary orbits, and an arrival hyperbola as Mars captures it. The art is entirely in the patching — making the velocity at the end of one conic become the correct starting velocity for the next.

The sphere of influence: where one body hands off to the next

The boundary at which you switch from "Earth only" to "Sun only" is the sphere of influence (SOI). It is not a physical surface — gravity has infinite range — but a bookkeeping shell inside which a given body's gravity dominates strongly enough that the next body can be treated as a small perturbation. The standard estimate comes from Pierre-Simon Laplace:

r_SOI = a · (m / M)^(2/5)

where a is the body's orbital semi-major axis around its parent, m is the body's mass, and M is the parent's mass. The 2/5 power comes from balancing the parent's tidal perturbation against the body's own gravity, not from a simple force balance (that balance — the gravity null point — gives a different, larger radius). Plugging in Earth's numbers:

a = 1.496 × 10^8 km,  m_Earth/M_Sun = 3.003 × 10^-6
r_SOI = 1.496 × 10^8 × (3.003 × 10^-6)^0.4 ≈ 9.24 × 10^5 km

So Earth's SOI is about 924,000 km in radius — roughly 2.4 times the average Earth–Moon distance of 384,400 km. (The Moon orbits well inside Earth's SOI, which is why lunar missions are themselves patched conic problems: Earth → Moon-SOI → Moon.) The values for the other planets span a huge range because both mass and distance vary.

Body	Semi-major axis a	Mass ratio m/M☉	SOI radius	SOI / body radius
Mercury	0.387 AU	1.66 × 10⁻⁷	1.12 × 10⁵ km	~46
Venus	0.723 AU	2.45 × 10⁻⁶	6.16 × 10⁵ km	~102
Earth	1.000 AU	3.00 × 10⁻⁶	9.24 × 10⁵ km	~145
Mars	1.524 AU	3.23 × 10⁻⁷	5.77 × 10⁵ km	~170
Jupiter	5.203 AU	9.54 × 10⁻⁴	4.82 × 10⁷ km	~674
Saturn	9.537 AU	2.86 × 10⁻⁴	5.46 × 10⁷ km	~906
Moon (around Earth)	384,400 km	0.0123 (m/M_Earth)	6.6 × 10⁴ km	~38

Note that Mars's SOI (≈577,000 km) is smaller than Earth's despite being farther out, because Mars is only about a tenth of Earth's mass. Jupiter's SOI, at nearly 48 million km, is so vast that it makes the giant planets superb gravity-assist partners.

The governing equations of each conic leg

Within a single SOI the motion obeys the two-body equation, whose energy form (the vis-viva equation) is the workhorse of the whole method:

v² = μ ( 2/r − 1/a )

Here μ = GM is the body's standard gravitational parameter, r the current distance, and a the semi-major axis of the conic (positive for an ellipse, negative for a hyperbola). Useful values of μ: Sun 1.327 × 10²⁰ m³/s², Earth 3.986 × 10¹⁴ m³/s², Mars 4.283 × 10¹³ m³/s². The orbital period of the elliptical legs follows Kepler's third law, T = 2π √(a³/μ).

The quantity that does the actual patching is the hyperbolic excess velocity, v∞. On an escape hyperbola the semi-major axis is negative, and the vis-viva equation at infinite range gives

v∞² = −μ/a = v² − 2μ/r   →   C3 ≡ v∞²

v∞ is the speed the craft keeps after climbing entirely out of the planet's well — the velocity left over at the SOI edge. Launch vehicles quote their capability as the characteristic energy C3 = v∞², in km²/s². The patch condition is a vector equation: the spacecraft's heliocentric velocity at the SOI must equal the planet's heliocentric velocity plus the spacecraft's planet-relative velocity.

v_helio = v_planet + v∞    (vector sum at the SOI boundary)

Because the SOI radius (≈10⁶ km) is tiny compared with the heliocentric orbit (≈10⁸ km), the standard simplification treats the patch as instantaneous and co-located: the spacecraft is taken to be at the planet's position with the planet's velocity at the moment of patching. This is the single biggest source of error, and the one real missions correct numerically.

Worked example: the Earth-to-Mars Hohmann transfer

Take the cheapest classic case — a Hohmann transfer ellipse tangent to both circular planetary orbits. Approximate the orbits as circular at r_Earth = 1.000 AU = 1.496 × 10¹¹ m and r_Mars = 1.524 AU = 2.279 × 10¹¹ m. The transfer ellipse has perihelion at Earth and aphelion at Mars, so its semi-major axis is

a_t = (r_Earth + r_Mars)/2 = (1.496 + 2.279)/2 × 10¹¹ = 1.888 × 10¹¹ m

The heliocentric speed at perihelion (departure) from vis-viva with μ_Sun:

v_peri = √[ μ_Sun (2/r_Earth − 1/a_t) ]
       = √[ 1.327e20 × (2/1.496e11 − 1/1.888e11) ]
       ≈ 32.7 km/s

Earth's own circular speed is v_Earth = √(μ_Sun/r_Earth) ≈ 29.8 km/s. So the spacecraft must be travelling 32.7 − 29.8 ≈ 2.9 km/s faster than Earth when it leaves the SOI. That excess is exactly the departure v∞:

v∞ ≈ 2.94 km/s   →   C3 ≈ 8.6 km²/s²

Now back inside Earth's SOI: from a 200 km circular parking orbit (radius 6,578 km, circular speed √(μ_Earth/r) ≈ 7.79 km/s), the speed needed at that radius to be on the escape hyperbola with this v∞ is

v_perigee = √(v∞² + 2μ_Earth/r) = √(2.94² + 2×3.986e5/6578) ≈ 11.4 km/s

The trans-Mars injection burn is therefore Δv = 11.4 − 7.79 ≈ 3.6 km/s. The heliocentric coast lasts half the transfer-ellipse period, T/2 = π √(a_t³/μ_Sun) ≈ 259 days — close to the ~7-month cruise of real Mars missions. At Mars, the symmetric arithmetic gives an arrival v∞ of about 2.65 km/s, which sets the arrival hyperbola and the capture or aerobraking strategy. Three conics, a handful of vis-viva evaluations, and you have a complete first-cut mission plan.

How the legs are actually assembled

In practice the patched conic recipe runs as a pipeline, and several pieces have their own classic algorithms:

• Pick a launch and arrival date. The departure and arrival v∞ both depend on where the planets are. Sweeping all date pairs and contouring the required C3 produces the famous porkchop plot — concentric ovals of constant launch energy that mission designers read like a topographic map to choose a launch window.
• Solve Lambert's problem. Given the Earth's position at departure, Mars's position at arrival, and the flight time, Lambert's problem returns the unique heliocentric conic connecting them. It generalises the Hohmann ellipse to any transfer time and is the engine behind every porkchop plot.
• Patch in the planet-centred hyperbolae. The heliocentric departure velocity minus Earth's velocity gives the departure v∞ vector; that fixes the escape hyperbola and the required injection burn. The same at the far end fixes the arrival hyperbola.
• Hand off to a real propagator. The conic solution seeds a full numerical integrator — JPL's MONTE or the open-source GMAT — that includes every planet, the Moon, solar radiation pressure, and general-relativistic corrections, then optimises the trajectory and schedules the trajectory-correction manoeuvres.

Gravity assists are just one more patched hyperbola

The patched conic framework extends effortlessly to gravity assists — the manoeuvre that flung Voyager across the solar system. A flyby is simply an extra hyperbolic leg inside the assisting planet's SOI. Energy conservation in the planet's frame means the spacecraft leaves with the same speed it arrived, v∞,in = v∞,out, but the velocity vector is rotated by the turn angle δ:

sin(δ/2) = 1 / (1 + r_p v∞² / μ)

where r_p is the closest-approach (periapsis) distance. When that rotated v∞ is added back to the planet's heliocentric velocity, the spacecraft's Sun-relative speed changes — it has borrowed orbital energy from the planet (the planet loses an utterly negligible amount, conserving total momentum). Voyager 2 chained four such patched hyperbolae — Jupiter (1979), Saturn (1981), Uranus (1986), Neptune (1989) — on the rare Grand Tour alignment that recurs only every 175 years, reaching Neptune on a launch energy that could barely have reached Jupiter directly.

Method	Bodies modelled per leg	Solution type	Cost to compute	Best used for
Patched conic	One at a time	Exact conic per leg, patched	Microseconds	First guess, mission surveys, teaching
Restricted 3-body (CR3BP)	Two massive + spacecraft	Numerical; Lagrange points, manifolds	Seconds–minutes	Low-energy transfers, halo orbits
Full N-body integration	All of them	Numerical, no closed form	Minutes–hours	Final navigation, ephemeris-accurate ops

History and the missions it flew

The ingredients are centuries old. Johannes Kepler established that orbits are conic sections (1609–1619); Newton derived them from the inverse-square law (1687); Laplace introduced the sphere-of-influence concept in the early 1800s. But the synthesis into a practical interplanetary design tool belongs to the dawn of the space age. Through the late 1950s and 1960s, engineers at the Jet Propulsion Laboratory and pioneers such as Krafft Ehricke and Derek Lawden worked out how to chain the conics for real spacecraft. The 1971 textbook Fundamentals of Astrodynamics by Roger Bate, Donald Mueller, and Jerry White — written for the U.S. Air Force Academy — codified the patched conic method for a generation of engineers and is still in print.

It flew everything in the early planetary program. Mariner 2 reached Venus in 1962 and Mariner 4 returned the first close-up images of Mars in 1965, both designed on patched conics. Pioneer 10/11, Voyager 1/2, Galileo (Venus-Earth-Earth Gravity Assist, 1989–1995), Cassini (Venus-Venus-Earth-Jupiter assists, 1997–2004), and New Horizons (Jupiter assist en route to Pluto, 2006–2015) were all first sketched as patched conics before numerical refinement. The method remains the universal opening move in astrodynamics.

Variants, limits, and the low-energy alternative

• Lunar and planetary capture sequences. A lunar mission patches Earth → Moon-SOI → Moon; a Jupiter orbit insertion patches the heliocentric arrival hyperbola into a Jupiter-centred capture ellipse. The bookkeeping is identical, just with more seams.
• Patched conic with gravity assist (PCGA). Adds flyby hyperbolae at intermediate planets; this is the design language of every Grand-Tour-class mission.
• Lambert-based broad search. Running Lambert's solver over a grid of dates is how trajectory teams discover unexpected windows and multi-flyby sequences (e.g. the Cassini VVEJGA route).
• Where it breaks: low-energy transfers. The patched conic worldview assumes a body's gravity is either dominant or negligible. Near the Lagrange points of the Sun–Earth or Earth–Moon system, two gravities are comparable and neither can be ignored. The weak stability boundary / invariant-manifold transfers — used by Japan's Hiten (1991) to salvage a lunar mission and by NASA's GRAIL probes — live precisely in this regime and require the circular restricted three-body model instead. They trade slower flight times for dramatically lower fuel.

Common misconceptions and subtleties

• The SOI is not a force-balance surface. Where Earth's and the Sun's gravitational forces on the spacecraft are equal lies about 260,000 km out — much closer than the 924,000 km SOI. The SOI is defined by relative perturbation magnitudes (the 2/5-power Laplace criterion), not raw force, because what matters is which gravity is the small correction to which.
• v∞ is not the launch speed and not the escape speed. It is the speed remaining after escape. A craft on a parabolic (just-barely-escape) trajectory has v∞ = 0. Any interplanetary mission needs v∞ > 0, costing energy beyond escape.
• A gravity assist does not violate energy conservation. In the planet's frame the speed is unchanged; the Sun-frame speed-up is real and comes from the planet's orbital energy. The planet is decelerated by an immeasurably tiny amount — total momentum and energy are conserved.
• The instantaneous patch is a fiction. Treating the SOI crossing as zero-time and the spacecraft as co-located with the planet introduces position errors of tens of thousands of kilometres at Mars. This is why no real mission flies the raw conic — it flies the numerically corrected version, with mid-course manoeuvres budgeted in advance.
• "Patched" means matching state, not forces. At the seam you match position and velocity vectors between the two conic frames; you do not attempt to make the gravitational accelerations continuous (they jump, because you've switched which body you're modelling). The discontinuity is the price of the approximation.