Small impulses.
Long consequences.

A local orbital flight lab, grounded in a restricted Newtonian model. Fictional missions; real integration. This is an educational experience, not a spacecraft engineering or navigation tool.

Your first flight

Start on 02 · Higher ground and choose Fly guided mission. In about 27 visible seconds, Kestrel launches into its stated insertion orbit, burns onto a transfer ellipse, coasts to apoapsis, and burns again to circularize. The clock advances about 2½ simulated hours. The director explains each decisive maneuver. Mission completion requires another 180 simulated seconds inside the orbit criteria.

For manual flight, select Launch probe. The simulation stays paused for planning. Adjust Prograde or Radial and inspect the gold trajectory and predicted apses. Execute maneuver applies the burn now, or arms it for the specified delay. Load flight solution suggests a departure maneuver; after a successful transfer injection, it schedules a circularization at the next apoapsis. Execute that plan to fly the full manual sequence. A zero-length burn cannot be executed.

Launching means releasing the probe at the mission’s specified position and velocity, not simulating a rocket ascent from the surface. Guided flight resets the selected mission. Take the controls cancels guidance and pauses the current flight. Cancel armed maneuver stops a pending burn and pauses flight. You can then edit the plan. Reset restores the selected mission and its budget.

The three missions

Read the scene

Cyan is the current coast orbit, calculated forward from the live state. The brighter trail records actual past positions. Dashed gold is the coast trajectory following your planned impulse. The gold reticle is the maneuver node. Green is the target altitude corridor. PE and AP mark periapsis and apoapsis of the planned orbit when a burn is set, otherwise the current orbit. Live telemetry always describes the current spacecraft.

The default camera gives an oblique view of the orbital plane. Plan view removes foreshortening, so circular orbits appear circular. All orbital positions use the same kilometer scale; the probe, reticle, atmosphere glow, and thin corridor have enlarged visual sizes for clarity. Earth’s surface uses the locally bundled 1024×512 NASA Blue Marble texture (434.3 KB). Credit: NASA/Goddard Space Flight Center Scientific Visualization Studio; Blue Marble data courtesy of Reto Stockli (NASA/GSFC) and NASA Earth Observatory. Recognizable geography comes from the source mosaic; the terminator and atmosphere are illustrative lighting, not a live Earth ephemeris. Prediction is limited to one period or 42,000 simulated seconds (whichever is shorter); open paths use a 22,000–24,000 second horizon and can leave the frame. The Fit button resets view zoom.

Equations and coherent units

State consists of planar position (x,y) in kilometers and velocity (vx,vy) in kilometers per second. Simulation time is seconds. A massless probe accelerates under a fixed central Earth:

a = −μ r / |r|³
ε = |v|²/2 − μ/|r|
aₒ = −μ/(2ε), rₚ = aₒ(1−e), rₐ = aₒ(1+e)

μ = 398600.435507 km³/s² is the Earth GM in JPL Solar System Dynamics’ DE440 astrodynamic parameter table. The model chooses a spherical surface radius of 6,371 km; all altitude readouts subtract it. Inverse-square gravity and elliptical-orbit terminology are explained in NASA’s Gravity & Mechanics chapter.

Velocity-Verlet advances in steps no longer than 2 s: position uses the old acceleration, then velocity uses the mean of old and new acceleration. Rendering never assigns positions. Prediction copies the state and runs the same integrator. Time acceleration repeats integration steps; it does not enlarge them. Partial steps land on maneuver times exactly.

Inputs are in m/s and convert to km/s inside the model. Prograde here means the transverse direction perpendicular to the radius, following orbital motion; radial points outward. This orthonormal local frame makes the cost Δv = √(prograde² + radial²). Away from an apsis, the transverse direction need not align with the full velocity vector. These ideal impulses have no duration, thrust curve, attitude, mass, or propellant equation.

Why a Hohmann transfer works here

For circular coplanar radii r₁ and r₂, the transfer semimajor axis is aₜ = (r₁+r₂)/2. Vis-viva gives the two velocities; the half-period gives the coast time:

Δv₁ = √[μ(2/r₁−1/aₜ)] − √(μ/r₁)
Δv₂ = √(μ/r₂) − √[μ(2/r₂−1/aₜ)]
t = π √(aₜ³/μ)

The transfer ellipse touches both circular orbits. The first impulse changes the opposite apsis; the arrival impulse matches circular speed. These expressions follow from the two-body energy equation above. See NASA’s Trajectories chapter and the circularization discussion in NASA’s Planetary Orbits chapter. This demo does not claim general optimality across every radius ratio or maneuver class.

Boundaries and limitations

There is no atmosphere or atmospheric drag, oblateness, Earth rotation in the dynamics, third-body gravity, radiation pressure, relativity, thrust duration, orbit-plane change, fuel mass, ephemeris, navigation noise, physical rendezvous target, or communications delay. The capture encounter begins at closest approach. It is an energy-capture lesson around Earth, not an interplanetary patched-conic mission.

A surface crossing ends flight at the first integrated state with radius at or below 6,371 km. With the maximum 2-second step, the impact position and time are approximate within a step. Outward motion beyond 120,000 km geocentric radius with nonnegative specific energy ends as departure. That radius is a chosen simulation boundary, not Earth’s physical sphere of influence. Being temporarily out of frame is not itself escape.

Orbit prediction and trails are bounded. Rendering uses Canvas 2D, a cached 440×440 sphere rendered from the local NASA texture, capped pixel density, slower redraws while paused, and reduced density after repeatedly slow predictions. Backgrounding the browser pauses flight. The NASA texture is included locally. No remote requests, external services, accounts, or downloads are needed at runtime.

Controls and recovery

Space pauses or resumes when the page is focused. R resets; G starts or takes over a guided flight. Native buttons activate with Enter or Space. Tab navigates every control; arrow keys adjust a focused slider. While typing in an input, flight shortcuts are suppressed. Escape closes the manual. Opening the manual pauses an active flight.

If the gold orbit crosses Earth, reduce retrograde thrust or add a safer radial/tangential change before committing. If a burn exceeds the delta-v budget, reduce it. An unreachable future node is rejected. Surface impact and departure show an explicit reset path. All missions can be reset at any time.

Verification

The included Node test suite checks circular-orbit stability and conserved quantities, second-order step convergence, analytical Hohmann relationships against numerical flight, radial/prograde effects, preview parity, fuel checks, scheduled burns, collision/escape, bounded trails, and mission success and failure. Automated tests import the same physics.mjs and flight.mjs modules used by this page. Browser QA is recorded separately in the delivered QA report, with screenshots.

Sources retrieved September 2, 2026. No quotations are required to use the manual.

Return to your flight →