How rockets work

A rocket accelerates by ejecting part of its own mass through a nozzle. The exhaust gains rearward momentum and the vehicle gains equal forward momentum.^[1] A chemical rocket carries its reactants, including oxidizer when its propellant chemistry requires one, so it does not need atmospheric oxygen and can operate in vacuum.^[5] A circular low Earth orbit requires about 7.7 to 7.8 kilometers per second of sideways speed. A launch vehicle must supply more delta-v than that because it also climbs and loses velocity to gravity and drag; the cited low Earth orbit examples on this page range from about 9 to 10.44 kilometers per second.^[19][20]

The physics is old. Isaac Newton stated the governing law in 1687, and Konstantin Tsiolkovsky published the mathematics of variable-mass rocket flight in 1903.^[17] The engineering remains difficult because every tank, engine, and kilogram of payload must be accelerated along with the propellant that will be burned later. Staging, high expansion-ratio upper-stage nozzles, and lightweight tanks are different responses to that same mass constraint.

Newton's third law#

Newton's third law describes the paired forces between the engine and its exhaust, while conservation of momentum gives the useful accounting. For a steady rocket engine, NASA writes the thrust equation as:^[5]

F = m-dot x Ve + (pe - pa) x Ae

Here m-dot is exhaust mass flow per second, Ve is exhaust velocity at the nozzle exit, pe is exit pressure, pa is ambient pressure, and Ae is exit area. The first term is momentum thrust. The second is pressure thrust. There is no inlet-air term because a rocket carries its working mass aboard.^[5]

The ambient-pressure term explains why a fixed engine and nozzle usually produce more thrust at altitude. At the same 109 percent rated power level, NASA lists the RS-25 at 418,000 pounds-force (1,859 kilonewtons) at sea level and 512,300 pounds-force (2,279 kilonewtons) in vacuum, about 23 percent more.^[13] The vacuum does not create energy. Lower outside pressure simply reduces the atmospheric force opposing the nozzle exhaust.

Thrust alone does not set acceleration. In a simplified vertical ascent, the net force is thrust minus weight and drag; a rocket leaves the pad only when upward thrust exceeds its weight.^[3] As propellant is consumed, mass and weight fall, so the same thrust can produce increasing acceleration. Guidance must also control the direction of the thrust vector.

Nothing about this requires air or ground to push against. A New York Times editorial of January 13, 1920, criticized Robert Goddard's proposal on that mistaken basis. On July 17, 1969, the paper corrected itself and acknowledged that a rocket can function in vacuum as well as in an atmosphere.^[16]

Specific impulse#

Specific impulse, written Isp, measures thrust per unit weight flow of propellant. NASA first combines momentum and pressure thrust into an equivalent exhaust velocity, Veq = F / m-dot, then divides by standard gravity:^[6]

Isp = F / (m-dot x g0) = Veq / g0

Standard gravity g0 is 9.80665 meters per second squared. Dividing a velocity by an acceleration leaves seconds, and the result is numerically the same in coherent metric and US customary units.^[6] The unit is easy to misread: an engine rated at 450 seconds does not burn for 450 seconds. Its equivalent exhaust velocity is 450 x 9.80665, about 4.41 kilometers per second.

Specific impulse is also not thrust. A small engine and a large engine can have the same Isp while moving very different amounts of propellant each second. Because the pressure term depends on ambient pressure, an engine's quoted sea-level and vacuum Isp are different operating points and should not be mixed in one comparison.^[5][6]

The rocket equation#

Tsiolkovsky's ideal rocket equation relates a burn's velocity change to equivalent exhaust velocity and mass ratio:^[1]

delta-v = Isp x g0 x ln(m0 / mf)

m0 is mass at ignition and mf is mass at burnout. For one stage, mf still includes its dry tanks, engines, residual propellant, carried upper stages, and payload. NASA divides launch mass into payload, propellant, and structure; the equation combines payload and structure in the final mass and cannot determine their shares by itself.^[7] It also assumes constant equivalent exhaust velocity and omits gravity, aerodynamic drag, and steering losses. Those effects must be handled by a trajectory model or a separate loss budget.^[1]

Worked example: one ideal stage#

Take an idealized stage with an ignition mass of 500 metric tons, a burnout mass of 50 tons, and a constant Isp of 350 seconds.

• Equivalent exhaust velocity: 350 x 9.80665 = 3,432 m/s.
• Mass ratio: 500 / 50 = 10.
• Ideal delta-v: 3,432 x ln(10) = 7,903 m/s, or 7.90 km/s.

That result is close to orbital speed but does not mean the stage can reach orbit. A launch must also pay gravity and drag losses, and the 50 tons remaining must include all dry hardware and payload. If the target ideal delta-v is 9.4 km/s with the same Isp, the inverse equation gives MR = exp(9,400 / 3,432) = 15.47. A 500-ton stage would then finish at only 32.3 tons and consume 467.7 tons, a 93.5 percent propellant fraction. This is why the equation's logarithm is called tyrannical: each extra increment of delta-v demands a multiplicative change in mass ratio.^[1][2]

Staging#

Staging improves mass ratio by discarding tanks and engines after they have done their work. Each stage has its own ignition mass, burnout mass, and Isp; the vehicle's ideal delta-v is the sum of the stage burns, not the result of putting the whole stack into one mass ratio.^[8]

NASA distinguishes two arrangements. In serial staging, a lower stage burns and separates before an upper stage continues, as on the Saturn V. In parallel staging, strap-on boosters burn alongside a sustainer and separate while the sustainer continues, as on the Space Shuttle.^[8] The Space Launch System combines parallel boosters with later serial separation, while Falcon 9 uses two stages in series.

Worked example: why the discard matters#

Consider a deliberately simplified two-stage vehicle with a 5-ton payload:

At first-stage ignition the stack is 355 tons. At first-stage burnout, before separation, it is 55 tons. The first-stage ideal delta-v is 300 x g0 x ln(355 / 55) = 5.49 km/s. Separation drops the 25-ton dry first stage, so the upper stage ignites at 30 tons and burns down to 10 tons. Its ideal delta-v is 450 x g0 x ln(30 / 10) = 4.85 km/s, for a two-stage ideal total of 10.33 km/s.

If the empty first stage stayed attached, the upper burn would instead run from 55 tons to 35 tons and add only 1.99 km/s. Discarding 25 tons raises the upper-stage contribution by 2.85 km/s in this toy case. The example is not a vehicle design or payload prediction; it isolates the arithmetic benefit of staging. A reusable stage must carry recovery hardware and reserve propellant, so recovery changes both m0 and mf.^[19] See reusable rockets for that system-level trade.

What a rocket equation calculator can show#

A useful calculator for this page should keep ideal propulsion arithmetic separate from trajectory assumptions.

The calculator should use g0 = 9.80665 m/s^2, reject nonpositive masses and any case where m0 <= mf, and display unrounded intermediate values before rounding the result. It should not claim to calculate payload capacity, max q, gravity loss, or orbit insertion from mass ratio alone. Those require vehicle aerodynamics, thrust history, guidance, launch site, and a numerical trajectory model.^[1][20]

Explore the rocket science lab#

Use the interactive lab below to follow a launch profile, inspect flow through a converging-diverging nozzle, step through a staged vehicle, compare liquid-engine cycles, and run ideal rocket-equation cases. Its calculator reports ideal delta-v rather than a simulated orbit, so interpret the result with the gravity, drag, steering, and reserve terms discussed on this page.

Propellant families#

Rocket propellants are chosen as a system, not by specific impulse alone. Density sets tank volume, storage temperature shapes ground operations, chemistry affects materials and hazards, and the engine cycle constrains pressure and restart behavior. The ranges below are representative rather than fixed limits; exact performance depends on mixture ratio, chamber pressure, nozzle, and operating altitude.^[9]

Conventional large solid motors are mechanically simple and can remain stored, but their cast grain fixes much of the thrust-time profile. Once ignited, they generally cannot be deeply throttled, shut down, and restarted like a liquid engine. A Space Shuttle solid rocket booster used 69.8 percent ammonium perchlorate, 16 percent atomized aluminum, 12 percent PBAN synthetic-rubber binder, and small fractions of iron oxide and curing agent; each booster produced about 2.65 million pounds-force at liftoff.^[18]

Hypergolic fuel and oxidizer ignite on contact. That removes a separate ignition system and supports rapid pulsing and restarts, which is useful for spacecraft maneuvering, but common hydrazine and nitrogen tetroxide combinations are toxic, reactive, and difficult to handle.^[9] RP-1 is dense and liquid near ordinary ground temperatures. Liquid hydrogen gives the highest specific impulse among the common chemical pairs in the table, but it must be stored near 20 kelvin and needs much more tank volume.^[9][13]

Representative densities from NASA's liquid-engine survey show the scale of that volume trade. Density varies with temperature and, for kerosene, composition, so these are engineering comparison values rather than universal constants.^[9]

In one NASA comparison sized to the impulse of three Space Shuttle main engines burning for 520 seconds, oxygen and hydrogen required 24 percent less propellant mass than the methane and kerosene cases. The hydrogen case still needed far greater tank volume. The numerical result belongs to that study, not every rocket, but it demonstrates why density can outweigh a specific-impulse advantage on a booster.^[9]

Maximum heat release is not necessarily maximum engine performance. The stoichiometric oxygen-to-hydrogen mass ratio is 8:1, while the RS-25 nominal mixture ratio is about 6:1.^[9][12] Fuel-rich operation lowers gas temperature and average exhaust molecular mass. The performance gain from lighter exhaust species can outweigh the temperature reduction, while lower temperature also eases thermal loads. The extra hydrogen is therefore a deliberate mixture-ratio trade, not fuel lost by accident.^[9]

Methane occupies the middle of the density and specific-impulse trade. It is about six times as dense as liquid hydrogen in NASA's comparison and has a boiling point much closer to that of liquid oxygen, which can simplify thermal design relative to a hydrogen system.^[9] LandSpace's Zhuque-2 became the first liquid oxygen and methane launch vehicle to reach orbit on July 12, 2023.^[4]

Engine cycles#

The central feed-system problem in a liquid rocket is moving propellant into a chamber whose pressure is already high. The RS-25's main chamber operates near 2,994 pounds per square inch, about 20.6 megapascals, so its pumps must deliver at still higher pressure.^[12] An engine cycle describes where pump power comes from and where the turbine-driving fluid goes.^[9]

No cycle is automatically best. A gas-generator engine can be lighter or easier to develop than a closed-cycle engine, while staged combustion can support higher chamber pressure and avoids the open cycle's main performance loss.^[9] An expander engine is attractive for restartable upper stages, but heat-transfer area limits pump power as thrust grows. Full-flow staged combustion spreads turbine work across more mass flow, which can lower turbine temperature and stress, but two preburner circuits make startup and shutdown control demanding.^[9][26]

Current examples make the distinctions concrete. SpaceX identifies Merlin as a gas-generator engine.^[23] Blue Origin identifies BE-4 as oxygen-rich staged combustion and BE-3U as open expander.^[22] Rocket Lab's Electron user guide describes batteries powering Rutherford's electric pumps.^[24] SpaceX's Raptor uses the full-flow staged-combustion arrangement, with fuel-rich and oxidizer-rich preburner circuits.^[27] These labels describe plumbing and power flow; propellant choice, chamber pressure, nozzle expansion, mass, service life, and cost still determine the engine's actual performance.

Cooling, turbopumps, and combustion stability#

Combustion gas is hotter than the metal chamber and nozzle can survive without cooling. In regenerative cooling, propellant flows through small passages in the chamber and nozzle walls before reaching the injector. The wall stays below its material limit, and the propellant carries absorbed heat onward into the engine rather than requiring a separate expendable coolant.^[15] Regeneration does not make cooling free: channel pressure drop, wall temperature gradients, and local hot spots remain design constraints.

Turbopumps exchange tank mass for machinery. They let thin-walled tanks remain at much lower pressure than the chamber, but the pumps must raise a large cryogenic flow to injection pressure. NASA's RS-25 history compares the high-pressure fuel turbopump's power with 28 locomotives and the high-pressure oxidizer turbopump with another 11.^[12] Pump inlet pressure must remain high enough to avoid cavitation, while turbine temperature, shaft speed, and discharge pressure must stay within limits during start, throttling, and shutdown.^[9][12]

Combustion instability is a different hazard. Pressure oscillations can couple with injection and heat release until loads and heat flux rise fast enough to destroy a chamber. NASA's F-1 program manager Sonny Morea said an unstable test could "burn through the thrust chamber in milliseconds." Engineers divided the F-1 injector face with copper baffles, then used small explosive charges in hot-fire tests to disturb combustion and verify that the oscillations damped out. NASA records 65 F-1 engines flying on 13 Saturn V vehicles without a combustion-instability failure.^[14] The lesson is not that every instability has the same cure. Injector geometry, chamber acoustics, propellant response, and operating point have to be tested together.

Nozzles and expansion ratio#

A converging-diverging nozzle converts gas enthalpy and pressure into directed kinetic energy. The familiar rule that a widening passage accelerates the exhaust applies only after the flow is supersonic. If the chamber-to-ambient pressure ratio is high enough to choke the nozzle, the sequence is:^[10]

The phrase "Mach 1 at the throat" therefore needs its condition: it is true for choked operation, not for every pressure ratio. Once choked, the ratio of exit area to throat area sets the ideal exit Mach number and pressure for a given exhaust-gas state.^[10]

A fixed nozzle is pressure-matched at only one ambient pressure:

Flow separation matters because it can become asymmetric and impose transverse side loads on the nozzle.^[28] A sea-level engine therefore accepts a smaller expansion ratio than a vacuum engine, which can carry a longer, wider bell without atmospheric separation during its intended burn.^[9][10]

NASA quantified the penalty while planning a short ground-test nozzle for the J-2X upper-stage engine. Reducing area ratio from an optimized 92:1 to 59:1 lowered guaranteed minimum vacuum Isp from 448 to 435 seconds and nominal vacuum thrust from 294,000 to 285,000 pounds-force.^[9] That comparison changes nozzle geometry while holding the engine concept fixed, so it isolates why upper stages value expansion ratio. A larger bell still adds mass and length, and a booster nozzle must also survive its overexpanded low-altitude operating range.

Steering and control#

A rocket must control both translation and rotation. Guidance determines the desired path, navigation estimates the actual state, and the control system commands forces and torques to reduce the difference. NASA summarizes the vehicle-level tasks as stability and steering.^[11]

Many launch vehicles gimbal one or more engines. Tilting the thrust vector away from the center of mass creates a torque; the RS-25 can gimbal through plus or minus 11 degrees.^[11][12] Other methods include differential throttling among engines, small vernier or reaction-control thrusters, and aerodynamic surfaces while enough atmosphere remains. Exhaust vanes used on vehicles such as the V-2 also deflect the jet, but lose energy and face severe heating.^[11]

Orbit is sideways speed, not altitude#

Crossing an altitude boundary and entering orbit are different tasks. An object can coast above 100 kilometers and fall back without completing an orbit. To stay up, it needs enough horizontal speed that Earth's surface curves away as fast as the object falls. The International Space Station travels about 7.7 kilometers per second at roughly 400 kilometers altitude and circles Earth about 16 times per day.^[25]

Gravity is still close to 90 percent of its surface strength at station altitude. Astronauts appear weightless because the station, crew, and loose objects share the same continuous free fall, not because gravity has disappeared. The launch problem is therefore dominated by horizontal kinetic energy, not by reaching a height called space. A Stanford 400-kilometer polar-orbit example assigns 7.67 kilometers per second to circular orbital speed but only 0.47 kilometers per second to an energy-equivalent altitude term.^[20]

The same kinetic energy must be removed during return. Atmospheric entry converts orbital energy into heat, shocked gas, radiation, and vehicle motion, which is why an orbital return vehicle needs a thermal-protection system. Motion after engine cutoff is covered in orbital mechanics.

Max q and the ride through the atmosphere#

Dynamic pressure is q = 0.5 x rho x V^2, where rho is local air density and V is speed relative to the air. It is not itself the total force on a vehicle; aerodynamic force also depends on reference area, shape, angle of attack, and aerodynamic coefficients. It is a useful common load scale. Early in ascent the air is dense but speed is low. Later, speed is high but density is low. Dynamic pressure peaks between those regimes at max q.^[3][20]

Artemis I reached max q 1 minute 10 seconds after liftoff.^[21] A throttle-capable vehicle can reduce thrust near that point, but throttling is one option rather than the definition of max q. The Space Shuttle main engines were designed to throttle from 67 to 109 percent of their original rated power level; ascent software reduced power to limit early structural loads and later reduced it again to keep crew acceleration below 3 g.^[12] Trajectory design must balance dynamic pressure, heating, acceleration, controllability, and gravity loss.^[20]

Failure chains and safety layers#

Rocket failures are often coupled. A feed-system problem can reduce chamber pressure, which changes thrust; asymmetric thrust can increase angle of attack, which raises aerodynamic load; that load can then exceed a structural or control limit. The response depends on the vehicle, flight phase, and available margin.

Mission protection and public protection are not the same layer. A vehicle computer may save a mission by shutting down one engine or changing guidance, but only if the launcher was designed with that redundancy and delta-v reserve. For public safety, launch operators establish air, sea, and ground hazard areas and use flight-safety systems that can render a vehicle non-propulsive if it leaves its allowed corridor.^[29] For a crewed capsule, a launch abort system is another independent layer: Orion's solid-motor system is designed to pull the crew module away from a failing launcher on the pad or during ascent, orient it, and permit parachute landing.^[30]

Gravity and drag losses#

Gravity and drag losses explain why circular orbital speed is not the launch vehicle's required ideal delta-v. ESA uses about 9.0 kilometers per second for a simplified low Earth orbit launcher example.^[19] A Stanford lecture obtains 10.44 kilometers per second for a specific 400-kilometer polar launch with no benefit from Earth's rotation.^[20] The figures are not competing constants; they use different mission and trajectory assumptions.

The Stanford polar case provides a useful loss-budget diagram:^[20]

In compact form, gravity loss is the time integral of g x sin(gamma) during powered flight, where gamma is flight-path angle above the local horizontal. Drag loss is the time integral of D / m. Both depend on the actual trajectory.^[20] Steering also costs performance when thrust is not aligned with the velocity direction, and reserve propellant and dispersions add operational margin that is not a physical loss term.

A shorter burn and an earlier pitch toward horizontal reduce gravity loss, but high thrust adds engine mass and can violate max-q, acceleration, or heating limits. Staying vertical longer clears dense air but spends more time with thrust opposing gravity. A gravity turn balances those effects by using gravity and controlled thrust-vector changes to bend the path while keeping angle of attack manageable.^[20]

Earth's rotation changes the launch-site boundary condition. At the equator the surface moves east at about 1,650 kilometers per hour, or 465 meters per second.^[2] At latitude phi, the eastward component is approximately 465 x cos(phi) meters per second before accounting for the desired orbital plane. An eastward low-inclination launch can use much of that speed; a polar mission cannot. This is why a single universal "delta-v to orbit" number is misleading and why calculator results must state launch site, orbit, and loss assumptions.

References#

1. Ideal Rocket Equation - NASA Glenn Research Center.
2. Basics of Space Flight, Chapter 14: Launch - NASA Science.
3. Four Forces on a Rocket - NASA Glenn Research Center.
4. Zhuque-2 Y2 launch successfully enters orbit - LandSpace, July 13, 2023.
5. Thrust Equations Summary - NASA Glenn Research Center.
6. Specific Impulse - NASA Glenn Research Center.
7. Mass Ratios - NASA Glenn Research Center.
8. Booster Staging - NASA Glenn Research Center.
9. Overview of Liquid Propellant Rocket Engine Systems and the J-2X - Richard O. Ballard, NASA Marshall Space Flight Center, 2011 (NASA Technical Reports Server).
10. Nozzle Design - NASA Glenn Research Center.
11. Rocket Control - NASA Glenn Research Center.
12. Space Shuttle Main Engine: The Relentless Pursuit of Improvement - NASA Marshall Space Flight Center / AIAA Space 2011 Conference (NASA Technical Reports Server).
13. SLS RS-25 Engine Fact Sheet - NASA, April 2025.
14. Solving Combustion Instability and Saving America's First Trips to the Moon - NASA, July 12, 2019.
15. 2024 Invention of the Year Winner: Thrust Chamber Liner and Fabrication Method - NASA.
16. 95 Years Ago: Goddard's First Liquid-Fueled Rocket - NASA, March 17, 2021.
17. Rocket History: 20th Century and Beyond - NASA Glenn Research Center.
18. The Space Shuttle - NASA.
19. Towards Reusable Launchers: A Widening Perspective - ESA Bulletin 87.
20. Launch Trajectories, AA284a Advanced Rocket Propulsion, Lecture 7 - M. Arif Karabeyoglu, Stanford University.
21. Artemis I Mission Timeline - NASA.
22. Blue Origin Engines - Blue Origin.
23. Falcon 9 - SpaceX.
24. Electron Payload User Guide, version 7.0 - Rocket Lab, 2024.
25. International Space Station Facts and Figures - NASA.
26. New Rocket Engine Combustion Cycle Technology Testing Reaches 100% Power Level - NASA, January 8, 2013.
27. Statement of Jeffrey Thornburg, SpaceX, before the House Armed Services Subcommittee on Strategic Forces - U.S. House of Representatives, June 26, 2015.
28. Flow Separation Side Loads Excitation of Rocket Nozzle FEM - Kurt B. Smalley et al., NASA Marshall Space Flight Center and University of Alabama in Huntsville, 2007 (NASA Technical Reports Server).
29. Safe to Launch - Federal Aviation Administration.
30. Final NASA Test Qualifies Orion's Abort System for Crewed Artemis Missions - NASA, April 6, 2022.