How rockets work

A rocket is a machine that moves by throwing part of its own mass out of a nozzle at very high speed: burning propellant accelerates hot gas out the back, and the reaction pushes the vehicle forward.^[1] Because a rocket carries its oxidizer along with its fuel, it works in the vacuum of space where turbine engines and propellers cannot.^[5] Reaching low Earth orbit means accelerating to about 7.8 kilometers per second, which forces orbital rockets to be roughly 85 to 90 percent propellant at liftoff.^[2]

The physics is old. Isaac Newton stated the governing law in 1687, and the Russian schoolteacher Konstantin Tsiolkovsky published the mathematics of rocket flight in 1903, arguing that the speed and range of a rocket are limited only by the exhaust velocity of the escaping gas.^[18] The engineering remains hard because the numbers are unforgiving, and nearly every feature of a modern launcher, from staging to the industry's current shift toward methane fuel, follows from that propellant constraint.

Newton's third law#

Newton's third law says that for every action there is an equal and opposite reaction. A rocket engine accelerates a jet of gas to between 2 and 4.7 kilometers per second, and the momentum leaving through the nozzle appears as thrust on the vehicle.

NASA writes the rocket thrust equation with two terms. Thrust equals the mass flow rate of exhaust multiplied by its exit velocity, plus the nozzle exit area multiplied by the difference between the pressure of the gas at the nozzle exit and the pressure of the surrounding air.^[5] There is no term for incoming air, because a rocket takes none aboard. That single absence is the whole reason a rocket works where a jet engine does not.^[5]

The pressure term is small compared with the momentum term, but it is not negligible, and it is why the same engine gets stronger as the air thins out. NASA rates the RS-25 at 418,000 pounds of thrust at sea level and 512,300 pounds in vacuum, a gain of about 23 percent from nothing but the drop in outside pressure.^[13]

Nothing about any of this requires pushing against air or ground. The push happens between the rocket and its own exhaust, the way a canoeist slides backward after heaving a stone forward. The point was once controversial. A New York Times editorial of January 13, 1920, scoffed at Robert Goddard's ideas on the mistaken grounds that a rocket needs air to push against; on July 17, 1969, the day after Apollo 11 lifted off for the Moon, the paper printed a correction saying that "it is now definitely established that a rocket can function in a vacuum as well as in an atmosphere."^[17]

Specific impulse#

Specific impulse, the standard measure of a rocket engine's efficiency, comes from a simple division. Divide thrust by mass flow rate and what comes out is a velocity. NASA calls it the equivalent velocity, and it folds the pressure term into a single number, so that thrust is simply mass flow multiplied by equivalent velocity.^[6] Specific impulse is that velocity divided by the standard gravitational acceleration, about 9.8 meters per second squared:

Isp = Veq / g0

The division by gravity is what leaves the answer in seconds, because specific impulse is really thrust divided by the weight flow rate of propellant rather than the mass flow rate, and the units of weight cancel against the units of force.^[6] Two consequences follow. The number is identical in metric and imperial units, which is one reason engineers like it. And it is not a duration of anything the engine does: an engine with a specific impulse of 450 seconds does not burn for 450 seconds. It exhausts at 450 times 9.8, or about 4.4 kilometers per second.

That conversion matters, because the rocket equation cares about exhaust velocity, not about seconds. When a hydrogen engine at 450 seconds is compared with a kerosene engine at 310, the real comparison is 4.4 kilometers per second against 3.0, and the difference compounds through a logarithm.

The rocket equation#

Tsiolkovsky's rocket equation says that a rocket's total possible change in velocity, called delta-v, equals the exhaust velocity multiplied by the natural logarithm of the mass ratio: the fueled mass divided by the empty mass.^[1] Written with specific impulse in place of velocity, it is delta-v = Isp x g0 x ln(MR).^[1] Performance depends on exactly two things: how fast you throw propellant out the back, and what fraction of the vehicle is propellant.

NASA groups a rocket's mass into three parts. The payload, the propellant, and the structure, meaning everything else including tanks, engines and pumps. Empty mass is payload plus structure, full mass is empty mass plus propellant, and the mass ratio MR is the second divided by the first.^[7] Two related figures fall out of the same accounting: the payload ratio, which engineers want large, and the structural coefficient, the structural mass divided by structure plus propellant, which they want small.^[7]

Two worked examples show how tight the margins are.

NASA's own runs the equation for a hydrogen and oxygen engine at about 350 seconds of specific impulse, aiming for the roughly 25,000 feet per second needed to reach a 200-mile orbit. The mass ratio comes out at about 10, which is another way of saying that 90 percent of the launch weight of an orbital rocket is propellant, and that at the current state of the art the payload is about 1 percent of it.^[1]

ESA's version is harsher because it counts the whole trip. Taking about 9,000 meters per second as the velocity increment needed to reach low Earth orbit and an average exhaust velocity of 4,000 meters per second, which is roughly the best that hydrogen and oxygen deliver, the surviving mass fraction is 10.54 percent. So 89.46 percent of liftoff mass is propellant, and everything else, meaning tanks, engines, structure, avionics and payload, has to fit in the remaining tenth. Assign 1 percent to payload and the vehicle itself must be built inside 9.54 percent of its own launch mass.^[20]

The logarithm is what makes this so difficult. Adding more propellant helps less and less, because the new propellant must itself be carried and accelerated before it is burned. With a kerosene engine at a vacuum specific impulse near 340 seconds, exhausting at about 3.3 kilometers per second, delivering the roughly 9.5 kilometers per second that orbit effectively requires once losses are counted demands a mass ratio near 18: for every kilogram of vehicle that reaches orbit, about 17 kilograms of propellant are burned on the way up.

Astronaut Don Pettit called this "the tyranny of the rocket equation." A car is about 4 percent fuel by mass and a loaded airliner about 40 percent, but an orbital rocket must be roughly 85 to 90 percent propellant, leaving only a few percent of liftoff mass for tanks, engines, avionics, and payload.^[2]

Staging#

Staging is the standard escape from the rocket equation's tyranny: throw hardware away in flight. Empty tanks are dead weight, so rockets are built in stages that drop off when spent, and each discard resets the mass ratio for what is left.

NASA distinguishes two arrangements. In serial staging a smaller stage rides on top of a larger one, the lower stage burns out and separates, and the upper stage then lights and carries the payload the rest of the way; the Saturn V staged twice this way on its climb to Earth orbit. In parallel staging, strap-on boosters ignite at liftoff alongside a central sustainer and are discarded when their propellant runs out while the sustainer keeps burning, as on the Space Shuttle.^[8] The Space Launch System and Falcon Heavy both use the parallel arrangement, and some launchers combine the two. Most current vehicles, including Falcon 9, reach orbit with two stages in series: the first supplies the initial 2 to 3 kilometers per second and falls away, while a smaller upper stage does the rest.

No launcher in service reaches orbit in a single stage. The mass fraction that would demand sits at the edge of what tanks and engines allow, and it gets harder still for a vehicle that also has to survive re-entry and fly again, which is ESA's stated reason for treating single-stage reusable launchers as a much harder problem than expendable stages.^[20] Recovering and reflying the discarded stages instead is the province of reusable rockets.

Propellant families#

Rocket propellants fall into five main families, and the choice among them shapes the whole vehicle.

Solid motors are simple and storable for years and deliver enormous thrust, but they cannot be throttled deeply or shut down once lit. A Space Shuttle booster was 69.8 percent ammonium perchlorate as oxidizer, 16 percent atomized aluminum powder as fuel, 12 percent polybutadiene binder, and small fractions of iron oxide catalyst and epoxy curing agent, cast as a single rubbery grain that produced 2.65 million pounds of thrust at liftoff.^[19] Hypergolic propellants ignite on contact with each other, which makes them dependable for spacecraft engines that must restart far from Earth, at the cost of extreme toxicity; NASA notes that eliminating the ignition system is exactly what buys thrust on demand and rapid pulsing.^[9] Kerosene is dense and easy to handle, but it deposits soot in engine passages (coking), a headache for reuse. Hydrogen offers the best efficiency of any common chemical fuel, about 450 seconds of specific impulse, yet it is absurdly bulky and has to be held at about minus 423 degrees Fahrenheit, roughly 20 kelvin, which is the cold end of the RS-25's stated operating range.^[13]

Bulk is what decides most vehicle designs, because efficiency and density pull in opposite directions.

NASA sized the trade for an impulse requirement equivalent to three RS-25 engines burning for 520 seconds. Hydrogen and oxygen need 24 percent less propellant mass than methane or kerosene, and yet the tanks come out so much larger that the vehicle grows, which is why hydrogen is more common on upper stages than on boosters. The lesson NASA draws is blunt: specific impulse is not everything, especially on a first stage.^[9]

Engines also do not burn their propellants in the proportion that releases the most heat. Stoichiometric combustion of hydrogen and oxygen is a mixture ratio of 8 to 1, but the RS-25 runs near 6 to 1.^[9][19] Running fuel rich lowers chamber and plume temperatures, and it leaves unburned hydrogen molecules in the exhaust. Because those molecules are light, they accelerate to a higher velocity than steam would, so the engine gains performance by deliberately wasting fuel.^[9]

Methane has been the choice for most new large engines of the 2020s. It burns more cleanly than kerosene, leaving less deposit behind in an engine meant to fly again. It is six times denser than liquid hydrogen and stays liquid at a temperature much closer to liquid oxygen's, which simplifies tanks and ground systems.^[9] Its efficiency lands between kerosene and hydrogen. And it can in principle be synthesized on Mars from carbon dioxide and water ice, an assumption built into SpaceX's Mars settlement architecture. The first methane-fueled rocket to reach orbit was LandSpace's Zhuque-2, on July 12, 2023.^[4] Several of the largest vehicles to enter service since then burn methalox, including Starship, New Glenn, Vulcan Centaur, and Zhuque-3, which reached orbit on its first flight in December 2025 and came within about 40 meters of a successful booster landing.^[27] Hydrogen has not gone away: the SLS core stage still burns it.^[13]

Engine cycles#

The central engineering problem in a liquid rocket is feeding propellant into a chamber that is already at enormous pressure. The RS-25 runs its main chamber at 2,994 pounds per square inch, roughly 200 atmospheres, and something has to push propellant in against that.^[12] Engine cycles are the different answers, and NASA classifies them by how the turbopumps get their power.^[9]

Gas-generator engines trade a little efficiency for simplicity: SpaceX's Falcon 9 payload user's guide describes the Merlin as a gas generator cycle engine with a single shaft carrying both the oxygen and the fuel pump, chosen over the more complex staged combustion.^[24] Staged-combustion engines route their turbine gas into the main chamber instead so nothing is wasted, which is why NASA calls the cycle high performance and high maintenance in the same breath.^[9] The oxygen-rich variant, long a Soviet and Russian specialty, is now American too: Blue Origin's BE-4 burns liquid oxygen and liquefied natural gas in an oxygen-rich staged combustion cycle for 550,000 pounds of sea-level thrust.^[23] Full-flow staged combustion existed only as a ground demonstrator when NASA surveyed the options in 2011; SpaceX's Raptor now flies it.^[9] At the other end of the scale, Rocket Lab's Rutherford replaces turbomachinery entirely with brushless DC motors and lithium polymer batteries, and puts nine on Electron's first stage plus one vacuum-optimized version on the second.^[25]

Cooling, turbopumps, and combustion stability#

Chamber gas in a large rocket engine runs at thousands of degrees, well above the melting point of the metal containing it. The usual solution is regenerative cooling: small channels built into the chamber and nozzle walls circulate fuel, or sometimes oxidizer, along the outside of the hot wall before that propellant is injected and burned.^[16] The coolant carries the heat back into the combustion process rather than dumping it, so the wall survives and almost none of the energy is lost. Falcon 9's Merlin uses a regeneratively cooled chamber with a milled copper alloy liner for exactly this reason.^[24]

The turbopumps that feed the chamber are among the most highly loaded machines ever built. The RS-25's pumps turn 580 times a second, close to 35,000 rpm, against roughly 9,000 for a stock car engine and 19,000 for a Formula 1 engine.^[14] The engine's high-pressure fuel turbopump alone produces as much power as 28 locomotives, and the high-pressure oxidizer turbopump the equivalent of 11 more.^[12]

The subtler danger is combustion instability, in which pressure waves inside the chamber reinforce one another and shake the engine apart. NASA's F-1 program manager Sonny Morea described what it did to a test engine: "once we had that instability, it would burn through the thrust chamber in milliseconds." Three years of work produced the fix, and it was mechanical rather than chemical: copper dividers, called baffles, set between the injector holes to break the injector face into compartments and disrupt the resonance. To prove it, engineers set off a small bomb inside a running F-1 to drive it deliberately unstable and watched the baffles damp the oscillation out. Between 1967 and 1973, 65 F-1 engines flew 13 Saturn V rockets with no combustion instability at all.^[15] Later designers attacked the same problem in the injector itself; SpaceX says it chose the Merlin's pintle injector for its inherent combustion stability.^[24]

Nozzles and expansion ratio#

A rocket nozzle converts heat and pressure into directed velocity. Rockets use a converging-diverging shape: the flow accelerates through a narrow throat, where it chokes at exactly Mach 1, and the throat area therefore sets the mass flow rate for the whole engine. Downstream of the throat the passage widens again, and the ratio of exit area to throat area, called the expansion ratio or area ratio, determines how far the supersonic flow expands, which in turn fixes the exit velocity, temperature and pressure.^[10]

Because the thrust equation contains the difference between exit pressure and ambient pressure, a nozzle is matched to the outside air at only one altitude, and NASA states the design constraint plainly: for a booster engine, the area ratio is limited by atmospheric pressure.^[9] That is why first-stage engines have short bells and modest area ratios, while an engine that never sees the atmosphere is free to carry an enormous one.

The penalty for cutting a nozzle short is easy to quantify. In NASA's design work on the J-2X upper-stage engine, dropping the area ratio from an optimized 92 to 1 down to 59 to 1 cost 13 seconds of guaranteed minimum vacuum specific impulse, 448 down to 435, and about 9,000 pounds of vacuum thrust.^[9] The Space Shuttle went the other way and gave the RS-25 an area ratio optimized for altitude rather than for the pad, accepting reduced sea-level performance because the solid boosters carried the vehicle through the thick air.^[9]

Steering and control#

A rocket is not simply pointed; it is balanced. NASA divides a launch vehicle into four systems, structure, payload, guidance, and propulsion, and gives the guidance system two jobs: keep the vehicle stable, and steer it.^[3][11] Every steering method works by generating a torque about the center of gravity.

Most modern rockets gimbal the engine. Swiveling the nozzle moves the thrust vector off the center of gravity and rotates the vehicle; the RS-25 gimbals through plus or minus 11 degrees.^[11][12] Older approaches survive in places. Early rockets such as the V-2 and the Redstone put small vanes directly in the exhaust stream to deflect it. The Atlas missile carried separate small vernier engines for control torque, an arrangement NASA says fell out of use because the extra propellant and plumbing cost too much weight. Movable fins at the rear still steer air-to-air missiles, where there is enough air for them to bite.^[11]

Orbit is sideways speed, not altitude#

Space is usually said to begin at the Karman line, 100 kilometers up, but altitude is the cheap part. Orbit is a speed. Newton illustrated it with a cannon on a mountain: fire a ball fast enough horizontally and its fall matches the curve of the Earth, so it falls forever without landing. The International Space Station does exactly this, moving sideways at roughly 7.7 kilometers per second some 400 kilometers up, which carries it around the planet about 16 times a day.^[26] Its crew float not because gravity is absent (it is about 90 percent of surface strength there) but because station and crew are falling together.

The comparison with suborbital flight shows the gap. Blue Origin's New Shepard crosses the 100-kilometer line at roughly 1 kilometer per second and falls back within minutes. An orbital vehicle needs nearly eight times that speed, and because kinetic energy grows with the square of velocity, that is around 60 times the energy per kilogram.

This is why orbital rockets dwarf suborbital ones, and why anything returning from orbit needs a heat shield: all that energy has to go somewhere on the way down. What happens after engine cutoff belongs to the field of orbital mechanics.

Max q and the ride through the atmosphere#

Aerodynamic force on a launch vehicle depends on dynamic pressure, which is half the air density multiplied by the square of the speed. Early in flight the vehicle is slow but the air is thick; later it is fast but the air is thin. The product peaks in between, roughly a minute into flight, and that peak is called max q. On Artemis I the timeline put maximum dynamic pressure at 1 minute 10 seconds after liftoff, about a minute before booster separation.^[22]

Vehicles are designed to survive that moment, and engines that can throttle are usually pulled back through it. The Space Shuttle's main engines were rated from 67 to 109 percent of rated power level, and the low end of that range was used twice per flight: once early in ascent when dynamic pressure peaked, and once near main engine cutoff to hold acceleration below 3 g for the crew. NASA's engineers described the requirement as minimizing structural loads on the vehicle early in flight and acceleration on the crew late in it.^[12] Dynamic pressure also shapes the trajectory before the vehicle ever flies: maximum dynamic pressure is one of the standard constraints an ascent trajectory is optimized against, along with maximum acceleration and maximum heating.^[21]

Gravity and drag losses#

Gravity and drag losses are the reason a launcher must deliver more than the 7.8 kilometers per second of orbital speed itself. ESA puts the velocity increment needed to reach low Earth orbit at about 9,000 meters per second once gravity and drag are accounted for.^[20] Trajectories that get no help from Earth's rotation cost more: one worked analysis of a polar launch totals 10.44 kilometers per second.^[21]

Gravity losses are the cost of the time spent thrusting upward against gravity before the trajectory bends over, and they are by far the larger of the two loss terms. Drag losses are the price of pushing through the lower atmosphere. That same polar-launch analysis splits its 10.44 kilometers per second into 7.67 for orbital velocity itself, 0.47 for the potential energy of the altitude gained, 2.2 for gravity loss, and only 0.1 for drag.^[21] The exact split depends on the trajectory and the vehicle, and gravity loss is the only large term that trajectory design can actually change, which is why launch vehicles fly a gravity turn, pitching over gently so that by staging most of the thrust is building horizontal speed rather than fighting gravity.^[21]

Earth itself refunds part of the bill. A pad near the equator is already moving at over 1,650 kilometers per hour relative to Earth's center, and that speed counts toward the roughly 28,000 kilometers per hour that orbit requires.^[2] It is why most missions launch toward the east over water, and why a high-inclination or polar orbit costs payload: the launch vehicle has to supply a much larger share of the orbital speed itself.^[2] The gap between what chemistry offers and what orbit demands stays narrow; that is why engineers chase every second of specific impulse.

References#

1. Ideal Rocket Equation - NASA Glenn Research Center.
2. Basics of Space Flight, Chapter 14: Launch - NASA Science.
3. Rocket Parts - NASA Glenn Research Center.
4. China's Landspace reaches orbit with methane-powered Zhuque-2 rocket - SpaceNews, July 12, 2023.
5. Rocket Thrust Equation - 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 (Space Launch System) RS-25 Core Stage Engine - NASA.
14. RS-25: The Clark Kent of Engines for the Space Launch System - NASA.
15. Solving Combustion Instability and Saving America's First Trips to the Moon - NASA, July 12, 2019.
16. A One-piece Liquid Rocket Thrust Chamber Assembly - NASA Technology Transfer Program.
17. 95 Years Ago: Goddard's First Liquid-Fueled Rocket - NASA.
18. Rocket History: 20th Century and Beyond - NASA Glenn Research Center.
19. The Space Shuttle - NASA.
20. Towards Reusable Launchers: A Widening Perspective - ESA Bulletin 87.
21. Launch Trajectories, AA284a Advanced Rocket Propulsion, Lecture 7 - M. Arif Karabeyoglu, Stanford University.
22. Artemis I Mission Timeline - NASA.
23. BE-4 Rocket Engine - United Launch Alliance fact sheet.
24. Falcon 9 Launch Vehicle Payload User's Guide - SpaceX.
25. Rocket Lab Celebrates 100th Rutherford Engine Build - SpaceNews.
26. International Space Station Facts and Figures - NASA.
27. China pushing for reusability milestone with Zhuque-3 launch and near-landing - NASASpaceFlight.com, December 2025.