Project Orion was the first engineering design study of a spacecraft powered by nuclear pulse propulsion, an idea proposed first by Stanisław Ulam during 1947. The project, initiated in 1958, envisioned the explosion of atomic bombs behind the craft and was led by Ted Taylor at General Atomics and physicist Freeman Dyson, who at Taylor's request took a year away from the Institute for Advanced Study in Princeton to work on the project.
By using energetic nuclear power, the Orion concept offered high thrust and high specific impulse (10 to 1,000 km/s) at the same time; the optimum combination for spacecraft propulsion. As a qualitative comparison, traditional chemical rockets (the Moon-class Saturn V or the Space Shuttle being prime examples) provide (rather) high thrust, but low specific impulse, whereas ion engines do the opposite. Orion would have offered performance greater than the most advanced conventional or nuclear rocket engines now being studied. Cheap interplanetary travel was the goal of the Orion Project. Its supporters felt that it had potential for space travel, but it lost political approval over concerns with fallout from its propulsion. The Partial Test Ban Treaty of 1963 is generally acknowledged to have ended the project.
During the late 1940s, Stanisław Ulam realized that nuclear explosions could not yet be realistically contained in a combustion chamber. Such a project did briefly exist, named Helios, but while its theoretical performance was similar to that of what would become the Orion, the lack of materials that could withstand the propulsion generating process meant that Helios never got beyond the drawing board.
Instead, the Orion design would have worked by dropping small shaped charge fission or thermonuclear explosives out the rear of a vehicle, detonating them 200 feet (60 m) out, and catching the blast with a thick steel or aluminum pusher plate.
Large multi-level high shock absorbers (pneumatic springs) were to have absorbed the impulse from the plasma wave as it hit the pusher plate, spreading the millisecond shock wave over several seconds and thus giving an acceptable acceleration speed. The long arm pistons proved one of the most difficult design features. Low pressure gas bags were also proposed as a primary shock absorber. The two sets of shock absorption systems were tuned to different frequencies to avoid resonance.
One aspect of the proposed vessel seems counter-intuitive today: because of the force involved in the thermonuclear detonations and the need to absorb the energy without harm, massive vessel designs were needed. Early designs had crew compartments and storage areas that were several stories tall, as opposed to contemporary chemical rockets whose height was almost all multi-stage fuel tanks with relatively little payload.
The 'base design' consisted of a 4000 ton model planned for ground launch from Jackass Flats, Nevada. Each 0.15 kt of TNT (600 MJ) (sea-level yield) blast would add 30 mph (50 km/h, 13 m/s) to the craft's velocity. A graphite based oil would be sprayed on the pusher plate before each explosion to prevent ablation of the surface. To reach low Earth orbit (300 mi), this sequence would have to be repeated about 800 times, like an atomic pogo stick.
Jerry Pournelle, who is acquainted with the project and its ex-team leader Freeman Dyson, has been quoted as saying that a single mission could have provided us with a large permanent moon base. Alternatively, an Orion could reach Pluto and return to Earth inside of a year. In the early 1960s a NASA study proposed a rocket launched hybrid for a Mars round trip.
In the 1954 Operation Castle nuclear test series at Bikini Atoll, a crucial experiment by Lew Allen proved that nuclear explosives could be used for propulsion. Two graphite-covered steel spheres were suspended near the test article for the Castle Bravo shot. After the explosion, they were found intact some distance away, proving that engineered structures could survive a nuclear fireball.
The Orion nuclear pulse drive combines a very high exhaust velocity, from 20 to 30 km/s, with meganewtons of thrust. Many spacecraft propulsion drives can achieve one of these or the other, but nuclear pulse rockets are the only proposed technology that could potentially deliver both (see spacecraft propulsion for more speculative systems). Specific impulse measures how much thrust can be derived from a given mass of fuel, and is the standard figure of merit for rocketry.
Since weight is no limitation, an Orion craft can be extremely robust. An unmanned craft could tolerate very large accelerations, perhaps 100 g. A human-crewed Orion, however, must use some sort of damping system behind the pusher plate to smooth the instantaneous acceleration to a level that humans can comfortably withstand – typically about 2 to 4 g.
The high performance depends on the high exhaust velocity, in order to maximize the rocket's force for a given mass of propellant. The velocity of the plasma debris is proportional to the square root of the change in the temperature (Tc) of the nuclear fireball. Since fireballs routinely achieve ten million degrees Celsius or more in less than a millisecond, they create very high velocities. However, a practical design must also limit the destructive radius of the fireball. The diameter of the nuclear fireball is proportional to the square root of the bomb's explosive yield.
The shape of the bomb's reaction mass is critical to efficiency. The original project designed bombs with a reaction mass made of tungsten. The bomb's geometry and materials focused the X-rays and plasma from the core of nuclear explosive to hit the reaction mass. In effect each bomb would be a nuclear shaped charge.
A bomb with a cylinder of reaction mass expands into a flat, disk-shaped wave of plasma when it explodes. A bomb with a disk-shaped reaction mass expands into a far more efficient cigar-shaped wave of plasma debris. The cigar shape focuses much of the plasma to impinge onto the pusher-plate.
For example, a 10 kiloton of TNT equivalent atomic explosion will achieve a plasma debris velocity of about 100 km/s, and the destructive plasma fireball is only about 100 meters in diameter. A 1 megaton TNT explosion will have a plasma debris velocity of about 10,000 km/s but the diameter of the plasma fireball will be about 1000 m..
The maximum effective specific impulse, Isp, of an Orion nuclear pulse drive generally is equal to:
where C0 is the collimation factor (what fraction of the explosion plasma debris will actually hit the impulse absorber plate when a pulse unit explodes), Ve is the nuclear pulse unit plasma debris velocity, and gn is the standard acceleration of gravity (9.81 m/s2; this factor is not necessary if Isp is measured in N·s/kg or m/s). A collimation factor of nearly 0.5 can be achieved by matching the diameter of the pusher plate to the diameter of the nuclear fireball created by the explosion of a nuclear pulse unit.
The smaller the bomb, the smaller each impulse will be, so the higher the rate of impulses and more than will be needed to achieve orbit. Smaller impulses also mean less g shock on the pusher plate and less need for damping to smooth out the acceleration.
The optimal Orion drive bomblet yield (for the human crewed 4,000 ton reference design) was calculated to be in the region of 0.15 KT, with approx 800 bombs needed to orbit and a bomb rate of approx 1 per second.
The following can be found in George Dyson's book  pg. 55 published in 2002. The figures for the comparison with Saturn V are taken from this section and converted from metric (kg) to US short tons.
|Ship mass||880 t||4,000 t||10,000 t||3,350 t|
|Ship diameter||25 m||40 m||56 m||10 m|
|Ship height||36 m||60 m||85 m||110 m|
|0.03 kt||0.14 kt||0.35 kt||n/a|
(to 300 mi Low Earth Orbit)
(to 300 mi LEO)
|300 t||1,600 t||6,100 t||130 t|
(to Moon soft landing)
|170 t||1,200 t||5,700 t||52 t|
(Mars orbit return)
|80 t||800 t||5,300 t||–|
(3yr Saturn return)
In late 1958 / early 1959 it was realized that the smallest practical vehicle would be determined by the smallest achievable bomb yield. The use of 0.03 KT (sea-level yield) bombs would give vehicle mass of 880 tons. However this was regarded as too small for anything other than an orbital test vehicle and the team soon focused on a 4,000 ton 'base design'.
At that time, the details of small bomb designs were shrouded in secrecy. Many Orion design reports had all details of bombs removed before release. Contrast the above details with the 1959 report by General Atomics which explored the parameters of three different sizes of hypothetical Orion spacecraft:
|Ship diameter||17–20 m||40 m||400 m|
|Ship mass||300 t||1000–2000 t||8,000,000 t|
|Number of bombs||540||1080||1080|
|Individual bomb mass||0.22 t||0.37–0.75 t||3.00 t|
The biggest design above is the "super" Orion design; at 8 million tons, it could easily be a city. In interviews, the designers contemplated the large ship as a possible interstellar ark. This extreme design could be built with materials and techniques that could be obtained in 1958 or were anticipated to be available shortly after. The practical upper limit is likely to be higher with modern materials.
Most of the three tons of each of the "super" Orion's propulsion units would be inert material such as polyethylene, or boron salts, used to transmit the force of the propulsion unit's detonation to the Orion's pusher plate, and absorb neutrons to minimize fallout. One design proposed by Freeman Dyson for the "Super Orion" called for the pusher plate to be composed primarily of uranium or a transuranic element so that upon reaching a nearby star system the plate could be converted to nuclear fuel.
The Orion nuclear pulse rocket design has extremely high performance. Orion nuclear pulse rockets using nuclear fission type pulse units were originally intended for use on interplanetary space flights.
Missions that were designed for an Orion vehicle in the original project included single stage (i.e., directly from Earth's surface) to Mars and back, and a trip to one of the moons of Saturn.
One possible modern mission for this near-term technology would be to deflect an asteroid that could collide with Earth. The extremely high performance would permit even a late launch to succeed, and the vehicle could effectively transfer a large amount of kinetic energy to the asteroid by simple impact. Also, an automated mission would eliminate the shock absorbers, the most problematic issue of the design.
Nuclear fission pulse unit powered Orions could provide fast and economical interplanetary transportation with useful human crewed payloads of several thousand tonnes.
Freeman Dyson performed the first analysis of what kinds of Orion missions were possible to reach Alpha Centauri, the nearest star system to the Sun . His 1968 paper "Interstellar Transport" (Physics Today, October 1968, p. 41–45) retained the concept of large nuclear explosions but Dyson moved away from the use of fission bombs and considered the use of one megaton deuterium fusion explosions instead. His conclusions were simple: the debris velocity of fusion explosions was probably in the 3000–30,000 km/s range and the reflecting geometry of Orion's hemispherical pusher plate would reduce that range to 750–15,000 km/s .
To estimate the upper and lower limits of what could be done using contemporary technology (in 1968), Dyson considered two starship designs. The more conservative energy limited pusher plate design simply had to absorb all the thermal energy of each impinging explosion (4×1015 joules, half of which would be absorbed by the pusher plate) without melting. Dyson estimated that if the exposed surface consisted of copper with a thickness of 1 mm, then the diameter and mass of the hemispherical pusher plate would have to be 20 kilometers and 5 million metric tons, respectively. 100 seconds would be required to allow the copper to radiatively cool before the next explosion. It would then take on the order of 1000 years for the energy-limited heat sink Orion design to reach Alpha Centauri.
In order to improve on this performance while reducing size and cost, Dyson also considered an alternative momentum limited pusher plate design where an ablation coating of the exposed surface is substituted to get rid of the excess heat. The limitation is then set by the capacity of shock absorbers to transfer momentum from the impulsively accelerated pusher plate to the smoothly accelerated vehicle. Dyson calculated that the properties of available materials limited the velocity transferred by each explosion to ~30 meters per second independent of the size and nature of the explosion. If the vehicle is to be accelerated at 1 Earth gravity (9.81 m/s) with this velocity transfer, then the pulse rate is one explosion every three seconds. The dimensions and performance of Dyson's vehicles are given in the table below
|Ship diameter (meters)||20,000 m||100 m|
|Mass of empty ship (metric tons)||10,000,000 t (incl.5,000,000 t copper hemisphere)||100,000 t (incl.50,000 t structure+payload)|
|+Number of bombs = total bomb mass (each 1MT bomb weighs 1 metric ton)||30,000,000||300,000|
|=Departure mass (metric tons)||40,000,000 t||400,000 t|
|Maximum velocity (kilometers per second)||1000 km/s (=0.33% of the speed of light)||10,000 km/s (=3.3% of the speed of light)|
|Mean acceleration (Earth gravities)||0.00003 g (accelerate for 100 years)||1 g (accelerate for 10 days)|
|Estimated cost||1 U.S. GNP (1968)||0.1 U.S. GNP|
Later studies indicate that the top cruise velocity that can theoretically be achieved by a thermonuclear Orion starship is about 8% to 10% of the speed of light (0.08-0.1c). An atomic (fission) Orion can achieve perhaps 3%-5% of the speed of light. A nuclear pulse drive starship powered by matter-antimatter pulse units would be theoretically capable of obtaining a velocity between 50% to 80% of the speed of light.
At 0.1c, Orion thermonuclear starships would require a flight time of at least 44 years to reach Alpha Centauri, not counting time needed to reach that speed (about 36 days at constant acceleration of 1g or 9.8 m/s2). At 0.1c, an Orion starship would require 100 years to travel 10 light years. The late astronomer Carl Sagan suggested that this would be an excellent use for current stockpiles of nuclear weapons.
A concept similar to Orion was designed by the British Interplanetary Society (B.I.S.) in the years 1973–1974. Project Daedalus was to be a robotic interstellar probe to Barnard's Star that would travel at 12% of the speed of light (0.12c). In 1989, a similar concept was studied by the U.S. Navy and NASA in Project Longshot. Both of these concepts require significant advances in fusion technology, and therefore cannot be built at present, unlike Orion.
From 1998 to the present, the nuclear engineering department at Pennsylvania State University has been developing two improved versions of the Daedalus design known as Project Ican and Project Aimstar.
The expense of the fissionable materials required was thought high, until the physicist Ted Taylor showed that with the right designs for explosives, the amount of fissionables used on launch was close to constant for every size of Orion from 2,000 tons to 8,000,000 tons. The larger bombs used more explosives to super-compress the fissionables, reducing fallout. The extra debris from the explosives also serves as additional propulsion mass.
Project Daedalus later proposed fusion explosives (deuterium or tritium pellets) detonated by electron beam inertial confinement. This is the same principle behind inertial confinement fusion. However, theoretically, it might be scaled down to far smaller explosions, and require small shock absorbers.
From 1957 until 1964 this information was used to design a spacecraft propulsion system called "Orion" in which nuclear explosives would be thrown behind a pusher-plate mounted on the bottom of a spacecraft and exploded. The shock wave and radiation from the detonation would impact against the underside of the pusher plate, giving it a powerful "kick". The pusher plate would be mounted on large two-stage shock absorbers which would transmit the acceleration to the rest of the spacecraft in a smooth manner.
During take-off, there were concerns of danger from fluidic shrapnel being reflected from the ground. One proposed solution was to use a flat plate of conventional explosives spread over the pusher plate, and detonate this to lift the ship from the ground before going nuclear. This would lift the ship far enough into the air that the first focused nuclear blast would not create debris capable of harming the ship.
A preliminary design for the explosives was produced. It used a shaped-charge fusion-boosted fission explosive. The explosive was wrapped in a beryllium oxide "channel filler", which was surrounded by a uranium radiation mirror. The mirror and channel filler were open ended, and in this open end a flat plate of tungsten propellant was placed. The whole thing was built into a can with a diameter no larger than 6 inches (15 cm) and weighed just over 300 lb (140 kg) so it could be handled by machinery scaled-up from a soft-drink vending machine (indeed, Coca-Cola was consulted on the design!).
At 1 microsecond after ignition, the gamma bomb plasma and neutrons would heat the channel filler, and be somewhat contained by the uranium shell. At 2–3 microseconds, the channel filler would transmit some of the energy to the propellant, which would vaporize. The flat plate of propellant would form a cigar-shaped explosion aimed at the pusher plate.
The plasma would cool to 14,000 °C, as it traversed the 25 m distance to the pusher plate, and then reheat to 67,000 °C, as (at about 300 microseconds) it hit the pusher plate and recompressed. This temperature emits ultraviolet, which is poorly transmitted through most plasmas. This helps keep the pusher plate cool. The cigar shaped distribution profile and low density of the plasma reduces the instantaneous shock to the pusher plate.
The pusher plate's thickness would decrease by about a factor of 6 from the center to the edge, so that the net velocity of the inner and outer parts of the plate are the same, even though the momentum transferred by the plasma increases from the center outwards.
At low altitudes where the surrounding air is dense, gamma scattering could potentially harm the crew and a radiation refuge would be necessary anyway on long missions to survive solar flares. Radiation shielding effectiveness increases exponentially with shield thickness (see gamma ray for a discussion of shielding), so on ships with mass greater than a thousand tons, the structural bulk of the ship, its stores, and the mass of the bombs and propellant would provide more than adequate shielding for the crew.
Numerous model flight tests (using conventional explosives) were conducted at Point Loma in 1959. On November 14, the one-meter model, called "Hot Rod" (or "putt-putt"), first flew using RDX (chemical explosives) in a controlled flight for 23 seconds to a height of 56 meters. Film of the tests has been transcribed to video shown on the BBC TV program "To Mars by A-Bomb" in 2003 with comments by Freeman Dyson and Arthur C. Clarke. The model landed by parachute undamaged and is in the collection of the Smithsonian National Air and Space Museum.
The first proposed shock absorber was merely a ring-shaped airbag. However, it was soon realized that, should an explosion fail, the 500 to 1000 ton pusher plate would tear away the airbag on the rebound. So a two-stage, detuned spring/piston shock absorber design was developed. On the reference design, the first stage mechanical absorber was tuned 4.5 times the pulse frequency whilst the second stage gas piston was tuned to 1/2 times the pulse frequency. This permitted timing tolerances of 10 ms in each explosion.
The final design coped with bomb failure by overshooting and rebounding into a 'center' position. Thus, following a failure (and on initial ground launch) it would be necessary to start (or restart) the sequence with a lower yield device. In the 1950s methods of adjusting bomb yield were in their infancy and considerable thought was given to providing a means of 'swapping out' a standard yield bomb for a smaller yield one in a 2 or 3 second time frame (or to provide an alternative means of firing low yield bombs). These days the yield of a standard device would be 'tuned down', as needed, 'on the fly'.
The bombs had to be launched behind the pusher plate fast enough to explode 20 to 30 m beyond it every 1.1 seconds or so. Numerous proposals were investigated, from multiple guns poking over the edge of the pusher plate to rocket propelled bombs launched from 'roller coaster' tracks .. however the final reference design used a simple gas gun to shoot the devices through a hole in the center of the pusher plate.
Exposure to repeated nuclear blasts raises the problem of ablation (erosion) of the pusher plate. However, calculations and experiments indicate that a steel pusher plate would ablate less than 1 mm if unprotected. If sprayed with an oil, it need not ablate at all (this was discovered by accident; a test plate had oily fingerprints on it, and the fingerprints suffered no ablation). The absorption spectra of carbon and hydrogen minimize heating. The design temperature of the shockwave, 67,000 °C, emits ultraviolet. Most materials and elements are opaque to ultraviolet, especially at the 340 MPa pressures the plate experiences. This prevents the plate from melting or ablating.
One issue that remained unresolved at the conclusion of the project was whether the turbulence created by the combination of the propellant and ablated pusher plate would dramatically increase the total ablation of the pusher plate. According to Freeman Dyson, during the 1960s they would have had to actually perform a test with a real nuclear explosive to determine this; with modern simulation technology, this could be determined fairly accurately without such empirical investigation.
Another potential problem with the pusher plate is that of spalling—shards of metal—potentially flying off the top of the plate. The shockwave from the impacting plasma on the bottom of the plate passes through the plate and reaches the top surface. At that point spalling may occur, damaging the pusher plate. For that reason, alternative substances (e.g., plywood and fiberglass) were investigated for the surface layer of the pusher plate, and thought to be acceptable.
If the conventional explosives in the nuclear bomb detonate, but a nuclear explosion does not ignite (a dud), shrapnel could strike and potentially critically damage the pusher plate.
True engineering tests of the vehicle systems were said to be impossible because several thousand nuclear explosions could not be performed in any one place. However, experiments were designed to test pusher plates in nuclear fireballs. Long-term tests of pusher plates could occur in space. Several of these tests almost flew. The shock-absorber designs could be tested at full-scale on Earth using chemical explosives.
But the main unsolved problem for a launch from the surface of the Earth was thought to be nuclear fallout. Any explosions within the magnetosphere would carry fissionables back to earth unless the spaceship were launched from a polar region such as a barge in the higher regions of the Arctic, with the initial launching explosion to be a large mass of conventional high explosive only to significantly reduce fallout; subsequent detonations would be in the air and therefore much cleaner. Antarctica is not viable, as this would require enormous legal changes as the continent is presently an international wildlife preserve. Freeman Dyson, group leader on the project, estimated back in the 1960s that with conventional nuclear weapons, each launch would cause on average between 0.1 and 1 fatal cancers from the fallout. Danger to human life was not a reason given for shelving the project – those included lack of mission requirement (no-one in the US Government could think of any reason to put thousands of tons of payload into orbit), the decision to focus on rockets (for the Moon mission) and, ultimately, the signature of the Partial Test Ban Treaty in 1963. The danger to electronic systems on the ground (from electromagnetic pulse) is insignificant from the sub-kiloton blasts proposed.
Orion-style nuclear pulse rockets can be launched from above the magnetosphere so that charged ions of fallout in its exhaust plasma are not trapped by the Earth's magnetic field and are not returned to Earth.
With special designs of the nuclear explosive, Ted Taylor estimated that it could be reduced tenfold, or even to zero if a pure fusion explosive could be constructed; however, a pure fusion explosive has yet to be successfully developed.
The vehicle and its test program would violate the Partial Test Ban Treaty of 1963 as currently written, which prohibited all nuclear detonations except those which were conducted underground, both as an attempt to slow the arms race and to limit the amount of radiation in the atmosphere caused by nuclear detonations. There was an effort by the US government to put an exception into the 1963 treaty to allow for the use of nuclear propulsion for spaceflight, but Soviet fears about military applications kept the exception out of the treaty.
One way around the restrictions of the treaty would be to use a form of the Project Daedalus fusion microexplosion rocket. Daedalus class systems use pellets of one gram or less ignited by particle or laser beams to produce very small fusion explosions with a maximum explosive yield of only 10–20 tons of TNT equivalent.
The launch of such an Orion nuclear bomb rocket from the ground or from low Earth orbit would generate an electromagnetic pulse that could cause significant damage to computers and satellites, as well as flooding the van Allen belts with high-energy radiation. This problem might be solved by launching from very remote areas, because the EMP footprint would be only a few hundred miles wide. The Earth is well shielded by the Van Allen belts. In addition, a few relatively small space-based electrodynamic tethers could be deployed to quickly eject the energetic particles from the capture angles of the Van Allen belts.
Assembling a pulse drive spacecraft in orbit by more conventional means and only activating its main drive at a safer distance would be a less destructive approach. The space elevator hypothetically provides an excellent solution, although existing carbon nanotubes composites do not yet have sufficient tensile strength. All chemical rocket designs are extremely inefficient (and expensive) when launching mass into orbit, however could be employed if the result was seen as worth the cost (for example, the alternative being the impact of an asteroid of size similar to that of the Cretaceous-Tertiary extinction event).
Adverse public reaction to any use of nuclear explosives of any type is likely to remain a stumbling block even if practical and legal difficulties are overcome.
A test similar to the test of a pusher plate occurred as an accidental side effect of a nuclear containment test called "Pascal B" conducted on 27 August 1957. The test's experimental designer Dr. Brownlee performed a highly approximate calculation that suggested that the low-yield nuclear explosive would accelerate the massive (900 kg) steel capping plate to six times escape velocity. The plate was never found, and Dr. Brownlee believes that the plate never left the atmosphere (for example it could have been vaporized by compression heating of the atmosphere due to its high speed). The calculated velocity was sufficiently interesting that the crew trained a high-speed camera on the plate, which unfortunately only appeared in one frame, but this nevertheless gave a very high lower bound for the speed.