Chemical rockets, such as those used on the Space Shuttle, are available to us today and have benefited from years of development. The U.S. Government, however, has done a great deal of research on Nuclear Powered Rockets over the years. Despite the research starting in the 1950s (when everything was better for being ‘atomic!’), it managed to survive into the 1970s – and was reborn (briefly) in the 1980s and early 1990s for SDI.

What is a nuclear rocket?

Any reaction drive engine which, as part of its motive energy, draws power from a static nuclear fuel. This could, in theory, be either fission or fusion based; however, at present, fission is much more feasible.

How would such a fission rocket (called a Nuclear Thermal Rocket or NTR) work?

Essentially, it’s a giant step backwards, all the way back to James Watt and the Steam Engine. Remember the two-spouted teakettle that spun when the water inside was boiled? Same thing, just…bigger. And, of course, more expensive. In a nuclear rocket, a propellant (usually hydrogen) is passed through the hot core of an operating nuclear reactor. It is heated by the warmth transferred to it from the fuel elements, and as a result, it expands (violently) and is expelled out the back of the rocket, just as it would have been had it been combusted. Instead of chemical oxidation, however, it is simply heated to high temperature (the higher the better) and released.

There have been at least two differing reactor technologies proposed for this task. The first was known as the Rover/NERVA variant. NERVA (which stands for Nuclear Engine for Rocket Vehicle Applications) was the subject of a whole line of test engines and reactors from 1955 up through around 1971, when work on nuclear spacecraft propulsion in the U.S. ceased.

A NERVA rocket is not much more than a solid-core Uranium-fueled reactor core, with many small passages through the core for hydrogen fuel to pass. The assembly would be surrounded with neutron reflectors to maintain the reaction. When in operation, the NERVA reactor would operate at a much higher power density than commercial reactors. It was designed as a limited-life expendable engine for both orbital transfer and interplanetary spaceflight applications; a lifespan of hours would have been fine, according to the design team. In fact, in 1972, Los Alamos National Laboratory designed a NERVA-based rocket for use in the Space Shuttle as the orbiter main engine. This design was projected to have a specific impulse of around 875 seconds, with a peak power output of approximately 72 kilonewtons, with a hydrogen fuel consumption of 8.5 kg/s and a lifetime of maximum 20 hours. Given that the existing Main Engines of the Space Shuttle have cycle times measured at around 15 minutes (the time it takes to reach MECO) this isn’t all that unrealistic.

Anyhow, back to history. The first testbeds were part of a project code-named KIWI. The first test, KIWI-A, was fired up in 1959. The reactor ran for several minutes at a power output of approximately 85 megawatts, but structural damage to the core from vibration and erosion required massive redesign. The KIWI tests continued as engineers struggled to cope with the fact that their desired core moderator (graphite) eroded quickly in the presence of hot hydrogen. Eventually, coated fuel elements and moderator rods diminished the erosion problem, but never eliminated it; it remained a ‘stopping block’ for the technology.

The tests culminated in 1968 with the activation of reactor Phoebus 2A . This was, according to project documents:

”…The most powerful nuclear reactor of any type ever constructed, with a design power level of 5,000 MW. Operations in June 1968 were limited to 4,000 MW due to premature overheating of aluminum segments of pressure vessel clamps. A total of 12.5 minutes of operations at temperatures of up to 2310 K included intermediate power level operations and reactor restart.”

The NERVA design goals were met in 1968, when prototype NERVA engine XE was constructed. The first to be downward-firing (others had fired up into the atmosphere, and concerns over contamination were high), it operated for a toal of 115 minutes at 1100 megawatts (MW). Apparently, tests could only last 10 minutes due to limited coolant water storage at the test stand. This was the last of the NERVA rockets, and was definitely a prototype; it was never installed into a vehicle.

In 1971, with public concern over nuclear power growing and clean-burning chemical alternatives for the Space Shuttle main engines available (plus the fact that an orbital infrastructure for weapons had failed to materialize due to treaty limitations) the entire nuclear propulsion concept was dropped.

It wasn’t to stay dead, however. In 1988 or thereabouts, it was reactivated with a new technology. The U.S. Air Force, instructed to build an orbital interceptor as part of the SDI program which needed high burst power engines and long shelf life, turned to fission. They included a new type of reactor, the Particle Bed Reactor (PBR) in the design, which would have raised the power output by several hundred percent while lightening the engine due to smaller shielding requirements and higher heat transfer allowing for smaller fuel elements, and thus lower cooling requirements as well. This project was named TIMBERWIND.

Timberwind never really progressed past the stage of a design study, and the papers were declassified in 1991 or so. One reason for this was the PBR design was never successfully finalized – the higher heat transfer in the core, while requiring a lower core temperature, also made the reactor dangerously easy to overrun. When that happened, the core swiftly achieved temperatures that would damage the lighter construction.

Debris contamination was another contributing factor. The holes in the particle bed (called the hot frit for reasons I cannot fathom) were extremely small in order to maximize surface area for heat transfer. However, any debris in the fuel flow would block these holes. This would not only starve the engine of thrust but would cause local disruptions of cooling (done by the hydrogen flow) on the reactor bed. This, in turn, would lead to more failures. As a result, any blockage would start a ‘chain reaction’ (much like the metal in a jet engine will contribute to its destruction if a front fan blade is broken) and the reactor core would become severely damaged.

The major advance of the PBR was its heat transfer. Due to using fuel elements shaped into small particles encased in shielding, rather than in large rods, the surface area of a cubic centimeter of reactor core fuel went from approximately 10 cm2 for the NERVA designs to as high as 95 cm2 in the TIMBERWIND reactors. The reasons for potential overrun of the reactor become clear when these numbers are compared.

The main problem with the NERVA design, fuel and moderator erosion, was never really conquered even in TIMBERWIND. As a scientist noted in a presentation in 1991:

"Corrosion was most pronounced in the mid-range region, about a third of the distance from the cold end of the fuel element. Fuel operating temperatures were lower here than the fabrication temperatures, hence thermal stresses were higher than at the hot end. Also, the neutron flux was highest in this region..."

"No fuel element geometry or fuel material ever totally solved the NERVA fuel element degradation problem. Mass loss of both uranium and carbon continued to limit service life by causing significant perturbation to core neutronics during the tests. Crack development in the fuel element coating was never completely eliminated.... Non-nuclear testing of coated fuel elements revealed an Arrhenius relationship between diffusion and temperature. For every 205 K increase in temperature (in the range 2400 to 2700 K), the mass loss increased by a factor of ten... resulting in loss of 20% of total uranium in approximately 5 hours of testing at 2870 K."

So, the outlook today for Nuclear spacecraft propulsion isn’t good, at least until/unless we can get a working fusion alternative. There were other proposed technologies for nuclear propulsion, including but not limited to:

And onward into the real twenty-first century we sail. Hopefully we’ll get one of these options, at least, working enough to play with before I’m dead.

Update 10-27-04
Thanks to spiregrain for pointing out the Pratt & Whitney TRITON engine to me! Apparently, our good thrust-obsessed friends at P&W have built a combination LANTR/EP engine. It is called TRITON to highlight its three operating modes: first, it can operate as a 'straight' NTR with an Isp of around 900 seconds, generating approx. 300-500 MW of power, in order to generate 15,000 to 30,000 lbs. of thrust with liquid hydrogen fuel. This would be its baseline spatial maneuvering mode. Second, it can operate in a 'higher power' mode as a LANTR engine by injecting LOX downstream of the exhaust, deriving a 200% increase in base thrust from the additional mass in the exhaust stream and the energy extracted from the combustion of the LOX and LH2. This would be used within planetary gravity wells, since it would dramatically increase fuel requirements (and hence bunkerage) but would not increase reactor output or radiation. Third, it contains helium/xenon in a closed loop running through the center of its fuel elements which acts as both an additional cooling mechanism and, when the reactor is idling, can be used in a closed Brayton cycle turbine generator system to produce electricity. If the reactor is run at approximately 1% load (100 kW), it can generate around 25 kW of usable electric power for spacecraft systems when not maneuvering. This from a system only about 4 or 5 meters in height. An exciting development! P&W claims to have solved the erosion problem completely by using tungsten cladding and a new element design, as well as by ensuring lower passthrough velocities for fuel/exhaust.


  1. NASA Jet Propulsion Laboratory presentation and materials on Advanced Propulsion Concepts, available at
  2. FAS articles on Timberwind and NERVA available at
  3. Horman, F.J., et al, "Particle Fuels Technology for Nuclear Thermal Propulsion," AIAA/NASA/OAI Conference on Advanced SEI Technologies, Cleveland, Ohio, 4-6 September 1991, Paper AIAA 91-3457. Referenced by FAS and retrieved for research.
  4. Aviation Week articles on NERVA and KIWI
  5. Bulletin of the Atomic Scientists for a small clarification on KIWI research
  6. Nuclear, "Pratt & Whitney nuclear rocket entry."

Log in or register to write something here or to contact authors.