Nuclear Thermal

Basically a Nuclear Thermal Rockets (NTR) is a nuclear reactor where the propellant is the coolant. And instead of the coolant being directed into a cooling tower, it instead exits out the exhaust nozzle, creating thrust.

They use the heat generated from a nuclear reaction to heat up propellant. The nuclear reaction is controlled by adjusting the amount of free neutrons inside the mass of fissioning material (like all nuclear reactors do, generally with reactor control drums).

As a side effect, if you have a cluster of several such engines it is vitally important to have distance and neutron isolation shields between them. Otherwise the nuclear reaction in each engine will flare out of control due to the neutron flux from its neighbor engines.

The fact that the propellant is also the coolant means that after a thrusting period is over, you still have to vent propellant through the reactor after you turn it off. Until the reactor goes cold.

Exhaust Velocity Limits on Nuclear Thermal Rockets

The exhaust velocity and specific impulse of NTR are proportional to the thermal levels inside the reactor. Which a fancy way to say "the hotter the reactor, the higher the exhaust velocity."

Which brings us to the exhaust velocity limit. Solid core NTRs use a solid-core nuclear reactor. Such reactors are made of matter. And as with all matter, if you raise the temperature, at some point it will get hot enough so that the reactor melts. Which means the core ain't solid any more. This is a bad thing, technical term is Nuclear meltdown, non-technical term is The China Syndrome. The molten remains of the reactor shoots out the exhaust bell like a radioactive bat from hell, killing anybody nearby and leaving the spacecraft without an engine.

To avoid this unhappy state of affairs, solid core NTRs are limited to a temperature of about 2,750 K (4,490° F), which translates into an exhaust velocity limit of about 8,093 m/s (with liquid hydrogen, double that if you've manage to figure out how to stablize monoatomic hydrogen). Some fancy high temperature designs can push that up to an exhaust velocity of about 11,800 m/s.

Lateral thinking rocket engineers had the brainstorm of "what if the reactor starts out molten in the first place?" This lead to the design of liquid-core NTR, with a temperature of 5,250 K and an exhaust velocity of 16,000 m/s.

Because rocket engineers can't resist turning it up to 11, they figured if liquid is good then gaseous should be even better. This is the open-cycle gas-core NTR, with an exhaust velocity of a whopping 34,000 m/s.

The major draw-back of open-cycle GCNTR is that there is no feasible to prevent any of the radioactive fission products and unburnt uranium from escaping out the exhaust. Which more or less makes the exhaust plume a weapon of mass destruction, and significatly increases the radiation exposure on the poor ship's crew.

Engineers tried to fix the radiation problem of the open-cycle GCNTR by making it closed-cycle; that is, preventing physical contact between the gaseous uranium and the propellant. This turned out to be an attempt to have your cake and eat it too. The entire point of gas core was to allow outrageous engine temperatures by not having any solid components inside the engine, but sadly baffles that prevent the uranium from mixing with the propellant are solid components. They managed an makeshift solution, but the price was the exhaust velocity was cut in half.

Solid Core

Solid Core NTR
3200° K
Exhaust velocity (H1)16,000? m/s
Exhaust velocity (H2)8,093 m/s
Exhaust velocity (CH4)6,318 m/s
Exhaust velocity (NH3)5,101 m/s
Exhaust velocity (H2O)4,042 m/s
Exhaust velocity (CO2)3,306 m/s
Exhaust velocity (CO or N2)2,649 m/s

Nuclear thermal rocket / solid core fission. It's a real simple concept. Put a nuclear reactor on top of an exhaust nozzle. Instead of running water through the reactor and into a generator, run hydrogen through it and into the nozzle. By diverting the hydrogen to a turbine generator 60 megawatts can be generated. The reactor elements have to be durable, since erosion will contaminate the exhaust with fissionable materials. The exhaust velocity limit is fixed by the melting point of the reactor.

Solid core nuclear thermal rockets have a nominal core temperature of 2,750 K (4,490° F).

Thrust is directly related to the thermal power of the reactor. Thermal power of 450 MWth with a specific impulse of 900 seconds will produce approximately 100,000 Newtons of thrust.

Specific impulse (and exhaust velocity) is directly related to exhaust temperature. A temperature of 2,300 to 3,100 K will produce approximately a specific impulse of 830 to 1,000 seconds.

As a general rule, solid core NTR have superior exhaust velocity over chemical rockets because of the low molecular weight of hydrogen propellant. Chemical LH2/LOX rockets actually run hotter than solid core NTRs, but the propellant has a much higher molecular weight.

Hydrogen gives the best exhaust velocity, but the other propellants are given in the table since a spacecraft may be forced to re-fuel on whatever working fluids are available locally (what Jerry Pournelle calls "Wilderness re-fuelling", Robert Zubrin calls "In-situ Resource Utilization", and I call "the enlisted men get to go out and shovel whatever they can find into the propellant tanks"). For thermal drives in general, and NTR-SOLID in particular, the exhaust velocity imparted to a particular propellant by a given temperature is proportional to 1 / sqrt( molar mass of propellant chemical ).

The value for "hydrogen" in the table is for molecular hydrogen, i.e., H2. Atomic hydrogen would be even better, but unfortunately it tends to explode at the clank of a falling dust speck (Heinlein calls atomic hydrogen "Single-H"). Another reason to avoid hydrogen is the difficulty of storing the blasted stuff, and its annoyingly low density (Ammonia is about eight times as dense!). Exhaust velocities are listed for a realistically attainable core temperature of 3200 degrees K for the propellants Hydrogen (H2), Methane (CH4), Ammonia (NH3), Water (H2O), Carbon Dioxide (CO2), Carbon Monoxide (CO), and Nitrogen (N2).

The exhaust velocities are larger than what one would expect given the molecular weight of the propellants because in the intense heat they break down into their components. Ammonia is nice because it breaks down into gases (Hydrogen and Nitrogen). Methane is nasty because it breaks down into Hydrogen and Carbon, the latter tends to clog the reactor with soot deposits. Water is most unhelpful since it doesn't break down much at all.

Dr. John Schilling figures that as an order of magnitude guess, about one day of full power operation would result in enough fuel burnup to require reprocessing of the fissionable fuel elements. (meaning that while there is still plenty of fissionables in the fuel rod, enough by-products have accumulated that the clogged rod produces less and less energy) A reprocessing plant could recover 55-95% of the fuel. With reprocessing, in the long term each totally consumed kilogram of plutonium or highly enriched uranium (HEU) will yield ~1E10 newton-seconds of impulse at a specific impulse of ~1000 seconds.

Dr. Schilling also warns that there is a minimum amount of fissionable material for a viable reactor. Figure a minimum of 50 kilograms of HEU.

One problem with solid-core NTRs is that if the propellant is corrosive, that is, if it is oxidizing or reducing, heating it up to three thousand degrees is just going to make it more reactive. Without a protective coating, the propellant will start corroding away the interior of the reactor, which will make for some real excitement when it starts dissolving the radioactive fuel rods. What's worse, a protective coating against an oxidizing chemical is worthless against a reducing chemical, which will put a crimp in your wilderness refueling. And trying to protect against both is an engineering nightmare. Oxidizing propellants include oxygen, water, and carbon dioxide, while reducing propellants include hydrogen, ammonia, and methane. Carbon Monoxide is neither, as the carbon atom has a death-grip on the oxygen atom.

Keep in mind that the oxidizing/reducing effect is only a problem with solid-core NTRs, not the other kinds. This is because only the solid-core NTRs have solid reactor elements exposed to the propellant (for heating).

As of the year 2019, solid core NTR development has switch focus to using Low Enriched Uranium (LEU) fuel. Mostly because the powers-that-be are hysterically afraid of Highly Enriched Uranium (aka "Weapons-Grade") falling into the Wrong Hands. HEU drastically increases the development costs and security regulations.


(ed note: most solid-core NTR rockets are designed to use liquid hydrogen, because that gives the best specific impulse. But in some cases, other propellants will have advantages despite having a lower specific impulse)


  • Water, ammonia, and carbon oxides have been found on the Moon and our current technology has the capability to extract and process these resources This will make reusable space transportation vehicles possible
  • Nuclear thermal propulsion ( engines can theoretically use any fluid as a propellant provided that significant core material degradation does not occur
  • Analysis is currently underway to understand how water and ammonia impact NTP engine performance
  • Water/ammonia NTP engine performance will help determine mission architectures that could be more beneficial than both chemical and hydrogen NTP

Challenges with Hydrogen

  • Hydrogen can provide a high specific impulse (Isp) of ~900 seconds in NTP engines.
  • Hydrogen has some unfavorable properties:
    • Low Boiling Temperature (20 K)
    • Low liquid density (7.1% to 8.6% that of water)
    • High specific heat capacity (3.5 times greater than that of water)

A Simpler Vehicle Architecture

  • Denser and non cryogenic propellant will reduce tank size and dry mass
  • Wider liquid phase temperature range does not require precise thermal management
  • Lower specific heat capacity will decrease the reactor power level given a thrust level or increase the thrust level given a reactor power level
  • This type of propellant
    • Yields lower Isp
    • Increases initial vehicle wetted mass resulting in longer burn times
    • If reusability is considered, partially tanked stages/vehicles can be sent from Earth
    • Decreases dry mass and thus cost

Energy Limited Propulsion Systems 1

  • Encompasses both NTP and electrical propulsion
  • The purpose of any transportation system is to get something from point A to point B
  • The Ideal Rocket Equation provides information about a single mission and focuses on propellant mass
  • Assuming propellant availability on the Moon and spacecraft reusability, a different parameter should be used
  • In electrical and NTP engines, the propulsion system introduces energy into the propellant rather than chemical engines where the energy is inside the propellant
  • This energy can be adjusted to meet mission parameters
  • Electrical propulsion is power limited (limited by wattage) while NTP is energy limited (limited by joules)

Energy Limited Propulsion Systems 2

  • Etot = ½mpropVe2 = ½mprop(Ispg0)2

  • Specific energy Esp = Etot / mdry

    • PER TRIP

    • In chemical propulsion, Esp = Etot / mprop

    • Scales with mdry in electrical propulsion systems

  • In NTP systems, 1 kg of 235U will produce 8.4 TJ of energy. Therefore, the total energy OF THE VEHICLE depends upon the total mass of useable 235U inside the NTR core.

  • can be divided by the number of desired flights to yield Etot while mdry remains constant.

  • In the NASA-AR HALEU NTP engine, based on BWXT’s reactor parameters, the total usable 235U mass will yield of about 30 TJ
  • By replacing mprop with (2Etot) / (Ispg0)2 from kinetic energy, the Ideal Rocket Energy Equation is obtained:

Energy Limited Propulsion Systems 3

The Ispopt concept can be explained with a Mars Transfer Vehicle for a conjunction class mission that has a ΔV of 4200 m/s and Etot of 3 TJ (1 TJ per engine).

  • Hydrogen NTP
  • Isp of 875 seconds
  • 4 stages
  • 5 SLS aggregation launches
  • Will be able to perform 1st mission after aggregation
  • 6 Lunar launches for reusability
  • Requires propellant grade water and electrolysis plants

  • Ammonia NTP
  • Isp of 360 seconds
  • 2 stages
  • 3 SLS aggregation launches
  • Requires additional 12 Lunar launches to perform 1st mission
  • 15 Lunar launches for reusability
  • Requires an ammonia scrubber which is already in the considered architecture.

This study will address the question of, "Is trading no electrolysis plants for more Lunar launches advantageous?"

Operation Time Limited Propulsion Systems

  • Total operational time is a function of both chemistry and temperature

  • is the hard limit inherent to NTP engines and another limit is the maximum design power

  • will provide an operational time of over 15.7 hours

  • Ammonia has not shown any significant adverse reactions with the baseline Tungsten and ZrC coating

  • Water is highly reactive at high temperatures, so Silicon Carbide (SiC) will need to be used for fuel element coatings

  • At NTP flow rates and pressures, this coating will only survive a single flight

  • Bringing Ts down will bring Isp down but increasing thrust up to the limit could bring the and limits closer together

Current NTP Engine

  • A current NTP engine design under consideration by NASA is from Aerojet Rocketdyne (AR). This is a low enriched uranium (LEU), hydrogen propellant, expander cycle design

  • Due to hydrogen’s small liquid temperature range, a boost pump with a boost turbine is required to avoid cavitation

  • This design works with supercritical hydrogen starting from State 3

  • The baseline design is the RL-10 with a thrust of 25,000 lbf

  • The NASA AR LEU NTP engine features

    • Reactor power of 550 MWt

    • Isp of 893 seconds

    • Mass flow rate of 12.9 kg/s

    • Area ratio of 386:1

Engine Models

Engine architectures consider:

  • Insights from NASA’s NTP Engine System/FMDP Conceptual TRD, of which BWXT and AR are providing support for engine, reactor and fuel design and analysis.

  • Both expander and bleed cycles

  • Engine transients

  • Maximum fuel temperature of 2850 K

  • Thrust levels:
    • 25 klbf
    • 15 klbf
    • at max

  • Line pressure losses

Engine Model (Ammonia)

Ammonia Expander Cycle Model Graphs

Water Bleed Cycle Model Graphs

Water NTP Results

For a given reactor power level using water:

  • About 2X the thrust of hydrogen-NTP can be achieved.

  • Isp values range between:
    • 250 seconds for sustainable cladding surface temperature (1400 K) and bleed cycle
    • 315 seconds for maximum cladding surface temperature (2400 K) and expander cycle

  • Multistage pump systems (up to 7 stages to prevent cavitation) for expander cycles are required depending on thrust and surface temperature criteria. This results in pressures as high as 700 atm and a bleed cycle would be more feasible.

  • Phase change will occur during engine transients and bleed cycle operation.

Ammonia NTP Results

For a given reactor power level using ammonia:

  • About 2X the thrust of hydrogen NTP can be achieved

  • Isp of
    • 345 seconds for bleed cycle
    • 360 seconds for expander cycle

  • Multistage pump systems (up to 4 stages to prevent cavitation) for expander cycles are required depending on thrust criteria This results in pressures of 270 atm for expander cycles

  • Phase change will occur during engine transients and bleed cycle operation

  • Ammonia was found to yield a more feasible architecture with higher performing Isp

  • Detailed results are currently being extracted from the models

(ed note: TL;DR: solid-core nuclear-thermal-rocket single-stage-to-orbit is pointless with a thrust-to-weight ratio under 20)

A few months ago, I spent some time describing some calculations of payload fraction that I derived to assist in the design of rocket vehicles. My motivation for getting into this type of work came about from my work on the X-33 rocket when I was an intern at the Skunk Works. I wondered how so many people could think that SSTO (single-stage-to-orbit) was a good idea when the mathematics argued against it.

Right after I joined NASA, in early 2000, I was in a group that was looking at some really advanced concepts, and somehow or another, we got looking at using (solid core) nuclear thermal rockets for an SSTO vehicle. At first blush, the whole idea seems to make sense. Nuclear thermal rockets offer almost twice the specific impulse (Isp) of chemical rockets, and if an SSTO doesn’t have enough Isp with chemical rockets, then surely nuclear rockets must be better, right?

Wrong. Super wrong.

Nuclear-thermal SSTO turns out to be one of the worst ideas anyone has ever come up with, for two simple reasons: hydrogen and the lousy thrust-to-weight ratio of nuclear thermal rockets. Those are the same two reasons that make NTR lousy or marginal for nearly any other space application as well, but this post will focus on the issues surrounding NTR SSTO.

In the case of any earth-to-orbit vehicle, you’ve got to have the thrust to get off the ground in the first place. Let’s assume that we’re dealing with a vertically-launched NTR SSTO. It has to have a vehicle thrust-to-weight ratio greater than one, and probably a fair bit better than that in the first place, just to get off the ground. So we can take those expressions that I derived before, assuming hydrogen as a propellant and the engine thrust-to-weight ratios that have been quoted by NTR proponents like Stan Borowski to quickly try to figure a payload fraction for an NTR SSTO.

We find the propellant-mass-sensitive term λ (derivation here) assuming the liquid hydrogen has a density of 71 kg/m3, ullage of 3%, a mixture ratio of zero, and a tank structural mass factor of 10 kg/m3. This gives us a value of 0.1452 for this term.

We find the gross-mass-sensitive term φ (derivation here) by assuming that the engine has a vacuum thrust of 15000 lbf, a weight of 5000 lbm, and vacuum thrust-to-weight of 3 to 1. I’m not even going to “ding” the engine for sea-level performance, since as we’ll see, it won’t even matter. With a vacuum T/W of 3 and the same for the sea-level T/W and an initial vehicle thrust-to-weight ratio of 1.25, and we’ll just say that the thrust structure doesn’t weigh anything either, the gross-mass-sensitive term comes out to be 0.4167.

We’ll also ignore any recovery hardware (wings, landing gear, TPS, etc) and say all that weighs nothing. We’ll assume that the engine has a vacuum Isp of 900 seconds and that it takes 9200 m/s of delta-V to get to orbit.

Plugging those numbers in the rocket equation gives us a mass ratio MR of 2.835 (very good!)

and a propellant mass fraction PMF of 64.73%.

Next we use the prop-mass-sensitive and gross-mass-sensitive terms, along with the propellant mass fraction to get the payload fraction mpayload/mgross (derivation here).

We start out with the final mass fraction (1 – prop mass fraction PMF) of 35.27%. It doesn’t get any better than that. Then we subtract the gross-mass-sensitive term φ (41.67%). Now we could stop right here, because we’re already negative (-0.064). That is to say, even before accounting for the issues with tankage, we’re already out of performance. The engines weigh too much. But we’ll keep going and subtract the product of the the propellant mass fraction PMF and prop-mass-sensitive term λ (0.6473*0.1452 = 0.0940) and we end up with a payload fraction of -0.1579.

So it’s a no-go with these engines. Our payload fraction is grossly negative and we’ve got nothing. It’s clear from the magnitude of the numbers that the engine thrust-to-weight ratio is the main culprit, although the “fluffy” liquid hydrogen tanks don’t help much either.

So what kind of engine performance would you have to have to get even a zero payload fraction? Well, I ran some rough calculations based on a variety of speculative vacuum T/W ratios for some putative NTR engine, at a few different values of specific impulse and plotted the results here:

The graph tells the story. To get payload fractions of zero (a launch vehicle of infinite size) you have to have a T/W at 900 sec Isp of over 10 (red circle). That’s more than three times the T/W that Stan Borowski projects for his sporty 15K NTR design, which he says will have a T/W of 3. So if you think that Stan or others can design an NTR that only weighs a third of what he thinks it will weigh, then you can dream about an NTR SSTO of infinite mass.

As for me, I’ve thought for some time that NTR was a really bad idea for almost every application for which it is considered. The SSTO application is probably the worst. (referring back to his earlier post, where he showed that for current NTR T/W ratios, and LH2 tank masses, NTRs only get a modest improvement in payload over a comparable LOX/LH2 system, but at much higher development cost.)

(ed note: Keeping in mind that a closed-cycle gas core NTR can have an Isp of 3,000 seconds or so.)


(ed note: 1I/'Oumuamua was that interstellar object that came streaking through the solar system in 2017. Everybody who was well read in science fiction novels were immediately reminded of Arthur C. Clarke's Rama. While the object is probably not an alien generation starship, you never know for sure until you examine the blasted thing. A pity it is traveling at 26.33 kilometers per second, we'll never catch it now.

Or can we?

The authors of the paper figured that using chemical rockets made about as much sense as using a arthritic tortoise to catch a cheetah. Nuclear thermal is more reasonable. They crunched the numbers for a variety of solid-core nuclear thermal rockets)


The first definite interstellar object observed in our solar system was discovered in October of 2017 and was subsequently designated 1I/’Oumuamua. In addition to its extrasolar origin, observations and analysis of this object indicate some unusual features which can only be explained by in-situ exploration. For this purpose, various spacecraft intercept missions have been proposed. Their propulsion schemes have been chemical, exploiting a Jupiter and Solar Oberth Maneuver (mission duration of 22 years) and also using Earth-based lasers to propel laser sails (1-2 years), both with launch dates in 2030. For the former, mission durations are quite prolonged and for the latter, the necessary laser infrastructure may not be in place by 2030. In this study Nuclear Thermal Propulsion (NTP) is examined which has yet to materialise as far as real missions are concerned, but due to its research and development in the NASA Rover/NERVA programs, actually has a higher TRL than laser propulsion. Various solid reactor core options are studied, using either engines directly derived from the NASA programs, or more advanced options, like a proposed particle bed NTP system. With specific impulses at least twice those of chemical rockets, NTP opens the opportunity for much higher ΔV budgets, allowing simpler and more direct, time-saving trajectories to be exploited. For example a spacecraft with an upgraded NERVA/Pewee-class NTP travelling along an Earth-Jupiter-1I trajectory, would reach 1I/’Oumuamua within 14 years of a launch in 2031. The payload mass to 1I/’Oumuamua would be around 2.5metric tonnes, but even larger masses and shorter mission durations can be achieved with some of the more advanced NTP options studied. In all 4 different proposed NTP systems and 5 different trajectory scenarios are examined.

2.2 NTR Propulsion Options

Four NTR options are considered here and are provided in Table 1.

Table 1 : NTR options with their performance values
NTR motorDescriptionRef.Mass
Isp (s)
Upgraded version of the Pewee studied
in the NASA NTR NERVA program
‘50s to early ‘70s
SNREBased on the Small Nuclear Rocket
Engine, studied in the Rover program
SLHCSquare Lattice Honeycomb[39]2500970147.5
SNTPParticle Bed Nuclear Thermal Rocket[40]800950196

It is assumed the s/c is transported to a LEO of 406km by a NASA Space Launch System (SLS) Block 2. It is currently envisaged that an SLS offers a 130metric tonne capability to LEO. We further assume that only one NTR motor is used, and LH2 propellant can be stored for significant durations with no-leakage and with a zero boil-off cryocooler [41], and further that there are 2 Staged LH2 tanks with an optimum mass ratio. Payload here is understood to mean the total mass of spacecraft after the engines and spent LH2 tanks have been jettisoned. This therefore represents the useful available spacecraft mass. Note howeve\a\r that this mass includes that of any heat shield which may be necessary if a Solar Oberth is involved.

We get Figure (1) for payload mass against ΔV budget.The ratio of dry stage mass to wet stage mass for both stages is assumed to be p=0.035 which was calculated using the same spacecraft/LH2 mass budgets provided in [41]and incorporates the tank insulation and cryocooler mass.

To generate Figure (2) for Ammonia (NH3) propellant a factor of 0.63 is applied to the specific impulses provided for LH2 above. The value of p for LH2 is retained. This is probably an overestimate as the requirements on storage/insulation would be less severe than for LH2, so this Figure (2) is probably a conservative estimate.

The data from Figures (1) & (2) will be used in Section (3.2) as input to generate flight times as a function of payload mass to 1I/‘Oumuamua.

2.3 Trajectory Scenarios

There are five scenarios considered here, which partly correspond to scenarios which have already been presented in the previous literature, but using chemical propulsion:

  1. Direct from Earth to 1I/’Oumuamua[28-30]
  2. From Earth to 1I using a Solar Oberth
  3. From Earth to 1I via Jupiter and Solar Oberth[29-31]
  4. From Earth to Jupiter to 1I using LH2
  5. Identical to (4) but using NH3

Figures (3), (4), (5) show examples of trajectory scenarios (2), (3), (4/5) respecitvely for illustration. Note that scenarios(1) & (2) have optimum trajectories every one Earth year, due to the position of the Earth with respect to 1I/‘Oumuamua.Scenarios (3), (4) & (5) have optimum trajectories approximately every Jupiter year, so around 12 Earth years, due to the alignment of Jupiter with ‘Oumuamua. Scenario (3) has optima in 2033, 2045, 2057 and so on. Scenarios (4) & (5) have launch optima in 2031, 2043, 2055 etc.

3. Results

3.1 Mission Flight Duration and ΔV Budgets for Different Trajectory Scenarios

The available ΔV for NTR is generally larger than that for chemical propulsion.

Figure (6) shows minimum flight duration againt launch date (based on yearly optima) for the years 2025 to 2045 and taking a direct trajectory (scenario (1) ). Four differentΔV budgets are allocated, 25km/s, 30km/s, 35km/s and 40km/s. The s/c is assumed to have already been placed in LEO of altitude 406km. Observe that for ΔV budgets of 25km/s and with launch years > 2035, direct trajectories from LEO to 1I have prohibitively long mission durations.

Minimum flight duration plots against Solar Oberth perihelion solar radial distance are shown for three different ΔV budgets for scenario (2) in Figure (7) and four ΔV budgets for scenario (3) in Figure (8). Scenarios (4) & (5) minimum flight duration against ΔV is shown in Figure (9). For Figures (6), (7), (8) & (9) it is assumed the s/c starts in an LEO of altitude 406km.

Even with the powerful Space Launch System (SLS) Block 2, it can be shown that for scenarios (1), (2) & (4/5), chemical propulsion cannot deliver ΔV’s of the magnitude needed for sensible flight durations but scenario (3) is achievable.

3.2 Payload Masses to ‘Oumuamua Achievable Using NTR

For direct trajectories, i.e. scenario (1), data from Figure (6) can be combined with Figure (1) to give Figure (10).

Data from Figures (7), (8) and (9) can be combined with Figure (1) to give Figures (11)-(13). Figure (11) is scenario (2) and note that the NERVA Pewee-class engine is excluded from this plot because it cannot achieve payload masses > 0kg since the required ΔV budgets are too high.

The results are summarised in Table 2. Regarding these Figures (10)-(13), it can be observed that there is a clear relationship between payload mass and flight time. Hence as one might expect, as flight time goes up so the ΔV reduces enabling higher payload masses (ref. Figure (1)). Generally of the NTR options, the NERVA Pewee-class NTR has the worst performance (in terms of longer flight times and lower payload masses) whereas the Particle Bed (SNTP) has the best performance. As also might be predicted, for those trajectories which employ a Solar Oberth Maneuver, the closer the Solar Oberth to the sun, the better the overall mission capability, though naturally the solar flux and consequent heat shield mass requirement increase.

Scenario (2) enables lower flight durations than scenario (1) but has lower payload mass capability.

Generally scenario (3) enables higher payload masses to 1I than scenarios (1) & (2) but it can also be seen that scenarios (1) & (2) trajectories have yearly optima as opposed to the 12-yearly optima for scenario (3). Furthermore, flight durations are lower for scenarios (1) & (2). Scenario (4) with LH2 tanks is shown in Figure (13) and offers better performance than scenario (3), but without a hazardous close approach to the sun. Scenario (5) with NH3 tanks is shown in Figure (14) and has lower payload masses compared to LH2 as would be expected.

Comparing Figure (13) (scenario (4) using a powered Jupiter GA with LH2 tanks) against Figure (10) (scenario (1), direct transfer), it appears that the former provides significantly higher payload masses, but this is only because the former has a generally lower ΔV mission profile. In fact for equivalent ΔV budgets, these two scenarios have the same payload mass but scenario (4) gives a significantly lower flight duration. So for instance, with ΔV= 25km/s, scenario (1) gives a duration of 37 years whereas scenario (4) with LH2 has a duration of 14 years. However there are two key disadvantages with scenario (4) and they are first that it involves a journey to Jupiter which requires the LH2 to be stored without significant leakage and with zero boil-off (so with a cryocooler) and second the launch optima are at twelve year intervals, between which trajectories are not viable.

Table 2 Results Summary
using Pewee-
class NTR
1Direct from
LEO to 1I
Every year2.9
2LEO to Solar
Oberth to 1I
Every year02.9
3LEO to
Jupiter to
Solar Oberth
to 1I
Every 12
years, 2033,
4LEO to
Jupiter to 1I
Every 12
years, 2031,
5LEO to
Jupiter to 1I
using NH3
Every 12
years, 2031,

4 Discussion

In this paper, we investigated the use of NTR for chasing interstellar objects, once they have left the inner solar system. We used the example of 1I/’Oumuamua for illustration.

We identified several advantages of using NTR for a mission to ‘Oumuamua and similar interstellar objects. First, due to the higher Isp, flight durations can be considerably reduced to < 15 years compared to > 20 years for chemical propulsion. The higher Isp also implies that generally the payload masses to 1I/’Oumaumua are considerably larger than for chemical propulsion. This translates to payload masses of 1000’s of kg as opposed to 100’s of kg for chemical.

In terms of the trajectories, direct trajectories are possible, which significantly reduce the complexity of missions. Direct trajectories also mean that the optimum launch windows arrive once a year, when the Earth and ‘Oumuamua are appropriately aligned. Second, in contrast to chemical propulsion, the arrival velocities are much lower approx. 18km/s, compared to 30km/s with chemical. Lower velocities allow for longer observation times during the encounter and thereby a higher science return. Trajectories without a Solar Oberth maneuver also have the advantage of a lower degree of uncertainty. One of the caveats of the Solar Oberth maneuver is that the errors or uncertainties in the burn at Perihelion have a disproportionate influence on the solar system escape trajectory. Also, the heat shield is not required, thereby saving mass. If a Solar Oberth is utilized, they can be farther from the sun (perihelia for the SO can be > 10Solar Radii) compared to chemical propulsion (where perihelia for the SO are typically< 10 Solar Radii), which reduce the requirements for the heat shield, as the solar irradiation per area diminishes with 1/r².

A major drawback of the NTR trajectories which employ a trip to Jupiter, is the high relative velocity of the s/c as it approaches 1I/’Oumuamua (from Table 2 around 40km/s).

However, though not studied, there is the potential for slowing down as the target is approached and even to perform a rendezvous, though this would be after a long flight duration and so contingent on nearly zero-leakage and zero boil-off LH2 tanks and cryocoolers.

The direct trajectory option (scenario (1)) would not require long LH2 storage durations and for reasons elucidated above is possibly one of the preferred options. However if mission duration is important, scenario (4), Earth-Jupiter-1I, is possibly the preferred choice.

One limitation of our study is that the payload is considered as a black box, and potential constraints from the spacecraft with its instrumentation are not taken into consideration. Such constraints might be related to compatibility issues between NTR and certain spacecraft instruments.

To summarize, our findings indicate that NTR for missions to interstellar objects would have a significant effect on the duration of a mission to such objects (trip times can be even cut in half) and allow for payload masses of an order of magnitude higher than for chemical propulsion. Hence, NTR would be a game changer for missions chasing interstellar objects, when they are on their way out of the solar system. Future work should explore the use of NTR for rendezvous and even sample return missions, which are feasible with NTR.

5 Conclusions

In this paper, we examined the use of Nuclear Thermal Propulsion (NTP) for missions to interstellar objects, exemplified by 1I/’Oumuamua. Four different proposed NTP options are analysed, ranging from NERVA-based designs to more advanced NTP. Using the OITS trajectory optimization tool, we find that NTP would allow for simpler and more direct, time-saving trajectories to 1I/’Oumuamua. Significant savings in terms of mission duration (14 years for a launch in 2031) are identified. Payload masses on the order of 1000s of kg, compared to 100s of kg using a Space Launch System launcher would be feasible. We conclude that NTP would be a game changer for chasing interstellar objects on their way out of the solar system, drastically reducing trip times and increasing payload masses. Future work should explore rendezvous missions using NTP as well as the feasibility of using NTP for reactive missions, where interstellar objects are discovered early.


Dangerous radiation. Overstuffed pantries. Cabin fever. NASA could sidestep many of the impediments to a Mars mission if they could just get there faster. But sluggish chemical rockets aren't cutting it — and to find what comes next, one group of engineers is rebooting research into an engine last fired in 1972.

The energy liberated by burning chemical fuel brought astronauts to the moon, but that rocket science makes for a long trip to Mars. And although search for a fission-based shortcut dates back to the 1950s, such engines have never flown. In August, NASA boosted those efforts when the agency announced an $18.8-million-dollar contract with nuclear company BWXT to design fuel and a reactor suitable for nuclear thermal propulsion (NTP), a rocket technology that could jumpstart a new era of space exploration.

"The strengths with NTP are the ability to do the very fast round trip [to Mars], the ability to abort even if you're 2 to 3 months into the missions, the overall architectural robustness, and also the growth potential to even more advanced systems," Michael Houts, principal investigator for the NTP project at NASA's Marshall Space Flight Center, told [Superfast Space Propulsion Concepts (Images)]

Superior gas mileage

NTP rockets would pull all that off by offering about twice the bang for the buck that chemical rockets do (exhaust velocity of 8,000 m/s as opposed to 4,400 m/s). Rather than burning fuel with oxygen, a nuclear fission reactor would serve as a powerful furnace, heating liquid hydrogen and expelling the resulting gas for thrust. How much oomph a rocket gets from its fuel depends largely on how fast it can hurl particles out the back, which in turn hinges on their mass. And NTP's single or double hydrogen atoms would be up to a dozen times lighter than chemical rocket outputs. 

That atomic bean counting could add up to significant time savings. "Nuclear thermal propulsion can enable you to get to Mars faster, on the order of twice as fast," said Vishal Patel, a researcher involved in subcontract work for BWXT at the Ultra Safe Nuclear Corp. in Los Alamos, New Mexico. "We're looking at nice 3- to 4-month transit times."

New tricks with an old technology

Unlike truly exotic propulsion proposals using antimatter or nuclear fusion, researchers have long considered nuclear fission rockets technologically feasible. Concrete development began with the Atomic Energy Commission's Project Rover in 1955 — three years before NASA's founding — and continued with the NERVA rocket prototype, which fired for nearly 2 hours straight during ground tests before budget cuts ended development in 1972. 

By then, NASA had already canceled Apollo 18 through 20, as well as Saturn V rocket production. When Mars plans followed suit, the multibillion-dollar NERVA project lost its main purpose, Houts said. The technology saw a brief revival in the late '80s and early '90s with the Space Nuclear Thermal Propulsion (SNTP) program, which also ran out of funding before flight testing.  

But now, with interest turning back toward Mars, past research is finding new life in current projects. 

"The key thing is, [the NERVA rocket] was extremely well documented," said John Helmey, project manager for BWXT's NTP project. "We aren't starting from scratch. We're building upon really good work that was done back in that time frame," he told Over the course of the contract, which extends through 2019, BWXT will develop conceptual designs focusing on fuel elements and the reactor core.

Three main challenges distinguish modern efforts from the legacy research.

Nuclear-testing rules have changed, said Jonathon Witter, BWXT NTP project chief engineer. The potential for trace levels of radioactivity in the engine exhaust means that engineers can no longer let clouds of hydrogen gas billow into the atmosphere. Instead, BWXT plans to test a trick developed at NASA's Stennis Space Center and combust the hydrogen gas with oxygen to make easy-to-catch water. Early, small-scale demonstrations will use non-nuclear hydrogen gas to test this exhaust-capturing method, but water from future nuclear tests could be decontaminated with off-the-shelf technology.

Engineers are also redesigning the fuel elements with new materials surrounding the uranium fuel particles, according to Witter. Rocket efficiency depends on temperature too, and BWXT expects that a ceramic and tungsten composite will allow for better operation at higher temperatures.

What's more, NERVA ran on 90 percent highly enriched uranium that would today qualify as weapons-grade. But because the fission process throws off more than enough heat, those levels are overkill, Patel said. BWXT's designs will harness material enriched to just below 20 percent (19.75%), putting it in the less-tightly regulated low enriched uranium (LEU) category. On top of allowing safer reactors, the modest levels of fissionable material could open the door to more public-private partnerships. (I'm sure that got Elon Musk's attention)

"The LEU thing really enables the idea that non-governmental entities can get in on this," Patel said. "It's potentially game changing."

But extensive design and many years of testing separate NTP's on-paper potential from single-season jaunts to the Red Planet, and nuclear space technology's history of false starts makes it a long shot for NASA's early Mars missions, currently scheduled for the 2030s. "It's one of several advanced propulsion options, Houts said. "There're a lot of good options that use chemical systems, and options that use electric propulsion."

Scott Hall, a developer of one such electric propulsion prototype that recently broke records at the University of Michigan, says he'd love to see any of these technologies get into space, but doesn't think it'll happen soon. 

"Optimistically, it will be 15 years," Hall said of his high-powered ion thruster, "and realistically it's probably more like 50… The process is just moving so slowly, and I imagine the nuclear guys are in a similar boat." 

But whether it takes one decade or 10, Houts thinks nuclear technology could transform space exploration. He cites Martian power plants and the possibility of spacecraft that refuel from naturally occurring resources like water or methane as examples of far-off possibilities. 

"What we're talking about is a first generation system. The systems beyond that could have extremely advanced capabilities," he said.

Nasty Methane Carbon Build-up

     Rip started to announce his name, rank, and the fact that he was reporting as ordered. Commander O’Brine brushed his words aside and stated flatly, “You’re a Planeteer. I don’t like Planeteers.”
     Rip didn’t know what to say, so he kept still. But sharp anger was rising inside of him.
     O’Brine went on, “Instructions say I’m to hand you your orders en-route. They don’t say when. I’ll decide that. Until I do decide, I have a job for you and your men. Do you know anything about nuclear physics?”
     Rip’s eyes narrowed. He said cautiously, “A little, sir.”
     “I’ll assume you know nothing. Foster, the designation SCN means Space Cruiser, Nuclear. This ship is powered by a nuclear reactor. In other words, an atomic pile. You’ve heard of one?”
     Rip controlled his voice, but his red hair stood on end with anger. O’Brine was being deliberately insulting. This was stuff any Planeteer recruit knew. “I’ve heard, sir.”
     “Fine. It’s more than I had expected. Well, Foster, a nuclear reactor produces heat. Great heat. We use that heat to turn a chemical called methane into its component parts. Methane is known as marsh gas, Foster. I wouldn’t expect a Planeteer to know that. It is composed of carbon and hydrogen. When We pump it into the heat coils of the reactor, it breaks down and creates a gas that burns and drives us through space. But that isn’t all it does.”
     Rip had an idea What was coming, and he didn’t like it. Nor did he like Commander O’Brine. It was not until much later that he learned that O’Brine had been on his way to Terra to see his family for the first time in four years when the cruiser’s orders were changed. To the commander, whose assignments had been made necessary by the needs of the Special Order Squadrons, it was too much. So he took his disappointment out on the nearest Planeteer, who happened to be Rip.
     “The gases go through tubes,” O’Brine went on. “A little nuclear material also leaks into the tubes. The tubes get coated With carbon, Foster. They also get coated with nuclear fuel. We use thorium. Thorium is radioactive. I won’t give you a lecture on radioactivity, Foster. But thorium mostly gives off the kind of radiation known as alpha particles. Alpha is not dangerous unless breathed or eaten. It won’t go through clothes or skin. But when mixed with fine carbon, thorium alpha contamination makes a mess. It’s a dirty mess, Foster. So dirty that I don’t want my spacemen to fool with it.

(ed note: now in a real solid-core NTR, nuclear fuel leaking from the reactor elements is a major malfunction)

     “I want you to take care of it instead,” O’Brine said. “You and your men. The deputy commander will assign you to a squadron. Settle in, then draw equipment from the supply room and get going. When I want to talk to you again, I’ll call for you. Now blast off, Lieutenant, and rake that radiation. Rake it clean.”
     Rip forced a bright and friendly smile. “Yes, sir,” he said sweetly. “We’ll rake it so clean you can see your face in it, sir.” He paused, then added politely, “If you don’t mind looking at your face, sir—to see how clean the tubes are, I mean.”
     Rip turned and got out of there.
     Koa was waiting in the passageway outside. Rip told him what had happened, mimicking O’Brine’s Irish accent.
     The sergeant-major shook his head sadly. “This is what I meant, Lieutenant. Cruisers don’t clean their tubes more’n once in ten accelerations. The commander is just thinking up dirty work for us to do, like I said.”
     “Never mind,” Rip told him. “Let’s find our squadron and get settled, then draw some protective clothing and equipment. We’ll clean his tubes for him. Our turn will come later.”
     He remembered the last thing Joe Barris had said, only a few hours before. “Joe was right,” he thought. “To ourselves we’re supermen, but to the spacemen we’re just simps.” Evidently O’Brine was the kind of space officer who ate Planeteers for breakfast.
     Rip thought of the way the commander had turned red with rage at that crack about his face, and resolved, “He may eat me for breakfast, but I’ll try to be a good, tough mouthful!”
     Commander O’Brine had not exaggerated. The residue of carbon and thorium on the blast tube walls was stubborn, dirty, and penetrating. It was caked on in a solid sheet, but when scraped, it broke up into fine powder.
     The Planeteers wore coveralls, gloves, and face masks with respirators, but that didn’t prevent the stuff from sifting through onto their bodies. Rip, who directed the work and kept track of the radiation with a gamma-beta ion chamber and an alpha proportional counter, knew they would have to undergo personal decontamination.

(ed note: in a real rocket, the tubes would be in vacuum, so the crew would need space suits. The tubes would also be close to the reactor. The reactor is not very radioactive if it is shut down, except for neutron activation.)

     He took a reading on the ion chamber. Only a few milliroentgens of beta and gamma radiation. That was the dangerous kind, because both beta particles and gamma rays could penetrate clothing and skin. But the Planeteers wouldn’t get enough of a dose to do any harm at all. The alpha count was high, but so long as they didn’t breathe any of the dust it was not dangerous.
     The Scorpius had six tubes. Rip divided the Planeteers into two squads, one under his direction and one under Koa’s. Each tube took a couple of hours’ hard Work. Several times during the cleaning the men would leave the tube and go into the main mixing chamber while the tube was blasted with live steam to throw the stuff they had scraped off out into space.
     Each squad was on its last tube when a spaceman arrived. He saluted Rip. “Sir, the safety officer says to secure the tubes.”
     That could mean only one thing: deceleration. Rip rounded up his men. “We’re finished. The safety officer passed the word to secure the tubes, which means we’re going to decelerate.” He smiled grimly. “You all know they gave us this job just out of pure love for the Planeteers. So remember it when you go through the control room to the decontamination chamber.”
     The Planeteers nodded enthusiastically.
     Rip led the way from the mixing chamber through the heavy safety door into the engine control room. His entrance was met with poorly concealed grins by the spacemen.
     Halfway across the room Rip turned suddenly and into Sergeant major Koa. Koa fell to the deck arms flailing for balance—but flailing against his protective clothing. The other Planeteers rushed to pick him up, and somehow all their arms and hands beat against each other.
     The protective clothing was saturated with fine dust. It rose from them in a choking cloud, was picked up, and dispersed by the ventilating system. It was contaminated dust. The automatic radiation safety equipment filled the ship with an earsplitting buzz of warning. Spacemen clapped emergency respirators to their faces and spoke unkindly of Rip’s Planeteers in the saltiest space language they could think of.
     Rip and his men picked up Koa and continued the march to the decontamination room, grinning under their respirators at the consternation around them. There was no danger to the spacemen since they had clapped on respirators the moment the warning sounded. But even a little contamination meant the whole ship had to be gone over with instruments, and the ventilating system would have to be cleaned.
     The deputy commander met Rip at the door of the radiation room. Above the respirator, his face looked furious.
     “Lieutenant,” he bellowed, “haven’t you any more sense than to bring contaminated clothing into the engine control room?”
     Rip was sorry the deputy commander couldn’t see him grinning under his respirator. He said innocently, “No, sir. I haven’t any more sense than that.”
     The deputy grated, “I’ll have you up before the Discipline Board for this.”
     Rip was enjoying himself thoroughly. “I don’t think so, sir. The regulations are very clear. They say, ‘It is the responsibility of the safety officer to insure compliance with all safety regulations both by complete instructions to personnel and personal supervision.’ Your safety officer didn’t instruct us and he didn’t supervise us. You better run him up before the Board.”
     The deputy commander made harsh sounds into his respirator. Rip had him, and he knew it. “He thought even a stupid Planeteer had sense enough to obey radiation safety rules,” he yelled.
     “He was wrong,” Rip said gently. Then, just to make himself perfectly clear, he added, “Commander O’Brine was within his rights when he made us rake radiation. But he forgot one thing. Planeteers know the regulations, too. Excuse me, sir. I have to get my men decontaminated.”
     Inside the decontamination chamber, the Planeteers took off their masks and faced Rip with admiring grins. For a moment he grinned back, feeling pretty good. He had held his own with the spacemen, and he sensed that his men liked him.
     “All right,” he said briskly. “Strip down and get into the showers.”
     In a few moments they were all standing under the chemically treated water, washing off the contaminated dust. Rip paid special attention to his hair, because that was where the dust was most likely to stick. He had it well lathered when the Water suddenly cut off. At the same moment, the cruiser shuddered slightly as control blasts stopped its spinning and left them all weightless. Rip saw instantly what had happened. He called, “All right, men. Down on the floor.”
     The Planeteers instantly slid to the shower deck. In a few seconds the pressure of deceleration pushed at them.
     “I like spacemen,” Rip said wryly. “They wait until just the right moment before they cut the water and decelerate. Now we’re stuck in our birthday suits until we land—wherever that may be.”

From Rip Foster Rides the Gray Planet by Blake Savage {Harold Leland Goodwin} (1952)


Nuclear Engine for Rocket Vehicle Applications. The first type of NTR-SOLID propulsion systems. It used reactor fuel rods surrounded by a neutron reflector. Unfortunately its thrust to weight ratio is less than one, so no lift-offs with this rocket. The trouble was inadequate propellant mass flow, the result of trying to squeeze too much hydrogen through too few channels in the reactor.

Thrust Power0.198-0.065 GW
Exhaust velocitySee Table
Thrust49,000 n
Engine mass10 tonne
T/W >1.0no
Exhaust Velocity8,093 m/s
Specific Impulse825 s
Thrust49,000 N
Thrust Power0.2 GW
Mass Flow6 kg/s
Total Engine Mass10,000 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power50 kg/MW
Resuable Nuclear Shuttle [+]
Propulsion SystemNERVA
Exhaust Velocity8,000 m/s
Specific Impulse815 s
Thrust344,000 N
Thrust Power1.4 GW
Mass Flow43 kg/s
Wet Mass170,000 kg
Dry Mass30,000 kg
Mass Ratio5.67 m/s
ΔV13,877 m/s
Widmer Mars Mission [+]
Propulsion SystemNERVA
Exhaust Velocity8,000 m/s
Specific Impulse815 s
Thrust580,000 N
Thrust Power2.3 GW
Mass Flow72 kg/s
Wet Mass400,000 kg
Dry Mass150,000 kg
Mass Ratio2.67 m/s
ΔV7,847 m/s
HELIOS 2nd Stage [+]
Propulsion SystemNTR Solid
Exhaust Velocity7,800 m/s
Specific Impulse795 s
Thrust981,000 N
Thrust Power3.8 GW
Mass Flow126 kg/s
Wet Mass100,000 kg
Dry Mass6,800 kg
Mass Ratio14.71 m/s
ΔV20,968 m/s
Atomic V-2 [+]
Propulsion SystemNTR Solid
Exhaust Velocity8,980 m/s
Specific Impulse915 s
Thrust1,050,000 N
Thrust Power4.7 GW
Mass Flow117 kg/s
Total Engine Mass4,200 kg
Wet Mass42,000 kg
Dry Mass17,000 kg
Mass Ratio2.47 m/s
ΔV8,122 m/s
Specific Power1 kg/MW


Pewee-class Engine
Exhaust Velocity9,200 m/s
Specific Impulse940 s
Thrust111,200 N
(25 klbf)
Thrust Power512 MWt
Mass Flow12.5 kg/s
Total Engine Mass3,240 kg
Uranium 235
Fissle Loading0.25 g U per cm3
Max Fuel Temp2940 K
Fuel Element
1.32 m
U-235 Mass36.8 kg
Chamber Pressure1000 psi
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Engine Length
(inc. skirt ext)
7.01 m
Nozzle Skirt
2.16 m
Nozzle Exit Dia1.87 m
Specific Power6.3 kg/MW
Longest Single
44.5 min
Total Burn
79.2 min
Num Burns4

The 25 kilo-pounds-force (25 klbf) "Pewee" solid-core nuclear thermal rocket was the smallest engine size tested during U.S. Project Rover. While small, a cluster of three is adequate for a typical Mars mission. Single engines were adequate for unmanned scientific interplanetary missions or small nuclear tugs.

A cluster of three Pewee-class engines were selected to be used with NASA's Design Reference Architecture (DRA 5.0) Mars Mission, but later designs replaced them with a cluster of three SNRE-class.

One source suggested that each engine would require a 2,150 kg anti-radiation shadow shield to protect the crew (6.45 metric tons total for a cluster of three), assuming an 80 meter separation between the engines and the habitat module and all the liquid hydrogen propellant tanks used as additional shielding.


The Small Nuclear Rocket Engine (SNRE) is from the report Affordable Development and Demonstration of a Small NTR Engine and Stage: How Small is Big Enough? by Stanley Borowsky et al (2015). The scientists wanted to promote the development of a right-sized solid core nuclear thermal rocket that was as small as possible, but no smaller.

The 111,200 N (25 klbf) "Pewee-class" from the U.S. Project Rover was the smallest Rover engine. A cluster of three was specified for the NASA DRA 5.0 reference, but Borowsky et al determined that was still a bit larger than was strictly necessary.

They looked at a 33,000 Newton (7.5 klbf) engine which was pretty much the smallest NTR possible due to limits on nuclear criticality. There is a minimum amount of fissionable fuel for a reactor, or it just cannot support a chain reaction. But it was a bit too small to do anything useful, even in a cluster of three. About all it was good for was an unmanned robotic science mission.

A 73,000 Newton (16.5 klbf) engine on the other hand could perform quite a few proposed missions. It hit the goldilocks zone, it was just right. Some researchers took designs from NASA's Design Reference Architecture (DRA 5.0) Mars Mission and swapped out the trio of Pewee-class engines for a trio of SNREs.

The engine uses a graphite composite core, because that allowed them to build on the expertise from the old NERVA program.

One source suggested that each SNRE-class would require a 2,000 kg anti-radiation shadow shield to protect the crew (six metric tons for a trio of SNREs), assuming an 80 meter separation between the engines and the habitat module.

The criticality-limited engine has a retractable section of the nozzle, the SNRE-class engine has a nozzle skirt that folds on a hinge (see diagrams below). These are strictly for launch purposes. The spacecraft is boosted in modular parts by several flights of launch vehicle, and assembled in orbit. By retracting/folding the engine nozzle the engine's overall length is reduced enough so that the engine, the liquid hydrogen fuel tank and a small mission payload can be crammed into the launch vehicle's payload faring. Once the spacecraft is assembled, the nozzles are unretracted/unfolded and permanently latched into place.

Criticality-limited Engine
Exhaust Velocity8,770 m/s
Specific Impulse894 s
Thrust33,000 N
(7.4 klbf)
Thrust Power145 MWt
Mass Flow3.8 kg/s
Total Engine Mass1,770 kg
Uranium 235
Max Enrichment93% U-235 wt
Num Fuel Elements260
Num Tie-tube
Fissle Loading0.6 g U per cm3
Max Fuel Temp2736 K
U-235 Mass27.5 kg
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power12.2 kg/MW
Engine Length6.19 m
Fuel Length
Pressure Vessel
0.877 m
Nozzle Exit
1.32 m
Longest Single
22 min
Total Burn
29.5 min
Num Burns2
SNRE-class Engine
Exhaust Velocity8,829 m/s
Specific Impulse900 s
Thrust73,000 N
(16.7 klbf)
Thrust Power367 MWt
Mass Flow8.4 kg/s
Total Engine Mass2,400 kg
Uranium 235
Max Enrichment93% U-235 wt
Num Fuel Elements564
Num Tie-tube
Fissle Loading0.6 g U per cm3
Max Fuel Temp2,726 K
U-235 Mass59.6 kg
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power6.5 kg/MW
Engine Length4.46 m
Fuel Length0.89 m
Pressure Vessel
0.98 m
Nozzle Exit
2.26 m
Longest Single
21.4 min
Total Burn
55 min
Num Burns5


Cermet NERVA
Exhaust Velocity9,120 m/s
Specific Impulse930 s
Thrust445,267 N
Thrust Power2.0 GW
Mass Flow49 kg/s
Total Engine Mass9,000 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power4 kg/MW

Cermet NTR are where the fissionable fuel elements are a composite mixture of fissionable ceramics and a metal matrix.

The problem with the original NERVA fuel elements was the blasted things were too fragile. They were rods of uranium oxide about as strong as a fine china dish. Under the vibrations of rocket flight the rods tended to snap in two. And now you've got live radioactive nuclear fuel spewing out the exhausts like a flying Chernobyl. True the rods were clad in metal to prevent them from eroding away, but the metal less like armor and more like a foil covering. They did nothing to stop the snapping. Making the cladding any thicker caused other problems.

Cermet made the fuel rods act more like solid bars of metal. The rods were basically (solid) foamy tungsten with fissionable uranium oxide trapped inside the bubbles. The tungsten skeleton ensured that the rods would laugh at the engine vibrations.

Cermet NERVA
Cermet NERVA
Exhaust Velocity9,810 m/s
Specific Impulse1,000 s
Thrust134,400 N
Thrust Power0.7 GW
Mass Flow14 kg/s
Total Engine Mass32,546 kg
Frozen Flow eff.73%
Thermal eff.96%
Total eff.70%
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power49 kg/MW

The NERVA (Nuclear Engine for Rocket Vehicle Application) system captures the neutronic energy of a nuclear reaction using a heat exchanger cooled by water or liquid hydrogen. The exchanger uses thin foil or advanced dumbo fuel elements with cermet (ceramic-metal) substrates, jacketed by a beryllium oxide neutron reflector.

The chamber temperature is limited to 3100K for the extended operational life of the solid fuel elements, which can be fission, fusion, or antimatter. At this temperature, the disassociation of molecular H2 to H significantly boosts specific impulse at chamber pressures below 10 atm.

A propellant tank pressurized to 2 atm expels the LH2 coolant into the exchanger without the need for turbopumps. This open-cycle coolant is expanded through a hydrogen-cooled nozzle of refractory metal to obtain thrust.

The efficiencies are 96% thermal, 76% frozen-flow (mainly H2 dissociation, less recombination in the nozzle), and 96% nozzle. A 940 MWth heat exchanger yields a thrust of 134 kN, and a specific impulse of 1 ksec, at a power density of 340 MW/m3.

Altseimer, et al., “Operating Characteristics and Requirements for the NERVA Flight Engine,” AIAA Paper 70-676, June 1970.

From HIGH FRONTIER by Philip Eklund
Composite/Cermet Comparison


The study author is trying to compare different types of solid-core NTR to figure out which is best. Spoiler Alert: the HEU-Cermet seems to have a slightly better performance than the others.

All the engines were designed to have 111 kiloNewtons of thrust (25 klbf, the same as the Pewee). The first variable was Low Enriched Uranium (LEU, 2%-20% 235U) or Highly Enriched Uranium (HEU 20%-100% 235U, which includes Weapons Grade). The second variable was old-style composite fuel elements and new-style cermet elements. These gave four engines to compare:

Case Identifiers
HEU 93.1%com93cer93
LEU 19.1%com20cer20

HEU-Composite (com93) is based on the SNRE design. The SNRE had much more moderator material added in an attempt to reduce the engine mass, and to reduce the thrust.

LEU-Composite (com20) is a reactor designed to use LEU fuel, instead of the HEU fuel used in the orginal NERVA. It is much more difficult to create a nuclear reaction with LEU fuel, but military is much happier preventing weapons-grade uranium being in civilian hands. To allow using LEU, the materials in the engine were swapped with materials that absorbed less neutrons. The amount of moderator was increased by adding more hydrogen.

HEU-Cermet (cer93) is the standard cermet NTR. It was actually tested in the 60s and 70s. The cermet fuel elements are uranium oxide in a tungsten matrix. It has less moderating material than the NERVA.

LEU-Cermet (cer20) is a more theoretical concept. In order to get away with using LEU, much like the com20 the engine has to be built out of materials that absorb less neutrons. You want more neutrons generating power by hitting uranium atoms and less neutrons wasted being absorbed by the engine, when LEU has fewer uranium atoms to be hit in the first place. In particular the tungsten in the cermet was to be isotopically enriched, i.e., have a higher proportion of tungsten-184 than the 31% you find in naturally occurring tungsten. Other alternatives are having the fuel elements use uranium nitride ceramic instead of the standard uranium oxide and/or using molybdenum instead of tungsten (or at least in the regions of the engine where the temperature was below molybdenum's melting point).

Engine Descriptions

The HEU-Composite fuel element has a width of 1.91 cm flat and 19 holes to simplify fabrication, but the 121 cm length is troubling (longest of all the four concepts). The LEU-Composite element has a wider width of 2.77 flat to mimimize neutron absorption.

The HEU-Cermet has 91 holes for maximum specific impulse. The LEU-Cermet has only 61 holes but this only cuts the specific impulse by a few seconds. The HEU-Cermet has the shortest fuel length (64 cm) of all four concepts, which is a plus.

The U-235 densities are interesting. The cer93 (HEU-cermet) is two orders of magnitude higher than com20 (LEU-composite): 4.560 to 0.063. Most of the design difference between the concepts can be traced back to the U-235 densities.

All of the engines except for HEU-Cermet (cer93) use tie-tubes that contain zirconium hydride moderator. This reduced the mass but adds technical risk (i.e., it might not be possible). Zirconium hydride is prone to swelling and phase changes (melting) at various temperature under nuclear irradiation, which is a bad thing. Understand that the tie-tubes are the framework holding the engine together. After shut-down it might require the engine to keep venting huge amounts of liquid hydrogen propellant just to cool off the ZrH tie-tubes. This will drastically reduce the net specific impulse. The tie-tube hydrogen coolant is supposed to be reused, but the tie-tubes leak. Actually pretty much everything leaks hydrogen, the blasted stuff can sneak in between the atoms of the container and escape.

ZrC40 is a 40% dense (60% porous) zirconium carbide insulator material developed during the Rover/NERVA program. It is used to prevent heat from the hot uranium fuel rods from overheating the zirconium hydride moderator in the center of the tie-rods. The fuel rod heat is supposed to be all used on the hydrogen propellant. Any leaking into the tie-rod is both lowering propellant heating efficiency and threatening the zirconium hydride with a lethal melt-down. The secret of making ZrC40 may have been lost when the NERVA program shut down, which would require redevelopment work.

The HEU-Composite (com93) uses ZrH1.6, zirconium hydride with a H/Zr ratio of 1.62. Com20 and cer20 bump that up to a ratio of 2.0. The 2.0 ratio may be unworkable, depending upon how well it can be cooled (nominally and during transients). And depending upon how well the hydrogen coolant can be prevented from leaking out of the tie-tube, which hinges on controlling the temperature and overpressure.

To reduce neutron capture (very important if you are trying to make do with weak LEU) the LEU-Composite (com20) uses SS-315 (marine grade stainless steel) for the tie-tube structure. Assuming the tie-tubes can be kept cool enough to prevent the steel from melting.

The LEU-Cermet reduces neutron capture by using molybdenum that has been enriched in 96molybdenum, which has an even lower neutron capture than SS-315. It also usess W5Re instead of W25Re for the fuel coating. The lower rhenium content reduces neutron capture. The drawback is W5Re is less ductile, making it more difficult to manufacture the fuel rods and increasing the risk of rupture.

In fact, the LEU-Cermet design has a bunch of materials enriched to remove the isotopes which are notorious for absorbing neutrons. The materials are tungsten, gadolinium, molybdenum, and rhenium. Tungsten is 98% 184W, gadolinium is 99% 160Gd, molybdenum is 98% 92Mo, and rhenium is 95% 187Re. It is unclear if these levels of enrichment could ever be practical, especially 184W. The other designs would be improved by using these enriched materials, but they can get by without them. The LEU-Cermet cannot.

For the two cermet designs, "W10" indicates that the slat region is filled with 10% tungsten and 90% void. The designs also have an axial reflector composed of beryllium oxide on the cold end (the coolant inlet). This adds neutron reactivity, and flattens the power profile.

Concept Performance

Reactor Neutronics

Each of the concepts is designed to have at least a 1% operational margin in Keff; that is, Keff > 1.01 from Beginning of Life (BOL) to End of Life (EOL). "Life" in the context of a nuclear reactor means that time from when a fresh bundle of fissionable fuel rods is installed to the time when a spent bundle of fuel rods full of nuclear poisons is removed from the reactor. When the engine is shut down after a burn (the control drums are set to quench the reactor) Keff < 0.95.

While each concept meet the requirements, the LEU-Cermet (cer20) has so little neutron margin that it may be impossible to make a design that actually works.

Table III above give fuel power peaking, or the factor that peak power is above average power. Which make a real difference when applying it to a reactor with very high power densities. Two kilowatts average power with a peaking power of 2.7 means the design just has to handle a peak power of 5.4 kilowatts. Easy. But if the average power is two gigawatts, designing it to handle a peak power of 5.4 GW is not trivial.

Reactor Thermal-Hydraulics

Table IV summarizes the core thermal-hydraulics, while tables V and VI show the temperature conditions at two potentially limiting core regions. Without a detailed thermal-structural analysis, it is hard to say which of these regions will be limiting for the entire design. The report makes a guess that it will be the peak fuel temperature location in table VI.

The cermet engines have much higher power densities than composite cores, the cermets actually have lower fuel delta-Ts due to the smaller lattice size of the holes and higher thermal conductivity. The max fuel temperatore of 2800 K is arbitrary and probably optimistic.

Rocket Performance

In Table VII the power deposition (P.D.) values represent direct nuclear heating (i.e., reactor energy transferred into propellant for thrust, which we want). Loss values represent heat transfer between components (reactor energy wasted by heating up engine components, which we do not want).

Table IX summarizes the key performance parameters of each system.

The reactor mass includes an internal radiation shield designed to keep doses "above" (i.e., in a direction opposite to the exhaust flow) to no more than 10 MRad(Si) gamma radiation and 1×1014 n/cm2 (>100 keV). This is a radiation level that motors and turbomachinery can tolerate, but not much else. Even adding separation that much radiation will heat up the propellant tanks until they pop and give the crew lethal doses. A much heavier shield will be needed for crew safety.

And if you cluster engines, you will need neutron isolation shields or neutrons from adjacent engines will cause nuclear flare-ups. In this case, the cermet cores will have an advantage because the side leakage is reduced by the core-internal high-Z shielding, a thicker radial reflector, and a power profile peaked closer to the radial centerline.

The engines specific impulses (Isp) are listed (multiply by 9.81 to get exhaust velocity in m/s). Two adjustments were made.

The "decay cooling" adjustement accounts for hydrogen flow required to prevent overheat after reactor shutdown. The engine concepts that use extra moderation (com93, com20) require significantly more cooling because of the need to keep the zirconium hydride cool.

The "peaking change" adjustment accounts for changes in the peaking factor cause by control drum movements required for burnup reactivity effects. These include fuel depletion and fission product accumulation over the assumed 10 hours of total thrusting time between BOL and EOL. It also includes the effect of 135xenon poisoning during the (presumed) 45 minute individual burns. Xenon-135 is the most powerful known neutron-absorbing nuclear poison with a half-life of 9.2 hours. Xenon poisoning may prevent the moderated systems (com93, com20) from restarting for a burn within a day or two of prior operation. This will limit mission flexibility.


  • There is relatively little difference between the engines. Not a surprise since they all have the same fuel temperature limit
  • The net Isp of the HEU-Cermet (cer93) is about 17 seconds higher than the others, and with a higher thrust-to-weight ratio as well
  • The LEU-Cermet (cer20) has two strikes against it: peaking factors and requirement for isotopal enrichment. These could be mitigated with a higher mass design.

NERVA Derivative

Exhaust Velocity8,085 m/s
Specific Impulse824 s
Thrust334,061 N
Thrust Power1.4 GW
Mass Flow41 kg/s
Total Engine Mass10,100 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power7 kg/MW


General Dumbo
Thrust Power14.0-4.6 GW
Exhaust velocitySee Table
Thrust3,500,000 n
Engine mass5 tonne
T/W >1.0yes
Dumbo (H2)
Exhaust Velocity8,093 m/s
Specific Impulse825 s
Thrust3,500,000 N
Thrust Power14.2 GW
Mass Flow432 kg/s
Total Engine Mass5,000 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Dumbo Model A
Engine mass0.7 tonne
Thrust400,000 n
Propellant mass flow52 kg/sec
Exhaust velocity7,700 m/sec
Engine Height0.6 m
Engine Radius0.3 m
Engine Volume0.2 m3
Dumbo Model B
Engine mass2.8 tonne
Thrust3,560,000 n
Propellant mass flow460 kg/sec
Exhaust velocity7,700 m/sec
Engine Height0.6 m
Engine Radius1.0 m
Engine Volume1.8 m3
Dumbo Model C
Engine mass2.1 tonne
Thrust400,000 n
Propellant mass flow48 kg/sec
Exhaust velocity8,300 m/sec
Engine Height0.6 m
Engine Radius0.4 m
Engine Volume0.3 m3

This was a competing design to NERVA. It was shelved political decision that, (in order to cut costs on the atomic rocket projects) required both projects to use an already designed NEVRA engine nozzle. Unfortunately, said nozzle was not compatible with the DUMBO active cooling needs. Dumbo does, however, have a far superior mass flow to the NERVA, and thus a far superior thrust. Dumbo actually had a thrust to weight ratio greater than one. NASA still shelved DUMBO because [a] NERVA used off the shelf components and [b] they knew there was no way in heck that they could get permission for nuclear lift-off rocket so who cares if T/W < 0? You can read more about Dumbo in this document.

Note that the "engine mass" entry for the various models does not include extras like the mass of the exhaust nozzle, mass of control drums, or mass of radiation shadow shield.

Pebble Bed

Pebble Bed
Exhaust Velocity9,530 m/s
Specific Impulse971 s
Thrust333,617 N
Thrust Power1.6 GW
Mass Flow35 kg/s
Total Engine Mass1,700 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power1 kg/MW

Particle bed / nuclear thermal rocket AKA fluidized-bed, dust-bed, or rotating-bed reactor. In the particle-bed reactor, the nuclear fuel is in the form of a particulate bed through which the working fluid is pumped. This permits operation at a higher temperature than the solid-core reactor by reducing the fuel strength requirements . The core of the reactor is rotated (approximately 3000 rpm) about its longitudinal axis such that the fuel bed is centrifuged against the inner surface of a cylindrical wall through which hydrogen gas is injected. This rotating bed reactor has the advantage that the radioactive particle core can be dumped at the end of an operational cycle and recharged prior to a subsequent burn, thus eliminating the need for decay heat removal, minimizing shielding requirements, and simplifying maintenance and refurbishment operations.

Project Timberwind

Project Timberwind was started in President Reagan infamous Strategic Defense Initiative ("Star Wars"). It was later transferred to the Air Force Space Nuclear Thermal Propulsion (SNTP) program. The project was cancelled by President William Clinton.

NTR Comparison
Engine Mass6,803 kg1,500 kg
Thrust (Vac)333.6 kN392.8 kN
Specific Impulse850 s1,000 s
Burn Time1,200 s449 s
T/W530 !!!

The idea was to make a nuclear-powered interceptor to destroy incoming Soviet ICBMs. The Timberwind NTR upper stage would have to make the NERVA engine look like a child's toy, with huge specific impulse and an outrageously high thrust-to-weight ratio. The project managers babbled about advances in high-temperature metals, computer modelling and nuclear engineering in general justifying suspiciously too-good-to-be-true performance. It was based on the pebble-bed concept.

Diameter4.25 m2.03 m8.70 m
Thrust (Vac)392.8 kN735.5 kN2,451.6 kN
Specific Impulse1,000 s1,000 s1,000 s
Engine Mass1,500 kg2,500 kg8,300 kg
Burn Time449 s357 s493 s


One of the design goals was to make the engine work with low enriched uranium (LEU) instead of weapons-grade uranium (HEU) like older designs. LEU is harder to make work in a nuclear thermal engine, but weapons-grade uranium in civilian hands make the military very nervous.

(12.5% U-235)
Engine Power1.76 GW
Mass flow rate25.15 kg/s
Exhaust Vel10,610 m/s
Isp1,082 sec
Thrust270,000 N
Engine Mass~2,900 kg

      The development of a low enriched, nuclear thermal rocket (NTR) has become a necessity for more effective deep space travel. An NTR provides a significantly higher specific impulse (Isp) than chemical rockets. Starting in the 1960s the United States began the Nuclear Engine for Rocket Vehicle Applications (NERVA) program. The overall design of the NTR described in this paper, INsTAR, incorporates many elements from the NERVA prototypes. While many NERVA engines incorporated a high enriched uranium fuel, the low enriched uranium fuel concepts have proven to be feasible. The overall design objective was to develop a full core design for an NTR that addresses the problems found in the Space Nuclear Thermal Propulsion Project core and improves upon the Isp and thrust to weight ratio observed during the project. The core design is characterized with respect to neutronics, thermal hydraulics, and propulsive performance. A critical component of the design was the study of materials utilized in the core and their compatibilities at high operating temperatures.


     Nuclear thermal propulsion (NTP) is a leading, inspace design to reduce travel time for astronauts and carry larger payloads than contemporary chemical rockets. Efforts to develop nuclear thermal rockets (NTRs) began in the 1960s. One such effort, the Nuclear Engine for Rocket Vehicle Applications (NERVA), was a nuclear thermal rocket engine development program that ended in 1972 (Ref. 1). An example of a rocket engine developed by the NERVA program is seen in Fig. 1. NERVA ultimately demonstrated that NTRs were a feasible technology for space exploration. One of the engine designs during this time met all requirements for a mission to Mars. Since then, it has been determined that NTRs are essential for regular trips to Mars and human exploration of further planets. Approximately twenty experiments of different types of nuclear reactors for space applications were completed.1

     Since the feasibility of low enriched uranium (LEU) for NTR applications has been demonstrated, two NERVA-inspired LEU concepts have been under investigation. These concepts are the Space Capable Cryogenic Thermal Engine (SCCTE) and the Superb Use of Low Enriched Uranium (SULEU). Both designs are based on the same geometry as the NERVA engine and use similar materials and components. The difference between these two concepts are in the fuel design. The SCCTE fuel design utilizes enriched Tungsten-184 ceramic-metallic fuel. The SULEU fuel design utilizes (U,Zr)C-Graphite composite fuel that has been the typical fuel design of previous NTRs. Both concepts utilize LEU fuel with an enrichment of 19.75% U-235. They also are designed to operate with a specific impulse of 900 seconds at a thrust of 35k lbf.3

     Particle bed reactor (PBR) concepts are used in space applications and utilize grain sized spherical particles composed of fuel and cladding. Thousands of fuel particles are arranged in an array and can be in many different configurations based on desired properties. Several layers of fuel elements create the reactor and gas, typically hydrogen, radially flows through the fuel elements to cool the system. The hydrogen is heated and expands as it is ejected axially. The fuel's very high surface-to-volume ratio is desirable as this allows for high heat transfer efficiencies and large power densities.1 The PBR's compact design and high power density are ideal attributes for space applications.

     The coolant and propellant used in the PBR design is liquid hydrogen. Liquid hydrogen has the lowest molecular weight of all substances, leading to a higher specific impulse than other propellants. It is cryogenic and therefore should be handled with extreme care and stored below 423 °F. Shielding and insulation are required to keep the propellant from evaporating or boiling. The shielding will protect the propellant from the radiant heat of the sun and the insulation will provide protection from other sources of heat like air friction during the flight through the atmosphere.4 Liquid hydrogen is the signature fuel used by the American space program and works well with the proposed NTR design, INsTAR.


     Specific impulse (Isp) and thrust to weight ratios (T/W) are important performance metrics for space exploration. The NTR is able to achieve an Isp exceeding twice that of a traditional chemical rocket, helping account for the lower T/W than chemical engines.1 The proposed design improves upon the NERVA derived Space Nuclear Thermal Propulsion (SNTP) engine. The earliest NTR designs achieved a T/W of 4:1 to 8:1 with SNTP achieving a T/W of 16.3:1. The SNTP design incorporates a PBR.1 The fuel elements were composed of high enriched uranium (HEU) with a tungsten (W) cladding that was primarily W-184. The W-184 cladding allows the fuel to operate at much higher temperatures while still providing a lower thermal neutron absorption cross-section.5

     The SNTP engine was selected as the design basis for the LEU NTR, INsTAR, for its high T/W and its Isp of 930 seconds. Despite this success, the engine endured failures that are addressed in the final design of INsTAR. Initial nuclear element test runs yielded power irregularities caused by fuel element failure. The fuel had migrated within the fuel bed, creating regions of maximum stress, not accounted for in previous thermal calculations. The hot frit had split, and the cold frit had warped due to stresses from fuel bed thermal expansion. Later particle bed nuclear tests had derived the melting temperature of the UC2 to be 2500K, below that of the operating temperatures of the LEU NTR.1

     The overall goal of this project is to improve on the issues that were experienced during the SNTP program and to expand on research that was done previously. Some of these issues included fuel migration and the inability to achieve the high temperatures desired.1 The fuel design used was chosen to prevent the issues seen in the SNTP engine and to increase thermal capabilities. The end goals of this project are to achieve an Isp of approximately 1000 sec. and a T/W between 8:1 and 10:1 with a pressure drop across the core that is less than 1000 psi. MCNP 6.1 (Ref. 6) and thermal hydraulics calculations were used to validate the core design. Compatibility of the materials at high temperatures was investigated. The rocket performance was characterized through Isp and T/W calculations.


     The proposed reactor is an improvement on the PBR design that emerged from the SNTP project. The system design is outlined in the following sections.

III.A. Fuel Pellet

     The fuel pellets consist of spherical particles with an internal kernel of uranium (U) metal enriched to 12.5% U-235. The kernel of U metal is clad in a 0.2 mm thick tungsten-rhenium alloy for a total outer diameter of 2 mm, as seen in Fig. 2. There is a 0.1 mm gap between the fuel and the cladding to accommodate for fission gases and to account for thermal expansion of the fuel as it is heated and then melts.

     The cladding is composed of 90% W-184 with 5% natural W and 5% rhenium (Re). Adding 5% Re to the cladding improves ductility and strength of the overall material and will improve the ability to spherically encase the fuel.7 Tungsten alone isn't ductile enough to successfully encase uranium during operation of INsTAR.

     The fuel particle design allows the U metal to operate in a molten state while the cladding is still in a stable solid state. This is because the melting temperature of U is 1135 °C (Ref. 8) whereas the melting temperature of W is 3422 °C (Ref. 9) and the melting temperature of Re is 3185 °C (Ref. 10). From a phase diagram, the melting temperature of the WRe alloy is about 3300 °C (Ref. 11). The material interaction we would potentially see between the cladding and fuel would be a recrystallization of the two materials. This shouldn’t be too much of a concern given that we are only burning for several minutes at a time. The fuel design allows the rocket to safely operate at temperatures up to approximately 3500 K.

III.B. Fuel Element Design

     The fuel particles will be arranged in a hexagonal close-packed design, as seen in Fig. 3, with the particles sintered together to form a fuel element which will prevent particle migration. The particles will form a cylinder with a conical shaped channel in the center to allow for the hydrogen gas to flow out towards the nozzle, as seen in Fig. 4. There is a hot frit that lines this channel, composed of porous graphite. The cold frit is also composed of porous graphite and lines the outside of the fuel particles. The fuel particles are surrounded by graphite moderator, as seen in Fig. 4.

III.C. Core Description

     The core consists of identical fuel elements surrounded by a beryllium reflector. The center of the fuel element is a conical-shaped channel allowing for flow of the H2 propellant, as seen in Fig. 4. The hot frit is surrounded by a 10 cm wide conical particle bed tapering to 5 cm consisting of the sintered fuel pellets described in the fuel pellet section. These particles are located between the cold frit and hot frit in Fig. 4. The particle bed is surrounded by a 15 cm wide hexagonal shaped graphite moderator. Fuel elements are arranged to fit 37 identical fuel elements inside of a beryllium (Be) reflector as shown in Fig. 5. Six beryllium control drums were added to the design to account for the excess reactivity. One third of each control drum is lined with boron carbide, which has a high neutron absorption cross section. The final dimensions of the core were 150 cm tall with a 140 cm diameter. The mass of the core is estimated to be approximately 1500 kg, with 310 kg of U metal. Crosssections of the core can be seen in Fig. 5 and Fig. 6.


     Several aspects of core performance were evaluated before a final design was reached. MCNP 6.1 (Ref. 6) was used to characterize the neutronic capabilities of the core. Thermal hydraulic calculations were completed to determine pressure drops, heat generation, and the maximum core temperature. Propulsion calculations were performed to determine Isp and the T/W. Both the neutronics and thermal hydraulics calculations required a packing efficiency, which is 74% for hexagonally close-packed systems.

IV.A. Neutronics

     Through neutronics calculations, it was determined that the original design had an extreme amount of excess reactivity. This resulted in the decision to reduce the enrichment of the uranium, making this design more favorable regarding uranium enrichment. The final design had a keff of 1.02134 ± 0.0006. Six beryllium control drums were added to the design for reactivity control. When the 120° arc of boron carbide is turned towards the core, the reactivity is reduced, and the reactor becomes subcritical. Fig. 7 shows the effect of the control drums on the reactivity.

IV.B. Thermal Hydraulics

     Thermal hydraulics calculations were used to ensure the feasibility of the system and were correlated with the neutronics calculations to ensure the design was consistent. The results of the thermal hydraulic analysis regarding temperature and pressure are summarized in Table 1 below. The maximum temperature within the system is 3301 K. The system requires a mass flow rate of 25.15 kg/s, resulting in a pressure drop across the core of 623 psi, which meets the design criteria. The total power was calculated to be 1.76 GW.

Core inlet and outlet parameters
ParameterCore InletCore Outlet
Temperature (K)333300
Pressure (psi)16231000

IV. C. Rocket Performance

     When a final design was decided upon, the Isp and T/W ratio was calculated, along with the Mach number at several sections of the nozzle. A schematic of a nozzle can be seen in Fig. 8. The hydrogen exits the core and mixes in a conical section between the core and nozzle which results in a homogeneous hydrogen temperature. From there, the gas enters the nozzle with a Mach number less than one and reaches a Mach number of one at the throat of the nozzle, as seen in Fig. 8. From here, the gas continues to speed up and exits the nozzle at a supersonic velocity and a Mach number greater than one. The gas exits the nozzle with a velocity of 10,610 m/s and a Mach number of 5.66. The Isp for this design is 1,082 seconds and the T/W is 9.5 for a thrust of 60,000 lbf. The Isp calculation does not account for the hydrogen dissociation that would occur at the operating temperature of this design. Accounting for hydrogen dissociation will increase the Isp due to a lowered average molecular weight. The Isp determined for this design is higher than several other design concepts. This is due to the fact that a higher maximum operating temperature is allowed because the tungsten-rhenium cladding has a high melting temperature. For comparison, the SNTP Isp was 930s, and the NERVA Pewee had an Isp of 901s.1


     This analysis has demonstrated that the preliminary design of the INsTAR rocket is feasible. Further optimization and design changes need to be implemented to increase the usefulness of the design, such as increasing the T/W. The Isp, T/W, and pressure drop across the core met design criteria and further support the feasibility of this design. An Isp of 1,082 seconds is higher than any other previous NTR designs. This high specific impulse is largely due to the innovative use of a tungsten-rhenium cladding, which allows the fuel to operate in a molten state and the propellant to be heated to a higher temperature than is otherwise possible. The T/W is larger than the values achieved by similar sized NTRs. For future work, determination of the hydrogen dissociation rate will allow for more accurate Isp calculations. At this point in time, from a material stand point, the design is compatible from all angles. There needs to be optimization for materials and the interactions that they undergo. This will come from testing and expanding on any new research that is available. Major design components that need to be addressed further include the cladding stability at very high temperatures and the flow of high velocity gases through particle fuel beds. In addition, the MCNP models have fuel elements which approximate the particle bed fuel as a porous medium. Though a design such as this one has not been fully operable, there are many things to learn and overcome. By addressing these minor issues, it will be possible to make a low enriched nuclear thermal rocket for deep space exploration. Our research into this NTR has proven that LEU nuclear thermal propulsion is possible with further optimization and testing. As a future goal, the hope is that we will eventually have a working rocket core to properly and safely achieve propulsion, expanding opportunities for deep space missions.


  1. R. A. HASLETT, “Space Nuclear Thermal Propulsion Program Final Report,” Grumman Aerospace Corporation (1995).
  2. NASA ON THE COMMONS, Flickr (2018).
  3. P. VENNERI and Y. KIM, “Advancements in the Development of Low Enriched Uranium Nuclear Thermal Rockets,” Energy Procedia, 131 (2017).
  4. B. DUNBAR, NASA – Liquid Hyrogen – the Fuel of Choice for Space Exploration, (2018).
  5. R. L. MACKLIN, D. M. DRAKE and E. D. ARTHUR, “Neutron-Capture Cross Sections of the Tungsten Isotopes 182W, 183W, 184W, and 186W from 2.6 to 2000 keV,” Oak Ridge National Laboratory (1982).
  6. T. GOORLEY,, “Initial MCNP6 Release Overview,” Nuclear Technology, 180, p. 298-315 (2012).
  7. M. S. EL-GENK, “A review of refractory metal alloys and mechanically alloyed-oxide dispersion strengthened steels for space nuclear power systems,” Science Direct (2004).
  8. Uranium – Element information, properties and uses | Periodic Table, (n.d.).
  9. Tungsten – Element information, properties and uses | Periodic Table, (n.d.).
  10. Rhenium – Element information, properties and uses | Periodic Table, (n.d.).
  11. U. RAVI KIRAN,, “Refractory metal alloying: A new method for improving mechanical properties of tungsten heavy alloys,” Journal of Alloys and Compounds, 709 (2017).
  12. C. NEWEY and G. WEAVER, Materials Principles and Practice, M. KEYNES, Open University, Materials Department (1991).

Pulsed Solid-core NTR

The pulsed nuclear thermal rocket is a type of solid-core nuclear thermal rocket concept developed at the Polytechnic University of Catalonia, Spain and presented at the 2016 AIAA/SAE/ASEE Propulsion Conference. It isn't a torchship but it is heading in that direction. Thanks to Isaac Kuo for bringing this to my attention.

As previously mentioned, solid core nuclear thermal rockets have to stay under the temperature at which the nuclear reactor core melts. Having your engine go all China Syndrome on you and shooting out what's left of the exhaust nozzle in a deadly radioactive spray of molten reactor core elements is generally considered to be a Bad Thing. But Dr Francisco Arias found a clever way to get around this by pulsing the engine like a TRIGA reactor. The engine can be used bimodally, that is, mode 1 is as a standard solid-core NTR (Dr. Arias calls this "stationary mode"), and mode 2 is pulsed mode.

Pulse mode can be used two ways:

Direct Thrust Amplification: Garden variety solid core NTRs can increase their thrust by shifting gears. You turn up the propellant mass flow. But since the reactor's energy has to be divided up to service more propellant per second, each kilogram of propellant gets less energy, so the exhaust velocity and specific impulse goes down.

But if you shift to pulse mode along with increased propellant mass flow, the reactor's effective energy output increases. So you can arrange matters in such a way that each kilogram of propellant still gets the same share of energy. Bottom line: the thrust increases but the specific impulse is not degraded.

Specific Impulse Amplification: This is really clever. For this trick you keep the propellant mass flow the same as it was.

In a fission nuclear reactor 95% of the reactor energy comes from fission-fragments, and only 5% come from prompt neutrons. In a conventional solid-core NTR the propellant is not exposed to enough neutrons to get any measurable energy from them. All the energy comes from fission fragments.

But in pulse mode, that 5% energy from neutrons could be higher than the 95% fission-fragment energy in stationary mode. The difference is that fission fragment energy heats the reactor and reactor heat gives energy to the propellant. And if the reactor heats too much it melts. But neutron energy does not heat the reactor, it passes through and directly heats the propellant.

The end result is that in pulse mode, you can actually make the propellant hotter than the reactor. Which means a much higher specific impulse than a conventional solid-core NTR which running hot enough to be right on the edge of melting.

Thermodynamics will not allow heat energy to pass from something colder to something hotter, so it cannot make the propellant hotter than the reactor. But in this case we are heating the propellant with neutron kinetic energy, which has zippity-do-dah to do with thermodynamics.

The drawback of course is that the 95% fission-fragment energy is increased as well as the neutron energy. The important point is by using pulsing you can use an auxiliary cooling system to cool the reactor off before the blasted thing melts, unlike a conventional NTR.

Apparently Dr. Arias' paper claims the pulsed NTR can have a higher specific impulse than a fission fragment engine. I am no rocket scientist but I find that difficult to believe. Fission fragment can have a specific impulse on the order of 1,000,000 seconds. But see SSailor67's comment below.

How Does It Work?

TRIGA reactor have what is called a large, prompt negative fuel temperature coefficient of reactivity. Translation: as the nuclear fuel elements heat up they stop working. It automatically turns itself off if it gets too hot. Technical term is "quenching."

Which means you can overload it in pulses. The TRIGA is designed for a steady power level of 100 watts but you can pulse the blasted thing up to 22,000 freaking megawatts. It automatically shuts off after one-twentieth of a second, quickly enough so the coolant system can handle the waste heat pulse.

Amplification Factor

The amount of amplification of thrust or specific impulse requires the value of N, or energy ratio between the pulsed mode and the stationary mode (pulsed mode energy divided by stationary mode energy). This can be calculated by the formidable equation

ΔT is the temperature increase during a pulse (in Kelvin), t is the residence time of the propellant in the reactor (seconds), and [ ΔT/t ] is the quench rate (K/sec). ΔT will probably be about 103 K (assuming propellant velocity of hundreds of meters per second and chambers about one meter long), t will probably be from 10-3 sec to 10-2 sec. This means [ ΔT/t ] will be about 105 to 106 K/s.

I'm not going to explain the other variables, you can read about them here.

Be that as it may, Wikipedia states that if you use standard reactor fuels like MOX fuel or Uranium dioxide, fuel heat capacity ≅ 300J/(mol ⋅ K), fuel thermal conductivity ≅ 6W/(K ⋅ m2), fuel density of ≅ 104kg/(m3), cylindrical fuel radius of ≅ 10-2m and a fuel temperature drop from centerline to cladding edge of 600K then:

N ≅ 6×10-3 * [ ΔT/t ]

This boils down to N being between 600 and 6,000.

Direct Thrust Amplification Details

Thrust power is:

Fp = (F * Ve ) / 2

Thrust is:

F = mDot * Ve

Specific Impulse is:

Isp = Ve / g0


Fp = Thrust Power (w)
F = Thrust (N)
Ve = Exhaust Velocity (m/s)
mDot = Propellant Mass Flow (kg/s)
Isp = Specific Impulse (s)
g0 = acceleration due to gravity (9.81 m/s2)

With a conventional solid NTR, thrust power is a constant. So if you wanted to increase the thrust by, for instance 5 time, you have to increase the propellant mass flow by 52 = 25 times and decrease the exhaust velocity by 1/5 = 0.2 times. Which decreases the specific impulse 0.2 times.

But a pulsed NTR can increase thrust power. So if you want to increase the thrust by 5 times, you increase the thrust power by 5 times, the propellant mass flow five times, and keep the exhaust velocity and specific impulse the same.

The limit on the increase in thrust power is N.

Specific Impulse Amplification Details

If in pulse mode the amplification factor is N, then the amplified specific impulse is:

IspPulse = IspS * sqrt[ (fn * N) + 1]


IspPulse = Specific Impulse in Pulse Mode
IspS = Specific Impulse in Stationary Mode
fn = fraction of the prompt neutrons (0.05)
N = energy amplification by pulsing the reactor
sqrt[x] = square root of x

So if N is between 600 and 6,000, the specific impulse will increase by a factor of 5.57 to 17.35. With a basic NERVA having a specific impulse of about 800 seconds, a pulsed version would have instead 4,460 to 13,880 seconds!


      The wiki math that estimates ~13k seconds Isp as the maximum is wrong. The math assumes cylindrical fuel geometry like a standard NERVA, but the paper's abstract says it has to be thin sheets. Money quote:

     "In addition, thin geometries of the fuel are mandatory to keep intimate content with the quenching coolant."

     If pulsing a cylindrical fuel element can get ~10,000s Isp, thin linear sheets probably can get 100x better due to more neutron contact, which would validate the "better Isp than a fission fragment" claim.

     Plugging that into the rocket equation, you get ~6,800km/s delta-V with a mass ratio of 2, and the possibility of high thrust, which is definitely a torch ship, albeit not one that can do a 1G brachistochrone to Pluto.

by SSailor67 (2021)

Russian Twisted Ribbon

These are from Russian Nuclear Rocket Engine Design for Mars Exploration by Vadim Zakirov and Vladimir Pavshook. The unique "twisted ribbon" fuel elements were developed in the Soviet Union, and continued development in Russia. The twisted ribbon surface-to-volume ratio is 2.6 times higher than that of the US NERVA fuel elements, which enhances the heat transfer between fuel and propellant.

The prototype RD-0140 engine was a pure rocket engine, while the nuclear power and propulsion system (NPPS) is a Bi-Modal NTR acting as an electrical power generator in between thrust periods. A spacecraft designed for a Mars mission would have three or four NPPS engines.

Twisted Ribbon Engines
Thrust (vac) (kN)35.2868
PropellantH2 + HexaneH2
Propellant Mass Flow (kg/s)~4~7.1
Specific Impulse (vac) (s)~900~920
Core outlet temparture (K)3,0002,800 to 2,900
Chamber Pressure (105 Pa)7060
U235 enrichment (%)9090
Fuel Composition(U,Nb,Zn)CU-Zr-C-N
Fuel Element FormTwisted ribbonTwisted ribbon
Generated electrical power (kW)N/A50
Working fluid for power loop
(% by mass)
N/A93% Xe + 7% He
Max temp for power loop (K)N/A1,500
Max press for power loop (105 Pa)N/A9
Working fluid flow rate (kg/s)N/A1.2
Thermal power - propulsion mode (MW)196340
Thermal power - power mode (MW)N/A0.098
Core length (mm)800700
Core diameter (mm)500515
Engine length (mm)3,700No Data
Engine diameter (mm)1,200No Data
Lifetime - propulsion mode (h)15
Lifetime - power mode (yr)N/A2
Mass (kg)2,000*1,800**

N/A = not applicable. * = including radiation shield and adapter. ** = reactor mass.

In the RD-0140 they added hexane to the liquid hydrogen propellant. Unfortunately pure hot hydrogen tended to erode the fuel elements and make the exhaust radioactive.

Twisted Ribbon Engine
Thrust power1,650 MW
Exhaust velocity9,420 m/s
Specific impulse960 s
Thrust330,000 N
Engine mass5,260 kg

The CIS engine developed jointly by the US/CIS industry team of Aerojet, Energopool and B&W utilizes a heterogeneous reactor core design with hydrogen-cooled ZrH moderator and ternary carbide fuel materials. The ZrH moderator, in the form of close-packed rods, is located between reactor fuel assemblies and is very efficient in minimizing the inventory of fissile material in the reactor core.

The CIS fuel assembly (shown in Figure 6) is an axial flow design and contains a series of stacked 45 mm diameter bundles of thin (~1 mm) "twisted ribbon" fuel elements approximately 2 mm in width by 100 mm in length.

The "fueled length" and power output from each assembly is determined by specifying the engine thrust level and hydrogen exhaust temperature (or desired Isp).

For the 75 klbf (330,000 N) CIS engine design point indicated in Figure 4, 102 fuel assemblies (each containing 10 fuel bundles) produce ~1650 MWt with a Isp of ~960 s.

For a 15 klbf (67,000 N) engine, 34 fuel assemblies (with 6 fuel bundles each) are used to generate the required 340 MWt of reactor power at the same Isp.

The fuel material in each "twisted ribbon" element is composed of a solid solution of uranium, zirconium and niobium ceramic carbides having a maximum operating temperature expected to be about 3200 K. The fuel composition along the fuel assembly length is tailored to provide increased power generation where the propellant temperature is low and reduced power output near the bottom of the fuel assembly where the propellant is nearing its exhaust temperature design limit. In the present CIS design a value of 2900 K has been selected to provide a robust temperature margin. During reactor tests, hydrogen exhaust temperatures of 3100 K for over one hour and 2000 K for 2000 hours were demonstrated in the CIS.

At 2900 K, an engine lifetime of ~4.5 hours is predicted.

The Aerojet, Energopool, B&W NTR design utilizes a dual turbopump, recuperated expander cycle. Hydrogen flowing from each pump is split with ~84% of the flow going to a combination recuperator/gamma radiation shield and the remaining 16% used to cool the nozzle. The recuperator/shield, located at the top of the engine, provides all of the necessary turbine drive power. The turbine exhaust cools the reactor pressure vessel and is then merged with the nozzle coolant to cool the moderator and reflector regions of the engine. The coolant then passes through borated ZrH and lithium hydride (LiH) neutron shields located within the pressure vessel between the reactor core and the recuperator/gamma shield, before returning to the recuperator where it heats the pump discharge flow. Exiting the recuperator the cooled hydrogen is then routed to the core fuel assemblies where it is heated to 2900 K.

The 75 klbf (330,000 N) CIS engine design point has a chamber pressure of 2000 psia (14,000 kpa), a nozzle area ratio of 300 to 1, and a 110% bell length nozzle resulting in a Isp of ~960 s.

(ed note: from the chart, the 75 klbf CIS engine has a thrust-to-weight ratio of 6.4. If my slide rule is not lying to me, that means the engine has a mass of 5,260 kilograms)

The same pressure and nozzle conditions were maintained for the 15 (67,000), 25 (110,000) and 50 klbf (220,000 N) engine design points with the resulting weight scaling indicated in Figure 4.

The approximate engine lengths for the 15 (67,000), 25 (110,000), 50 (220,000) and 75 klbf (330,000 N) CIS engines are 4.3 m, 5.2 m, 6.5 m, and 7.6 m, respectively.

Low Pressure NTR

Low Pressure NTR
Engine Mass835 kg
Full Thrust49,000 newtons
Full T/W6.0
Full Isp1,210 sec
Single-H Thrust9,800 newtons
Single-H T/W1.2
Single-H Isp1,350 sec

This is from Low Pressure Nuclear Thermal Rocket (LPNTR) concept (1991)

This is a theoretical concept, but it has enough impressive advantages over conventional solid-core NTRs that it is well worth looking into. The engine has a specific impulse of up to 1,350 seconds (exhaust velocity 13,200 m/s) which is virtually the theoretical maximum for solid-core NTR. It also is very lightweight plus much more reliable. The latter is due to the absence of certain heavy and fault-prone components (those with moving parts) required for solid-core.

Solid-core NTRs commonly use liquid hydrogen as propellant, since that is the propellant with the sweet spot of low molecular weight and convenience. The lower the molecular weight, the higher the specific impulse and exhaust velocity.

There is one propellant with an even lower molecular weight, but it is anything but convenient. Monatomic hydrogen has half the molecular weight of molecular hydrogen so it has a much higher performance. A pity it explodes like a bomb if you give it a stern look. In his novels Robert Heinlein calls monatomic hydrogen "Single-H", and handwaves really hard that future engineers will figure out some way to stablize the dire stuff. Sorry Mr. Heinlein, we need a real-world solution here.

Heating molecular hydrogen to above 3,000 Kelvin will dissociate it into single-H. Sadly at the high pressures commonly used in solid-core reactors, the temperature and the propellant mass flow would combine into a heat flux high enough to destroy the reactor. Remember the difference between heat and temperature: temperature is an interesting number but it is the heat joules that ruin the reactor.

Dr. Ramsthaler said "Ah, but what if we designed the engine to use low pressure?" Then we can make single-H at a heat flux low enough for the reactor to survive, allowing our specific impulse will climb to amazing levels. A standard NERVA has an engine pressure of 31 bar (450 pounds force per square inch), the LPNTR only has a pressure of 1 bar (14.5 psia). This means the LPNTR has a heat flux that is 50-to-one less than the NERVA.

The drawback is the low pressure will drastically reduce the propellant mass flow, which reduces the thrust (because thrust = propellant mass flow times exhaust velocity). This problem can be addressed with clever engineering. Dr. Ramsthaler thinks it is possible to push the engine up to a thrust-to-weight ratio of 1.2. The Monatomic-H MITEE tries the same low-pressure trick, but only at a thrust-to-weight ratio of 1.0.

Everything comes at a cost. The engine can do a T/W ratio of 6.0 at full thrust, but this means the specific impulse is only 1,210 seconds. If you shift it into temperatures that allow dissociation to create Single-H, the T/W ratio is only 1.2 but the Single-H makes a specific impulse of 1,350 seconds. So the engine has two gears.

In addition, a low pressure engine means it does not need turbopumps to create high pressure. Turbopumps are penalty-weight, turbopumbs need complicated plumbing to supply the energy needed to spin the little darling, and turbopumps contain several points of mechanical failure with all their moving parts. Good riddance to bad rubbish. The natural propellant tank pressure is enough for the LPNTR to operate.

Also the low heat flux means the engine only needs an exhaust nozzle that is very short compared to a NERVA. 50-to-one less than the NERVA, remember?

Dr. Ramsthaler's secret is a reactor with a radial outflow core: it maximizes propellant mass flow at low pressure but high temperature. Remember:

  • High temperature is needed to make Single-H and crank up the specific impulse to 11, er, ah, 1,350 seconds
  • Low pressure counteracts the high temperature so the heat level is not high enough to melt the reactor
  • Maximizing propellant mass flow counteracts the low pressure so the thrust-to-weight ratio is at least 6.0

For standard NERVA and related solid-core NTRs, at low pressure the critical flow is where the propellant exits the core. The propellant enters the top of the cylindrical core, is heated inside the core, and exits the core at the bottom. Then it enters the exhaust nozzle.

Dr. Ramsthaler's design uses a spherical core. The propellant enters the center of the core, is heated inside the core, and exits the core from its surface. Given the 120 flow outlet holes on the surface, the engine has almost 50% flow area at the exit of the core.

The design can accommodate almost any kind of nuclear fuel elements: pebbles, plates, whatever.

Safety and reliabily was Dr. Ramsthaler's primary goal. But his solution to control of the nuclear reactor raises eyebrows.

Conventional NERVA engines use control drums to control the criticality in the nuclear reactor. Spin the drums so the neutron reflector face the nuclear fuel elements and the reactor fires up. Spin the drums so the neutron poison faces the fuel elements and the reactor shuts down like a blown-out match.

As it turns out, the liquid hydrogen propellant is a pretty good neutron moderator all by itself. The spacecraft engineer has to be careful about feeding propellant into a dry hot reactor. Otherwise neutron transients will build into full-fledged runaway nuclear oscillations and your reactor will go all Chernobyl on you. The addition of the moderator changes the nuclear characteristics of the reactor.

Anyway Dr. Ramsthaler looked at the way the propellant altered the reactor behaviour and wondered if careful propellant control could replace the control drums. Control drums are penalty-weight, control drums require electricity, and control drums contain several points of mechanical failure with all their moving parts. Using propellant to control the reactor would happily reduce the engine mass even more, and increase the engine reliabilty.

The hydrogen propellant is injected into the center of the spherical core, remember? This turns out to be the perfect location for the hydrogen to moderate the neutrons flux, where the neutrons are thickest. The hydrogen turns worthless fast neutrons into reactor-grade thermal neutrons which maintain the fission chain reaction.

The dry reactor just sits there, its nuclear characteristics are such that no chain reaction can happen. But as soon as the liquid hydrogen fills the center, the reactor goes critical and starts generating large amounts of thermal energy by the miracle of nuclear fission.

But just in case the reaction gets out of hand, there is a rod of neutron poison that can be slammed into the center of the core to scram the engine.

Dr. Ramsthaler figures with such low engine mass, the spacecraft could afford to have seven engines. This would allow thrust vectoring by throttling engines instead of the mechanical nightmare of gimbaled engines. All together now: engine gimbals are are penalty-weight, engine gimbals require hydraulics, and engine gimbals contain several points of mechanical failure with all their moving parts. Get rid of them.

Rob Davidoff points out that the above gimbal-less scheme will do yaw and pitch thrust vectoring just fine. But it is incapable of performing roll vectoring. A spacecraft using such a scheme will have to rely upon its reaction control system (attitude jets) for rolls.

In addition, a cluster of seven engines would allow the spacecraft to lose up to two engines and still limp through the mission ("two-engine-out" capability). Instead of total mission failure and all the crew dying.

LPNTR advantage IMEO for Mars mission
+ shield
500 KM
Earth Orbit
Ref NERVA85042.68841400
Adv NERVA92563.37131037
3200 K
3600 K
3600 K
Dual mode

The Ref mission is a Mars mission that ends with the spacecraft in a huge ecliptic orbit around Terra. This will require lots of energy when you want to reuse the spacecraft. The 500 KM Earth Orbit mission is the Mars mission, using extra propellant and delta V to end with the spacecraft in a nice circular orbit for easy spacecraft reuse.

You can see how the Initial Mass in Earth Orbit (IMEO) nicely drops as the engine Isp increases. And how using a dual-mode engine with the Single-H mode drops the IMEO by 77 metric tons of propellant compared to the single-mode engine.

Nuclear Thermal Rockets (NTR’s) once seemed like a fantastic idea to the Sci-Fi greats who were involved in the early days of the post-War Space Age.

In the UK Arthur C Clarke wrote of a two-stage hybrid nuclear rocket “Prometheus” for Moon missions in his “Prelude to Space(written in 1947, but published in 1951). To get the required performance Clarke fudged a little for the air-breathing nuclear Ramjet Beta component, hoping a high temperature system could be created to provide the required thrust. To get the required performance for the Moon return-mission, the Alpha component depended on methane heated to a high enough temperature to mostly dissociate into carbon and hydrogen and increase the exhaust velocity with a high atomic hydrogen fraction.

In the USA Robert Heinlein wrote many tales with Nuclear Thermal Rockets – his 1947 “Space Cadet” featured monatomic hydrogen sub-orbital rockets. By 1950 the first hard performance data of high temperature reactors was available and the first nuclear thermal rocket designs were appearing in the journals. His most detailed discussion was in “The Rolling Stones(1952) about a family on the Moon buying a second hand NTR spaceship to take to Mars, then the Asteroids and Saturn. The characters mention tanking up on monatomic hydrogen for the long range destinations. Back in the infancy of cryogenics, monatomic hydrogen seemed potentially stable over long periods of time. Using just H, rather than H2, increases the exhaust velocity of an NTR by 50% or so.

Atomic hydrogen doesn’t like being by itself and quickly recombines into molecular hydrogen all by itself. However once molecular hydrogen gas is hot enough, it starts breaking apart and this fact can improve exhaust velocity. That’s the basic physics idea behind the Low-Pressure Nuclear Thermal Rocket (LPNTR) which feeds the gas into a really hot reactor, but at a low enough pressure to minimise the recombination of the hot atomic hydrogen. The improvement is significant – the Specific Impulse jumps from NERVA’s 850-925 seconds to 1210-1350 seconds.

However making hydrogen is a non-trivial exercise – about 120 MJ/kg when making it from steam. Chemically free hydrogen is rare anywhere but the Sun and the Gas Giants. And one other place. Titan, which has an atmosphere that’s 0.1% hydrogen. While that doesn’t sound like much, it’s readily extracted and liquefied for a lot less energy than cracking it out of water ice. The total mass of H2 is about 675 billion tonnes. The delta-vee to launch to the Earth-Moon system from the surface of Titan is about 4.8 km/s – round it to 5 km/s to account for gravity losses. In energy terms roughly 12.5 MJ/kg. Using the high-thrust mode of the LPNTR, the mass-ratio is ~1.54, meaning the rocket can be mostly payload. For the long cruise back home, the reactor’s fission-product decay heat coupled by heat exchanger to a thermophotovoltaic system can run a cryo-cooler to keep the hydrogen chilled.

In terms of energy expenditure, sourcing hydrogen from Titan makes more sense than anywhere else in the Solar System. Hohmann Transfer Orbit launch windows open more frequently for Saturn-Earth, every 12.4 months, versus 26 months for Mars-Earth, thus more frequent delivery opportunities. The Transfer time is 6 years, but this can be scheduled for. Faster elliptical (~3 years), parabolic (~2 years) and hyperbolic orbits are possible for higher hydrogen expense, but for a well established automated delivery schedule the Hohmann transfer is sufficient.


Exhaust Velocity9,221 m/s
Specific Impulse940 s
Thrust67,000 N
Thrust Power0.3 GW
Mass Flow7 kg/s
RemassLiquid Hydrogen
Exhaust Velocity6,347 m/s
Specific Impulse647 s
Thrust184,000 N
Thrust Power0.6 GW
Mass Flow29 kg/s
RemassHydrogen + Oxygen
Uranium 235
ReactorSolid Core
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
SpecialLow-High Gear
Nuclear DC-X NERVA
High Gear
Exhaust Velocity9,810 m/s
Specific Impulse1,000 s
Thrust/Engine1,112,000 N
Thrust5,560,000 N
Thrust Power27.3 GW
Mass Flow567 kg/s
Specific Power7 kg/MW
Low Gear
Exhaust Velocity5,900 m/s
Specific Impulse601 s
Thrust/Engine3,336,000 N
Thrust16,680,000 N
Thrust Power49.2 GW
Mass Flow2,827 kg/s
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power4 kg/MW
Number Thrustersx5
Total Engine Mass199,600 kg
Uranium 235
SpecialLow-High Gear
Wet Mass460,000 kg

LOX-augmented Nuclear Thermal Rocket. One of the systems that can increase thrust by lowering Isp, in other words Shifting Gears. This concept involves the use of a "conventional" hydrogen (H2) NTR with oxygen (O2) injected into the nozzle. The injected O2 acts like an "afterburner" and operates in a "reverse scramjet" mode. This makes it possible to augment (and vary) the thrust (from what would otherwise be a relatively small NTR engine) at the expense of reduced Isp

Bi-Modal NTR

Bimodal NTR Solid (NASA)
Propulsion SystemNTR Solid Bimodal
Exhaust Velocity8,980 m/s
Specific Impulse915 s
Thrust/Engine66,667 N
Number Thrustersx3
Thrust200,000 N
Thrust Power0.9 GW
Mass Flow22 kg/s
Total Engine Mass6,672 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Wet Mass80,000 kg
Dry Mass26,830 kg
Mass Ratio2.98 m/s
ΔV9,811 m/s
Specific Power7 kg/MW

A useful refinement is the Bimodal NTR.

Say your spacecraft has an honest-to-Johnny NERVA nuclear-thermal propulsion system. Typically it operates for a few minutes at a time, then sits idle for the rest of the entire mission. Before each use, one has to warm up the reactor, and after use the reactor has to be cooled down. Each of these thermal cycles puts stress on the engine. And the cooling process consists of wasting propellant, flushing it through the reactor just to cool it off instead of producing thrust.

Meanwhile, during the rest of the mission, your spacecraft needs electricity to run life support, radio, radar, computers, and other incidental things.

So make that reactor do double duty (that's where the "Bimodal" comes in) and kill two birds with one stone. Refer to below diagram. Basically you take a NERVA and attach a power generation unit to the side. The NERVA section is the "cryogenic H2 propellant tank", the turbopump, and the thermal propulsion unit. The power generation section is the generator, the radiator, the heat exchanger, and the compressor.

Warm up your reactor once, do a thrust burn, stop the propellant flow and use the heat exchanger and radiator to partially cool the reactor to power generation levels, and keep the reactor warm for the rest of the mission while generating electricity for the ship.

This allows you to get away with only one full warm/cool thermal cycle in the entire mission instead of one per burn. No propellant is wasted as coolant since the radiator cools down the reactor. The reactor supplies needed electricity. And as an added bonus, the reactor is in a constant pre-heated state. This means that in case of emergency one can power up and do a burn in a fraction of the time required by a cold reactor.

Pretty ingenious, eh?

Most nuclear thermal rockets do not need heat radiators because they get by with open-cycle cooling. But bimodal engines do need radiators, which makes sense with a few moment's thought. While running in power-generation mode the rocket is not thrusting. No thrust means no rocket exhaust. And no rocket exhaust means no handly plume of gas to use for open-cycle cooling. So you need a physical radiator to take care of the waste heat created by electrical power generation.

An even further refinement is the Hybrid BNTR/EP option. This is where the electrical power output has a connection to an Ion Drive. This is a crude form of Shifting Gears: trading thrust for specific impulse/exhaust velocity. So it can do low-gear NTR thrust mode, high-gear ion-drive thrust mode, and no-thrust electricity generation mode while coasting.

And the Pratt & Whitney company went one step better. They took the Bimodal NTR concept and merged it with the LANTR concept to make a Trimodal NTR. Called the Triton, it uses a LANTR engine to allow Shifting Gears. So it can do low-gear NTR-Afterburner thrust mode, high-gear NTR thrust mode, and no-thrust electricity generation mode while coasting.


5.3.1 Bimodal NTR Mission Concept

An option to Reference Mission Version 3.0 (DRM 3.0) that utilizes bimodal NTR transfer vehicles in place of the expendable NTR stages is being evaluated. A common "core" stage, used on cargo and piloted vehicles alike, is outfitted with three 15 klbf bimodal NTR engines capable of providing up to 50 kilowatts of electrical power (kWe) using any two engines The bimodal core stage is not jettisoned after the Trans-Mars Insertion (TMI) maneuver but remains with the cargo and piloted payload elements providing midcourse correction (MCC) propulsion and all necessary power during transit. Near Mars, the bimodal stage separates from the aerobraked payloads and performs its final disposal maneuvers. A key difference between Reference Mission 3.0 and the bimodal option is the absence of the aerobraked LOX/methane (CH4) Trans-Earth Insertion (TEI) stage which is replaced by an "all propulsive" bimodal NTR-powered Earth Return Vehicle (ERV) illustrated in Figure A5-7.

The bimodal stage LH2 tank is slightly shorter than the expendable TMI stage tank at 19 meters and has a maximum LH2 propellant capacity of ~51 tons with a 3% ullage factor. A turbo-Brayton refrigeration system is located in the forward cylindrical adaptor section to eliminate LH2 boiloff during the lengthy (~4.3 year) ERV mission. A 12 kWe Brayton refrigeration system is included to remove the ~100 watts of heat flux penetrating the 2 inch MLI system in low-Earth-orbit where the highest heat flux occurs. Enclosed within the conical aft radiator section of the bimodal core stage is a closed Brayton cycle (CBC) power conversion system employing three 25 kWe Brayton rotating units (one for each bimodal reactor) which operate at ~2/3 of rated capacity, thus providing an "engine out" capability. The turbine inlet temperature of the He-Xe working gas is ~1300 K and the total system specific mass is estimated to be ~30 kg/kWe.

A mass comparison of the bimodal NTR transfer vehicles and the Reference Mission Version 3.0 vehicles is shown in Table A5-1.

The mass values assume a "2-perigee burn" Earth departure scenario. Overall, the bimodal approach has a lower "three-mission" initial mass than Reference Mission 3.0. In addition, the bimodal approach can reduce the operational complexity of the mission (eliminates solar array deployment/retraction) as well as eliminating the need for an aerobrake and injection stage for the Earth Return Vehicle.

5.3.2 All Propulsive" Bimodal NTR Option Using TransHab

Another option to the Reference Mission 3.0 under consideration is the use of a bimodal NTR stage to propulsively capture all payload elements into Mars orbit. This "all propulsive" NTR option provides the most efficient use of the bimodal engines which can supply abundant power to the spacecraft and payloads in Mars orbit for long periods. Propulsive capture into the reference "250 km by 1 sol" elliptical Mars parking orbit also makes possible the use of a standardized, reduced mass "aerodescent" shell because of the lower payload entry velocity (~4.5 km/s) encountered. From this orbit, the triconic aerobrake mass varies by only ~400 kg for a 20 ton increase in payload mass (see Section 3.3.3).

The attractiveness of the "all propulsive" bimodal NTR option is further increased by the utilization of the lightweight, inflatable "TransHab" module discussed in Section 3.1. The substitution of TransHab for the heavier, hard-shell habitat module introduces the potential for propulsive recovery of the Earth Return Vehicle in Earth orbit and its reuse on subsequent missions. TransHab use also allows the crew to travel to and from Mars on the same bimodal transfer. In Mars orbit, the crew transfer vehicle rendezvous with the "unpiloted" habitat lander which is now delivered as a cargo element by the bimodal stage. The absence of crew from the bimodal habitat lander eliminates the need for outbound consumables and engine crew radiation shields and allows it to carry off-loaded surface habitation and science equipment previously carried on the cargo lander.

A three-dimensional image of the bimodal transfer vehicle used on the piloted mission is shown in Figure A5-8. The TransHab is ~9.7 meters long and inflates to a diameter of ~9.5 meters. Its total mass is ~24.3 metric tons which includes the crew and their consumables. The total length and initial mass of the piloted transfer vehicle is ~54 meters and ~141 metric tons, respectively. A smaller, "in-line" propellant tank is used on the bimodal transfer vehicles that deliver the ~46 metric ton habitat and ~54 ton cargo landers into Mars orbit. The habitat and cargo transfer vehicles are ~56 meters long and have a LEO mass of ~129 metric tons and 144 metric tons, respectively.

Dual-mode Fission
Exhaust Velocity9,810 m/s
Specific Impulse1,000 s
Thrust124,700 N
Thrust Power0.6 GW
Mass Flow13 kg/s
Total Engine Mass33,000 kg
Thermal eff.94%
Total eff.94%
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power54 kg/MW
Thermal Electrical eff.19%
Electrical Power60 MWe

When struck by a thermal neutron, a fissile nuclide splits into two fragments plus energy. For example, the fission of the 235U atom produces 165 MeV of energy plus 12 MeV of neutral radiation (gammas and a couple of fast neutrons). The fast neutrons must be thermalized by a low Z moderator (a surrounding blanket of about 80 cm of D2O, Be, liquid or gas D2, or CD4), which returns enough thermal neutrons to the core to sustain the chain reaction. (Thermal neutrons diffuse through the reactor like a low pressure gas.) Alternatively, a molybdenum neutron reflector can be used. Much of a reactor’s mass is constant, regardless of power level. Therefore, nuclear power sources are more attractive at higher power levels.

The 650 MWth system illustrated is dual mode, which can either generate electricity, or directly exhaust coolant for thrust. It uses a fast reactor with fuel tubes interspersed with cooling tubes. The coolant is lithium, which for electrical power is passed to a potassium boiler at 1650 K. The potassium vapor is passed to a static (AMTEC) or dynamic (turbine) heat engine for power generation (60 MWe), or heats hydrogen in a heat exchanger for thrust (125 kN at a specific impulse of 1 ks). The thermal efficiency is 19% if closed-cycle (for power generation) or 94% if open-cycle (for thrust).

From High Frontier by Philip Eklund
Pebble-bed Fission Reactor
Exhaust Velocity9,810 m/s
Specific Impulse1,000 s
Thrust172,700 N
Thrust Power0.8 GW
Mass Flow18 kg/s
Total Engine Mass58,000 kg
Thermal eff.94%
Total eff.94%
Uranium 233
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power68 kg/MW
Electrical Power60 MWe

This is a graphite-moderated, gas-cooled, nuclear reactor that uses spherical fuel elements called "pebbles". These tennis ball-sized pebbles are made of pyrolytic graphite (which acts as the moderator), interspersed with thousands of micro fuel particles of a fissile material (such as 235U).

In the reactor illustrated, 360,000 pebbles are placed together to create a 120 MWth reactor. The spaces between the pebbles form the "piping" in the core for the coolant, either propellant or inert He/Xe gas.

The design illustrated can is dual mode. It can operate either as a generator for 60 MWe of electricity, or act as a solid-core thruster using hydrogen propellant/coolant expelled at a specific impulse of 1 ksec. When used as a thruster, it offers a slight increase in specific impulse but significant acceleration benefits over traditional fission reactors. Moreover, the high temperatures (up to 1900 K) allow higher thermal efficiencies (up to 50%).

From HIGH FRONTIER by Philip Eklund


MInature ReacTor EnginE. MITEE is actually a family of engines. These are small designs, suitable for launching on existing boosters. You can find more details here.

Exhaust Velocity9,810 m/s
Specific Impulse1,000 s
Thrust14,000 N
Thrust Power68.7 MW
Mass Flow1 kg/s
Total Engine Mass200 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power3 kg/MW

The baseline design is a fairly conventional NTR. Unlike earlier designs it keeps the fuel elements in individual pressure tubes instead of a single pressure vessel, making it lighter and allowing slightly higher temperatures and a bit better exhaust velocity.

Monatomic H
Monatomic-H MITEE
Exhaust Velocity12,750 m/s
Specific Impulse1,300 s
Thrust2,350 N
Thrust Power15.0 MW
Mass Flow0.18 kg/s
Total Engine Mass200 kg
Uranium 235
ReactorSolid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power13 kg/MW

This advanced design works at a lower chamber pressure so that some of the H2 propellant disassociates into monatomic hydrogen, although the chamber temperature is only slightly greater. The drawback is that this reduces the mass flow through the reactor, limiting reactor power.

Exhaust Velocity17,660 m/s
Specific Impulse1,800 s
Thrust1,700 N
Thrust Power15.0 MW
Mass Flow0.10 kg/s
Total Engine Mass10,000 kg
Uranium 235
ReactorSolid Core
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power666 kg/MW

The use of individual pressure tubes in the reactor allows some fuel channels to be run at high pressure while others are run at a lower pressure, facilitating a hybrid electro-thermal design. In this design cold H2 is heated in the high pressure section of the reactor, is expanded through a turbine connected to a generator, then reheated in the low pressure section of the reactor before flowing to the nozzle. The electricity generated by the turbine is used to break down more H2 into monatomic hydrogen, increasing the exhaust velocity. Since this is a once through system there is no need for radiators so the weight penalty would not be excessive.

Liquid Core

Liquid Core 1
Exhaust Velocity16,000 m/s
Specific Impulse1,631 s
Thrust7,000,000 N
Thrust Power56.0 GW
Mass Flow438 kg/s
Total Engine Mass70,000 kg
Uranium 235
ReactorLiquid Core
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power1 kg/MW
Liquid Core 2
Exhaust velocity14,700 to 25,500 m/s

Nuclear thermal rocket / liquid core fission. Similar to an NTR-GAS, but the fissionable core is merely molten, not gaseous. A dense high temperature fluid contains the fissionable material, and the hydrogen propellant is bubbled through to be heated. The propellant will be raised to a temperature somewhere between the melting and boiling point of the fluid. Candidates for the fluid include tungsten (boiling 6160K), osmium (boiling 5770K), rhenium (boiling 6170K), or tantalum (boiling 6370K).

Liquid core nuclear thermal rockets have a nominal core temperature of 5,250 K (8,990°F).

The reaction chamber is a cylinder which is spun to make the molten fluid adhere to the walls, the reaction mass in injected radially (cooling the walls of the chamber) to be heated and expelled out the exhaust nozzle.

Starting up the engine for a thrust burn will be complicated and tricky, shutting it down even more so. Keeping the fissioning fluid contained in the chamber instead of escaping out the nozzle will also be a problem.


Exhaust Velocity19,620 m/s
Specific Impulse2,000 s
Thrust20,000 N
Thrust Power0.2 GW
Mass Flow1 kg/s
Total Engine Mass1,000 kg
Uranium 235
ReactorLiquid Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power5 kg/MW
Propulsion SystemLARS
Exhaust Velocity10,300 m/s
Specific Impulse1,050 s
Thrust11,000,000 N
Thrust Power56.6 GW
Mass Flow1,068 kg/s
Total Engine Mass9,000 kg
Uranium 235
Thrust DirectorNozzle
Wet Mass226,000 kg
Dry Mass45,000 kg
Mass Ratio5.02 m/s
ΔV16,623 m/s

Nuclear thermal rocket / liquid annular reactor system. A type of NTR-LIQUID. You can find more details here

The molten fissioning uranium is held in tubes which are spun to provide centifugal gravity. This keeps the uranium from escaping out the exhaust, mostly. Seeded hydrogen propellant is injected down the spin axis where it is heated by the nuclear reaction then escapse out the exhaust nozzle.

These engines have a specific impulse ranging between 1,600 to 2,000 seconds, and an internal temperature between 3,000K and 5,000K

Centrifugal NTR

Centrifugal NTR
1,500 to
1,800 sec
Exhaust Velocity
14,700 to
17,700 m/s
PropellantsHydrogen (LH2)
Ammonia (NH3)
Methane (CH4)
Water (H2O)


     The Centrifugal Nuclear Thermal Rocket (CNTR) is one of few designs that could enable extremely rapid missions to Mars using currently available technologies. McCarthy conducted the first conceptual study of high performance nuclear thermal propulsion (NTP) and published his findings in 1954[1]. High performance NTP was further investigated by Princeton researchers in the early 1960s[2] and continued by other researchers throughout the 60s, 70s, and 80s. An interagency panel conducted in 1991 demonstrated the potential of liquid core nuclear rockets, such as the Liquid Annular Reactor System (LARS), to reach temperatures up to 5000 K and specific impulses (Isp) up to 2000 s[3]. A more recent study of versatile NTP asserts that a similar propulsion system, the Centrifugal Gas Core Reactor (CGCR), can reach an estimated Isp of 1800 s[4]. An Isp of that magnitude significantly reduces travel times and consequently health risks to flight crews[4]. The CNTR seeks to build on the work of previous liquid core NTP systems and aims to reach an Isp in the range of 1500 s to 1800 s while using hydrogen as the propellant. However, the CNTR is not limited to hydrogen and can instead utilize other volatiles such as ammonia, methane, or water at about half the Isp of hydrogen[4]. This flexibility expands the CNTR’s mission range considerably by providing propellant storability and the potential for directly using volatiles available in-situ.

CNTR Design and Propellant Flow Path

     Before closely examining specific interfaces of the CNTR system, it is important to understand the CNTR design and propellant flow path. The CNTR utilizes a matrix of centrifugal fuel elements (CFEs) like the 19 CFE core block configuration displayed in Fig. 1. Each CFE consists of a molten uranium fuel source contained in a fritted centrifuge. The fritted surface is located along the length of the centrifuge and allows for radial propellant flow into the molten uranium. The 19 CFE configuration is tentative as further studies of the system’s neutronics identify more mass and energy efficient CFE sizes and layouts. Another possible layout involves using several solid core rods to preheat the propellant to temperatures around 1200 K to 1700 K before travelling into a single, larger CFE where the propellant is “superheated” to the desired temperature above 5000 K. This configuration could possibly improve propellant separation and heat transfer, but, for the purposes of this study the 19 CFE matrix will be discussed.

     The propellant flow in the CNTR is multipurposed. Before being heated by the molten uranium fuel source and ejected for thrust, the propellant is used to cool various system components and power the centrifuge rotation. The propellant flow path is depicted in Fig. 2. For clarity, the propellant being used in the following description is hydrogen but can be replaced by the other mentioned propellants. The hydrogen exits its storage vessel as a gas, enters the inlet plenum at the top of the assembly, and flows around the various external components of the system. This brings the propellant closer to the CFEs while also cooling the moderator block, radial reflectors, core drums, and other system components outside the CFEs to an acceptable temperature under 800 K. Then the hydrogen flows towards the nozzle end of the configuration, enters the gas inlet manifold, cools the nozzles, penetrates the moderator block via access paths below the centrifuges, and travels axially along the outside of the centrifuges.

     The propellant then passes the turbines, generating power for the centrifuge. Past the turbines, the hydrogen path is rerouted by the redirection disk into perforations at the top of the centrifuge, as shown in Fig. 2. The hydrogen pushes axially downward, this time between the centrifuge outer wall and fritted silicon carbide layer. The hydrogen then passes through the fritted surface into the liquid uranium. Inside the liquid uranium, the hydrogen will be heated to a goal temperature exceeding 5000 K before entering the center cavity created by the centrifuge rotation. The hydrogen will then exit at a high exhaust velocity out the nozzle and create thrust.

CFE Temperature Profile

     Each CFE undergoes nuclear fission to heat a propellant that is ejected out the nozzle, generating thrust. Thrust is related to exit velocity by equation [1], and exit velocity is related to exit temperature by equation [2].

Here, F, , Ve, pe, po, Ae, Me, γ, R, and Te are thrust, mass flow rate, exit velocity, exit pressure, free stream pressure, exit area, exit Mach, specific heat ratio, gas constant, and exit temperature, respectively. Thus, maximizing the propellant temperature increases the exit velocity and thrust.

     By utilizing a primarily liquid core system, the total heat transfer to the propellant is much higher than in a solid core system system. A primarily liquid core element can push propellant temperatures to values exceeding 5000 K. While 5000 K is above the boiling point of molten uranium, the desired outlet pressure and centrifuge pressure of the system is expected to exceed the vapor pressure of the molten uranium, reducing uranium vaporization and increasing uranium condensation. Fig. 3 shows the desired temperature profile within a single CFE cross section.

Molten Uranium Leakage and Entrainment

     While using liquid core elements increases the maximum temperature of the system, it poses many unique challenges such as uranium loss from leakage and entrainment. To counteract liquid uranium leakage from the nozzle, each fuel element is contained within a centrifuge, creating a CFE. The centrifuge rotates at a constant angular velocity and forms a vortex that displaces the liquid uranium towards the outer edges of the CFEs. This also creates a pocket for propellant flow at the center of each CFE. However, while the propellant flows radially from the outer edge of the CFE into the center of the CFE, it is possible that some uranium liquid droplets, or even small amounts of uranium vapor, can become entrained in the propellant flow. To reduce uranium loss from entrainment, the ends of the CFE will be tapered. This rotating tapered region acts as a net, catching the liquid uranium droplets and uranium vapor as depicted in Fig. 4. This net area will be maintained at a lower temperature than the rest of the CFE, allowing condensation of the uranium vapor. The centrifugal force will push the condensed uranium vapor and liquid uranium droplets back into the main liquid uranium region, reducing or possibly eliminating uranium entrainment.

     Maintaining the desired temperature distribution and reducing entrainment to acceptable levels will be a challenge due to the complex discrete-continuous fluid dynamics and lack of material properties in the extreme temperatures of the system. To begin to address the complex system, a series of experiments purposed with breaking down the complex propellant-uranium interface will be discussed in the following sections. These experiments focus on creating a reliable propellant flow channel in the center of each CFE, understanding propellant dynamics within the uranium, and quantifying the effects of temperature profiles within each CFE.

Material Analogs

     As discussed, a series of simplified experiments for development of the CNTR are outlined in the following sections. To reduce safety complications and costs, early experiments will use more readily available materials comparable to the final hydrogen-uranium system. Initially, gaseous nitrogen will replace the hydrogen as the propellant, and water will replace the uranium as the liquid medium. While these materials are not the best representations of the final system, the material analog serves multiple purposes. The primary reason is that conducting initial experiments with the nitrogen-water system will be safer while also providing key information about the propellent-liquid interface physics. Additionally, nitrogen and water are easily obtained and stored. This is essential to the workflow of the project, as it is likely that many revisions will need to be made over an extended time period.

     The initial results obtained from these experiments illuminates the behavior of discrete-continuous fluid dynamics that can also be applied to the hydrogen-uranium interface, albeit with some inaccuracies. The system behavior can then be modeled with a computational fluid dynamic (CFD) software package such as OpenFOAM or ANSYS Fluent and validated with the nitrogen-water system. These models can then be adjusted to a more advanced, testable system, such as nitrogen and molten tin or gallium. This will more accurately reflect a gas and molten metal interface and provide information on the setup, testing, and performance of the hydrogen-uranium system.


     There are many interactions in the proposed system and using an iterative testing scheme to add layers of complexity will aid in long term progression of the design. The first task will focus on accurately predicting surface profiles when rotating a CFE at different angular velocities. Increasing angular velocity will cause an increase in centrifugal pressure. In an environment with gravity, the centrifugal pressure will combat hydrostatic pressure and create a parabolic profile as seen in forced vortex behavior. In an environment without gravity, centrifugal pressure is the only force acting on the liquid. Thus, the liquid surface profile created is uniform because the centrifugal pressure is independent of the height of the liquid.

     During use in a rocket, however, the thrust acting on the system will create an artificial “gravity” that will likely be weaker than earth’s gravity, but still generate some hydrostatic pressure. As described previously, the bottom portion of the centrifuge will be designed as a conical section to act as a net, capturing some of the entrained uranium liquid or vapor created by turbulent flow of the propellant through the uranium layer, but may also help reduce the amount of uranium flowing out of the system due to the forces generated by thrust.

     This initial experiment will provide data on water surface profiles and facilitate the creation and validation of a CFD model in ANSYS Fluent. If this model can reliably predict the surface profiles generated, then the model can be applied again but with a system that more accurately reflects the uranium-propellant system. Additionally, this model can become the baseline for future models that will monitor entrainment based on different propellant inlet characteristics. Another area of interest will be the startup and shutdown processes for the centrifuge.

     A prototype testing apparatus has been designed, and materials are being collected for assembly. The current prototype is displayed in Fig. 5. The outlet is currently oriented upward to prevent water from pouring out of the centrifuge prior to startup. The taper on the upper end of the cylinder forms the conical condensation section that will be used to reduce entrainment but is not particularly important for the initial experiment. The water in the centrifuge is expected to form a surface profile according to the following forced vortex equation, equation [3].

The variables z, z0, ω, r, and g represent height of fluid at a certain radius, lowest point of the vortex, angular velocity, radius, and gravitational acceleration, respectively.

     An example of case of this system with an angular velocity of 50 rad/s is shown in Fig. 6. Equation [3] predicts that at an angular velocity of 50 rad/s, the height of the fluid at the centrifuge wall relative to the bottom of the vortex should be 31.8 cm. A CFD model of this system constructed in ANSYS Fluent using a coupled k-ω Shear Stress Transport turbulence equation and a volume of fluids equation predicts the height of the fluid at the centrifuge wall relative to the bottom of the vortex to be 27.0 cm, as shown in Fig. 7.


     The experiments to follow aim to create a one-way flow for radial propellant injection along the fritted surface. The current proposed method involves using a hydrophobic paper or spray. Once the propellant is reliably injected into the liquid with minimal backflow, the resulting bubbling or jetting behavior will be examined and quantified. Current models on bubble versus jet behavior in liquid core fuel elements are sparse in literature. The conceptual analysis conducted by Princeton in the early 1960s assert that the experimentally observed bubble shape in this sort of system is a spherical cap[2]. The current proposed method involves using a clear centrifuge and some visualization techniques and equipment to evaluate the bubble behavior. Varying the pore size and density, propellant entry velocity, and the method of application for the hydrophobic materials are all expected to affect the bubble behavior and size. Bubble behavior and size are important aspects of CFD multiphase calculations and obtaining this data will be essential to the accuracy of the models. These results will help deduce propellant action mechanisms in the propellant-uranium interface.

     The final experiment to be discussed in this summary centers around the system behavior in increasing temperatures. Changing temperature profiles within the system will help quantify the heat transfer between the propellant and the liquid medium. It will also change the density and viscosity of the gaseous and liquid phases which will affect the discrete-continuous fluid interface although it is unclear to what degree. The effects of temperature change will be especially prominent with the molten metals due to the dramatic changes in viscosity. Other factors to consider will be bubble behavior, flow patterns, and exhaust temperatures for the propellant. The results from this task are essential, as there is little data on propellant-uranium interactions at high temperatures. The data collected with the nitrogen-water experiment will be essential to predicting the behavior of the uranium as the temperatures begin to reach levels of 3000 K – 5500 K.


     In summary, the CNTR is a complex system with many design challenges, particularly in the propellant-uranium centrifuge interface. Currently, there are several assumptions that will need to be confirmed or adjusted based on experiment results. One of the most important challenges in the CNTR is minimizing entrainments. The described basic experiments will help illuminate the physics of the system and begin to answer some prominent questions. One of these questions is how much vaporization can be expected? Another is how much cooling and area will be needed to capture any uranium that has vaporized? Will the vapor pressure be smaller than the centrifuge and outlet pressure and if so can the model be reduced to only three phases: gaseous hydrogen, solid uranium, and liquid uranium? All these questions are essential to the formation of accurate models and the efficacy of the CNTR system. While the system does have some complex interactions, the overall concept is simple. By iteratively improving experiments such as the ones proposed in this paper, an accurate model representing the CNTR can be used to predict and optimize the final system’s behavior. This will accelerate the CNTR’s development and result in reduced mission times and increased mission success rates.


     This work was supported by Universities Space Research Association, NASA Marshall Space Flight Center, and Argonne National Laboratory.


  1.      J. MCCARTHY, “Nuclear Reactors for Rockets,” American Rocket Society, Vol. 24, No. 1, p. 36, (1954)

  2.      S. T. Nelson, “Conceptual Design Study of a Liquidcore Nuclear Rocket,” Report No. 665, Princeton University, (1963).

  3.      J. S. CLARK, P. MCDANIEL, S. HOWE, I. HELMS, M. STANLEY, “Nuclear Thermal Propulsion Technology: Results of an Interagency Panel in FY 1991,” NASA-TM-105711, NASA, (1993)

  4.      M. HOUTS, C. R. JOYNER, J. ABRAMS, J. WITTER, P. VENNERI, “Versatile Nuclear Thermal Propulsion,” IAC 70, Washington D.C., Oct. 21-25, 2019. International Astronautical Congress (2019)


Droplet Core

Droplet Core
Reactor inner diameter1 m
Reactor outer diameter2 m
Reactor inner length3 m
Reactor outer length4 m
Engine length13 m
(no shadow shield)
(with shadow shield)
Engine mass
(no shadow shield)
6,800 kg
Engine mass
(with shadow shield)
21,200 kg
Engine pressure500 atm
Internal temp6,000K
Isp2,000 sec
Exhaust velocity19,600 m/s
Engine power1,500 MWth
Thrust333,000 N

The data is from Droplet Core Nuclear Rocket (1991).

The main draw-back is that developing such an engine will be just as hard as developing a gas core nuclear thermal engine. But it has much lower performance. So why bother?

This propulsion system straddles the line between liquid-core and vapor-core. Much like how vapor-core straddles the line between liquid-core and gas-core. Instead of the uranium fuel being in the form of gaseous vapor, it is instead in the form of a fog of droplets.

Droplet core engines have a specific impulse between 1,500 and 3,000 seconds and an internal temperature between 5,000K and 7,000K. The specific impulse is enhanced because the nuclear energy is strong enough to dissociate some (20%) of the hydrogen molecules of propellant into atomic hydrogen. The propellant flow rate can be between 1 to 1,000 kilograms per second.

The temperature depends upon the pressure inside the chamber. The design shown assumes a pressure of 500 atmospheres, where the melting point of uranium is 1,400K and the boiling point is 9,500K. This is enough to heat the hydrogen propellant to 6,000K and gives a specific impulse of 2,000 seconds.

The chamber is about one meter in diameter and three meters tall.

At the top molten uranium with a temperature of around 2,000K is injecting through the unfortunately named "atomizer." In this case the term has nothing to do with nuclear physics, but more to do with Victorian perfume spray bottles. The droplets are from five to ten microns in size, and enough are sprayed into to create a critical mass. The upper 1.5 meters of the chamber is clad in neutron reflectors, so about 70 to 80% of the power generated occurs here. The next meter has only partial neutron reflectors, and the lower half meter has no neutron reflectors at all. Naturally the neutron flux is highest in the part with the most reflectors.

In the upper half of the chamber hydrogen propellant bleeds in from the walls, but in the lower half high pressure tangential jets spray a flood of hydrogen. Like vapor-core and open-cycle-gas-core the frantically fissioning uranium is intimately mixed with the hydrogen propellant. This gives an almost three orders of magnitude improvement on heat transfer area (i.e., about a thousand times better than a solid-core nuclear engine). The propellant is heated not only by heat radiation, but also by heat conduction of hydrogen gas in direct contact with the uranium drops. A whopping 30% to 40% of the fission energy is transferred to the propellant.

The tangential spray in the lower half of the chamber does two things: [1] help keep the blasted uranium drops from splattering on the walls and [2] create a vortex that will assist capturing uranium so it can be re-used instead of losing it out the exhaust nozzle. That stuff is both deadly and expensive, you don't want any un-burnt uranium escaping. The report calculates that the uranium loss will be less than 50 kilograms per mission.

About half a meter from the bottom of the chamber the tangential hydrogen jets are replaced with molten lithium-6 jets. The vortex makes the hot uranium drops hit the relatively cool lithium layer. This chills the uranium so the drops mix with the lithium. The mixture is captured at the bottom and sent to a fuel separator. The unburnt uranium is sent back to the top for another trip through the chamber while the lithium is sent back to the lithium jets.

The engine has a very high thrust-to-weight ratio. A 1,500 MWth engine with 333,000 Newtons of thrust would have a T/W of 5.0. Though actually that drops to 1.6 once you add the radiation shadow shield so the crew doesn't die. If my slide rule is not lying to me, this means the described engine has a mass of 6.8 metric tons with no radiation shield, and a mass of 21.2 metric tons with (or a shield mass of 14.4 metric tons).

This particular engine would have about 20 kg of uranium in the reaction chamber at any given time, and 100 kg total fuel. As mentioned before the report predicts it will lose about 50 kg out the exhaust nozzle over an entire mission.

Vapor Core

Vapor Core
Thrust Power1.6 GW
Exhaust velocity9,800 to
11,800 m/s
Thrust330,000 n
Propellant mass flow30 kg/sec
Reactor thermal power1,400 to
1,800 MW
Total engine mass6.83 tonne
Fuel element mass total1.35 tonne
Forward reflector mass0.60 tonne
Aft reflector mass0.51 tonne
Radial reflector mass2.47 tonne
Radiation shield mass0.9 tonne
Total reactor mass5.83 tonne
Misc. engine
component mass
0.9 tonne
Uranium Hexafluoride
ReactorVapor Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power4 kg/MW

This is sort of an intermediate step in learning how to design a full-blown Gas Core Nuclear Thermal Rocket. It is basically a solid core NTR where the solid nuclear fuel elements are replaced by chambers filled with uranium235 tetrafluoride vapor. The engine is admirably compact with a nicely low critical mass, and an impressive thrust-to-weight ratio of 5-to-1. However the specific impulse / exhaust velocity is only slightly better than a solid core.

In other words, the system is not to be developed because it has fantastic performance, but because it will be an educational step to building a system that does.

The specific impulse is around 1,280 seconds and the internal temperature is between 6,000K and 8,000K.

The uranium fuel is kept physically separate from the hydrogen propellant, so the exhaust is not radioactive.

A 330,000 newton thrust NVTR would have a core with almost 4,000 fuel elements, with a core radius of 120 cm, core height of 150 cm, and 1,800 MW. Criticality can be achieved with smaller cores: a core volume five times smaller with radius of 60 cm, height of 120 cm, and power of 360 MW.

Data is from Conceptual Design of a Vapor Core Reactor Rocket Engine for Space Propulsion by E.T. Dugan, N.J. Diaz, S.A. Kuras, S.P. Keshavmurthy, and I. Maya (1996).

ForwardBeryllium oxide15 cm0.60 tonne
AftC-C Composite25 cm0.51 tonne
RadialBeryllium oxide15 cm2.47 tonne
CORE: 2000 fuel elements
Radius0.5 m
Height1.5 m
Fuel channel per element12 to 32
Hydrogen channel per element12 to 32
Critical mass20 kg
Hydrogen pressure100 atm
UF4 pressure100 atm
Fuel center temperature4,500 K
Design Values
Pump Flowrate (Total)75.20 lbm/s
Pump Discharge Pressure3,924 psia
Pump Efficiency80.01%
Turbopump RPM70,000 RPM
Turbopump Power (each)9,836 HP
Turbine Inlet Temperature481 deg-R
Turbine Pressure Ratio1.69
Turbine Flow Rate (each)33.77 lbm/s
Reactor Thermal Power1,769 MW
Fuel Element and Reflector Power1,716 MW
Nozzle Chamber Temperature5,580 deg-R
Chamber Pressure (Nozzle Stagnation)1,500 psia
Nozzle Expansion Area Ratio500:1
Vacuum Specific Impulse (Delivered)997.8 sec
Heat Loads
Nozzle-con (total)30.05 MW
Nozzle-div (total)22.97 MW
Reflector (total)35.0 MW
Typical NVTR Engine Parameters
Nozzle Area Ratio500
Fuel Pressure100 atm
Average Fuel Temperature4000 K
Maximum Element Heat Flux420 W/cm2
Nomial Element Length150 cm
Fuel Volume Fraction0.15
Coolant Volume Fraction0.15
Moderator Volume Fraction0.70
Fuel Element Power0.9 MWt
Element Heat Transfer Area2141 cm2
Reactor Core L/D1.5
Fuel Channel Diameter0.142 cm
Fuel Channel Sectional Area0.0158 cm2
Total Fuel Channel Area Per Element0.505 cm2
Fuel Element Sectional Area3.464 cm2
Element Diameter (across flats)2.2 cm
Coolant Channel Diameter0.142 cm
Coolant Channel Sectional Area0.0158 cm2
Total Coolant Channel Area Per Element0.505 cm2
Core Volume1.2 m3
Core Volume Density1,500 MW/m3
Fuel Element Mass, Total1.35 MT
Forward Reflector Mass0.60 MT
Aft Reflector Mass0.51 MT
Radial Reflector Mass2.47 MT
Radiation Shield Mass0.90 MT
Total Reactor Mass5.83 MT
Misc. Engine Components Mass0.9 MT
Total Engine Mass6.83 MT
Engine F/W5.0


Colloid-Core NTR
PropellantLiquid Hydrogen
FuelUranium 235 + Zirconium Carbide
Thrust445,000 N
Low Specific Impulse
Exhaust Velocity11,300 m/s
Specific Impulse1,150 sec
Uranium Fuel Loss20 kg/min
Engine Temp3,620°K
Engine Pressure76 atm
High Specific Impulse
Exhaust Velocity11,800 m/s
Specific Impulse1,200 sec
Uranium Fuel Loss28 kg/min
Engine Temp3,950°K
Engine Pressure200 atm

This is from Research on Uranium Plasmas and their Technological Applications (1971). Page 29: The Colloid-Core Concept—A Possible Forerunner for the Gaseous Core.

Gas-Core nuclear-thermal-rocket engines are attractive since they are free from the exhaust velocity limits of solid-core engines. But great galloping galaxies, are those gas-core engines ever an engineering nightmare to design!

So Drs. Stefanko and Dickson of the Westinghouse Electric Corp. Astronuclear Laboratory had the thought it might be worth-while to develop some kind of forerunner engine to the full gas core. An engine whose thrust and specific impulse-exhaust velocity was only mid-way between solid-core and gas-core, but was a heck of a lot easier to design. This will allow spacecraft with nuclear engines superior to solid core NTR to be made now, and give the engineers some breathing room to figure out how to make a real live honest-to-Johnny gas-core engine.

Chemical rocket engines have an exhaust velocity that maxes out at about 4,400 m/s. A good solid-core NTR can double that, about 8,800 m/s. A good open-cycle NTR can manage an exhaust velocity of something like 40,000 m/s without even trying.

Stefanko and Dickson figured out an engine that can manage an exhaust velocity of about 11,800 m/s, with a thrust of around 445,000 Newtons. Which was more than a solid-core, less than a gas-core, and much simpler to engineer.

The researcher found two major difficulties in engineering gas-core NTR: nuclear fuel loss and large engine size. Both of these are because the core is gaseous.

Nuclear Fuel Loss

Since rocket engines need a hole in the bottom to let the exhaust out, the challenge is to allow all of the exhaust to escape but none of the uranium or plutonium. This is tricky since both the uranium fuel and the hydrogen propellant is gaseous, at an incredibly high temperature + pressure, moving at high velocity, and being held in the functional equivalent of an upside-down coffee mug.

You want the propellant to escape because that makes the thrust, which is the entire function of a rocket engine. You do NOT want the uranium fuel to escape because [1] it turns the exhaust plume into a cloud of glowing radioactive death and [2] that stuff is expensive, you can't afford to let any un-burnt uranium to get away.

Solid-core NTR prevent the fuel from escaping by making it non-gaseous, at a pathetically low temperature + pressure, not moving at all, and functionally bolted to the upside-down coffee mug so it can't fall out. The price is a lack-luster exhaust velocity.

Closed-cycle gas-core NTR try to confine the gaseous, high temperature + pressure + velocity uranium fuel inside a "lightbulb" to prevent it from fallout out of the upside-down coffee mug. The difficulty is to make a lightbulb (composed of matter) that can withstand the incredibly high temperature and pressure, and allow the heat radiation to leave the lightbulb so it can warm up the propellant. This is trying to have it both ways, since the entire point of a gas-core nuclear rocket was to have the core be gaseous. Having a non-gaseous lightbulb is breaking the rules. In order to prevent the lightbulb from instantly exploding and vaporizing the pressure and the temperature have to be lowered to the point where the exhaust velocity is about half that of an open-cycle gas-core. And even then the engineering challenge of designing an indestructible lightbulb is heart-breaking.

Open-cycle gas-core NTR don't even try to totally confine the uranium, they just try to minimize the loss. They use various clever tricks with vortexes and magnetic fields, but nothing matter-free is going to be 100% successful at confining that radioactive crap.

Large Engine Size

Nuclear fuel will produce about as much energy as a sleeping hippopotamus if you don't have a critical mass. Among other things, you need a large enough amount of uranium inside a small enough volume.

Getting enough uranium inside a small enough volume is relatively easy, if the stuff is solid or liquid. Unfortunately gas is the exact opposite of what we need, it want to get a little uranium inside a volume as possible (this is similar to the way that one small cat can expand itself so it takes up the entire freaking bed). To cram enough gaseous uranium into a given volume requires pressure, lots of it. This raises the engine mass, because pressure vessels whose wall are too thin tend to explode.

Even after you've squeezed the uranium gas as much as possible, you'll find it still lacks the density required to have a critical mass unless the reaction chamber is huge. This increases both the engine's volume size and mass size.

Bottom line is that gaseous uranium has a high critical mass requirement.

So, both of the problems are a direct outcome of the fact that the fuel is gaseous. The researchers asked the question: is to possible to make is not quite gaseous and still get good rocket performance? What if instead of being gaseous, it was instead a colloid?

In this concept, the fuel is not a gas so much as it is a dust-storm. This will allow the fuel to have a density several orders of magnitude (i.e., about a thousand times) as compared to an actual gas.

The nuclear fuel loss problem can be addressed by using the large density difference between the dust particles of uranium and the free atoms of hydrogen.

The large engine size problem is drastically reduced because using fuel particles instead of gaseous fuel allow a critical mass is a much reduced volume. It also dramatically lowers the required pressure, to about half of that required by an open-cycle gas core.

Result: an engine with an exhaust velocity of about 11,800 m/s, with a thrust of around 445,000 Newtons. Which was more than a solid-core, less than a gas-core, and much simpler to engineer.

This concept sounds similar to the pebble-bed reactor engine, but it isn't. The pebble-bed uses a fluidized-bed configuration, and the particles are not completely suspended. This means the pebble-bed fuel is in physical contact with the reactor walls, encouraging them to melt. Which is never a happy occasion.

The concept also sounds similar to the liquid-core reactor engine, but it ain't that either. The liquid core engine uses bubbles of propellant passing through a liquid film of nuclear fuel. This presents flow distribution problems that have been shown to be insurmountable (as of the report date of 1971).

The colloid-core engine uses fluid mechanics to separate the fuel from the propellant in the exhaust stream. The nuclear fuel is suspended in a confined fuel zone (2 in the diagram) to keep it in a critical mass. As mentioned before the pressure required for critical mass was about half that required for a standard open-cycle gas-core, which is good news for the engineers.

According to their mathematical models, the fluid mechanics did a fantastic job of preventing the uranium dust from escaping out the exhaust nozzle. Unfortunately the dust got hot enough to leak uranium vapor from the particle surface, and fluid mechanics didn't do zippidee-doo-dah to stop the vapor from escaping. And the amount of vapor escaping was very significant.

To deal with the uranium vapor problem, they replaced the pure uranium dust fuel with dust composed of a uranium—zirconium carbide compound. A mix of one part uranium carbide to ten parts zirconium carbide (UC2 + 10 ZrC). Zirconium carbide is an extremely hard refractory ceramic material which is highly corrosion resistant and a member of the ultra high temperature ceramics group. Zirconium carbide is already used as a coating for nuclear reactor fuel elements because it has a low neutron absorption cross-section and weak damage sensitivity under irradiation. But the main figure of merit is its low vapor pressure, i.e., it does not vaporize worth a darn.

Mind you, there is still some uranium fuel loss. But it is measured in kilograms per minute, instead of kilograms per second.

Figure 2 above shows the specific impulse given the engine static temperature, the engine pressure in ATMs, and the uranium+zierconium carbide fuel mix. Just ignore all the fuel mixes except for UC2 + 10 ZrC. This figure is mildly interesting.

Figure 3 above shows how much uranium is lost out the exhaust nozzle in kilograms per minute, given engine static temperature, engine pressure in ATMs, and fuel mix. Again ignore all the fuel mixes except for UC2 + 10 ZrC. This figure is mildly interesting.

This is the interesting figure. The curves show specific impulse and uranium fuel loss; given static temperature of engine and engine pressure. The chart assumes UC2 + 10 ZrC fuel, and 445,000 Newtons of thrust (100,000 lbs)

Point "A" (green lines) has specific impulse of 1,150 sec, uranium loss rate of 20 kg/min, static temperature of 3,620°K, and centerline static pressure of 76 atm.

Point "B" (blue lines) has specific impulse of 1,200 sec, uranium loss rate of 28 kg/min, static temperature of 3,950°K, and centerline static pressure of 200 atm.


Implications for the Colloid Core Reactor

Table 1 Flow Parameters
SymParameterProjected Reactor
Router radius of vortext0.30 m
haxial length of chamber0.30 m
ṁ/Atotal gas flow rate into chamber over A17.7 kg/m2 sec
ρggas density1.1 kg/m3
μgas viscosity0.04 centripoise
ρpmaterial density of particle8 gm/cm3
Dparticle diameter10μ
Mtotal powder mass in vortex30 kg
εvoid fraction0.95
rradius0.13 m
ωangular velocity450 rad/sec

     As an example of possible extrapolation from our present experiment to operating conditions of the colloid core reactor, a hypothetical engine will be assessed, with thrust of 20,000 lb force (90,000 Newtons), and specific impulse of 1000 lbf-sec/lbm (Isp 1000 sec, Ve 9,810 m/s). A mass flow rate of 10 kg/sec would then be required. Taking hydrogen as operating medium and assuming a temperature of 3300°K, the dynamic viscosity of the gas flow in the reactor would be about 0.04 centipoise, and gas density at a median cavity pressure of 100 atmospheres would be about 1 kg/m3. A uranium carbide alloy, (1U—lOZr)C (Uranium 235 + Zirconium Carbide), has been proposed as a possible fuel for the colloid core reactor because of its low uranium equilibrium vapor pressure. Material density of this fuel would be about 8 gm/cm3, near the density of zinc. For convenience in relating the behavior of the particulate fuel to our observations of zinc particles, the fuel particle diameter will be taken as 10μ. For this example, the reactor radius is taken as 30 cm, with an axial length of 30 cm, and an inlet velocity of 100 m/sec is given for tangential gas injection.

     Critical mass calculations for a cavity radius of 30 cm and axial length of 12 cm indicate a uranium mass requirement of about 3 kg. Critical mass requirement for the longer chamber used in the present example is probably lower than the 3 kg figure, but for a conservative estimate, 3 kg uranium mass, for a total fuel load of 30 kg, will be assumed. Applying the parameters to Eq. (2-4), inner bed radius, angular velocity, and void fraction can be calculated. The various extrapolated flow parameters for the reactor are compared with experimental conditions in Table 1. Note that the particle volume fraction (1 - ε) required by this fuel load is well below the 10% value attained in our present experiments.

     The power output required by the engine performance specification is about 600 mw, with a reactor power density of about 104 mw/m3. The average temperature differential between particles and gas is derived in the Appendix, indicating a temperature difference on the order of 20°K between fuel particles and propellant gas for this reactor. The total pressure drop needed to fluidize the particle bed under the prescribed conditions is about 33 atm.

     If the zinc particle loss rate given in Fig. 5 is linearly scaled to the larger gas flow rate per unit area under consideration for the reactor, a fuel loss of about 30 gm/sec is predicted, for a uranium loss rate of 3 gm/sec. Vaporization loss for uranium has been estimated at roughly 100 gm/sec at an operating temperature of 3300°K, so that particulate loss will be small in comparison to vaporization.


     Our experiments with two component vortex flows, when extrapolated to the operating range of a colloid core reactor, project a reasonable degree of optimism concerning the ultimate feasibility of such a system. The projected reactor size is modest, no problem appears in fluidizing and rotating the required fuel load, and reactor power density is well within the heat-transfer capacity of the fluidized bed. Rough estimates of particulate fuel losses indicate that uranium loss rate will be governed primarily by materials problems, rather than aeromechanical considerations. Many problems concerning the generation and transfer of power within a rotating fluidized bed are yet to be answered, but the basic concept of maintaining a fluidized vortex flow of high-particle density to provide nuclear propulsion appears to be sound.

Gas Core

Remember, all nuclear thermal rockets are using nuclear energy to heat hydrogen propellant for rocket exhaust. The hotter the reactor core, the more the propellant is heated, and the higher the specific impulse and exhaust velocity. That means the rocket has more delta-V go travel to more distant places, and also can carry more payload.

The problem is that the reactor is made out of matter, and above a certain temperature the reactor melts. Go higher and the reactor vaporizes into gas. Solid-core nuclear thermal rockets keep the temperature below the melting point, which means they top out at a specific impulse of 1,200 seconds or so. Admittedly this is better than the pathetic 450 seconds you can squeeze out of a conventional chemical rocket. But it is still not high enough to really open up the exploration of the solar system.

So crazy engine designers started to look into reactors that were molten under normal operation. Or even gaseous.

If you allow the uranium to reach a temperature where it melts you can get up to a specific impulse of 2,000 seconds or so. This is a liquid-core nuclear thermal rocket. You spin the reaction chamber around the thrust axis to make the hot bubbling liquid uranium stick to the chamber walls instead of escaping out the exhaust.

But if you want to crank it up to the max you have to let the uranium reach temperatures where it vaporizes into white-hot gas. This can get up to a whopping 3,500 seconds of specific impulse.

The drawback is trying to keep all that expensive and deadly gas from shooting out the exhaust bell. Which isn't easy.

Closed-Cycle gas-core NTR try to have it both ways. They enclose the nuclear fury of gaseous uranium in solid quartz-crystal containers to keep the exhaust non-radioactive. Which is counter-productive since the whole idea was to let everything vaporize for maximum heat output. The end result is the specific impulse will be about half of what it could be.

Open-cycle gas-core NTR just let it all hang out. Radioactive fission-products vapor escapes out the exhaust making it very unhealthy to be anywhere near the rocket when it is thrusting. But it has the maximum specific impulse. Since that enriched uranium is hideously expensive you want to at least make a cursory effort to keep it in the reaction chamber as long as possible. You do not want un-burnt uranium escaping, you want it all burnt in the reaction chamber. The general rule is that as long as the hydrogen mass expelled is 25 to 50 times a high as the uranium mass expelled the uranium losses are within acceptable limits. The various open-cycle designs use different strategies to make the hydrogen to uranium exhaust flow ratio as high as possible.

Closed Cycle

Gaseous Core NTR closed 1
Exhaust Velocity20,405 m/s
Specific Impulse2,080 s
Thrust445,000 N
Thrust Power4.5 GW
Mass Flow22 kg/s
Total Engine Mass56,800 kg
Uranium Hexafluoride
ReactorGas Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power13 kg/MW
Gaseous Core NTR closed 2
Thrust Power0.6 to 231 GW
Exhaust velocity10,800 to 31,400 m/s
Thrust117,700 to 14,700,000 n
Engine mass30 to 300 tonne
Engine T/W0.4 to 5.0
Operating Pressure400 to 1600 atm
NASA report nuclear lightbulb
Thrust Power3.7 GW
Engine Power4.6 GW
Exhaust velocity18,300 m/s
Thrust409,000 n
Engine mass32 tonne
Engine T/W1.3
Operating Pressure500 atm
Propellant mass flow22.3 kg/s
Liberty Ship
Propulsion SystemNuclear Lightbulb
Exhaust Velocity30,000 m/s
Specific Impulse3,058 s
Thrust/Engine5,340,000 N
Number Thrustersx7
Thrust37,380,000 N
Thrust Power560.7 GW
Mass Flow1,246 kg/s
Total Engine Mass378,000 kg
Uranium Hexafluoride
Wet Mass2,700,000 kg
Dry Mass1,600,000 kg
Mass Ratio1.69 m/s
ΔV15,697 m/s
Specific Power0.67 kg/MW

Closed-cycle gaseous core fission / nuclear thermal rocket AKA "Nuclear Lightbulb". Similar to an open-cycle gas core fission rocket, but the uranium plasma is confined in a fused quartz chamber. It is sort of like a child's classic Easy-Bake Oven. Except that there is propellant instead of cake mix and the light bulb is full of fissioning uranium instead of electricity.

You can read more about this on the Unwanted Blog in the posts here, here, and here.

The good news is that unlike the open-cycle GCNR it does not spray glowing radioactive death there is no uranium escaping in the exhaust. The bad news is that the maximum exhaust velocity is halved, as is the Delta-V. Yes, I did ask some experts if it was possible to make some kind of hybrid that could "shift gears" between closed and open cycle. Sorry, there would be no savings over just having two separate engines.

The maximum exhaust velocity is halved because you are trying to have it both ways at once. The higher the propellant's temperature, the higher the exhaust velocity and rocket Delta-V. Unfortunately solid core nuclear reactors have a distressing habit of vaporizing at high temperatures (as do all material objects). Gas core reactors are attempting to do an end run around this problem, by having the reactor start out as high temperature vapor. But adding the quartz chamber is re-introducing solid material components into the engine, which kind of defeats the purpose. The only thing keeping this from utter failure is the fact that quartz does not heat up as much due to the fact it is transparent.

Even with the restrictions, it seems possible to make a closed-cycle GCNR with a thrust to weight ratio higher than one. This would allow using the awesome might of the atom to boost truely massive amounts of payload into Terra orbit, without creating a radioactive wasteland with every launch. See the GCNR Liberty Ship for an example. The Liberty Ship can boost in one launch more payload than any given Space Shuttle does in the shuttles entire 10 year operating life. Then the Liberty Ship can land and do it again.

The ideal solution would be to somehow constrain the uranium by something non-material, such as a mangnetohydrodynamic force field or something like that. Alas, currently such fields can only withstand pressures on the order of the breeze from a flapping mosquito, not the 500 atmospheres of pressure found here. But since researchers are working along the same lines in their attempts to make a fusion reactor, this may change. And I did find a brief reference to something called an "MHD choke" in reference to slowing the escape of uranium into the exhaust stream of an open-cycle gas core rocket. I'm still trying to find more information on that.

The high pressure is to ensure the uranium vapor is dense enough to sustain a fission reaction.


The nuclear Cargo Orbital Transport Vehicle (COTV) concept analyzed combined the desirable features of the chemical COTV and the electrical COTV — high thrust and high specific impulse, respectively. The stage, shown on Figure A-21, has a nuclear gas core, light bulb-shaped engine with a theoretical specific impulse of 2250 seconds and a thrust level of 890,000 newtons. The component mass breakdown is given in Table A-3.

Although such a system could meet the short trip time requirement for personnel transfer and the high performance requirement for cargo transfer, the development risks and the presence of nuclear materials in LEO eliminated this system from further consideration.

VCR light bulb fission
VCR light bulb fission
Exhaust Velocity19,620 m/s
Specific Impulse2,000 s
Thrust56,400 N
Thrust Power0.6 GW
Mass Flow3 kg/s
Total Engine Mass72,566 kg
Uranium Hexafluoride
ReactorGas Core
RemassSeeded Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power131 kg/MW

Most fission reactors avoid meltdown, but the vapor core reactor (VCR) runs so hot (25000 K) that its core vaporizes.

At this temperature, the vast majority of the electromagnetic emissions are in the hard ultraviolet range. A “bulb” transparent to this radiation, made of internally-cooled a-silica, bottles the gaseous uranium hexafluoride, while letting the fission energy shine through.

The operating pressure is 1000 atm. The UF6 fuel is prevented from condensing on the cooled wall by a vortex flow field created by the tangential injection of a neon “buffer” gas near the inside of the transparent wall.

In a generator mode, the UV uses photovoltaics to generate electricity. In a propulsion mode, the UV heats seeded hydrogen propellant, which exits at a specific impulse of 2000 seconds.

From High Frontier by Philip Eklund
NASA Report

The pictured engine is the "reference engine" described in the report Studies of Specific Nuclear Light Bulb and Open-Cycle Vortex-Stabilized Gaseous Nuclear Rocket Engines. I will give the high-lights from the report, but for all the boring nitty-gritty details you'll have to read the report yourself. Please note that as always, "down" is the direction the thrust is going (to the right in the blueprint) and "up" is the opposite direction (to the left). I apologize for the use of imperial units instead of metric. The report is in imperial, and it is too much of a pain to change everything.

The important statistics: Specific impulse is 1870 seconds (18,300 m/s), thrust 409,000 newtons, engine mass 32,000 kg, thrust-to-weight ratio 1.3.

The other stats. The total volume of the reaction chambers (cavities) is 170 cubic feet. There are seven cavities, each six feet long. The cavity pressure is 500 atmospheres. The specific impulse is 1870 seconds (report says can theoretically be from 1500 to 3000 seconds). The total propellant flow (including seed and nozzle transpirant coolant flow) is 49.3 pounds per second. The thrust is 92,000 pounds. The engine power is 4600 megawatts. The engine weight is 70,000 pounds. If one can design a variable-throat-area nozzle (instead of fixed-area) this will result in a major decrease in the required chamber pressure during startup.


The basic configuration is seven separate unit cavities surrounded by moderator-reflector material in between each cavity (beryllium oxide) and surrounding the entire cavity array (graphite). Each cavity is 6.0 feet long and the total volume of all seven cavities is 169.8 cubic feet. The cavity pressure is 500 atmospheres due to criticality and fuel density considerations.


In each lightbulb, a critical mass of gaseous uranium creates thermal radiation. The thermal radiation can pass through the transparent quartz crystal walls of the lightbulb, but the uranium vapor cannot. This means no lethal uranium enters the exhaust. Hydrogen propellant flowing over the lightbulb is heated to high temperatures by the thermal radiation and is expelled out the rocket nozzles, producing thrust. The hydrogen is "seeded" with tungsten dust because it too is ordinarily transparent to thermal radiation. The seeding makes it opaque, and allows it to be heated. Seven "lightbulbs" are used instead of one, since that increases the total lightbulb radiating area by about 2.2 times.

Transparent quartz walls

The transparent quartz wall of the lightbulb contains lots of coolant channels. This is because the quartz is mostly transparent to thermal radiation, but not totally. And fissioning uranium produces an awful lot of thermal radiation. I told you that nuclear lightbulb designers were trying to have it both ways. The coolant channels are marked "circumferential coolant tubes" in the diagram below.

Inside a lightbulb

Inside the lightbulb, neon buffer gas is used to create a vortex ring to suspend the gaseous nuclear fuel (a "radial inflow" vortex). The vortex ring looks like an elongated donut (I know it looks like two separate blobs above, that's due to the fact the diagram is a cross-section). One of the important jobs done by the neon buffer gas is to prevent the 42,000°R uranium plasma from making contact with the lightbulb walls. This would be very bad, as the walls would be instantly vaporized. The neon passes along the lightbulb walls, bends round the end caps, then travels down the long axis of the lightbulb (right down the center of the vortex ring). When it reaches the fore end cap, it is removed from the lightbulb through a port (marked "thru-flow" in diagram above).

The removed neon is very hot, and contains unburnt uranium and fission products. It is cooled by mixing with low-temperature neon, which condenses the unburnt uranium vapor into hot liquid uranium. The liquid uranium is separated from the neon by a centrifuge and sent back into the vortex (at point marked "fuel injection"). The neon is cooled further then it too is sent back into the vortex (at point marked "buffer gas injection"). While examining the blueprint, I noticed that the centrifuges, and indeed the entire uranium fuel delivery system is conspicuous by its absence. Probably classified.

Note that the centrifuges is a neat solution to the problem of fission fragments clogging up the fuel. In essence, this design has its own built-in nuclear fuel reprocessing plant. Of course the nasty fission fragments will have to be stored and eventually disposed of.

Lightbulb dimensions

The total volume inside all the lightbulbs is 84.9 cubic feet, which is 12.1 cubic feet per lightbulb. The radius of the uranium fuel containing region is 85% of the radius of the transparent wall. While the fissioning uranium fuel has a core temperature of 42,000° Rankine, the outer surface is only at 15,000° Rankine.

Propellant flow in a lightbulb

The propellant is assumed to exit with a temperature of 80% of the fuel temperature, or 12,000° Rankine. This is because the quartz transparent walls will reflect about 15% of the thermal radiation back inside. By some compilcated reasoning that you will find in the report, the total thermal radiation from the lightbulbs is 4.37 x 106 BTU/sec. The hydrogen propellant has an "enthalpy" of 1.033 x 105 BTU/pound at 12,000°R. So by dividing the two, one discovers that the entire engine can support a propellant flow rate of 42.3 pounds per second, which means 6.07 lb/sec for each of the seven cavities.

If that last paragraph confused you, let me explain. As a simple example, if a pound of hydrogen at 5°R contains 2 BTUs ("enthalpy"), and the engine puts out 6 BTU per second, then obviously the engine can heat up 6 / 2 = 3 pounds of hydrogen per second. Why do we care? If you multiply the propellant flow rate by the exhaust velocity you will discover the engine's thrust value. And that's a number we do care about.

The tungsten dust that the propellant is seeded with has a particle diameter of 0.05 microns. The seed density is 1.32 x 10-2 lb/ft3, which is about 3.9 percent of the inlet propellant density. This can probably be reduced if tungsten dust was in the form of thin flat plates instead of spherical particles.

The hydrogen propellant enters the pressure shells from the fore end (see "Primary Circuit Inlet" in pressure shell diagram below). A bit is bled off from small H2 flow ports in order to pressurize the interior of the shells, circulating to provide coolant to the engines and machinery. But most of it is fed into the turbopump, then injected into the cavities. Since the fore end of each cavity is almost blocked off by the butt end of the lightbulb, there is only a narrow rim to inject the hydrogen.

In the diagram to the right, you can see how the propellant is fed from the pink pipe into the pink-and-gold wedge-shaped injectors. I presume there are three injectors per cavity, spraying into the clear area between the transparent wall's coolant manifolds and buffer gas injectors.

Uranium fuel

The total fissioning uranium in all seven vortexes be about 25.2 pounds of uranium (about 3.6 pounds per cavity). You would ordinarily need more to ensure nuclear criticality, but the required amount is brought down by the beryllium oxide neutron reflector encasing each cavity. The average uranium fuel density is 0.409 lb/ft3. The total density of the neon-uranium mix inside the vortex is about 0.56 lb/ft3. A unit of neon gas will spend about 3.8 seconds inside the cavity. A unit of uranium will spend about 19 seconds inside the cavity. This implies a uranium fuel flow rate of 0.19 lb/sec per cavity.

According to my slide rule, if the array of seven cavities is producing 4,600 megawatts, it means that the array is burning a miniscule total of 0.055 grams (0.00012 pounds) of uranium fuel per second (0.0079 grams per cavity per second). It still needs the full 3.6 pounds per cavity to be present in order to burn the fraction of a gram.

The theoretical maximum specific impulse possible is 2230 seconds. Due to this designs incomplete expansion, transpiration coolant flow in the nozzle, presence of tungsten seeding, and friction losses the specific impulse is reduced to 84% or 1870 seconds. Total propellant flow (allowing for tungsten seeds and transpiration cooling) is 49.3 lb/sec. This would result in a thrust of 92,000 pounds force. For complicated reasons you can find in the report, this implies that the exhaust nozzles are 0.0875 feet in diameter at the throat expanding to 2.04 feet diameter at the exit.

Uranium refueling

Careful readers may have noticed how the description avoids mentioning the details on how one gets the uranium into the lightbulbs. This is because it is quite a difficult problem, and each of the proposed solutions has drawbacks. The basic problem is old reliable: all the atomic fireworks inherent in 235U will happen if you merely let too much of it accumulate in one place. You have to store it diffuse and somehow bring it together in the lightbulb.

Method #1 Store it as uranium hexafluoride gas. This would be in large tanks of low pressure (i.e., low density) and with the tanks full of neutron absorbing foam. Pump enough into the lightbulb, a chain reaction will start, and well before the reaction reaches 13,000°R the uranium will have separated from the fluorine.

The problem is that now you have the insanely dangerous task of dealing with 13,000°R fluorine gas. At room temperature the blasted stuff will violently react with any element in the known universe except helium and neon. A temperature of 13,000°R makes it about 13,000 times as deadly. It will explosively corrode away anything solid in its path like molten lead on facial tissue. Chemist Derek Lowe sarcastically notes that "At seven hundred freaking degrees, fluorine starts to dissociate into monatomic radicals, thereby losing its gentle and forgiving nature." You can read more about the suicidal risk of dealing with hot fluorine in his amusing blog post.

Method #2 Store it as sub-critical chunks of uranium, melt them, and inject the molten uranium into the lightbulb. Uranium melts at 1403°K, which is difficult but not impossible. The plan is to somehow turn the molten uranium into a sort of aerosol mist suspended in hot neon.

The problem is that the molten uranium wants to plate itself all over the melter and the aerosol spray equipment. Which is annoying if the material in question is something like lead, but disasterous if the material is radioactive and fissionable.

Method #3 is to store the uranium cold as finely divided dust. As dust it is pumpable, injectable, and it will not plate over everything. Inside the lightbulb the uranium dust will be rapidly heated to vaporization by the nuclear reaction. This method does not have any major problems, except for the common problem of how to protect the transparent wall from being vaporized by the heat.

Again, the uranium delivery system seems to be totally missing from the blueprint. The only bit present is the short stub of the injector at the top of each lightbulb.

Pressure shells

The entire engine is encased in two nested pressure shells constructed of filament-wound fiberglass. The inside of the inner shell is pressurized to 500 atmospheres. Hydrogen propellant enters through a 0.5 foot diameter duct at the fore end (aka "Primary Circuit Inlet"). There are seven 0.4 foot diameter holes in the aft end for the engine nozzles, one at zero degrees off-axis, the other six at 60°. The pressure shell can be separated into two parts along the flange at the point of maximum diameter, to allow an engineer or waldo manipulator access to the engine interior. This point is also where the rear structural grid protrudes from the interior, this is where the engine bolts onto the structural frame of the spacecraft to transmit the engine thrust.

If you look at the large blueprint, you will see that parts of the rear structural grid penetrate the cavities to support the end-caps of the quartz lightbulbs.

Coolant system

The plumbing for the coolant system is rather complicated (translation: I don't understand it all). Click for larger image. You can use this diagram along with the large blueprint to attempt to puzzle out what all the pipes are for. Basically the propellant enters the system through the "Primary circuit inlet" (at lower left of plumbing diagram, and in the pressure shell diagram above) and leaves the system via the "Propellant injection" arrow, where the propellant is heated by the lightbulbs in the cavity and jets out the exhaust nozzles. In between, the propellant frantically threads its way over every single other engine component in a desperate attempt to cool them off.

Cross sections

Here are a set of cross sections through the cavities. The one on the left is zoomed in on the cavity interior, the other two gradually zoom out.

UAC Report

The information comes from a series United Aircraft Corporation reports written mostly by Thomas L. Latham. There are more reports than the ones I've used.

The reference design had seven cells with six surrounding the center cell. The entire engine was sized to fit into the Space Shuttle cargo bay. It was also sized at 4.6 gigawatts, 409,000 Newtons, and a specific impulse of 1,860 seconds in order to avoid the need for external heat radiators. At this level no radiators are required for the moderator or pressure vessel, open-cycle cooling will suffice. Above a specific impulsle of 1,860 seconds radiators will be needed or the engine will melt.

If the specific impulse is above 2,500 seconds the nozzle throats will require their own cooling system.

The hydrogen propellant is seeded with tiny tungsten particles due to the unfortunate fact that hydrogen is transparent to the frequencies emitted by the nuclear reaction. Otherwise the chamber walls would be heated instead of the propellant, which is the exact opposite of what we want. The fissioning U235 or U233 fuel also emits ultraviolet light that degrades the transparency of the enclosing quartz "lightbulb." The researchers were experimenting with seeding the uranium with something that would turn the UV into infrared in order to protect the quartz. Happily the ionizing radiation does expose the degraded quartz to a radiation damage annealing effect that restores transparency to some extent.

The fuel is in the form of Uranium Hexafloride.

The average dose rate in the filament-wound fiberglas pressure vessel was calculated to be 0.17 mrad/sec. This would allow about six full-power runs of 1000-sec duration (about 17 minutes) before the total dose became 1000 mrad, the estimated allowable dosage before degradation of the laminate strength commences.


     This is basically the propellant, passing from the propellant tanks to be heated by the nuclear light bulbs, and then rushing through the exhaust nozzles to provide thrust. Along the way it provides some cooling for various items.
     Starting at the tank, the primary hydrogen pump sends it through a H2-H2 heat exchanger for preheating (and providing additional heat rejection for the Secondary Hydrogen Circuit). It passes through a H2-Ne heat exchanger to cool off the neon gas in the Neon And Fuel Circuit. It passes through the Fuel And Neon Separator. A turbine then sends it through the Solid Moderator and End-Wall Liners. Somewhere along the line it is seeded with tungsten microparticles so the hydrogen will be heated by the nuclear light bulbs.
     Finally it experiences extreme Direct Heating from the nuclear light bulbs, and exits through the exhaust nozzles.
     Basically the coolant system. It runs cooling hydrogen over the pressure vessel, nozzles, flow divider, tie rods, liner tubes, and the transparent walls of the quartz light bulbs (during shutdown it also cools the Fuel And Neon Separator).
     The now-hot hydrogen passes through a H2-H2 heat exchanger to give the heat to the space radiator. The lukewarm hydrogen passes through a second H2-H2 heat exchanger to cool down further and preheat the propellant hydrogen.
     The Neon Make-Up supply keeps the neon pressure in the circuit at the required level. The uranium-235 Fuel Make-Up keeps the amount of fuel droplets in the circuit at the required level. Both are fed into the Fuel Cavity in the interior of the quartz light bulbs to create the furious nuclear reaction (unless engine shut-down is in progress, then the Fuel Control Valve closes to shut off the uranium). The reaction provides the direct heating to the Primary Hydrogen Propellant Circuit. Some of the neon goes through the Cavity Bypass Flow.
     Only a fraction of the uranium undergoes fission. So the neon/uranium that comes out of the Fuel Cavity is sent through the Neon and Fuel Separator to strain the uranium out of the neon gas. The neon is cooled which makes the uranium gas condense into liquid droplets. The two are separated by a centrifuge. The neon is cooled further by the H2-Ne heat exchanger.
     The neon goes to the Neon Pump, the uranium goes to the Fuel Pump and the cycle begins anew.

For additional details see Ref. 5 (Nuclear studies of the nuclear light bulb rocket engine).

Neon supply is the Neon Make-Up supply, keeping the neon pressure in the circuit at the required level. It is fed into the Fuel Cavity (Unit Cavity) tangentally just inside the quartz light bulb Transparent Wall. This creates the neon-uranium vortex.

The Fuel distillation canister is the Fuel Make-Up. It is fed by the Fuel Pump into the fuel injection duct, introducing it into the Fuel Cavity (Unit Cavity). This creates the furious nuclear reaction inside the quartz light bulb, providing the direct heating to the Primary Hydrogen Propellant Circuit.

The mixture of hot neon, unburnt gaseous uranium fuel, and fission products exits the Fuel Cavity via the Exhaust Duct (about two meters long). Not shown is how cool neon is introduced into the entire length of the exhaust duct to [1] cool the exhaust from 6550 K to 1500 K, [2] prevent the exhaust from severely damaging the exhaust duct, [3] condense the gaseous uranium into liquid uranium droplets, and [4] ensuring that the uranium droplets condense inside the neon gas, instead of on the walls of the exhaust duct causing a nuclear reaction.

The 1500 K neon-uranium droplet flow is sent to the Neon and Fuel Separator (Separator) where the two are isolated by a centrifuge. The neon is cooled by the H2-Ne heat exchanger and goes to the Neon Pump. The uranium fuel goes to the Fuel Pump. Alternatively the uranium is distilled to separate out the silicon seeding and the uranium is deposited in the fuel distillation canister.

Values for weight flow rates, temperature, and volume flow rates are indicated at various stations in the system.

In the Neon and Fuel Separator, the seven exhaust duct inlet pipes from the seven nuclear light bulb unit cavities enter from the left. They enter two inlet plenums: four inlet pipes on the top plenum and three on the bottom. Each plenum has an injection slot delivering the gas mix into the separator cavity, with a velocity of 500 m/s at a steep angle designed to spin the gas. The spin centrifugally separates the uranium from the neon, at about 100,000 g's. The uranium is harvested by uranium collector tubes on the separator wall, while the neon is harvested by an outlet pipe on the separator's long axis. The separator cavity and uranium collector tubes have to be maintained at or above 1,500 K, or the uranium will condense on them. This will not only clog the thing up, but if enough uranium plates out it will accumulate a critical mass with regrettable results.


Startup Sequence

  1. Fill hydrogen ducts and neon system from storage to a pressure equal to approximately 20 atmospheres
  2. turn on neon recirculation pump
  3. inject fuel until critical mass is reached
  4. increase power level and adjust flow rates and cavity pressure to maintain criticality and limit component temperatures to tolerable level
  5. inject propellant seeds when 10 percent of full power is reached
  6. increase power to desired operating leve

The paper looks at two "power ramps", going from cold to full power in 60 seconds or a more leisurely 600 seconds. Below a temperature of 15,000°R the fusing uranium is heating up the hydrogen propellant mainly by convection. Above 15,000°R the uranium heats the propellant by infrared thermal radiation.

Since convection does such a pathetic job of transfering heat, most of the fission energy goes to heating up the uranium dust instead of the propellant. In about five seconds flat the uranium reaches 12,000°R, and vaporizes from dust into red-hot gas. Then at 15,000°R thermal radiation takes over and the uranium temperature rises more slowly (which you can see by the way the curve starts flattening out). At 60 or 600 seconds (depending upon which power ramp you used) the uranium is at the nominal temperature of 45,000°R. It won't rise any higher unless the engine is exploding or something rude like that.

As previously mentioned the hydrogen propellant is pretty much transparent to thermal radiation, which is most unhelpful. Normally the infrared will shoot right through the hydrogen without heating it up. So tungsten dust is seeded into the propellant to soak up the thermal radiation and heat the propellant by conduction. Any thermal radiation that misses the seeding will hit the far wall of the propellant chamber, which is also the beryllium oxide moderator (BeO) helping to keep the uranium fissioning. The thermal heating of the BeO is nothing but wasted energy but the seeding is doing the best it can. The BeO is designed so it can handle up to 2,400°R.

Since the BeO moderator outweighs the uranium dust by several orders of magnitudue, it takes far longer to heat up. As you can see from the graph the uranium fuel starts heating up after only 0.03 seconds but the BeO doesn't even start heating until 10 seconds, about 300 times longer. The uranium gets up to nominal temperature in 60 seconds but the BeO takes 300 seconds. And the BeO only gets up to 2,400°R while the uranium is smokin' at 45,000°R. That is for the 60 second ramp. The 600 second ramp has both the uranium and BeO all warmed up at the same time, only because 600 seconds gives the BeO time to catch up.

However, the shorter 60 second ramp is desireable, because the 600 second ramp wastes precious propellant. Take the propellant mass required for a standard 20 minute burn at full power. The 60 sec ramp requires an additional 2.7% propellant as startup wastage. The 600 sec ramp requires a whopping 27% additional, which is totally unacceptable. What, do I look like I am made of propellant? The paper says it might be possible to reduce the ramp time down to 6 seconds, in the interest of reducing the propellant startup wastage even further (presumably to 0.27%).

The critical mass of uranium-235 fuel required in the quartz tubes increases during the ramp up. It requires 18.6 pounds at zero power up to 30.9 lbs at full power. For the 60 second ramp up full power initially happens at 28.2 lbs, but rises to 30.9 lbs at 300 seconds. This is because at 60 seconds the BeO moderator has only warmed up about two-thirds of the way to its max temperature. Apparently once the BeO is fully warmed up the critical mass rises.

When the paper was read, one of the attendees was skeptical about pressure. Specifically if the pressure of the uranium/neon mix is not the exact same as the pressure of the hydrogen propellant, the pressure differential will shatter the quartz tube like dropping an old-school incandescent lightbulb on a concrete floor. The paper authors insisted that the two pressures could be balanced rapidly enough to prevent that unhappy state of affairs. They say that a differential of two or three atmospheres will shatter the blasted tube, so they want to keep the diff under 2/3rds atm. Yikes, I didn't know that! That would instantly ruin the propulsion system, and spray everybody and everything close by with fissioning uranium.



Shutdown Sequence

  1. Close Fuel Injection Control Valve (turn off the uranium)
  2. Begin Linear Decrease in Propellant Flow Rate (propellant flow past light bulbs to exhaust nozzles)
  3. Begin Linear Increase in Radiator Flow Rate (flow from coolant heat exchanger to radiator)
  4. Maintain Secondary Circuit (flow of hydrogen coolant) and Cavity Neon Flow (buffer gas flow inside the quartz light bulbs) at Full Power Value.

Once the engine shut-down sequence is initiated, it takes six seconds for the power level to drop to zero. It only takes 0.8 seconds for power level to drop to 0.01 of full power, during which time the contained uranium fuel drops from the steady-state level of 13.65 kg down to 11.5 kg.

Squeezing The Last Drop Report



The nuclear light bulb engine is usually considered to be comprised of seven separate unit cavities (Fig. la). Each cavity contains a region of gaseous uranium which heats seeded hydrogen propellant by thermal radiation. An internally cooled transparent wall is located between the fuel and the propellant regions (Fig. lb). The gaseous nuclear fuel is isolated from the transparent wall by a neon (or helium) vortex. This neon flow passes out through ports located on the centerline of the end wall of each cavity to a fuel recycle system in which fuel entrained in the neon is condensed to liquid form, centrifugally separated from the neon, and pumped back into the fuel region.

In the reference engine described in Ref. l (Fig. 2), each of the seven cavities has a length of 6 ft. The total volume of all seven cavities is equal to that for a single cylindrical cavity having a diameter of 6 ft and a length of 6 ft. The total amount of fuel contained within the seven cavities is about 14 kg, and the power is about 4600 Mw. The critical mass of 14 kg is less than that for a single-cavity reactor having the same total cavity volume because of the beneficial effect of the moderating walls between the unit cavities. The total pressure in the cavity is estimated to be 500 atm. The total propellant flow rate of 49 lb/sec is heated to 12,000 R, which provides a specific impulse of 1870 sec. The resulting engine thrust is, therefore, 92,000 lb. The total weight is estimated to be 70,000 lb and is made up of the following component weights: moderator (graphite and beryllium oxide), 27,000 lb; pressure vessel, 30,000 lb; turbopumps, 3000 lb; and miscellaneous (including the fuel recycle system), 10,000 lb.

The nuclear light bulb engine offers the possibility of perfect containment of the gaseous nuclear fuel because of the presence of the internally cooled transparent wall between the fuel and the propellant and because of the incorporation of a fuel recycle system in the engine.


Although the prime purpose of this paper is not to review in detail the status of the research to determine the feasibility of the nuclear light bulb engine, some comments on the status of this research are pertinent. Key results of work in five technical areas are outlined in Fig. 3, and comments on each of these technical areas are given in the following paragraphs.

Fig. 3
  • R-F radiant energy source developed with radiant heat flux of 49 kw/in2 (T* = 10,860 R)
  • Arc radiant energy source used to heat seeded gas to temperature of 3000 to 4000 R; goal of 6000 R set for this summer
  • Techniques developed for making structures from thin-walled (0.005-in.) fused silica tubing
  • Dynamitron tests indicate that radiant energy absorption in fused silica walls due to nuclear-induced coloration will be unimportant
  • Engine power appears to be controllable by controlling flow rate of nuclear fuel

A major effort at UARL has been devoted to developing radiant energy source configurations which can be used in experiments to determine the characteristics of the fuel region and the transparent-wall regions of a nuclear light bulb engine (Ref. 2). This work has concentrated on the development of techniques for depositing large quantities of radio-frequency energy in small volumes of argon gas. In one test a total of 160 kw of energy was added to an ellipsoidal volume having a length of 2 in. and a minor diameter of approximately 0.5 in. at an operating pressure of 19 atm. Of the total energy deposited, 131 kw was radiated through the surrounding transparent wall. The energr radiated in this test corresponds to a flux of 49 kw/in.2, which is equal to the flux from a blackbody radiating at 10,860 R. For comparison, the flux at the surface of the sun corresponds to a blackbody radiating temperature of 10,400 R. In other individual tests to date: total powers up to 254 kw have been added to the discharge; 82% of the deposited energy has been radiated through the transparent wall; and maximum pressures of 22.5 atm have been employed.

In another experiment (Ref. 3), a dc arc radiant energy source has been used to heat argon gas seeded with carbon particles (which simulates the seeded hydrogen propellant in a full-scale engine) to temperatures between 3000 and 4000 R. It is anticipated that simulated propellant-exit temperatures of greater than 6000 R will be attained by this sunmer.

The calculations of Ref. 1 indicate that the internally cooled transparent walls in the full-scale reference engine should be made from silica tubing having a wall thickness on the order of 0.005 in. Therefore, a portion of the SNSO-sponsored program (Ref. 4) has been devoted to developing techniques for making models from such thin-walled tubing. It appears practical to make such models using a technique in which the tubes are "potted" into their supporting structure so as to provide a partially flexible joint to minimize the possibility of structural failure due to unequal thermal expansion. Recent information from radiation damage tests — see following paragraph — indicates that it may be possible to increase the allowable thickness of the transparent wall in the reference engine to between 0.010 and 0.015 in., which simplifies wall fabrication.

Tests of fused silica specimens in the Dynamitron electron accelerator at the NASA Langley Research Center (Ref. 5) have indicated the existence of a radiation bleaching phenomenon which reduces the coloration due to nuclear irradiation. This coloration would cause absorption of thermal radiation in the transparent wall, particularly in an absorption band centered at 0.21 microns. As a result of this radiation bleaching phenomenon, it appears that nuclear-induced coloration of fused silica will be unimportant at the fluxes which will be present in a full-scale engine.

Studies have also been conducted (Ref. 6) to determine the characteristics of various components of the reference nuclear light bulb engine (see Fig. 2). Particular emphasis has been placed on the control system used during start-up, steady operation, and shutdown. It appears that engine power can be controlled relatively easily by adjusting the flow rate of the nuclear fuel injected into the cavities.

Recent results of other investigations conducted under Contract SNPC-70 are described in Refs. 7 through 9.


Three factors which might influence the maximum specific impulse of a nuclear light bulb engine are illustrated in Fig. 4. These factors are: the necessity of employing a space radiator, which will result in an increase in engine weight at high values of specific impulse; the necessity of minimizing absorption of thermal radiation in the transparent buffer gas region; and the necessity of providing transpiration cooling for the exhaust nozzle.

In a nuclear light bulb engine, approximately 15% of the energy created is deposited in the structure of the engine, either due to the absorption of neutrons and gamma rays, convective heating of the structure, or fission product decay energy. For low values of specific impulse, all of this energy can be removed by the hydrogen propellant before it is injected into the annular propellant regions, where it is further heated by thermal radiation. For instance, for the reference engine shown in Fig. 2, the temperature of the hydrogen at injection into the cavity (approximately equal to the maximum moderator temperature) is approximately 4000 R, and the specific impulse is 1870 sec. If the specific impulse is limited to approximately l400 sec, the maximum moderator temperature will be only 2500 R; such a reduction in maximum material temperature would make the engine relatively easy to construct. Conversely, the maximum material temperature must increase as the specific impulse increases, since the hydrogen temperature rise corresponding to absorption of 15% of the energy increases as specific impulse increases. At some point it is easier to remove this energy by the use of a space radiator than to permit an increase in the hydrogen inlet temperature and the corresponding material temperature. (A space radiator is shown removing a portion of the energ in the sketch in Fig. 4.) It appears at present that a space radiator will be desirable for values of specific impulse greater than 2100 to 2400 sec. Also, if a space radiator is employed, the minimum weight of the engine will be increased. Since high engine weights are very undesirable for many space missions, such a factor may be extremely important in evaluating the usefulness of a nuclear light bulb engine. It appears from present studies that the weight of a nuclear light bulb engine, because of the necessity of a space radiator at high values of specific impulse, will exceed 105 lb at a specific impulse of between approximately 2100 and 2500 sec. The choice of 105 lb as a maximum desirable weight is arbitrary, and should be mission dependent, but appears logical based on mission studies conducted at UARL.

For energy to be transferred to the hydrogen propellant, it is necessary that the radiating temperature at the edge of the nuclear fuel be greater than the propellant-exit temperature corresponding to the specific impulse desired. It has usually been assumed that this radiating temperature is 1.25 times the hydrogen propellant-exit temperature. Since the neon (or helium) buffer gas immediately surrounding the gaseous nuclear fuel must be at approximately the same temperature as the edge of the fuel, it is necessary that this buffer gas remain transparent at edge-of-fuel temperatures (Fig. 4). The absorption coefficients of helium and neon increase very rapidly above the temperatures at which the buffer gases first start to ionize. Neon and helium were chosen as candidate buffer gases because they have the highest ionization potentials of all known substances (21.6 and 24.6 electron volts, respectively). Any energy which is deposited in the buffer gas must be removed by convection of the buffer gas out of the cavity region; otherwise, the temperature of the buffer gas would increase in order to reradiate the absorbed energy. Detailed studies are now being conducted at UARL to determine the magnitude of the energy deposited in the buffer gas region as a function of edge-of-fuel temperature, and the limiting value of this energy deposition which can be removed by convection of the buffer gas. Although this analysis is not yet complete, it appears that the maximum specific impulse of a nuclear light bulb engine because of this limitation will be between approximately 2300 and 3000 sec.

The most difficult region of the nuclear light bulb engine to cool is the exhaust nozzle throat. Transpiration cooling is required to protect the wall from convective heating, and the small holes required for transpiration cooling eliminate the possibility of having high-reflectivity surfaces, and hence reducing radiant heating of the nozzle walls. An extensive analysis to determine the amount of transpiration coolant required to protect typical exhaust nozzles against over-heating due to convective and radiant heat loads is presented in Ref. 10. Some of the results from this analysis are given in Fig. 5. The spread in the information in Fig. 5 results from the wide range of assumptions which have been employed in the analysis as to the details of the convective and radiant heating processes and the effective enthalpy rise of the transpiration coolant fluid. It can be seen from this figure that the transpiration coolant flow necessary to protect the exhaust nozzle from convective heating is relatively small and is relatively independent of specific impulse. However, the transpiration coolant flow necessary to protect the exhaust nozzle from radiant heating increases very rapidly with increases in specific impulse; this transpiration coolant flow varies as approximately the ninth power of specific impulse. The limiting specific impulse for any set of nozzle assumptions occurs approximately at the point where the transpiration coolant flow required to protect the nozzle from radiant heat transfer exceeds that required to protect the nozzles from convective heat transfer. This occurs at a specific impulse between approximately 2500 and 5000 sec.


Studies have been conducted at UARL to determine the initial mass in earth orbit (IMEO) and the resulting direct operating cost associated with using various propulsion systems for various missions (Ref. ll). The results of IMEO calculations for a mission velocity requirement of 75,000 ft/sec are given in Fig. 6. (The direct operating costs are approximately proportional to IMEO for this mission.) Such a velocity increment corresponds to a typical manned Mars mission with return of all major components to earth orbit. As noted in Fig. 6, the values of IMEO associated with using the reference nuclear light bulb engine are approximately 20% of those associated with using a solid core nuclear rocket engine and approximately 5% of those associated with using a chemical rocket engine for the same mission. Also, the IMEO associated with the use of a nuclear light bulb engine having a specific impulse of approximately 1100 sec would be approximately the same as that for a solid core nuclear rocket engine. (The increased specific impulse of a nuclear light bulb engine relative to a solid core engine to obtain "break-even" performance is due to the lower thrust-to-weight ratio of the nuclear light bulb — see Fig. 7.) Thus nuclear light bulb engines with values of specific impulse only slightly greater than 1100 sec (possibly 1400 sec) might provide sufficient advantage to warrant their development, particularly because of the resulting simplification of material problems. Also, the advantage to be gained by increasing the specific impulse of a nuclear light bulb engine to values substantialJy higher than that for the reference engine (1870 sec) are less than at lower values of specific impulse. Thus for missions of the type considered, the limitations discussed in conjunction with Figs. 4 and 5 may not be important. However, for higher-velocity-increment missions, the payoff resulting from the use of higher values of specific impulse will become more important.


The information discussed in preceding sections is summarized in Fig. 7 where it is used to indicate "technology boundaries" for the nuclear light bulb engine. The "degree of difficulty" in building a nuclear light bulb engine is specified in Fig. 7 on the basis of the fuel radiating temperature, T*. This fuel radiating temperature is related to specific impulse by assuming that the propellant-exit temperature is 80% of T*. For an engine of the size of that shown in Fig. 2, the total power is also determined by the fuel radiating temperature. This power then determines the mass flow of hydrogen propellant which can be heated, and hence the thrust which can be obtained. The values of engine thrust-to-weight shown in the top part of Fig. 7 were calculated in this manner. The spread in the values of thrust-to-weight ratio for high values of fuel radiating temperature reflect an uncertainty in the weight associated with the space radiator. The lower values of thrust-to-weight ratio correspond to construction techniques analogous to those employed for electric propulsion devices. However, the space radiators on a nuclear light bulb engine are required to operate for only short periods of time because of the relatively high thrust-to-weight ratios available. For these short operating times, it is permissible to employ thin structures since small holes created by meteoroids will result in a leakage rate of coolant fluid which is small relative to the engine flow rate. Thus the higher values of thrust-to-weight ratio shown in Fig. 7 are believed to be pertinent for a nuclear light bulb engine.

The moderator temperatures shown in Fig. 7 are indicative of the degree of difficulty which will be encountered in maintaining the structural integrity of the moderator region. If the moderator temperature is not required to exceed 2500 R, then fabrication of the moderator from lower-maximum-temperature materials is possible. As the moderator temperature increases, the resulting material temperatures increase. As indicated in Fig. 7, it will probably be desirable to employ a space radiator when the moderator temperature exceeds a value between 4500 and 6000 R. This will result in increasing the engine weight for higher fuel radiating temperatures, leading to engine weights greater than 105 lb for fuel radiating temperatures greater than 18,000 to 24,000 R, which corresponds to values of specific impulse between approximately 2100 and 2500 sec. Also indicated in Fig. 7 are the values of fuel radiating temperature (and corresponding specific impulses) above which the energy deposition in the buffer gas is presently believed to be too large and above which the nozzle transpiration coolant flow rates will be unacceptable.

The break-even specific impulse with a solid core nuclear rocket (approximately 1100 sec) was obtained from Fig. 6 for the case of a velocity increment approximately corresponding to that for a manned Mars mission. The reference engine appears to be a suitable compromise between an engine which will "break even" with the solid core nuclear rocket engine and an engine which will be subject to the difficulties inherent in approaching the maximum specific impulse of a nuclear light bulb engine. However, reducing the specific impulse to a value of l400 to 1500 sec would result in a nuclear light bulb engine which is still significantly better than a solid core engine (see Fig. 6).

The solid core system is used for comparison only. It is recognized that solid core technology is currently available, and gas core technoloy is further in the future. Therefore, some margin in performance above the break-even point is required to justify development of the gas core rocket.


l. G. H. McLafferty and H. E. Bauer, "Studies of Specific Nuclear Light Bulb and Open- Cycle Vortex-Stabilized Gaseous Nuclear Rocket Engines," United Aircraft Research Laboratories Rept. F-910093-37 (Sept. 1967)

2. W. C. Roman, "Experimental Investigation of a High-Intensity R-F Radiant Energy Source to Simulate the Themal Environment in a Nuclear Light Bulb Engine," United Aircraft Research Laboratories Rept. J-910900-4 (Sept. 1970)

3. J. F. Klein and W. C. Roman, "Results of Experiments to Simulate Radiant Heating of Propellant in a Nuclear Light Bulb Engine Using a D-C Arc Radiant Energy Source," United Aircraft Research Laboratories Rept. J-910900-1 (Sept. 1970)

4. P. G. Vogt, "Development and Tests of Small Fused Silica Models of Transparent Walls for the Nuclear Light Bulb Engine," United Aircraft Research Laboratories Rept. J-910900-3 (Sept. 1970)

5. G. E. Palma and R. M. Gagosz, "Optical Absorption in Transparent Materials During 1.5 Mev Electron Irradiation," United Aircraft Research Laboratories Rept. J-990929-1 (Sept. 1970)

6. H. E. Bauer, R. J. Rodgers and T. S. Latham, "Analytical Studies of Start-Up and Dynamic Response Characteristics of the Nuclear Light Bulb Engine," United Aircraft Research Laboratories Rept. J-910900-5 (Sept. 1970)

7. J. F. Jaminet and A. E. Mensing, "Experimental Investigation of Simulated-Fuel Containment in R-F Heated and Unheated Two-Component Vortexes," United Aircraft Research Laboratories Rept. J-910900-2 (Sept. 1970)

8. T. S. Latham and H. E. Bauer, "Analytical Studies of In-Reactor Tests of a Nuclear Light Bulb Unit Cell," United Aircraft Research Laboratories Rept. J-910900-6 (Sept. 1970)

9. N. L. Krascella,."Analyt1cal Study of the Spectral Radiant Flux Emitted from the Fuel Region of a Nuclear Light Bulb Engine," United Aircraft Research Laboratories Rept. J-9l0904-l (Sept. 1970)

10. G. H. McLafferty, "Limitations on Gaseous Nuclear Rocket Isp Due to Nozzle Coolant Requirements," J. Spacecraft and Rockets, Vol. 3, 1515 (1966)

11. R. R. Titus, "Evaluation of NLB for Space Missions," United Aircraft Research Laboratories Rept. J-170917-1 (Feb. 1971)


Directed To: George McLafferty

1. With regard to the first technique, it seems to me that the window properties are really critical for this technique; I do not see how the cooling of this window below the melting point of quartz is achieved.

Response: (Mr. McLafferty) We have been through a fairly extensive analysis of all of the factors which influence the heat load on the transparent walls and the cooling of the walls. The radiation is probably the factor that worries you most. We have looked at the spectrum coming out of the fuel and the transmission spectrum of the walls and have calculated that the radiant energy absorption is on the order of 1% of the total energy created; this, along with some oonvective heating of the walls, sets the thickness of the walls of the tubes at approximately 5 mils. The cooling of the walls by the hydrogen flowing through the tubes is not particularly difficult. The problem is more that of conducting that energy through the silica wall; however, if that becomes a problem, we will change to a BeO wall which has a wider transmission spectrum and a higher thermal conductivity.

2. The other question is in regard to the seeding technique. I understand why this is needed because hydrogen gas even up to 10,000 Kelvin absorbs relatively little radiation; this is well known. However, the calculations we have heard about in this program neglect the effect of evaporation of the particulates. lt seems to me this would be an important factor.

Response: (Mr. McLafferty) It tums out that tungsten particles after they have vaporized form a gas that is very rich in spectral lines and that the high pressures of the surrounding gas broaden these lines to give a fairly continuous absorption. It appears that something on the order of 1 to 5% by weight of tungsten particles is enough to give the required absorption. There is a decrease in absorption as the particles vaporize, but it still looks like the absorption of the gas is adequate.

Response: (Mr. Ragsdale) The engine thermal radiation calculations that l have discussed for the open-cycle, porous wall engine do include the effect of particulate evaporation. In the calculations, the hydrogen propellant and solid-seeds increase in temperature as the mixture flows radially inward from the cavity wall. The vapor pressure equations of the solid-seed material are in the computer calculation and when the mixture reaches the evaporation temperature, the particles do evaporate in the calculations. At that point, the mixture opacity is no longer determined by the solid-seed absorption cross sections, but is instead calculated from the opacity of the gaseous seed material. From that point on, the opacity of the seed-vapor/hydrogen mixture follows the proper opacity law of the mixture. Very near the fuel region, temperatures sufficiently high are reached, say 20,000°K; the dominant absorber is the hydrogen gas itself, though the opacity contribution of the seed-vapor is still included in the calculation.

Pulsed Close Cycle

This is from Pulsed Plasma-Core Rocket Reactors (from Research on Uranium Plasmas and their Technological Applications page 52) (1970)

This is actually quite clever. Dr. Winterberg was trying to address the two main problems with open-cycle gas core reactors: preventing unburnt U235 from escaping out the exhaust nozzle, and dealing with wear and tear on the engine from the horrifically high operating temperatures. His solution was to pulse the reaction.

Now remember nuclear fission 101: when a thermal neutron crashes into a uranium 235 nucleus, the nucleus is split into fission fragments, and nuclear energy is released. Oh, and it also emits several neutrons, which keep the chain reaction going.

You want to burn as much of the U235 as possible, that stuff's expensive. If you can't burn it all in the rocket chamber, the next best thing is to try and catch the unburnt U235 before it escapes out the exhaust and re-use it.

where ΔNu/Nu is the percent of U235 that was successfully burnt in a fission reaction, σf is the fission neutron cross section, φ is the neutron flux, and τ is the fuel confinement time (or lifetime in the reactor if engine is a solid core NTR).

In other words, improving the amount of U235 burnt means increasing the the amount of uranium atoms getting in the way of neutrons, increasing the number of neutrons for the uranium to get in the way of, and increasing how long the uranium atoms are playing demolition derby with the neutrons. Which is kind of obvious if you think abou it.

So if a bog-standard nuclear power reactor had a neutron cross section σf of 10-22 cm3, a neutron flux φ of 1015, and was turned on τ of 103 seconds (16.6 minutes); then the ΔNu/Nu u235 burnup would be 0.0001 or 0.01%.

Open-cycle gas core nuclear rockets are really bad at confining the fuel for any reasonable length of time. τ is really low. To make up for this you have to increase the neutron cross section or the neutron flux. Or both. Increasing the neutron cross section means drastically increasing the chamber pressure to make the U235 cloud more dense, which means more mass for a heavy-duty pressure chamber, which sends the engine's thrust-to-weight ratio gurgling down the toilet. Increasing the neutron flux means more neutron heating of the engine, or even enough neutron heating to actually vaporize the engine.

Dr. Winterberg noted that while increasing the neutron cross section is probably out of the question, there might be a way to manage an increase in neutron flux. The neutron heating of the engine relies upon duration. The longer the engine is exposed to the neutron flux, the hotter it gets. Reduce the exposure time and you reduce the engine temperature rise. In other words: Pulse the reaction. You can use a fantastically high neutron flux as long as the duration of the flux is short enough so that the engine does not overheat. Wait for the engine to cool off then you can pulse again.

Taken to extremes, you'll have the equivalent of an Orion drive, where the reaction is less a slow energy release and more like an Earth-Shattering Kaboom. A bomb in other words. But Dr. Winterberg saw there was a lot of performance improvement possible using pulses much less violent than bomb-level. More to the point, improvements that would allow the engine to get away with having very short confinment times.

The engine will use a high neutron flux to pulse a series of "soft" nuclear detonations. This will have the following advantages:

  • The high neutron flux will increase the U235 burnup rate to the point where you can get away with a shorter required fuel confinement time. The short pulse will ensure that the neutron flux does not vaporize the engine.
  • The higher temperatures created in the reaction chamber will increase the exhaust velocity and specific impulse something wonderful. Again the short pulse will prevent the temperatures from damaging the engine
  • Pulse operation allows starting the chain reaction from a uranium-propellant mixture at high density with a small critical mass. This allows the reaction chamber and the rest of the engine to be smaller than other gas-core designs.
  • Using pulsed operation allows using a dynamic system to separate the fuel from the propellant, meaning to prevent uranium from escaping out the exhaust so the engine will be more closed-cycle than open cycle. Details to follow.

Figure (a)

The reactor vessel is surrounded by a conventional nuclear reactor (not shown). It is designed to be powered up then powered down rapidly, to create an intense pulse of neutron flux (something like φ = 1014/cm2 sec) inside the reactor vessel.

Valve V opens, and into the reactor vessel is injected a slug of hydrogen propellant, containing a subcritical piece of U235. The valve snaps shut behind the slug.

Note that the U235 is off-center inside the slug, further away from the exhaust nozzle than most of the propellant. This is so when the U235 explodes, most of the hydrogen propellant will be blown out the exhaust nozzle before any of the fissioning U235 reaches the nozzle exit.

The U235 is still fully embedded in the propellant, none of the U235 is exposed. This is so all the nuclear explosion energy hits the propellant, instead of frying the interior of the reactor vessel.

A tiny "trigger" piece of U235 is injected at high velocity down pipe T.

Figure (b)

When the trigger enters the subcritical U235, the surrounding nuclear reactor simultaneously pulses an intense neutron flux. The assembly becomes prompt critical and a small nuclear explosion ensues. This heats the hydrogen propellant which is pushed out the exhaust nozzle, creating thrust. Propellant on the other sides of the explosion protecting the reactor vessel from thermal radiation.

Figure (c)

Just before the unburnt U235 and fission fragment cloud escapes through the exhaust nozzle, the nozzle plug P closes the nozzle. The nozzle plug has to close at a rate of about 104 cm/sec, which can be done with a plug driven by pressurized gas. The hot propellant / unburnt U235 / fission fragment cloud is trapped inside the reactor vessel. This is sucked out of the reactor vessel through pipe E.

The gases are sent through a heat radiator to be cooled off. Then they enter a fuel-propellant reprocessing plant. This separates the three ingredients. The fission fragments are disposed of. The hydrogen propellant is sent to the propellant tank. The unburnt U235 is carefully fabricated into subcritial fuel masses, being very careful not to let a critical mass accidentally accumulate. The subcritical masses are sent to the fuel storage unit.

Open Cycle

The open-cycle gas core engine has a radioactive exhaust, there is no getting around it. So the first thing you have to do is estimate the radiation hazard and ensure the crew has adequate radiation shielding.

The second thing to do is find a design that does not wastefully allow expensive un-burnt uranium to escape out the tailpipe. Again the general rule is that as long as the hydrogen mass expelled is 25 to 50 times a high as the uranium mass expelled the uranium losses are within acceptable limits. The various open-cycle designs use different strategies to make the hydrogen to uranium exhaust flow ratio as high as possible.

Crew radiation dose from plume of Gas-Core rocket

In the open-cycle gas-core nuclear rocket concept the heat source is fissioning uranium gas. This released heat is radiated to and absorbed by the hydrogen propellant, The heated propellant is exhausted through a nozzle, producing thrust. The fission fragments that are formed and the unfissioned uranium fuel are also exhausted into the vacuum of space. As the plume is formed, the crew is exposed to gamma radiation from the fission fragments in the plume.

The radiation dose to the crew from the fission fragments in the plume can be separated into two components. Component one results from the fact that there is a microscopic amount of plume material that has sufficient kinetic energy to flow back towards the vehicle. Some of this material will strike and stick to the vehicle. Since this material will contain fission fragments, these gamma radiation sources will stay with the crew throughout the entire trip and this dose could represent a significant source of radiation. Masser(3) has estimated this dose and has concluded it would be less than 10-3 rem for a typical manned Mars mission.

Component two of the dose results from the fission fragment distribution throughout the entire plume volume and is potentially much larger than component one. Since the plume contains over 99 percent of the exhausted material, 99 percent of the fission fragments will be in the plume. It is the purpose of this paper to estimate the radiation dose rate and total dose to the crew from the fission fragments in the plume for four specific missions to the planet Mars.

Another source of radiation is caused by the delayed decay of the fission fragments that are passing through the nozzle. This includes delayed neutrons which can cause secondary fissioning and gamma's. This source, however, has not been included. There is another radiation source associated with the gas-core reactor, that of the reactor core. This radiation source, along with solar radiation, must be ultimately considered when total dose rates to the crew are evaluated. This study, however, is concerned only with that part of the total radiation problem that arises from the fission fragments in the plume volume.

3. Masser, C. C., "Radiation Hazzard from Backflow of Fission Fragments from the Plume of a Gas-Core Nuclear Rocket," Research on Uranium Plasmas and Their Technological Applications, SP- 236, 1971, NASA, Washington, D.C.

(ed note: the equations used to draw these graphs are in the document. I didn't bother to include them since they involve calculus. The radiation doses in the graphs give spacecraft designers the radiation shielding requirements)

1. For the most probable fission fragment retention, time of 100 seconds, and crew nozzle separation of 100 meters, the radiation dose varied from 170. to 36. rem for the 80 and 200 day round trip times respectively. Five centimeters of lead shielding would reduce the radiation dose by two orders of magnitude, thereby protecting the crew. The increase in vehicle weight would be insignificant. For example, a shield of five centimeters thickness and four meters in diameter would add 7120 kilograms to the vehicle gross weight of 0.94 million kilograms. Also additional attenuation is available In the form of liquid hydrogen propellant, spacecraft structure, nuclear fuel, equipment, and stores.

2. For the trip times included in this analysis the total radiation dose to the crew is proportional to the energy required for the mission. Therefore, within the ranges used in this analysis one can estimate the crew radiation dose by knowing the energy needed for the mission.

3. For the crew-nozzle separation of 100 meters, approximately 50 percent of the plume radiation is received from the first 0.1 kilometer into the plume. This percentage is increased to 90 percent for 1 kilometer and 100 percent for 100 kilometers into the plume.

4. For an 80 day round trip to Mars, with a crew-nozzle separation distance of 100 meters, the radiation dose varied from about 0.5 to 1670. rem for fission fragment retention times of 10,000 and 10 seconds, respectively.

5. For all cases, increasing the crew distance from 100 to 200 meters from the nozzle exit reduced the unshielded radiation dose by half.

General Open Cycle
Open Cycle
Propulsion SystemGas Core NTR
Exhaust Velocity35,000 m/s
Specific Impulse3,568 s
Thrust3,500,000 N
Thrust Power61.2 GW
Mass Flow100 kg/s
Total Engine Mass200,000 kg
Uranium Hexafluoride
ReactorGas Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power3 kg/MW
Engine mass30-200 tonne
T/W 11.9 to 1.8
Open Cycle 2
Propulsion SystemGas Core NTR
Exhaust Velocity50,000 m/s
Specific Impulse5,097 s
Thrust5,000,000 N
Thrust Power0.1 TW
Mass Flow100 kg/s
Uranium Hexafluoride
RemassLiquid Hydrogen
Specific Power2 kg/MW
Engine mass30-200 tonne
T/W 17.0 to 2.5
Open Cycle 3
Thrust Power GW
Exhaust velocity25,000 to 69,000 m/s
Thrust19,600 to 108,000 n
Engine mass40 to 110 tonne
T/W0.05 to 0.10
Operating Pressure400 to 2000 atm
Open Cycle MAX
Exhaust Velocity98,000 m/s
Specific Impulse9,990 s
Thrust3,000,000 N
Thrust Power0.15 TW
Mass Flow31 kg/s
Total Engine Mass15,000 kg
Uranium Hexafluoride
Propulsion SystemGas Core NTR
Exhaust Velocity35,316 m/s
Specific Impulse3,600 s
Thrust3,500,000 N
Thrust Power61.8 GW
Mass Flow99 kg/s
Uranium Hexafluoride
RemassLiquid Hydrogen
Wet Mass433,000 kg
Dry Mass268,000 kg
Mass Ratio1.62 m/s
ΔV16,943 m/s

Gaseous core fission / nuclear thermal rocket AKA consumable nuclear rocket, plasma core, fizzer, cavity reactor rocket. The limit on NTR-SOLID exhaust velocities is the melting point of the reactor. Some engineer who obviously likes thinking "outside of the box" tried to make a liability into a virtue. They asked the question "what if the reactor was already molten?"

Gaseous uranium is injected into the reaction chamber until there is enough to start a furious chain reaction. Hydrogen is then injected from the chamber walls into the center of this nuclear inferno where it flash heats and shoots out the exhaust nozzle.

The trouble is the uranium shoots out the exhaust as well. This not only makes the exhaust plume dangerously radioactive but it also wastefully allows expensive unburnt uranium to escape before it contributes to the thrust.

The reaction is maintained in a vortex tailored to minimize loss of uranium out the nozzle. Fuel is uranium hexaflouride (U235F6), propellant is hydrogen. However, in one of the designs, U235 is injected by gradually inserting into the fireball a long rod of solid uranium. The loss of uranium in the exhaust reduces efficiency and angers environmentalists.

In some designs the reaction chamber is spun like a centrifuge. This encourages the heavier uranium to stay in the chamber instead of leaking into the exhaust. This makes for a rather spectacular failure mode if the centrifuge's bearings seize.

You can find more details here.

The thermal radiation from the fission plasma is intended to heat the propellant. Alas, most such engines use hydrogen as the propellant, which is more or less totally transparent to thermal radiation. So the thermal stuff goes sailing right through the hydrogen (heating it not at all) then striking the reaction chamber walls (vaporizing them).

To remedy this sorry state of affairs, gas-core designers add equipment to "seed" the propellant with something opaque to thermal radiation. Most of the reports suggest tungsten dust, with the dust size about the same as particles of smoke, about 5% to 10% seeding material by weight. The seeding absorbs all but 0.5% of the thermal radiation, then heats up the hydrogen propellant by conduction. The chamber walls have to cope with the 0.5%.

Most of the reports I've read estimate that the reaction chamber can withstand waste heat up to 100 megawatts per square meter before the chamber is destroyed. For most designs this puts an upper limit on the specific impulse at around 3,000 seconds.

However, if you add a heat radiator to cool the reaction chamber walls and the moderator surrounding the reaction chamber, you can handle up to about 7,000 seconds of specific impulse. The drawback is the required heat radiator adds lots of mass to the engine. A typical figure is of the total mass of a gas core engine with radiator, about 65% of the mass is the radiator.

Another fly in the ointment is that the proposed seeding materials turn transparent and worthless at about the 10,000 second Isp level. To push the specific impulse higher a more robust seeding material will have to be discovered. Since current heat radiators cannot handle Isp above 7,000 seconds, robust seeding is not a priority until better radiators become available.

Yet another challenge is that 7% to 10% of the fission plasma power output is not in the form of thermal radiation, but instead neutrons and gamma rays. Which the propellant will not stop at all, seeded or not. This will penetrate deep into the chamber walls and moderator (since gamma-rays are far more penetrating than x-rays), creating internal waste heat.

Sub 3,000 Isp designs deal with radiation heat with more regenerative cooling. Higher Isp need even more heat radiators.

Most designs in the reports I've read use 98% enriched uranium-235 (weapons-grade). The size of the reaction chamber can be reduced somewhat by using uranium-233 according to this report.

The reaction chamber size can be reduced by a whopping 70% if you switch to Americium-241 fuel according to this report. The drawback is the blasted stuff is $1,500 USD per gram (which makes every gram that escapes un-burnt out the exhaust financial agony). The short half-life means there is no primordial Americium ore, you have to manufacture it in a reactor via nuclear transmutation. The report estimates that for a 6 month brachistochrone trajectory the spacecraft would need about 2,000 kilograms of the stuff. Which would be a cool three million dollars US. I'm sure the price would drop if dedicated manufacturing sites were established to create it.

If used for lift off it can result in a dramatic decrease in the property values around the spaceport, if not the entire country. An exhaust plume containing radioactive uranium is harmless in space (except to the crew) but catastrophic in Earth's atmosphere.

Amusingly enough, this is the best match for the propulsion system used in the TOM CORBETT: SPACE CADET books. However the books are sufficiently vague that it is possible the Polaris used a nuclear lightbulb. According to technical advisor Willy Ley, "reactant" is the hydrogen propellant, but the books imply that reactant is the liquid uranium.



The temperature limitations imposed on the solid core thermal rocket designs by the need to avoid material melting can be overcome, in principle, by allowing the nuclear fuel to exist in a high temperature (10,000 — 100,000 K), partially ionized plasma state. In this so-called "gaseous- or plasma-core" concept, an incandescent cylinder or sphere of fissioning uranium plasma functions as the fuel element. Nuclear heat released within the plasma and dissipated as thermal radiation from its surface is absorbed by a surrounding envelope of seeded hydrogen propellant that is then expanded through a nozzle to provide thrust. Propellant seeding (with small amounts of graphite or tungsten powder) is necessary to insure that the thermal radiation is absorbed predominantly by the hydrogen and not by the cavity walls that surround the plasma. With the gas core rocket (GCR) concept Isp values ranging from 1500 to 7000 s appear to be feasible [Ref. 26]. Of the various ideas proposed for a gas core engine, two concepts have emerged that have considerable promise: an open cycle configuration, where the uranium plasma is in direct contact with the hydrogen propellant, and a closed-cycle approach, known as the "nuclear light bulb engine" concept, which isolates the plasma from the propellant by means of a transparent, cooled solid barrier.

Porous Wall Gas Core Engine

The "open cycle," or "porous wall," gas core rocket is illustrated in Fig. 9. It is basically spherical in shape and consists of three solid regions: an outer pressure vessel, a neutron reflector/moderator region and an inner porous liner. Beryllium oxide (BeO) is selected for the moderator material because of its high operating temperature and its compatibility with hydrogen. The open cycle GCR requires a relatively high pressure plasma (500 — 2000 atm; 1 atm = 1.013 × 105 N/m2 ) to achieve a critical mass. At these pressures the gaseous fuel is also dense enough for the fission fragment stopping distance to be comparable to or smaller than the dimensions of the fuel volume contained within the reactor cavity. Hydrogen propellant, after being ducted through the outer reactor shell, is injected through the porous wall with a flow distribution that creates a relatively stagnant non-recirculating central fuel region in the cavity. A small amount of fissionable fuel (1/4 to 1 % by mass of the hydrogen flow rate) is exhausted, however, along with the heated propellant.

Because the uranium plasma and hot hydrogen are essentially transparent to the high energy gamma rays and neutrons produced during the fission process, the energy content of this radiation (~7—10% of the total reactor power) is deposited principally in the solid regions of the reactor shell. It is the ability to remove this energy, either with an external space radiator or regeneratively using the hydrogen propellant, that determines the maximum power output and achievable Isp for the GCR engines. To illustrate this point, an open cycle engine with a thrust rating of 220 kN (50,000 lbf) is considered. We assume that 7% of reaction energy Prx reaches the solid, temperature-limited portion of the engine and that the remainder is converted to jet power at an isentropic nozzle expansion efficiency of ηj. Based on the realtionships between Isp, reactor power, and propellant flow rate (ṁp) given below.

(ed note: elsewhere in this website, ṁ is called "m-dot")

0.93·Prx(MW) = 4.9×10-6·F(N)·Isp(s) / ηj

0.93·Prx(MW) = 4.9×10-5·ṁp(kg/s)·Isp2(s) / ηj

a 5000 s engine generating 7500 MW of reactor power will require a flow rate of 4.5 kg/s at rated thrust. If the hydrogen is brought into the cavity at a maximum overall operating temperature of 1400 K, no more than 1.2% of the total reactor power (~17% of the neutron and gamma power deposited in the reactor structure) can be removed regeneratively (ṁp cp ΔT ≈ 90 MW). Total removal requires either (1) operating the sold portions of the engine at unrealistically high temperatures (>11,000 K at ṁp = 4.5 kg/s) or (2) increasing the propellant flow rate substantially to 36.8 kg/s (at 1400 K), which reduces the engine's Isp to 1750 s. "Closed cooling cycle" space radiator systems have been proposed [Ref. 27] as a means of maintaining the GCR's operational flexibility. With such a system, adequate engine cooling is possible even during high Isp operation when the hydrogen flow is reduced. Calculations performed by NASA/Lewis Research Center [Ref. 28] indicate that specific impulses ranging from 3000 to 7000 s could be attained in radiator-cooled, porous wall gas core engines.

The performance and engine characteristics for a 5000 s class of open cycle GCRs are summarized in Table 4 for a range of thrust levels. The diameter of the reactor cavity and the thickness of the external reflector/moderator region are fixed at 2.44 m and 0.46 m, respectively, which represents a near-optimum engine configuration. The engine weight (Mw) is composed primarily of the pressure vessel (Mpv); radiator (Mrad); and moderator (Mmod).

Table 4
Characteristics of 5000 s Porous Wall Gas Core Rocket Engines







  1. For a hydrogen cavity inlet temperature of 1400 K and a heat deposition rate that is 7% of the reactor power, the ratio of radiated to total reactor power is a constant equal to 5.8%.
  2. The weight of the spherical pressure vessel is based on a strength-to-density value of 1.7×l05 N-m/kg [Ref. 29] which Is characteristic of high strength steels.
  3. Used in these estimates is a radiator specific mass of 145 kg/MW [Ref. 28] which is based on a heat rejection temperature of 1225 K and a radiator weight per unit surface area of 19 kg/m2
  4. Density of BeO is 2.96 mT/m3.

By fixing the engine geometry in Table 4 the mass of the BeO moderator remains constant at 36 mT. However, the pressure vessel and radiator weights are both affected by the thrust level. While the radiator weight increases in proportion to the extra power that must be dissipated at higher thrust, the reason for the increase in pressure vessel weight is slightly more subtle. For a constant Isp engine an increase in thrust is achieved by increasing both the reactor power and hydrogen flow rate. In order to radiatively transfer this higher power to the propellant, the uranium fuel temperature increases, necessitating an increase in reactor pressure to maintain a constant critical mass in the engine. Accommodating this increased pressure leads to a heavier pressure vessel. (In going from 22 kN to 440 kN, the engine pressure rises from 570 atm to 1780 atm).

As Table 4 illustrates, the moderator is the major weight component at lower thrust levels (<110 kN) while the radiator becomes increasingly more important at higher thrust. At thrust levels of 220 kN and above, the radiator accounts for more than 50% of the total engine weight. There is therefore a strong incentive to develop high temperature (~1500 K) liquid metal heat pipe radiators that could provide significant weight reductions in the higher thrust engines.

Table 4 also shows an impressive range of specific powers (alphas) and engine thrust-to-weight ratios for the thrust levels examined. The F/Mw ratio for the 22 kN engine is over two orders of magnitude higher than the 5000 s nuclear-powered MPD electric propulsion system proposed in the Pegasus study [Ref. 30]. For manned Mars missions the higher acceleration levels possible with the GCR can lead to significant (factor of 5) reductions in trip time compared to the Pegasus system.


(ed note: calculated estimates of gas core nuclear rocket engine weights for specific impulses ranging from 3000 to 7000 seconds and for engine thrusts ranging from 4400 to 440,000 newtons. Contains useful equations for calculating the mass of various engine components.)


Virtually all existing or proposed rocket propulsion engines can be categorized as either high-thrust systems or high-specific-impulse systems. What is really needed for fast interplanetary travel is both characteristics, namely a high specific impulse (3000 sec or greater), and an engine thrust-weight ratio that is in the range from to 10-1. The characteristics of a gas-core nuclear rocket engine are examined in this study to see how closely it meets these requirements.

Calculations were carried out to estimate gas-core engine weights for specific impulses ranging from 3000 to 7000 seconds and thrust levels from 4.4×103 to 4.4×105 newtons. A vapor-fin space radiator operating at 1100 K was incorporated into the engine system to dispose of waste heat not regeneratively removed by the hydrogen propellant. The total engine weight was composed of the individual weights of the radiator, the reactor moderator-reflector materials, the pressure shell, the nozzle, and the propellant turbopump. The study produced the following results and conclusions:

1. Gas-core engines have the potential of producing a specific mass in the range 0.6 to 0.02 kilogram of weight per kilowatt of thrust power.

2. For a specific impulse of 5000 seconds and a thrust of 1.1×105 newtons, engine weight is estimated to be 91 000 kilograms. This weight is composed of about equal proportions of radiator, moderator, and pressure shell weights. For the entire range of specific impulses and thrust levels of this study, engine weight varied from 35 000 to 380 000 kilograms.

3. Engine weight increases with increasing specific impulse and with increasing thrust level. For a given specific impulse, higher thrust-weight ratios occur at higher thrust levels because engine weight does not increase as fast as the thrust does.

Figure 1(a) illustrates schematically how this basic notion might be translated into a rocket engine. It is not unreasonable to picture this kind of engine as a nuclear "sun" with the central fireball and surrounding gas flow contained within a chamber' surrounded by structural materials. The analogy is not exact, of course, because the heat generation is due to nuclear fission rather than fusion. However, in both cases tha amount of energy that can be generated in, and released from, the fireball is essentially unlimited. There is, however, a limitation on how much energy can be absorbed by the hydrogen and turned into thrust without overheating the cavity wall or the exhaust nozzle. It is the amount of energy that reaches various solid, temperature-limited regions of the engines that ultimately limits the power generation and therefore the specific impulse.

The proposed reactor shown in figure 1(a) is basically spherical. It is composed of an outer pressure vessel, a region of heavy-water reflector, a high-temperature beryllium moderator region, an inner heavy-water moderator, and finally a porous or slotted cavity liner. Approximately 7 to 10 percent of the reactor power is deposited in these solid regions of the reactor due to attenuation of high-energy gamma and neutron radiation. This heat is removed either by a coolant in an external space radiator loop, or regeneratively by the hydrogen propellant before it enters the central reactor cavity, The beryllium region is operated at a temperature of about 1300 K and the radiator at 1100 K.

Uranium metal would have to be injected into this high-pressure region. Once inside the cavity, the uranium vaporizes and rises to temperatures sufficient to thermally radiate the energy that is generated by the fissioning uranium. A possible fuel injection technique might consist of pushing a thin rod of solid uranium metal at a high velocity through a shielded pipe (perhaps made of cadmium oxide) that penetrates the moderator. Some cooling of the uranium fuel and the shielded passage may be required to remove the heat that would be generated in the fuel as it passes through the moderator region. A 100-kilogram force would be required to drive a 0.15-centimeter diameter wire into a cavity with a pressure of 5.07×107 newtons per square meter. As it enters the cavity, the uranium instantly vaporizes and rises in temperature to about 55 000 K. Reactor startup could be achieved by first establishing the hydrogen flow. Next uranium particles would be blown into the dead cavity region to achieve nuclear criticality. The power would then be increased to a level sufficient to vaporize the incoming uranium rod.

The seeded hydrogen is heated solely by absorbing the thermal radiation from the fissioning uranium fireball. The cavity walls receive only about 1 or 0.5 percent of the thermal radiation from the fireball. This wall protection is accomplished by introducing about 1 percent by weight of a seeding material such as graphite or tungsten particles into the hydrogen. This same technique is used in the nozzle region to reduce the hydrogen radiation heat load and the hydrogen temperature near the nozzle wall to tolerable levels. Seed concentrations of about 1 to 10 percent are required here. Figure 1 shows that some cold hydrogen can be introduced through the nozzle walls directly from the plenum at the downstream end of the engine if it is required. This would tend to reduce the specific impulse.


The specific impulse of a gas-core rocket engine is limited by the fraction of the reactor power that reaches the solid, temperature-limited portions of the engine, and by how that heat is removed. It is an unavoidable characteristic of the nuclear fission process that about 7 to 10 percent of the energy release is high-energy gamma and neutron radiation that will go through the hydrogen gas but be stopped in the surrounding solid reactor structure.

This energy that is deposited in the moderator can be regeneratively removed by the incoming hydrogen propellant. There is, however, a limit to how much heat the hydrogen can accommodate. For a 3000-second specific impulse engine, 7 percent of the reactor power will heat all the hydrogen propellant to 2800 K before it enters the reactor cavity. To achieve a higher specific impulse would require the solid parts of the engine to operate at an unrealistically high temperature. If the reactor materials, including the porous cavity wall, were limited to a little over 1000 K and if only regenerative cooling were used, the specific impulse would be limited to 2000 seconds.

Higher specific impulses are possible by using an external radiator to reject part of the moderator heat to space. The radiator is shown schematically in figure 1. To bring the hydrogen into the reactor cavity at 1000 K for a specific impulse of 5000 seconds would require that the hydrogen remove no more than about 1 percent of the reactor power from the moderator, as shown in figure 2. The remaining 6 to 9 percent would have to be removed by the radiator loop.

It appears that the ultimate limitation on specific impulse of a gas-core engine will depend on the ability to absorb the thermal radiation from the fuel in the hydrogen so that the cavity wall and the nozzle wall do not receive an excessive heat flux. Based on current estimates of the optical absorption and emission properties of the gases involved, a recent Lewis in-house study indicates that the maximum specific impulse is in the range 5000 to 7000 seconds.


The engine weight analysis used for this study is the same as was presented in reference 4, except for the addition of a space radiator and the elimination of a specific equation for fuel volume as a function of the hydrogen-to-uranium-mass-flow ratio (how many units of hydrogen propellant are expended in the exhaust before one unit of uranium is lost). The engine weight is taken to be the sum of the individual weights of the moderator, pump, nozzle, pressure shell, and radiator:


An initial series of calculations were made to select a "best" cavity diameter and moderator thickness combination. This preliminary optimization was done at values of specific impulse (5000 sec) and thrust (4.4×104 N) that are centered in the ranges covered in this study. One cavity diameter and one moderator thickness were selected on this basis, and then held constant for all subsequent variations of specific impulse and thrust. Thus, after this initial reactor optimization, the moderator weight was not a variable in this study.

Engine Pressure

In order to calculate the weights of the nozzle, turbopump, and pressure shell, it was necessary to calculate the pressure required to have a critical mass in the engine. This was obtained from the following equation:


where P is the reactor pressure in atmospheres, Mc, is the critical mass in kilograms, F is the engine thrust in newtons, Isp is the specific impulse in seconds, Dc is the reactor cavity diameter in meters, and VF is the fraction of the reactor cavity filled with fuel. Equation (2) is more general than the form used in reference 4 where a specific relation between fuel volume fraction and hydrogen-touranium-mass-flow ratio was used to eliminate VF from equation (2). The present study was carried out for a fuel volume fraction of 0.25. Recent fluid mechanics experiments using air/air indicate that this value should be attainable for hydrogen-to-uranium-flow ratios in the range 100 to 400.

Nozzle, Turbopump, And Pressure Shell


where the component weights are in kilograms, F is thrust in newtons, Isp is specific impulse in seconds, P is reactor pressure from equation (2) in atmospheres, and Rs is the inside radius of the pressure shell in meters.

The radiator weight estimate was based on a recent study of a vapor-fin for space power systems. The vapor-fin design would weigh 290 kilograms per megawatt of radiated power, based on operating the radiator at 945 K. For this study it was assumed that the same radiator, or at least one of the same weight per unit surface area (19 kg/m2 of plan form area), could be operated at 1100 K. This gives a weight of 145 kilograms per megawatt of radiated power:


Equations (2) to (6) were used to obtain the weight of each engine component. Equation (1) was used to obtain the total engine weight. For this study, calculations were carried out for specific impulses of 3000, 5000, and 7000 seconds, and for engine thrusts from 4.4×103 to 4.4×105 newtons.

It may be necessary to operate the radiator at a pressure less than that of the reactor cavity in order to keep the lightweight vapor-fin design. For example, the pressure stress in the radiator tube walls would range from 10.14×107 to 50.7×107 newtons per square meter for internal tube pressures ranging from 10.14×106 to 5.07×107 newtons per square meter, respectively. This same pressure stress range could be reduced by a factor of 3 by increasing the tube wall thickness such that the overall radiator weight would increase by about 20 percent. In an actual engine design, one might not choose to do this, but instead operate the radiator at a lower pressure than that of the reactor. This would then require a pump to increase the radiator discharge pressure to that inside the reactor pressure vessel.


The engine weight results are presented and discussed in this section. First, the effect of varying the cavity diameter and the moderator thickness is presented. Based on these results, one cavity diameter and one moderator thickness are selected for the remainder of the calculations. For this fixed engine geometry, the effect of thrust level on engine weight is determined for a specific impulse of 5000 seconds. Next, the effect of specific impulse on engine weight is presented over a range of thrust levels. Finally, these results are presented in terms of a parameter commonly used to describe lowthrust propulsion devices, engine specific mass, which is the ratio of engine weight to thrust power (in kg/kW).

Effect of Cavity Diameter and Moderator Thickness

Changes in cavity diameter or in moderator thickness cause two effects on engine weight. One effect is that the weight of moderator material is changed. The other effect is that the uranium density required for criticality is changed. This changes the required reactor pressure, which, in turn, results in a change in the pressure shell weight.

These two influences on engine weight tend to oppose each other. For example, reducing the moderator thickness reduces the moderator weight, but increases the pressure required for criticality. Thus, there is some optimum moderator thickness that gives a minimum engine weight. It is possible, however, that the engine pressure at this minimum-weight geometry would be unrealistically high, so that one might choose to operate at some near but off-optimum configuration that has a somewhat lower pressure.

Engine weight was calculated for five combinations of cavity diameter and moderator thickness. The results are shown in figure 3. The critical mass requirements are listed in table I. These engine weight calculations were carried out for a specific impulse of 5000 seconds and a thrust level of 4.4×104 newtons. Both of these values are centered within the ranges covered in this study.

Cavity diameters of 2.4, 3.6, and 4.9-meters were used with a constant moderator thickness of 0.76 meter. Moderator thicknesses of 0.61, 0.76, and 0.91 meter were used with a constant cavity diameter of 3.6 meters. Within these ranges, reductions in either parameter caused a decrease in engine weight but an increase in engine pressure. A cavity diameter of 2.4 meters with a moderator thickness of 0.76 meter produced the lightest engine, which weighed 64 000 kilograms. The reactor pressure for this engine was 7.8×107 newtons per square meter.

Further reduction of cavity diameter below 2.4 meters would probably have produced a slightly lighter engine, but at the expense of an extremely high pressure. This is shown in figure 4. On the basis of these results, a 2.4-meter cavity diameter and a 0.76-meter moderator thickness were selected as representing a near-optimum engine configuration. The remaining calculations were carried out using this one engine geometry.

Effects of Thrust Level on Engine Weight

Higher thrust requires a heavier engine. The component weights are shown in figure 5 for engine thrust varying from 0 to 1.1×105 newtons at a specific impulse of 5000 seconds. For a thrust below about 5×104 newtons, the radiator weight is not too important, compared to the moderator and the pressure shell weights. At a thrust of 1.1×105 newtons, the radiator, pressure shell, and moderator each contribute about one-third of the total engine weight.

For higher thrusts, the radiator weight begins to dominate. This is shown in table II. At a thrust of 2.2×104 newtons, the radiator only contributes 6400 kilograms to the total engine weight of 51 000 kilograms, or about 12 percent. At a thrust of 2.2×105 newtons, the radiator accounts for 64 000 kilograms out of 133 000 kilograms, or almost 50 percent. This indicates that for thrusts above 2.2×105 newtons, at this specific impulse of 5000 seconds, significant weight reductions can be achieved if higher temperature radiators can be developed. For example, the radiator weight could be cut in half by operating at 1300 K instead of 1100 K. All the calculations of this study were done for a radiator temperature of 1100 K.

Effect of Specific Impulse on Engine Weight

Higher specific impulses require heavier engines, at a given thrust level. This is shown in figure 6. For a thrust of 4.4×104 newtons, engine weights of 50 000, 64 000, and 73 000 kilograms are required for specific impulses of 3000, 5000, and 7000 seconds, respectively.

At a specific impulse of 3000 seconds, a radiator may not be necessary. If the hydrogen propellant enters the reactor at 2800 K, it can regeneratively remove all the gamma and neutron heat deposition from the moderator region. This produces a lighter engine, as shown by the dashed curve in figure 6. Whether one would actually choose to operate the moderator at a little over 2800 K in order to achieve the lower weight would depend on a number of factors, such as the particular mission involved and the effect of moderator temperature on engine reliability and life. The solid curves in figure 6 are based on a hydrogen cavity inlet temperature of 1400 K. Table III lists the percent of reactor power that must be radiated away for this temperature.

Gas-Core Specific Mass

For low-acceleration systems such as electric thrusters, it is useful to characterize the propulsion device in terms of its specific mass. This parameter α is in kilograms of powerplant weight per kilowatt of thrust power. It can be related to the engine thrust-weight ratio as follows. The thrust power, or jet power as it is sometimes called, is given by 1/2 (F×Isp×g), which is simply the kinetic energy in the jet exhaust. SP Using this relation, the specific mass α, in kilograms per kilowatt, is


where the specific impulse is in seconds and the engine thrust-weight ratio is dimensionless.

Figure 7 shows the results of the present study presented on this basis. The specific mass of a gas-core engine varies from a high of 0.6 to a little less than 0.02 for specific impulses from 3000 to 7000 seconds and thrust levels from 4.4×10+3 to 4.4×10+5 newtons. Higher specific impulse or higher thrust produces a lower, and therefore better, specific mass.


An analysis has been carried out to determine the characteristics of a low-thrust, high-specific-impulse, gas-core, nuclear rocket engine. The latest information on reactor critical mass requirements, radiant-heat-transfer properties, and fluid mechanics were used. For specific impulses above 3000 seconds, it was necessary to incorporate a space radiator as an engine system component. Engine weight was calculated for specific impulses ranging from 3000 to 7000 seconds, and for thrust levels from 4.4×103 to 4.4×105 newtons. Radiator weight estimates were based on an operating temperature of 1100 K. The calculations indicate the following results:

1. Gas-core engines have the potential of producing a specific mass in the range 0.6 to 0.02 kilogram of weight per kilowatt of thrust power.

2. For a specific impulse of 5000 seconds and a thrust of 1.1×105 newtons, engine weight is estimated to be 91 000 kilograms. This weight is composed of about equal proportions of radiator, moderator, and pressure shell weights. For the entire range of specific impulses and thrust levels of this study, engine weight varied from 35 000 to 380 000 kilograms.

3. Engine weight increases with increasing specific impulse and with increasing thrust level. For a given specific impulse, higher thrust-weight ratios occur at higher thrust levels because engine weight does not increase as fast as the thrust does.


This is from Mini Gas-Core Propulsion Concept by R. E. Hyland (1971) and A Study Of The Potential Performance And Feasibility Of A Hybrid-Fuel Open Cycle Gas Core Nuclear Thermal Rocket by Lucas Beveridge (2016).

As previously mentioned open-cycle gas core engines solve the "reactor got so hot it vaporized" problem by starting out with the reactor already vaporized. The primary problem is how to prevent the uranium gas from prematurely escaping out the exhaust nozzle, but the secondary problems are pretty bad as well.

To start the uranium fissioning, you need a certain amount of uranium at a certain density surrounded by enough neutron reflectors to kick stray neutrons back into play. Sadly, by definition, gaseous uranium has a much lower density than solid uranium. As it works out, for the engine to require a non-outrageous critical mass of uranium and a non-outrageous reaction chamber volume, the core pressure has to be very very high. Which will require a massive pressure vessel. Which makes the engine mass skyrocket. Which savagely cuts into the available payload mass and seriously degrades the engine's thrust-to-weight ratio.

Oh, calamity and woe! How can this problem be remedied?

Isp1,600 sec
Exhaust Velocity15,696 m/s
Thrust450 N
Reactor Mass2,200 kg
Pressure Shell Mass3,100 kg
Radiator Mass4,930 kg
Total Mass10,230 kg
Outer Diameter1.22 m
Cavity Diameter0.61 m
233U Plasma Diameter0.43 m
233U Plasma Mass1.42 kg
233U Plasma Power
4.5 MW
233U Driver Power
15.9 MW
Total Engine Power
20.4 MW
Radiator Alpha310 kg/MW
Pressure51 MPa
Exhaust Temp4,000 to
5,000 K


Robert Hyland pondered the problem until the question arose "is it really necessary for all the uranium to be gaseous?"

Hyland's solution was to embed a small solid core reactor in the walls of the chamber, as sort of a reactor layer. This is called the "driver core." It is far enough from the furious heat raging inside the chamber so it wouldn't melt. The driver core produces heat, but the important part is it produces neutrons. This makes the interior of the chamber so neutron-rich that the gaseous uranium does not have to be under such high pressure. In other words, the extra neutrons from the driver core lower the required critical mass of uranium gas inside the chamber.

This allows the engine to get away with using a much less massive pressure vessel, which lowers the engine mass, which reduces the payload reduction and increases the thrust-to-weight ratio.

Hyland said "I shall call him 'Mini-Gas Core.'" Lucas Beveridge called it the hybrid-fuel engine, since it uses both solid and gaseous uranium.

Hyland scaled this to have an engine power of 20.4 MW, which implied a meager thrust of only 450 N. He thought it might be useful for unmanned space probes.

So part of the total engine power is produced by the driver core (233U Driver Power or Psolid) and part of the total engine power is produced by the uranium plasma inside the chamber (233U Plasma Power or Pgas). Only Pgas is used to heat up the propellant to create thrust. Most of Psolid is just waste heat, a fraction of it is used to create neutrons to supercharge the uranium plasma. So heat radiators will be needed to get rid of the 15.9 megawatts worth of Psolid waste heat.

Ptotal = Pgas + Psolid = 20.4 MW

εgas = Pgas / Ptotal = 0.221

εsolid = Psolid / Ptotal = 0.780

εgas is the ratio of power in the 233U Plasma to the total. Hyland's Mini-Gas Core has a εgas of 0.221, or only 1/5th of the power is in the plasma. Beveridge found that was too low, and was focusing on a Low-ε engine with εgas = 0.51 and a High-ε engine with εgas = 0.673.


Engine Common
Thrust300,000 N
Total Engine
Power (Ptotal)
3 GW
Engine Mass<36,000 kg
Low-ε Engine
Isp1,600 sec
Exhaust Vel15,700 m/s
High-ε Engine
Isp1,950 sec
Exhaust Vel19,100 m/s
Inert Mass36,000 kg
Payload Mass
(incl. crew)
62,800 kg
Dry Mass98,800 kg
Propellant Mass33,620 kg
Wet Mass132,420 kg
Mass Ratio1.34
Exhaust Vel19,100 m/s
ΔV5,590 m/s
Initial Accel2.27 m/s
(0.23 g)

Remember that εgas is the ratio of power in the 233U Plasma to the total. Hyland's Mini-Gas Core has a εgas of 0.221, or only 1/5th of the power is in the plasma. Beveridge focused on a Low-ε engine with εgas = 0.51 and a High-ε engine with εgas = 0.673.

Beveridge found that it was not optimal if εsolid is larger than 0.50, that is, if more than 50% of the total engine power comes from the driver core. Hyland's design had εsolid = 0.780, or almost 80%. This means the Hyland's driver core needed more cooling than the cavity wall.

The obvious solution won't work. Rockets in general use cold propellant to cool off engine components. So one would think the solution is to cool off the driver core with propellant, then send it into the chamber to be superheated by the uranium plasma. But since Hyland's engine only had about 20% of the total energy generated by the uranium plasma, the plasma would not significantly heat the propellant more than the driver core already had. Bottom line is the performance would be about the same as a garden-variety NERVA solid core reactor, but with an engine that was much more expensive.


In Hyland's engine the driver core produces 78% of the power. Since the driver core is a solid-core reactor, it cannot go above 3,000K or it will melt. Since the uranium plasma is only 22% it can only heat the propellant about 500K more for a total exhaust temperature of 3,500K. Which is about the same as a bog-standard solid-core NTR.

But if both the driver core and the uranium plasma produce 50% of the power, then the gas core can add about 2,400K more for a total exhaust temperature of 5,400K which is much better than a solid-core NTR. But wait! There's more! Above a temperature of about 5,000K, molecular hydrogen propellant dissociates into monoatomic hydrogen (single-H). This could increase the exhaust velocity and specific impulse by up to a factor of 1.4 (i.e., √2).

To avoid that unhappy state of affairs, you have to use the un-obvious solution of using a heat radiator to cool the driver core. The trouble is that heat radiators add literally tons of penalty-mass to the engine. You will have to dial down the total engine power to control the heat radiator mass. The end result would be an engine with about the same mass as a standard nuclear-electric propulsion engine (NEP), fractionally more thrust, and drastically less specific impulse (Isp of 2,000 sec instead of 6,000 sec.) In which case it would be more advantageous to use NEP.

Since neither of those solutions works, Beveridge found a third option. Design the engine so that the driver core power is less than 50% of the total. This means the driver core can be cooled by propellant, and the uranium plasma will most certainly heat the propellant more than the driver core did. A heat radiator is used to cool the chamber from uranium plasma heat. Bottom line: high specific impulse and high power.

Beveridge did comparison studies on a pure open-cycle gas core, a Low-ε hybrid engine with εgas = 0.51 and a High-ε engine hybrid with εgas = 0.673. For comparison purposes they were all scaled to have a power level of 3 gigawatts. Unsurprisingly the low-ε had the lowest critical mass, the pure open-cycle had the highest, and the high-ε was somewhere in the middle.

Exhaust Velocity17,658 m/s
Specific Impulse1,800 s
Thrust17,800,000 N
Thrust Power0.2 TW
Mass Flow1,008 kg/s
Total Engine Mass127,000 kg
Uranium Hexafluoride
ReactorGas Core
RemassLiquid Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power1 kg/MW
Thrust Power0.495GW
Exhaust velocity22,000 m/s
Specific Impulse2,200 s
Thrust45,000 n
Engine mass66,000 kg
Diameter5 m
Length5 m
Fuel Temp20,000° R
Propellant Temp10,000° R

Gaseous core coaxial flow fission / nuclear thermal rocket.

The basic problem of gas core nuclear rockets is ensuring that the hot propellant escapes from the exhaust nozzle, but the nuclear fuel does not. In this concept, the propellant and fuel are kept separate by a velocity differential. That is, a central, slow moving stream of fission fuel heats an annular, fast moving stream of hydrogen.

Yes, the uranium jet is aimed straight at the exhaust nozzle. But they figured the uranium loss would be acceptable as long as 25 to 50 times as much hydrogen propellant escapes compared to uranium fuel (measured by mass).

No, the concept does not work very well. In theory the difference in velocity should keep the uranium/plutonium and the hydrogen separate. Unfortunately the velocity differential at the boundry between the propellant and fuel generates shear forces. The fast hydrogen strips off uranium atoms from the slow fuel plume like a carpenter's plane (laminar and turbulent mixing processes). This means the hydrogen to uranium escape ratio drops below 25.

The concept seems to have been abandoned.

This is from Estimates Of Fuel Containment In A Coaxial Flow Gas-Core Nuclear Rocket (1970).

Again the idea is to have all the furiously hot hydrogen propellant go shooting through the exhaust nozzle, while trying to prevent from escaping as much as possible of the dangerously radioactive and hideously expensive uranium fuel. The point of the paper is to use computer simulations to draw graphs predicting how much fuel will escape given a specific propellant-to-fuel flow ratio. The end result of the calculation is the contained fuel mass (how much fails to escape) in the form of a dimensionless number called the "fuel volume fraction." This is the fraction of the cavity volume occupied by fuel.

The analysis uses a coaxial free-jet computer model along with custom-made eddy viscosity equations, neither of which they reveal in the paper. They assume a smooth inlet velocity profile. They also assume that the cavity is a cylinder with the diameter equal to the length.

They plot how the fuel volume fraction varies with different flow ratios, fuel radius, and fluid density. They looked at propellant-to-fuel flow ratios from 10 to 100, fuel-to-propellant density ratios from 1.0 to 4.7, and fuel-to-cavity radius ratios from 0.5 to 0.7. The predicted results more or less agrees with data from previous physical experiments. "More or less" is defined as within ±30%.

The vaporized uranium fuel stream in the axis is surrounded by the lighter, faster moving hydrogen propellant stream. The coaxial flow should in theory contain the fuel and keep it from escaping even at very high fuel temperatures. In practice though the containment is less than perfect. The large difference in velocity between the fuel and the propellant streams causes turbulent mixing. The goal is to predict the contained uranium fuel mass for various flow ratios (i.e., how much uranium does NOT escape out the tail-pipe).

You need to know the contained fuel mass:

  • So that the engine can be designed such that the contained full mass is above critical mass. Otherwise there is no nuclear fission and the engine just looks stupid sitting there with sputtering hydrogen flatulence.

  • So that the engine design can be optimized, selecting a engine parameters that push the fuel volume fraction as close to maximum as possible. You don't want to waste uranium, that stuff is more expensive than ink-jet printer ink.

  • So that the engine be designed such that the specific impulse and thrust meets the propulsion system requirements. Otherwise the project boss will be very angry.

A target engine design would have a desired fuel volume fraction of 0.20 and a propellant-to-fuel flow ratio of 50. The analysis indicates this can be achieved with a fuel-to-propellant density ratio of 1.0 and a fuel-to-cavity radius ratio of 0.7.


In theory an open-cycle gas-core NTR can achieve a specific impulse greater than 1,500 seconds (exhaust velocity greater than 14,700 m/s) and thrust on the order of 2,000,000 Newtons. The slower-moving uranium fuel stream is at about 55,000°C, the faster moving hydrogen propellant stream is heated by the uranium to about 5,500°C. According to one reference a desirable engine should contain enough fuel to give a fuel volume fraction of at least 0.20 at a propellant-to-fuel flow ratio of 50 or greater.

A solid rod of uranium is inserted into the engine where is is vaporized by fission heating to form the fuel vapor cloud. The analysis assumes that downstream of plane A-A the fuel is completely vaporized and flowing nearly parallel.

Flow Model

Figure 2 shows the mathematical model.

Continuity equation:

Momentum equation:

Mass diffusion equation (Schmidt number assumed to be Sc = 0.7):

Intitial and boundary conditions are:

Eddy Viscosity

For the region near the inlet:

and downstream

Having said that, the location of x12 is where ε1 = ε2. With the present calculation the cavity is far shorter than x12, so you can ignore equation (5).

Inlet Velocity Profiles

One of the references actually put a plane of porous material at plane A-A to smooth out the inlet velocity. Otherwise the turbulence reduces fuel containment.

The equation for smooth inlet velocity profile is:

The report generalizes the inlet velocity profile by using the variable RB (inlet velocity profile half-radius) which they set equal to the upstream buffer radius from one of the references. So RB/RF values (ratio of inlet buffer radius to fuel stream radius) in the reference of 1.14, 1.22, and 1.3 correspond to the radius ratios RF/RC (ratio of fuel stream radius to cavity radius) of 0.7, 0.6, and 0.5.

Fuel Volume Fraction

The fuel mass is calculated as the fuel volume fraction VF. This is the fraction of the cavity volume occupied by pure fuel vapor if it were gathered into a central volume at its original temperature and cavity pressure. Cavity volume is defined by planes A-A and B-B, and the streamline through r=RC at inlet (see figure 2).

The fuel volume fraction is:

With a known pure fuel density (for a specific cavity pressure and average temperature), VF is a direct meaure of the fuel mass contained inside a full-sized heated engine.


As mentioned above, the experiental data from the references were for radius ratios RF/RC (ratio of fuel stream radius to cavity radius) of 0.5, 0.6, and 0.7. Also from the references the experimental data for fuel-to-propellant density ratio was from 1.0 to 4.7.

In the graph, the white circles are experimental data for density ratio 1.0 and the black circles are experimental data for density ratio 4.7. The x-circle is for a desired engine design with a fuel volume fraction of 0.20 and a flow ratio of 50.

The upper curved line is drawn by the report's equation for density ratio of 1.0. The lower curved line is drawn by the equation for density ratio of 4.7.

The point of the report is that all but four of the 23 experimental data points fall within ±30% of the equation's curves.


By using cross-plotting and other clever mathematical tricks, the three figures 7a, 7b, and 7c can be collapsed into one graph: figure 8.

The report notes that eighty percent of the experimental data points fall within ±30% of the calculated correlating curve. So the equations appear to be close to predicting reality.

Vortex Confined
Vortex Confined
Vortex Confined
Exhaust Velocity19,620 m/s
Specific Impulse2,000 s
Thrust50,400 N
Thrust Power0.5 GW
Mass Flow3 kg/s
Total Engine Mass114,116 kg
Frozen Flow eff.75%
Thermal eff.70%
Total eff.53%
Uranium Hexafluoride
ReactorGas Core
Vortex Confined
RemassSeeded Hydrogen
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorNozzle
Specific Power231 kg/MW

The hotter the core of a thermodynamic rocket, the better its fuel economy. If it gets hot enough, the solid core vaporizes.

A vapor core rocket mixes vaporous propellant and fuel together, and then separates the propellant out so it can be expelled for thrust. Energy is efficiently transferred from fuel to propellant by direct molecular collision, radiative heat, and direct reaction fragment deposition.

The open-cycle arrangement illustrated accomplishes this by spinning the plasma mixture in a vortex maintained by tangential injection of preheated propellant from the reactor walls. The denser material is held to the outside of the cylindrical reactor vessel by centrifugal force. The fuel is subsequently cooled in a heat exchanger and recirculated for re-injection at the forward end of the reactor, while the propellant is exhausted at high velocity.

The plasma source can be fission, antimatter, or fusion.

For fission reactions, the outer annulus of the vortex is high-density liquid uranium fuel, and the low-density propellant is bubbled through to the center attaining temperatures of up to 18500 K. A BeO moderator returns many reaction neutrons to the vortex. Prompt feedback actuators maintain a critical fuel mass in spite of the turbulent flow of water or hydrogen propellant. Since the core has attained meltdown, reaction rates must be maintained by fuel density variation rather than with control rods or drums.

For antimatter reactions, swirling liquid tungsten (about 4 cm thick) is used instead of uranium, for absorbing anti-protons.

For fusion reactions, it is the propellant that is cooler and higher in density, and thus it is the reacting fuel ball that resides at the center of the vortex.

N. Diaz of INSPI, 1990.

From High Frontier by Philip Eklund
Wheel Flow

This is from Wheel-flow gaseous-core reactor concept (1965).

John Evvard figures this gas-core rocket will have (like all the others) an upper limit of about 3,000 seconds of specific impulse, exhaust velocity of about 29,400 m/s. The design is trying to increase the propellant to fuel mass flow ratio to something between 25 and 50. Since uranium has something like 238 times the molecular weight of hydrogen increasing the mass flow ratio is very hard to do.

The brute-force approach does not work. If you increase the engine pressure to 2,000 psi with a partial-pressure ratio of 80, preventing the reaction chamber from exploding will increase the reactor mass to something between 250,000 to 500,000 pounds. With that penalty weight the propellant load will have to exceed 500,000 to 1,000,000 pounds to capitalize on the increase specific impulse the engine enjoys over a conventional solid-core NTR. And even then the fuel mass flow ratio would be below 25. So this is a dead end.

So the standard solution is to somehow make an incredibly high hydrogen-uranium volume flow ratio.

There are numerous schemes to increase the volume flow.

The vortex-confined GCR makes a vortex of gaseous uranium (sort of a smoke ring) with the center hole aligned with the thrust axis. Hydrogen is injected around the outer edge of the vortex, travels radially across the furiously fissioning uranium being heated all the way, enters the hole in the center of the smoke ring, turns 90 degrees and goes rushing out of the hole and out of the exhaust nozzle.

The pious hope was that the centrifugal forces acting on the heavier uranium atoms would counteract the diffusion drag of the inwardly moving hydrogen. Sadly the drag produced by the flowing hydrogen is so great that it carries along too much of the valuable uranium.

The coaxial-flow reactor was another idea that failed even harder. The uranium gas in the center moved really slow while the hydrogen gas around the rim moved really fast. The regrettable result was the velocity difference caused shear forces which allowed the dastardly hydrogen to drag uranium along with it right out the exhaust nozzle.

John Evvard had a fresh idea: the Wheel-Flow Confined GCR.

The problem with the vortex confined GCR was that the hydrogen moves through the uranium. This allows the hydrogen to drag along some uranium. The problem with the coaxial-flow is that though the hydrogen doesn't move through the uranium, it is moving at a vastly different velocity. This causes shear forces that allow the hydrogen to drag along some uranium.

So Evvard tried to find a geometry where the hydrogen does not move through the uranium and it moves at the same velocity as the uranium.

In the Wheel-flow there is a cylinder of gaseous fissioning uranium in the center of the chamber, spinning around its long axis.

Hydrogen is injected at the outer surface of the cylinder and moves along the surface, not moving through the uranium. This avoid the vortex-confined GCR's problem. The hydrogen moves at the same velocity as the uranium gas cylinder. This avoids the coaxial-flow GCR's problem. The uranium and the hydrogen rotate as one, as if they were a solid wheel.

After one rotation of the cylinder the hydrogen is good and hot. It then exits tangentally from the chamber into an array of exhaust nozzles. And there is your thrust.

Uranium will be lost due to fission and some unavoidable diffusion into the hydrogen. Fresh uranium will be injected from the two end walls, entering the long axis of the uranium cylinder. The end wall will also rotate to match the wheel, to avoid stirring up turbulence.

The main drawback is that the boundary layer between the hydrogen and uranium is unstable. Any blob of uranium entering the hydrogen blanket will be accelerated outward by simple boyancy. This could possibly be stablized by an axial magnetic field. The fissioning uranium is more ionized than the hydrogen so the magentic field will grab the uranium more firmly.

Since the temperature inside the reaction chamber is hot enough to vaporize any material object the ions are moving like microscopic bats from hell. You'd think the high uranium molecular velocities would make the uranium cloud instantly explode to fill the chamber. Luckily the mean free path of individual atoms is a microscopic 10-7 meters or less (one micrometer, about the length of a bacteria). Since the hot uranium atoms cannot move further than the span of a typical e coli germ without crashing into other atoms their effective speed is slowed down about the same as the wheel rotational velocity.

The report is a little vague about this design. It says that if the wheel-flow engine is used in a gravitational field, the spinning cylinder of fissioning uranium might settle to the bottom of the chamber, which is bad. However, unless you were using an open-cycle gas core nuclear engine spraying radioactive death from the nozzle as an aircraft engine I don't see the application.

The idea seems to be that while some hydrogen is injected around the uranium gas cylinder for coolant, most of the propellant hydrogen goes across the top along the axial line. I guess the propellant lowers the gas pressure enough to levitate the uranium cylinder.

In the standard wheel design, the end walls will have to be cooled since they are exposed to the fury of fissioning uranium. This can be avoided by bending the uranium gas cylinder so the ends meet, converting the cylinder into a torus donut shape. Since it is now a ring there are no end walls and no need to cool them.

The problem is that the end walls were where the fresh uranium was injected, and it is unclear how to refresh the torus.

This design makes a bit more sense. It uses a torus of uranium gas. The rocket rotates around the thrust axis to make artificial gravity. This pulls the torus outward, making it expand. Meanwhile the propellant hydrogen is roaring down the thrust axis, being heated and expelled out the exhaust nozzle. This lowers axial gas pressure and pulls the torus inward, making it contract.

Between the artificial gravity and Bernoulli's principle the torus of uranium is held in place.

Of course there is still the unsolved problem of how to refresh the torus.

MHD Driven Rotation
MHD Driven Rotation
T/W ratio1:1211:1121:135
mass flow ratio
Mass flow
Radiator area
Fission Power
Mass Schedule
(metric tons)
Cavity Struct1246942
Cryogenic Magnet12898128
Turbo-electric Gen999634482
TOTAL ENGINE5,6904,0104,110

This is from Gas-Core Nuclear Rocket With Fuel Separation by MHD-Driven Rotation (from Research on Uranium Plasmas and their Technological Applications page 155) (1970)

The major problem with open-cycle gas core NTR is keeping the blasted uranium from escaping out the exhaust nozzle. Or at least only escaping after all the expensive uranium has been burnt in nuclear fission.

The researchers noted that injecting gas tangentially at high velocity (Figure 1 (a)) would confine the U235 fuel to the outer region of the cavity yet allowing the propellant to diffuse radially into the center region was the great hope. The supersonic rotation would develop high centrifugal force, forcing the heavy U235 to the outer while permitting the light hydrogen into the middle. Nope, this was tested and it don't work no-way, no-how (though they put it "this configuration does not result in effective separation of the two gases").

The researchers noted the next great hope was vortex-stabilization. The idea here was to exploit the high stability of the rotation flow instead of centrifugal force. The flow was subsonic, so the centrifugal force was negligible. A donut-shaped vortex of U235 floats in the center of the chamber, while hydrogen propellant flows around the edge of the chamber. The idea is that the hydrogen will stay in the outer regions due to the inherent stability of the rotation flow.

This didn't work either.

Looking more closely at the scheme of injecting gas tangentally, it was noticed that the centrifugal separation worked best in the part of the flow that resembled solid-body rotation. That is, rotating as if it was a solid brick of matter instead of rotating gas. So investigation focused on making the U235 and hydrogen gas rotate as if it was a solid body.

The trouble with tangential gas injection is it develops into what is called an "inviscid vortex flow field". I don't know that that is either, but apparently it makes TGI about as effective as trying to push spaghetti uphill, when it comes to making a centrifuge. So a different means of rotating the gas will be needed.

In Figure 1 (b) the entire cavity chamber will be spun mechanically to induce the centrifugal effect. Alas, in order to separate the uranium from the hydrogen you need about 1,000,000 gs of centrifugal force. No known construction material can withstand that sort of force so the chamber would explode like a bomb. To keep it from exploding you'd have to apply an external gas pressure of at least 10,000 atmospheres, which is rather excessive. What's worse is the friction loss from the external gas would become prohibitively large. Another failed concept.

The study authors had an idea. How about spinning the uranium and hydrogen using electromagnetic forces? See Figure 1 (c)

This uses the magic of Magnetohydrodynamics.

If you send electrical current J in a direction parallel to the thrust axis (z-axis), a magnetic field B will be created radially at 90° (r direction). This is called J×B. The important point is the magnetic field will be pushing the gas towards the chamber walls, allowing the gases to rotate as if it was a solid body. Rotating such that the blasted U235 separates from the propellant, and is kept away from the exhaust nozzle.

No chambers rotating so fast they explode, no friction loss from external gas at 10,000 atm, this design looks like it would actually work.

The report presents a prototype engine.

The cavity is typically seven to eight meters in both diameter and length. The radial mangetic field is produced by cryogenically cooled magnetic coils, and has a strength of about 1.0 tesla near the cavity walls. The coils draw a negligible amount of power. The total electrical current is about 20,000 amps, flowing through 200 pairs of segmented electrodes. The chamber pressure is 20 to 60 atmospheres. The maximum tangential rotational velocity is about 1.6 to 1.8 kiometers per second.

The hydrogen propellant is introduced along the centerline of the cavity and flows through the center at low velocity. The propellant is heated mainly by radiation, plus a bit by conduction on the part of any propelant that brushes the hot uranium. Hydrogen propellant is regrettably mostly transparent to infrared heat rays, which is why most gas core designs seed the hydrogen with microscopic tungsten particles or something. In this design they seem to be relying upon uranium atoms for seeding, since some will diffuse into the propellant (about 0.00015 mole fraction).

The chamber temperature varies from 10,000K at the chamber walls (in the hot uranium) to 6,000K at the centerline (in the propellant). This averages out to a specific impulse of approximately 1,770 seconds, which is actually pretty good. The thrust is about 40 metric tons or 392 kiloNewtons.

The liquid hydrogen first flows through the superconducting magnetic coils to help keep them at cryogenic temperatures. It then passes through a pump to pressurize it to about 640 atmospheres. Now the propellant travels through the moderator-reflector wall heat exchanger, simultaneously cooling it off and getting real hot. The hot hydrogen enters the first-stage turboelectric generator, which supplies half the required MHD power. The hydrogen leaves the turbogenerator at about 107 atmospheres of pressure. It then passes through the moderator-reflector heat exchanger a second time, and enters the second-stage turboelectric generator. This produces the other half of the required MHD power. Finally the propellant enters the chamber to be superheated and exhausted to create thrust.

The propellant only removes about 3% of the waste heat in the moderator-reflector. The other 97% of the waste heat is removed by a liquid-metal cooling system which expells the heat out of a heat radiator with a surface temperature of 1,200K. A fraction of this is diverted to an auxiliary power generator for other power needs.

Fissioning Plasma Core


     This study investigated the development of a system concept for space power generation and nuclear electric propulsion based on a fissioning plasma core reactor (FPCR) with magnetohydrodynamic (MHD) power conversion system, coupled to a magnetoplasmadynamic (MPD) thruster. The FPCR is a liquid-vapor core reactor concept operating with metallic uranium or uranium tetrafluoride (UF4) vapor as the fissioning fuel and alkali metals or their fluorides as working fluid in a closed Rankine cycle with MHD energy conversion. Candidate working fluids include K, Li, Na, KF, LiF, NaF, etc. The system features core outlet temperatures of 3000 to 4000 K at pressures of about 1 to 10 MPa, MHD temperatures of 2000 to 3000 K, and radiator temperatures of 1200 to 2000 K. This combination of parameters offers the potential for low total system specific mass in the range of 0.4 to 0.6 kg/kWe. The MHD output could be coupled with minimal power conditioning to the variable specific impulse magnetoplasma rocket (VASIMR), MPD thrusters or other types of thruster for producing thrust at very high specific impulse (Isp= 1500 to 10,000 s).


     Solid core reactors provide the path for minimum risk for generating nuclear space power in the coming decade. These reactors can be expected to achieve evolutionary improvements in their performance based on modest extrapolations of current fuel technology. In contrast, liquid, gas, and vapor core reactors offer a path for extraordinary improvements in performance, and have the highest potential for reducing overall system specific mass. Some key features of gas core reactors include:

  • Increased launch safety and flexibility and decreased cost through in space assembly, fueling, and refueling
  • Power is generated at high temperatures and heat is rejected at high temperatures-providing flexibility in minimizing the radiator size and weight
  • Power scaling and choice of conversion technique are practically unlimited-power level and quality can be accommodated by adjusting fuel circulation rate and the average fuel exit temperature
  • For surface power generation, a completely passive, gravity driven, ultrahigh bumup and very long life system is achievable
  • Very low fuel development cost, testing, and facility requirements

     Liquid and gaseous core reactors impose minimum geometrical constraints on the fuel configuration while providing a high temperature heat source plus a powerful ionization source. These features lead to the inherent technical advantages of energy conversion temperatures that are not limited by fuel integrity limits and to enhancement of working fluid conductivity by the ionization processes. High conductivity of the fissioning plasma/working fluid permits the use of magnetohydrodynamic (MHD) energy conversion in these systems.

     This study examined the nuclear electric propulsion (NEP) options for a fissioning plasma core reactor (FPCR) concept. This reactor model was based on earlier, detailed studies of gaseous and liquid-vapor core reactor concepts with MHD. Also examined was the specific mass performance of this system to illustrate the potential cost savings of multimegawatt class FPCR-MHD-NEP systems.


     The baseline FPCR design selected for this study is a UF4 fueled, fissioning plasma core reactor with MHD power conversion. Previous analysis indicates that in a typical FPCR-MHD system, 200 MWe power could be supplied by a MHD generator that is fed by more than 1100 MW thermal power in a 3 m3 reactor core with a 50 cm Be0 reflector region. The UF4/KF working fluid mixture enters the reflector as liquid at an inlet temperature of 1800 K at a pressure of 5 MPa. The mixture boils in the reflector and enters the core at 1900 K. A separate loop of KF cools the reflector and then enters the core. A slipstream of this loop is used to cool the cavity wall before entering the core. The reactor outlet temperature is 4000 K. The reactor outlet stream is directed through a nozzle and into the MHD channel as pictured in Figure 1. The MHD generator operates at an inlet Mach number of 3, a Hall parameter of 2.7, and a plasma conductivity of 60 mho/m. The magnetic field strength is 4 T. The exhaust of the channel is directed to a diffuser and into the main heat exchanger. In the heat exchanger, the KF component condenses first and is directed to pumps. The UF4 is further cooled in a second heat exchanger and then routed to the pumps.

     A system diagram is illustrated in Figure 2 for a 3 m3 gas FPCR with MHD energy conversion with net cycle efficiency of 22%. Other system parameters include a neutron flux level of 1016 n/cm2sec, nuclear enhanced electric conductivity of 60 mho/m, and a magnetic field of 4 Tesla. These technical features lead to design advantages that include high working fluid and heat rejection temperatures, high fuel utilization (burnup up to 200,000 MWd/MT), elimination of fuel fabrication, testing, and verification, simplified fuel management, inherent hot spot compensation, and flat power density profiles. Operationally, these design advantages lead to mission benefits and improved safety. Benefits include shorter trip times and less initial mass in low earth orbit (IMLEO), reduced radiator mass, the potential for reactor refueling and reuse due to the fluid nature of the fuel, and a zero probability of accidental criticality on launch since the reactor is launched unfueled.


     The mass of six different reactodgenerator designs ranging in power from 100 to 210 MWe with core pressures of 3 and 4 MPa and core volumes between 2 to 3 m3 were estimated in order to determine how system mass scales with generator power level and to arrive at an optimum design for further study as a baseline reference case. The method used to estimate specific mass was to calculate the mass of each component for each specific case (unique power level). Where applicable, three estimates of component mass were calculated for different levels of technology available for application at present or in the near-term. The first and largest mass estimate in each case is for current proven or established technology available to serve the function of the specified component. The second estimate is for present-day, advanced technology (state-of-the-art), while a third estimate, if listed, is calculated based on the development and utilization of emerging technologies that can be expected to be available in time for application in the FPCR or similar propulsion system. The results of these calculations are shown in Table 1 and Figure 3

     Mass estimates for the reactor include both the core and reflector regions. The core was modeled as a uniform cylinder with radius and height based on criticality and power requirements for core volume. Be0 reflector/moderator regions were modeled as an annulus with thickness of 50 cm and a top and bottom disk 25 cm thick. MHD generator mass was estimated based on the equation for a line MHD generator specific mass from Rosa (1968). The pressure vessel necessary to contain the reactor system was modeled as two separate pressure vessels, an inner (thickness 3 cm) and an outer (thickness 2 cm). A differential pressure is maintained between the two vessels with 3 to 4 MPa inside the inner vessel, 1.5 to 2 MPa inside the outer vessel. To account for piping and other structures, 15% was added to the estimated weight of the vessel. Several different alloys at different stages of development were identified for application as a reactor pressure vessel including stainless steel, TZM (Mo99/Ti0,9/Zr0.1), Mo-Ti, and Ti-Al alloys. The area of the radiators was calculated from heat exchanger rejection rate based on the reactor power and the assumed efficiency for the system. For established, off-the-shelf technology an estimate of 8.2 kg/m2 was obtained from Cropp et al. (1990) for high temperature radiators. For advanced, state-of-the-art technology using SiC reinforced Ti with carbon/graphite composite fins, an estimate of 5.5 kg/m2 was used based on the work of Begg et al. (1989).

Shielding Calculations and Mass Estimates

     Estimates of shield mass ranged from 12 to 15.5 MT. These were based on a study of shielding design to determine the amount of shielding necessary to reduce neutron and gamma ray dose rates to levels below that of background radiation, for which the payload/crew must be shielded for separately. A variety of shielding designs were analyzed to determine the optimum design for reducing the dose rate to acceptable levels and also to minimizing the size and hence the mass of such a shielding design.

     The general Monte Carlo transport code, MCNP-4C, was used to calculate the dose rate at various distances from the power generating system. The geometry model for these calculations is illustrated below in Figure 4. The core region is modeled as a cylinder with L/D=1 and volume equivalent to that of the core for each case studied. A 50 cm reflector/moderator region of BeO is modeled surrounding the core with another 25 cm above and below the core region. The shield region adjacent to the reactor consists of two parts, a flat circular region that covers the entire reactor bottom including the reflector region and a hemispherical part with radius equal to the core radius abutted to the first part. The first shield is laminated, with a 5 cm layer thick of concentric hafnium and copper rings followed by an 8 cm thick layer of lead. Because of its high neutron absorption, the hafnium layer spans the core region, while lower density copper covers the reflector/moderator region.

     Liquid hydrogen tanks for propulsion and refrigeration were modeled between the reactor/power generation complex and the payload/habitable regions and lying along the boom connecting the two modules. Although the liquid hydrogen is not very dense (~0.1 g/cc) there is a considerable amount present (50 ton is commonly referenced). The payload/habitable region is modeled beyond the liquid hydrogen tanks with a diameter of four meters. It is covered with a 5cm layer of boron carbide (B4C) for capturing any remaining thermal neutrons that reach this end.

Both fission neutrons (prompt and delayed) and gamma rays (prompt and decay) from fission were modeled. The neutron sampling distribution was taken as a Watt fission spectrum. The distribution of gamma rays from fission was taken from Lamarsh (1983). The shield mass was calculated in each specific case (power level) based on the dimensions of the reactor modeled (see Table 1).


     An ultra-high power nuclear electric propulsion system requires simultaneous optimization of the thruster and power generation system. The MHD output power should be conditioned to directly feed the thruster current and voltage needs. Both MHD channels and magnetoplasmadynamic (MPD) thrusters tend to produce and consume high currents at relatively low voltages. This could potentially eliminate or at least, dramatically reduce the power management and conditioning needs in these systems. The high breakdown voltages that are needed to establish the arcs in MPD thrusters could be generated with a compact power storage system. The direct coupling of MHD output power to generate plasmoid pulses in MPD could produce a few grams to multiple kilograms of thrust at very high specific impulse (Isp = 1500 to 10,000 s). In essence, the coupled MHD-MPD propulsion system operates like a transformer converting the energy of very high mass flow rate and relatively low velocity of the FPCR plasma to very low flow rate and extremely high velocity of the thruster plasma. The order of magnitude reduction in specific mass and very high specific impulse of FPCR-MHD-MPD system could lead to a dramatic reduction in the cost of interplanetary as well as interstellar missions.

Variable Specific Impulse Magnetoplasma Rocket (VASIMR)

     The flexibility to vary power and energy characteristics inherent in a space nuclear reactor make it ideal for providing power for vehicle systems operation and propulsion as well as mission surface power. Traditionally, this advantage of nuclear power was hampered by the inflexibility and fixed performance of electric propulsion systems. However, the equally variable performance characteristics of VASIMR make it a natural match for the FPCR-MHD power system. The analogy might be made to the advantage of a multi-gear bicycle over a fixed single-gear bicycle in that it can use a low gear when climbing a hill (escaping from planetary gravitational bodies with high thrust, low Isp) and a more efficient, higher gear on a hi-way (fuel savings during the acceleration phases of a mission with low thrust, high Isp).

     In addition to offering variable thrust operation allowing for optimum performance of an NEP system based on the specific requirements and constraints of a mission, VASIMR also offers other features. The use of magnetic nozzles instead of mechanical nozzles and electrodes reduces the erosion and other life limiting factors for the thruster. Due to the designed "leaky" nature of the applied magnetic field, the plasma system is not as much susceptible to instability as other magnetically confined plasma system.


     Gaseous and liquid-vapor core reactors can potentially provide the highest reactor and cycle temperature among all existing or proposed fission reactor designs. This unique feature makes this reactor concept a very natural and attractive candidate for very high power (10 to 1000 MWe) and low specific mass (0.4 to 5 kg/kWe) nuclear electric propulsion applications.

     The exceptional specific mass performance of an optimized FPCR-MHD-MPD/VASIMR system could lead to a dramatic reduction in the cost and duration of manned or robotic interplanetary as well as interstellar missions. The FPCR-MHD-MPD/VASIMR system could enable very efficient Mars cargo transfers or short (<8 month) Mars round trips with less IMLEO. The system could also enable highly efficient lunar cargo transfer and rapid missions to other destinations throughout the solar system.

RD-600: Soviet bimodal GCNTR
RD-600 bimodal GCNTR
Thrust5,880,000 N
Specific Impulse2,000 sec
Chamber Pressure500 kg/cm2

This is astonishing. A rocket expert living in Russia, Denis Danilov, stumbled over this in their research. This is a proposal for a Soviet gas core nuclear thermal rocket project. That ran from 1963 to 1973. And was bimodal!

The fact that the Soviet Union actually had a real live gas core project that ran for ten years and employed 90 researchers got my attention.

The fact it was a gas-core BIMODAL engine made my jaw drop. I have never ever seen a proposal for a bimodal nuclear engine that used anything other than a solid-core nuclear engine. According to this document (page is in Russian, use Google Translate if need be), in power generation mode it uses a separate circuit for the uranium which has no connection with the outside world. Yes, this adds more points of failure, but on the plus side it allows one to generate power without spraying fissioning uranium out the exhaust like it does in thrust mode.

While in thrust mode there is a second MHD power generator around the exhaust nozzle, to harvest a bit of thrust to make electricity. In coast mode the uranium is diverted from the rocket engine altogether, entering a closed cycle which energizes the main MHD generator.

I did spot a single sentence in one of the documents that seems to imply an important advantage to gas core MHD power generation.

In the following documents:

  • Type A NTR: NERVA style solid core nuclear thermal rocket
  • Type V NTR: open-cycle gas core nuclear thermal rocket
  • tf: Tonnes-Force. 1 tf = 9806.65 Newtons

This is from a thread in the Kerbal Space Program forum by DDE:

Dear Mr. Chung,

I have been a user of your website for quite a while. However, I’ve stumbled onto a virtually unknown piece of Red Atomic Rockets history that I’d like to share with you. I’ll stick mostly to direct translation of sources to avoid putting my spin onto it.

So, there I was trawling through the full list of Energomash rocket engines at, trying to make sense of their classification scheme. I knew RD-1xx were kerolox, RD-2xx were hypergolic, RD-3xx involved fluorine and had to be given a wide berth, RD-4xx were early, solid-core NTRs, RD-5xx used peroxide, and RD-7xx switched from kerolox to hydrolox on the fly to maximize total Δv. I was trying to find out what the heck an RD-6xx was, thinking it may have been the bimodal NTR branch.

There was only the RD-600. What has a vacuum thrust of 600 tonne-forces (5,880,000 N), an Isp of 2000 sec and chamber pressure of 500 kg/cm2?

I was hooked already, and the Russian Wikipedia, messy as it is, answered me with an article about the solid-core twisted-ribbon RD-0140 ( that contained a timeline apparently copied from, which is a webpage of the Department of physical mechanics of the Moscow physico-technical university. Translation follows:


To produce a gas-core nuclear rocket motor with high specific impulse (>3000 sec) for missions to planets of the Solar system.


     1957 – Initiation of research as suggested by V.M. Ievlev and approved by I.V. Kurchatov, M.V.Keldysh and S.P. Korolev (by then known as the Three Ks as a result of their ICBM work – ed.)
     1953 – Government decree on research into “cruise missiles propelled by ramjets exploiting nuclear energy”
     1955 – formation of research group at Air Industry Ministry NII-1 (currently Keldysh Research Centre – ed.). Led by V.M.Ievlev (K.I.Artomonoc, A.S.Koroteev et al). Objective: development of NTRs “type A” (Isp=850-900 sec) and “type V” (up to 2000 sec).
     1956 – Government decree on “creation of a long-range ballistic missile with an atomic engine”. Chief designers: overall – S.P.Korolev, engine – V.P.Glushko, reactor – A.I.Leypunskyi; personnel recruitment and training at the Moscow Aviation Institute – N.N.Ponomarev-Stepnoi.
     1958 – Government decree on initiating NTR research; overall command relegated to M.V.Keldysh, I.V.Kurchatov and S.P.Korolev
     1958 – construction of reactor test stand and “hot lab” commences at Ministry of Defence Proving Grounds №2 (Semipalatinsk nuclear testing site)
     1964 – combined Central Committee of the Communist Party of the Soviet Union and Government of USSR decree on initiating construction of the Baikal firing complex at the Semipalatinsk NTR test site.
     1966 – creation of 11B91, an A-type NTR (non-GRAU designation RD-0140 – ed.). Scientific supervision – Keldysh Centre (V.M.Ievlev), manufacturing – Chemical Automatics Design Bureau (A.B.Konopatov), fuel elements – Perm research and technological institute (I.I.Fedik)
     1968 – development of RD-600 GCNR; scientific supervision by the Keldysh Centre, lead design by NPO Energomash under V.P.Glushkon; thrust 6 MN, Isp 2000 sec
     1968 – Government decree on development of RD-600 GCNR and construction of the Baikal-2 test stand
     1970 – NPO Energomash and the Keldysh Centre complete a draft proposal for a 3.3 GWt gas-core powerplant, EU-610
     1972 – criticality of an IVG high-temperature research reactor at Baikal (N.N.Ponomarev-Stepnoy)
     1978 – criticality of the first 11B91 NTR, Image caption: high-temperature GCNR variant (“Type V”)


     1955 – commencement of work on a Type A NTR (SCNR) at Los-Alamos under the Rover program
     1960 – conceptual development of a Type V NTR (GCNR) by Weinstein, Kerrebrock (MIT) and Los-Alamos; Isp=(600-2000) sec
     1963 – development of Nuclear Engine for Rocket Vehicle Applications (NERVA) by Westinghouse and Los-Alamos
     1962-1968 – experiments in hydrodynamics, plasma stability, heat physics and radiation of uranium plasma, optical qualities of hydrogen, neutron calculations of reactor criticality
     1973 – cessation of TR research



     1985 – Los-Alamos and NASA conduct systemic analysis of Lunar missions, concluding that resumption of research on Type V systems is crucial (twofold cost and flight duration reduction). Equipment and systems are preserved in Los-Alamos and Nevada (Keldysh Centre and Semplatinsk).
     1989 – President Bush announces the Space Exploration Initiative – a manned mission to Mars by 2018 (see Russian space program). NTRs assumed as baseline by NASA and Los-Alamos. DOE/NASA task force on NTRs formed.
     1991 – GCNR conference in Los-Alamos
     1992 – research into stability, neutrons, displacement, quantitative modelling, MHD (presumably, magnetohydrodynamic generators or magnetohydrodynamics in general – ed.)
     2005 – China and Kazakhstan declare research into spaceborne nuclear power a priority

The GCNR program in the US has been unsuccessful due to “lack of experimental data on thermophysical qualities of substances and the calculating power for modelling high-temperature hydrodynamics and turbulence (from MIT report by R.Patrick & Kerrebrock). The USSR has resolved these issues with participation of the Department of physical mechanics

Both the US and USSR would relegate large rocket projects to a triad of R&D Centre-University-Test Ground, e.g. Los-Alamos-MIT-Nevada and Keldysh Centre-Moscow Institute of Physics and Technology-Semipalatinsk

Key development aspects:

  • Handling and operation of a gaseous fuel element
  • Thermophysics of nuclear fuel and reaction mass
  • Vortex and magnetic hydrodynamics
  • Radiative and convective heat and mass exchange
  • Thermal protection of reaction chamber walls and the egress canal
  • Achieving GCNR criticality
  • Achieving stable GCNR operation despite high mobility of fission fuel

GCNR parameters:

  • Pressure – 1000 atm
  • Temperature: fuel 40-60 thousand K, reaction mass 8-10 thousand K
  • Molten uranium at 1500-2300 K
  • High-pressure hydrogen at up to 2800 K
  • Chemically aggressive environment due to alkali metals at up to 2800 K
A.S.Koroteev, E.E.Son. Development Nuclear Gas Core Reactor in Russia: AIAA-2007-0035 (behind paywall), 45th AIAA Aerospace Sciences Meeting and Exhibit, 2007, Reno, Nevada.

By then I was thoroughly intrigued and began to go up and down Yandex search results while trying to filter out the mentions of the RD-600 turboshaft engine. I’ve stumbled upon the questionably legal (as is most of Russia’s internet) online version of a collection of Glushko’s works, published by Energomash in 2008 with a total run of 250 books, that contains a few interesting official documents of his own authorship.

August 18 1963

to the research and development council of the USSR State commission on defence technology
on prospective R&D at OKB-456

The research and development conducted at OKB-456 have led to the following conclusions:

1. Further development of oxygen and RFNA engines at OKB-456 is inexpedient, as most of the expected pathways of rocket development are better served by other engine types. This does not mean that oxygen and RFNA motors are undeserving of further development, as occasionally they are highly fit for purpose. For example, when performance takes a back seat to propellant liquidity range, RFNA engines are quite applicable. Liquid oxygen, non-toxic and cheap, is also quite successful, despite difficulties associated with its low boiling temperature…

(sections 2, 5, 10, 11, 15 missing – ed.)

3. At the present stage the development of high-powered liquid propellant rocket motors for surface-to-surface missiles and space boosters at OKB-456 relies entirely on UDMH and N2O4.

The use of storeable, hypergolic fuel components already mastered by the chemical industry has ensured the development of highly effective motors for R-36, 67S4, R-56 and the first stage of UR-500 (a.k.a. original variant of Proton – ed.) in accordance with the Decrees of CC of CPSU and Government of USSR.

These engines range, in thrust from 12 (vacuum) to 600 (sea level) tf, and in specific impulse at sea level from 272 to 300 sec, in vacuum up to 325 sec.

4. OKB-456’s current plan outlines the development of a gas-core nuclear motor with liquid hydrogen as reaction mass, with thrust of 200-600 tf (RD-600) and specific impulse of 2000 seconds.

The creation of such an engine would be a true revolution in rocket science due to the dramatic leap in specific impulse.

Further development of this engine class can allow, with time, specific impulses in the 2500 sec range, allowing creation of booster rockets with an order of magnitude greater payload than regular chemical-powered ones.

6. For boosters intended for first or second space velocities (orbital and planetary escape trajectories respectively – ed.) the optimal design is as follows: the chemical motors on the first stage loft the second stage carrying the GCNR to the minimum safe altitude dictated by the contamination of the exhaust by fission products…

7. The low-altitude loft stage in the abovementioned GCNR-based system should rely on the use of chemical rockets using high-density storeable propellants (UDMH and N2O4), as under such conditions the fuel is more energetically efficient than kerolox and, due to its chemical stability and hypergolicity, more convenient. An UDMH-N2O4 first stage can achieve Isp of 300-320 sec and a mass ratio of 1.18 (e.g. 8D420 engines).

8. Use of a GCNR makes the development of an entirely reusable booster more realistic by allowing the use of an airbreathing first stage…

9. In case of two-way missions to the planets and their moons using a GCNR the payload mass can be further increased by using a third stage with refireable electric rockets with Isp in the 10000-20000 sec range. (here and henceforth bold added as emphasis – ed.)

12. (sales pitch for high-energy, storeable lander propellants, including UDMH-N2O4, late RD-5xx series motors burning H2O2 and pentaborane, and the RD-550 using H2O2 and beryllium hydride (!); sly suggestion to cease development of all cryogenic chemical rockets – ed.)

13. The preceding deliberations on GCNR application can only be realized once there is confidence of the possibility of creation of said motor, backed by experimental research, including stand tests of a single fuel element motor with a gas core reactor, in a state approaching operational (plasma temperatures up to 30000 K, pressure up to 500 atm)…

14. The prospective development plan of OKB-456 has been developing for the last few years along the lines laid out in sections 1-13 and currently ranges out to 1970. In accordance with it:

d) Key developments are

I) High-powered UDMH-N2O4 motors: for first stages of R-36 and 67S4 (8D723, 8D724), for first stages of UR-500 and R-56 (PD43), for second and third stages of R-56 (11D44, 8D724) and a high-performance motor with ASL thrust of 600 tf and specific impulse of 300 sec ASL and 323 sec in vacuum for the first stage of heavy boosters (8D420);

II) Upper stage motors of 10-12 tf using: UDMH-N2O4 (8D725, Isp=325 sec), H2O2-pentaborane (11D11, Isp=375 sec), H2O2-beryllium hydride (RD-550, Isp=400-460 sec, under investigation);

III) Gas-core nuclear rocket with liquid hydrogen reaction mass with thrust 200-600 tf (RD-600, Isp=2000 sec) as second-stage engine…

16. (section on weaker chamber pressure in US chemical rockets leading to poorer performance at comparable mass – ed.)

OKK-456 chief designer, academician GLUSHKO

Archive 1727 (123-130)

Backtracking a bit, I’ve found an earlier memo containing some of the points missing from the above report

May 6 1963


9. The above deliberations on expedient applications of liquid-fuel rocketry using storeable and cryogenic propellants, solid-core and gas-core NTRs and electric rockets can only be realized once the creation of GCNR is proven feasible in the coming years, backed by experimental research, including stand tests of a single fuel element motor with a gas core reactor, in a state approaching operational (plasma temperatures up to 30000 K, pressure up to 500 atm). Creation of the testbed facility and an experimental single fuel element motor is planned for late 1965, with tests commencing in 1966. First experimental results are likely to be produced by 1967. Should development outcomes prove favourable, a flight-ready GCNR (RD-600) can be deployed by 1970.

OKK-456 chief designer, academician GLUSHKO

Archive 1727 (66-71)

Here’s the economic aspect:

April 29 1969

comrade ABRAMOV I.I.

Re: development of Type V NTR

In accordance with the Decree of CC of CPSU and Government of USSR 524-215 of June 19 1964 the Ministry has initiated the program “Development of RD-600 nuclear rocket motor” at an estimated cost of 20 million roubles.

Over 1960-1968 Energomash and related organizations have conducted a range of design, theoretical, production and experimental studies the results of which are found in the following reports:

  1. Draft project of RD-600 GNCR (original sent to MGM on September 30 1964)
  2. Report on preliminary project of a testbed loop-type motor with a single gas-core fuel element (original sent to MGM on April 9 1965)
  3. Report on progress on RD-600 NTR for 1966 (original sent to MGM on December 31 1966)
  4. Report on progress on RD-600 NTR for 1967 (original sent to MGM on January 4 1968)
  5. RD-600 nuclear rocket motor. Abbreviated results of theoretical and research studies (original sent to MGM on December 18 1968); said report is equivalent in content to a complete pre-draft project; it has been approved in that capacity by Central Research Institute of Machine Building in correspondence of April 15 1969.

Over the process of developing the RD-600 a phased approach to development of Type V nuclear rocket motors has been determined to be expedient, including the initial development of a testbed motor with a single gas-core fuel element, alongside the requisite testing facilities.

The above-mentioned phased approach is approved by the Decree of CC of CPSU and Government of USSR №388-146 of May 24 1968. In particular, the decree outlines the development of a high-performance on-board powerplant based around the single gas-core fuel element design. The technical objective set by the Central construction bureau of experimental machinebuilding (currently RKK Energiya, earlier KB-1, the late Sergei Korolev’s outfit – ed.) calls for a powerplant developing 3 mln kWt (sic), which is achievable by using a reactor of a Type V NTR with a thrust of approximately 50 tf.

Due to the above I request your permission to close the program titled Development of RD-600 nuclear rocket motor”, to disburse the actual expenses of 9325 thousand roubles, and to commence a program titled “Development of an experimental testbed reactor with a single gas-core fuel element and a draft project of a high-performance powerplant” in accordance with the draft project file attached to my letter of April 1 1969, with an estimated cost of 2416 million (sic) roubles.

Chief designer GLUSHKO

Archive 82/125 (36-37)

One of the last documents in the collection is an overview of Glushko’s entire NTR business at the height of its glory. Believe me, most of the more informative materials on Soviet rocketry, such as Gubanov’s memoirs about Energiya-Buran, are this dry and technical.

July 26 1973

Outline of development of nuclear rocket motors at KB Energomash

Increase of the specific impulse, being one of the cardinal directions of rocket motor design, has driven the effort to exploit the energy of nuclear fission in this rapidly developing field of technologies. The successes of national rocket design in 1950-1955 have allowed the Physico-energetic institute of the Ministry of medium machinebuilding (cover name for Soviet nuclear armament and energy agency; similar to the Ministry of general machinebuilding seen earlier – ed.) to put forward the concept of integrating a solid-core nuclear reactor into a rocket engine. Based on PEI’s proposal, relying on a uranium-graphite reactor heating up hydrogen, KB Energomash began ongoing research work on solid-core nuclear rockets (Type A) in 1956. Research in cooperation with PEI over 1956-1958 revolved around neutron flux and thermal calculations and covered a variety of thrust levels (tens to hundreds of tf), reaction masses and reactor types (by moderator material, fuel distribution, et cetera). In 1958 Energomash formed a permanent design taskforce focused on nuclear motors, evolving in 1961 into a full-fledged NTR design section. From the start the unit was led by R.S.Glinik; some of the first members ncluded E.M.Matveev, G.L.Lioznov, V.Ya.Sirotkin, K.K.Nekrasov and V.N.Petrov. Based on research by PEI and Energomash along with parallel investigation of rocket propulsion through fission energy by the Keldysh Centre, in 1958 the CC CPSU and Government of the USSR issued a joint decree that, among others, outlined he development of a draft design of a high-thrust Type A NTR using ammonia as reaction mass. The draft design was completed in 1959 with assistance from PEI (neutron flux) and the Keldysh Centre (thermal physics, fuel rod design, engine dynamics research), yielding two variants, RD-401 with a water moderator and RD-402 with a beryllium moderator. RD-402 produced superior performance, at 168 tonne-force vacuum thrust, 428 sec Isp, weight to power ratio 22 kg/tf of thrust (note that the original refers to it as specific power – ed.) and a length of 6760 mm.

The draft proposal presented the following innovations: proof of applicability of heterogenous multi-fuel rod reactors for thrust up to 500 tf; application of a solid beryllium moderator; fuel rod design providing high reaction mass temperature (up to 3000 K) with minimal fluctuations along the cross-section of the fuel element; homogenized reactor control system using absorber gas in isolated canals; closed-cycle turbopump feed, with the working fluid heated by dedicated fuel rods; steering via gimballing the engine; multi-nozzle design dramatically shrinking the length of the motor; mounting of the booster pump in the outboard section of the rocket’s tankage; proof of the necessity of cooling most of the components due to atomic radiation. Although overall the energetics of the RD-402 were relatively unimpressive, the design and development allowed many of the complex issues in construction of Type A NTRs to be exposed and solved.

By 1960 positive outcomes of use of liquid hydrogen in rocket design and the availability of requisite tankage technology have allowed to push forward with the draft design of a Type A NTR using hydrogen reaction mass; the CC CPSU and Government of the USSR authorized it in 1960. As a result by 1962 Energomash in cooperation with a range of other units, under general leadership of the Keldysh Centre and with reactor design input by PEI, developed the draft design of the RD-404.

RD-404 was rated for 200 tf vacuum thrust, 950 sec specific impulse and weight-to-power ratio of 45 kg/tf of thrust including radiation shielding for the tankage, and was 7770 mm long.

RD-404’s design included a number of design solutions developed specifically for Type A NTRs. After researching a range of moderator options, the final design used an optimized arrangement of beryllium into autonomous cells mated with the fuel rods. A modified sectioned design of the fuel rod was substantiated and implemented, with the initial graphite-coated zone and a post-heat section coated in a metal-carbide composite leading to average hydrogen temperatures in the 3000 K range. A liquid control “rod” system using mercury was conceived and researched. An engine control system relying on specialized pyroautomatics was also developed.

A complex study was conducted in order to optimize weight-to-power ratio while also developing rocket-engine system design. Research into start cycle and throttling was performed, along with the terminal stage and shutdown cycle using the remnant heat of a subcritical reactor; reverse systems were developed to minimize post-shutdown thrust.

Engine subsystem design, tankage protection and engine-to-rocket coupling were investigated with regards to reactor radiation effects.

A steering system using vanes on nozzle edges was also tested and implemented. Initiation calculations very verified by a set of experiments on fuel rod material resistivity, component durability, physical assembly tests, gas-dynamic tests of the multi-nozzle design, tests of the mercury control system, et cetera.

A considerable contribution to Type A NTR design was on part of a Keldysh Centre taskforce led by V.M.Ievlev and including K.I.Artamonov, R.B.Akopov, V.N.Bogin, A.I.Gori, V.A.Zaitsev, G.V.Konyukhov, E.P.Terekhov.

The development of RD-401, -402 and -404 essentially led to establishing the key principles of Type A NTR and subsystem design, along with many aspects of manufacturing, testing and operation. Over the course of these projects the organizations involved brought up new creative collectives while laying the groundwork for extensive industrial cooperation. (standard canned Communist phrases, if you can’t tell. – ed.)

The resultant draft projects of RD-401, -402 and -404 were investigated by a number of highly competent expert panels and technical councils, and were rated highly by them.

Based on existing experience, in 1962-1963 a draft design was carried out for a mid-range NTR referred to as RD-405, with 30-40 tf thrust, liquid hydrogen reaction mass, specific impulse of 900-950 sec, weight-to-power ratio 55 kg/tonne-force of thrust including tankage shielding. The reactor used a zirconium hydride moderator, beryllium reflectors and fuel rods similar to those of RD-404.

In actual operation requiring multiple firings the average specific impulse of a SCNR will be degraded by several tens of seconds due to lengthy reheating from standby state.

Limited possibility of any further increase of specific impulse in Type A NTRs has been the key reason for cessation of research at Energomash in 1963 and relegation of this development topic to a different OKB (see: RD-0140 – ed.)

It was decided that Energomash would instead focus itself on the much more promising gas-core NTR (Type V), which could lead to a revolutionary leap in aerospace design.

The research at the Keldysh Centre in 1958-1963 under the overall supervision of V.M.Ievlev, covering principal schemes of gas-core reactors, gas-core fuel elements and Type V NTRs overall, had substantiated the plausibility of a highly energetic engine design, which justified initiating of design development at the construction bureau. A significantly greater specific impulse of Type V designs compared to Type A, making possible the design of spacecraft with qualitatively new capabilities, along with a number of operational advantages (absence of fissionables in the engine during manufacture, possibility of fissionable removal after tests, thus making in-situ servicing and refueling more plausible) make Type V systems exceptionally promising, despite the obvious challenges and the considerable development costs.

From the very beginning of work on Type V NTR at Energomash in accordance with the Decree of CC CPSU and the Government of the USSR under general scientific supervision of the Keldysh Centre had two key directions: development of an actual high-thrust rocket motor, and the development of a testbed motor used to test off the key functioning principles of the full-scale engine, with a range of theoretical and practical problems being combated along the way and a dedicated test facility being constructed. It should be noted that the development of Type V engines benefitted immensely from the expertise accumulated when developing Type A NTRs.

The RD-600 was designed in 1964-1968, rated for 600 tf of thrust at 2000 sec specific impulse, with weight-to-power ratio of ~100 kg/tonne of thrust including shielding, and a total length of 14000 mm. The reaction mass is liquid hydrogen doped by lithium. The engine includes a multi fuel element gas-core reactor with a solid moderator and reflector (beryllium, beryllium oxide, graphite), gaseous fuel elements with a central moving stream of fission fuel, and a closed circuit for nuclear fuel, complete with a condenser, separators, pump and fission products removal system. Stabilization of flow in gaseous fuel elements is performed by magnetic solenoids powered by a unipolar electric generator. Research conducted by the Keldysh Centre, PEI and a number of other institutes covered a wide array of topics around designing the principle layout of the motor and optimization of key aggregates and systems in an effort to maximize the specific impulse. Research included neutron flux measurements on physical assemblies, modelling of parallel flows, studying the effects of longitudinal magnetic fields on flow of conductive media, investigating radiative heat transfer and thermal protection of the structure from high-intensity heat, studying resilience of various construction materials when immersed in liquid uranium, seeking production and testing methods of temperature-resistant porous materials, et cetera.

The executed research proved the principal possibility of producing an NTR with uniquid qualities through use of a gas core reactor, while revealing the extraordinary challenges of organizing the operation, requiring new and specialized materials and manufacturing technologies, associated with such an engine. It was proved that a range of experiments simulating the key operational processes would be crucial for the development of a GCNR, but would require specialized testbeds and a reactor stand facility.

Creation of such a facility and a testbed motor was outlined by a Decree of CC CPSU and the Government of the USSR of 1968, and had taken centre stage in the work of Energomash and its associates since 1964; it is even more important now, as the theoretical solutions in gas-core reactor design take a backseat to the need for practical testing.

In accordance with the aforementioned decree in 1970 another design study was completed, this time of a spaceborne electric powerplant dubbed EU-610, with an electric output of circa 3.3x106 kWt, specific power 0.7x105 kWt/kg/sec (sic), relative mass 18.7 g/kWt, length 10000 mm. The powerplant is based around Keldysh Centre’s proposed improvements to the gaseous stream fuel element design. A significant increase in magnetic field intensity coupled with special endcaps have enabled the creation of a “dead space” for fuel cooldown, allowing he creation of a reactor with a single fuel element, with an order of magnitude less power than the RD-600, and without a circulation contour for fission fuel. Further draft proposals outlined the use of this reactor core as a standardized design, producing an NTR with a thrust of 50-60 tf.

High parameters of hydrogen plasma heated up by a GCNR make it especially appealing for powerplants that use high-efficiency heat-to-electricity conversion using magnetohydrodynamic generators. A draft proposal for such a system was drawn up by Energomash, the Keldysh Centre, PEI and IPPE.

Over the development of the draft a significant volume of testing of key processes on model systems and of neutron flux profiles on physical assemblies were performed.

A major role in theoretical and calculative work in designing the Type V system and buttressing the experimental developments was played by I.M.Ievlev and the laboratory under his leadership, namely K.I.Artamonov, N.N.Borisov, A.Ya.Goldyn, A.I.Gorin, M.M.Gurfink, A.M.Kostylev, V.N.Krylov, H.H.Kuznetsov, V.M.Matyshin, A.V.Moskolyov, O.I.Novoznov, A.B.Prishleptsov, A.A.Pavelyev, S.S.Preobrazhensky, E.P.Terekhov, R.A.Fedotov, A.A.Shirokov.

The development of the EU-610 powerplant will unlock considerable capabilities in achieving astronautic and general fusion power objectives.

Nuclear rocket motors and nuclear powerplants of Type V, possessing quantitatively and qualitatively greater capabilities compared to chemical and Type A nuclear rockets, are intended to ensure further progress in rocket and aerospace technology development.

Glushko V.P., Glinik R.A.

Archive 82/179 (72-80)

And finally, here’s an essay from a defunct… site, credited to an Aleksandr Valeryevich Khoroshikh at horoshih-aleksander at; it has images, citations, technical details, and it offers a saddening finale to the whole story.

Gas-core nuclear rockets

The idea of using a nuclear motor in a rocket dates back to the dawn of that field [1]. An ammonia NTR produced a specific impulse comparable or superior to hydrogen-oxygen chemical motor [2], without requiring sophisticated cryogenics and bloated tankage to contain the low-density hydrogen. In particular, there was a program to develop an R-7-style missile, with an NTR sustainer stage surrounded by six kerolox strap-ons [2]. (see – ed.)

At the time it was assumed that a nuclear motor would become the core of a successful intercontinental ballistic missiles, ensuring considerable funding of that sphere. However, once chemical motors (especially ones using heptyl-amyl (official codenames for UDMH-N2O4 – ed.)) achieved performance that satisfied the military, the concept of an NTR-based ICBM was abandoned. However (sic), the USSR initiated a space program that included a manned Lunar, and later a Martian mission. One should not forget Nikita Khrushchev’s announcement of the Soviet moonshot. This reinvigorated NTR development.

An ICBM could only utilize a solid-core NTR, with all other options (liquid or gaseous fuel in a hollow reaction chamber) inevitably leading to some of the fuel escaping and contaminating the environment [4] Furthermore, as mentioned above, reaching America turned out to be within the capabilities of chemical rockets. This relegated NTRs to the single role of a high-efficiency motor for upper stages of boosters and for interplanetary craft.

Interplanetary missions are especially dependent on engine specific impulse, as the requisite Δv approaches tens of kilometers per second. In this aspect, gas-core NTRs are particularly outstanding, capable of exhaust velocities comparable to electric rockets [5] while also developing thrust comparable to chemical motors. Unlike electric rockets, they can achieve the requisite speed in comparatively short order, rather than months. This in turn allows rapid transit through the Earth’s radiation belts, subjecting the crew to much less radiation. Also, quite notably, not only the minimum energy Hohmann transfers, but “fast-track” trajectories, like parabolic orbits, become possible.

“The decision to develop NTRs and spaceborne nuclear electric powerplants based on gas-core nuclear reactors was formulated by Energomash chief academician V.P.Glushko in 1963 and was later approved by a Decree of CC CPSU and the Government of the USSR. By then the scientific corps at Energomash had six years of experience in designing and developing SCNRs. Theoretical research into GCNRs had been conducted since 1957 under the leadership of a USSR Academy of sciences corresponding member V.M.Ievlev at the Scientific institutes of heat processes (later the Keldysh Centre). Only two countries, USSR and USA, have attempted to tackle this technology, comparable in complexity to controlled thermonuclear fusion and requiring colossal financial expenditures,” as [6] describes the beginning of GCNR research.

Energomash’s lead unit on gas-core reactors and derived NTRs was a section led by R.A.Glinik. Solving the problem involved in design required cooperation with numerous institutes (primarily from the aerospace and nuclear industries) and the country’s leading universities under the overall scientific leadership of the Keldysh Centre. Considerable support was lent from the country’s leading scientists, such as academicians M.D.Millionschikov, A.A.Bochvar, Ye.P.Velikhov [6]. One of the key participants was B.I.Katorgin, who was also involved in development of RD-560 (H2O2 and beryllium hydride) and RD-600 (GCNR) engines. In addition to developing the particular systems, this development provided fundamental information on flow dynamics of pseudo-liquefied powdered fuels and combustion products within the chambers, on feeding said fuels into the combustion chamber and igniting them with decomposition products (in RD-560), along with gaseous flow dynamics (in RD-600). This work formed the basis of the candidate thesis Katorgin defended at the Bauman Moscow State Technical University in 1967 [7].

The designers encountered a range of principle difficulties. Here is a lis of some of them: [6]

  1. Operation of a gaseous fuel element
  2. Achieving criticality in a gas-core reactor
  3. Achieving stable functioning of the gas-core reactor
  4. Maintaining functionality of components and subsystems at extreme temperatures
  5. Ensuring resistance of construction materials to corrosion
  6. Thermal protection of the nozzle and MHD generator
  7. Separation of fission products in closed-cycle GCNRs

In 1963-1973 the GCNR and gas-core reactor unit of Energomash included approximately 90 people. That period saw intense experimental and production work on preparing reactor testing that was due to launch in 1975. However, in 1974 Energomash began developing the RD-170/171 – a high-performance kerolox rocket engine for the Energiya-Buran system (along with the Zenit booster, and later for ULA Atlas V in the form of RD-180, for Antares and Angara in the form of RD-190 and for Soyuz-2.1v in the form of RD-193. So much for abandoning cryogenic fuels! – ed.), which caused GCNR research to be halted and the relevant section reduced to 30 people. Over eight years the funding was only sufficient for on-paper studies, resulting in a considerable loss of technological, industrial and experimental groundwork. [5,6]

Starting in 1982 full-scale development was resumed, and the restored design and development unit spent two years recovering the technological and experimental base. However, in late 1989 funding was cut almost entirely. Neither did any of the programs in the United States reach even the small-scale demonstration experiment stage.

It was expected that the GCNR would consist of one or several reaction chambers surrounded by a neutron moderator-reflector. The nuclear fuel inside the chambers would be suspended in a plasma state, without contacting the chamber walls, in a quantity sufficient for a self-sustained chain reaction. The reaction mass would flow through the gap between the fissioning plasma and the chamber walls. Reaction mass heating is through radiative energy transfer, with average temperature at chamber egress reaching 104 K. Absorption of radiated heat also provides thermal protection for the chamber walls.

The key problem in developing the gas-core reactor was minimizing the loss of fission fuel, which had to be kept within tens of percent of reaction mass flow. Acceptable loss level was to be ensured by laminarization of the inbound reaction mass flow, profiling the field of its initial velocities, an external magnetic field, appropriate choice of working materials, and chamber geometry. Loss of fission fuel was to be compensated by its further input in either liquid form (at 1500 K) or as a paste-like powder mix with a NaK eutectic.

Spaceborne powerplants were designed along both open-cycle and closed-cycle lines. If the working fluid is ejected through a rocket nozzle, then the system is an open-cycle rocket motor. The working fluid if hydrogen that, for the purposes of increased radiative absorption and electric conductivity is doped by NaK and Li vapours along with tungsten powder; this also helps achieve acceptable reactor wall temperature. An NTR of such a design would possess extremely high specific characteristics (Isp on the order of 2000-3000 sec). If the system is designed to eject the hydrogen through a high-efficiency MHD generator, then it is an open-cycle powerplant.

A closed-cycle powerplant still uses the MHD as a power converter, but all working elements are cycled through isolated loop. In this case we gain a nuclear electric powerplant of impressive efficiency (30-40%) and of low specific mass and working medium expenditure. The additives to the working mass are, among other aspects, intended to improve interaction with the MHD generator. In addition to the reactor and MHD generator, the design inevitable has to include refrigeration, separation and pumping system. The working medium is a mixture of NaK steam and helium. The excess heat is dumped into space via radiators. The power produced can be utilized for a variety of purposes, not the least to power an electric rocket.

And advantage of gas-core systems over solid-fuel rods in closed-loop powerplants is the ability to considerably extend uninterrupted operation via continuous input of fresh fission fuel to replace the extracted reaction products.

The conceptual design of a nuclear engine-powerplant system for a manned Mars expedition is the latest-dated, and encompasses all past experience. The design is based around a combined single cavity solid-and-gas-core transforming reactor massing 57.5 t. Gross heat output 2.14 GWt. Solid fuel assemblies, arranged in a ring around the reactor vessel and mounted on an input-extraction system, provide the necessary level of neutron flux for criticality at start-up, before the fuel is introduced to the gaseous fuel element cavity. As the fission fuel is introduced and accumulated in the central cavity, i.e. as a plasma zone appears and the gaseous fuel element is formed, the solid fuel rods are retracted from the active zone, and the reactor becomes a pure gas-core system.

Thanks to the transforming design the system has two modes of operation:

  • thrust (gas-core) mode developing 17 tf (according to other sources, 600 tf [8]) with an Isp of 2000 sec – for boost and deceleration stages of the trajectory;
  • energetic (solid-core) mode developing 200 kWt of electric power for supplying the needs of a spacecraft with no loss of working medium – for coast stage of the trajectory. This mode involved the operation a closed-cycle gas turbine circuit with a He-Xe mixture as working medium, thermal energy conversion at 20% and radiative cooling via the Brayton cycle.

In thrust mode, electric energy is generated by a 25 MWt MHD generator integrated into the nozzle, with electrodes and excitement busses oriented down the nozzle’s throat. [6,9]

RD-600 SCHEMATIC [10] (Figure 1)

Power block layout: 1 – drive electromotors; 2 – feed-screw; 3 – retractable solid fuel rods; 4 – radiation shield; 5 – coaxial coils; 6 – reaction zone; 7 – reactor structure; 8 – solenoid; 9 – carbon fiber reinforcement coiling; 10 – solenoid heatshield; 11 – lateral moderator-reflector; 12 – high-temperature molybdenum bulkhead; 13 – integrated MHD generator; 14 – supersonic expander nozzle; 15 – front endcap; 16 – fuel rods (graphite with dispersed uranium carbide); 17 – rear endcap; 18 – channels filled with 3He (!? – ed.) (reactor control system actuators); 19 – electrodes for the multipolar Faraday MHD generator

A layout of a Mars Expeditionary Complex using a block of two above-described gas-core nuclear systems is depicted in Figure 2. At assumed payload of 150 tons for this mission type, approximate mass of the complex in Earth orbit would be 520-540 t depending on launch date. For comparison, an SCNR results in 730-800 t and chemical rockets in 1700-2500 t.

MEC LAYOUT (Figure 2) (the awful quality is inherited from the original; curiously, it’s annotated in both English and Russian. The scale at the bottom-right is 10 m. Going left-to-right/aft-to-bow, the spacecraft consists of a pair of engines, a V-shaped radiator array around a load-bearing truss that also contains the lithium tanks. Everything between that and the truss section further to the right are hydrogen tanks (jettisonable? There are three distinct sets); the foremost section costs of an orbital habitat with an Earth Return Vehicle stuck to one side and a very vaguely-depicted Mars Landing Vehicle on the other. – ed.)

Gas-core reactor and derived GCNR development strategy was based around three incremental phases. The initial stage involved the still-operational unique testing facility based around an Impulse Graphite Reactor (IGR) at the Semipalatinsk nuclear range, Kazakhstan. It involved brief (up to 5 sec) operational tests of reduced-scale modes of gaseous fuel elements, up to 100 mm in diameter and 250 mm in length.

The second phase would involve constructing a new IGR-type reactor called Nephrite, capable of testing samples thrice the size and for an order of magnitude greater durations.

The final stage would involve a full-scale prototype testbed gas-core or, more precisely, combined solid-and-gas-core reactor dubbed Lampa (before you ask, no other mention of lightbulbs or quartz in anything I’ve found – ed.), with an active zone capable of housing a “dead zone” type, self-contained gas-core fuel element.

The last two stages would take place at the Baikal-2 stand complex, also at Semipalatinsk. Baikal-2 has had significant research invested into it, with considerable attention paid to safety concerns, primarily radiological and nuclear; in particular, the system was built entirely for closed cycle.

The preparation for first stage of practical testing of a scale model of a gaseous fuel element in the IGR reactor took the most time and resources. The experimental ampula, containing a model of a fuel elements and all the requisite systems, was to be located into the vertical canal at the centre of the reactor. Over the course of the experiment a displacement-type system was to introduce the fission fuel into the working chamber, located in the middle of IGR’s own active zone. The furl could be a paste composed of small particulate powder of uranium and alkaline metals, or liquid uranium heated before introduction into the chamber. The fuel introduction tract had highly effective and compact neutron shielding in order to prevent overheating of fuel and the surrounding container. Primary dimensions of the interior of the working chamber: diameter 80 mm, length 240 mm. The uranium-containing stream, once injected into the chamber, would be hit by the intense neutron stream, heat up, vaporize and ionize. Radiation from the plasma heated up the working medium. The conical inner wall of the entry section of the working chamber was composed of a high-melting-point alloy. The wall was made permeable to allow injection of hydrogen and helium alongside the fuel. This prevented the formation of a recircularization zone in the fuel vaporization are, and inflow turbulence. The incoming hydrogen, on the other hand, provided a coaxial boundary layer that isolated the chamber walls from the primary uranium plasma stream.

The cylindrical section of the working chamber had an ablative coating along its interior, providing reliable protection to the outer structure, including in cases of metallic uranium condensing on the ablative material (by the ablation forcing uranium back into the primary stream).

Upon exiting the chamber the high-temperature stream of reaction mass was to enter the condenser. The walls of the condenser had rows of slots used to inject gaseous hydrogen for the purpose of dilution. Furthermore, the interior of the condenser also had the anti-uranium coating. To reduce heat flow from uranium while passing through the condenser, the exterior of the condenser was also equipped with neutron shielding. The resultant gas mixture containing fission products would then be forced through a transonic nozzle towards the filtration system located near the bottom neutron shield of the reactor. Large particles would be intercepted by the inertial traps, while smaller ones would be caught in metal-ceramic filter cartridges. The use of a transonic nizzle would stabilize the pressure in the working chamber should hydraulic resistance of filtering cartridges change over the course of the experience. Gaseous products would then be routed to the test stand exhaust containment system. To limit head generation and filter heat-up it was also equipped by stationary lateral and bottom neutron shielding.

Once the fission fuel was exhausted and the test was completed, the shutdown was performed by cooling the heat-producing fuel residue in the ampule filter with a stream of gas.

The experimental ampule (Figure 3), produced at a pilot plant, had a diameter of 185 mm and a length of 6500 mm, and included the following components: fission fuel feed system, working chamber, condenser, and filter. This, along with communication systems, measurement sensors and overall assembly elements, was packaged into the airtight shell. It was assumed that the requisite supply of the fission fuel would be loaded into the ingress tract of the ampule immediately before commencing the operation. After the test, all solid and liquid products are retained within the filter. Therefore, radiation safety across all stages of operation is ensured through localization and containment of the fission fuel and the bulk of the fission products within the ampule’s interior. Leakage of radioactive substances into the environment was completely excluded.

The central canal of the IGR impulse reactor, which would hold the experiment ampule, had a water-cooled airtight shell separating it from the uranium-graphite cladding of the active zone. The upper part of the ampule mounted the connections to the test stand’s communications.

Considerable attention was paid to safety measures preventing damage of IGR reactor core and radioactive contamination of testing facilities in case of possible failure modes of functional subsystems within the experimental ampule.

Two complete test packages with miniaturized fuel elements were completed and ready for delivery to the test facility (Figure 5) (missing here, see same at – ed.). Specialized test stand equipment kits and expended experimental radioactive material handling and transport equipment were already dispatched there. In addition to the experimental ampule, a draft design of an advanced gas-core fuel element with a “dead space” cooling area and magnetic stabilization was also completed



  1. Б. Е. Черток "Ракеты и люди. Книга 4 Лунная гонка"
  3. Первушин "Битва за звёзды. Космическое противостояние"
  4. Л. Гильберг "Покорение неба", стр. 325-326
  6. "Двигатель", "Газофазные ядерные двигатели для космических аппаратов", 5 1999 г.
  9. "Двигатель", "Газофазные ядерные двигатели для космических аппаратов", 6 1999 г.

I’ve tracked down the citations. 1 is Boris Chertok’s somewhat less technical autobiographical work covering the Soviet rocket program since he was hunting for leftover V-2s alongside Sergei Korolev; it’s available at 2 and 8 are the same master list of Energomash engines I started from. 3 and 4 are aerospace tech populariser paperbacks. 5 is a dead link to the same site I got Glushko’s correspondence from. 7 and 10 are obsolete links to a periodical that’s behind a paywall; the up-to-date links are and

6 and 9 are from a journal with the telling name Engines, a two-part article about GCNRs by Grigory Lioznov, an Energomash engineer whom we’ve heard about earlier; they’re available online at and Aside from several more images, I’ve gleaned two very interesting sections not covered above:

Miniaturization and reduction in mass of GCNRs are facilitated by:

Use of uranium-233 fission fuel

•Maximization of use of metallic beryllium, including large-crystal beryllium, in the moderator-reflector assembly, with the rest composed of graphite

•Maximization of use of metals with improved isotope composition and high melting temperatures in the design of reaction chamber interior, and of high-durability titanium alloys and reinforcing carbon composites in the reactor frame.

•Use of hyperconductive aluminium (0.9999 purity) in high-current magnetic stabilization, MHD excitation and turbopump power feed systems due to its ability to conduct up to 50-100 A/mm2 when cooled by liquid hydrogen, while developing less than a tenth of resistivity of copper.

It is obvious that the temperature extremes in operation of many GCNR components and the highly chemically aggressive environment (molten uranium, high-pressure hydrogen, alkali metals) required significant materials science investigations. As a result, high melting point alloys based on tantalum-tungsten-hafnium as well as niobium were developed for the fission fuel feed system. Certain areas of the reactor vessel have necessitated the development of heat-resistant porous materials based on tungsten and molybdenum, for the high-temperature fuel filters – on nickel and nichrome.

Estimated parameters of a gas-core fuel element
Pressure in reaction chamber, kgf/cm2200
Uranium expenditure, g/sec200
Hydrogen expenditure in the reaction chamber, g/sec10
Velocity of fuel when entering the reaction chamber, m/s1.7
Power, kWt1,000
Share of vaporized uranium in the egress flow, %80
Temperature of uranium plasma, K8—10×103(unclear – ed.)
Thermal neutron flow, neutrons/cm2/sec1015(unclear – ed.)

Final reflections? Just… holy hell. You never know if that old Soviet closet has a skeleton or a suit of Mobile Infantry powered armour in it.

Yours sincerely,
Denis Danilov


These acronyms are odd looking because the original words are in Russian.

  • GFTE: gas-phase fuel rod
  • GMFR: gas-phase nuclear reactor
  • SFSF: structure in high-temperature
  • TFTS: solid-phase fuel assemblies
  • YACEU: radioactive nuclear power and nuclear space power plants
  • YARD: nuclear rocket engine

The motor power plant of the open circuit (Fig. 1) includes a single-cavity reactor with an annular output channel and a gas-phase fuel rod (GFTE) with a stagnant nuclear zone of nuclear fuel. Zone stabilization is carried out using a powerful external solenoid. The use of such a scheme for environmental reasons is possible only on spacecraft, but not on carriers launched from the Earth.

In order to provide energy to various consumers, including a solenoid and an electric pump drive, the unit was supposed to use a combination of a nozzle and an MHD generator. In addition to the circuit difference, the YARD and YACEU differ in the degree of use of the gas flow energy in the MHD generator: in the first case, no more than 2% is converted into electricity, and in the second - 30 ... 40%.

In installations of a closed circuit (Fig. 2), the energy converter is an MHD generator, and all the working components circulate in a circuit that does not have connection with the external environment (so it does not unnecessarily spray fissioning uranium and fission fragments all over tarnation). In this case, we obtain YACEU, which has a very high efficiency (30:40%), low values ​​of the specific gravity of the converter and specific consumption of the working fluid. Additives introduced into the working fluid, among other things, are designed to promote MHD-interaction. In addition to the gas-phase reactor and the MHD generator, refrigerators, separators and pumps must be present in the design. The working fluid is NaK vapor mixed with helium. Released excess heat is discharged into outer space using emitters. The energy produced is used for various purposes, one of its consumers may be an electric rocket engine.

The advantage of using GFNR in closed circuits, in which gaseous gas is used instead of solid fuel rods, is the fundamental possibility of ensuring a very long-term operation due to appropriate fueling instead of nuclear reaction products removed from the circuit to the external environment

(Nyrath note:solid fuel rods become clogged with nuclear reaction products (nuclear poisons) when 15% of the fuel has been burnt (nuclear fission). A clogged rod will not support fission. The rod has to be removed and taken to a reprocessing plant. For this reason NASA's reusable nuclear shuttle has an engine life of only 10 Terra-Luna round trips before disposal, even though the rods still contain 85% of the expensive uranium-235 unburnt. Extracting the rods for reprocessing is too dangerous in NASA's eyes, they just send the nuclear shuttle into a graveyard orbit.

The same limit applies to a nuclear power plant. Ground based plants periodically halt operations when the solid fuel rods become clogged. They then spend a few months carefully opening the reactor, removing the clogged rods, replacing them with fresh rods, then carefully reassembling the reactor. The clogged rods are sent to a reprocessing plant to extract the unburnt U235 and using it to fabricate fresh rods. The rest of the rod goes to a long term nuclear waste disposal site.

If I am reading the above sentence correctly; they are implying that the fuel gas, after one pass through the MHD generator, can be passed through an on-board refinery. The product is already gaseous, as opposed to solid fuel rods, which simplifies refining. In the refinery the nuclear reaction products clogging the gas can be filtered out, and the unburnt U235 can be sent on another pass through the MHD generator. The alternative is removing the gas "from the circuit to the external environment", i.e., wastefully jettisoning the gas into deep space along with all its expensive unburnt U235.

Now that my nose has been rubbed in the fact, I realize that a nuclear lightbulb gas core engine should also require an on-board reprocessing plant. But in all the nuclear lightbulb documents I've read, a re-processor is conspicuous by its absence. I'm going to have to review the documents.)

The fact that in closed circuits the requirement for the removal of nuclear fuel from the reactor together with the working fluid is less stringent than in the open is also significant. This allows us to consider a more simple organization of processes that allow a greater degree of mixing of nuclear fuel and working fluid. In this case, there is no need for magnetic stabilization — the plasma zone from the stagnant turns into a jet zone. The use of several such zones (multi-cavity reactor) improves the overall size characteristics of the SFSF.

It is known that there is a definite relationship between the thermal power of the reactor and the possibilities of providing an acceptable temperature condition for the structural elements. Research has found that the optimal thermal capacity of an open-circuit GFNR should be no less than 2 GW, and a closed one — 300 MW (with a pressure in the working chamber of about 1000 kgf / cm2).

The conceptual development of a nuclear propulsion system to support the Martian expedition is the latest in time, incorporating all previous experience. The installation is based on a combined single-cavity gas-phase-solid-phase reactor of a transformable structure weighing 57.5 tons (Fig. 3). Thermal power of the reactor is 2.14 GW. Solid-phase fuel assemblies (TFTS), placed in a ring around the central cavity of the reactor and equipped with drive mechanisms, provide the necessary level of neutron flux and criticality at start-up when there is no nuclear fuel in the cavity of a gas-phase fuel element. With the supply and accumulation in the central cavity of a nuclear fuel, i.e. the formation of the plasma zone and the formation of a gas-phase fuel element, TFTS from the active zone are extracted.

Thanks to the transformable design, the installation can operate in two modes:

  • Thrust Mode: a 17 tonne-force motor (gas phase) with a specific impulse of 2000 seconds
  • Coast Mode: energy (solid-phase) with an electric power of 200 kWe to meet the internal needs of the spacecraft without spending the working fluid. This mode is provided by a closed gas turbine circuit with a helium-xenon mixture as a working fluid, converting thermal energy into electricity with efficiency % and the discharge of excess heat through the heat radiator (Brighton cycle)

In thrust mode of operation, the power supply is provided by a 25 MW multi-pole MHD generator built into the nozzle with electrodes and field buses oriented along the nozzle-forming elements.

Nuclear Salt Water
20% UTB
FuelUranium Tetrabromide
20% enriched
2% UTB solution
Exhaust Velocity66,000 m/s
Specific Impulse6,728 s
Thrust12,900,000 N
Thrust Power425.7 GW
Mass Flow195 kg/s
Plenum Radius3.075 cm
Plenum Length65 cm
Total Engine Mass33,000 kg
ReactorGas Core
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorPusher Plate
Specific Power77.5 kg/GW
90% UTB
(missing items same as above)
FuelUranium Tetrabromide
90% enriched
Exhaust Velocity4,725,000 m/s
(1.575% c)
Specific Impulse482,140 s
Thrust Power30,600 GW
(30.6 TW)
Mass Flow3 kg/s
Specific Power1.1 kg/GW

This concept by Dr. Zubrin is considered far-fetched by many scientists. The fuel is a 2% solution of 20% enriched Uranium Tetrabromide in water. A Plutonium salt can also be used.

Just to make things clear, there are two percentages here. The fuel is a 2% solution of uranium tetrabromide and water. That is, 2 molecules of uranium tetrabromide per 100 molecules of water.

But the uranium tetrabromide can be 20% enriched. This means that out of every 100 atoms of uranium (or molecules of uranium tetrabromide), 20 are fissionable Uranium-235 and 80 are non-fissionable uranium. If it is 90% enriched, then 90 atoms are Uranium-235 and 10 atoms are non-fissionable. As a side note, 90% enriched is considered "weapons-grade".

The fuel tanks are a bundle of pipes coated with a layer of boron carbide neutron damper. The damper prevents a chain reaction. The fuel is injected into a long cylindrical plenum pipe of large diameter, which terminates in a rocket nozzle. Free of the neutron damper, a critical mass of uranium soon develops. The energy release vaporizes the water, and the blast of steam carries the still reacting uranium out the nozzle.

It is basically a continuously detonating Orion type drive with water as propellant. Although Zubrin puts it like this:

As the solution continues to pour into the plenum from the borated storage pipes, a steady-state conditions of a moving detonating fluid can be set up within the plenum.

He also notes that it is preferable to subject your spacecraft to a steady acceleration (as with the NSWR) as opposed to a series of hammer-blow accelerations (as with Orion).

The controversy is over how to contain such a nuclear explosion. Zubrin maintains that skillful injection of the fuel can force the reaction to occur outside the reaction chamber. He says that the neutron flux is concentrated on the downstream end due to neutron convection. Other scientists are skeptical.

Naturally in such a spacecraft, damage to the fuel tanks can have unfortunate results (say, damage caused by hostile weapons fire). Breach the fuel tubes and you'll have a runaway nuclear chain reaction on your hands. Inside your ship.

The advantage of NSWR is that this is the only known propulsion system that combines high exhaust velocity with high thrust (in other words, it is a Torchship). The disadvantage is that it combines many of the worst problems of the Orion and Gas Core systems. For starters, using it for take-offs will leave a large crater that will glow blue for several hundred million years, as will everything downwind in the fallout area.

Zubrin calculates that the 20% enriched uranium tetrabromide will produce a specific impulse of about 7000 seconds (69,000 m/s exhaust velocity), which is comparable to an ion drive. However, the NSWR is not thrust limited like the ion drive. Since the NSWR vents most of the waste heat out the exhaust nozzle, it can theoretically produce jet power ratings in the thousands of megawatts (meaning it is not power limited, like other nuclear propulsion). Also unlike the ion drive, the engine is relatively lightweight, with no massive power plant required.

Zubrin suggests that a layer of pure water be injected into the plenum to form a moving neutron reflector and to protect the plenum walls and exhaust nozzle from the heat. One wonders how much protection this will offer.

Zubrin gives a sample NSWR configuration. It uses as fuel/propellant a 2% (by number) uranium bromide aqueous solution. The uranium is enriched to 20% U235. This implies that B2 = 0.6136 cm-2 (the material buckling, equal to vΣfa)/D) and D = 0.2433 cm (diffusion coefficent).

Radius of the reaction plenum is set to 3.075 centimeters. this implies that A2 = 0.6117 cm-2 and L2 = 0.0019. Since exponential detonation is desired, k2 = 2L2 = 0.0038 cm-2. Then k = U / 2D = 0.026 cm-1 and U = 0.03.

If the velocity of a thermal neutron is 2200 m/s, this implies that the fluid velocity needs to be 66 m/s. This is only about 4.7% the sound speed of room temperature water so it should be easy to spray the fuel into the plenum chamber at this velocity.

The total rate of mass flow through the plenum chamber is about 196 kg/s.

Complete fission of the U235 would yield about 3.4 x 1012 J/kg. Zubrin assumes a yield of 0.1% (0.2% at the center of the propellant column down to zero at the edge), which would not affect the material buckling during the burn. This gives an energy content of 3.4 x 109 J/kg.

Assume a nozzle efficiency of 0.8, and the result is an exhaust velocity of 66,000 m/s or a specific impules of 6,7300 seconds. The total jet power is 427 gigawatts. The thrust is 12.9 meganewtons. The thrust-to-weight ratio will be about 40, which implies an engine mass of about 33 metric tons.

For exponetial detonation, kz has to be about 4 at the plenum exit. Since k = 0.062 cm-1, the plenum will have to be 65 cm long. The plenum will be 65 cm long with a 3.075 cm radius, plus an exhaust nozzle.

Zubrin then goes on to speculate about a more advanced version of the NSWR, suitable for insterstellar travel. Say that the 2% uranium bromide solution used uranium enriched to 90% U235 instead of only 20%. Assume that the fission yield was 90% instead of 0.1%. And assume a nozzle efficency of 0.9 instead of 0.8.

That would result in an exhaust velocity of a whopping 4,725,000 m/s (about 1.575% c, a specific impulse of 482,140 seconds). In a ship with a mass ratio of 10, it would have a delta V of 3.63% c. Now you're talkin...

Ken Burnside: In my game universe, the engineers call the pumps that feed Uranium Tetrabromide solution into the reaction chamber "Wileys", reputedly after the engineer who first made them safe to use and maintain.

More than likely, it's after the coyote of the same name...

Winchell Chung: An appropriate name for what are basically atomic squirt-guns.

From a thread in SFConSim-l (2002)
Zubrin NSWR
Zubrin NSWR
Exhaust Velocity78,480 m/s
Specific Impulse8,000 s
Thrust8,696,900 N
Thrust Power0.3 TW
Mass Flow111 kg/s
Total Engine Mass495,467 kg
Frozen Flow eff.80%
Total eff.80%
Uranium Tetrabromide
ReactorGas Core
Remass AccelThermal Accel:
Reaction Heat
Thrust DirectorPusher Plate
Specific Power1.45 kg/MW

The illustration shows the vision of Robert Zubrin: a rocket riding on a continuous controlled nuclear explosion just aft of a nozzle/reaction chamber.

The propellant is water, containing dissolved salts of fissile uranium or plutonium. These fuel-salts are stored in a tank made from capillary tubes of boron carbide, a strong structural material that strongly absorbs thermal neutrons, preventing the fission chain reaction that would otherwise occur.

To start the engine, the salt-water is pumped from the fuel tank into an absorber-free cylindrical nozzle. The salt-water velocity is adjusted as it exits the tank so that the thermal neutron flux peaks sharply in the water-cooled nozzle.

At critical mass (around 50 kg of salt water), the continuous nuclear explosion produces 427 GWth, obtaining a thrust of 8600 kN and a specific impulse of 8 ksec at a thermal efficiency of 99.8% (with open-cycle cooling). Overall efficiency is 80%.

Robert Zubrin, "Nuclear Salt Water Rockets: High Thrust at 10,000 sec ISP," Journal of the British Interplanetary Society 44, 1991.

You need much more propellant than fuel, 22,000 times more in the case of the Zubrin without open cycle cooling, and 44,000 times more if open cycle cooling is used.

The Zubrin drive exhaust (without open cycle cooling) contains 108 kg/sec of water, but only about 5 grams/sec of uranium.

(This is from a quick calculation: mass flow equals the Zubrin thrust (8.7 meganewtons) divided by the exit velocity (80 km/sec) = 108 kg/sec. But the fissioning energy can be estimated from the Zubrin total power of 427 GW divided by the energy content of Uranium 235 of 83 TJ/kg.)

Dr. Zubrin responded, and he defends the performance of the Zubrin drive as depicted in the game (as high thrust & high specific impulse rocket with low mass and low radiators).

1). In U235 fission, only about 2% of the energy goes into neutrons (unlike D-T fusion).

2). The design uses a pusher plate or open nozzle, like an Orion drive. Or magnetic confinement (since most of the energy is released as a plasma). Therefore, the opportunity to absorb heat is low.

3) Many of the neutrons that are intercepted would sail through the pusher plate, rather than be absorbed as waste heat.

4) No lithium should be in the outer water, because this would poison the fission reactions.

5). Because the design does not use a heat engine cycle, the radiators could be far hotter than ones in the game. He suggested graphite at 2500 K°. That would drop the required radiating area by a factor of 40 (2.5 to the fourth power), which means that the radiator could be the first wall itself.

Dr. Zubrin went on to say the chief disadvantage is the expense of the fuel (like He3-D and antimatter drives).

Philip Eklund, from a discussion on the High Frontier Yahoo group about the NSWR drive in High Frontier

     "So anyway, we were passing through the outer Kirkwood Gap, totally the ass end of nowhere. I'm trying to catch some rack time, XO has the conn, nice boring trip to Europa." The CO of U.S.N.A.S. Saskatchewan tipped back another shot of Scotch and continued his story. "Totally routine, right? No problems at all. So then, all of the sudden, the whole ship gets racked. Meteoroid. Big one, too, maybe a centimeter across."

     The captains seated around the table, two Americans including Fitzthomas, an Indian, three Chinese, and the South African, all clucked and groaned.

     "Well, we got lucky and it missed the crew compartment, but by the time I get to command the chief engineer is screaming over the intercom that it holed tank one, busted three tubes, and we've got nuke juice pooling and we have to dump the tank. Problem is, we're running at top speed and if we dump the tank, we don't have enough propellant to stop at Europa. We'd have to ride all the way out to Neptune, sling around, and hope someone from the inner solar system has dispatched a tanker to intercept us on the return trip, and we don't have near enough consumables for that."

     "So what did you do, mate?" said the South African.

     "I told the chief he had to fix the tank or we'd all starve before we could stop the damn ship. Well, he screams some more that we don't have time, and I tell him his choice is fix the tank or die real fast in a runaway, because we're not going to die slow in the void. So he grabs a crew, stuffs them into suits, and crawls out onto the tank. They punch some holes in it to let the juice drain instead of pool, but it's still leaking like a f***** and the water's evaporating and leaving uranium crusted all over everything. So he radios command and says, 'It's still leaking, and all this uranium crud is going to accumulate into a critical mass somewhere, so we still have to drop the tank.' Meteoroid busted open three valves, you see. No way to stop the leak. And I tell him again, that's no good, and by now astrogation has confirmed it and the XO has tallied up the consumables and I know for sure we don't have enough for an unscheduled trip to Neptune.

     "So he says something about how he's not a miracle worker, and I tell him he damn well has to be. Lo and behold, he and his crew go ahead and do something crazy and it works."

     "What was that?" said the South African.

     "They take torches to the tank. The plug up the broken pipes as best they can, and then they go ahead and cut away the smashed cells. Just cut it off and jettison it into space, and suddenly the propellant that's still leaking is leaking right into space. We have lousy flow through the tank and the braking burn is going to be real tricky, but we can make Europa. I put the chief up for a commendation medal for figuring that out on the fly and saving our asses."

     The other captains nodded their approval at the chief's quick thinking. Good chiefs prevented accidents; great ones prevented disasters.

     "Is the chief's name Mr. Scott, by any chance?" said one of the Chinese captains.

     Commander George Allen, New Jersey's full blooded Cherokee XO, drifted into the command deck from astrogation, where he'd been monitoring the final approach to Hektor. He took his place at the copilot station and put on his headset. Fitzthomas toggled his direct channel to Allen's station.

     "How was the approach?"
     "We wasted too much propellant before the chain reaction started. I think Pennai should inspect the nozzles and pumps before we get underway again."
     "What does Pennai say about it?"
     Pause. "Pennai thinks the fuel is dirty."
     "Is it?"
     "It was certified 90% enriched at Roosevelt Station."
     "Is there any way to test it here?"
     "No sir. Not without a centrifuge."
     "How does Pennai know, then?"
     "Some engineering technobabble about neutron flux and reaction rate. I couldn't follow a tenth of what she said."
     Fitzthomas considered that. "Have her inspect the pumps and nozzle alignment. If they pass, then we might have a fuel problem."

     "Captain," said Allen, "Thought you'd like to know: Pennai just inspected the entire fuel line. Everything there is in order."
     "So what are you telling me, George?"
     "I think we have dirty fuel."
     "What's her recommendation?"
     "She wants to drain the tanks and top up with the good stuff. But I can't—"
     "Write that request, I know. The CO has to. Where's Pennai now?"
     "Racked out. She has the midwatch tonight."
     "After her watch tonight, she has four days of leave."
     "Sir, she's supposed to be OOW all day Wednesday."
     "I'll take that shift. She was right, we were wrong. She deserves to be rewarded. When I get back I'll write up a request and have it to the fuelmaster by tomorrow AM."

     "Do you have to return to your ship?"
     "Yeah. Dirty fuel, God damn it. Wait until I get my hands on the fuelmaster at Roosevelt."

     (Admiral Castro said) "Anyway, I saw your chief engineer's report. I passed it back to Fleet. The fuelmaster at Roosevelt Station is going to have a lousy day tomorrow. There's also a bulletin going out to the entire fleet. Everyone who tanked up at Roosevelt near the same time you did should keep a close eye on his reaction rate. Your Lieutenant Pennai might be up for a commendation letter in her file."

     Duvalier left Ortiz main engineering and vaulted down the access tube to the reactor room. The tube ran down the ship's spine, surrounded by megaliters of water enriched with uranium salts in highly complex tanks made of neutron absorbing material. In his head, he knew the tube was the safest part of the ship, shielded from the worst the universe could throw at it by dozens of meters of water. In his head, he knew the fuel, so long as it didn't pool into a critical mass somewhere in the thousands of kilomters of pipes on all sides of him, emitted only low intensity alpha rays which couldn't penetrate his own skin, let alone the aluminum skin of the pressure tube. It was all perfectly safe, so far as anything in space could be safe. He knew that in his head.

     His balls, however, hadn't gotten the memo. His testicles tried to crawl up into his body every time he climbed through the hatch.

From The Last Great War by Matthew Lineberger (not yet published)
Lithium Salt Water

      My friend Robert Zubrin published a paper about Nuclear Salt Water Rockets back in 1991.
     Here, you took water along with nuclear salts, and stored them in neutron absorbing tanks, exhausting the water through a nozzle that allowed the salts to achieve criticality. You could achieve very high exhaust velocities that way!

     Another approach is possible. Namely, putting Lithium-6 Deuteride in water, along with neutron multiplying materials, and exposing the fusion fuels to a high neutron flux in the rocket engine. Similar exhaust velocities can be achieved. Without the radioactive byproducts of the Nuclear Salt Water Rocket approach.
     There are several advantages relative to conventional NTR designs. As the peak neutron flux and fission reaction rates occur outside the vehicle, these are far greater than what is possible when built into the vessel. A contained reactor can only allow a small percentage of its fuel to undergo fission at any given time, otherwise it would overheat and meltdown or explode in a runaway fission chain reaction. The fission reaction in an NSWR is dynamic and because the reaction products are exhausted into space it doesn't have a limit on the proportion of fission fuel that reacts.

     NSWRs are a hybrid between fission reactors and fission bombs.
     Due to their ability to harness the power of what is essentially a continuous nuclear fission explosion, NSWRs would have both very high thrust and very high exhaust velocity. The rocket would be able to accelerate quickly as well as be extremely efficient in terms of propellant usage. Zubrin proposed one design that generates 13 meganewtons of thrust at 66 km/s exhaust velocity. Another design achieves 4,700 km/s and uses 2,700 tonnes of highly enriched uranium salts in water to propel a 300 tonne spacecraft up to 3.6% of the speed of light.

     This basically solves the problem of spaceflight when done with Lithium-6 Deuterium Jetter Cycle process.
     That cycle may be sustained by a high flux of neutrons in the engine core. The High Flux Isotope Reactor is a model for the type of engine I'm talking about.

     Here neutrons are focused into a central tube through which water passes. The water moderates the neutrons passing through it. A beryllium reflector keeps the neutrons in the reactor. Lithium-6 Deuteride suspended in the water absorbs the neutrons and supports Jetter Cycle fusion. Helium gas, neutrons and steam are the only exhaust products.
     NSWRs share many of the features of Orion propulsion systems, except that NSWRs generate continuous thrust and work on much smaller scales than the smallest feasible Orion designs.

     A single stage system I described previously, with one crewman and five passengers seated in a capsule beneath a 30 cubic meter propellant tank, would mass 1,585 kg and carry 30,000 kg of water salted with 6LiD (mass ratio of 19.9), passing through a high neutron flux region to produce controlled thrust. With a 4,700 km/sec exhaust speed, the vehicle is capable of achieving (a delta-V of) 14,062.86 km/sec! Enough for a one gee boost of 16.6 days!!
     With 4 days of boost combined with 4 days of slowing down the ship can cruise at one gee a distance of 1,171.28 million km (torchship brachistochrone trajectory). This is sufficient to fly to any celestial body out to Jupiter and back. Reducing acceleration after planetary escape to 0.416 gees increases boost time to 10 days per leg, 40 days per round trip, and increases range to 6,075.15 million km. This is sufficient to take us to all celestial bodies in the solar system, including Pluto, Haumea, Makemake, and Eris.

One gee
0.42 gee

     Building a ship such as this, and flying it back to the moon, would be an appropriate opening to the second round of the space age. In less than a month such a ship could do a 'grand tour' of all the planets out to Jupiter.
     On 24 July 2014 The Grand Tour would be:
Grand TourDistDays

     The hop from Mercury to Jupiter and back to Earth would occur at 41% earth normal gravity to conserve fuel.
     A hop from Earth to moon would take 3.75 hours! Then a hop to Mars, taking 3.05 days. Then off to Venus 4.25 days. Thence to Mercury, 1.88 days. Then the long haul to Jupiter 10.14 days at reduced gravity. Then back to Earth again at reduced gravity 11.12 days.
     The five passengers brought along would pay $100 million each — which would pay for the entire development programme. The promotion of the power source used in the rocket, which would be developed along with the engine, would also pay dividends for the lucky passengers.

     The reaction conditions for a power plant is quite different than that for a rocket exhaust.
     A 1000 MW steam turbine for the US power plant Ravenswood Unit 3 was cross-compound with 16.6 Mpa, 538C steam conditions with 44.7% thermal efficiency. These steam conditions give the density of Li-6 Deuteride needed for a given neutron flux to maintain 1000 MW electrical output. A closed cycle steam system can be quite compact for this unit, with no exhaust to the atmosphere except helium, but even that can be mined from the closed cycle system. Lithium-6 Deuteride is added to the water tank to maintain power levels.
     1000 MW operating continuously and selling power for $0.10 per kWh produces $876.6 million per year revenue. Discounted at 6.25% per year over 30 years this revenue stream is worth $11.75 billion. Profits of $10 billion are possible for each unit sold. Half these profits from the first unit divided among the passengers, translates to $1 billion returns for the first power plant installed — with 50% revenue flowing to the money investors — and 50% flowing to the company — so, the trip plus $1 billion is what each passenger gets for their $100 million. They can even sell forward their seat as the vehicle nears flight readiness, and increase their revenue.

     This small ship described here massing 31.6 metric tons at lift off, with a take off acceleration of 2 gees, requires a mass flow rate of 142 grams per second with an exhaust velocity of 4,300 km/sec. This ship during lift-off produces an exhaust jet power of 1.31 trillion watts! The ability to reliably sustain this level of power safely and reliably in a compact space, is an obvious promotion of the company's capacity to build reliable compact and safe nuclear power units. Five 1000 MW turbines installed each year generate $50 billion in free cash flow, when monetized at discount rates.

     What do you with the money? Build bigger ships!
     The ships scale to any size. You open the solar system with them, and depopulate the Earth.

Fission Fragment Type

Fission Fragment

George Chapline
Specific Impulse1,000,000 sec
Exhaust velocity9,810,000 m/s

This is from Fission Fragment Rockets — A Potential Breakthrough

All of the other nuclear thermal rockets generate heat with nuclear fission, then transfer the heat to a working fluid which becomes the reaction mass. The transfer is always going to be plagued by inefficiency, thanks to the second law of thermodynamics. What if you could eliminate the middleman, and use the fission heat directly with no transfer?

That what the fission fragment rocket does. It uses the hot split atoms as reaction mass. The down side is that due to the low mass flow, the thrust is minuscule. But the up side is that the exhaust velocity is 3% the speed of light! 9,810 kilometers per second, that's like a bat out of hell. With that much exhaust velocity, you could actually have a rocket where less than 50% of the total mass is propellant (i.e., a mass ratio below 2.0).

The fission fragment is one of the few propulsion systems where the reaction mass has a higher thermal energy than the fuel elements. The other notable example being the Pulsed NTR.

Dr. Chapline's design use thin carbon filaments coated with fission fuel (coating is about 2 micrometers thick). The filaments radiated out from a central hub, looking like a fuzzy vinyl LP record. These revolving disks were spun at high speed (1 km/sec) through a reactor core, where atoms of nuclear fuel would undergo fission. The fission fragments would be directed by magnetic fields into an exhaust beam.

The drawback of this design is that too many of the fragments fail to escape the fuel coat (which adds no thrust but does heat up the coat) and too many hit the carbon filaments (which adds no thrust but does heat up the filaments). This is why the disks spin at high speed, otherwise they'd melt.

Dusty Plasma
Thrust22 N
Thrust Power0.2 GW
Mass Flow1.00e-06 kg/s
Specific Power55 kg/MW
Thrust344 N
Thrust Power2.6 GW
Mass Flow2.30e-05 kg/s
Specific Power3 kg/MW
Exhaust Velocity15,000,000 m/s
Specific Impulse1,529,052 s
Total Engine Mass9,000 kg
Uranium 235
ReactorGas Core
MHD Choke
Remass AccelFission-Fragment
Thrust DirectorMagnetic Nozzle

Rodney Clark and Robert Sheldon solve the problem with their Dusty plasma bed reactor (report).

You take the fission fuel and grind it into dust grains with an average size of 100 nanometers (that is, about 1/20th the thickness of the fuel coating in dr. Chapline's design). This does two things [A] most of the fragments escape and [B] the dust particles have such a high surface to volume ratio that heat (caused by fragments which fail to escape) readily dissipates, preventing the dust particles from melting.

The dust is suspended in the center of a reaction chamber whose walls are composed of a nuclear moderator. Power reactors will use beryllium oxide (BeO) as a moderator, but that is a bit massive for a spacecraft. The ship will probably use lithium hydride (LiH) for a moderator instead, since is only has one-quarter the mass. Probably about six metric tons worth. The dust is suspended electrostatically or magnetically by a containment field generator. The dust is heated up by radio frequency (RF) induction coils. The containment field generator will require superconductors, which will probably require a coolant system of its own.

The dust particles are slow and are relatively massive, while the fission fragments are fast and not very massive at all. So the magnetic field can be tailored so it holds the dust but allows the fission fragments to escape. Magnetic mirrors ensure that fragments headed the wrong way are re-directed to the exhaust port.

One valuable trick is that you can use the same unit for thrust or to generate electricity. Configure the magnetic field so that the fragments escape "downward" through the exhaust port and you have thrust. Flip a switch to change the magnetic field so that the fragments escape upward into deceleration and ion collection electrodes and you generate electricity. As a matter of fact, it is so efficient at generating electricity that researchers are busy trying to adapt this for ground based power plants. But I digress.

The dust is only sufficient for a short period of critical nuclear reaction so it must be continuously replenished. The thermal energy released by fission events plus heat from collisions between fission fragments and dust grains create intense heat within the dust cloud. Since there is no core cooling flow, the reactor power is limited to the temperature at which the dust can radiatively cool itself without vaporizing. The interior of the reaction chamber walls will protected by a mirrored (95% reflection) heat shield attached to a heat radiator. The outer moderator layer will have its own heat shield.

Clark and Sheldon roughed out a propulsion system. It had six tons for the moderator, 2 tons for radiators and liquid metal cooling, 1 ton for magnets, power recovery, and coils, for a grand total of 9 tons. The reaction chamber will be about 1 meter in diameter and 10 meters long. The moderator blanket around the chamber will be about 40 centimeters thick. The thrust is a function the size of the cloud of fissioning dust, and is directly related to the power level of the reactor. There is a limit to the maximum allowed power level, set by the coolant system of the reaction chamber. Clark and Sheldon estimate that only about 46% of the fission fragments provide thrust while the rest are wasted. See the report for details.

In the table, the 550AU engine is for a ten year journey to the Solar gravitational lensing point at 550 astronomical units (so you can use the sun as a giant telescope lens). The 0.5LY engine is for a thirty year trip to the Oort cloud of comets. These are constant acceleration brachistochrone trajectories, the 550AU mission will need a reactor power level of 350 MW and the 0.5LY mission will need 5.6 GW. Don't forget that the engine power is only 46% efficient, that's why the table thrust values are lower.

Werka FFRE
First Generation
Exhaust Velocity5,170,000 m/s
Specific Impulse527,013 s
Thrust43 N
Thrust Power0.1 GW
Mass Flow8.00e-06 kg/s
Total Engine Mass113,400 kg
Plutonium 239
ReactorGas Core
MHD Choke
Remass AccelFission-Fragment
Thrust DirectorMagnetic Nozzle
Specific Power1,020 kg/MW
Propulsion SystemWerka FFRE
Wet Mass303,000 kg
Dry Mass295,000 kg
Mass Ratio1.03
ΔV138,336 m/s

Robert Werka has a more modest and realistic design for his fission fragment rocket engine (FFRE). He figures that a practical design will have an exhaust velocity of about 5,200,000 m/s instead of his estimated theoretical maximum of 15,000,000 m/s. His lower estimate is still around 1.7% the speed of light so we are still talking about sub 2.0 mass ratios. Collisions between fission fragments and the dust particles is responsible for the reduction in exhaust velocity.

Incidentally the near relativistic exhaust velocity reduces radioactive contamination of the solar system. The particles are traveling well above the solar escape velocity (actually they are even faster than the galactic escape velocity) so all the radioactive exhaust goes shooting out of the solar system at 0.017c.

The dusty fuel is nanometer sized particles of slightly critical plutonium carbide, suspended and contained in an electric field. A moderator of deuterated polyethylene reflects enough neutrons to keep the plutonium critical, while control rods adjust the reaction levels. The moderator is protected from reaction chamber heat by a heat shield, an inner layer composed of carbon-carbon to reflect infrared radiation back into the core. The heat shield coolant passes through a Brayton cycle power generator to create some electricty, then the coolant is sent to the heat radiator.

The details of Werka's initial generation FFRE can be found in the diagram below. The reaction chamber is about 5.4 meters in diameter by 2.8 meters long. The magnetic nozzle brings the length to 11.5 meters. The fuel is uranium dioxide dust which melts at 3000 K, allowing a reactor power of 1.0 GW. It consume about 29 grams of uranium dioxide dust per hour (not per second). Of the 1.0 GW of reactor power, about 0.7 GW of that is dumped as waste heat through the very large radiators required.

The second most massive component is the magnetic mirror at the "top" of the reaction chamber. Its purpose is to reflect the fission fragments going the wrong way so they turn around and travel out the exhaust nozzle. Surrounding the "sides" of the reaction chamber is the collimating magnet which directs any remaining wrong-way fragments towards the exhaust nozzle. The exhaust beam would cause near-instantaneous erosion of any material object (since it is electrically charged, relativistic, radioactive grit). It is kept in bounds and electrically neutralized by the magnetic nozzle cage.


(ed note: this is from a Russian website, so please forgive the stilted translation)

Modern chemical and nuclear jet-propulsion engines (CJPE and NJPE respectively) have a relatively small velocity of combustion products ( heated work body, in NJPE) discharged from the engine nozzle, and may therefore be used only for flights in space or for launching automatic stations to planets of the Solar system to receive information from them, which may take many years.

A new jet-propulsion engine is critically needed for effective mastering of interplanetary space when the time of flight to any planet and back to the Earth does not exceed one year. The proposed NJPE equipped with an installation producing nuclear fuel (NF) for its operation can be a version of such an engine.

Both the engine itself and the NF-producing installation are based on the new principle of interaction of neutrons with the substance and of using the effect of involving thermonuclear neutrons in the fission reaction, which allows for an increase in the speed of nuclear reactions of ten to a hundred thousand times and thus, correspondingly, for a reduction in density of substances engaged in the reactions.

The NJPE developed on the basis of these principles has the following characteristics:

  • the chain nuclear fission reaction is carried out in a deep vacuum, and high-energy products of nuclear reactions may therefore leave the active zone (AZ) and the engine nozzle with practically no loss of its kinetic energy at the speed of about 10,000 km/s and create a jet thrust; the proposed engine is much less loaded by its thermal mode than modern NJPE because less than 5% of the nuclear energy released remains in its AZ whereas in present-day NJPE virtually all 100 percent remains in the AZ. Its power may therefore reach tens and hundreds of GWts.

  • nuclear fuel for engine operation, except for the time needed for bringing all its systems to full capacity, will be produced aboard the spacecraft from natural uranium. Thus, a certain quantity of safe compact metallic uranium will be in the missile instead of massive and explosion- hazardous fuel tanks.

  • the NJPE will operate during the time of the entire flight, ensuring practical elimination of weightlessness and the maintenance of comfortable conditions for astronauts, while the average flight speed may reach up to 1000 km/s.

  • since fission products leave the engine nozzle at a high speed, the radiation background when the engine is in operation both in the vicinity of the spacecraft and in interplanetary space will not increase because these products will escape the limits of the Solar system in less than 10 days.

  • potential of the engine allows in principle for bringing an interplanetary spacecraft to any planet of the Solar system in less than 20 days or for boasting an interstellar probe at 1/2 light speed.

Since the design is a fundamental development, it requires considerable capital and human resources and is not of commercial interest at present, its implementation may require an international program.

Afterburner Fission Fragment

Engine Mass
107,000 kg
Engine Mass
(mod oil)
91,000 kg
Engine Mass
268,961 kg
Reactor Power2.5 GW
Thrust4,651 N
Thrust Power730 MW
32,000 sec
313,900 m/s
Mass Flow
3.12×10-5 kg/s
Mass Flow
0.0179 kg/s
Mass Flow
0.018 kg/s

Robert Werka has apparently figured out a new configuration for his fission-fragment rocket engine (FFRE). The report is here.

As with most engines that have high specific impulse and exhaust velocity, the thrust of a FFRE is pathetically small. Ah, but there is a standard way of dealing with this problem: shifting gears. What you do is inject cold propellant into the exhaust ("afterburner"). The fission fragment exhaust loses energy while the cold propellant gains energy. The combined exhaust velocity of the fission fragment + propellant energy is lower than the original pure fission fragment, so the specific impulse goes down. However the propellant mass flow goes up since the combined exhaust has more mass than the original pure fission fragment. So the thrust goes up.

Now you have an Afterburner fission-fragment rocket engine (AFFRE).

As you are probably tired of hearing, this means the engine has shifted gears by trading specific impulse for thrust.

Shifting Gears
FFRE527,000 sec43 Newtons
AFFRE32,000 sec4,651 Newtons

The heart of the engine is a standard "dusty plasma" fission fragment engine. A cloud of nanoparticle-sized fission fuel is held in an electrostatic field inside a neutron moderator. Atoms in the particles are fissioning like crazy, spewing high velocity fission products in all directions. These become the exhaust, directed by a magnetic nozzle.

The AFFRE alters this a bit. Instead of a cylindrical reactor core it uses half a torus. Each end of the torus has its own magnetic nozzle. But the biggest difference is that cold hydrogen propellant is injected into the flow of fission fragments as an afterburner, in order to shift gears.

In the diagram above, the magnetic nozzles are the two frameworks perched on top of the reactor core. It is a converging-diverging (C-D) magnetic nozzle composed of a series of four beryllium magnetic rings (colored gold in the diagram). Note how each frame holding the beryllium rings is shaped like an elongated hour-glass, that is the converting-diverging part. The fission fragment plume emerges from the reactor core, is squeezed (converges) down until it reaches the midpoint of the magnetic nozzle, then expands (diverges) as it approaches the end of the nozzle. At the midpoint is the afterburner, where the cold hydrogen propellant is injected.

The semi-torus has a major and minor radius of 3 meters. The overall length of the engine is 13 meters. The reactor uses 91 metric tons of hydrocarbon oil as a moderator. This means the heavy lift vehicle can launch the engine "dry" with no oil moderator. In orbit the oil moderator can be easily injected into the reactor, at least easier than building the blasted thing in free fall out of graphite bricks.

Fission Sail

Fission Sail

Antimatter-Driven Sail

Antimatter Sail to 240 AU
TripTerra⇒Kuiper Belt
Trip Distance240 AU
Trip Time10 years
Mission ΔV116,800 m/s
Nat15,000 atoms/fission
Exhaust Velocity6.7×104 m/s
Isp6,800 sec
Thrust0.81 N
Power Plant
Electrical output400 We
Specific Mass16 kg/kW
Mass Schedule
Carbon Sail0.7 kg
Power Plant6.4 kg
Antihydrogen30.45 milligram
Antmatter Storage9 kg
Inert Mass16 kg
Payload10 kg
Dry Mass26 kg
Uranium fuel109 kg
Wet Mass135 kg
Initial Accel 0.006 m/s2

This is from NIAC Phase I Progress Report Antimatter Driven Sail for Deep Space Missions (2004)

They initially wanted to design a spacecraft that could transport a few kilograms of instrument package to another star system. Since that was a daunting goal, they decided to do baby-steps first and design for the less demanding mission of sending a payload 250 AUs to the Kuiper Belt in ten years (preferably braking to a halt but fly-by is acceptable). Even this is way beyond the current state of the art.

Their final design can send a 10 kilogram instrument package 250 AUs in 10 years using 30 milligrams of antimatter. That is a lot of antimatter, but the US could produce it in as little as 40 years with current particle accelerators. The average velocity was figured to be 116.8 km/s.

The design could send 10 kg of instruments to Alpha Centauri in 40 years using 17 grams of antimatter. A previous JPL study had concluded that kilograms of antimatter would be needed, the sail design uses only a fraction of that. Of course 17 gm of antimatter would take about 23,000 years to produce with current particle accelerators, more efficient antimatter production methods would be needed.

The sail is made of graphite and carbon-carbon fiber, infused with a tiny amount of uranium. It is subjected to a misting of antiprotons. These induce uranium atoms to fission, with the recoil pushing the sail. Since this is nuclear powered, the sail does not have to be kilometers in diameter, five meters will do.

The critical factors are:

  • The momentum transfered to the sail from the antimatter-induced fission has to be maximized (if the fission does not give the sail enough thrust you'll need outrageous amounts of antimatter)
  • Technology needs to be developed to store anti-hydrogen micro-pellets within solid-state integrated circuits (antimatter+matter goes boom! Antimatter containment system cannot weigh tons, it has to be very low mass)
  • Technology needs to be developed to create a high specific-mass electrical-power supply based on antimatter fission conversion (power supply cannot weigh tons either. Neither can the fuel supply)

The critical factor is "Nat", or number of uranium atoms ejected per fission. Nat has a remarkably strong impact on the number of antiprotons required for the mission.


The carbon-carbon sail contains uranium. A uranium atom undergoes fission when impacted by an antiproton. If only two fission fragments are released, then the momentum is determined by the velocity and mass of one of the fragments.

The fission fragments vary in mass, but they can be approximated by assuming the average fission product is palladium-111, mass is 1.8×10-25 kg/atom. The fission releases 190 MeV, which would give the palladium-111 atom an exhaust velocity of 1.39×107 m/s. This corresponds to a specific impulse of around 1.4 million seconds (i.e., dividing by 9.81).

Now, if neutral atoms of uranium are blown off with each fission event in addition to the palladium-111 fragments, the energy is distributed to all the ejected particles. Increasing the number of neutral uranium atoms (increasing Nat) will increase the thrust on the sail, at the cost of reducing the specific impulse. Yes, it's our old friend "shifting gears" again.

This is important. If you have a fixed-thrust rocket engine, and a given deltaV for the mission, you can calculate the optimal exhaust velocity and specific impulse. For our 116.8 km/s Kuiper mission this come out to about an exhaust velocity of 6.7×104 m/s (Isp of 6,800 seconds).

So by controlling Nat, one can control the exhaust velocity.

Nat is controlled by altering the velocity of the antiprotons, in other words the antiproton beam energy. The faster they go, the deeper they penetrate the uranium layer, and the more uranium atoms are ejected by each fission event. This is done by electrostatically biasing the sail relative to the antihydrogen container. This can easily raise the antiproton beam energy to 100 keV, which will allow them to penetrate the uranium to a depth of 355 nanometers below the surface.

As Nat increases, so does the consumption of uranium fuel. For the Kuiper belt mission, the uranium layer will have to have a thickness of 293 microns.

The researchers did a parametric study using Nat (number of atoms ejected per fission) as a free parameters.

Figure 1 shows the specific impulse as a function of Nat. We have assumed that the energy released in fission is equally distributed among the atoms ejected.

Figure 2 shows the dependence of the mass of antimatter required to perform the mission. The figure clearly depicts a minimum for Nat equal to around 15,000. This corresponds to a specific impulse of almost 7500 sec.

Figure 3 shows the masses of the uranium fuel and the spacecraft as a function of Nat. For Nat equal to 1, the specific impulse is over a million seconds and the fuel mass is small compared to the ship mass. As the specific impulse gets below 10,000 s, the mass of the fuel begins to dominate the ship mass.

For the Kuiper mission, the optimum value for Nat, i.e. the value where the number of required antihydrogen atoms was a minimum, occurred at Nat = 15,000 atoms per fission. So this value was used for the system studies.

The sail is 5 meters in diameter. It composed of a carbon-carbon layer 15 microns thick (34 g/m2), coated with a uranium fuel layer 293 microns thick. The total mass of uranium fuel is 109 kilograms.

Since the sail is so thin it does not need any heat radiators. It can dissipate the waste heat passively by black-body emmission. Assuming an emissivity of 0.3 for the carbon and uranium, the steady state temperature would be 570° C, well below uranium's melting point.

Antihydrogen storage unit

The antimatter storage is held 12 meters away from the sail by four tethers.

The storage is an array of small chips that resemble integrated circuit chips but are not. They are a series of tunnels etched in a silicon subtrate. Each tunnel is a sequence of electrodes. Each pair of electrodes holds a tiny pellet of antimatter fuel, in the form of solid antihydrogen. Each pellet has about 1015 antihydrogen atoms, charged to about 10-11 coulombs.

Each tunnel has 67 electrode pairs holding 67 antihydrogen pellets. Each 4 cm long chip has 100 tunnels, so each chip contains 1.6×1019 antihydrogen atoms. There are roughly 2,000 chips in the storage unit. Total number of antihydrogen atoms is 1.8×1022 or 30.45 milligrams. The entire storage unit has a mass of about 9 kilograms.

As pellets are ejected toward the sail, the remaining pellets in a tunnel are shuffled towards the sail-end like a bucket brigade.

My back of the envelope calculation says that if the entire storage array failed the resulting blast will be roughly the same as a 93 kiloton nuclear bomb, a little less than a W76 thermonuclear warhead.


The spacecraft requires electricity. Since a nuclear reactor or RTG is too heavy, and solar power is pretty worthess when you get further from the Sun than the asteroid belt, something new is needed. Since the ship is already carrying the most powerful fuel in the universe, the study authors tried to design an Antimatter Fission Conversion (AFC) power plant.

Some of the antimatter storage tunnels point rearward at the AFC uranium cone. The antimatter causes uranium atoms to fission. The scintillator converts the moving fission fragments into light (finding a scintillator material with the proper properties will be difficult). The photovoltaic cells convert the light into electricity. The AFC has a liquid lithium heat exchanger that removes the waste heat to a beryllium heat radiator, otherwise the blasted AFC would melt.

The spacecraft power requirements were estimated as 400 watts (same as the Voyager spacecraft). Voyager was powered by three MHW-RTG with a combined mass of 113.1 kg. The AFC will have a mass of only 6.4 kg, specific mass of 16 kg/kW (6.4 / 0.4).

The over-all efficiency of the AFC was estimated to be about 4.4%, so 2×1014 antiprotons per second are needed for 400 W of electrical power. The power is generated on demand when communications back to Terra are indicated. The beryllium heat radiator is in two dametrically opposed sheets, edge-on to the sail. They have a radiating surface area of 3.5 m2 and a temperature of 620° C.


The researchers studied the effect of increasing the spacecraft's dry mass (power plant, structure, payload, etc). The result is in Figure 7. Mpbar (mg) is the mass of antimatter fuel in milligrams. Mu (kg) is the mass of uranium fuel in kilograms. Mship (kg) is the wet mass of the spacecraft in kilograms. Delta dry mass (kg) scale is the increase in the ship's dry mass (baseline ship has a dry mass of 26 kg). Mass scale is in kilograms or milligrams depending upon the weight unit in parenthesis.

So if the dry mass is doubled (delta dry mass = 26 kg, total dry mass = 52 kg) then mass of ship only rises to Mship = 280 kg and the amount of antimatter fuel rises to Mpbar = 60 mg.

The baseline ship is where delta dry mass = 0 kg. Mpbar = 30.45 milligrams, Mu = 109 kilograms, Mship = 135 kg.

The minimum acceleration level was decided to be 0.006 m/s2. If my slide rule is not lying to me, that implies a thrust of 0.81 Newtons.


This is from LASL nuclear rocket propulsion program (1956) and Propulsion Systems for Space Flight by William R. Corliss (1960). It is called a "consumable nuclear rocket", a "Fizzer", a "Fizzing Bomb", or a "Burning Wall" rocket.

This is totally insane. Thank the stars it was never developed. This is sort of a mash-up of a solid-core NTR, a gas-core NTR, a chemical solid rocket, and atomic Primacord. Think of it as a giant nuclear-powered sparkler from hell. It is from those innocent days when the rocket designers wouldn't recognize a bad idea even if it they tripped over it.

Note that the propellant is lithium hydride, presumably a convenient way to hold hyrogen since cryogenic tanks need refrigeration equipment and all sorts of extra stuff.

The propellant is NOT lithium deuteride. There is another name for a fission reaction next to a slab of lithium deuteride, it is "thermonuclear weapon." Lithium hydride creates thrust. Lithium deuteride will just go off like an H-bomb, vaporizing the payload and anything else nearby.


There are three basic types of nuclear rockets. They may be classified by the manner in which the nuclear reaction is contained (Table 5-1).

The so—called “heat-transfer nuclear rocket” is no more than a nuclear-reactor core through which propellant is passed and heated. Figure 5-14 illustrates this type schematically. The second type is the consumable nuclear rocket. Here the nuclear fissioning occurs directly in the working fluid. Both fission fuel and fission products are released in the exhaust. This type is conceptually midway between the heat transfer reactor and an atomic bomb.

An example of the solid type is shown in Fig. 5-21. The U—235 fuel is concentrated in a rod running the length of the rocket. A good neutron absorber, such as cadmium, surrounds the fuel. Around this there is a thick layer of propellant, perhaps lithium hydride. To fire the rocket, a section of the cadmium sheath is pulled off the bottom of the fuel cylinder causing the neutron chain reaction to begin. In the region where cadmium is absent, the nuclear reaction will flash the solid materials into a high temperature vapor. The anisotropic expansion of the vapor will produce a forward thrust on the rocket. The average molecular weight of the vapor will be dictated by the relative proportions of uranium and propellant in a given cross section. Usually the ratio of masses will be about 100 kg of propellant for each kilogram of fuel. Once the reaction begins, it will proceed up the body of the rocket as the neutron temperature rises high enough to make the cadmium an ineffectual neutron absorber. Unfortunately, no way has yet been found to prevent the reaction from traveling at velocities exceeding 100 m/sec. The consequence of this fact is that inordinately long rockets would be needed to obtain reasonable accelerations and burn times. For manned systems, it is desirable to keep accelerations below 10 g. Burn times between 10 and 100 sec are satisfactory for many launching missions. If this fundamental problem of slowing down the reaction can be solved, the solid, consumable nuclear rocket is simple and attractive in operation.


Another class of rockets is generally typified by the fizzling bomb concept. At a sufficiently high rate of reaction, a substantial energy release can be achieved by an explosive fission reaction in which the recoiling mass of the reactor itself furnishes the impulse to drive the vehicle. Highly moderated reactors should be used, both for low critical mass and to provide inexpensive material to be ejected. Possible methods lunge all of the way from a single shot, through multiple explosions, to a continuous reaction analogous to a solid propellant chemical rocket. All such schemes are characterized by the incompatibility of reasonable accelerations and economical use of active materiel. Roughly speaking, the time of an explosive nuclear reaction is equal to the shock wave transit time across the reacting zone. Unless some ingenious cushion is built in, this time also characterizes the impulse given to the missile. If this impulse is to give the missile a velocity increment of several thousand feet per second, the resultant accelerations are fairly fantastic.

In principle, a slower reaction with reasonable fissionable material economy could be achieved with a gaseous reactor that retains preferentially the fissionable fuel, but no schemes yet proposed seem workable.

Thermal and fast neutron reactors, separated fission products, direct use of fission fragments to heat gas, self-heating mixtures of moderator and fuel (both thrown away), thrust from fission fragment momentum, alpha particle recoil, and electron and ion accelerators were considered for vehicle propulsion systems.

The LASL, UCRL and Lockheed discussions generally pertained to the use of unconventional means for the production of thrust, such as fissioning gaseous systems, burning "cigarettes" or internally burning reactors, or the use of radiation from bomb bursts to heat a working fluid (BATO) . Martin proposed to obtain thrust from thermonuclear processes achieved in a transient (pulsed) shock wave heated system. In general, none of these imconventional methods offered any real hope of successful achievement in the near future.

UCRL reported it has concluded the fizzling and gaseous reactors seem to have one order of magnitude of impossibility inherent in their makeup and that they therefore do not appear promising for application to rocket use.


This section has been moved here

Antimatter Energy

This section has been moved here

Radiation Shielding

This section has been moved here

Solid Core

This section has been moved here

Gas Core

This section has been moved here

Plasma Core

This section has been moved here

Antimatter Bottle

This section has been moved here

Beam Core

This section has been moved here

Positron Ablative

This section has been moved here


This section has been moved here

Fusion Fuels

This section has been moved here


This section has been moved here


This section has been moved here


This section has been moved here


This section has been moved here

Inertial Confinement

This section has been moved here

Electrostatic Inertial

This section has been moved here

Magnetic Confinement

This section has been moved here

Linear Fusion

This section has been moved here

Toroidal Fusion

This section has been moved here

Direct Fusion Drive

This section has been moved here

Flow-Stabilized Z-Pinch

This section has been moved here

Magnetio Inertial Confinement

This section has been moved here

Fusion Engines

This section has been moved here

Pure Fusion Engines

This section has been moved here

Afterburner Fusion Engines

This section has been moved here

Dual-Mode Fusion Engines

This section has been moved here

Breeder Fusion Engine

This section has been moved here

Nuclear Magnetic Spin Alignment

This section has been moved here

( STARFIRE Fusion Afterburner )

This section has been moved here

( AV:T Fusion )

This section has been moved here

( THS Fusion Pulse )

This section has been moved here

( Epstein Drive )

This section has been moved here


This section has been moved here


This section has been moved here

Orion Thrust and Isp

This section has been moved here

Orion Environmental Impact

This section has been moved here

Atomic Manhole Cover

This section has been moved here


This section has been moved here

Zeta-Pinch Fission

Mini-Mag Orion

This section has been moved here

Z-pinch Microfission

This section has been moved here

n-Li6 Microfission

This section has been moved here

Zeta-Pinch Fusion

This section has been moved here

PuFF Pulsed Fission Fusion

This section has been moved here


This section has been moved here

Inertial Confinement

This section has been moved here

Magneto Inertial Fusion

This section has been moved here

Thio MTF

This section has been moved here


This section has been moved here

Slough FDR

This section has been moved here

Plasma Jet MIF

This section has been moved here

Antimatter Bottle

This section has been moved here

Antimatter catalyzed

This section has been moved here


This section has been moved here


This section has been moved here

Radioisotope Positron

This section has been moved here

Atomic Rockets notices

This week's featured addition is SPIN POLARIZATION FOR FUSION PROPULSION

This week's featured addition is INsTAR

This week's featured addition is NTR ALTERNATIVES TO LIQUID HYDROGEN

Atomic Rockets

Support Atomic Rockets

Support Atomic Rockets on Patreon