Engine Room


What's in the engine room? The control console for the reactor and the control console for the propellant pumps. Plus the controls to the remote waldoes/robot who fix things in the radioactive section. Damage control equipment and related items.

Remember, outside the engine room hatch will be a decontamination booth. And I'm sure over the hatch will be mounted an alarm with a red rotating light, so you don't have to put your ear on the bulkhead to hear Astro say "Oh SH*****T!!!". Past the hatch will be a short corridor, with a dog-leg bend in it, so you can get in but radiation cannot get out (radiation has to travel in straight lines, but crewmen can zig-zag). Be sure you are wearing your dosimeter.

The shadow shield will be in the floor, with the engine(s) below that.

Around the engine deck will be auxiliary propellant tanks. The main tanks provide propellant, which is also used to cool the reactor and keep it from melting. In case something happens to the main tanks, the auxiliary tanks give Astro a few precious seconds of coolant time so he can scram the reactor. The engine deck will also have some kind of Geiger counter to warn the crew of a radiation leak. There will also be controls for the ship's power plant, whether the power is tapped from the propulsion system or from a separate unit.

Emergency Buttons

There will also be controls for the flow rate of propellant, and pyrometer (temperature) gauges with helpful red bands labeled "DANGER - MELTDOWN IMMINENT!". There will be a large red "discoverer" button which activates the "there's been a nuclear oopsie" alarms, right next to the even bigger button marked "SCRAM", which is the emergency reactor shutdown.

But there will be one even bigger red button, marked "PANIC!". This one activates the explosive bolts and JATO units that jettison the reactor out the stern of the spacecraft, hopefully removing the spacecraft from the lethal radius. The reactor isn't going to explode like a nuclear weapon, it will just act like a midget neutron bomb, utterly destroying all life on the ship except for the cockroaches.

Atomic Rocket Engineer Control Console

This amusing example of 1960's style user interface design is from NUCLEAR SPACE PROPULSION by Holmes F. Crouch (1965). This complements the Pilot's console from the same book. This design assumes that it is for a solid-core nuclear thermal rocket. Mr. Crouch decided that controlling the rocket's trajectory while simultaneously juggling the power levels of the nuclear reactor was a little too much to ask of a single human being, so he split it into two jobs. Each subsystem has too many displays, control functions, and automatic interlocks.

According to Mr. Crouch, there are four independent subsystems involved with flying a nuclear thermal rocket:

  1. Thrust vectoring (Engine exhaust nozzle)
  2. Spacecraft orientation and stability (Attitude jets)
  3. Heat generation for specific impulse (Reactor)
  4. Propellant flow (Turbopump)

The pilot will be controlling the thrust vectoring and spacecraft orientation, the engineer will be controlling heat generation and propellant flow. So the pilot is flying the rocket, while the engineer is flying the reactor and turbopump.

Front and center is the Engine Scanner. This would display the outputs of the various closed-circuit TV scanners, infared mappers, microphone pickups, lights, and periscopes built into the propulsion system to allow the engineer to keep tabs on everything. In addition, multichannel recorders would provide visual displays of engine data, calibration curves, and nuclear radiation profiles. The data plotters would allow the engineer to spot any dangerous trends. The engineer would be particularly interested in displaying computations of fission product inventory and the changing center-of-mass due to propellant sloshing and consumption.

Below the scanner is the thrust demand indicator. This displays the thrust mode order given by the pilot via their thrust mode selector. Again, this is basically a glorified Engine Order Telegraph from the age of steam. The pilot uses it to tell the engineer what sort of thrust is required. It is then the engineer's job to juggle the reactor control rod and the propellant turbines to produce what is requested.

The engineer has two control levers. The right directs the reactor, the left directs the turbopump. These are one-axis levers with stepped notches for shim settings and vernier knobs for trim settings (i.e., a dial on the top of the lever for fine tuning). What it boils down to is that the right lever controls the rocket's exhaust velocity, and the left lever controls the rocket's propellent mass flow. The constraint is that the engineer cannot set the two levers to values that will cause the reactor to melt or the turbopump to explode. Or "flood the engine", that is, feed so much propellant that the reactor can't heat it all so the exhaust is lukewarm.

More on Engineering Tasks, and Math

I'm about to go into more detail about the engineer's task, with equations and everything. If you feel your eyes starting to glaze over, skip past this section.

So, say the pilot calls for an acceleration of 0.1 meter per second (about 0.01 g). By keeping careful tabs on everything that leaves the ship (which is mostly a matter of tracking propellant mass expended), the engineer knows the ship's current mass.

F = A * Mi


  • F = Thrust (Newtons)
  • A = ship Acceleration (m/s) {divide by 9.81 for Gs}
  • Mc = ship's Current Mass (kg)

If the good ship Polaris has its tanks topped off, it has a current mass of 181,000 kilograms. To have an acceleration of 0.1 m/s will require a thrust of 0.1 * 181,000 = 18,100 Newtons. Now, to generate those Newtons:

F = mDot * Ve


  • F = Thrust (Newtons)
  • mDot = Propellant Mass Flow (kg/s) { left lever }
  • Ve = Exhaust Velocity (m/s) { right lever }

So the engineer has to set the turbopump (propellant mass flow) and reactor temperature (exhaust velocity) such that the two produce a thrust of 18100 Newtons. With the added constraint that there is an upper limit on what level of propellant mass flow the turbopump can put out (before it explosively delaminates) and an upper limit on the reactor temperature (before it either melts or has a criticality accident). There is also a limit on how many kilograms of propellant that the reactor can heat up to full temeperature in one second. Try to feed more and the exhaust will be lukewarm, causing the thrust to suffer.

Ve is proportional to the square root of the reactor temperature. This means that to double the exhaust velocity, you have to raise the reactor temperature four times. The approximate equation is:

Qe = (Ve / (Z * 129))2 * Pw


  • Qe = engine reaction chamber temperature (Kelvin)
  • Ve = exhaust velocity (m/s)
  • Z = heat-pressure factor, varies by engine design, roughly from 1.4 to 2.4 or so.
  • Pw = mean molecular weight of propellant, 1 for atomic hydrogen, 2 for molecular hydrogen

I have some very shakey figures that I calculated myself (and are therefore suspect) that suggest a NERVA/Dumbo style solid core have a maximum temperature of around 2,300° K, a Cermet core up to 2,800° K, and a pebble bed up to around 3,000° K. I saw another figure that implied a theoretical maximum of about 4,700° K, but I don't trust that figure.

So if the engineer wanted to keep the engine at a conservative 2,000° K, and the engine had a heat-pressure factor of 1.9, that would produce an exhaust velocity of

Qe = (Ve / (Z * 129))2 * Pw
solve for Ve
Ve = sqrt(Qe / Pw) * (Z * 129)
Ve = sqrt(2000 / 2) * (1.9 * 129)
Ve = 30 * 245.1
Ve = 7,400 m/s

With that exhaust velocity, to get 18,100 Newtons you'll have to set the turbopumps to do 18,100 / 7,400 = 2.5 kg/s.

As mentioned before, you generally don't care what the ship's acceleration is, except during lift-off and landing. But it is also important if you are in a combat situation, or docking, or otherwise have to change the ship's velocity in a hurry.

Associated with the right lever reactor controller, there would be readouts of reactor period, neutron flux level, control drum setting including vernier tube setting, net reactivity worth, rate of temperature change, and temperature levels.

Associated with the left lever turbopump controller, there would be readouts of pump speed, discharge pressure, propellant flow rate, amount of propellant onboard, core bypass valve positions, and quality of propellant going into the reactor core.

Finally would be the dire warning of the "risk-of-excursion" light, at the very top ("excursion" is a technical term for "nuclear reactor criticality accident" ). This alerts the engineer that the reactor is at risk of a criticality and they had better do something about it fast! The "reactor period" readout displays the dangerous decreasing reactor period. The smaller the reactor period, the more rapid the change in reactor power level. The neutron flux readout shows the current reactor power level. If it goes too high, the reactor emits a deadly burst of neutrons and everybody dies.


Reactor Dampers

Nuclear fission reactors are throttled by controlling the amount of neutrons available in the core. "Thermal neutron" reactors need slow moving neutrons, "Fast neutron" reactors use fast ones. Basically this means that a thermal neutron reactor has the fission fuel elements embedded in a "moderator", which turns fast neutrons into thermal neutrons. In both designs, the reactor is generally encased in a neutron reflector. This intercepts neutrons on their way out of the reactor and sends them back in, "kicking them back into play" so to speak. This lowers the amount of fission fuel you need in order to sustain a chain reaction.

The amount of free neutrons available is controlled by the dampers. In a NERVA engine, these are rods of cadmium or other neutron poison inserted or removed from the reactor to control the chain reaction. Make sure they are not warped. A non-automated set-up will have the dampers positioned by hand using a "multiplying vernier" and a "danger gauge". The gauge tells you how hot the reaction is. Verniers tell how far a given damper has been inserted into the reactor. Pull the dampers out slowly until the danger gauge tells you the reaction is at the desired level.

Side note: a vernier is what they used to use back in the ancient days before digital read-outs. Standard gauges have a pointer that runs along a scale. A vernier is a clever way to increase the accuracy of the scale.

In some NERVA designs, instead of rods of cadmium inserted into the body of the reactor, they instead have drums imbedded in the neutron reflector around the body of the reactor. One side of the drums are composed of neutron reflector, the other side is composed of neutron absorbers (AKA "neutron poison"). If the reflector side of the drums are facing the reactor, enough neutrons are reflected to sustain the chain reaction. If the absorber side is facing, it sucks up the neutrons and there are not enough neutrons to sustain the reaction. The control drums are rotated to partial positions in order to throttle the reaction to desired levels.

In both cases you want a damper fail safe, so if anything happens the failure modes will tend to make the dampers slam in to the full "quench reaction" position.

There will be damper safeties. When no acceleration is expected for some time, these lock the dampers in the full-quench position.

Thrust Runup Transients

This is the single most deadly problem in nuclear thermal rocket design. Among designers, more attention has been given to this subject, "Engine Dynamic Stability," than to any other aspect of thrust control. The following description is for a solid core reactor, and is adapted from NUCLEAR SPACE PROPULSION by Holmes F. Crouch (1965).

Say the engineer gets the command from the pilot to prepare for a burn. The engineer "preconditions" the turbopump and the reactor so they are able to generate thrust on command, they are in the state called "idling." The turbopump is idling with its discharge bypassed back to its suction side. The propellant lines into the reactor have been chilled down and bleed close-loop flow is maintained. The reactor is nuclearly critical but the core is "dry", it has no propellant in it.

The fun starts when you try to actually make this contraption generate thrust.

You are changing from a dry core to a wet core, one that has propellant flowing through it. The core is both very hot with heat and very hot with neutron radiation. And the liquid hydrogen propellant is both very cold (we are talking a freaking minus 252° Celsius here) and also a pretty good neutron moderator. So when the cold propellant hits the hot reactor two things happen, and both are rather drastic. There is a nuclear disturbance and a thermal disturbance. Both create transients oscillations that can destroy the engine if they are not controlled. See the top graph to the right.

The nuclear transient is due to the unfortunate fact that the liquid hydrogen propellant is a pretty good moderator. The dry reactor has the control drums set so the core has enough thermal neutrons to sustain a moderate chain reaction. The flood of propellant abruptly skyrockets the number of thermal neutrons, which abruptly skyrockets the chain reaction. The chain reaction "thermostat" automatically reacts to the sudden rise in neutron flux, and quickly rotates the control drums to cool off the reaction (in top diagram at right, at point "Control Drums Set Back").

This causes a flux undershoot as the drums over-correct (in top diagram, at point "Flux Undershoot").

The thermostat then over-corrects while attempting to bring the chain reaction up to the desired level, which causes a flux overshoot. If the engine designer has done their job correctly, these oscillations die down to a steady flux at the desired level. If they have not done their job correctly, the runaway oscillations increase until the reactor melts or has a criticality accident. The runaway nuclear oscillations would also make runaway thermal oscillations, compounding the disaster.

The thermal transient occurs because the initial flood of sub-zero propellant extracts heat from the idling reactor core faster than the heat can be resupplied. This lowers the core temperature. The temperature drops until the heat from the rising chain reaction increases enough to stop it (in top diagram at right, at point "Minimum Temp"). If the temperature oscillation is not controlled, they could create pressure oscillations in the propellant. These pressure oscillations can be transmitted upstream where they could damage the turbopump. And the pressure oscillations can affect the moderating characteristics of the propellant, which would affect the nuclear oscillation.

What makes this so deadly is the fact that the thermal oscillations can affect the nuclear oscillation and vice versa. If they start to amplify each other you are doomed.

There are equations for these transients, but most of them are differential equations so I'm not going to bore you with them. You can find them in Dr. Crouch's book.

A Kiwi nuclear thermal rocket test (1965). This was a deliberate destruction test. The safety experiment was designed to obtain basic reactor shutdown information for use in predicting the behavior of nuclear rocket reactors under a wide range of accident conditions. It detonated with a nuclear explosion yield equivalent to 2.1 tons of TNT, and reached a maximum temperature of 4,250° K, which vaporised 5% of the reactor core fuel rods, of which 68% was dispersed as fallout with a specific activity of 1015 fissions/gram for refractory nuclides like Zr-95.


Gimbals are one technique of thrust vectoring, allowing the thrust to go off center. The pilot uses this to yaw and pitch. If there are multiple engines with enough separation from the spacecraft's axis, gimbaling can be used roll as well.

It can also be used in emergencies if during thrust the rocket "falls off its tail." That is, if the center of gravity unexpectedly shifts, and the engines can gimbal by enough degrees off center, they can compensate.

In a 1969 US Atomic Energy Commission pamphlet on NERVA, it had this to say:

A critical structural problem arises at the junction between the engine and rocket body, however. Across this junction must be transmitted the entire engine thrust that is conveyed upwards from the nozzle through the engine exoskeleton. Ordinarily, such a joint would present no engineering difficulties. But in this case, the joint must be flexible. All big rockets have their engines mounted on gimbals that permit the "driver" to steer them.

Gimballing chemical engines is relatively easy because they are lightweight. But, in a nuclear rocket, the heavy reactor replace the empty combustion chamber of the chemical rocket. Despite the added weight, suitable flexible joints now have been designed.

Dr. Crouch is unsure about this. He notes that unlike a chemical rocket, the poor gimbal is trying to swing an entire reactor so the actuation loads will be large. Even worse: the gimbal is subject to cryogenic temperatures, nuclear radiation, and thrust loads. Simultaneously. He is of the opinion that a more optimal solution would be secondary injection, jetavators, or a movable nozzle.

Having said that, here are the gimbals designed for the NERVA project.

In the images here, the gimbals are the network of blue rods emerging from the bottom of a hydraulic piston. The wasp-waist in the center is the pivot. The green "mushroom caps" are the control-drum actuators, perched on top of the control drums protruding from the reactor.

Note that the gimbals are a nasty trouble spot. If one jams, that engine cannot be steered. The proximity to the dangerously radioactive engine makes repair difficult. This is a good place for a waldo.


Control of these large vehicles during powered flight is expected to use some means of thrust vector control, TVC. Thrust vector control allows the alignment of the vehicle thrust with vehicle center of gravity (CG) to maintain straight line flight or to induce vehicle steering as desired. The objective of this paper is to define the requirements and conceptual design of thrust vector control systems for NTR applications. Several different vehicles will be examined to explore vehicle and mission influences on TVC design.

A reaction control system (RCS) can also manage vehicle attitude, and is essential during times when engines are not thrusting. While the NTR engines are thrusting, the use of thrust vector control can avoid most use of RCS propellant. Otherwise, RCS propellant consumption could become excessive in some cases, especially off-nominal situations such as thrust asymmetry due to engine malfunction. In a multiengine vehicle, differential engine thrusting could be used to steer the vehicle, as an alternative to TVC. But this method would not work if one of three engines failed. Definition of the capabilities of a TVC system also helps define the requirements of the RCS.

Gimbaling of engines is the most common method of thrust vector control for large rockets.

Gimbal Bearing

Typical configurations includes a gimbal bearing attached to the engine that allows two axis rotation of the engine (e.g., yaw and pitch rotation) but prevents rotation about the axis of the engine. The gimbal bearing carries most of the engine thrust load.

Gimbal Ring

Two actuators, attached to the engine and vehicle structure can be mounted 90° apart to provide full two-axis gimbaling. Instead of a gimbal bearing on the engine axis, gimbal motion can be achieved by the use of a gimbal ring that is external to the engine, as in a gyroscope mount. One gimbal actuator would be attached to the ring, the other attaches to the engine, 90° apart.

Thrust Vector Trim

An alternative method of engine steering was considered during the Nuclear Engine for Rocket Vehicle Application (NERVA) program (Ref. 4). This method, designated Thrust Vector Trim (TVT) uses one engine attach point with a fixed ball and socket joint and two moveable engine attach points that translate parallel to the rocket engine axis. The fixed and moveable points are located 120° around the engine axis. By actuating the moveable points fore and aft, the engine thrust axis can be directed within a cone of operation. The advantage of this configuration is the elimination of the engine gimbal bearing; there is only the need for a flexible connection to carry propellant into the engine. The disadvantage of this configuration is that the mechanisms that provide the engine movement must each continuously carry about one third of the engine thrust load. The engine attach points must similarly be designed to react the same thrust loads. This method may be attractive for relatively low thrust engines. Mechanical engine steering concepts are shown in Figure 1.

Other thrust vector control methods include the use of paddles or vanes to alter the rocket engine exhaust to direct the net thrust vector. These methods are problematic because the thrust directing elements must withstand the high temperatures in the rocket exhaust.

Invisible Blast

The countdown was still at ten seconds when we were startled by a blast of light. For a moment, we wondered if Olympus had also met with catastrophe. Then we realized that someone was filming the take-off, and the external floodlights had been switched on.

During those last few seconds, I think we all forgot our own predicament; we were up there aboard Olympus, willing the thrust to build up smoothly and lift the ship out of the tiny gravitational field of Phobos, and then away from Mars for the long fall sunward. We heard Commander Richmond say "Ignition," there was a brief burst of interference, and the patch of light began to move in the field of the telescope.

That was all. There was no blazing column of fire, because, of course, there's really no ignition when a nuclear rocket lights up. "Lights up" indeed! That's another hangover from the old chemical technology. But a hot hydrogen blast is completely invisible; it seems a pity that we'll never again see anything so spectacular as a Saturn or a Korolov blast-off.

From "Transit of Earth" by Arthur C. Clarke (1971)


The Power Plant will probably be some species of space nuclear reactor. Clever designers will use the Bimodal NTR concept to get double duty from a single reactor.



The various controls, tongs, and remote control "waldoes" will reach around or penetrate the anti-radiation shadow shield, and there may be auxiliary lead baffles. Peeking around the baffles is how Rhysling lost his sight in Heinlein's "The Green Hills of Earth".

The various controls, tongs, and remote control "waldoes" will reach around or penetrate the anti-radiation shadow shield, and there may be auxiliary lead baffles. Peeking around the baffles is how Rhysling lost his sight in Heinlein's "The Green Hills of Earth".


Remember that the shadow shield will be in the floor, with the engine below that. Closed-circuit TV monitor will help Astro see what he is doing, but if they are damaged, he'll have to make do with mirrors and/or doing it by touch. What he really needs is one of Tom Swift Jr.'s Giant Robots, which were designed to do maintenance inside nuclear power plants. There is more about robots here.

For external repairs, the chief engineer might use something similar to the amazing Canadarm 2, which is currently on active duty on the International Space Station. Unlike the first Canadarm, this one is not attached at either end. Instead, either end can plug into special sockets ("power data grapple fixtures") built at strategic spots on the surface of the station. Canadarm 2 can literally walk on the surface of the station to where it is needed, moving end-over-end like a giant metal inch worm. The main limitation is that each "step" must end at a socket, but this is due to power and control signal issues. A more advanced version might be self contained enough to not require sockets, just hand-holds or other protrusions that it could grab.

Canadarm 2 is quite large, 17.6 meters (57.7 feet) long when fully extended. On your atomic rocket, one would use arm(s) long enough to reach any spot on the radioactive engine.


Outside but adjacent to the engine deck will be the maintenance shop and storage for replacement parts. On larger ships this might be the place for damage control central.

Rather than carry many extra tons of spare parts, which might or might not be used, the ship is equipped with extensive workshops to repair or manufacture the required parts as they become needed by the maintenance crews. Each workshop features a large number of automated machine tools and precision autofacs that hold the specifications of all parts in memory. Only the most sensitive and specialized parts, such as the microscopic computer nodes and chips, are held in storage.

The microgravity stations are sealed transparent bays with built-in gloves. Items to be repaired are slipped inside through a zippered opening, and can then be disassembled without fear of small parts flying off. Tools can also be left floating free inside the enclosure rather than having to be tethered. Machine tools are housed within similar enclosed bays for safety and tidiness reasons; articulated arms hold the part while it is being machined.

(ed note: I will observe that enclosures make sense from the standpoint of eye safety. You do NOT want to have a free-floating metal shaving getting into your eye. However, it is not a good idea to have one's tools untethered if unexpected spacecraft accelerations can occur.)

The machinery, decks, and walls will have lots of access panels to allow the engineers to make repairs. If the machinery is large, there will be ladders and catwalks. The lighting will also be bright and harsh. The walls will have tool bays, spare parts lockers, fire control gear, and similar items.

The engineers will also have lots of interesting things in their tool boxes.

The engineering deck is likely to be a noisy place, with all the heavy machinery. Engineers might need ear-plugs, and good lip-reading skills. If the situation is more high-tech, they will wear combination radio/hearing protection headsets.

The engineer has a logbook as well.

Atomic Rockets notices

This week's featured addition is Pulsed Solid Core Nuclear Thermal Rocket

This week's featured addition is Convair Electria spacecraft

This week's featured addition is the Lewis Research Center Mars Mission

Atomic Rockets

Support Atomic Rockets

Support Atomic Rockets on Patreon