NERVA

In principle, the design of a nuclear thermal rocket engine is quite simple: a turbopump would force hydrogen through a nuclear reactor that would heat it to very high temperatures. Complicating factors were immediately apparent. The first was that a means had to be found of controlling reactor temperature and power output. The second was that a means had to be devised to hold the propellant. The only practical means of storing hydrogen was in liquid form, and this required temperatures below 20 K (−253.2 °C). The third was that the hydrogen would be heated to a temperature of around 2,500 K (2,230 °C), and materials would be required that could both withstand such temperatures and resist corrosion by hydrogen.[16]

For the fuel, plutonium-239, uranium-235 and uranium-233 have been considered. Plutonium was rejected because while it forms compounds easily, it could not reach temperatures as high as those of uranium. Uranium-233, as compared to uranium-235, is slightly lighter, has a higher number of neutrons per fission event, and a has high probability of fission, but its radioactive properties make it more difficult to handle, and in any case it was not readily available.[17][18] As to structural materials in the reactor, the choice came down to graphite or metals.[17] Of the metals, tungsten emerged as the frontrunner, but tungsten was expensive, hard to fabricate, and had undesirable neutronic properties. To get around its neutronic properties, it was proposed to use tungsten-184, which does not absorb neutrons.[19] On the other hand, graphite was cheap, actually gets stronger at temperatures up to 3,300 K (3,030 °C), and sublimes rather than melts at 3,900 K (3,630 °C). Graphite was therefore chosen.[20]

To control the reactor, the core was surrounded by control drums coated with graphite or beryllium (a neutron moderator) on one side and boron (a neutron poison) on the other. The reactor's power output could be controlled by rotating the drums.[21] To increase thrust, it is sufficient to increase the flow of propellant. Hydrogen, whether in pure form or in a compound like ammonia, is an efficient nuclear moderator, and increasing the flow also increases the rate of reactions in the core. This increased reaction rate offsets the cooling provided by the hydrogen. Moreover, as the hydrogen heats up, it expands, so there is less in the core to remove heat, and the temperature will level off. These opposing effects stabilize the reactivity and a nuclear rocket engine is therefore naturally very stable, and the thrust is easily controlled by varying the hydrogen flow without changing the control drums.[22]

LASL produced a series of design concepts, each with its own codename: Uncle Tom, Uncle Tung, Bloodhound and Shish.[23] By 1955, it had settled on a 1,500 MW design called Old Black Joe. In 1956, this became the basis of 2,700 MW design intended to be the upper stage of an ICBM.[17]

Engine maintenance assembly and disassembly (E-MAD) facility

Nuclear reactors for Project Rover were built at LASL Technical Area 18 (TA-18), also known as the Pajarito Site. The reactors were tested at very low power before being shipped to Jackass Flats in the Nevada Test Site. Testing of fuel elements and other materials science was done by the LASL N Division at TA-46 using various ovens and later the Nuclear Furnace.[24]

Work commenced on test facilities at Jackass Flats in mid-1957. All materials and supplies had to be brought in from Las Vegas. Test Cell A consisted of a farm of hydrogen gas bottles and a concrete wall 1 meter (3 ft) thick to protect the electronic instrumentation from radiation produced by the reactor. The control room was located 3.2 kilometers (2 mi) away. The reactor was test fired with its plume in the air so that radioactive products could be safely dissipated.[17]

The reactor maintenance and disassembly building (R-MAD) was in most respects a typical hot cell used by the nuclear industry, with thick concrete walls, lead glass viewing windows, and remote manipulation arms. It was exceptional only for its size: 76 meters (250 ft) long, 43 meters (140 ft) and 19 meters (63 ft) high. This allowed the engine to be moved in and out on a railroad car.[17]

The "Jackass and Western Railroad", as it was light-heartedly described, was said to be the world's shortest and slowest railroad.[25] There were two locomotives, the remotely controlled electric L-1, and the diesel/electric L-2, which was manually controlled but had radiation shielding around the cab.[17] The former was normally used; the latter was provided as a backup.[26] Construction workers were housed in Mercury, Nevada. Later thirty trailers were brought to Jackass Flats to create a village named "Boyerville" after the supervisor, Keith Boyer. Construction work was completed in the fall of 1958.[17] NASA planned to develop a community of 2,700 people, with 800 dwellings and their own shopping complex by 1967.[27]

President John F. Kennedy (right) visits the Nuclear Rocket Development Station on 8 December 1962 with Harold Finger (left) and Glenn Seaborg (behind)

By 1957, the Atlas missile project was proceeding well, and the need for a nuclear upper stage had all but disappeared.[28] On 2 October 1957, the AEC proposed cutting its budget.[29] Two days later, the Soviet Union launched Sputnik 1, the first artificial satellite. This surprise success fired fears and imaginations around the world. It demonstrated that the Soviet Union had the capability to deliver nuclear weapons over intercontinental distances, and contested cherished American notions of military, economic and technological superiority.[30] This precipitated the Sputnik crisis, and triggered the Space Race.[31] President Dwight D. Eisenhower responded to by creating the National Aeronautics and Space Administration (NASA), which absorbed the NACA.[32]

NACA had long been interested in nuclear technology. In 1951, it had begun exploring the possibility of acquiring its own nuclear reactor for the aircraft nuclear propulsion (ANP) project, and selected its Lewis Flight Propulsion Laboratory in Ohio to design, build and manage it. A site was chosen at the nearby Plum Brook Ordnance Works,[33] NACA obtained approval from the AEC, and construction of the Plum Brook Reactor commenced in September 1956.[34] Abe Silverstein, the director of Lewis, was particularly eager to acquire control of Project Rover.[35]

Donald A. Quarles, the Deputy Secretary of Defense, met with T. Keith Glennan, the new administrator of NASA, and Hugh Dryden, Glennan's deputy on 20 August 1958,[35] the day they after Glennan and Dryden were sworn into office at the White House,[36] and Rover was the first item on the agenda. Quarles was eager to transfer Rover to NASA, as the project no longer had a military purpose.[15] Responsibility for the non-nuclear components of Project Rover was officially transferred from the United States Air Force (USAF) to NASA on 1 October 1958,[37] the day NASA officially became operational and assumed responsibility for the US civilian space program.[38]

Project Rover became a joint NASA–AEC project.[37] Silverstein, whom Glennan had brought to Washington, DC, to organise NASA's spaceflight program,[39] appointed Harold Finger to oversee the nuclear rocket development as head of NASA's Office of Space Reactors.[15] Anderson had doubts about Finger's suitability for the job. He felt that Finger lacked enthusiasm for it. Glenn met with Anderson on 13 April 1959, and convinced him that Finger would do a good job.[40] On 29 August 1960, NASA created the Space Nuclear Propulsion Office (SNPO) to oversee the nuclear rocket project.[41] Finger was appointed as its manager, with Milton Klein from AEC as his deputy.[42] Finger was also the Director of Nuclear Systems in the NASA Office of Advanced Research and Technology.[43] A formal "Agreement Between NASA and AEC on Management of Nuclear Rocket Engine Contracts" was signed by NASA Deputy Administrator Robert Seamans and AEC General Manager Alvin Luedecke on 1 February 1961. This was followed by an "Inter-Agency Agreement on the Program for the Development of Space Nuclear Rocket Propulsion (Project Rover)", which they signed on 28 July 1961.[43] SNPO also assumed responsibility for SNAP, with Armstrong becoming assistant to the director of the Reactor Development Division at AEC, and Lieutenant Colonel G. M. Anderson, formerly the SNAP project officer in the disbanded ANP Office, became chief of the SNAP Branch in the new division.[42] It soon became apparent that there were considerable cultural differences between NASA and AEC.[15]

The High Energy Rocket Engine Research Facility (B-1) (left) and Nuclear Rocket Dynamics and Control Facility (B-3) (right) at NASA's Plum Brook Station in Sandusky, Ohio, were constructed in the early 1960s to test full-scale liquid hydrogen fuel systems in simulated altitude conditions.

SNPO Headquarters was co-located with AEC Headquarters in Germantown, Maryland.[41] Finger established branch offices at Albuquerque, New Mexico, (SNPO-A) to work liaise with LASL, and in Cleveland, Ohio, (SNPO-C) to coordinate with the Lewis Research Center, which was activated in October 1961. In February 1962, NASA announced the establishment of the Nuclear Rocket Development Station (NRDS) at Jackass Flats, and in June an SNPO branch was established at Las Vegas (SNPO-N) to manage it. By the end of 1963, there were 13 NASA personnel at SNPO Headquarters, 59 at SNPO-C and 30 at SNPO-N.[43] SNPO staff were a combination of NASA and AEC employees whose responsibilities included "program and resource planning and evaluation, the justification and distribution of program resources, the definition and control of overall program requirements, monitoring and reporting of progress and problems to NASA and AEC management, and the preparation of testimony to Congress."[44]

Finger called for bids from industry for the development of the nuclear engine for rocket vehicle application (NERVA) based upon the Kiwi engine developed by LASL.[45] The award was scheduled for 1 March 1961, so that the decision to proceed could be made by the incoming Kennedy administration.[46][47] Eight companies submitted bids: Aerojet, Douglas, Glenn L. Martin, Lockheed, North American, Rocketdyne, Thiokol and Westinghouse. A joint NASA–AEC board evaluated the bids. It rated North American's bid as the best bid overall, but Westinghouse and Aerojet had superior bids for the reactor and engine respectively when they were considered separately.[48] After Aerojet promised NASA administrator James E. Webb that it would put its best people on NERVA, Webb spoke to the selection board and told them that while he did not wish to influence their decision, North American was deeply committed to Project Apollo, and the board might consider combining other bids.[49] On 8 June, Webb announced that Aerojet and Westinghouse had been selected.[47] Aerojet became the prime contractor, with Westinghouse as the principal subcontractor.[50] Both companies recruited aggressively, and by 1963, Westinghouse had 1,100 staff working on NERVA.[48]

In March 1961, President John F. Kennedy announced the cancellation of the aircraft nuclear propulsion project just as NASA's Plum Brook reactor was nearing completion,[51] and for a time it seemed that NERVA would soon follow. NASA estimated its cost at $800 million (although AEC reckoned that it would be much less),[52] and the Bureau of the Budget argued that NERVA made sense only in the context of a crewed lunar landing or flights further into the Solar System, to neither of which had the administration committed. Then, on 12 April, the Soviet Union launched Yuri Gagarin into orbit on Vostok 1, once again demonstrating their technological superiority. A few days later, Kennedy launched the disastrous Bay of Pigs Invasion of Cuba, resulting in yet another humiliation for the United States.[53] On 25 May, he addressed a joint session of Congress. "First," he announced, "I believe that this nation should commit itself to achieving the goal, before this decade is out, of landing a man on the moon and returning him safely to the earth." He then went on to say: "Secondly, an additional 23 million dollars, together with 7 million dollars already available, will accelerate development of the Rover nuclear rocket. This gives promise of someday providing a means for even more exciting and ambitious exploration of space, perhaps beyond the moon, perhaps to the very end of the Solar System itself."[54]

Technicians in a vacuum furnace at the NASA Lewis' Fabrication Shop prepare a Kiwi B-1 nozzle for testing.

The first phase of Project Rover, Kiwi, was named after the New Zealand kiwi bird.[17] A kiwi cannot fly, and the Kiwi rocket engines were not intended to do so either. Their function was to verify the design, and test the behavior of the materials used.[20] The Kiwi program developed a series of non-flyable test nuclear engines, with the primary focus on improving the technology of hydrogen-cooled reactors.[68] In the Kiwi A series of tests conducted between July 1959 and October 1960, three reactors were built and tested. Kiwi A was considered a success as a proof of concept for nuclear rocket engines. It demonstrated that hydrogen could be heated in a nuclear reactor to the temperatures required for space propulsion and that the reactor could be controlled.[69]

The next step was the Kiwi B series of tests, which commenced with Kiwi B1A on 7 December 1961. This was a development of the Kiwi A engine, with a series of improvements. The second test in the series, Kiwi B1B on 1 September 1962, resulted in extreme structural damage to the reactor, with fuel module components being ejected as it was ramped up to full power. The subsequent full-power Kiwi B4A test on 30 November 1962, along with a series of cold flow tests revealed that the problem was vibrations induced as the hydrogen was heated when the reactor was brought to full power that shook the reactor apart (rather than when it was running at full power).[70] Unlike a chemical engine that would likely have blown up after suffering catastrophic damage, the nuclear rocket engine remained stable and controllable even when tested to destruction. The tests demonstrated that a nuclear rocket engine would be rugged and reliable in space.[71]

Kennedy visited Los Alamos on 7 December 1962 for a briefing on Project Rover.[72] It was the first time a president had visited a nuclear weapons laboratory. He brought with him a large entourage that included Lyndon Johnson, McGeorge Bundy, Jerome Wiesner, Harold Brown, Donald Hornig, Glenn Seaborg, Robert Seamans, Harold Finger, Clinton Anderson, Howard Cannon and Alan Bible. The next day, they flew to Jackass Flats, making Kennedy the only president to ever visit a nuclear test site. Project Rover had received $187 million in 1962, and AEC and NASA were asking for another $360 million in 1963. Kennedy drew attention to his administration's budgetary difficulties, and asked what the relationship was between Project Rover and Apollo. Finger replied that it was an insurance policy, and could be used in the later Apollo or post-Apollo missions, such as a base on the Moon or a mission to Mars. Weisner, supported by Brown and Hornig, argued that if a Mars mission could not occur before the 1980s, then RIFT could be postponed to the 1970s. Seamans noted that such an attitude had resulted in the Sputnik crisis and a loss of American prestige and influence.[73]

Inside the E-MAD

In January 1963, Anderson became chairman of the United States Senate Committee on Aeronautical and Space Sciences. He met privately with Kennedy, who agreed to request a supplemental appropriation for RIFT if a "quick fix" to the Kiwi vibration problem that Seaborg promised could be implemented. In the meantime, Finger called a meeting. He declared that there would be no "quick fix". He criticised LASL's management structure and called for LASL to adopt a project management structure. He wanted the case of the vibration problems thoroughly investigated, and the cause definitely known before corrective action was taken. Three SNPO staff (known at LASL as the "three blind mice") were assigned to LASL to ensure that his instructions were carried out. Finger assembled a team of vibration specialists from other NASA centers, and along with staff from LASL, Aerojet and Westinghouse, conducted a series of "cold flow" reactor tests using fuel elements without fissionable material.[74][75] RIFT was cancelled in December 1963. Although its reinstatement was frequently discussed, it never occurred.[55]

A series of minor design changes were made to address the vibration problem. In the Kiwi B4D test on 13 May 1964, the reactor was automatically started and briefly run at full power with no vibration problems. This was followed by the Kiwi B4E test on 28 August in which the reactor was operated for twelve minutes, eight of which were at full power. On 10 September, Kiwi B4E was restarted, and run at full power for two and a half minutes, demonstrating the ability of a nuclear rocket engine to be shut down and restarted.[70] In September, tests were conducted with a Kiwi B4 engine and PARKA, a Kiwi reactor used for testing at Los Alamos. The two reactors were run 4.9 meters (16 ft), 2.7 meters (9 ft) and 1.8 meters (6 ft) apart, and reactivity measurements were taken. These tests showed that neutrons produced by one reactor did indeed cause fissions in another, but that the effect was negligible: 3, 12 and 24 cents respectively. The tests demonstrated that nuclear rocket engines can be clustered, just as chemical ones often are.[71][76][77]

NERVA nuclear rocket engine

SNPO chose the 330,000-newton (75,000 lbf) Kiwi-B4 nuclear thermal rocket design (with a specific impulse of 825 seconds) as the baseline for the NERVA NRX (Nuclear Rocket Experimental). Whereas Kiwi was a proof of concept, NERVA NRX was a prototype of a complete engine. That meant that it would need actuators to turn the drums and start the engine, gimbals to control its movement, a nozzle cooled by liquid hydrogen, and shielding to protect the engine, payload and crew from radiation. Westinghouse modified the cores to make them more robust for flight conditions. Some research and development was still required. The available temperature sensors were accurate only up to 1,980 K (1,710 °C), far below what was required. New sensors were developed that were accurate to 2,649 K (2,376 °C) , even in a high-radiation environment. Aerojet and Westinghouse attempted to theoretically predict the performance of each component. This was then compared to the actual test performance. Over time, the two converged as more was understood. By 1972, the performance of a NERVA engine under most conditions could be accurately forecast.[78]

The first test of a NERVA engine was of NERVA A2 on 24 September 1964. Aerojet and Westinghouse cautiously increased the power incrementally, to 2 MW, 570 MW, 940 MW, running for a minute or two at each level to check the instruments, before finally increasing to full power at 1,096 MW. The reactor ran flawlessly, and only had to be shut down after 40 seconds because the hydrogen was running out. The test demonstrated that NERVA had the designed specific impulse of 811 seconds (7.95 km/s); solid-propellant rockets have a maximum impulse of around 300 seconds (2.9 km/s) while chemical rockets with liquid propellant can seldom achieve more than 450 seconds (4.4 km/s). Executives at Aerojet and Westinghouse were so pleased they took out a full-page ad in the Wall Street Journal with a picture of the test and the caption: "On to Mars!" The reactor was restarted on 15 October. Originally this was intended to test the nozzle, but that was dropped as it was close to its design maximum of 2,270 K (2,000 °C). Instead, the turbopump was tested. The engine was powered up to 40 MW, the control drums were locked in place, and the turbopump was used to keep the power steady at 40 MW. It worked perfectly. The computer simulations had been correct, and the whole project was ahead of schedule.[79][80]

ETS-1 at Test Cell C

The next test was of NERVA A3 on 23 April 1965. This test was intended to verify that the engine could be run and restarted at full power. The engine was operated for eight minutes, three and a half of them at full power, before the instruments indicated that too much hydrogen was going into the engine. A scram was ordered, but a coolant line became clogged. Power increased to 1,165 MW before the line unclogged, and the engine shut down gracefully. There were fears for the integrity of the tie rods that held the fuel clusters together. They were supposed to operate at 473 K (200 °C), with a maximum of 651 K (378 °C). The sensors recorded that they had reached 1,095 K (822 °C), which was their own maximum. Laboratory tests later confirmed that they might have reached 1,370 K (1,100 °C). There was also what appeared to be a hole in the nozzle, but this turned out to be soot. The robust engine was undamaged, so the test continued, and the engine was run for thirteen minutes at 1,072 MW. Once again, the test time was limited only by the available hydrogen.[79][80]

Testing of NASA's NERVA NRX/EST (Engine System Test) commenced on 3 February 1966.[81] The objectives were:

1. Demonstrate the feasibility of starting and restarting the engine without an external power source.
2. Evaluate the control system characteristics (stability and control mode) during startup, shutdown, cooldown and restart for a variety of initial conditions.
3. Investigate the system stability over a broad operating range.
4. Investigate the endurance capability of the engine components, especially the reactor, during transient and steady-state operation with multiple restarts.[82]

The NRX/EST was run at intermediate power levels on 3 and 11 February, with a full power (1,055 MW) test on 3 March, followed by engine duration tests on 16 and 25 March. The engine was started eleven times.[81] All test objectives were successfully accomplished, and NRX/EST operated for nearly two hours, including 28 minutes at full power. It exceeded the operating time of previous Kiwi reactors by nearly a factor of two.[82]

The next objective was to run the reactors for an extended length of time. The NRX A5 was started up on 8 June 1966, and run at full power for fifteen and a half minutes. During cooldown, a bird landed on the nozzle and was asphyxiated by the nitrogen or helium gas, dropping onto the core. It was feared that it might block the propellant lines or create uneven heating before being blown out again when the engine was restarted, so the Westinghouse engineers rigged a television camera and a vacuum hose, and were able to remove the bird while safely behind a concrete wall. The engine was restarted on 23 June and run at full power for another fourteen and a half minutes. Although there was severe corrosion, resulting in about $2.20 of reactivity lost, the engine could still have been restarted, but the engineers wanted to examine the core.[83][84]

An hour was now set as the goal for the NRX A6 test. This was beyond the capacity of Test Cell A, so testing now moved to Test Cell C with its giant dewars. NRX A5 was therefore the last test to use Test Cell A. The reactor was started on 7 December 1966, but a shutdown was ordered 75 seconds into the test due to a faulty electrical component. This was followed by a postponement due to inclement weather. NRX A6 was started up again on 15 December. It ran at full power (1,125 MW) with a chamber temperature of over 2,270 K (2,000 °C) and pressure of 4,089 kilopascals (593.1 psi), and a flow rate of 32.7 kilograms per second (4,330 lb/min). It took 75.3 hours to cool the reactor with liquid nitrogen. On examination, it was found that the beryllium reflector had cracked due to thermal stress. The test caused the abandonment of plans to build a more powerful NERVA II engine. If more thrust was required, a NERVA I engine could be run longer, or it could be clustered.[83][84]

With the success of the A6 test, SNPO cancelled planned follow-on tests A7 and A8 and concentrated on completing ETS-1. All previous tests had the engine firing upwards; ETS-1 would permit an engine to be reoriented to fire downward into a reduced-pressure compartment to partly simulate firing in the vacuum of space. The test stand provided a reduced atmospheric pressure of about 6.9 kilopascals (1.00 psi) – equivalent to being at an altitude of 60,000 feet (18,000 m). This was done by injecting water into the exhaust, which created superheated steam that surged out at high speeds, creating a vacuum.[85][86]

NERVA control room

ETS-1 took longer for Aerojet to complete than expected, partly due to shrinking budgets, but also because of technical challenges. It was built from pure aluminum, which did not become radioactive when irradiated by neutrons, and there was a water spray to keep it cool. Rubber gaskets were a problem, as they tended to turn into goo in a radioactive environment; metal ones had to be used. The most challenging part was the exhaust ducts, which were required to handle much higher temperatures than their chemical rocket counterparts. The steel work was carried out by Allegheny Technologies, while the Air Preheater Company fabricated the pipes. The work required 54,000 kilograms (120,000 lb) of steel, 3,900 kilograms (8,700 lb) of welding wire and 10.5 kilometers (6.5 mi) of welds. During a test the 234 tubes would have to carry up to 11,000,000 litres (3,000,000 US gal) of water. To save money on cabling, Aerojet moved the control room to a bunker 240 meters (800 ft) away.[85]

The second NERVA engine, the NERVA XE, was designed to come as close as possible to a complete flight system, even to the point of using a flight-design turbopump. Components that would not affect system performance were allowed to be selected from what was available at Jackass Flats to save money and time, and a radiation shield was added to protect external components.[87] The test objectives included testing the use of ETS-1 at Jackass Flats for flight engine qualification and acceptance.[88] Total run time was 115 minutes, including 28 starts. NASA and SNPO felt that the test "confirmed that a nuclear rocket engine was suitable for space flight application and was able to operate at a specific impulse twice that of chemical rocket system[s]."[89] The engine was deemed adequate for Mars missions being planned by NASA. The facility was also deemed adequate for flight qualification and acceptance of rocket engines from the two contractors.[89]

The final test of the series was XE Prime. This engine was 6.9 meters (23 ft) long, 2.59 meters (8 ft 6 in) in diameter, and weighed approximately 18,144 kilograms (40,001 lb). It was designed to produce a nominal thrust of 246,663 newtons (55,452 lb_f) with a specific impulse of 710 seconds (7.0 km/s). When the reactor was operating at full power, about 1,140 MW, the chamber temperature was 2,272 K (2,000 °C), chamber pressure was 3,861 kilopascals (560.0 psi), and the flow rate was 35.8 kilograms per second (4,740 lb/min), of which 0.4 kilograms per second (53 lb/min) was diverted into the cooldown system.[1] A series of experiments were carried out between of 4 December 1968 and 11 September 1969, during which the reactor was started 24 times,[86] and ran at full power for 1,680 seconds.[1]