Physicists love explosions. Twelve years after the end of Project Orion I was cooking breakfast for my father, in British Columbia, on a pressurized kerosene stove that sprung a leak and became engulfed in flames. As I dropped everything and turned to run, there stood Freeman. "Oh good! An explosion!" he exclaimed.
Physicists also love simplified, abstract models. A mathematical model need not correspond exactly to actual physics to provide insight into the real world, and a physical experiment that does not correspond exactly to a mathematical model may still indicate whether the model is on the right theoretical track or not. The development of nuclear weapons, ever since the first test at Alamogordo, was driven by this interplay between theory and experiment, numerical models advanced the design of test devices that advanced the design of numerical models leading to new generations of bombs. "Nine times out of ten we got about three-quarters of the way with computing, but we always had rude surprises when we did experiments," says physicist Bud Pyatt, still engaged today in weapons-effects work descended from Project Orion.
In early 1958, Ted Taylor and his colleagues faced the same challenge that had faced the pioneers at Los Alamos: how to combine numerical modeling, intuition, and limited experimental evidence into a design that stood the best chance of working on the first try. They were operating in a vacuum, propelled by a million dollars from ARPA, without precedent or constraint. Until August 1958 physicists had no firsthand experience with nuclear explosions in space. The behavior of a nuclear explosion in a vacuum should be easier to predict than the behavior of a nuclear explosion in the atmosphere, yet, when the AEC and DOD did conduct high-altitude nuclear tests at the edge of space, many people—if not the Orion physicists—were quite surprised. "Setting off explosions in space is so complex you just can't understand it," says Ted. "Nobody can understand it. All you have to do to prove that is set somebody down and ask them questions until you ask them something where they say, 'I don't know.' You don't have to go very far."
A nuclear explosion in a vacuum instantaneously transforms a certain amount of cold matter into an extremely hot, unconstrained gas. How does the initial shape and density of the cold material affect the distribution of high-temperature, high-velocity material that results? This was one of the first questions that Freeman examined in 1958. Assuming Orion to be a sequence of events leading to Mars or Jupiter and back, the initial expansion of the propellant—material placed around, or near, the bomb—was a good place to start.
This did not mean designing bombs. That was the domain of Livermore and Los Alamos, the AEC's two authorized bomb-designing labs. Ted and his colleagues might have ideas about what kind of bombs to use for Orion, but it was none of General Atomic's business to specify their design. "What was finally agreed on," says Moe Scharff, "is that only the labs should work on weapons design, but other entities could work on weapons effects. The labs—Livermore, Los Alamos, Sandia—were willing to have outside companies work on the effects of their products, as long as they didn't dabble in the products themselves. In the end the position was taken that the weapon ends with its outer skin or envelope, so if you then put something around it, to utilize the output, OK, even if that something was less than a meter away. That's the way it finally worked out. Don't try to fiddle around inside this energy-producing mechanism, but once the energy is produced, if you want to transform it into some other kind of energy, be our guest. As long as you check carefully with us, regularly."
Freeman Dyson had learned most of what he knew about nuclear weapons in one day at the end of the summer of 1956. Not yet an American citizen, he could work on a civilian fission reactor like TRIGA, but not on civilian fusion power ("Project Sherwood") or anything to do with bombs. Toward the end of the summer, after the design of TRIGA had been completed, the AEC, in a panic over a perceived shortage of tritium, assigned General Atomic to design a tritium-producing reactor, code-named "Project August" because the design had to be completed within three weeks. To enlist Freeman's help required a special clearance from the AEC. "The AEC of course has a logical explanation for this absurd situation," Freeman reported on August 26. "Their regulations say that secret information may be given to a foreigner only when this is necessary to the national defense. Obviously, if the information is not vital military information, it cannot be necessary to the national defense to give it to me. Therefore I can have the important military secrets but not the unimportant civilian secrets. It is the craziest joke I ever heard."[140]
Within a few days of receiving his temporary clearance, Freeman received an invitation to stop by Los Alamos on his way home to Princeton in September, for which the necessary additional clearances were being arranged. "I finally managed to get here," he wrote from the Lodge at Los Alamos on September 20. "Yesterday came the news that my clearance has been approved. So I flew to Albuquerque in the evening and this morning took the little 5-seater plane which comes up here and lands on a little mesa not much wider than the runway, perfectly flat on top and with a deep canyon on each side." Freeman had only two days in Los Alamos before flying to New York. "I took the last of my 14 rabies injections in La Jolla yesterday," he added, the result of a trip to Tijuana where he had been bitten by a rabid-looking dog. "Today I spent absorbing all the information I could at tremendous speed. My clearance is good for everything, Sherwood machines and all kinds of bombs."[141]
"To my amazement they simply stuffed me with all their information about bombs," says Freeman. "I hadn't asked for that, I wasn't particularly interested in bombs. They wanted to tell me everything they'd been doing, as if they'd just been burning to talk about this to somebody—all the designs that they had done and what they were planning to do. So I listened to all this, I didn't do anything, and came back here to Princeton and resumed the normal life. But it was very useful when it came to Orion." Freeman now believes the mysterious invitation to Los Alamos was organized out of a concern that Livermore might get to him first.
Freeman's analysis of nuclear explosions in a vacuum, resulting in a series of three short papers titled Free Expansion of a Gas, was central to the feasibility of Orion. It was also central to the feasibility of directed-energy nuclear weapons, and led directly from Orion to a project code-named "Casaba-Howitzer," described as "a one-shot version of Orion, like Orion except without any ship." Casaba-Howitzer, conceived by Moe Scharff while still at Livermore, would be resurrected many years later as the basis for the "Star Wars" space-weapons program, known as the Strategic Defense Initiative or SDI. "Whereas Orion directed a dense plasma at relatively low velocity at a wide angle, this was to direct a lower-density plasma at a higher velocity and a narrower angle," Scharff explains. "Orion was a space vehicle. Casaba-Howitzer could be considered space weaponry. It could even have been things carried aboard an Orion, for example, if Orion was a battleship."
Casaba-Howitzer's descendants remain under active investigation and Scharff is unable to give any further details beyond the origins of the name. "They had been naming things after melons and the good ones were gone already. They were on a melon kick that year. The one connection was seeds—many of those melons have seeds, like the particles we were projecting." Casaba-Howitzer was derived directly from Orion, and later versions of Orion drew heavily on Casaba-Howitzer's experimental and theoretical results. Funding for Casaba-Howitzer kept the Orion team going after funding for Orion dwindled out. But there was a costly side to the bargain—a shroud of secrecy that has lingered long after any plans for battleship Orion were shelved. Conversely, if we ever decide to build something like Orion, it will be the continued work on directed-energy weapons—and how to protect surfaces against them—that will allow us to pick up where Project Orion left off.
Pulse
unit for a 10-meter-diameter Orion vehicle: yield approximately 1
kiloton,
weight 311 pounds, with between 2,000 and 3,000 charges required for a
voyage
to Mars and back. As the nuclear device explodes, the initial burst of
energy
is confined by the radiation case and channeled toward the propellant
slab.
Anything in the near vicinity of a nuclear explosion gets vaporized into a plasma—a cloud of material so hot that its atoms are stripped of their electrons—that cools as it expands. It was a simple mathematical problem to draw some conclusions relating the shape and density of the initial object that gets vaporized to the shape and density of the resulting cloud of gas. "The model should be simple enough so that the hydro-dynamical equations can be integrated exactly," Freeman explained. "A real cloud of gas will not have precisely the density-distribution of the model, but still one may expect the behavior of a real cloud to be qualitatively similar to that of the model."[142] Freeman set up the equations and the numbers were run on General Atomic's IBM 650 card-programmed calculator, one of the workhorse machines that had handled many of the early bomb and blast-wave calculations at Los Alamos and had not yet been superseded by the IBM 704 that General Atomic acquired in 1959.
According to Freeman's model, something originally in the shape of a cigar expands into the shape of a pancake, and something originally in the shape of a pancake expands into the shape of a cigar. This was "very directly relevant to the expansion of a bomb," he explains. "If you have something that starts in the form of a pancake and you heat it up to a very high temperature it will expand more sideways along the axis, and less at the edges. The pressure gradient is highest along the axis, so then after a while, since the velocity is highest along the axis, it becomes cigar-shaped. So you get inversion, something that begins like a pancake becomes like a cigar, and something that begins as a cigar becomes a pancake, if you just let it expand freely. It goes roughly with the square root, if you start with a pancake where the ratio of the diameter to thickness is ten, then it will end up as a cigar where the ratio of the length to the diameter is square root of ten, roughly speaking. That would be quite helpful, of course, if you had a real Orion, to start out with a pancake and it will produce then a jet that is collimated within 20 degrees or so quite nicely. The fact that it's so easy to make an asymmetrical explosion may still be classified, for all I know."
The right pancake in the right place can focus a significant fraction of the bomb's output into a narrow jet of kinetic energy, directed constructively at the pusher plate of a nearby spaceship—or destructively at something else. The thinner the pancake, the narrower the jet. In the early days of Orion, with a huge pusher plate as the target, the propellant was assumed to be a thick slab of something light and cheap like polyethylene; later versions of Orion, with smaller pusher plates, required a thinner slab of higher-density material, such as tungsten, to focus the bomb's energy into a narrower cone. Exactly how narrow remains a secret, though a look at the later configurations of Orion permits a guess. This is one of the reasons that detailed design information about Orion, such as the exact standoff distance between the pulse unit and the pusher plate, remains classified, even after forty years have passed.
As the jet of propellant is targeted more narrowly in space, its impact against the pusher plate is spread out more widely in time. The result is more effective horsepower and a softer ride. "In the end we did come up with some designs that were very tight in their angular distribution of momentum," says Bud Pyatt, without mentioning specific numbers, but revealing that "you had to have it pointing at the center of the pusher plate, it couldn't even be five degrees off without stressing the shock absorber too much."
Pyatt, then twenty-six years old, arrived at General Atomic just as Freeman was leaving in September 1959. "It was an exciting period," he says. "When people asked, 'What are you doing?' — 'Well, I'm working on a spaceship propelled by nuclear weapons exploding a few hundred meters away.' They would look at you with this very strange look." Pyatt spent his first eighteen months assisting astrophysicist John C. Stewart in a detailed study of opacities of light elements and then began to focus, under the guidance of Burt Freeman, on improving the pulse unit design.
Adapting weapons-design codes from Livermore and Los Alamos, Pyatt and his colleagues explored a series of refinements to Orion, in far more detail than the first approximations made in 1958 and 1959. "A typical complete calculation for a given pulse system and the subsequent interaction with the pusher requires about two man-years of effort and 50 hours of computer time on an IBM-7090," he explained in 1963.[143] As new computers, improved computer codes, and experimental data came in, confidence increased that the pulse units could be made to perform better than had been hoped for in the original design. "No one had really looked into this with the detail that we examined the design of the pulse propulsion system," Pyatt explains. "The weapons lab people were perfectly happy to stop once they figured out that the bomb worked. All the emphasis had been on bomb physics, not on the analysis of what the emanations of the bomb might do, particularly the hydrodynamics. Some of the stuff Freeman Dyson did, that I followed up on, was very exciting."
"The idea was to have a variable density in the pulse propulsion system. In the same way that one designed two-stage bombs, we had a radiation channel and then a plate was accelerated by the radiation pulse." In cross section, the Orion pulse units resemble an old-fashioned television: the bomb sits at the neck of the picture tube, surrounded by a melon-shaped radiation case; the conical picture tube contains the channel filler; the face of the picture tube is the pancake of propellant. "And now one can go in here," says Pyatt, pointing to the pancake, "and start shaping this and controlling its density in such a way that as it expands outward it's going to have the right density."
The load on the pusher—which for the sake of shock absorbers and passengers should be spread out as widely as possible in time—is governed by the local density of the propellant cloud multiplied by the square of its velocity. "Freeman Dyson wrote a beautiful report on that the summer he was here before I came, and it proved perfectly correct," Pyatt continues. "The velocity is the velocity; you can affect the average and the peak, but you cannot affect the distribution. But you can affect the density distribution, greatly, by controlling the initial density in the propellant slab. Not only does a flat plate expand as a long cigar, if you build in a density profile through the flat plate, that is, a lower density in the front or a higher density and then a lower density in the middle or something, it remembers this. The idea was to mitigate the shock. You didn't want to change the total impulse. The impulse was the impulse; you had to have that to make the whole concept work. But you could certainly control the pressure, if you could stretch it out, spread it out over time, but first of all just direct it—within a half angle of ten degrees or so. And that was highly classified at the time."
Any detailed discussion of how to direct the propellant comes dangerously close to certain details concerning the design of hydrogen bombs. "The technology associated with two-stage devices—radiation channels driving the implosion of secondary devices and bombs—was the technology that we exploited to get the momentum direction," Pyatt explains. "And I'm sure there could have been much more done with this. We never went as far as we could have in designing the pulse system so that it controlled the delivery of the impulse in such a way that it was complementary to the shock absorber. And that would have made the shock absorber problem a heck of a lot easier than just living with a rapid rise and almost exponential decay."
"Later, in the very early days of the first Apple Macintosh I ever owned," says Pyatt, "I ran a design program that let me play with that in my spare time." In 1959, General Atomic owned less computing power than $300 will buy today. "Most of the work that we did on the pulse propulsion system was a full-up, two-dimensional hydrodynamics," says Pyatt. "General Atomic never had a machine that was fast enough, but there were machines that the Air Force made available. I remember endless nights of traveling, and this was before 1-5, you still had a lot of Highway 101 to go up to Hughes Aircraft in Los Angeles. We would get time there, on an IBM 7090. That was a fairly big machine, it probably had 64,000 words of storage on it." This is the memory of the first IBM PC introduced in 1981, and half the memory of the first Apple Macintosh introduced in 1984. "On my desk today I have far more computing power than I ever had when I was working on Orion," adds Pyatt, "with machines that filled the whole basement of the big building at Convair."
There are three surviving legacies of Project Orion: people, documents, and codes. Many of Orion's people came from Livermore and Los Alamos, and so did the codes. "In a moment of weakness, Johnny Foster had agreed with Ted that if they would send someone up to Livermore for a couple months to work with one of their design groups he would let them have one of the big two-dimensional codes that they had developed," Pyatt recalls. "I was elected to do that and I spent a couple of months up at Livermore working with Bill Schultz and brought down the so-called Coronet code. Coronet was the two-dimensional radiation transport code that allowed Livermore to put Los Alamos to shame for ten years in the efficiency of the design of their two-stage devices. We converted that to be a design tool for calculating the behavior of the pulse propulsion system." The Orion version, named MOTET, incorporated improvements that have remained at the heart of the weapons design and verification business ever since.
On July 9, 1962, a 1.4-megaton thermonuclear bomb was exploded 400 km above Johnston Island in the South Pacific. AEC and DOD officials were caught off guard by the spectacular and unexpected aftereffects of the Starfish "event," including eye damage to observers within the exclusion zone, illumination of the nighttime sky 1,400 miles away, and a major disturbance of the Van Allen radiation belts. The Orion group was able to model what had happened, after the fact, using MOTET, concluding that "its close prediction of what was actually observed in the test provided strong verification of the code."[144] This was viewed as a milestone for Orion, even though Starfish was not an Orion test.
In addition to MOTET for expansion and SPUTTER for ablation, the mathematical group at General Atomic came up with codes such as BUMP, for impulse; BAMM, for dynamic response; PRESS, for pusher plate stress; BETELGEUSE, for pusher plate vibration; POGO, for shock absorber behavior; HAYO, computing propellant mass and number of charges per given maneuver; TRIP estimating fuel requirements; and OROP and OROPLE, the Orion Optimization codes. Completed in December 1964, OROP and OROPLE embodied the mathematical relationships among 106 different design and performance parameters defining any given Orion vehicle. The accumulated wisdom of six years of work on all aspects of Orion was distilled into sixty-one pages of Fortran code written for the IBM 7044.[145]
OROP and OROPLE were evolutionary dead ends, codes so specialized that they went extinct when their host project came to an end. More adaptable Orion codes are still going strong. Soon after Project Orion was terminated in 1965, Pyatt and a core group of Orion physicists left General Atomic to form their own independent company, Systems, Science, and Software, or "S-Cubed." The new group included Burt Freeman, Charles Loomis, and Moe Scharff who brought with them the latest generations of programs descended from those first brought to General Atomic from Livermore and Los Alamos in 1958 and 1959. S-Cubed was subsequently merged into Maxwell Technologies in San Diego and recently transferred to SAIC (Science Applications International), where questions first raised by Orion are still under investigation, using the latest versions of computational tools developed forty years ago, such as the Sputter ablation code. Originally constructed by Charles Loomis and Burt Freeman, Sputter can be applied to questions ranging from whether an Orion ship can survive repeated explosions to whether directed-energy weapons can be counted on to cause hostile mechanisms to fail. "Sputter became the Zeus code, which became—I can't even give you all of the legacy," says Pyatt. "That basic code was really the heritage of Orion that S-Cubed exploited, in all of the work that we've done on the ablation problem. I used it extensively on the earliest laser weapon effects calculations. We're using it today."
Orion, with its unanswered questions about ablation and how to fine-tune the free expansion of a gas, has always remained in the back of Pyatt's mind. "I don't think there's anything I've seen that would raise any questions about what we were saying forty years ago," he says. "As to precise details, we'd probably come up with different answers today." He remains convinced that the optimism of 1959 was technically sound. "It would have worked. Even in my dotage, I'm a true believer." When asked about a specific aspect of the pulse unit design, he answers: "The answer is yes, but I can't go into any details on it. It works." So at the time of Orion, I suggest, it was a guess, and then later it was tested? "Yes, for other reasons. It wasn't an Orion test."
In answer to a less specific question, he elaborates: "I think there is absolutely no doubt—and we did some experiments later; still quasi-classified, related to Casaba-Howitzer—that the propulsion system would have worked. We knew what we were doing in designing it. We could send 85 percent of the momentum in one direction that we wanted it to go in, and there were enough experiments—and there have been enough experiments—done on the protection of the pusher plate, to have no doubt that it would have worked. Between those two things there is a tremendous amount of engineering detail to be worked out, but I think it was engineering detail. It could have worked. Now, could it have been done economically, could it have been done in time? Those were all different questions, but I think all of those things could have been solved. Today, people ask me, 'Was it really a joke, Pyatt, or was it serious?' It was dead serious. If we wanted to do it, if there were any good reason for wanting to have high specific impulse and high thrust at the same time, we could go out and build Orion right now. And I think it would make a lot of sense."