Damage to the cable was strictly cosmetic, and we wrapped some electrical tape over the wound. It still worked perfectly, but we were too embarrassed to deliver it to the plant in that condition, so we called ROLM Military Computers in Cupertino, California, and ordered a new unit. They sent it promptly, and this was an improved example of the cable. The Army had apparently given them grief about the weight of their parallel-digital cables, so ROLM had built new ones using a smaller gauge of wires. It was slender and svelte, not as clumsy and bulky-looking as the original cable, and each individual wire was tested for continuity and lack of cross-talk. Once the new cable passed all the tests, we installed it in to the power plant.
However, not everything was perfect. The problem we soon found with the new cable was that, thanks to the slimmer design, there was a reduced cross-sectional area of the wires used. A digital computer signal is sent over an electrical circuit, and as such it needs a transmission wire and a return wire, or ground. In a wired digital circuit of many parallel paths, such as a data-bus or an interconnection cable as we were using, a common ground connection is used by all signals. In our original cables, the wire gauge was overkill, with more copper than was necessary for the low-current digital signals, and a single ground wire was big enough to carry all the signals firing at once. With the wire size reduced to a gauge that was still good for the signal current, the smaller ground wire was incapable of returning more than a few signals at the same time. If, by pure chance, most of the sent bits at a particular instance happened to be ones, then the signal voltage would drop below the logic threshold, and at the receiving end they would be interpreted as zeros. The reduced-diameter cable would test perfect if every wire and return was actuated individually, but under certain, seemingly rare conditions of data structure, it could fail.[284]
We sent home an emergency request for the rat-bitten cable to be returned to us, re-installed it, and experienced no further data drop-outs. The electricians at the E. I. Hatch Nuclear Power Plant found our predicament amusing. The celebration party never happened, but the laughter we generated still rings in my ears.
The lesson of the Rat Cable Problem gave me a sense of sympathy for the engineers at General Electric as they watched their dreams of robust reactor systems, designed to work under the worst conditions imaginable, crumble to pieces at Fukushima I when the unimaginable happened. Trying to build something that will work perfectly for all time is a noble goal, but it is simply impossible. The issue that causes failure can be as grand as an earthquake and tsunami or as mundane as a brown rat.
Admiral Rickover’s nuclear submarine power plant was such a design. It was radical in every detail, and a great number of innovations were necessary to make it work. It started with the idea of using water as both the coolant and the neutron moderator. That would work, but it would mean that the fuel had to be enriched with an unnatural concentration of uranium-235. Some material would have to be devised that would hold the fuel in place with water running among small, cylindrical rods of uranium, and it could not parasitically absorb neutrons. Every neutron was precious. It also had to be able to withstand high temperature and be strong enough to hold the reactor core together without dominating the space. Zirconium fit the list of requirements, but there were no zirconium mines, refineries, or fabrication techniques. Rickover had to invent it all from scratch. He came up with the idea of control rods made of hafnium, which was another material that was not available at the hardware store. His exotic machine was entirely successful, driving his submarines, catapulting the United States Navy further into world domination, and it did not harm a single sailor. Rickover’s system test program was, without question, as rigorous and complete as could be accomplished.
The civilian nuclear-power industry, also starting from scratch in a very small, experimental step-off, was not the sort of enterprise that could develop new metals and radical designs of things that had never been built before. The utilities that were bold enough to try nuclear power were pleased when the naval reactor technology was declassified and turned into a stationary power plant at Shippingport, Pennsylvania, in 1957. It was a small plant, generating only 60 megawatts of electricity, but it never had a problem in 25 years of service. Seeing this as a good sign, the United States and eventually the world eventually stopped experimenting with different reactor concepts and settled on Rickover’s submarine unit as a standard for how nuclear power should be applied to the need for reliable, clean power.
Was this a good idea, or did the world’s utilities fall into a trap? The other radical idea for nuclear power, such as the liquid-metal-cooled fast breeder, had also been tried many times with consistently unfavorable results, just as Rickover had predicted long ago, when the Navy wanted to try it in submarines. The liquid-metal technology, as he pointed out, was expensive, prone to disasters, and extremely difficult to repair when something broke. Having a coolant that would catch fire when exposed to air did not seem right. Rickover was correct on that observation. Why would he be wrong about the pressurized-water reactor?
There have been trillions of problem-free watt-hours generated by scaled-up Rickover plants, but there may be a problem area that was not evident when submarine reactors were tiny, 12-megawatt machines, but that was revealed when the Rickover model was enlarged multiple times over for industrial use. The reactor core, the uranium fuel pellets lined up in zirconium tubes and neatly separated from each other, is terribly sensitive for such an otherwise robust machine. Let the coolant come off the fuel for a few minutes, even with the reactor shut completely down, and the entire, multi-billion-dollar machine is in irreversible jeopardy. The high-temperature zirconium alloys in an overheated reactor core oxidize, losing their metallic strength, generating explosive hydrogen gas, and contributing to high-pressure conditions in the isolated reactor vessel. The delicately structured core collapses, and the soluble fission products are able to mix with the escaping remnants of the coolant. It has happened as recently as 2011. To start the destruction sequence requires a lot of bad luck and human intervention. It is part of the nuclear power plant that does not easily forgive errors or dampen out mistakes. This part of the reactor design, the orderly matrix of thin tubes filled with fuel, is a weakness in an otherwise robust system.
There have been many engineering fixes and modifications to correct these problems, but ironically, these fixes can then present new issues, as they are complex add-ons, cluttering up an otherwise simple design with a maze of pipes and hundreds of additional valves, tanks, electrical cables, pumps, turbines, filters, re-combiners, and compressed-air tubing. Most of the plumbing in a nuclear plant has nothing to do with generating electricity. It is part of the fix that keeps the reactor core from melting down under unusual circumstances. These complex light-water-reactor designs, the boiling-water reactor and the pressurized-water reactor, have been pursued with such enthusiasm over the past sixty years, one could assume that there is no other reasonable way to build a civilian power reactor. Alternate designs, such as the graphite, as was used in Windscale, and the liquid-metal-cooled reactor at Fermi 1, have proven impractical and have fallen away.[285] If only there were a proven reactor design that was in no danger of melting the fuel, collapsing the core structure, and generating hydrogen, it would solve many problems that bedevil the current crop of world-wide power reactors.[286]
284
We were not the only ones bitten by the ground-return problem. The S-100 “Altair” data bus (IEEE696-1983), introduced in 1974, was an extremely popular design feature used in many micro-computer applications in the late seventies through the eighties, but it suffered from too few grounds.
285
Another alternate design, the Canadian CANDU reactor, continues on. Although its coolant/moderator, heavy water, is extremely expensive, the concept of a power reactor that runs on natural uranium with constant refueling has been attractive to countries such as India and China. A byproduct of CANDU power generation is bomb-grade plutonium-239.
286
The nuclear-power industry is acutely aware of these weak points and has not been sitting idle. The newer Generation 3 reactors, such as the Westinghouse AP1000, address the issues pointed out here with several innovations. The emphasis is on passive systems, not requiring electricity, to keep water covering the fuel and preventing core damage. Two of these reactors are currently being built at the Vogel Nuclear Power Plant near Augusta, Georgia. The nuclear division of Westinghouse is now a Japanese-owned company.