In the early days of nuclear reactor design, physicists realized that it would be possible to build reactors which would produce more fissile[i] material than they consumed. Nature has not supplied a generous quantity of fissile isotopes, so reactors have to be designed either to operate with natural uranium[ii] (possible, but not desirable, for various reasons), with enriched uranium, or with plutonium. It is costly either to enrich natural uranium or to produce plutonium by using thermal reactors.[iii] Breeder reactors combine the role of power production with that of fissile material production and offer the potential of a virtually unlimited fuel supply. It is possible to build breeder reactors which can be fueled with a fissile material and operated to produce electric power as in any conventional reactor; at the end of the operating cycle, the reactor fuel will contain more fissile material than it had when initially started. This is not exactly a free lunch, but it is a very good deal.
Although work has been done to develop a thorium based light water breeder reactor, that design has not progressed to the point where it has been of great interest. Instead, the concept of breeding U-238 to produce Pu-239 has become the focus of most design and experimentation. Even in conventional light water power reactors, there is a degree of plutonium production which happens because the reactors are loaded with only 3-4% U-235. The remaining uranium is U-238 which can absorb some of the excess[iv] neutrons, and after a couple of beta decays, will produce Pu-239. This production contributes positively to the performance of the fuel, since it adds some fissile material to replace consumed U-235.
Thermal reactors[v] do not convert U-238 to Pu-239 very efficiently because the neutron capture cross-section is energy[vi] dependent. As the neutron flux becomes harder (more energetic), the capture cross-section increases and that increases the rate of production of Pu-239. It therefore turns out that the way to design a reactor to optimize breeding is to minimize neutron moderation. The resulting design is known as a fast breeder reactor.
When a Pu-239 nucleus is caused to fission by a fast neutron, the average production of fission neutrons is about 3.04 per fission. This is much more than the approximately 2.4 produced by the fission of U-235 when it captures a thermal neutron.
The general design characteristics of any fast breeder are that they can accept fuel which has been recovered from prior cycles of fast breeder cores; they are loaded to enrichments which are typically 3 to 8 times those used in light water reactors; they contain few light elements (which would soften the neutron spectrum); the cores contain a mixture of fissile and fertile isotopes; and they have both radial and axial blankets, consisting of depleted U-238. As these reactors are operated, the fissile isotopes in the core are consumed, but are replaced by the transmutation of U-238 into Pu-239. Neutrons which leak from the core reach the surrounding blanket areas where they may be captured to form additional Pu-239, which begins to fission as the fissile concentration increases. After a period of time, the core reaches the end of its designed lifetime and is removed and replaced. This may be done in stages (partial removal and partial replacement). The reason that the fuel has to be replaced is due to fission product buildup in the fuel and ultimately to metallurgical considerations in the fuel cladding.
When I joined Babcock & Wilcox in 1966, the state of commercial US reactor development was that a number of thermal and breeder research reactors had been built and operated and a few relatively small thermal power reactors had been built largely as pilot plants and demonstrations. For example, B&W had built a good number of low power research reactors, the propulsion reactor for the nuclear powered ship Savannah[vii], and the small Consolidated Edison Thorium Reactor. Other players in the nuclear reactor industry were Westinghouse, General Electric, Combustion Engineering, and General Atomic, each having entered the market in a similar manner.[viii] The government was pursuing reactor development for various military and civilian purposes. It built a number of research reactors, including some fast breeders, such as EBR-I, EBR-II, TREAT, and Fermi-1. Within the first year of my employment, I found myself working on a government sponsored fast breeder reactor design effort. There were parallel programs going on involving helium, steam, and sodium cooled breeders. B&W had the steam cooled breeder reactor (SCBR) contract.
At about that time, work was underway to determine what had caused the fuel melt which had recently damaged the Fermi-1 reactor. I recall seeing a description of a twisted piece of metal which was found to have blocked one of the sodium flow channels. After a rather long time, the metal was identified as cladding which had been added to a structure in the bottom of the reactor, which was designed to redirect the flow of the sodium upwards into the core. The design drawings did not include cladding for the flow director, but someone in the field thought it would be a “good idea,” so he had the cladding fabricated and riveted on. The resulting accident was a precursor of the uphill battle that would face US designed breeder reactors in the next decade.
The time frame was the late 60’s and we were working with a computer which had discrete transistors. Minor calculations were done with sliderules and adding machines. Our only core modeling computer codes were written for thermal reactor calculations and were known as 4 group (diffusion theory) codes. That is, they divided all neutrons into 4 lethargy groups.[ix] This is a satisfactory and efficient way to treat water moderated thermal reactors, but it is not at all desirable for use with fast breeders, since the latter are specifically designed to prevent neutron moderation and to maintain the hardest possible energy spectrum. The actual distribution of neutron energies in a fast breeder is quite wide and it is therefore important to have a large number (on the order of 100) of energy groups built into the model. This point is insignificant, except to demonstrate that the design early design studies were being conducted with unsophisticated modeling.
When the various design reports were finished, the government had the task of eliminating two of the design categories and to then proceed with research on the one believed to be the most promising. As a nonparticipant in the selection process, I can only say that we were hoping it would be the SCBR. The one thing we hoped would carry a lot of weight in the decision was the higher breeding ratio that we argued could be obtained with the SCBR. Each of the three designs had advantages though and the final choice was the Liquid Metal Fast Breeder Reactor (LMFBR).
The liquid metal concept had a number of very attractive features: sodium has a small neutron absorption cross-section (so it will not act as a neutron poison); it is a poor moderator; and it has a high heat transfer coefficient. Sodium has several drawbacks, but the one of greatest concern is its well known chemical reactivity, especially when it comes in contact with water. It also becomes radioactive and must be kept hot enough that it will not solidify.
The outstanding heat removal properties of liquid sodium offer an attraction in the form of enhanced safety. If the cooling system is designed as a “pot,”[x] it can be configured in such a way that it can cool the reactor (from the point of a recent scram), even without forced pumping. This basically means that, if something were to go very wrong, the reactor would scram and could then dissipate its decay heat, even if there was no forced circulation of the liquid sodium.
Once the choice was made, new contracts were let and all parties started to work on the LMFBR.
Having probably made the correct decision, the government design research programs turned to parametric studies of the, sodium cooled, design. By the time we reached this stage, I had learned how to do the core design work and found myself in the very enjoyable job of designing LMFBR cores full time. We got finally got better core physics codes and could be much more confident of our calculations. One of the particularly important parametric studies was to determine the optimum shape of the core. In a light water reactor, the core is a slightly tall cylinder, but that design was not necessarily optimum for a breeder. The base design is the compact cylinder, which is one that has a height that is approximately equal to the diameter. By designing cores that are taller and shorter, the behavior characteristics could be compared and an optimum identified.
The compact cylinder is the design which minimizes neutron leakage from the active core.[xi] If the reactor did not have a blanket, the compact cylinder would be attractive from a neutron conservation perspective. But in the presence of radial and axial blankets, neutrons which are lost from the core have a high probability of being absorbed by U-238 in the blankets. A tall core would obviously increase leakage in the radial direction, while a short one would increase leakage in the axial direction. The result of the parametric study was that the best performance could be obtained by a very short core, which we called the “pancake” design. The pancake core was also attractive because shorter fuel assemblies had desirable mechanical and coolant properties (higher stiffness, ease of fabrication and handling, fewer cooling concerns, etc.). Since the early days of these parametric studies, breeder cores have remained somewhat squat, but not to the same extent as the first designs.
Before the fun was over, I spent some time studying sodium flow channel blockages (of the kind that happened in the Fermi-1 accident). The basic finding was that partial blockages of up to 70% would not cause sodium boiling, but sudden blockages beyond that point would cause boiling, even if there was a very quick scram. One curious thing came out of this study: the temperature rise in the flow channel would be less, if the design were less inherently self-protecting (due to such things as Doppler coefficient reactivity feedback). This was based on the assumption that the reactor protection system would function properly and scram the reactor. With less reactivity feedback, the reactor would simply scram sooner, thereby reducing the amount of energy dumped into the coolant flow channel.
The breeder design work that I was involved with ended suddenly when government funding dried up. Work stopped and my company got out of the breeder business permanently. As I recall, the program was canceled primarily because of opposition, from the Carter administration, to fuel reprocessing. The LMFBR could not serve its intended purpose unless its discharged fuel could be reprocessed for recovery of plutonium. The Carter people believed that the generation of plutonium by the United States would increase the danger of nuclear weapons proliferation. This decision killed US work on the breeder, but did not stop other countries from pursuing their own programs. Russia, Japan, France, and others did not ban reprocessing and did not stop work on their breeder reactor programs.
For years after the first “end” of the breeder program, various projects were funded at national labs and elsewhere. In 1970 the Clinch River Breeder Reactor (CRBR) project was funded as a demonstration of the viability of the breeder concept. Funding went up and down, on and off throughout the next 13 years, when it was finally abandoned. CRBR was killed so many times that it is impossible to know when it was officially alive and when it was pretending to be dead. Somewhere around 1980, I recall walking into the office of a guy at the Department of Energy and asking him about photovoltaic research. His door had a sign on it indicating that this was the “photovoltaic section.” He laughed and told me that what they were really doing was working on CRBR, which was supposed to be dead at that point.
The fate of CRBR was due not only to the strong opposition from the Carter administration, but also to the variable economic appeal of breeders. CRBR was initiated relatively early in the development of breeder technology and suffered from old age before it was born. When the price of uranium was high, breeders were appealing and when the price went down breeders looked inefficient. The stops and starts and variation in funding of CRBR kept it from making rapid progress and reaching its goal of demonstrating the merits of LMFBR technology. The fact is that breeders have not become standard reactor designs in the countries where they have been built and operated. For this reason, the cancellation of CRBR was probably not as much of a loss as an energy alternative as it was a disappointment to those who wanted to see the technology developed. The ultimate success of breeders will be determined by their ability to economically provide energy and fuel for standard light water reactors. A revival will first require a change in direction which results in resumed construction of light water reactors in the United States. If that happens, the remaining requirement will be for a source of fissile materials that is less expensive than enriched uranium. That will depend on the future economics of producing reactor fuel via uranium enrichment as compared to the cost of developing reliable breeders and fuel reprocessing facilities.
[i] Fissile isotopes are those which can be caused to fission by the absorption of a thermal neutron. There are a number of fissile isotopes, but the ones of greatest concern in reactor design are U-235, and Pu-239. U-233 also has reactor applications, but those are beyond the scope of this discussion. Thermal neutrons are those which have been slowed down by collisions with moderator nuclei and have velocities which are in thermal equilibrium with their surroundings.
[ii] Natural uranium has a U-235 enrichment of 0.7204%.
[iii] The plutonium produced by the US for military purposes has come from graphite moderated thermal reactors at the Hanford, Washington site and from heavy water moderated reactors at the Savannah River, South Carolina site.
[iv] Each fission must produce enough neutrons to cause at least one additional fission. Those neutrons which do not cause fissions are either lost by leakage, or absorbed. Absorptions may occur in structural material, control rods, coolant, or fertile material, such as U-238.
[v] “Thermal reactors” are those which use neutron moderators to “soften” the neutron spectrum via collisions with relatively light weight moderator nuclei. The most common moderators are water, heavy water, and carbon.
[vi] In this case, “energy” is primarily a term used to describe neutron velocity. Neutrons are born at high energy and may slow down if they impart energy to nuclei by collisions.
[vii] B&W also codesigned the NS Otto Hahn with Germany.
[viii] There were a few other very minor players, but they had little importance.
[ix] Neutron energy is referred to by designers as “lethargy.” A neutron gains lethargy as it slows down by moderation. Lethargy is an inverse of energy.
[x] A configuration in which the core is completely below the level of all coolant pipes and which is capable of convection movement of the liquid sodium.
[xi] A fast breeder has an active core, which is loaded with fuel containing both fissile and fertile isotopes, and blankets, which are loaded with depleted uranium.