example, requires a detailed commercial design. The probability of various accidents depends more on the details of design than on the reactor’s generic characteristics. The only design that has been completely assessed for safety by the standards of the United States is the light water reactor.
Light water reactors can be designed for conversion of thorium to 233U. Although a light water reactor operated on a Th-U fuel cycle with reprocessing could achieve conversion ratios of about 0.7, the initial fissile inventory would require highly enriched uranium. (This highly enriched fuel may be ruled out by regulations to safeguard the fuel cycle.) The lifetime fuel requirements of a light water reactor on this fuel cycle could be 50–60 percent lower than those of an LWR on the once-through cycle, but the reactor would have to operate some time before enough 233U accumulated for reprocessing. Preliminary studies suggest that Th-U fueling of light water reactors would be uneconomical;32 however, the relatively modest changes required represent the most immediate opportunity to begin learning the engineering of Th-U fuel cycles. Spectral-shift-control reactors (SSCR’s) are essentially similar to pressurized-water reactors (PWR’s), one of the two light water reactors in use today. The coolant/moderator is changed during operation from heavy water to ordinary light water as the fissile content of the fuel in the core decreases. The development of this reactor consists mostly of conceptual studies, although a small pilot plant has been operated in Belgium.33
The light water breeder reactor, in spite of its name, is actually an advanced converter or break-even thermal breeder. Its design goal is to convert enough fertile material to fissile material to completely reload the core after accounting for fuel cycle losses. A demonstration LWBR achieved criticality in 1977, and experimenters anticipate that the reactor will be fully demonstrated by 1985.
The CANDU, an advanced converter designed in Canada to operate efficiently on natural (unenriched) uranium, supplies about 10 percent of the kilowatt-hours (kWh) generated by nuclear power in North America. This reactor employs heavy water as the reactor’s moderator and coolant, and allows on-line refueling. The introduction and installation of heavy water reactors offers a relatively near-term opportunity in the United States to improve uranium efficiency on the once-through fuel cycle. The use of slightly enriched uranium oxide, perhaps 1.0–1.2 percent 235U, would reduce the fuel requirements of a heavy water reactor 40 percent below the fuel requirements of a comparable light water reactor on a once-through fuel cycle. A CANDU-type reactor could also be designed to