In addition to the Generation III+ designs of commercial reactor vendors, the DOE is sponsoring R&D on advanced reactor systems at national laboratories and universities (see Box 8, p. 59). Two are thermal reactors and three are fast reactors that would use plutonium-based fuels. One goal of these designs — known as Generation IV — is greater safety. However, there is no basis for assuming that any of the five designs now under study would be significantly safer than today’s nuclear power plants.
First, Generation IV designs have little or no operating experience, so detailed computer models would be needed to accurately predict their vulnerability to catastrophic accidents. However, this project is still in its infancy, so developing and extensively validating computer models for each design will be a formidable task.
Second, all the designs use coolants that are highly corrosive under normal operating conditions, and will therefore require advanced structural materials that can perform well in extreme environments. This is true even for the Very High Temperature Reactor (VHTR), which uses inert helium gas as a coolant, as low levels of impurities in the coolant would be highly corrosive at the operating temperature of 1,000°C.138 Development of these advanced materials is speculative, and failure to meet the performance goals would translate into lower safety margins and higher operating costs.
Third, to reduce costs, Generation IV designs aim to reduce safety margins wherever possible. This is at odds with the fundamental concept of defense-in-depth, in which backup safety systems compensate for uncertainties in the performance of the main safety systems.
For example, one Generation IV goal is to eliminate the need for off-site emergency response plans, which are a critical component of defense-in-depth strategy. The confidence to take such an unprecedented step can come only from a wealth of operating experience, which is lacking for the new designs. And any new design will have to undergo the “break-in” phase of the aging curve, according to which higher failure rates are expected at the beginning and end of a plant’s lifetime. Accidents at U.S. reactors have conformed to this curve.139
Fourth, the Sodium-cooled Fast Reactor (SFR) and Lead-cooled Fast Reactor (LFR) have inherent safety problems because of their coolants. Lead-bismuth coolant is less reactive and has a higher boiling point than sodium coolant. However, it is extremely corrosive, and when irradiated produces highly volatile radioisotopes (polonium-210 in particular) that would be a challenge to contain even under normal operating conditions.
As noted, the use of liquid sodium as a coolant presents serious safety challenges. According to a 2002 Department of Energy report,140
It is also true that sodium as a reactor coolant has two major drawbacks: its chemical reactivity, and its positive void coefficient of reactivity in most plutonium-fueled applications. . . . There have been small sodium leaks (and small fires) at essentially every sodium-cooled reactor plant built; in some cases, several of them. These incidents, though, do not disqualify the coolant from further use.
The “void coefficient of reactivity” indicates how the reactor’s power output would change if steam bubbles (or voids) form in the coolant. Power increases if the coefficient is positive. Thus, if the core overheats and the liquid metal coolant boils, the reactor’s reactivity and power will rise rapidly. This intrinsic positive feedback can lead to a rapid increase in power and disrupt the core, while reducing the amount of time operators have to take mitigating action.
The NRC requires that reactors have a prompt negative feedback response to any increase in reactivity.141 Therefore, the NRC could not license an SFR with a positive sodium void coefficient under today’s guidelines.
Nonetheless, the NRC could make an exception. NRC staff concluded in the 1990s that “a positive void coefficient should not necessarily disqualify a reactor design,” provided the risk to the public remained low.142 Scientists at Argonne National Laboratory often argue that the EBR-II — an experimental SFR in Idaho that operated from 1961 to 1994 — was a “passively safe” reactor that shut itself down after a safety test, despite its positive void coefficient. However, the shutdown relied on expansion of the reactor’s metal fuel elements as they heated, which is not “prompt inherent nuclear feedback,” and cannot be relied on to compensate for increases in reactivity.
Design changes can reduce or eliminate the positive void coefficient in fast reactors. For instance, the 4S is designed to maintain a negative void coefficient over its entire operating cycle. However, such changes usually increase the amount of reactivity in control systems, and therefore raise the severity of other types of reactivity accidents.143 Whether there is an optimal design for fast reactors that can make their overall risk acceptable is far from clear.
Perhaps even more serious than the positive void coefficient is that, unlike most light-water reactors, fast reactors are not in their most reactive configuration under normal operating conditions. This means that an event that causes the core to become more compact — such as a core meltdown — could substantially raise reactivity, resulting in a rapid power increase that could vaporize the fuel and blow the core apart.144 Such an explosion — dubbed a “hypothetical core disruptive accident” — would be similar to the explosion of a very small nuclear fission weapon, with a yield comparable to that produced by a ton of TNT.
These problems are already severe for SFRs that use only mixtures of plutonium and uranium. However, the DOE ultimately plans to adapt its advanced recycling reactor to use fuels that also contain the highly radioactive actinides neptunium, americium, and curium (see p. 69), which tend to increase the severity of these reactivity problems. Designing cores for such reactors that can both effectively fission these actinides and be acceptably safe will be a major challenge.
Some new reactor designs represent the next evolutionary step for nuclear power, incorporating features intended to make the plants safer and more economical. These features, however, are largely untested in the field or have very limited operating experience. Other new reactor designs have operated only in cyberspace and have never experienced the trials and tribulations of real-world operation. The gremlins hiding in their designs have not yet been exposed, let alone exorcised.
138 See, for example, C. Cabet, A. Monnier, and A. Terlain, “Corrosion of high-temperature alloys in the coolant helium of a gas-cooled reactor,” Materials Science Forum 461-464 (2004): 1165–1172.
140 Nuclear Energy Research Advisory Committee and Generation IV international forum, “Generation IV roadmap: Description of candidate liquid-metal-cooled reactor systems report,” GIF-017-00, December 2002, p. 34.
141 The NRC’s General Design Criterion 11 states that, “the reactor core and associated coolant systems shall be designed so that in the power operating range the net effect of the prompt inherent nuclear feedback characteristics tends to compensate for a rapid increase in reactivity.” 10 CFR Part 50, Appendix A.
142 NRC, “Issues pertaining to the advanced reactor (PRISM, MHTGR and PIUS) and CANDU 3 designs and their relationship to current regulatory requirements,” SECY-93-092, p. 23.
Lisbeth Gronlund is a senior scientist and co-director of the Global Security Program at the Union of Concerned Scientists (UCS) in Cambridge, Massachusetts and a research affiliate in the Program on Science, Technology, and Society at the Massachusetts Institute of Technology. David Lochbaum is Director of the Nuclear Safety Project at the UCS. Edwin Lyman is a senior staff scientist in the Global Security program at the UCS. The text above is an excerpt from Lisbeth Gronlund, David Lochbaum, and Edwin Lyman, “Nuclear Power in a Warming World: Assessing the Risks, Addressing the Challenges” (Union of Concerned Scientists, December 2007); it is reproduced here for non-profit educational purposes.
var idcomments_acct = ‘c90a61ed51fd7b64001f1361a7a71191’;