Edwin S. Lyman, PhD

Scientific Director

Nuclear Control Institute

October 19, 2000

Overview

In the summer of 2000, the U.S. and Russia signed an agreement to each dispose of thirty-four metric tons (MT) of plutonium withdrawn from nuclear weapons programs. The U.S. intends to implement the agreement in part by converting twenty-five MT of plutonium into nuclear fuel (known as mixed-oxide or MOX fuel) and irradiating it in the Catawba and McGuire nuclear plants, operated by Duke Energy. The MOX fuel will be substituted for a portion of the low-enriched uranium (LEU) fuel that these reactors currently use. DOE chose Catawba and McGuire in 1999 because Duke Energy was a member of the only qualified consortium to submit a bid for the MOX contract.

Under the agreement, both sides are committed to taking "all reasonable stepsto begin operation of disposition facilities" no later than December 31, 2007. The current plan is to load two MOX test assemblies in McGuire Unit 2 in October 2003 and to load the first full batch of MOX fuel in Catawba Unit 2 in October 2007. After that, MOX fuel will be loaded during each refueling outage at all four Catawba and McGuire units until the entire quantity of plutonium is used, which is projected to take about fifteen years. At the height of the campaign, up to 40% of the cores of these plants will consist of MOX fuel. The city of Charlotte, NC is in close proximity to the four reactors that will use MOX fuel, all within twenty miles of downtown.

In 1999, the Nuclear Control Institute (NCI) released a study documenting that Duke Energy's plan to use MOX fuel in Catawba and McGuire would increase the risk to the public of injury in the event of a severe nuclear accident --- that is, an accident in which the fuel melts and the reactor containment building ruptures, releasing large quantities of radioactive materials into the environment. This is because compared to reactors using only LEU fuel, reactors using MOX fuel contain greater quantities of certain highly radiotoxic elements, such as plutonium, americium and curium.

NCI found that if Duke Energy carries out its plan to replace 40% of the LEU fuel now used in Catawba and McGuire with MOX fuel, the number of individuals who would die of cancer as a result of exposure to radiation following a severe accident would increase by 25%. Thousands of additional deaths could result. NCI concluded that this increase in risk associated with MOX use was unacceptable.

In response to NCI's report, DOE acknowledged that the number of deaths that would occur from a severe accident at Duke Energy's plants would be higher when MOX fuel is used, but argued that the increase was unimportant in light of information provided by Duke Energy showing that the overall accident risk was low.

However, an analysis has been recently released by the U.S. Nuclear Regulatory Commission (NRC), the federal agency that regulates nuclear power plants, that calls into question the level of protection to the public from potential accidents at Catawba and McGuire. This new NRC study has found that pressurized-water reactors (PWRs) with "ice condenser" containments, a category that includes Catawba and McGuire, are "substantially more sensitive to early containment failure" than other types of PWR containments. This means that in the event of a severe accident in which the reactor fuel melts, the risk that the reactor containment will rupture and large releases of radioactive materials into the environment will occur is significantly greater at Catawba and McGuire than at PWRs with other types of containment.

The implications of this result for the MOX fuel program are clear. The four

Catawba and McGuire units are among only nine PWRs in the U.S. that are exceptionally vulnerable to early containment failure. It is precisely this type of accident that could cause expulsion of the greater quantities of plutonium and other actinides in MOX cores into the environment. Therefore, it is highly imprudent to load MOX fuel into these reactors.

The results of NRC's analysis make clear that DOE's conclusions regarding the safety of the MOX program are outdated and need to be revised. For example, the overall probability of containment failure at McGuire in the event of a severe accident estimated in the NRC report was seven times greater than the value that DOE used in its Surplus Plutonium Disposition Environmental Impact Statement (EIS). According to DOE's obligations under the National Environmental Policy Act (NEPA), it must carry out a supplemental EIS in the event that "significant new ... information relevant to environmental concerns" becomes available concerning actions that it has proposed. The new NRC study clearly qualifies as "significant new information."

Vulnerabilities of Ice Condenser Containments

Nuclear power plants in the U.S. are required to have robust reactor containment buildings. The main purpose of these structures is to prevent the release of large quantities of radioactive materials in the event of a reactor core meltdown. In the aftermath of the 1986 Chernobyl accident in the former Soviet Union, the nuclear industry maintained that such a severe accident could never happen in the U.S. because U.S. reactors, unlike the Chernobyl reactor, were equipped with containments.

However, not all containments offer equal protection. Most pressurized-water reactors (PWRs) in the U.S. have "large dry" containments, which are typically massive concrete structures with walls several feet thick. Catawba and McGuire, on the other hand, are among a handful of PWRs worldwide with "ice condenser" containments. These are typically thin steel shells that have only half the volume and failure pressure of large dry containments. To compensate for the reduced strength of their containment buildings, ice condenser plants are equipped with "ice beds." These consist of baskets filled with blocks of ice that are supposed to cool and condense steam flowing past them during a core-melt accident, reducing the threat that the containment will become overpressurized and rupture from the rapid generation of steam.

However, even if the ice condensers do work as they are supposed to (which in itself is a questionable proposition), containment failure can still occur as a result of the combustion of hydrogen gas, which would be generated in large quantities during severe accidents when the metal cladding on fuel rods reacts with coolant water. During the Three Mile Island 2 (TMI-2) accident in 1979, a large amount of hydrogen was released to the containment and burned, although the pressure increase did not lead to rupture of TMI-2's large dry containment. The ice condensers not only cannot reduce the risk of hydrogen combustion but also can actually increase it, because they divide the containment volume into small compartments where hydrogen gas can more readily reach explosive concentrations.

The seriousness of this issue is clear from the following data on the strength of containment buildings. The pressure that can be generated in the containment from hydrogen combustion can typically reach a value of about 110 pounds per square inch (psi). The average failure pressure of U.S. large dry containments is around 113 psi, whereas for ice condenser containments it is around 63 psi. Therefore, hydrogen burns can easily overpressurize and rupture ice condenser containments.

For this reason, after the TMI-2 accident, NRC required that ice condenser plants install hydrogen igniters, which are operator-initiated, AC-powered devices that are designed to burn hydrogen at a controlled rate before it reaches an explosive concentration.

However, the risk of hydrogen explosions in ice condensers has not been eliminated entirely by this requirement, since the hydrogen igniter systems now in use require AC power to operate. Therefore, in the event of a simultaneous loss of both off-site and on-site AC power supplies, known as a station blackout (SBO), hydrogen control is lost.

Earlier this year, the Nuclear Regulatory Commission (NRC) released a report that analyzed the risk of containment failure during severe accidents at reactors with "ice condenser" containments. The report, entitled Assessment of the DCH [Direct Containment Heating] Issue for Plants with Ice Condenser Containments, NUREG/CR-6427, finds that "no ice condenser plant is inherently robust to all credible hydrogen combustion events in a SBO accident." It also concludes that "ice condenser plants are at least two orders of magnitude [one hundred times] more vulnerable to early containment failure than other U.S. PWRs" as a result of hydrogen explosions during core melt accidents. This study, which was performed by Sandia National Laboratories (SNL) in Albuquerque, calculated that for accidents in which the hydrogen igniters were not available, such as SBOs, the probability that the containment would rupture as a result of hydrogen combustion is 34% for Catawba and 58% for McGuire. Using the same methodology, previous NRC studies found that the risk of containment failure at large dry containments is less than 0.1%.

SNL found that certain SBO accidents --- namely, those in which the reactor coolant system remains at high pressure at the time that the reactor vessel is breached by molten fuel --- the probability of early containment failure as a result of detonation of pre-existing hydrogen is nearly 100% for both Catawba and McGuire. This means that if one of these sequences were to occur, there would be little difference between the ice condenser plants and nuclear plants without containments like Chernobyl.

NRC and the nuclear industry continue to argue that accidents as severe as an SBO are so unlikely that the weakness of ice condensers is not a high-priority concern. However, an SBO actually occurred at the Vogtle plant in Georgia in 1990, during which the plant lost all off- and on-site power supplies for 35 minutes. Other plants have come quite close to an SBO. For instance, in 1996 Catawba lost off-site power for more than a day with one of the two emergency diesel generators unavailable. That means it was only one generator away from an SBO. NRC estimates that at that time, there was a 0.2% chance that the core of the reactor would have been damaged. In light of the SNL study, it is now known that this corresponded to a nearly one in a thousand chance of a Chernobyl-type accident.

According to Duke Energy's own data, provided to the NRC in its Individual

Plant Examination (IPE) submittals (probabilistic risk assessments done by licensees, without peer review), McGuire has a relatively high probability of experiencing an SBO. Factoring in this probability, NRC obtained a containment failure probability given core damage of 13.9% for McGuire. This result is nearly seven times greater than the value of 2.4% reported by Duke in the McGuire IPE.

Although this value exceeds NRC's guideline that containment failure probability should not exceed 10%, NRC argues that it is "consistent with a general objective" of 10%. However, this result does not take into account "external events" such as earthquakes or tornadoes. A tornado caused a loss of off-site power at the Davis-Besse plant in 1998, and one of the diesel generators became inoperable afterward. Such events are associated with a much higher SBO risk than internal transients. Therefore, the fraction of core damage scenarios that are also associated with SBOs would be much higher if external events were included.

For example, according to Duke Energy's own IPE data, the probability of an earthquake causing an SBO at Catawba is over ten chances per million per year. According to a recent NRC proposal, any accident sequence that had a probability of more than one chance per million per year would have to have an early containment failure risk of less than 10%. Catawba, with an early containment risk of 34%, would be in violation of this guideline based on the seismic risk alone.

Station blackout can also occur as a result of sabotage, which hasn't been taken into account in the analysis. For instance, during a recent NRC force-on-force exercise at the Oconee plant, also owned by Duke Energy, mock attackers were able to cut off-site power (this is always assumed to be the case, because the power lines are not protected), defeat the security force and cause core damage. However, the probability of a sabotage-induced SBO cannot be quantified. Therefore, the best line of defense in this case is to ensure that the containment will not fail.

The SNL report concludes that "all [ice condenser] plants, especially McGuire, would benefit from reducing the station blackout frequency or some means of hydrogen control that is effective in station blackouts," noting that the latter course would reduce early containment failure probabilities "by more than an order of magnitude in all plants and especially McGuire." However, according to the report, "previous cost/benefit studies generally do not justify the expense in providing hydrogen control in SBO because ... the SBO probability is a small fraction of the core damage frequency ...". This assumption has now been called into question.

NRC is in the process of reviewing its regulations on combustible gas control. NRC staff have recently proposed a requirement that ice condensers provide a means for controlling hydrogen in station blackouts unless it can be shown that the probability of a station blackout is acceptably low.

Meanwhile, Duke Energy has learned of the ice condenser report and is already raising doubts about its validity. Duke met with NRC staff on September 28 and vigorously opposed the idea that the installation of new equipment for controlling hydrogen gas accumulation in SBOs might be necessary.

Conclusions

Given the fact that the Duke Energy's ice condenser plants Catawba and McGuire in particular, are among the U.S. PWRs most vulnerable to an early containment failure, it is foolish to allow MOX to be used in them, especially since the consequences of such an accident would be increased with MOX fuel in the core. Even if NRC ultimately requires installation of new equipment in ice condenser plants, the uncertainties in risk calculations render questionable any probabilistic conclusions about their ultimate safety.

Return to What's New Page