Nuclear Control Institute Comments on the
Department of Energy's
Supplement to the Surplus Plutonium Disposition
Draft Environmental Impact Statement
June 28, 1999
Comments on the Environmental Impact Analysis (Section 5 and Appendix P)
The Nuclear Control Institute (NCI) has long urged the Department of Energy (DOE) to conduct a thorough, accurate and honest assessment of the environmental, safety and health risks associated with irradiating mixed-oxide (MOX) fuel derived from warhead plutonium in existing light-water reactors. NCI believes it is essential that such information be given considerable weight in the development of DOE's strategies both for disposition of U.S. excess plutonium and for cooperation with Russia on their own disposition program. Reactor safety issues have not been given the consideration that they warrant in the formulation of disposition policy, as evidenced by the selection of the MOX-immobilization "dual track" in 1996 based on the rudimentary environmental analysis and flawed calculations of the 1996 Storage and Disposition of Weapons-Usable Fissile Materials Programmatic Environmental Impact Statement (S&D PEIS).
The calculations of severe accident consequences contained in the Supplement to the Surplus Plutonium Disposition Environmental Impact Statement appear to be somewhat improved compared with those presented in the S&D PEIS. However, the overall analysis remains grossly incomplete and inadequate. DOE's final analysis must be strengthened to improve its credibility both with the public and with the Nuclear Regulatory Commission (NRC), which in spite of its relatively late start in examining the safety issues associated with DOE's MOX plan now appears to be taking a more thoughtful approach than DOE. If DOE continues to refuse to address seriously the full array of MOX safety issues, it will be inviting regulatory delays when license amendments to use MOX are sought by the Duke Cogema Stone and Webster (DCS) consortium in the future.
1) Beyond-Design-Basis Accident Analysis
The results of the beyond-design-basis accident analysis contained in the Supplement are substantially different from those provided by DOE in the S&D PEIS. This is apparent from the information provided in Table 4.28-9. Yet, there is no discussion in the text that explains the reasons for the different results of the two calculations. In addition, the table is misleading in not mentioning the fact that the S&D PEIS results were obtained for a full MOX core, while the Supplement calculations are based on a 40% MOX core.
The S&D PEIS calculations, which are cited in the Draft Surplus Plutonium Disposition EIS, indicate that for three out of four severe accident scenarios considered, the number of latent cancer fatalities (LCFs) that would result would be 3%-7% smaller for a full MOX core than for an LEU core. For the remaining accident scenario evaluated (late containment failure), the number of LCFs would be 8% higher for a MOX core.
The calculations in the Supplement give nearly diametrically opposite results. The three accident scenarios which were found originally to have less severe consequences for MOX cores than for LEU cores are now shown to have more severe consequences, with increases in LCFs of 1%-15% relative to LEU cores. In contrast, the one accident which was found in the S&D PEIS to have more severe consequences for MOX cores than for LEU cores, late containment failure, is now predicted to have less severe consequences for MOX cores at Catawba and McGuire, but more severe consequences for North Anna.
For North Anna, at first glance it appears that the result for late containment failure (a 9% increase in LCFs) is essentially unchanged from the S&D PEIS prediction of an 8% increase. However, taking into account the fact that the S&D PEIS results were obtained for a full MOX core, while the Supplement calculations are based on a 40% MOX core, it is clear that the new calculations indicate a MOX impact 2.5 times more severe than that implied by the S&D PEIS results.
The revised results provided in the Supplement are consistent with those estimated by NCI in a report released in January 1999 (attached), which found for a generic light-water reactor that the number of LCFs resulting from a severe accident with early containment failure or bypass would be approximately 28% greater for a 1/3-MOX core than for an LEU core as a result of radiation exposures incurred within one week after the accident. The chief difference between the NCI calculation and that of the Supplement is that the latter assumed that americium-241 (Am-241) would be removed from the plutonium via an aqueous separation process (so-called "plutonium polishing") prior to its fabrication into MOX fuel. However, at the time the NCI report was written, the baseline plan was to use only dry processing of plutonium feed, which would not remove the americium. NCI is revising its analysis to consider the effect of americium removal on its results. Preliminary results indicate that for a 40% MOX core with americium removal, the predicted number of excess LCFs is about 25% smaller than that originally estimated (for a 33% core without americium removal) or an increase of about 21% compared to an LEU core. Therefore, NCI's estimate and DOE's upper bound estimate are moving closer together.
However, many problems remain with DOE's analysis and presentation of data. These will have to be corrected and/or explained more fully in the final document. These include:
a) The results of calculations of population doses resulting from severe accidents are presented in the Supplement without sufficient detail to permit verification by independent analysts. The modeling of population dose in computer codes like MACCS 2 depends strongly on assumptions such as the time period of exposure considered, the cleanup standards, details of the evacuation and a whole host of other parameters. In general, the uncertainties associated with these calculations grow larger as longer time periods are considered. DOE must provide all the input parameters used in the calculations to facilitate independent public review.
Such information may shed light on some of the divergent results between sites, such as the reason why the MOX/LEU LCF ratios are smaller for Catawba and McGuire following a late containment failure accident, but larger for North Anna. These differences may be due to the use of results of the Independent Plant Evaluation (IPE), which have not been thoroughly reviewed by NRC. Because different utilities used different assumptions in developing their IPE submissions, the results may not be consistent for different plants. For instance, the frequencies of early containment failure at Catawba and McGuire given in the Supplement are smaller than that of North Anna, despite the fact that Catawba and McGuire have ice-condenser containments which are inherently more prone to failure in severe accident conditions.
Also, the reasons for the wide variation in MOX/LEU ratios depending on the particular type of severe accident must be discussed. NCI's analysis did not find such a large difference between early containment failure and containment bypass accidents.
b) There is an obvious error in the calculations in the Supplement which must be corrected in the final version. It is apparent from a comparison of population doses and LCFs in Tables 4.21-10 to 4.21-12 that a risk coefficient of 5x10-4 LCF/person-rem was used for all the calculations. This is inappropriate because it assumes a dose and dose-rate effectiveness factor (DDREF) of 2 is applicable for the entire affected population. However, this is clearly not the case, because many exposures following a severe accident will involve high doses delivered in short periods of time at rates far exceeding the threshold below which a DDREF of 2 is believed to be applicable (i.e. below 10 millirem per minute, according to UNSCEAR). DOE must revise its calculations so that the number of LCFs expected among those experiencing higher doses and/or dose rates are properly estimated using a DDREF of 1.
c) The calculations employ the MACCS 2 code developed by Sandia National Laboratories. NCI discovered a major error in this code which has a large impact on calculations of the consequences of severe accidents. Sandia altered the code and provided a corrected version to NCI. DOE should also use the corrected version for its final calculations.
d) The MOX/LEU ratios for fission product core inventories are remarkably similar to those used in the S&D PEIS, when adjusted for the different MOX core fraction. This leads one to surmise that Oak Ridge National Laboratory did not recalculate all fission product ratios for input into the Supplement, but only those for the actinides, and used the AP-600 ratios for the fission products. NCI has pointed out that the S&D PEIS ratios are not appropriate for use in the Supplement because they were obtained from an analysis of the Westinghouse AP-600 LWR, a reactor that has not been built and will not be used for plutonium disposition, rather than from an analysis of the designs of the existing reactors that will use MOX. Moreover, some of the fission product ratios are of questionable validity, such as that for Cs-134. The ratio of the inventory of Cs-134 in a 40% MOX core to that in a full LEU core is given as 0.85 in Table K-2 of the Supplement. This corresponds to a full MOX to full LEU ratio of 0.63, which is close to the value of 0.65 originally used in the S&D PEIS. NCI has been unable to reproduce such a low MOX/LEU ratio for Cs-134 in repeated ORIGEN-S runs. The value obtained by NCI is 0.96. (Incidentally, the value for this ratio given in the 1975 NRC Generic Environmental Impact Statement on the use of Mixed-Oxide Fuel in Light-Water Reactors [GESMO] is also 0.96.)
2) MOX Fuel Performance and Severe Accident Issues
The Supplement is silent on the question of MOX fuel performance, and in particular makes no mention of serious unresolved issues associated with the potentially inferior behavior of MOX fuel in certain severe accidents such as reactivity insertion accidents (RIAs) and loss-of-coolant accidents (LOCAs). These issues will surely be prominent in MOX licensing proceedings.
The Supplement assumes that all accident frequencies will remain unchanged by the substitution of MOX for LEU in existing LWRs, and references statements to this effect in the 1995 plutonium disposition study by the National Academy of Sciences (NAS). However, the NAS discussion was very general and did not examine in detail the following issues. These questions must be addressed in the Supplement so that the public can be informed about the numerous unresolved issues associated with MOX fuel performance in severe accidents.
MOX fuel produced via the MIMAS process, which will be the one used by the DCS consortium, is heterogeneous. It contains plutonium clusters (some of which have diameters of several hundred microns) which act as "hot spots," achieving much higher local burnups than occur in LEU fuel. For a fuel rod with an average burnup of 50 MWD/kg, the plutonium-rich clusters in MIMAS fuel achieve burnups of up to 200 MWD/kg.
The locally high burnups in plutonium-rich clusters result in the formation of high-porosity regions which allow fission gas to escape from the interior of fuel pellets. In addition, MOX fuel has a thermal conductivity approximately 10% lower than LEU fuel, resulting in centerline temperatures about 50 C greater. These two effects cause greater fission gas releases to occur in MOX fuel than in LEU fuel at similar average burnups. Above about 35 MWD/kg, the fission gas release in MOX fuel rods rises to several times that of LEU fuel at the same burnup. Another troubling observation (from recent experiments at the Halden reactor in Norway) is that while fission gas release in LEU fuel ceases when the fuel temperature is lowered below the threshold of onset, the same is not true for MOX fuel.
The increased fission gas release and higher temperature of MOX fuel rods can affect the severity of some accidents such as RIAs and LOCAs. The Rep-Na7 RIA test at the Cabri reactor in France, performed on a fuel rod that had been irradiated for four annual cycles and had a burnup of 55 MWD/kg, resulted in a "very severe failure" which caused the test channel to become almost completely blocked. This failure was unique because the fuel cladding did not have any important corrosion, unlike the LEU rods which failed in the same test series. As a result, according to those who conducted the experiment, "a MOX effect must be considered in this case."
NCI acknowledges that the plan of DCS is initially to irradiate MOX fuel for only two 18-month cycles, whereas some LEU assemblies are now irradiated for three 18-month cycles. However, the Supplement should detail the exact fuel management scheme that will be used and specify the maximum MOX assembly and rod burnups that will occur under this scheme.
The maximum burnup to which DCS is initially seeking authorization to take MOX fuel, 50 MWD/kg, is above the maximum MOX rod burnup that is currently permitted in France (about 47 MWD/kg), and is in a region where the rods' resistance to RIAs is clearly in question. Moreover, DCS refuses to preclude eventually irradiating MOX fuel to the same maximum burnup to which it currently irradiating LEU (with maximum rod burnups well over 60 MWD/kg). It is acknowledged in France that the current generation of MIMAS fuel must be modified and improved before such high burnups can be achieved. DCS should specify in detail how it is going to take into account future fuel modifications in its fuel qualification program.
The issue of MOX fuel performance in LOCAs is one which NRC has highlighted as a concern. Increased fuel and cladding temperature due to the lower thermal conductivity and higher average linear power of MOX assemblies, as well as the possibility of fuel-clad gap reopening due to the increased fission gas release, could enhance the clad oxidation rate during a LOCA and increase the severity of the accident. DOE should address this concern and its proposed LOCA mitigation strategy in the Supplement.
There are also disturbing indications that the fission gas release dynamics of MOX fuel may lead to enhanced releases of volatile and semi-volatile radionuclides during the early stages of core degradation compared to LEU fuel. This could have an effect on the consequences of some accidents, both design-basis and beyond design-basis.
3) Spent Fuel Management
The Supplement claims that the MOX program will not "impact spent fuel management" at the reactor sites, even though it predicts that additional spent fuel assemblies will be generated over the course of the campaign. However, the heat generation of spent MOX fuel will be greater than that of spent LEU fuel. NCI's calculations indicate that for two-cycle spent MOX fuel, the heat generation rate will be more than twice that of two-cycle LEU fuel soon after discharge and will remain at that level for many years. The Supplement should discuss how DCS can accommodate this incremental heat loading in their existing spent fuel storage facilities.
In summary, NCI believes that DOE cannot make credible or defensible decisions on a plutonium disposition strategy without a much more complete analysis of outstanding reactor safety issues associated with MOX use. Only then can the risks and benefits of various disposition strategies accurately be determined. In our view, the uncertainties and risks associated with reactor irradiation of MOX are significant enough to warrant a reevaluation of the "dual track" strategy. More serious consideration should be given to utilization of an all-immobilization approach to achieve the "spent fuel standard," so that the risks of MOX irradiation can be avoided.
Edwin S. Lyman, PhD
What's New Pu Disposition Page Home Page