pure metallic waste is of a lower priority than that containing significant amount of cellulosic materials due to cellulose’s potential for producing complexing agents when degraded.

2.7.6. Inventory evolution

As mentioned in the introduction to this chapter, it is also necessary to look at the implications of assumptions, uncertainties and errors (which are often implicit and unstated) in any inventory in relation to their potential impacts on the repository design and the safety assessment.

2.7.6.1. Assumptions

Depending on the maturity of the national waste management programme, the extent of knowledge concerning wastes can vary enormously. However, what must be stressed is that inventories must be improved based on continuous feedback from ongoing safety analysis studies. For example, in the case where a safety analysis notes that certain nuclides dominate the dose arising from a given repository design, further effort can be devoted to identifying precisely which wastes contain the nuclides of concern. Following further analysis of, e.g., the chemical form of the radionuclides in question, it can be determined whether these wastes really do dominate or if it is merely an artefact of the assumptions that have been made in the safety analysis.

This can be illustrated in the following examples:

1.For the SA of a LLW repository, a base assumption was that all nuclides are instantly available for release from the waste, i.e., no account of the retention of

nuclides in the waste forms was taken into account. In the preliminary safety assessment, it was seen that 108mAg was, unexpectedly, one of the dominant

nuclides. On review of its origin, it was seen that this was arising from Ag/Cd/In control rods which would retain the 108mAg; therefore, this was taken into account

for the actual SA.

2.In the recent safety assessment for an ILW repository in Japan (JAEA, 2007), it was speculated that selenium in bituminous, high-nitrate wastes could be present in the poorly retarded selenate form rather in the highly retarded selenide form. If this were so, then selenium would be a significant component in the calculated dose. This has led to the conclusion that, in the next round of inventory definition, the form of the Se in these wastes should be better defined so that a more precise potential Se-induced dose can be calculated.

2.7.6.2. Errors

Errors occur in all work, but these can be reduced (and managed) by a strict quality assurance (QA) regime, e.g., by checking calculations (a generally straightforward exercise), confirmation of results by so-called benchmarking exercises (which can be an expensive exercise) and rigorous assessment of the origin of input data (what are the uncertainties associated with these data?), etc.

In the case of inventory definition, two examples can be given to highlight these points:

Situation 1: the misuse of the ORIGEN code:

For the generation of radionuclide inventory data, a worldwide accepted methodology is to use the ORIGEN fuel depletion code to calculate inventories in spent fuel (and the associated cladding). Originally, this code required a mainframe to run and, as a result, it was available to only a limited number of users and involved long running times. Since the evolution of desktop computers, the code has become widely available and now produces results in minutes instead of hours. This has resulted in its extensive use in many applications and, as a result, it is now being used outside its original design envelope.

As noted above, ORIGEN correctly calculates (with uncertainties) inventories in spent fuel and fuel cladding materials, but it was not designed to calculate inventories of actinides and fission products (resulting from U and Np impurities) in fuel channels and reactor pressure vessel components. Further, in areas where the neutron flux changes over small distances (e.g., in the control rods) or where there is difficultly in modelling the neutron fluxes (e.g., in the pressure vessel and in the bioshield), the use of a simple code like ORIGEN is strictly precluded.

In the example of fuel channels and core components, the fission product and actinide inventories resulting from trace impurities can result in their underestimation by a factor of between 5 and 40 (Von Gunten et al., 1999). This can lead, in the case of surface or near-surface disposal – where individual radionuclide activities are used to limit the amounts of radionuclides being emplaced in the repository – to the situation where these limits could potentially be unwittingly exceeded.

For these types of calculations, a different code must be used where the neutron fluxes can be varied (a code with at least three group cross-sections) and also where the crosssections can be weighted with ‘‘infinite dilution flux’’, i.e., outside fuel and hence no resonance absorption.

Situation 2: incorrect half-life data:

For all inventories, it is necessary (to avoid undue conservatism in the safety assessment) to include radionuclide decay, especially when performing calculations for the long timescales of concern to a HLW/SF/MOX repository. While examining uncertainties in their SF calculations, the Finnish implementer, Posiva, noted that the 79Se half-life value had been incorrectly determined in the original laboratory study and, further, the original data were never independently confirmed (see details in Vieno and Nordman, 1999). The original half-life of 6.5 104 years was modified to 1.1 106 years, significantly increasing the time where 79Se, a safety-relevant radionuclide, would be present in the repository. Although, in Posiva’s case, this had a minimal impact on the overall performance of the SF repository, it clearly indicated the potential for problems in other national inventories.