Derivation of consistent sorption and diffusion parameters and their uncertainties for compacted MX-80 bentonite

SKB is currently preparing license applications related to the deep repository for spent nuclear fuel and an encapsulation plant. Recommended values for the diffusion-available porosity (ε), effective diffusivity (De) and distribution coefficient (Kd), as well as the associated uncertainties are derived and documented in a SKB technical report (Ochs, M., Talerico, C., 2004. SR-CAN: Data and uncertainty assessment. Migration parameters for the bentonite buffer in the KBS-3 concept. SKB Technical Report TR-04-18, SKB, Stockholm) for a total of 38 elements and oxidation states. Based on these results, this contribution focuses on issues of consistency and on quantification of the uncertainties associated with each parameter. The importance of consistency is twofold. First, the above parameters are conditional in nature. To assure consistency of the conditions assumed for parameter derivation versus those considered in the safety analysis, all parameters have to be derived explicitly for the expected in situ conditions, which especially in case of Kd include the specific bentonite porewater composition. Extrapolation of sorption data from experimental to the PA-relevant conditions was done with the help of thermodynamic sorption models or through semi-quantitative scaling factors. De for all radionuclides and ε were evaluated as a function of density for positively, negatively and uncharged species. Second, any set of parameters (ε, De, Kd) selected for a given element needs to be internally consistent. To this end, each selected parameter set was used to calculate apparent diffusivities (Da). These were then compared to independent, experimental Da values. Uncertainties in input parameters can be related to two fundamentally different sources: (i) the uncertainties of the underlying experimental data and models, including uncertainties introduced by scaling to reference PA conditions; (ii) parameter uncertainties caused by uncertainties in the conditions expected for PA. The impact of such variability was evaluated by specifically deriving input parameters for the reference PA conditions as well as for several alternative sets of PA conditions.