This paper presents a summary of the results obtained in the framework of the SPA (Spent fuel disposal Performance Assessment) project. The project was undertaken by ENRESA, E; GRS, D; IPSN, F; NRG, NL; SCK.CEN, B and VTT, FIN between May 1996 and April 1999. Devoted to the study of spent fuel disposal in various host rock formations (clay, crystalline rocks and salt formation), it notably had the objective to evaluate the long-term performance of different repository systems and to identify the most influential elements. The variety of concepts, sites and scenarios considered in the framework of this project provides a wide range of information from which some general conclusions can be drawn. Focusing on the work done in the case of granite host rock formations, this paper describes the various approaches adopted and states the main sources of differences. It particularly stresses the differences related to the geosphere and biosphere modelling. For the geosphere modelling, ENRESA, GRS and VTT use one dimensional discrete approaches to model the migration of contaminants through the geosphere taking into account for matrix diffusion, whereas IPSN uses a three dimensional continuum approach based on a single porosity model. The comparison of the biosphere conversion factors shows the high influence on the calculated radionuclide dose contributions that can results from biosphere modelling assumptions. It notably points out the differences existing between a simplified “water drinking” approach as implemented by VTT and a more classical one in which a wider range of exposure pathways are taken into account. With regards to the results obtained, several common trends can be identified. Among the fission and activation products, 129 I is found to play a leading role in most of the calculation cases. The contribution of 14 C, 36 Cl, 79 Se, 126 Sn and 135 Cs also appears to be potentially high but the variability from one participant to another or from one calculation to another may be sometimes large. Among heavy nuclides, 226 Ra is commonly recognised as one of the most important dose contributor. Among the others heavy nuclides, daughter nuclides in the tail of the decay chains (notably 229 Th, 230 Th and 231 Pa) are also of relative importance.

Performance assessment of a deep geological disposal requires the definition of a series of assumptions of very different natures. The particular features of the site studied, the amount and the characteristics of the waste to be disposed of, or the definition of the various assumptions and parameters used to build the transfer model must be specifically defined. The multiple sources of differences related to the assumptions made by the teams involved in this international exercise make the interpretation difficult when a broad comparison of results obtained is considered. The work performed and the results obtained within the recent SPA project (Spent fuel Performance Assessment) however enabled to draw some general conclusions on the main radionuclides and parameters that potentially govern the radiological impact of a spent fuel geological repository. The particular importance of 129 I was notably highlighted as well as the possible role of transuranic elements. For 129 I, matrix degradation rate and dilution were found to have a predominant influence on the calculated dose. For transuranic elements, disposal system was found to delay the arrival of activity into biosphere over a very long period of time. A relatively wide range of breakthrough times were however observed depending on sorption modelling in geosphere in particular. As for transuranic relative dose contribution, a determining influence was played by solubility limit values.

The main objective of the EVEREST project is the evaluation of the sensitivity of the radiological consequences associated with the geological disposal of radioactive waste to the different elements in the performance assessment. Three types of geological host formations are considered: clay, granite and salt. The sensitivity studies that have been carried out can be partitioned into three categories according to the type of uncertainty taken into account: uncertainty in the model parameters, uncertainty in the conceptual models and uncertainty in the considered scenarios. Deterministic as well as stochastic calculational approaches have been applied for the sensitivity analyses. For the analysis of the sensitivity to parameter values, the reference technique, which has been applied in many evaluations, is stochastic and consists of a Monte Carlo simulation followed by a linear regression. For the analysis of conceptual model uncertainty, deterministic and stochastic approaches have been used. For the analysis of uncertainty in the considered scenarios, mainly deterministic approaches have been applied.