The Role of Matrix Diffusion in Transport Modelling in a Site-specific Performance Assessment: Nirex 97

摘要

Rock-matrix diffusion (RMD) is recognised as a potentially significant process that can act to retard the transport of radionuclides dissolved in groundwater flowing through fractures, and accordingly is included in performance assessments. Clearly, it is important that confidence can be built in the approach undertaken to modelling a transport process such as RMD. The work undertaken by the Nirex Safety Assessment Research Programme (NSARP) to determine relevant RMD properties is discussed in this paper. This work led to the development of the Nirex 97 [1] model of RMD, and assists in building confidence in the approach adopted in that performance assessment. The NSARP has adopted two approaches to investigate RMD: laboratory-scale experimental measurements and observations of RMD in natural geochemical systems. Quantification of the RMD properties of rocks has been achieved through a laboratory-scale aqueous phase diffusion experimental programme, which allows precise and accurate measurements of the diffusive properties of rock cores on the scale of 10-100 cm~3. It is recognised, however, that the process of acquiring and preparing the samples might affect their RMD properties: the importance of any such changes needs to be assessed. Natural analogue studies have been used to support the laboratory data by demonstrating that RMD operates over the time scale of interest in a performance assessment. However, larger uncertainties are associated with the quantification of diffusion coefficients from observations in such natural geochemical systems, because of uncertainties in both the time scale over which the process has operated, and the diffusive mobility and retardation properties of the trace element being studied. Further, it is difficult to prove that diffusion alone is responsible for the observed distribution of radionuclides in such natural analogues. Natural analogue studies therefore are not used a priori to quantify rock properties for input to assessment calculations; rather they provide information that, qualitatively, can be used to justify the operation of the RMD process. Information derived from NSARP studies was used to develop the simple model of RMD used in Nirex 97. This model incorporates various approximations, such as all the fractures in the BVG have the same properties and are arranged regularly. However, in a real system the fractures would be arranged irregularly and their properties would vary from fracture to fracture, and within a single fracture. Variant numerical modelling studies have therefore been undertaken, to assess the adequacy of the Nirex 97 approach. It was found that the approach was acceptable, and conservative.