Modelling radionuclide transport in Brittle Fault Zones using a particle-based downscaling approach

A new Discrete Fracture Network (DFN) concept has been defined and implemented for possible use in futurethe next SKB (Sweden) and POSIVA (Finland) Safety Assessment (SA) studies of deep geological disposal of nuclear waste. In this improved DFN, deformation zones (BFZ, Brittle Fault Zones) are explicitly modelled as zones of individual fractures stochastically generated according to an exponential distribution of fracture centres around the core of each fault.

Due to the high computational cost, it is not feasible to use this DFN model to carry out all the particle tracking transport calculations required in a typical SA study. Thus, upscaling and downscaling approaches have been tested to investigate how the computational burden of the transport calculations can be reduced while preserving the signature of the underlying small-scale heterogeneity structure. The exercise has been carried out considering a single synthetic BFZ. First, the DFN model has been upscaled to an Equivalent Continuous Porous Medium (ECPM) and particle trajectories have been computed (this step is described in detail in a companion abstract). These particles trajectories have then been used to run radionuclide transport calculations using the time-domain particle tracking code MARFA (Painter and Mancillas,2013). Sub-grid heterogeneity is considered using a downscaling approach, in which sub-segment-based values of groundwater residence time and transport resistance are drawn from libraries and the correlation in velocity between adjacent segments is taken into account by using a Markov-type approximation (Painter and Cvetkovic, 2005). Specific to the current study is that DFN cubical models, representative of fracture intensity and orientation at both the core of the BFZ and at a distance of approximately one decay constant from the core of the BFZ, have been used to build the sampling libraries. The comparison between the results of the downscaled transport calculations (i.e. radionuclide breakthrough curves and statistics of residence time and transport resistance) and those obtained using the "synthetic reality" (i.e. the radionuclide transport calculations performed over the explicit DFN model) is here used to discuss the basis for a robust methodology to be used in future SA studies.