Monte-Carlo simulation of radionuclide migration from a nuclear waste repository in the fractured crystalline rock formation

Crystalline rocks are one of the potential host rocks for an engineered nuclear waste repository (NWR). However, crystalline rock formations contain extensive fracture networks, which are challenging to characterise hydrogeologically (Neuman, 1987). Thus, numerical models incorporate fracture networks and hydraulic heterogeneity in a statistical manner. Conventional discrete fracture network (DFN) models need to define the fracture’s orientation, aperture and roughness, which are themselves uncertain and challenging to characterise even at a laboratory scale (Neuman, 1987; Cvetkovic et al., 2004). Furthermore, field data on the hydraulic and transport properties of fracture networks remain rare (Neuman, 1987). To tackle this, geostatistical principles can be employed, approximating fractured rock mass permeability as a stochastic effective continuum permeability field (Neuman, 1987). To represent the important role of preferential pathways, subsequent probabilistic radionuclide (RN) transport studies are essential (Cvetkovic et al., 2004). The present study adopted the Gaussian autocovariance function to approximate fractured granitic rock permeability fields with log-normal distribution derived from semivariograms and simulated the transport of radionuclides (Neuman, 1987). The study considered a two-dimensional domain of an NWR consisting of buffer, intact granite rock and fractured granitic rock with multiple realisations of fractured rock permeability to account for uncertainty. The granitic rock’s correlation length, mean and standard deviation of permeability were derived from packer tests (Neuman, 1987). The transport mechanisms of the retarding and mobile radionuclides (Sr-90, Cs-135, I-129, Cl-36) considered were advection, diffusion, sorption and decay (Poller et al., 2004). The RN transport was simulated in the open-source finite element code OpenGeoSys (OGS) for one million years using the Monte-Carlo framework (Bilke et al., 2019). The simulations indicated that Sr-90 and Cs-135 sorbed onto the buffer due to their retarding nature and did not reach the geological barrier in significant concentrations. Besides, Sr-90 decayed faster due to its shorter half-life, whereas Cs-135 strongly sorbed onto the buffer due to its high retardation coefficient. However, the mobile radionuclides (I-129 and Cl-36) were transported into fractured rock mass. The mean and confidence intervals of mobile radionuclides within the crystalline rock were observed to be within the safe dosage limit of the biosphere. Overall, the study quantified the uncertainty in dosage rates, and the proposed framework reduced the computations of transport simulations in an NWR to a greater extent than traditional DFN models. Additional studies are essential to improve the computational efficiency for large-scale three-dimensional modelling, and an increased number of realisations to gain confidence.