Atsushi Sawada

Evaluation of uncertainties originating from the different modeling approaches applied to analyze regional groundwater flow in the Tono area of Japan.

Qualitative evaluation of the effects of uncertainties originating from scenario development, modeling approaches, and parameter values is an important subject in the area of safety assessment for high-level nuclear waste disposal sites. In this study, regional-scale groundwater flow analyses for the Tono area, Japan were conducted using three continuous models designed to handle heterogeneous porous media. We evaluated the simulation results to quantitatively analyze uncertainties originating from modeling approaches. We found that porous media heterogeneity is the main factor which causes uncertainties. We also found that uncertainties originating from modeling approaches greatly depend on the types of hydrological structures and heterogeneity of hydraulic conductivity values in the domain assigned by modelers. Uncertainties originating from modeling approaches decrease as the amount of labor and time spent increase, and iterations between investigation and analyses increases.

Analysis of Open Crack in Rock by X-ray CT

Crack aperture distribution in rock induced by splitting is estimated by the X-ray CT method. This is a promising method that measures the crack aperture from X-ray CT images. The fundamental formulae for crack aperture measurement were described first as well as the necessary noise reduction and the data processing techniques. It was found that crack aperture distribution was accurately evaluated. It was also shown that this newly developed nondestructive method by X-ray CT is effective for estimating crack aperture in rock.

Study on the Estimation Error Caused by Using One-Dimensional Model for the Evaluation of Dipole Tracer Test

In-situ tracer tests are a valuable approach to obtain parameters for a performance assessment of nuclear waste repository. A one-dimensional model is simple and is commonly used to identify radionuclide transport parameters by fitting breakthrough curves simulated using the model to those obtained from tracer tests. However, this method can increase uncertainty and introduce errors in the estimated parameters. In particular, such uncertainties and errors will be significant when evaluating parameters for tests conducted in a dipole (two-dimensional) flow field between injection and withdrawal wells. This paper describes a numerical analysis investigation into the effects of various experimental conditions on parameters estimated using a one-dimensional model for cases involving tracer tests in a two-dimensional fracture plane. Results show that longitudinal dispersivity tends to be overestimated by the one-dimensional model analysis. This overestimation is the result of several factors: smaller pumping rate, larger dipole ratio, stronger heterogeneity of the fracture hydraulic conductivity, and greater orthogonally-oriented background groundwater flow. Such information will help us to better plan and design tracer tests at underground research laboratories located in both Mizunami in central Japan and Horonobe in northern Japan. Understanding appropriate experimental conditions will help decrease the uncertainty in the results of tracer tests.Copyright

Radionuclide migration analysis using a discrete fracture network model

This paper describes an approach for assessing the geosphere performance of nuclear waste disposal in fractured rock. In this approach, a three-dimensional heterogeneous channel-network model is constructed using a stochastic discrete fracture network (DFN) code. Radionuclide migration in the channel-network model is solved using the Laplace transform Galerkin finite element method, taking into account advection-dispersion in a fracture network, matrix diffusion, sorption in the rock matrix as well as radioactive chain decay. Preliminary radionuclide migration analysis was performed for fifty realizations of a synthetic block-scale DFN model. The total radionuclide release from all packages in the repository was estimated from the statistics of the results of fifty realizations under the hypothesis of ergodicity. The interpretation of the result of the three-dimensional network model by a combination of simpler one-dimensional parallel plate models is also discussed.