In this paper, we present a linked multicontinuum and crack tensor approach for modeling of coupled geomechanics, fluid flow, and solute transport in fractured rock. We used the crack tensor approach to calculate effe...In this paper, we present a linked multicontinuum and crack tensor approach for modeling of coupled geomechanics, fluid flow, and solute transport in fractured rock. We used the crack tensor approach to calculate effective block-scale properties, including anisotropic permeability and elastic tensors, as well as multicontinuum properties relevant to fracture-matrix interactions and matrix diffusion. In the mod- eling, we considered stress dependent properties, through stress-induced changes in fracture apertures, to update permeability and elastic tensors. We evaluated the effectiveness and accuracy of our multicon- tinuum approach by comparing our modeling results with that of three independent discrete fracture network （DFN） models. In two of the three alternative DFN models, solute transport was simulated by particle tracking, an approach very different from the standard solute transport used in our multicon- tinuum modeling. We compared the results for flow and solute transport through a 20 m x 20 m model domain of fractured rock, including detailed comparison of total flow rate, its distribution, and solute breakthrough curves. In our modeling, we divided the 20 m ~ 20 m model domain into regular blocks, or continuum elements. We selected a model discretization of 40 x 40 elements （having a side length of 0.5 m） that resulted in a fluid-flow rate equivalent to that of the DFN models. Our simulation results were in reasonably good agreement with the alternative DFN models, for both advective dominated transport （under high hydraulic gradient） and matrix-diffusion retarded transport （under low hydraulic gradient）. However, we found pronounced numerical dispersion when using larger grid blocks, a problem that could be remediated by the use of a finer numerical grid resolution, while maintaining a larger grid for evalua- tion of equivalent properties, i.e. a property grid overlapping the numerical grid. Finally, we encountered some difficulties in using our approach when element sizes were so small that only one or a few fractures intersect an element-this is an area of possible improvement that will be pursued in future research.展开更多
基金conducted within the context of the international DECOVALEX-2011 ProjectFinancial support for Berkeley Laboratory authors was provided by the UK Nuclear Decommissioning Authority (NDA) to the Lawrence Berke-ley National Laboratory through the National Energy Technology Laboratory, under the U.S. Department of Energy Contract No. DE-AC02-05CH11231funded by the UK Engineering and Physical Sciences Research Council (EPSRC),the UK Nuclear Decommissioning Authority (NDA), and Serco, through CASE studentship award 08002638
文摘In this paper, we present a linked multicontinuum and crack tensor approach for modeling of coupled geomechanics, fluid flow, and solute transport in fractured rock. We used the crack tensor approach to calculate effective block-scale properties, including anisotropic permeability and elastic tensors, as well as multicontinuum properties relevant to fracture-matrix interactions and matrix diffusion. In the mod- eling, we considered stress dependent properties, through stress-induced changes in fracture apertures, to update permeability and elastic tensors. We evaluated the effectiveness and accuracy of our multicon- tinuum approach by comparing our modeling results with that of three independent discrete fracture network （DFN） models. In two of the three alternative DFN models, solute transport was simulated by particle tracking, an approach very different from the standard solute transport used in our multicon- tinuum modeling. We compared the results for flow and solute transport through a 20 m x 20 m model domain of fractured rock, including detailed comparison of total flow rate, its distribution, and solute breakthrough curves. In our modeling, we divided the 20 m ~ 20 m model domain into regular blocks, or continuum elements. We selected a model discretization of 40 x 40 elements （having a side length of 0.5 m） that resulted in a fluid-flow rate equivalent to that of the DFN models. Our simulation results were in reasonably good agreement with the alternative DFN models, for both advective dominated transport （under high hydraulic gradient） and matrix-diffusion retarded transport （under low hydraulic gradient）. However, we found pronounced numerical dispersion when using larger grid blocks, a problem that could be remediated by the use of a finer numerical grid resolution, while maintaining a larger grid for evalua- tion of equivalent properties, i.e. a property grid overlapping the numerical grid. Finally, we encountered some difficulties in using our approach when element sizes were so small that only one or a few fractures intersect an element-this is an area of possible improvement that will be pursued in future research.
