14:15 〜 14:30
[AGE27-03] A parallel Green element method simulator for modeling groundwater flow in discrete fracture networks
キーワード：Green element method, discrete fracture network, groundwater flow numerical simulation
The discrete fracture network (DFN) is usually the main groundwater flow pathway for fractured rocks with low matrix permeability. An effective way to investigate the groundway flow in complex DFNs is using a numerical simulator given a reasonable hydrogeological conceptual model. Accordingly, this study developed a new simulation tool using the Green element method (GEM) for modeling groundwater flow in DFNs. GEM is based on the green function that reflects the system response to a stimulus from a singularity. Using Green’s second identity, the terms containing the areal integral of the hydraulic head Laplacian and Green function Laplacian can be replaced by line integrals along a triangular cell boundary. This study derived analytical expressions of these line integrals to facilitate calculation. By doing so, the mass exchange between two grid cells belonging to the same fracture plane or the mass exchange via the intersection segment where multiple fractures intersect can be reasonably estimated.
The GEM numerical simulator was written in python. In particular, a multi-processing module was developed to boost the computational efficiency to overcome python’s poor computational efficiency. Several verification cases have been evaluated for the robustness of the GEM code. The first case considered a single fracture and was verified against the analytical solution, whereas the other two benchmark examples were taken from the literature and were evaluated against numerical solutions obtained by the commercial software FracMan 7.9. The GEM solution was successfully verified against the analytical solution for the single fracture case, with the hydraulic head root-mean-square-errror (RMSE) less than 2×10-10 m. For the two benchmark examples, GEM solutions are very close to those obtained from FracMan, with the best RMSE of less than 10-3 m and decreasing with grid refinement. As for the computational efficiency, the second benchmark example reduced the computation time by 164% after activating the multi-processing module. In summary, this study has successfully developed a GEM code for solving the groundwater flow in DFNs. Future studies will consider the anisotropic and heterogeneous fracture properties and the mixed and complex boundary conditions.
The GEM numerical simulator was written in python. In particular, a multi-processing module was developed to boost the computational efficiency to overcome python’s poor computational efficiency. Several verification cases have been evaluated for the robustness of the GEM code. The first case considered a single fracture and was verified against the analytical solution, whereas the other two benchmark examples were taken from the literature and were evaluated against numerical solutions obtained by the commercial software FracMan 7.9. The GEM solution was successfully verified against the analytical solution for the single fracture case, with the hydraulic head root-mean-square-errror (RMSE) less than 2×10-10 m. For the two benchmark examples, GEM solutions are very close to those obtained from FracMan, with the best RMSE of less than 10-3 m and decreasing with grid refinement. As for the computational efficiency, the second benchmark example reduced the computation time by 164% after activating the multi-processing module. In summary, this study has successfully developed a GEM code for solving the groundwater flow in DFNs. Future studies will consider the anisotropic and heterogeneous fracture properties and the mixed and complex boundary conditions.