Multilevel Graph Partitioning for Three-Dimensional Discrete Fracture Network Flow Simulations

We present a topology-based method for mesh-partitioning in three-dimensional discrete fracture network (DFN) simulations that takes advantage of the intrinsic multi-level nature DFN. DFN models are used to simulate flow and transport through low-permeability fractured media subsurface by explicitly representing fractures as entities. The governing equations numerically integrated on computational meshes generated interconnected networks. Modern high-fidelity require high-performance computing multiple processors where performance scalability depends partially obtaining high-quality partition mesh balance work-loads minimize communication across all processors. structure naturally lends itself various graph representations, which can be thought coarse-scale representations mesh. Using this concept, we develop two applications multilevel partitioning algorithm In first, project based topology onto second, DFN-based projection is initial condition further refinement compare these methods with standard using graph-based metrics (cut, imbalance, time), computational-based (FLOPS, iterations, solver total run time. mesh-based comparable terms metrics, but time required obtain several orders magnitude faster partitions. computation-based show between both so, combination, partitions than partition. Moreover, uses DFN-partition solution provided cut imbalance values were close fraction turn, hybrid outperformed other