Modelling spatially variable fracture size in discrete fracture networks

Natural mechanical discontinuities in rock vary in scale from microscopic fractures to kilometer scale faults. When modeling structures using the discrete fracture network (DFN) method, the large fractures typically play a dominant role in many DFN applications such as flow modeling or kinematic analysis. However, it is sometimes desirable to include smaller-scale fractures around areas of interest such as boreholes and excavations. Computational cost prohibits a kilometer-scale model with centimeter scale fractures throughout the entire domain. A common compromise uses nested models, where smaller fractures are only generated around areas of interest. However, this can significantly increase the complexity of the modeling workflow. This paper presents a new method for creating DFNs with spatially variable size ranges within a single fracture generation run, crucially whilst allowing other primary fracture parameters to be spatially variable. This approach permits increased model complexity without incurring workflow complexity, allowing the DFNs to be consistent with conceptual models across multiple scales and locations, ultimately improving model performance. For example, the minimum size of fractures in the DFNs can increase proportionally to their distance from areas of interest - a feature absent from standard modeling techniques. Example applications of the method are presented to validate the approach.