A new method for conditioning DFN models to field data such as outcrops, tunnel traces and borehole fracture logs is presented. The method generates DFNs that honor local observations whilst maintaining the specified fracture size and intensity statistics.

This conditioning workflow has two main steps. First, an unconditioned DFN is generated and all fractures intersecting specified sampling objects such as boreholes and rock faces are removed. The second step is to replace the removed fractures with conditioned fractures. The generation of conditioned fractures starts by defining a probability field of all potential fracture centers that would fit the observed data whilst honoring the limits of the fracture size model. Next, the potential fracture center points are weighted according to the underlying fracture size distribution. Conditioned fractures are then placed at a center point stochastically sampled according to these weights. This process is repeated until all observed data have an associated fracture.

This conditioning algorithm can match multiple observations with a single fracture or include negative constraints for preventing modelled fractures from intersecting sampling objects at locations where no traces are observed. In addition, conditioning is possible with grown DFN models which generate fractures dynamically, allowing for interactions such as terminations to occur. This functionality means the conditioning method provides a powerful tool for generating more geologically-realistic DFNs that can capture complex fracture patterns and match observed field data.

This content is only available via PDF.
You can access this article if you purchase or spend a download.