Building a DFN
A Discrete Fracture Network is a synthetic population of fractures whose statistics match the ones you measured. It is how a fracture study stops being a description of one rock face and becomes something you can propagate into the volume behind it.
If the Interpretation tree has no Discrete Fracture Network node at all — as opposed to one whose commands are greyed out — this feature is not on your licence. See Feature availability.
Everything in the previous four pages stands on its own without it.
This page is the shape of the workflow and the judgement calls in it. The mechanics are in the DFN user guide and, for this outcrop-driven route specifically, DFN: Matching Interpretations.
Why not just type the parameters in
You can build a DFN by typing a mean orientation, a K, a length distribution and a P32 for each set. Nothing is wrong with that, and for a sensitivity study it is the right thing to do.
The outcrop-driven route is different in that every one of those numbers comes from something you measured, and stays tied to it:
| Typed in | Driven from the outcrop | |
|---|---|---|
| Mean direction & K | You choose | Fisher fit to your orientation groups |
| Trace-length distribution | You choose | Fitted from the measurements themselves |
| Intensity (P32) | You choose | Derived from observed P21 over the real outcrop |
| Spatial placement | Uniform | Biased toward where you actually mapped |
| Provenance | None | Recorded per set, and survives save/reload |
| Validation | None | Scored trace-against-trace, with a report |
That last row is the reason to care. A typed DFN cannot be wrong, because there is nothing to check it against. A conditioned one can be — and finding out that it is, is the useful result.
The sequence
Your fracture sets from step 3 are the input.
1. Generate. Select one or more orientation groups → Create DFN from Selected Groups. Each group becomes one fracture set. VRGS shows a preview of the Fisher statistics per group before committing — read it, and back out if a set has a K that says it is not really one set.
- Groups with fewer than 3 measurements are skipped.
- The domain comes from the project's map limits (the data extents), so a project with no real 3D extents will refuse rather than build a flat domain.
- Every set starts at a placeholder P32 = 0.3. That is a starting value, not a measurement, and the next step is what replaces it.
2. Condition the intensity. Ctrl-click the DFN, the outcrop mesh and the trace group → Override P32 from Selected Traces…. VRGS measures P21 from your traces over the real mesh area and converts it to P32 stereologically.
This is the step that connects the model to the rock. Skip it and you have a network with your orientations and an arbitrary number of fractures in it.
3. Condition the placement (optional). Ctrl-click the DFN and the mesh → Set Region Mask Source from Selection, then switch each set's Spatial Model to Region Mask and Regenerate. Fractures are then placed with a falloff away from the mapped surface, instead of uniformly through the domain — appropriate when your mapping covers part of the volume and you do not want to claim knowledge of the rest.
4. Validate. Ctrl-click the DFN, the mesh and the trace group → Score DFN against Selected Traces…. The network is cut with the outcrop, and the synthetic traces that result are compared against the ones you mapped, on three axes: length distribution (Kolmogorov–Smirnov), orientation (Fisher mean and K), and intensity (P21). A visual report opens.
The score is not there to be maximised. A DFN that scores poorly on length but well on orientation and intensity is telling you something specific — usually that the exposure truncated the long fractures, which is censoring rather than a model failure.
Validating Against Outcrop explains what to change when each axis fails, and which failures are real.
5. Iterate. Change what the report tells you to change, Regenerate, and score again. Show Outcrop-Constrained Fractures Only is worth knowing here: it hides everything that would not leave a trace on the exposure, which is the only part of the network directly comparable with what you mapped.
Where to go next
Once the network is conditioned, the analysis tools open up — connectivity graphs and topology, fracture–fracture and mesh intersections, slicers and cross-sections, propagation under a stress tensor, and permeability. All of it is in the DFN user guide.
Wrap-up
That is the whole workflow: traces off the model, orientations out of the traces, sets out of the orientations, intensity out of the traces and the mesh, and a network out of all of it — checked back against the outcrop it came from.
Worth collecting as you go:
- Analysis reports — pull the stereonets, rose diagrams, intensity charts and DFN validation figures into one document and export it to PDF.
- Getting Data In and Out — every export route, including GeoJSON for the whole interpretation.
- Virtual Field Trips — if the audience for the study is people rather than a reservoir model.