by Dr Steve Rogers, Senior Principal Geoscientist at WSP
Designing a vent raise is never just about excavation, it’s about understanding and managing the geotechnical risks that come with it. For many years, kinematic assessments have been the go-to method for evaluating potential wedge failures that might develop along the walls or face of a raise. These conventional methods typically focus on identifying the largest possible wedge that could form, without considering the actual location of intersecting structures or the feasibility of the wedge forming. While this is useful for conservative designs, it often falls short when justifying the largely unsupported configurations commonly used in many raises.
This is where Discrete Fracture Network (DFN) modeling provides a contrasting approach. A DFN model is a discrete model of individual structures whose geometry (size, orientation, intensity) is defined stochastically from field data to build networks of structures that can be interrogated. Unlike traditional methods, DFN-based techniques don’t just look for the ‘worst-case’ wedge. Instead, they offer a probabilistic and spatially realistic picture of what’s actually likely to occur. By leveraging structural data from pilot boreholes, DFN models can:
- Reproduce the anticipated structural network along the entire raise length;
- Identify the spatial distribution and size of potentially unstable wedges;
- Predict where kinematic overbreak is most likely to occur.
The result? A much clearer, defensible, and data-driven assessment of raise stability that supports better design decisions and risk communication.
In today’s risk-based mining environment, this approach can be critical. Whether applied stochastically or deterministically, DFN modeling provides a robust framework for evaluating and managing geotechnical uncertainty, helping operators move beyond the ‘rule of thumb’ and into evidence-based design.
Three approaches to building vent raise DFN models
When it comes to building vent raise DFN models, there’s no one-size fits-all approach. The method you choose depends on your objective, whether it’s a quick first-pass evaluation or a detailed design-level assessment. Here are three common approaches:
1. Deterministic models
Fractures observed from the raise pilot borehole are projected outward with a conservative 25 m (82 ft) radius (i.e they have a known location and orientation in space, and a large arbitrary size). These fractures are analyzed to identify any problematic wedges. This is the simplest approach and considers that only the fractures seen in the pilot borehole are important. Of course, when the pilot hole path and the raise center line diverge, this deterministic approach does diminish in value.
2. Unconditioned stochastic models
As an alternative to the deterministic approach, a stochastic DFN can be generated using logged fracture intensity and orientation data from the pilot borehole, with fracture sizes guided by available mapping data. This model is then searched for potential wedge instabilities. It represents the most realistic depiction of the structural network over a prescribed length of raise, matching the global properties for different domains, but not the local properties of the network. Multiple realizations of the stochastic models are generated, producing a probabilistic distribution of unstable wedge sizes.
3. Conditioned stochastic models
This is a hybrid approach that combines the best of the first two methods. A stochastic DFN is generated, but any fractures intersecting the shaft-line borehole are replaced with actual logged structures at their true locations and orientations. This allows the model to capture both deterministic features and inferred (stochastic) structures, matching both the global properties of the fracture network and the location of observed structures in the pilot hole. With these options, there’s a workflow for every situation, from rapid assessments to rigorous design investigations. It’s about moving from possible to probable.
Visualizing and quantifying risk
One of the powers of DFN vent raise analysis is that running multiple realizations allows the wedges to be visualized and interrogated. Since this is a stochastic approach, we’re often more interested in probabilistic results, such as the 75th percentile or 95th percentile of a property, for design purposes.
By running a large number of realizations, we can post-process results to generate distributions for a range of geometric wedge properties. Typical properties include:
- Individual wedge apex distance: provides a probabilistic assessment of how far wedges extend away from the raise, informing decisions for unsupported raises and the length of rock bolts needed to anchor wedges.
- Individual wedge volume/mass: what is the volume or mass of individual wedges that may become free?
- Cumulative wedge volume: gives an idea of the total unstable volume to be handled.
By using the more realistic modeling of the DFN approach, we can really begin to work on what is probable, not just possible. This allows us to make more reliable designs, which protect parts of the raise that may be vulnerable, while leaving unsupported those zones anticipated to show minimal loss.
For more information: Get in touch with Steve on LinkedIn
