Key Takeaways
- Fracture spacing and frequency logging is one of the best-validated AI applications on the whole poster. It has a published head-to-head comparison against expert human logging.
- Datarock Core (part of IMDEX) uses a neural network to detect and classify fractures directly from core photography.
- The Carrapateena copper-gold mine study validated this against expert geotechnical logging. It used a real deposit and a real comparison, not a synthetic benchmark.
- This is a “Yes”. It is the clearest example on the list of AI matching, not just approximating, expert human judgment in a specific geotechnical task.
TL;DR
To log fracture spacing with AI, run your core photos through a neural-network tool such as Datarock Core. It marks each fracture along the core and returns spacing and frequency logs automatically. Geotechnical engineers then review the output and compute RQD and fracture frequency for design.
What Is Fracture Spacing?
Fracture spacing is the distance between natural breaks along a drill core. Fracture frequency is the inverse view: the number of natural fractures per metre. Both describe how broken the rock mass is.
These numbers matter because they control rock mass strength. They feed standard geotechnical parameters such as RQD (Rock Quality Designation) and fracture frequency (FF). RQD is the percentage of a core run made up of intact pieces at least 10 cm long. Engineers use these values in rock mass classification, slope design and support design.
Logging fracture spacing by hand is labor-intensive and subjective. A geologist counts and classifies fractures along the core, and must separate natural breaks from those caused by drilling or handling. As a result, different loggers can produce meaningfully different results on the same core.
How Do I Automate Fracture Spacing Logging With AI?
The workflow runs from good images, through the model, to engineering review.
Capturing images for fracture spacing analysis
Start with consistent imagery. The model reads standard RGB photos, so lighting, resolution and tray layout all affect its accuracy. Our post on core photography for AI covers how to standardize this step.
Detecting fractures with a neural network
Datarock Core uses a neural network trained to detect and classify fractures directly from core photos. It returns fracture locations, spacing and classification along the core length. It does this at a consistency and speed that manual logging cannot match.
What sets this apart from many “AI in mining” claims is the evidence. Datarock published a direct comparison of its AI fracture analysis against expert geotechnical logging at the Carrapateena copper-gold mine in South Australia. The AI output held up against the human benchmark on a real operating deposit.
Reviewing results and computing RQD
Finally, geotechnical engineers use the output to compute RQD and fracture frequency for the geotechnical block model. They can spot-check or override the AI classification where core condition is ambiguous. Heavily broken or altered zones are the usual candidates for review.
Limits of AI Fracture Spacing Logging
The Carrapateena result is strong, but it comes from one deposit type. Carrapateena is an iron-oxide copper-gold system. Other settings, such as porphyry copper systems with intense clay alteration, can look quite different in photos. Therefore, validate on your own core before relying on the output.
Photos also have limits. A model sees only the core surface, so it cannot always tell a natural fracture from a mechanical break. Keep a human review step for those cases. The resulting logs feed downstream work such as bench slope design and rock strength prediction, so errors carry forward if left unchecked.
FAQs
How does AI measure fracture spacing?
A neural network scans core photographs and marks each visible fracture along the core. Software then converts those positions into spacing and frequency per interval. Engineers review the output before using it in design.
Is AI fracture spacing logging as accurate as a geologist?
In the published Carrapateena comparison, Datarock’s AI output held up against expert geotechnical logging. It is also more consistent, because the same model applies the same rules to every tray. Accuracy on your deposit still needs local validation.
What is the difference between fracture spacing and RQD?
Fracture spacing is the distance between fractures along the core. RQD summarizes the percentage of a core run made of intact pieces at least 10 cm long. Both come from the same fracture log, so AI detection supports both.
Can AI tell natural fractures from drilling breaks?
Partly. Models can learn visual cues that separate natural fractures from mechanical breaks. However, ambiguous zones still benefit from a geotechnical engineer’s review.
Try It With Geocluster
Will AI fracture logging hold up for your specific rock mass and alteration style? That is a comparative literature question, especially if your deposit differs from Carrapateena’s setting. Geocluster can help you work through it quickly. Read how the Geocluster Research Harness works to see how.
