Key Takeaways

  • Clay mineral prediction matters because clay content is a first-order driver of flotation, filtration and tailings behaviour. Getting it wrong in the block model has real downstream cost.
  • Corescan’s HCI and CSIRO’s HyLogger are the two dominant systems for measuring clay mineralogy from core at scale.
  • CSIRO’s MyLogger automates the interpretation of HyLogger spectra into a first-pass geological log. It is a released, freely available tool, not a research prototype.
  • This is a real “Yes” for AI today, not an emerging or speculative category.

TL;DR

For clay mineral prediction with AI, scan your core with Corescan or HyLogger and let automated interpretation classify the spectra. Tools like CSIRO’s MyLogger turn that output into a first-pass log without waiting for a spectral specialist. You then load the result as point data into your geomet block model.

What Is Clay Mineral Prediction?

Clays are fine-grained sheet silicates such as kaolinite, montmorillonite and illite. They often form through alteration or weathering of the host rock. Even small amounts can change how ore behaves in the plant.

Clay mineral prediction means estimating which clays are present, and how much, in every block of the deposit before it is mined. That estimate lets planners flag problem ore early. For example, swelling clays raise slurry viscosity, consume flotation reagents and slow down thickeners and filters. Therefore, an accurate spatial clay model is directly valuable rather than a nice-to-have.

How Do I Do Clay Mineral Prediction With AI?

The pipeline has three steps: acquire spectra, interpret them automatically, then load the result into the block model.

Scan core to measure clays

The data-acquisition side is well established. Corescan’s Hyperspectral Core Imager and CSIRO’s HyLogger both scan core at high spatial resolution. They detect the spectral signatures of specific clay minerals that are difficult to tell apart by eye. Our guide to hyperspectral clay models covers this scanning layer in more detail.

Automate the interpretation step

Historically, a specialist had to interpret hyperspectral mineralogy. That was a real bottleneck, because spectral mineralogy is not part of standard geoscience training. CSIRO’s MyLogger, released in 2023 and funded through the MinEx CRC, closes much of that gap. It is an automated workflow that combines CSIRO’s The Spectral Geologist (TSG) mineral groupings with its Data Mosaic boundary detection. As a result, it converts HyLogger thermal-infrared data into a first-pass geological log. It identifies major lithological boundaries within about a metre of manually logged contacts, according to the peer-reviewed MyLogger paper.

Feed clay mineral prediction into the block model

Finally, take the automated clay and alteration classification and load it as point data into your geomet block model. Estimate clay content per block with your usual geostatistical or machine-learning method. This skips a manual interpretation step and keeps the clay model in step with the drilling.

How Clay Models Connect to the Plant

A clay model is only useful if someone acts on it. In the plant, clay-rich feed changes how AI flotation control has to respond and how the thickening circuit performs.

Upstream, clays often sit inside the same domains as your alteration proxies, so the two models should agree.

Keep expectations realistic. Automated interpretation gives a first pass, not a final answer. A geologist should still review the classification, especially for new deposit types.

FAQs

Why does clay mineral prediction matter for processing?

Clays raise slurry viscosity, consume flotation reagents and slow down thickening and filtration. They also affect how tailings settle and consolidate. Knowing where clays sit in the block model lets planners blend or schedule around problem ore.

Which instruments measure clay minerals in drill core?

Corescan’s Hyperspectral Core Imager and CSIRO’s HyLogger are the two dominant core-scanning systems. Both use reflectance spectroscopy to detect clays such as kaolinite, montmorillonite and illite. Portable SWIR spectrometers are a cheaper option for point measurements.

What does CSIRO MyLogger do?

MyLogger turns HyLogger thermal-infrared data into a first-pass geological log automatically. It combines CSIRO’s TSG mineral groupings with Data Mosaic boundary detection. As a result, geologists get a rapid log without needing a spectral specialist.

Is AI clay mineral prediction reliable enough for mine planning?

It is reliable enough to use as a first-pass input, and the scanning technology is mature. However, check the automated output against logging and lab mineralogy, especially early on. Treat it as a strong signal that a geologist validates.

Try It With Geocluster

Comparing Corescan against HyLogger for your deposit type and clay minerals of concern takes real sourced research. So does understanding what MyLogger can and cannot do for you. That comparative work is exactly what Geocluster is designed to run.