How would you summarize the last couple of months for Life Cycle Geo?

Our 16-person team, mostly geochemists and data scientists, recently welcomed our first chemical engineer. We have been expanding our AI-based workflows, and have found that machine learning-based applications for water management seem to be among the most promising and in demand. 

How are you applying machine learning on the material side of the business and in building block models?

We have been refining an orebody knowledge tool for about five years in collaboration with the Colorado School of Mines and are now in discussions to create a more advanced version with a tier-one provider. Traditionally, block models only estimate ore volumes, but the modern mine also needs to track volumes and locations of materials based on their environmental, geo-metallurgical, and geotechnical properties. Our tool uses machine learning to integrate these datasets, acting as a front end to a block model.

Building a block model starts with drilling and collecting assay data for every interval. High-value measurements (like acid-generating potential, mill versus heap-leach performance, or rock weathering rates) are costly and time-consuming and thus collected at a frequency insufficient for block modeling. We created a workspace where specialists classify materials in a drill hole context and use machine learning to project these properties to the exploration assay database for every drillhole interval. The process, known as block model uplift or upsampling, produces complete datasets ready for platforms like Leapfrog or Datamine.

Can you tell us about your role in Ivanhoe Electric’s Santa Cruz PFS?

Our focus has been on water and materials to help them secure permits. We began by establishing baseline (pre-mining) water quality and analyzing groundwater well data from both the mine area and the surrounding region. By using an unsupervised machine learning approach, we gained a deeper understanding of the range of pre-mining water qualities. As Ivanhoe Electric drilled the ore body and nearby access areas, we collected solid samples and tested them for environmental characteristics such as acid-generating and neutralizing potential. Operationally, we also forecast the types of water quality they will need to manage during production, ensuring it stays within the mine footprint.

Do you see differences between large and small companies implementing AI?

We have noticed that more tier-one miners are open to using AI and machine learning to manage both materials and water. While smaller mining companies can also benefit from it, they usually implement it to a specific project or stage in the mine life cycle. 

How long does it take to implement AI in a mining project?

It depends on the problem and the dataset. Some projects can be implemented in a week; others take a month or two. There is also an iterative element and it is imperative that clients stay involved for project success. Too often, machine learning is approached as a one-and-done exercise: “Here is our data, build us a model, and we are done”. However, this approach rarely works. We get the best results when clients share site knowledge and their domain expertise, which really helps shape our conceptual model. We apply an iterative and continuous improvement process, review results together with our clients, and further refine the approach. When companies see how it impacts the bottom line, trust grows, and they are more willing to take the next step and develop additional tools.

Can you share examples of AI’s impact in mining?

In one case, identifying the sources and types of onsite water led to major savings in long-term water treatment planning. In another, during a mine closure scenario, a company believed it would need to manage a vast water footprint; using a machine learning approach, we found that much of their impacted water was naturally present before mining began, allowing them to significantly shrink their water treatment footprint, which will likely save them eight figures.

How do recent changes in project approvals translate into opportunities for Life Cycle Geo?

It is encouraging to see some recent momentum on speeding up project approvals, and we are seeing projects that had stalled start to move again. More projects mean more water and materials to manage, creating more opportunities for us to help.