A Date with Data
Technology firms help miners find love at first byte
Facing the challenge of increasing production without opening new mines in the Americas, Freeport-McMoRan turned to artificial intelligence, starting at their Bagdad mine in Arizona. Data access was not the problem— a data warehouse stores sensor data collected on a second-by-second basis from the company’s trucks, shovels and stationary machines. The problem was creating and deploying an AI model at scale. McKinsey took the reins creating a model that boosted production by 5-10% and mitigated US$1.5-2 billion of losses to build a new processing facility. “The project taught us to be more receptive to what the data was telling us,” said Bertrand Odinet, chief information officer and chief innovation officer at Freeport-McMoRan in the McKinsey case study.
Data has long been a buzzword in the industry, much like AI, with vendors touting it as the key to optimizing operations, reducing energy consumption, cutting CO2 emissions, and boosting productivity. Without a doubt, data is the backbone of the industry’s advancements; it powers autonomy, AI, machine learning models, and nearly every modern shift in mining. “One of the biggest mistakes we see across the mining industry is doing things the way they have always been done, or worse, relying on “experience” to make decisions in the field without considering the data behind the decision. Sometimes our gut lies to us and the data disproves our initial reaction to a problem we are trying to solve,” said Keaton Turner, founder, president and CEO of Turner Mining Group.
However, data alone is not enough. The collection, type, manipulation and application of data will provide the benefits vendors promise. Data simply provides the ‘what.’ Technology providers in the Western US are working to use data to supply the ‘why’ and ‘how.’
Swipe Left
Data in the mining industry is like dating; just because there are billions of bits does not mean you will strike gold with every byte. Many vendors increased the number of sensors in hopes of collecting more data. Looks, however, are not everything: “Adding more sensors increases data volume, but not all data is useful, and if data is not enhancing decisions, the effort may not be worthwhile. To be effective, data must be captured, analyzed and actionable for decision makers,” said Derek Cooper, vice president US and Canada at Hexagon.
Data amount is subject to goldilocks conditions. “Geotechnical data collected by mining operations is rarely reliable due to poor quality, insufficiency, or the amount of time it takes to conduct a traditional survey. By the time the data collection is finished the data is already obsolete”, warned Ravi Sahu, CEO at Strayos.
To fix this, Strayos partnered with Wingstra, a drone manufacturing company. “Using drones for surveying, particularly those like Wingtra, which can map large mine sites quickly, significantly accelerates the data collection process—up to five times faster than traditional surveys—while enhancing quality and resolution- so that the data is reliable”, continued Sahu.
“In the past, the industry collected large amounts of data that often went unused. Now, the focus is on parsing data for insights that aid operational decision making; merely having data is not enough.”
Derek Cooper, Vice President, US and Canada, Hexagon
That’s my type
The key is not the amount of data, but rather knowing exactly what type of data you need. High-quality, unbiased data are best for training models (the AI kind). VerAI Discoveries, an AI-based mineral asset generator, leverages AI and machine learning algorithms to analyze geophysical data profiles— from magnetic, gravimetric, electromagnetic and seismic sources— of known economic ore bodies. By training their algorithms on this data, VerAI enhances the probability of identifying new economic mineral deposits.
Here is Lorraine Godwin, vice president commercial, on why unbiased data is so important: “Our improved success rate is primarily attributed to our rigorous approach to data utilization and AI model refinement. We prioritize high-quality, unbiased geophysical data. This approach minimizes interpretation errors and ensures our AI models are trained on the most accurate and reliable information available. By continuously advancing our AI algorithms through rigorous validation, we achieve higher success rates across different geological settings and mineral types.”
Made to perform
Once you find the right data match, it is all about performance. With the right data foundations in place, technology firms across the Western US are tailoring their models to specific deposits, thereby boosting productivity and saving valuable time. When given the proper data training, AI and machine learning (ML) are transforming the industry from the ground up.
Maptek is helping mining companies understand the true value of their resources by using AI to dig deeper. “We are witnessing significant AI adoption with Maptek DomainMCF, where ML analyzes drill holes or other sample data to create unbiased resource models. By considering structural elements and assessing data comprehensively, our cloud-based technology saves geologists from labor-intensive and time-consuming data manipulation. This innovation allows clients to generate preliminary models in hours instead of months,” said Clayton Fritz, sales manager of North America at Maptek.
Micromine is also playing the field with ML, streamlining the resource geologist’s workflow with Micromine Origin Grade Copilot, launched in November 2023. Ben McDonald, mining solutions manager, explained how the model leverages data to speed up resource estimation: “By leveraging advanced neural networks, Grade Copilot learns complex patterns in geological data to create comprehensive and robust models swiftly and autonomously. What once took weeks to accomplish manually can now be achieved in hours, and in some cases, mere minutes, freeing up valuable time for higher-level thinking.”
MST Global is all about creating a safe space, allowing clients to feel the connection, literally. “Core to our mission is ensuring seamless connectivity for equipment and personnel, especially in the challenging underground environment where traditional GPS tracking is not feasible,” said Jon Larson, general manager – Americas.
This takes the form of the AXON suite, providing network infrastructure and the Helix platform, allowing for advanced visualization and tracking.
“Miners often face the daunting task of turning extensive data into actionable insights.”
Rob Hardman, President and General Manager, Maptek
Taking things slow
Some players, like Asterra, were a bit late to the game, having kept their eyes on the stars. Originally, the firm’s propriety technology was designed to find water on Mars. The company, however, understands that good relationships require time and patience: “When entering new markets, spending time getting to know user base is a requirement,” said James Perry, executive vice president marketing and Earthworks Division.
Asterra took their time with the courting phase. Perry continued: “Through travel and interviews, we broadened our understanding of what our solution could achieve in the mining industry. This effort helped us evolve our solution, making our data timelier and more useful.”
To win over partners, Asterra became a down-to-earth firm, using synthetic aperture radar in the microwave L-band spectrum. “We identified an entry point where we could monitor and help maintain tailings dams. Since our technology can penetrate below ground, we provide the first line of sight for detecting moisture that might be leaching or piping through these dams.”
Performance anxiety
In the world of data-driven solutions, it is easy to feel overwhelmed, especially when every provider seems to have a different idea of what is best. The lack of standardization among technology providers only adds to the complexity. As Matt Blattman, director of technical services at Hecla Mining, pointed out in Seequent’s ‘Beyond the Hype: How technology can drive mining operations’ performance’ Insights Paper: “We would like to standardize, but within our own company, with four different mines, four different types of deposits and four different mining methods, it is hard to find something that fits every situation.”
This challenge has led data solution providers to fine-tune their approaches, ensuring they are not just offering a one-size-fits-all solution, but something more personalized. As Sahu puts it: “During onboarding we understand the specific problems and KPIs the client wants to address. Within the first six months, we achieve at least one KPI improvement. The AI model becomes more calibrated to the specific site as we gather more data—around 5,200 data sets are required to achieve accuracy.”
This gradual, get-to-know-you approach allows operators to dip their toes in before taking the plunge. And it is paying off: “When clients see the value, they are more inclined to extend their contracts and integrate our platform further into their operations,” Sahu added.
Life Cycle Geo, a firm specializing in ML services, observed a similar trend, said Tom Meuzelaar, the owner: “Model performance improves as operators collect more data. However, gaining buy-in and trust from operators when it comes to ML is typically a gradual process. Operators need to see tangible benefits—such as time and cost savings— and understand how a machine learning approach improves upon established practices before fully committing to new models.”
Ultimately, finding the right solution is like finding the right partner—it is a journey that requires exploring different options and assessing how well they align with your needs. Sometimes, you need to go on a few 'data dates' before you find the perfect match. In the end, the key to a lasting relationship is finding a partner that not only meets your needs today but is adaptable enough to grow with you into the future.
Article header image courtesy of Strayos