Weir Minerals digital monitoring center. (Photo: Weir Minerals)

AI-based technologies are ushering in the next level of operational excellence, delivering benefits at different stages of the mining process

By Carly Leonida, European Editor

In recent years, artificial intelligence, or AI, seems to have become a catchall term that’s used to describe any kind of smart or autonomous capability in mining equipment and systems. It’s often (incorrectly) hailed as a silver bullet to a range of common industry problems, from safety to declining ore grades, and I challenge you to find an equipment or technology focused press release that doesn’t include the term. Rather than churn out yet another article which glorifies AI without actually explaining what it is or the value it can deliver in mining and metals specifically, let’s start our exploration with a few definitions…

AI, according to computing behemoth, IBM, is a term that describes “technology that enables computers and digital devices to learn, read, write, talk, see, create, play, analyze, make recommendations, and do other things humans do.” AI combines computer science with datasets to solve complex problems and enable specific actions or responses.

There are many subsets of AI, the most well-known being machine learning (ML). ML focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy. “Through the use of statistical methods, algorithms are trained to make classifications or predictions, and to uncover key insights in data mining projects. These insights subsequently drive decision making within applications and businesses, ideally impacting key growth metrics,” states IBM on its webpage. Neural networks is a sub field of ML, and deep learning (DL), which is often confused with ML, is actually a subset of neural networks.

So, what does all of this mean for the mining industry? Put simply, AI-based techniques and technologies can be used to extend the capabilities of human operators and teams. Computers and algorithms are good at, for example, identifying patterns or outlying data points in vast datasets. They are good at instigating and/or replicating certain actions in machines or systems perfectly and inexhaustibly and, when combined with optimization techniques, they are very good at identifying optimal solutions or decisions over different time horizons given a range of complex and sometimes conflicting objectives and constraints. Most importantly, they are good at doing these things very quickly.

AI isn’t going to steal anyone’s job. Ultimately, what it will do (and is already doing to a degree) is augment the capabilities of companies that have traditionally made all of their investments and decisions based on human insights and perceptions. It will allow them to do more with the people and resources they already have, drive greater efficiency in different processes and functions, make better choices across value chains, and instill agility and flexibility in operations and organizations.

With this in mind, let’s look at some areas across the value chain where AI is delivering benefits today.

Exploration: The Ability To See What People Cannot

At the very start of the value chain and the mine lifecycle sits exploration, a discipline that is key to delivering the minerals and metals projects upon which we all rely, but one that also faces mounting pressures.

Discovery rates for important metals are declining (see Figure 1). In its annual analysis of major copper discoveries, in 2022, S&P Global found that: “Although a significant amount of copper has been added to the 1990-2021 total discovered compared with the period in our 2021 analysis, the downward trend in rate and size of major discoveries over the past decade continues. All the new copper came from older, well-developed discoveries from the 1990s. In fact, we have only been able to identify three additional discoveries over the past five years, which added only 5.6 Mt.”

This speaks not only to the need for an evolution in exploration strategies and investments, but also the need for new exploration techniques. Boston-based VerAI Discoveries Inc. describes itself as an “AI-based mineral asset generator” which is pioneering a new approach to the search for valuable minerals.

“As significant findings decline, there’s a pressing need for a paradigm shift in our approach to exploration,” said Lorraine Godwin, vice president, commercial, at VerAI. “Tomorrow’s economic discoveries will be deeper or in areas of cover where there’s no visible outcrop at surface. Prospecting alone is not enough to connect very sparse information, particularly in covered terrain. The question we need to ask ourselves is: how can we find new economic deposits in areas of vast covered terrain and open new search spaces?”

To do this requires a systematic approach using georeferenced data which is objective and measurable, together with technology like AI, which is also objective and measurable and can learn and improve, to help explorers see what they otherwise cannot. Furthermore, a sustainability aspect comes into play when companies can accurately identify the starting points for their exploration programs. This leads to reduced site investigation, minimized ground disturbance, and fewer required drill holes.

“AI and ML are great tools to discover what experts cannot see locked in a vicious cycle of expert-based limited hypotheses,” said Godwin. “The mining and exploration industry is in the early stage of experimenting with this groundbreaking technology. However, there is much to watch and learn from how the technology is being integrated and propelling other sectors, such as medicine, finance, insurance, security, etc. These are all performing very complex discovery challenges that, until recently, were driven only by human experts’ knowledge.”

VerAI’s Artificial Intelligence Targeting Platform detects concealed mineral deposits, while systematically improving the probability of success. This also shortens the time to discovery for its partners, and ultimately, helps meet the supply demand for critical minerals for the energy transition.

Godwin explained: “The industry standard probability of finding minerals needed for the green energy transition is 1:1,000. VerAI is on track to reduce this to 1:10 by using AI technology to find metals under covered terrain.”


Figure 1 — Number of significant mineral deposit discoveries worldwide by commodity 1975-2019. (Source: MinEx Consulting, 2020)

Better Blast Measurements, Better Reconciliation

A little further down the value chain, Hexagon announced in February 2024 that it’s partnering with Augment Technologies, an AI specialist based in Western Australia, to help mines more accurately measure blast movements. The partnership will use a combination of block model data, AI, bespoke movement models and measured 3D movement data to create a solution that enables mines to maximize ore yield and optimize operational efficiencies.

Augment Technologies uses a physics engine powered by an AI algorithm to create a Muckpile Block Model that is continuously improved through a ML process. This utilizes vast amounts of blasting data to ensure that the model’s controlling parameters and simulated physics are as accurate as possible, resulting in a bespoke solution for each customer. Meanwhile, Hexagon’s MinePlan Block Model Manager solution enables users to design, populate, manage and share block models while centrally managing sample points, variables and outputs associated with orebody data.

Hexagon said the combination of these two solutions will allow customers to view and manage the Muckpile Block Model that retains all the data and fidelity of the grade control model, with extremely high accuracy and resolution. Users will have the option of incorporating Hexagon’s Blast Movement Monitors as an additional measure for blast movement and for training the AI model, with complete transparency into all the data inputs and output. They can also combine operational data with insights from the Hexagon Block Model Manager API to help optimize upstream and downstream processes.

James Dampney, vice president, resource optimization for Hexagon’s Mining division, spoke to the benefits of this integration: “Ore loss, dilution and misclassification cost mines millions of wasted dollars a year,” he said. “Our partnership with Augment Technologies will help mines to optimize digging locations and downstream handling of ore, resulting in valuable processing efficiency and reductions in energy consumption.

“Customers will save training time and operation time by remaining in the same software used to model their ore. The incorporation of an industry-first block model manager provides auditability and traceability to reduce errors while managers and corporate stakeholders will see time-stamped changes of the block model.”

Mineral Processing: Precise Plant Control

In the mineral processing space, Weir recently acquired SentianAI, a software developer that uses advanced AI algorithms that continuously learn and adapt to dynamic processes within a mine, providing continuous improvement and optimization over time. This technology is focused on helping miners to use less energy, reduce their water consumption, and minimize their waste. The AI framework utilizes reinforcement learning (a type of ML), which means that it’s continuously learning and adapting to changing circumstances based on the latest data ingested.

Martin Rugfelt, director of intelligent operations for Weir Minerals,, spoke to E&MJ about this move: “Weir and SentianAI mirror each other’s approach: both are focused on harnessing digital technologies to optimize entire circuits and processes, rather than simply improving the performance of individual pieces of equipment,” he explained. “This is why the acquisition made perfect sense. Weir has libraries of product knowledge and application data, coupled with decades of success in process optimization services, and SentianAI has a track record of taking large datasets and successfully applying AI to improve mining processes.”

Once integrated, SentianAI’s technology will extend and expand Weir’s current capabilities and turn current process optimization services into real-time digital solutions. Some operators are already using advanced process control (APC) solutions, and, in time, Weir’s digital solution will sit on top of these and use AI to do both cloud- and edge-based process optimization. This will enable customers to monitor and tailor their approach to meet their specific objectives.

“The merging of SentianAI’s capabilities and Weir’s knowledge opens the door for a new generation of supervisory control systems focused on process optimization,” said Rugfelt. “These will, in time, cover the entire processing flowsheet, with the initial focus being comminution and the mill circuit.”

There are many simple AI solutions available on the market, but the combination of advanced AI software, industry experience and customer intimacy are still rare.


There are many complex problems, such as scheduling, that require the use of sophisticated algorithms across the mine value chain and across different time horizons.

“Different customers, operating in various markets, have a range of priorities and requirements,” Rugfelt said. “For instance, the cost of energy might determine whether an operation is viable, in which case Weir’s digital solutions can be harnessed to improve process efficiency and reduce energy consumption and, ultimately, ensure it’s profitable. It’s important that digital solutions are flexible and customizable enough to help miners overcome the full spectrum of their challenges.”

Reinforcement learning is an advanced form of AI that’s part of the neural networks-based ML tree. It’s based on letting an AI-powered agent (which, in layman’s terms, is just a software-based entity) explore and learn what works best during different operating conditions. It has the capability to handle the very complex conditions that occur in a control system, which is not the case for many other forms of AI. Over time, the model ingests more data and gets a fuller perception of the environment, while also adapting to drift and changes.

“The system can be applied in stages as data quality improves and operator confidence increases — going from making recommendations to fully autonomous control,” Rugfelt explained. “This allows the operator to build up data and confidence before committing to fully autonomous control.”

SentianAI’s platform has delivered value at various stages in processing plants, from crushing to flotation, but its integration into Weir’s broader digital ecosystem will focus on process optimization.

Rugfelt explained: “The Weir digital team has been focused on improving equipment performance and availability and product quality — known as overall equipment effectiveness (OEE). And, broadly speaking, these solutions have delivered impressive results in terms of availability and efficiency, regardless of whether the equipment-enabled is an Enduron high-pressure grinding roll or Warman slurry pump, just to take two examples. Now, with the additional AI capabilities that SentianAI brings to the table, we anticipate those already impressive results improving further in terms of sustainable productivity.”

The team will also use SentianAI’s capabilities to broaden Weir’s offering to full mill circuits. For instance, SentianAI recently optimized a SAG mill by implementing precise control of input variables, machine speeds and grinding parameters to enhance throughput and process efficiency in the grinding circuit. By optimizing these factors, the concentrator plant can achieve improved mineral separation and maximize the recovery of valuable minerals, leading to increased operational efficiency and profitability.

“The SAG mill grinding process was controlled by a fixed operational ‘recipe,’ which can lead to sub-optimal performance, increased energy costs, higher long term machine costs and potential disruptions to downstream processes,” said Rugfelt.

Ultimately, AI will play a vital role in addressing many of the most pressing challenges facing the mining industry. It will help miners increase production by getting more out of diminishing ore grades, enabling them to meet the demand required for the transition to a low carbon economy. From an environmental perspective, harnessing AI for process optimization will help miners reduce their environmental footprint and energy usage.

Planning and Scheduling: Smarter Decision Making

One of the fastest growing applications for AI, which can add huge amounts of value, is in optimized and automated decision making, both at the operational and value chain levels.

Louis Okada sales and marketing manager for at Polymathian, a Deswik company, explained: “If we define AI-based technologies with the intent of aiding decision making, then the field of industrial mathematics can offer the mining industry extraordinary outcomes. There are many complex problems that require the use of sophisticated algorithms across the mine value chain and across different time horizons. Combining innovative digital solutions and industrial mathematics empowers mining professionals to make the best decisions, every time.”

They do this in a number of ways… First, by automating the decision-making process to eliminate errors introduced by humans and with the ability to run continuously, 24/7. Second, by solving incredibly complex problems that are often impossible to tackle with manual or less sophisticated software tools, like Excel. And third, by providing outcomes that are mathematically guaranteed to be ‘the best’ solution for a given set of inputs, constraints and objectives.

“By applying these techniques in the operational space, there are opportunities to enable fully autonomous mining operations,” said Okada. “This means operations can significantly improve their productivity, react dynamically to changing conditions, achieve higher compliance to strategic priorities, and take humans out of hazardous situations. Shift operators can adopt the role of a supervisor or orchestrater, allowing them to focus on higher level priority tasks, and this also helps with personnel resourcing by allowing teams to do more with the valuable people that they have.”

Large mining operations are often divided into logical working groups, which can make it difficult to see what effect decisions in each area have on the overall mining system. Optimization and simulation technologies provide greater transparency to all facets of a business, which would otherwise be difficult due to the volume and complexity of the information that needs to be processed and presented in an easy-to-understand manner.

At a value chain level, solutions such as BOLT from Polymathian, allow mining companies to approach planning and management activities from a holistic/global perspective. Traditionally large value chains are planned in two halves — production and marketing (supply and demand) — but due to the scale of these operations, it can be difficult to plan from an integrated perspective.

Okada explained: “It can be hard to see how decisions may have an impact across the value chain, and over time. BOLT allows mines to understand how decisions in areas, such as environmental, social and governance (ESG), have an influence on the whole system. The name of the game is to achieve better ESG outcomes while maintaining profitability, but this is difficult to do without the use of sophisticated algorithms.”

In addition to BOLT, which caters for value chain planning from strategic to operational time horizons, Polymathian also offers ORB for dispatch optimization at underground mining operations and RACE for complex heavy haulage rail optimization. The algorithms these solutions are based on use commercial mathematical solvers, like Gurobi, as well as cloud-based technologies and software written and owned by Polymathian.

One of the world’s largest coal mines located in Australia recently deployed BOLT. The operation produces coking coal and blends multiple products on site before railing them to port. Unlike other mining operations, which allow products to be blended to specification further down the supply chain, this mine required on-spec products ready to be loaded at the train load out.

The mine required an automated solution that would prescribe the best set point for the coal handling and preparation plant (CHPP) and produce a precise plan to tell the trucking fleet how much of each feed source was required for each product specification. BOLT was deployed to automate the customer’s daily material movements and processing decisions. It now produces detailed plans for the trucking fleet 10 times faster than the previous method to ensure the correct coal grade ratios are delivered daily to the CHPP. The model is able to acknowledge the system constraints, what is possible and what isn’t, and develops plans accordingly up to 16 months into the future.

“Do you see the mining and metals industry leaning more heavily upon AI for operational excellence in the future?” E&MJ asked.

“Absolutely, we’ve already delivered a huge increase in value to our customers,” Okada concluded. “That’s value that is not only in profit margins, but in the ability to upskill workforces and give owners more options to pursue. Like the other industrial revolutions that have helped to accelerate human progress, leaning on technologies in AI can help to catapult the mining industry to become safer, more productive and accessible for everyone. Those who don’t adopt it will, in time, be left behind.”