The operation and optimization of mining operations - from pit to port - is a multi-disciplinary process that employs significant resources, tools and expertise. Over the last few decades, the mining industry has seen significant productivity improvements: new technologies in the areas of measurements, communications, planning, and automation have been transforming the way both mining and processing are being done.

Processing plants have been transformed by measurement and control solutions, arguably more so than mining operations themselves, due to their continuous, less distributed nature. On many minerals processing plants and metal refineries worldwide the days of manually setting pump speeds and valve openings are something of the past. Automated control systems allow operators to put certain control loops into auto, who can then hand menial stabilization tasks over to these systems. Over time, smarter control systems have been developed, moving from single-input-single-output (SISO) control loops to multivariable control (MIMO). PID controllers started being employed in more innovative ways, including gain scheduling and feedforward control, while fuzzy logic started being used to deliver now rather familiar expert systems. Model Predictive Control (MPC) systems, many of which were originally developed for the oil and gas industry, also started being applied to multivariable control problems in the minerals processing space and are still increasing their footprint today.

In a manner not too dissimilar to other processing industries, these advanced control systems have been transforming the operation of mining processes — unlocking levels of process stability that many in the days of manual and analog control would not have considered possible. From density control in grinding circuits to the control of flotation cell levels, these systems (when done well) can make the faster valve or pump changes required to reach and maintain process variable setpoints, and remain within bands that operators and metallurgists give them.

The improvements in stabilizing control have started making it possible for operating personnel at different levels to decide how they want to run their process, enter their setpoints via SCADA into their control systems and (if feasible) the process is steered there. Behind the processes, equipment and control systems, there are people making hard, complex decisions — all the time, every day. From the mine to the final product, every mine is abuzz with decision-making on multiple levels, informed by a wealth of hard-earned experience: from mine planning to maintenance, from geologists to control room operators.

Interactions between variables and processes on this decision-making (or we can call it optimization) level are often incredibly complex and nuanced — with each person often making a different decision in any given situation. Feedback on this decision-making level is usually relatively slow, which means that these decisions cannot be quickly corrected and should therefore be made well in the first place. The importance of these decisions is exacerbated by the fact that these decisions directly and significantly determine not only the financial performance of the plant and overall value chain, but its impact on the environment and the safety of people too.

To reduce human error here, people have tried applying stabilizing control systems to this decision-making layer. While it has been shown to work in industries with a consistent feed/raw material, such as oil & gas, mining processes pose a different challenge to real-time decision-making. Minerals and metals are unfortunately not uniformly dispersed throughout the crust of the earth. Material properties vary significantly — even in the same ore body, both on a micro (e.g. mineralogy) and macro (e.g. ore hardness and -size) level. When it comes to guiding or automating decision-making, to optimizing processes in the mining sector in real-time, stabilizing control systems are very quickly out of their depth.

A helpful analogy here is that of a human body. Our bodies have functions that are automated. A person does not stop breathing, nor does their heart stop beating when they think about something else. The functions that keep us alive are automated - similar to a plant being kept stable. However, above this automated layer is a decision-making layer. Voluntary muscle movements and other actions are driven by decision-making based on personal objectives and responses to the outside world — and these guide the automated layer. When someone starts running, their breathing and heart rate adapt automatically.

In partnership with the mining industry, we have made it our mission to help the mining industry make better decisions. To see how we can provide people and collective operations with something that gives them better visibility and guides them in their daily decision-making.

We have met men and women on-site who, through years of on-the-ground experience, have learnt to “read” their processes. Who can take a single look at the texture of a flotation cell’s froth and know what to tweak to improve the grade. Or feel how a pump is vibrating and know what they have been feeding into the SAG Mill. At the same time, though, we have seen that very often, operating personnel struggle to consistently make the right decisions. Each shift has a different crew, with their own way of running things. People often work in silos, with limited communication (sometimes even enmity) between different departments or parts of the mine/process.

As such, we have asked ourselves: what if we could digitally capture a proper understanding or experience of these processes — something that can relatively quickly “learn” the complex dynamics and interactions of a process? And what if we could offer this to the mining industry, to guide people of different levels in making better decisions — proactively, in real-time, every day?

For many years this is what we have been doing: providing the mining industry with Optimization Applications that use advanced technologies able to “understand” processes on an optimization or decision-making level. Where control systems are configured and tuned to determine how to pursue given setpoints, our Optimization Applications determine what these setpoints should be in the first place, to achieve operational targets or KPIs. A practical example here is finding, at every moment, the operating point that simultaneously maximises a grinding circuit’s throughput and the flotation circuit’s recovery, and minimises the environmental impact of these processes per kilogram of metal produced — given the ever-changing material properties of the ore being fed into the Mill. We employ a combination of first principle and machine learning modelling techniques to capture this complexity in our Digital Process Models: the brains behind our Optimization Applications.

In a manner similar to how operational personnel make decisions that span process units, our Optimization Applications draw on data from the mine to the final product and enable proper value chain optimization. It “learns” and can predict what influence the feed material properties have on each process step, which means that it can proactively provide the automated control systems of each subsequent process (e.g. flotation and leaching) with the setpoints and limits it would need to achieve optimal operation for this given feed. The same applies to the impact of each process on the next.

The Optimization Applications’ recommendations are provided to the relevant people — whether maintenance personnel, metallurgists, operators — as advisory or decision support. But mines can also opt to implement these optimal recommendations automatically, and pass them directly to the lower level control systems.

In this way, our Optimization Applications guide the “where we should go”, while the control systems then determine the “how we should get there”, similar to the human body analogy. The following diagram shows how this control-and-optimization hierarchy works:

Figure 1: How IntelliSense.io’s Optimization Applications operate in the decision-making layer, guiding operational personnel and stabilising control systems on where to operate the process for optimal performance.

When operating on the decision-making layer, it is essential to incorporate operational workflows — the way people of different teams work together. While control systems are automated and, per definition, not integrated with human workflows (beyond receiving limits and setpoints), IntelliSense.io’s Optimization Applications are configured to fit into a site’s unique way of working. Our Applications run on brains.app, IntelliSense.io’s real-time decision-making platform, which is integrated with data sources across the whole value chain — serving as a single source of truth for performance monitoring and decision-making for all levels of the mine. This includes the monitoring and reporting of how well the lower-level control systems are stabilizing the process and tracking the Optimization Applications’ recommended setpoints.

While we have focused on the connection between process optimization and plant control, it is worth recognizing how fundamentally different control and optimization in the mine is from the plant. Minerals processing and metallurgical plants are mostly continuous processes that are well-instrumented with both sensors and actuators and are therefore adapted to automated control. In contrast, in the mine space, you deal with discrete, stop-and-start “processes” involving spatial data, human actions and moving assets.

We believe that, to do pit to port optimization, the mining industry needs an optimization or decision support system that spans across the worlds of mine and plant, and combines them into a single, interoperable database and workflow. As such, IntelliSense.io’s Optimization Applications cover the whole value chain. Whilst on the plant side it provides recommendations to control systems, on the mine side these recommendations are provided as schedules to the likes of fleet management, drill-and-blast and stockpile management systems/personnel. By connecting our Applications across the value chain in our platform, brains.app, we are excited about breaking down silos and being a single source of truth and guidance for both mine and plant.

The mining industry has been rapidly transforming, with automation and stabilizing control (especially in processing plants) leading the way. At the core of the mining value chain, though, people make decisions about where and how to operate their sections. At IntelliSense.io we help mining companies unlock significant value by empowering people on all levels to consistently make better decisions, regardless of shift — to truly optimize their operations, and look after people and the environment in the process.

This content was first published as an IntelliSense.io White Paper. Feel free to pose any questions you might have to the author at niel.knoblauch@intellisense.io. For more information on how IntelliSense.io solves industry pain points, visit our website.