Autonomous mining is already here
Autonomous is a big thing currently not only in the automotive industry, but also in other industries including mining. Autonomous haul trucks have existed in mines already for quite some time, actually well before Google Cars were even imagined. Since then, we have also seen the introduction of autonomous drill rigs, loaders, and lately also the “biggest robot in the world” running on rails launched by Rio Tinto.
Introducing autonomy to the mining process provides many advantages. A huge benefit is increased safety, as more people can be removed from underground and other hazardous places to instead run the daily operations from a clean and safe control room. A second benefit is productivity, as more trivial tasks are automated hence freeing up people to do higher level planning and optimization work.
Levels of autonomy – from cars to processing plants
When it comes to vehicles, five levels of autonomy have already been clearly defined. At lowest level of autonomy, level 1, we see automation like cruise control, where the automation does a single task and always requires human supervision. On highest level, level 5, the vehicle is fully autonomous and no human intervention of any kind is required, allowing even the steering wheel to be removed. You can find a car related autonomy framework here.
Interestingly, in minerals processing, there is no generally accepted framework for the different levels of autonomy of the plant. At the outset, it may seem that such a classification would be unnecessary, since the plants are fixed and - unlike cars and other mobile equipment – do not move around.
However, the situation is quite the contrary and classification of autonomy levels would be needed. Today, minerals processing plants require constant daily intervention by humans to adapt to changes in plant feed, disturbances in operating conditions, and degrading equipment health. These plants can easily employ hundreds of people whose daily job is to monitor the process and the plant, and to take care of maintenance activities. At a fully autonomous plant, such an army of people would not be necessary or at least their role would be very different. Between the two extremes, clearly there must be levels of increasing autonomy.
Hence, I will here classify minerals processing plant autonomy into five levels as follows:
Level 1: Regulatory controls
Characteristics: Operator is responsible for reacting to disturbances and finding the optimal set-points at each time. Maintenance is reactive and asset availability low. There is high variability in production rate, quality, and yield.
Level 2: Advanced Process Control
Characteristics: APC software finds optimal set-points and automatically compensates for process disturbances and variability in feed. New, added advanced instrumentation gives greater visibility and control over the process.
Level 3: Intelligent equipment
Characteristics: Majority of operator tasks have been automated. Minor process or feed disturbances are autonomously compensated by the equipment or other intelligent actuators. Operation modes that are harmful to equipment health are quickly detected and corrected.
Level 4: Analytics & AI
Characteristics: Predictive and preventive maintenance practices provide high asset availability. PID loops are constantly analyzed and optimized (e.g., bin > feeder > crusher > conveyor). AI analyses the process and advises the operator on optimal set-points beyond what Advanced Process Control is capable of, especially in highly multivariate and non-linear response scenarios, where feedback or feed-forward loops are long and complex.
Level 5: Full auto-pilot
Characteristics: Advanced Process Control has been further augmented by yet additional sensors & analyzers. Highly skilled operator is supported by AI and a hi-fidelity, dynamic, real-time process simulator (Digital Twin). Deep subject matter expertise is constantly available remotely and can collaborate over the same data with the operator. Maintenance is 100% predictive and preventative, and no unplanned shutdowns or other major interventions are needed.
Having clear automation classifications provides a framework against which to plan for increasing levels of plant autonomy. Also, the actions are additive in the sense that achieving lower levels of autonomy provide a basis and platform on which to build higher levels of autonomy. Without a clear understanding of what must come first, one may set out to improve plant autonomy by implementing actions on a shaking or even a non-existing foundation.
Autonomous minerals processing plant in action
Is a fully autonomous plant a day dream or something that can become a reality already within a few years? Indeed, there are already many established and new promising technologies developed in Metso that can help work towards this aim. Below are some select examples:
Advanced Process Control: Metso OCS-4D
PID loop analyzers: Metso Expertune
Predictive maintenance. Metso Metrics
Remote expert support: Metso Performance Centers
Chief Digital Officer