Use of data and AI is today seen as one solution for unleashing this potential. By collecting, analyzing, and acting on process, machine, and sensor data, mining companies hope to identify and control the root causes of poor process performance. There is great promise and a lot at stake in turning the processing plant intelligent.
Three levers for intelligent processing
Essentially, there exist three levers for data-driven process improvement:
(1) Stabilize the process by reducing variability
(2) Optimize the process against the constraints (e.g. for higher throughput or improved energy efficiency)
(3) Maximize equipment availability and uptime
Advanced process control is a set of well-established technologies to address the first two levers. Taking control of the operations with an expert system that can handle the multivariate inputs and the often non-linear correlations between process variables is a must in the complex environment of a minerals processing plant. Once the process is under control, it can be optimized to be closer to the constraints to yield maximum performance.
However, the third lever – maximizing equipment availability and uptime – often gets too little attention. Yet, no matter how streamlined your process is, if your assets are down, you will lose a lot of production. This is where cloud-based IoT and AI come in.
Metso’s intelligent minerals processing pilots on fast track
This year, we are piloting Metso’s cloud-based IoT solution at multiple minerals processing plants in the USA, South America, Africa, and Australia. The record so far is at an African mine, where our remote condition monitoring solution has been up and running for a year already.
The results are quite promising. By analyzing data from three connected Metso NordbergTM MP crushers, we have been able to identify certain failure modes and predict some failures ahead of time. I believe that it will not be long before we can start predictively maintaining these crushers based on actual and forecasted component wear.
Going forward, Metso’s machines will become even better performing and reliable than they already are today. We will be able to collect and analyses data from an increasingly wide variety of machines within the comminution circuit. We will also be able to predict – and prevent – a lot more of the different equipment failure modes. The machine designs will improve at a radically faster pace when the engineers who design and build the machines can see data from the machines and how they perform in the field in real-world conditions. Furthermore, we will be able to increase our customers’ crushing circuit performance, e.g. by optimizing crushing efficiency and by retaining a more consistent particle size distribution.
To learn more about the digitalized minerals processing plant of the future, see this slideshow presented in May–June 2018 at the CIM 2018 Convention in Vancouver, Canada; Mines and Technology Europe in Helsinki, Finland; and SAIMM’s Digitalization in Mining Conference in Johannesburg, South Africa.
Roadblocks being removed from intelligent processing
Predictive maintenance is already a proven concept in mobile mining fleets such as haul trucks, draglines, and many types of underground equipment, such as LHDs and drills. It has taken surprisingly long for this concept to find its way to processing as well.
One of the major roadblocks has been the non-standardized nature of the processing plants. While mobile fleets are often standardized around one vendor and type, virtually every mining company in the world runs a mixed fleet in their processing plant. The plants consist of machines from multiple suppliers and are of starkly different ages and models, built according to a custom flowsheet design. This has slowed the adoption of data, AI, and IoT in the fixed plant. But all of that is about to change now.