Applications in Mining
Tech businesses such as Google and Amazon are already gaining benefit from BDA innovations because they have less trouble handling Big Data in the process of innovation. Although the adoption of technology is relatively slow in the mining industry, operational challenges such as declining ore grades, deeper location of ore bodies, environmental and social constraints, and stakeholder pressures are demanding fast changes towards digital capabilities. The following topics explore the application of digital technologies, especially Advanced Analytics methods, in the mining process analytics, control, and improvement initiatives.
Mineral Process Analytics
One of the most promising fields for the application of advanced analytics in mining is mineral processing. Not surprisingly, the massive amount of realtime operational data is a rich source of value creation in the Big Data era.
The mining company Barrick Gold developed an integrated analytics platform using an operational process model applied in the management and optimization of daily operations and as a planning tool connected to the geology resource model. The company's main objective is to maximize the gold extraction based on an active mine to mill integration. The process model was built using a digital plant template configured with target data from plant schedule and process inputs. The digital template receives realtime data and transforms it into operational insights that are then used to develop predictive models. Operational data are classified in different operating models (e.g., running, down, idle, trouble) that provide the desired level of detail to explore improvement opportunities. Additionally, the team advocates that the higher quality of data obtained by the configuration of operating mode event frames allows the application of advanced analytics tools and methods. For example, to access a more reliable particle size value over time, a soft sensor was built using machine learning tools [49].
Another mining giant that demonstrates a digital footprint and has heavily invested in advanced analytics technology for mineral processing plants is Freeport-McMoRan. Its machine learning program is one of three major initiatives aimed to increase the company's copper production by 30%. Supported by the consulting company McKinsey, the team started to unlock insights and boost operational performance approaching data mining in an agile way. In the earlier stages, the team investigated data from the Bagdad plant mill, looking for patterns that revealed opportunities for improvements. Then, a machine learning model was employed to assess the actual mill performance against that expected by the local team. Before the project, the technical and operational team believed that there was only one type of ore feeding the plant, which resulted in a single procedure to adjust the mill's forty-two parameters. However, a mindset shift occurred when the mill's data were used to model the material feeding characteristics and appointed to seven different types of ore. Motivated by this initial finding, which suggested a 10% production improvement by adjusting mill's parameters based on the different ore types, an artificial intelligence model was developed to prescribe mill control settings according to the ore characteristics and plant equipment sensors, raising cooper production from that ore. In order to reach the next improvement level, the team introduced a new artificial intelligence algorithm to maximize cooper output [50].
