Metals and mining companies are investing in digital technologies across the value chain, from operations to procurement to sales and marketing. Unfortunately, many of these investments have fallen far short of their potential.
For any new technology introduction, whether it’s new sensors or leveraging existing data, the biggest risk is data going out of calibration and a lack of uncertainty measurement in the input data, leading to flawed decision recommendations. After all, these sensors and input data systems are the ‘data lifeblood’ of AI systems.
This is a significant deficit, especially considering the enormous benefits digital technology can bring to the sector. By accelerating digital transformation using the new frameworks such as Objective-driven AI, metals and mining companies can boost throughput, simplify processes, lower costs, improve metal recovery and yield, and reduce supply chain complexity.
Data and Explainable Models for Managing Variability in Mining
The mining industry faces the unique challenge of working with ever-changing inputs due to the dynamic nature of its operating landscape.
Unlike other industries such as manufacturing with stable assembly line inputs, mining deals with fluctuating ore properties and varying operating conditions that demand distinct processing strategies for optimal mineral recovery.
What’s new is how miners are treating their data. Mining has never had access to as much data for decision-making as it does now. Twenty years ago, the industry couldn’t have imagined the wealth of information available today. This data has become a truly valuable asset.
However, effectively harnessing its potential requires addressing variability and ensuring explainability in data usage models. Trust comes from knowing the AI will tell you if it’s unsure of the outputs and allow the operator to override with confidence.
This comes from adopting the fundamentals of data uncertainty, and an uncertainty-driven modeling approach. The AI essentially creates an understanding of how the world works, which it learns through self-supervised learning.
While this is no easy feat, a handful of companies have succeeded, and the results are promising.
Accelerating Mining Toward a Successful Digital Future
Many miners around the world have successfully employed IntelliSense.io applications to achieve outcomes such as an additional $2M in gold production from selective mining using 3D block models of stockpiles, a 1% increase in recovery for flotation circuits, a 3% increase in metal recovery in solvent extraction, a 16% reduction in chemical use in thickeners, and more.
Beyond these quantitative benefits, there are qualitative advantages like increased visibility of operational units within large circuits (such as identifying underperforming individual flotation cells), improved stockpile management (reducing dumping errors and better blending decisions), and more proactive decision-making with real-time optimized recommendations and alerts.
Interested in learning how this can be applied to your operation?
