Abstract:

Ore Blend Optimization:

While a high mineral recovery percentage may incidentally be reached, companies are usually recovering a lower percentage of minerals than they would like, most likely through the changing of the ore blend characteristics being processed throughout the planned cycle. Many mining companies currently utilize certain strategies for choosing a blend to feed to their mills, such as prioritizing grade or throughput. However, these strategies may not consider what specific blends have worked in the past, and there is no consistent progress or continuous improvement. With many variables involved in the recovery process, such as material properties and mill operating setpoints, it is difficult to say what the effects will be of changing one or more variables on the final recovery; therefore, making the mineral recovery optimization challenge feel like a biased guessing game.

Identifying the necessary steps throughout the Mining process to optimize mill performance:

Mining companies have a lot of information at their fingertips allowing for opportunities to collect data. This data may originate from the organization’s fleet management system, crusher input tracking, mill operating measurements, and more. In today’s technological climate, there are a lot of different options available to make use of this data.
Mineral recovery is highly dependent on what material is processed, and the settings within the process itself. By tracking what material is fed into the mill (such as crusher inputs) and the operating parameters of the mill equipment, an archive of historical data can be amassed. This historical data can be fed into machine learning models to pinpoint the best possible blends to feed into the mill based on what stockpiles are currently in the stockpile yard.

Demoing next-generation technologies:

New technology is enabling companies the ability to learn from their historical data and optimize their mineral recovery for them to make the most out of their mining operations.

Many tools are available to help ingest, store, analyze, and learn from your data. Cloud technology such as Microsoft Azure and the Power Platform comes highly recommended for these needs. Power Apps can provide an intuitive user interface for ingesting data, and Azure SQL Server can be used to store large amounts of data at a low cost.
Power BI is a solid tool to connect to many kinds of data sources (such as fleet management system data, historians, and manual planning inputs) and for helping generate insights from visualizations and slicing data.
Machine Learning (ML) solutions require specialized developers to create. For an ML algorithm to generate accurate predictions, it requires good-quality, curated data to be tied together logically. Crushing data and mill operating data are strong candidates to serve as a basis for an ML initiative.
More than ever, Data Engineers have become an integral part of Mining Engineers’ daily operations, this collaboration results in smarter mine operations utilizing the treasure chest of historical data to the businesses advantage

Menu

Michael Matrick