Updated Benavente Correlation for Estimating Grinding Media Consumption Rates

Procemin 2013


The concern regarding grinding media consumption rates is as old as the invention of the tumbling mill. Over the years many materials and grinding media shapes have been tested. Through all this, steel grinding balls have proved to be the most effective media for comminution in tumbling mills.

There have been several approaches to be able to predict the wear of the ball charge in any given mill, all of them based on limited empirical evidence. The mining industry still utilizes the Bond Abrasion test that was developed in the 1960´s. As shown later in this publication, this approach shows error greater than 60%.

More recently, in 2007, Radziszewski from Mc Gill University has proposed a different total grinding media wear model; this model is based on decoupling the effects of the abrasion, corrosion and impact wear mechanisms. This new decoupled model results in an error of +/- 17%, which is an improvement in comparison to Bond model, but still exhibits an undesirably high degree of error.

Also in 2007, Benavente, at the time with Moly-Cop Peru, presented an empirical model relating both ore properties and operational conditions that would affect the wear performance of the grinding media. For the database referred in this article, Benavente’s model shows an average error of ± 12% which represents a higher degree of improvement in comparison with previous approaches.

The current paper presents recent research conducted by Moly-Cop Peru to further improve the capability of the empirical model originally developed by Benavente to estimate grinding media wear rates. The work performed included extensive determinations of the Bond Abrasion Index for different types of ores, collected at industrial grinding applications where the corresponding operational information was made available.

The results of the research enabled us to build a more robust and confident model for estimating the grinding media wear rate. The average error of this new model is in the order of  ± 9% which is a significant improvement over the models previously described.

>  Updated Benavente Correlation for Estimating Grinding Media Consumption Rates