Adaptation of Coal Grain Analysis to Improve Flotation Yield Estimation

Yield estimation from resource data is notoriously difficult. Factors impacting on achieved yield include:

• CPP feed size distribution which directly affects washability distribution;
• amount and nature of dilution;
• actual flotation response (subject of this report).

The CPP feed size distribution has a major impact on resulting washability, and that impacts on the output of gravity separation processes. Numerous projects over the life of NERDDP and ACARP have addressed this issue, and laboratory pre-treatment methods are reasonably well developed to generate particle size distributions of practical significance.

Any amount of 'pure' dilution, ie sinks 2.0 lumps and grains, can be readily catered for by simple yield compensation. Some forms of dilution may however include intermediate density material which will report to CPP product. To cater for that material, pre-treatment and washability testing needs to be included in resource assessment programs. If the dilution material breaks down to significant ultra-fines (< 0.1 mm), it is likely to have a significant impact on flotation separation performance due to entrainment, potential coating of coal particles, and/or interactions with collector and frother.

A major ongoing issue with CPP yield prediction, is how to accurately predict flotation yield. Flotation product is generally of low ash, so the yield achieved by flotation not only translates directly to CPP yield, but also allows larger particle separation processes (particularly DMC circuits) to be operated at higher density cutpoints, thus further increasing CPP yield. Consequently, flotation yield generally also has a secondary, and sometimes very significant, impact on CPP yield. Conversely, if flotation yield is over-estimated at the time of resource assessment, yield loss in practice may be further compounded by compensating DMC yield loss to meet product quality requirements. Predicted flotation yield will also have a large impact on product total moisture due to the high moisture in filter cake, and there have been cases where DMC cutpoints (and hence overall yield) has had to be decreased in order to decrease product ash to counteract the higher overall product total moisture.

Modelling of density separation processes (eg DMC) is reasonably straight forward and accurate, however existing methods of predictive modelling of flotation yield are severely limited, because flotation response is not directly related to particle size or density.

This project explores the role that Coal Grain Analysis (CGA) can play in better understanding the process and predicting flotation outcomes from resource data. Four full-scale flotation circuits were sampled and each of the feed, concentrate and tailings samples were subjected to CGA, which inherently allowed investigation of flotation response by size. For these circuits, reagent dosages were varied to measure the response of the different grain types. The results were assessed considering traditional free flotation and fine float sink test data.

This project has demonstrated further application of CGA as a tool to provide a more fundamental basis for modelling flotation performance. The CGA data provide a very clear picture of where valuable product coal is being lost to tailing, thus providing a tool for flotation cell and flotation reagent suppliers to target improved performance.