The thin line between accuracy and bias in characterisation of ore grindability
F Faramarzi1,4, R Morrison2 and T G Vizcarra3
1. Global Senior Mining Industry Consultant, Dassault Systemes, Brisbane Qld 4000.
Email:
2. Honorary Professor, JKMRC, The University of Queensland, Brisbane Qld 4068.
Email:
3. Senior Consultant, JKTech Ltd Pty, Brisbane Qld 4068. Email:
4. Alumnus, JKMRC, The University of Queensland, Brisbane Qld 4068.
Email:
ABSTRACT
In comminution, grindability or ‘hardness’ indices are used to describe ore resistance to ball milling. The most commonly used measure is the Bond ball mill work index (BBMWi). The implicit assumption underlying this test is that the distribution of mineral hardness within the sample is uniform, or at least falls within a narrow range of variation. This paper considers cases where that assumption is not valid.
In actual industrial practice, feed is subjected to multiple breakage mechanisms and several stages of classification. These may introduce biases in the hardness of material that is actually processed by the ball mill. That is, the actual hardness of feed presented to the ball mill may differ from the fresh circuit feed, especially when transition/weathered ore types are processed. Such ore types tend to exhibit variable breakage characteristics, and separation of their different hardness components is likely during different stages in comminution processes. The result is that harder minerals will tend to concentrate in the ball mill recirculating load. Another example is when crushed SAG pebbles (which are generally comprised of harder minerals) are diverted to the ball milling stage with the aim of improving SAG mill throughput. In this case, ball mills will process material which may be significantly harder than the circuit feed. In both cases, the ball milling stage may become an operational bottleneck. For these milling strategies, results from standard test work conducted on fresh ore samples should be treated with caution.
In addition to proposing a model for describing attrition breakage behaviour, this paper also tests one approach to measuring the extent and impact of ore hardness variability not reflected in standard test work protocols. It presents the results and discusses implications from a laboratory test method, developed at the Julius Kruttschnitt Mineral Research Centre (JKMRC), to investigate the distribution of hardness within samples that would each otherwise be considered ‘uniform’. The results from testing three different samples characterised by standard and modified Bond ball tests confirmed that up to a 10 per cent under-estimation in evaluating ore grindability is possible. Under-estimation of ball milling feed hardness increases the risk of ball mills with limited capacity during periods of processing transitional ore domains or when a modified circuit configuration is required for debottlenecking and improvement purposes.