Every day, often for 24 hours a day, seven days a week, clear across Australia massive attritor mills are pounding and crushing various mining products into bite-size chunks. A typical application is to produce pulverised fuel (size 100 microns or smaller) from crushed coal. The problem is that attritor mills are also pounding themselves into oblivion at the same time.

Figure 1. Before and after images of wear in the second stage of a multiple-stage coal attrition mill the second-effect hammer plates completely erode away.

At a cost of $1.5m, an attritor mill is no mean investment, so any means of reducing their ability to hammer themselves to uselessness in short order would save considerable amounts of money in direct costs and lost productivity.

Engineers at CSIRO Energy & Thermofluids Engineering have targeted these beasts of attrition, particularly impact crushers, to look at designing to improve wear rates.

Figure 2. Small-scale physical model of an attrition mill, manufactured from transparent plastics.

Among the variety of machine types in common service, including cone crushers, ball mills and ring-ball mills, are high-speed machines in which the size reduction takes place through blows from rotating hammers. Impact crushers respond rapidly to changes in plant load, and crushing usually occurs in an airflow used to transport the finer sized fractions from the machine (or through a secondary classifier stage), which can also provide some drying of the pulverised material.

These machines are designed for use on comparatively non-abrasive material, such as brown coals or limestone. However, in Australia they are used at a number of plants that run on abrasive high ash-content black coal. As a result, the mills require frequent refurbishment to address excessive wear. At large plants, it is not unusual for mill maintenance costs to run to several millions of dollars or more per annum.

This simple premise sums up why, over many years, solutions to the excessive wear in attritor mills have not been readily identified. This excessive wear is often highly localised or, in multistage machines, occurs dominantly in a particular stage. In these cases, the problems are often related to poor internal flow patterns of solids and gas. The observation scenario has been compounded by the option of trial remediation, which is often problematic as observing the internal flows in these machines is difficult and costly to undertake on plant that is in service. CSIRO has overcome these obstacles to progress through a combination of what some call the esoteric science of computational fluid dynamics (CFD) and complex multiphase small-scale modelling techniques.

Figure 3. Strobe-lit 'frozen' view of particle flows inside a model mill in the pneumatic conveying-loop test rig at CSIRO.

Either small-scale physical modelling based in a laboratory alone or combined with computational studies, provide a cost-effective means of understanding the internal flows and testing alternative solutions.

Modelling, either physical or numerical, is applied at CSIRO to gain an understanding of internal flow patterns. For physical experiments, this is done with the models running in our pneumatic conveying test loop, and is carried out using various flow visualisation techniques, including steady or strobe-lit global and sheet illumination, or high-speed video.

For CFD modelling, meshes can be produced with the aid of CAD geometric information. Then solutions can be proposed, manufactured on our in-house machine centres, and tried in the laboratory or on computer. A variety of changes can be evaluated quickly and at small incremental cost. Typical solutions call for revised internal flow deflectors, changing the size, shape and orientation of the hammers, or other comparatively simple changes. In those cases, the model, either computational of physical, can be modified then tested and the changes evaluated.

More fundamental changes, say to shift the basic crushing mechanism from hammer-particle to particle-particle impact, require careful redesign and may require on-site testing with real materials to confirm the effectiveness of proposed changes.

This purpose-built dilute phase pneumatic conveying test loop is situated at CSIRO’s Highett laboratories in Melbourne, which can accommodate a range of model mills, typically produced from clear plastics to enable internal flows to be observed. The models are designed and manufactured at CSIRO.

In practice, these models often quickly relate observed poor flow paths within the model with high localised wear rates found in full-scale machines, providing CSIRO engineers the opportunity to propose and check useful modifications at the laboratory scale, before they are implemented on site. The pneumatic conveying rig has also been used to model flows and wear of cyclonic separators.

It is important that the interaction between particles and airflow and between particles and the hammers within the machine is accounted for and correctly modelled. The airflow rates and particle densities, shapes and sizes in a physical model must be carefully selected, so that the flows have correct dynamic scaling and so that particles follow correctly scaled trajectories without destroying the model. Larger particles are less sensitive to airflows, and their trajectories are largely determined by hammer and wall impacts, while finer particles are much more affected by the internal airflow. This is illustrated by comparison of trajectories of different particle size fractions predicted in a CFD study of gas and particle flows inside a mill, as shown below.

If internal flow problems are not readily apparent by direct observation, it is possible to cover model surfaces with abradable coatings, such as paint, observe regions of high local wear, compare these to full-scale observations, and perform comparative evaluations of different possible solutions.

CFD can be used to study air and particle paths within machines. For machines that have a high flow rate of air or other gas, it is important to model the flow of both the solids and gas phases, and their coupling. Present CFD techniques do not incorporate breakage models for the process of breaking large particles into smaller ones. This is important, both from the point of view of energy consumption, and to allow the more correct modelling by allowing the particle size population to vary as it passes through the machine. To address this, new methods are being developed at CSIRO to allow breakage models to be incorporated into the CFD models, thus allowing comminution to be directly modelled. In addition, different breakage models can be comparatively evaluated.

Our clients to date are NRG, Queensland Alumina Ltd and Alcoa. As a part of a project to redesign impact crushers at NRG’s Gladstone power station, a half-scale metal model of one stage of a multi-stage impact crusher is being built to install in parallel with a full-scale mill on-site, following laboratory-scale modelling at CSIRO. This will provide direct confirmation of proposed design modifications and their effect on wear, using run-of-mine coal.