Prediction of Washability as a Function of Breakage

The washability and recovery of coal is governed by two processes: breakage and liberation of stone from coal. The efficiency of liberation is coal type and particle size dependent. Particle size distribution is a function of the material properties of the coal and the breakage energy imparted to the coal as it is mined, handled and processed. If both processes, breakage and washability, can be modelled independently, it should be possible to develop an integrated model to incorporate both kinds of data.

The objective of the project is to develop a capability for simulating washability as a function of breakage within a commercial breaker.

Coal Breakage Testing

The project was conducted at BHP Coal's Saraji Mine in central Queensland, Australia. Samples of different coal and rock types were collected from the Dysart seam and subjected to breakage testing at a series of impact energies on different size fractions. Results provided information on the range of material properties likely to enter the breaker station and the relationship between block size, impact energy and daughter particle size distributions after breakage (the "breakage maps") that were needed for the simulations.

The daughter particles generated from the breakage tests for two "end member" coal types were subjected to float-sink analyses at a range of densities. The ash yields of the density fractions were also determined. In total, 171 separate washability tests were conducted to provide data points for entry into the washability model. The resulting data provided information on the distribution of ash yields by particle size after a given breakage event that were needed to produce the "ash maps" used in the washability simulations.

A detailed audit of the Bradford Breaker at Saraji mine was conducted to obtain sizing and washability information on the breaker feed, undersize, product and reject material. These data were used to assist model development and simulation.

Existing numerical models for size reduction in a crusher were modified and applied to the Bradford Breaker. In addition, the model was refined to handle multiple coal and rock types in the feed. The most exciting aspect of the project was the development of a capability to simulate both breakage and liberation, in this case for a Bradford Breaker, under different operating conditions and machine configurations while being able to account for variable feed compositions. To incorporate washability predictions into the model it was necessary to first describe the manner in which the different feeds re-distribute or liberate ash in simple impact breakage processes like those experienced in the Bradford Breaker. Simulations were carried out separately for the different coal and rock components and the predicted individual products and rejects were then combined to form the overall product and reject of the circuit.

Given the feed size distribution and the breakage data that are generated from a single particle breakage tests, the models can predict:

1. The Reject rate from the circuit;
2. The rock contents in the Product and Reject;
3. The size distributions of the Product and the Reject.

The model algorithm was programmed in Fortran and incorporated as a DLL(dynamically linked library) with a user-friendly Excel based interface. This software allows the user to assess the impact of changes in either coal or energy delivered through change in breaker design (diameter, length, number of lifters, and throughput) on the final breaker reject and product sizing. A copy of the program containing test data for the Saraji project is included with this report.

This project has demonstrated that coal washability can be modelled as a function of breakage and that the results of the simulations are in agreement with the audit data from a commercial breaker. For a given feed coal of known size distribution and composition, the proportion of product and reject after breakage in a Bradford Breaker can be accurately simulated. In addition, the size distributions of and the amount of rock or ash in product and reject can also be simulated.