Measuring, modelling and understanding the mechanical behavior of bagasse

In the Australian sugar industry, sugar cane is smashed into a straw like material by hammers before being squeezed between large rollers to extract the sugar juice.\ud The straw like material is initially called prepared cane and then bagasse as it passes through successive roller milling units. The sugar cane materials are highly\ud compressible, have high moisture content, are fibrous, and they resemble some peat soils in both appearance and mechanical behaviour. A promising avenue to improve the performance of milling units for increased throughput and juice extraction, and to reduce costs is by modelling of the crushing process. To achieve this, it is believed necessary that milling models should be able to reproduce measured bagasse behaviour. This investigation sought to measure the mechanical (compression, shear, and volume) behaviour of prepared cane and bagasse, to identify limitations in currently used material models, and to progress towards a material model that can predict\ud bagasse behaviour adequately. Tests were carried out using a modified direct shear test equipment and procedure at\ud most of the large range of pressures occurring in the crushing process. The investigation included an assessment of the performance of the direct shear test for measuring bagasse behaviour. The assessment was carried out using finite element modelling. It was shown that prepared cane and bagasse exhibited critical state behavior similar to that of soils and the magnitudes of material parameters were determined. The measurements were used to identify desirable features for a bagasse material model. It was shown that currently used material models had major limitations for reproducing bagasse behaviour. A model from the soil mechanics literature was modified and shown to achieve improved reproduction while using magnitudes of\ud material parameters that better reflected the measured values. Finally, a typical three roller mill pressure feeder configuration was modelled. The predictions and\ud limitations were assessed by comparison to measured data from a sugar factory