Predicting mill speed

Each year, milling engineers undertake a process of setting milling units for the following crushing season. This process has significant implications for the achievement of the factory crushing rate target and for the achievement of milling train performance targets in terms of total pol extraction and final bagasse moisture content. To achieve both rate and performance targets, an estimate of the effect of the mill settings on mill speed is made, either implicitly or explicitly. It is possible to use millfeeding theory to predict mill speed. The industry currently has access to the Jenkins and Murry mill feeding theory for this purpose. This theory, while functional, does not accurately predict the effect of various parameters on mill-feeding performance over a wide range of possible values. A new mill-feeding model has been developed using a computational model. This feeding model has been extensively tested using the results of experiments conducted on specific milling units over a wide range of parameter values and using measurements of feeding performance from many Australian milling units. While the computational model does perform better than the Jenkins and Murry model, it is not as useful for routine use since it is dependent on the solution from an expensive commercial finite element analysis package, making it impractical for factory use and slow in terms of solution time. To overcome this limitation, a new empirical feeding model was developed based on outputs from the computational model over a wide range of parameter values. The new empirical feeding model has been shown to accurately predict the computational model outputs over that wide range of parameter values, making it a potential successor to the Jenkins and Murry model. The new empirical feeding model can potentially be useful to milling engineers to predict mill speed and to improve the ability to meet crushing rate and performance targets.