The upgraded ball mill size reduction model uses separate orespecific
parameters and machine parameters; it would be expected
to provide a better predictability for the model to simulate ore
property changes. To demonstrate the difference in predictions
between the perfect mixing ball mill model (Eq. (6)) and the
upgraded model, the perfect mixing ball mill model in JKSimMet
was used to fit the Sample A data, then using the fitted Sample A
parameters to predict Sample B product size distribution. Since
the default appearance function was used in the perfect mixing
model, JKSimMet made a scaled change in the fitted model parameters
ri/di
⁄
, according to the Bond work indices of Samples A and B.
Fig. 8 displays the predicted size distributions by the two models.
The Sum-of-Squares (SSQ) between the predicted and the measured
size distribution for the perfect mixing ball mill model in this
case is 108, and for the upgraded model the SSQ is 57. The smaller
SSQ indicates a closer match between the predicted