Simulation is increasingly being used to analyse mineral processing operations and explore efficiency options. Engineers that use simulation are rightly concerned with the quality and applicability of the underlying models. However, many of these models have a significant empirical component, and the best structure for this component may not be apparent. There is an opportunity to improve the quality and application range of these models by exploring potential structures for this empirical element._x000D_
One approach to this problem is the use of structural optimisation, which is a formalisation of the model development process. This method searches through a family or hyper-structure of candidate models for a model that is optimal with respect to some measure of model quality._x000D_
This method has been applied to the modelling of an industrial flotation circuit. The result was a model that was a significant improvement over a number of standard models reported in the literature. Structural optimisation provides a method to compare potential models in a systematic and efficient way, and hence to improve the structure of process models.