“Mass Balancing and Inference” Short Course

In order to optimize a mineral processing plant there are nine steps required:
1. Collect data (measurement)
2. Collate data (into a database)
3. Flowsheeting
4. Mass Balance
5. Inference
6. Determine Operational Parameter
7. Model fit
8. Simulate
9. Optimize
Mass balancing is often taught using least squares minimization. Inference is generally not taught and in the absence of inference, model fitting is rendered very difficult, which is why mineral processing simulation is generally much more difficult than it should be.
In this course we focus on mass balancing and inference. Whilst mass balancing is taught initially from a conventional least squares minimization approach, the course then discusses a probabilistic or information theory approach. Information theory is a relatively new area of mathematics which is only available to participants through this course. By using information theory one can infer ore properties at each stream in great depth (i.e. multi-mineral particle distributions); thereby providing the basis for later steps (particularly modelfitting).