The simulation of ore-dressing plants

An efficient and automated ore-dressing plant simulator has been developed.In this simulator, stream variables can be used to describe a large number of unique solid particles that have a wide range of properties. Consequently, a large number of simultaneous equations need to be solved.This paper demonstrates how particles can be categorized and the mass balance for complex recycle systems rapidly achieved. The simulator that has been developed uses unit modules that are linked together by an executive computer programme. The executive, by treating each module in turn, obtains a set of simultaneous equations that are solved iteratively. The sequence in which the modules are treated is determined by partitioning and tearing algorithms that select tear streams, thus rendering the system acyclic. The algorithms are designed to choose the tear set (as there are usually several) that will allow convergence in the fewest possible iterations. This choice is based on a criterion that can be extended to apply to all the convergence methods considered. A simultaneous convergence technique has been developed that is applicable specifically to linear ore-dressing plants, but can also be applied to non-linear systems. This technique, which is called the reduced Newton method, is shown to exhibit local convergence under reasonable conditions and is generally superior to direct substitution and to the bounded Wegstein methods.