Modelling of a Pb-Zn Flotation Circuit with a View toward Automatic Control

Sensors are now available in mineral processing circuits to monitor pulp flowrate and density, as well as ore characteristics and other process variables. The amount of information provided by these instruments is considerable but in many cases poorly exploited. This paper gives the results of the application of time series analysis techniques to signals generated by the sensors installed on an industrial Pb dezincing flotation circuit in New-Brunswick. Autocorrelation and cross correlation functions appear to be interesting tools for a preliminary circuit analysis, however the calibration of single input single output discrete transfer functions using collected sensors signals leads to conflicting results. For this reason a phenomenological approach, based on the kinetic and transport properties of the flotation process, is also considered. It has several advantages over the empirical transfer function modelling but also limitations in its applicability to industrial processes. A combination of the transfer function and kinetic modelling approches appears to be a promising avenue for better process modelling and control.