Digitalization of metallurgical processes means increasing the amount of process analytical instrumentation. Frequent measurements of product streams’ quality creates a data overload that becomes more and more difficult to handle. With usage of various statistical techniques it is possible to build calibration models based on spectral data, calculate concentration levels of desired elements, monitor process behavior based on residue values. A procedure has been developed in SIMP program that enables extracting new process data concerning impurities previously unknown from on-line data of X-Ray Fluorescence analyzers.
Identification of impurities in process streams without comprehensive and time-consuming laboratory analyses is needed in order to enable online control and increase the productivity of the processes. A drawback of conventional laboratory analyses is that disturbances such as impurities are often not detected unless they are specifically looked for. Advanced mathematical treatment of on-line process analytical data can issue warnings even for new types of disturbances that have not been encountered before
The feasibility of the procedure was proven by demonstrations. Based on demonstrations it is possible to create a software tool to estimate the process state from the feed variables and a predictive indication of final product quality deviations can be given to the operators. The mathematical methods employed are suitable for a wide range of measurement types but are most useful for complex spectral data. Since XRF analysers can be used for both solid and liquid samples, the methods find applications in both hydrometallurgy and pyrometallurgy. This gives a background to increase plant productivity and operability with means of new types of process control tools.
Outotec, Lappeenranta University of Technology