Traceability in continuous ore beneficiation processes using process mineralogy signatures

Sammanfattning: Traceability is the means to identify and follow real or imaginary lots through a process chain. It gives the opportunity to back-track a chain of events, or to predict process outcomes given the origin of a lot. Traceability can be used in different areas, e.g., in the middle of 1990´s traceability was a hot subject when different cases of food-carried diseases were exposed, but traceability is also used in other process industries to follow material and products during the processing. There is, however, a lack of traceability in continuous processes to compare with batch processes. The reason is that it is more complicated to identify a lot, and reach good traceability, in continuous process just because of they are contiguous. In this work, the continuous ore beneficiation process at LKAB (Luossavaara-Kiirunavaara AB, Sweden) Malmberget has been investigated in detail. The purpose is to trace the ore through the grinding sections by parameters and signatures like particle mineralogy, mineral associations and particle texture. To follow the material through the different process steps, analytical methods like optical microscopy and Particle Texture Analysis (PTA) are used. In paper I, different traceability methods to achieve traceability in continuous processes are explained. The advantages and disadvantages are presented for each method. In paper II, is showing the relations between the materials that comes into the grinding circuits and data collected from the PTA analysis is subjected to multivariate data analysis. The combination of automated process mineralogy and multivariate analysis is unique, and is first presented in this paper. The study in paper III is on an apatite-iron ore deposit at Malmberget, Sweden, and characterises an ore body both mineralogically and texturally in a quantitative manner by using different analytical methods. Paper IV explains how external flows are re-routed into the ordinary flowsheet and what impact they have. Paper V focus on the same data from previous paper by using multivariate data analysis. Comparisons between the PTA and QEMSCAN are discussed in paper VI. The differences come into play, when the data is further processed. In particular the liberation analyses seem to be dependent on the algorithms used, and their tolerance limits.