Lab-scale error analysis on X-ray fluorecence sensing for bulk ore sorting

Abstract X-Ray Fluorescence is one of the sensing technologies with potential in bulk ore sorting application. However, its reliability is often questioned since X-Ray Fluorescence does not sufficiently penetrate and errors are associated with surface measurements representing the volumetric grade. In this paper, error analysis was conducted on the XRF measurements on bulk ore materials at lab scale. Three main sources of the errors were identified, termed as instrumental error, heterogeneity error and the sampling error. Among the errors, the heterogeneity error was identified as the most significant due to its irrelevance with measuring the concentration and the unpredictability of that value. The influence of the errors on the conventional linear regression approach was investigated in terms of the sorting results and economics. Results showed that the linear regression approach was expected suitable for cases when sufficient analysis interval time could be ensured. It this was allowed for, the applicability of XRF in the bulk ore sorting was fundamentally justified.