Evaluation of complex copper ore sorting: Effect of optical filtering on particle recognition

Abstract

In mixed and multi-component mineral systems, the reflectance spectra of individual minerals interfere with the reflectance/absorbance features of other minerals and random spectra can form. This contradictory case causes difficulty in implementing ore sorting to certain minerals. Hence, filtering out certain wave lengths may prevent this situation which makes it difficult for an ore sorter to perform a selective and efficient separation. The scope of this study is to investigate the feasibility of the optical filtering to enhance the differences in the reflectance features of mineral grains, and to compare the possible separation performances of the ore sorters under the best filtered and unfiltered conditions.Within the experimental studies, a new experimental setup which runs its uniquely coded software was designed. Additionally, an adjustable filter cartridge apparatus was developed enabling to mount the wavelength specific filters in front of the conventional sensors which support universal f-mount type lenses. A representative sample batch consisting of 31 grains of complex copper ore was prepared. Filtered and unfiltered images of each grain in between 450 and 1650 nm wavelength range were acquired carefully with the designed experimental setup. Following the data extraction by imaging studies, each particle was weighed, ground, and analyzed to determine chemical compositions. Thus, the reflectance values and chemical analysis data of each particle with each filter were obtained. Principal component analysis (PCA), partial least square analysis (PLS) and stepwise multiple linear regression (SMLR) methods were used to evaluate the resulting data in order to establish meaningful relationships between dominant chemical contents and reflectance values of each individual particle. Hence, the effect of using optical filters in altering the differences among the reflectance values of complex copper ore grains has been determined. Following the determination of best filters and wave lengths, projected product qualities of a conjectural optical sorting operation integrated with optical filters have been analytically investigated. It was concluded that an alteration in the relationship between % sulphide content and reflectance values was observed over 1500 nm wavelengths in comparison to the unfiltered and other filtered conditions. Evaluation of the overall results showed that up to 73.30% of Au, 96.14% of Ag, and 86.24% of CuFeS2 in feed can be recovered when a long-pass 1500 nm filter was used.