An Optical Sensor for On-Line Analysis of Phosphate Minerals
A study has been conducted to investigate methods for rapid determination of BPL content in phosphate ores using optical sensing. technology. In this investigation, two approaches have been considered. The first approach involves the use of an optical fiber sensor to determine the relative reflected intensity of white light incident on a phosphate sample. The relative reflected intensity is inversely related to the BPL content. The second approach involves the use of a U.V. light source to create photoluminescence in a phosphate sample. The intensity of the photoluminescence is directly related to the BPL content.
Results of this study indicate that both approaches are capable of providing an indication of BPL content; however, the photoluminescence technique is limited by the low intensity of the resulting fluorescence output, and by the difficulty in differentiating the fluorescence output from the normal reflectance emanating from the sample. Furthermore, the photoluminescence technique is not able to eliminate the effect of phosphate color variation on the resulting calibration curve. The reflected light technique is also affected by variations in the phosphate mineral color; however, a procedure for normalizing the data has been developed which greatly reduces this effect.
A calibration curve has been developed using samples from IMC’s Kingsford, Four Comers and Noralyn operations. A statistical analysis of this curve indicates that the reflected light technique is capable of analyzing a typical amine concentrate or tailing sample within approximately ±30% relative error for a 90% prediction band (i.e., the analysis of 9 samples out of 10 will fall within ±30% of the true assay). Included within this relative error is the error associated with the chemical assays used as standards. A detailed analysis of these assays indicates that in some cases this error can be as high as 25%.
Previous experience with sensor calibration indicates that the relative error of the optical sensor can be greatly reduced by including additional points in the calibration curve and by generating separate calibration curves for concentrate and tailing. In fact, a video-based offline analyzer, which has been extensively calibrated and tested on site, is currently being used at the Texasgulf phosphate operation in North Carolina to analyze amine concentrate samples within ± 1% relative error.
Preliminary tests on slurry samples using the reflected light technique indicate that the accuracy of wet analysis is comparable to that of dry analysis. However, since these tests were conducted using a stationary sample, further work is needed to determine how the sensor will respond when the slurry is in motion. Based on the results obtained from this study, it appears that future work should be focused on two areas. In the long term, the reflected light technique shows sufficient promise to warrant further development. Future work in this area should concentrate on adapting the sensor to slurry applications and testing it in an on-line configuration. In the short term, the normalization procedure developed in the present work indicates that the video-based sensor developed for North Carolina phosphate could be adapted for Florida phosphate. The current success of this sensor in North Carolina warrants the installation and testing of this device in Florida.