Correction of Estimates of Mine-Model Block Grades at Cyprus Pima Mine

Ore estimation at Cyprus Pima mine begins with estimating average copper grades for mine model blocks. Accuracy of the estimates has been checked four times in the last several years by comparing the estimates with arithmetic averages of corresponding blasthole (BH) assays. The data file now contains 3.567 estimated blockgrades and more than 10,000 blasthole assays. These data were compared by straight-line regression analysis at appropriate intervals as the body of data grew. Arithmetic averages of the estimates differed each time from arithmetic averages of the BH averages by nearly the same amount. More significantly, regression curves of the new total sets of data did not change position appreciably showing that copper grade estimates of blocks in all grade ranges from the lowest to the highest grade ores remained in error by nearly the same amount. Other studies revealed that core hole assays of the dominant ore type are biased and that that bias almost equals the error in estimation. Discovery of constancy in error indicated that regression analysis could be a means for correcting estimates of average block copper grades. A correction for estimated block grades was developed. It required determining which regression is appropriate for the particular problem, and a mechanism that is demonstrably sound. This paper describes the problem and the solution: No comparable study has been, found in the literature. Two other applications of regression analysis to metal grade comparisons are presented. Each adds a bit of support to a generalization which was suggested by correction of average block copper grades, namely, when performance of an estimate is to be evaluated, the variable taken to be the yardstick for the performance should be regressed on the estimate.