When iron ore is crushed and split into lump and fines components, there are systematic differences between the lump and fines grades. Generally the lump product is richer in iron and lower in the other minerals, compared to the fines product._x000D_
The grade differences between lump and fines, together with the lump percentage, are referred to as the lump algorithm'._x000D_
Mine estimates of grade, based on exploration or blasthole drilling, relate to head grade. The lump algorithm' predicts lump and fines product grades from the head grade. This knowledge is important when developing mine plans. Different ore types can be expected to have systematically different lump algorithms._x000D_
Previously, lump algorithms for specific ore types have been estimated by running the crusher with a single ore type over a sample period. This method is costly, interferes with production, and yields results of limited statistical power._x000D_
It has been found that weighted-least-squares (WLS) multiple regression can be used to ascribe lump algorithms to different ore types, provided sample periods have sufficiently varied ore type mixes._x000D_
This study discusses the method and its application to the analysis of Mining Area C crusher production records to estimate the separate lump algorithms, which could be applied to the different ore types. The analysis has enabled the set up of lump algorithms in the production system, to improve the grade control of lump and fines product._x000D_
Some other relevant applications of the WLS multiple regression approach are also considered.