Predicting Nickel Recoveries at the Thompson Mill from Feed Mineralogy

Manitoba Mining and Milling operations consist of three underground mines (Thompson T1, T3 and Birchtree), feeding the Thompson Mill. Three years ago, a collaborative effort involving Mines Exploration, Mines Project Development Group, Mill Technical Services and Vale Base Metals Technical Excellence Center was initiated to establish and verify the cause of high variability in concentrate quality and Ni recovery at the Thompson Mill. One outcome of this work was a program to systematically characterize the mineralogy and metallurgical performance of current and future orebodies feeding the mill. Mineralogy results show that high value enabling ores typically have a high pentlandite to pyrrhotite ratio and benign silicate mineralogy (mostly quartz and feldspar from sedimentary host rocks), while metallurgically challenging ores contain some combination of pyrite and/or hexagonal (non-magnetic) pyrrhotite and/or elevated quantities of talc (with associated carbonates) derived from altered ultramafic host rocks. The metallurgical performance of these two ore types has been simulated by an Excel particle flotation model, which calculates the relative floatability of particles based on the particle’s mineral composition and mineral locking determined from MLA images. The relationship between these mineral properties measured by MLA and metallurgical performance was established using 13 monthly composite samples from Thompson Mill. The model was then validated with results of laboratory flotation experiments on individual orebodies submitted for evaluation. The validated model can be used to predict plant metallurgical performance based on individual ore bodies feeding the mill. Individual orebody’s grade-recovery curves are assumed to be additive, and can be weighted in proportion to the mill feed to predict mill recovery over the same time period. The ore mixing can be done with a simple Excel spreadsheet and comes with a graphical output for easy visualization of the resulting grade-recovery curve.