Some Insights into the Process Design of Ferronickel Smelters

Once upon a time, the process engineer developing the design criteria for a new saprolite ferronickel smelter had it easy: most ores were around 2.5% Ni, with a Fe/Ni ratio of 4 to 5 and a SiO2/MgO ratio of about 1.6; to use the French word: garnierites. The ferronickel grade selected was the conventional 20 to 25% Ni and the impurity levels could be predicted from well established [C], [Cr] and [Si] vs (FeO) graphs. However the process engineer can no longer depend on statistics to design smelters: Ni grades are decreasing while the goethite contents are increasing; SiO2/MgO ratios vary from 1 to 3; and Al2O3 concentrations are reaching levels that cannot be ignored. Mechanistic models must be developed to select or predict essential parameters such as kiln and furnace reductions, ferronickel grade, slag losses, metal and slag liquidus, operating temperatures, etc. But these models must be simplified so that they can be used on a routine basis; and they must be calibrated against a wide range of commercial operations in order to yield the most realistic predictions possible. The present paper proposes one possible approach, using FactSage as a basic source of thermodynamic data. The predictions thus obtained are compared to actual performance for two well documented operations, as well as to the predictions made with other tools that might be more accessible.