Abstract:
The Anglo American Platinum Converter Plant produces a copper-nickel sulphide converter matte which is
slow-cooled in ingots over several days. During the process, the formation of alloy platelets, containing
the majority of the PGM’s and Au, occurs. The alloy forms a magnetic fraction in the bulk matte which can
be liberated when the matte is crushed and milled. The alloy platelets are then separated via a magnetic
separation process in the Magnetic Concentration plant. The quality of the converter matte is dependent
on the reaction and cast temperatures, bulk matte composition and cooling rate of each ingot, which
define the microstructure of the slow-cooled matte that is produced as the final product of the ACP
process.
The current mould size used in the ACP slow-cool aisle is ~10 tonnes (maximum 14.84t), designed for 2 full
ingots to be poured from every tap from a full ladle with a capacity of ~20t. This is the ideal situation,
however, during normal plant operation ladles develop a build-up of material on the inside (or skull) that
reduces the ladle active volume. This results in a large number of half ingots being produced, and, given
the smaller ingot size, suboptimal cooling conditions arise and subsequently poor quality Waterval
Converter Matte (WCM) is produced because of rapid cooling.
In an attempt to match the converter blow size, the ladle size and the slow-cool mould size in future, a
larger mould size of 15 ton was specifically constructed to determine if the smelter converted matte can
be cast into the larger mould size. Therefore a larger size ingot of 15t with a maximum capacity of 20.6t
was also used as part of the trials in order to determine if cooling rates in the larger sized ingot
necessitated longer cooling times, and would therefore negatively affect the platinum pipeline and
working inventory. The data obtained from these trials were then used to develop and validate a CFD
model, specifically developed to simulate the cooling process.
From the heat loss data it was calculated that the bulk of the heat lost from the ingots is through the top
surface. After the first 12 hours of cooling, approximately 89% of the heat lost from the ingot is via the top
surface and this increases to 96% after 24 hours. It can therefore be concluded that the cooling rates of
the ingots can be easily manipulated by changing the thermal insulation of the lid that is placed over the
ingot after casting.
