6
explaining 75.69% and PC2 explaining 16.06%. Figure 6
presents the PCA biplot, showing collectors and perfor-
mance metrics along PC1 and PC2. PC1 is positively corre-
lated with Cu recovery and Cu distribution in the +210 μm
fraction and negatively with Cu distribution in the +75 μm
and +150 μm fractions. This suggests that high PC1 scores
indicate high recovery and higher losses in the +210 μm
fraction, while low PC1 scores indicate higher losses in the
+75 μm and +150 μm fractions. PC2 is positively correlated
with Cu distribution in the –75 μm fraction and negatively
with Cu distribution in the +210 μm fraction, indicating
that high PC2 scores correspond to higher losses in the
–75 μm fraction and lower losses in the +210 μm fraction.
Collectors like C1 and C11 perform well with high recov-
ery and low losses in the +75 μm and +150 μm fractions,
despite slightly higher losses in the –75 μm fraction. In
contrast, C10 and C12 are less efficient, with higher losses
in the –75 μm fraction and lower overall recovery. C9 is the
least efficient collector, with the lowest recovery and highest
losses in the +75 μm fraction. Collectors like C8 and C13
show high recovery but need optimization to reduce losses
in the +210 μm fraction.
The Hierarchical Clustering Analysis (HCA)
The HCA grouped the collectors based on their perfor-
mance metrics into four distinct groups, as shown in the
dendrogram in Figure 7, revealing the clustering process
and the similarity between collectors. The analysis iden-
tified four distinct groups. Group 1, which includes col-
lectors C1, C11, C2, and C3, features high similarity and
some of the best performers like C1 and C11, exhibiting
high recovery rates and low copper losses in the +75 μm
and +150 μm fractions. Group 2, including C4, C13,
and C8, also contains high-performing collectors, such as
C8 and C13, which demonstrate high recovery but need
optimization to reduce losses in the +210 μm fraction.
Group 3, comprising C5, C7, and C6, shows moderate
performance with varying recovery rates and copper dis-
tributions, indicating a mix of efficiency and inefficiency.
Group 4, consisting of C9, C10, and C12, includes the
least efficient collectors, such as C9 and C10, which have
low recovery rates and high copper losses, particularly in
the +75 μm fraction.
CONCLUSIONS
This study introduced a novel methodology that combines
metallurgical and statistical analysis to classify and select
flotation collectors effectively. By evaluating thirteen col-
lectors from various chemical families in Clariant’s product
portfolio, the study aimed to identify the most effective col-
lectors for copper recovery while minimizing copper losses
in flotation tailings.
The metallurgical analysis revealed that collectors C1,
C8, C11, and C13 demonstrated superior performance in
Figure 6. PCA biplot of Cu recovery and Cu distribution by size in the tailings for all tested
collectors
Previous Page Next Page