XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 2557
plagioclase, and hornblende, which together make up
77.5% (w/w) of the ore. Ore samples were collected in the
processing plant directly after the final milling stage, being
its particle size distribution suitable for flotation (d80,3 =
170 µm). Aliquots of 500 g were split using Retsch PT100
to be used as feed for each test. Prior to each flotation test,
the ore was added to 300 ml of water and ground in a labo-
ratory scale rod mill for 3 minutes at 96 rpm to reactivate
particle surfaces. Froth flotation tests were conducted in a
Magotteaux ® bottom-driven flotation cell with 33% pulp
density, an impeller speed of 450 rpm, and an air flow
rate of 5 l/min. The conditioning scheme is illustrated in
Figure 2.
Within time intervals of [0 60], [60 180], and [180
420], three concentrates were collected. The froth was
scraped off every five seconds. The filtered flotation prod-
ucts were dried overnight at 50°C to 60°C in a drying oven,
and mass recoveries were calculated based on dry weights.
All samples were split and then analyzed with a Bruker S1
TITAN 800 portable XRF (pXRF). In order to translate
pXRF results into mineralogical compositions based on the
element-to-mineral conversion methodology, a few samples
were examined using XRD and the Mineral Liberation
Analyzer (MLA)(Fandrich et al. 2007).
Experimental Conditions
The DoE workflow proposed for the batch flotation is
shown in Figure 4. Three stages make up the workflow:
(1) screening (2) investigation and modelling and (3) veri-
fication and optimization. The screening phase’s purpose
is to eliminate process parameters that have no significant
effect on the response variables. Fractional factorial design is
the most common design type utilized for screening. In the
investigation and modelling phase, a full-factorial DoE is
carried out varying the significant process parameters after
the irrelevant ones have been screened out. Testing every
possible combination of the independent process param-
eters and how it affects the response variables is the aim.
Both main and two-factor interaction effects are evaluated.
To determine whether the assumed model has a curvature
a centre point is added to the design. A model assumption
(linear, second order…) can be made and verified in the next
phase. Central composite designs are extensions of the fac-
torial experimental designs. They are composed of a partial
factorial or full factorial experimental design and a so-called
CCD star with a central experiment whose extreme settings
go beyond those of the factorial experimental design. The
central composite design composes our workflow as part
of the verification of the assumed model in the case of a
second-order model assumption. It is assumed that the full
factorial experimental design with centre points already
performed is sufficiently ample to estimate curvature effects
(Ahn 2015). Once the model is established and verified, the
process can be numerically optimized.
Two experimental phases were carried out follow-
ing the screening stage, where it was determined that the
frother dosage had no significant effect on the relevant tar-
get variables and as consequent fixed at a dosage of 25 g/t
during the further investigation. The variables in the first
stage (Table 2), which can be thought of as the reference
blank test, were the pH and the dosage of sodium oleate.
The effect of the colloidal silica dosage, pH and sodium
oleate dosage were the process parameters investigated in
the second stage (Table 3), which was carried out for each
colloidal silica suspension.
The experimental plans were created and evaluated
using a shiny R platform that was developed in-house. This
platform presented in IMPC2022 is open source and can
be found in (Ben Said et al. Asia-Pacific 2022).
Table 2. Levels of variables investigated for the blank tests
(22 full factorial design)
Parameter Unit
Factor Level
Minimum Maximum
pH — 8 10
Collector dosage g/t 100 400 Figure 2. Flotation flow chart of low grade scheelite ore
plagioclase, and hornblende, which together make up
77.5% (w/w) of the ore. Ore samples were collected in the
processing plant directly after the final milling stage, being
its particle size distribution suitable for flotation (d80,3 =
170 µm). Aliquots of 500 g were split using Retsch PT100
to be used as feed for each test. Prior to each flotation test,
the ore was added to 300 ml of water and ground in a labo-
ratory scale rod mill for 3 minutes at 96 rpm to reactivate
particle surfaces. Froth flotation tests were conducted in a
Magotteaux ® bottom-driven flotation cell with 33% pulp
density, an impeller speed of 450 rpm, and an air flow
rate of 5 l/min. The conditioning scheme is illustrated in
Figure 2.
Within time intervals of [0 60], [60 180], and [180
420], three concentrates were collected. The froth was
scraped off every five seconds. The filtered flotation prod-
ucts were dried overnight at 50°C to 60°C in a drying oven,
and mass recoveries were calculated based on dry weights.
All samples were split and then analyzed with a Bruker S1
TITAN 800 portable XRF (pXRF). In order to translate
pXRF results into mineralogical compositions based on the
element-to-mineral conversion methodology, a few samples
were examined using XRD and the Mineral Liberation
Analyzer (MLA)(Fandrich et al. 2007).
Experimental Conditions
The DoE workflow proposed for the batch flotation is
shown in Figure 4. Three stages make up the workflow:
(1) screening (2) investigation and modelling and (3) veri-
fication and optimization. The screening phase’s purpose
is to eliminate process parameters that have no significant
effect on the response variables. Fractional factorial design is
the most common design type utilized for screening. In the
investigation and modelling phase, a full-factorial DoE is
carried out varying the significant process parameters after
the irrelevant ones have been screened out. Testing every
possible combination of the independent process param-
eters and how it affects the response variables is the aim.
Both main and two-factor interaction effects are evaluated.
To determine whether the assumed model has a curvature
a centre point is added to the design. A model assumption
(linear, second order…) can be made and verified in the next
phase. Central composite designs are extensions of the fac-
torial experimental designs. They are composed of a partial
factorial or full factorial experimental design and a so-called
CCD star with a central experiment whose extreme settings
go beyond those of the factorial experimental design. The
central composite design composes our workflow as part
of the verification of the assumed model in the case of a
second-order model assumption. It is assumed that the full
factorial experimental design with centre points already
performed is sufficiently ample to estimate curvature effects
(Ahn 2015). Once the model is established and verified, the
process can be numerically optimized.
Two experimental phases were carried out follow-
ing the screening stage, where it was determined that the
frother dosage had no significant effect on the relevant tar-
get variables and as consequent fixed at a dosage of 25 g/t
during the further investigation. The variables in the first
stage (Table 2), which can be thought of as the reference
blank test, were the pH and the dosage of sodium oleate.
The effect of the colloidal silica dosage, pH and sodium
oleate dosage were the process parameters investigated in
the second stage (Table 3), which was carried out for each
colloidal silica suspension.
The experimental plans were created and evaluated
using a shiny R platform that was developed in-house. This
platform presented in IMPC2022 is open source and can
be found in (Ben Said et al. Asia-Pacific 2022).
Table 2. Levels of variables investigated for the blank tests
(22 full factorial design)
Parameter Unit
Factor Level
Minimum Maximum
pH — 8 10
Collector dosage g/t 100 400 Figure 2. Flotation flow chart of low grade scheelite ore