2896 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
also significant (p =0.047). None of the other factors were
found to be statistically significant.
COMPLEX ORE METHODOLOGY
While the model ore case study provided a good under-
standing of the entrainment in model conditions, ores are
always complex in practice, with particle properties such as
liberation, size, shape, and association influencing flotation
response. In order to account for ore and particle complex-
ity in entrainment studies, flotation tests were performed
on a porphyry copper ore using the same downscaled lab
version (6 liters) of FLSmidth’s nextSTEP ™ mechanical
flotation cell. Experiments were also planned following a
full factorial Design of Experiment (DoE) approach, but
having only 2 input variables (Vt and Jg) at 2 levels and a
mid-point. Each set of conditions was repeated thrice. Pulp
density (30% solids content), collector concentration (5 g/
tonne of AERO 3894, technical grade), and frother con-
centration (3 g/tonne of X-133, technical grade), and pH
(11.3) were kept constant for all tests. Four concentrates
were collected at cumulative intervals of 20, 60, 120, and
240 seconds. Process conditions for the complex ore are
summarized in Table 2.
All the fractions (four concentrates and tailings) were
sampled and analyzed scanning electron microscope (SEM)-
based automated mineralogy for detailed particle character-
istics. SEM analysis were performed on FEI Quanta 650F
scanning electron microscope equipped with two Bruker
Quantax X-Flash 5030 EDX detectors. The overall elec-
tron beam accelerating voltage of the SEM was 25kV and
Extended BSE Liberation Analysis measurement mode was
used. For each particle in the sample, this analysis provided
information on size, shape, mineral composition surface,
overall mineral composition and association. These particu-
late properties were used to train logistic regression (LR)-
based particle separation models (PSMs) that quantify the
recovery probabilities of each particle in a sample. A PSM
was trained for each experimental setup using particle data
collected from the flotation products following the proce-
dure introduced in Pereira et al. (2021b).
Several gangue minerals were used as proxies for
entrainment indicators in this case. These included quartz,
orthoclase, illite, muscovite, and biotite. Using particle
data, recovery probabilities of the gangue minerals allowed
for the understanding of gangue minerals at different
hydrodynamic conditions. Settling velocities of the afore-
mentioned minerals were computed at 10μm and 30μm to
obtain an indication of the order in which these minerals
should be entrained. The settling velocities (U) were calcu-
lated in a non-turbulent water environment using the fol-
lowing equation, where g is the gravitational acceleration,
Dp is the particle size, ρp and ρw are the particle and water
densities respectively, and C is the drag coefficient.
U C
4gD
3t
.5
p p w
w
0 t t
=
-^h F
C is assumed to be 0.47 for orthoclase and quartz given
their rounder shape, as compared to 0.82 for the micas (illi-
tie, muscovite, and biotite). The calculated settling veloci-
ties are shown in Table 3.
RESULTS
Recovery probability for all fully liberated particles (95%
surface liberation) of quartz, orthoclase, illite, muscovite,
and biotite at different hydrodynamic conditions are shown
in Figure 8. The highest recovery probability among the
gangue minerals was found to be that of biotite in most
cases while the least for illite. Increasing the Jg resulted in
an increase in recovery probability of all the indicated min-
erals in both cases, although the difference is small. Increase
in Jg also contributed towards the entrainment of coarser
particles, which is evident from the particle size at which
the recovery probability approaches zero. This size is higher
at higher Jg values. Increasing Vt also increased the recov-
ery probability of all gangue minerals, and promoted the
entrainment of coarser particles.
Figure 9 shows the recovery probability of the major
gangue minerals at a particular process conditions (Jg
=0.5 cms–1, Vt =5.5 ms–1), divided into different size
Table 2. Process conditions for the copper ore experimental campaign
Variable Minimum Center-Point Maximum
Superficial gas velocity (Jg) 0.40 cms–1 0.45 cms–1 0.50 cms–1
Impeller tip speed (Vt) 4.20 ms–1 4.85 ms–1 5.50 ms–1
Table 3. Settling velocities of fully liberated particles for
various gangue minerals at size 10μm and 30μm
Mineral
Particle Settling Velocity (ms–1)
@10 μm @30 μm
Illite 0.0168 0.0291
Muscovite 0.0168 0.0291
Biotite 0.0183 0.0316
Orthoclase 0.0208 0.0361
Quartz 0.0214 0.0371
also significant (p =0.047). None of the other factors were
found to be statistically significant.
COMPLEX ORE METHODOLOGY
While the model ore case study provided a good under-
standing of the entrainment in model conditions, ores are
always complex in practice, with particle properties such as
liberation, size, shape, and association influencing flotation
response. In order to account for ore and particle complex-
ity in entrainment studies, flotation tests were performed
on a porphyry copper ore using the same downscaled lab
version (6 liters) of FLSmidth’s nextSTEP ™ mechanical
flotation cell. Experiments were also planned following a
full factorial Design of Experiment (DoE) approach, but
having only 2 input variables (Vt and Jg) at 2 levels and a
mid-point. Each set of conditions was repeated thrice. Pulp
density (30% solids content), collector concentration (5 g/
tonne of AERO 3894, technical grade), and frother con-
centration (3 g/tonne of X-133, technical grade), and pH
(11.3) were kept constant for all tests. Four concentrates
were collected at cumulative intervals of 20, 60, 120, and
240 seconds. Process conditions for the complex ore are
summarized in Table 2.
All the fractions (four concentrates and tailings) were
sampled and analyzed scanning electron microscope (SEM)-
based automated mineralogy for detailed particle character-
istics. SEM analysis were performed on FEI Quanta 650F
scanning electron microscope equipped with two Bruker
Quantax X-Flash 5030 EDX detectors. The overall elec-
tron beam accelerating voltage of the SEM was 25kV and
Extended BSE Liberation Analysis measurement mode was
used. For each particle in the sample, this analysis provided
information on size, shape, mineral composition surface,
overall mineral composition and association. These particu-
late properties were used to train logistic regression (LR)-
based particle separation models (PSMs) that quantify the
recovery probabilities of each particle in a sample. A PSM
was trained for each experimental setup using particle data
collected from the flotation products following the proce-
dure introduced in Pereira et al. (2021b).
Several gangue minerals were used as proxies for
entrainment indicators in this case. These included quartz,
orthoclase, illite, muscovite, and biotite. Using particle
data, recovery probabilities of the gangue minerals allowed
for the understanding of gangue minerals at different
hydrodynamic conditions. Settling velocities of the afore-
mentioned minerals were computed at 10μm and 30μm to
obtain an indication of the order in which these minerals
should be entrained. The settling velocities (U) were calcu-
lated in a non-turbulent water environment using the fol-
lowing equation, where g is the gravitational acceleration,
Dp is the particle size, ρp and ρw are the particle and water
densities respectively, and C is the drag coefficient.
U C
4gD
3t
.5
p p w
w
0 t t
=
-^h F
C is assumed to be 0.47 for orthoclase and quartz given
their rounder shape, as compared to 0.82 for the micas (illi-
tie, muscovite, and biotite). The calculated settling veloci-
ties are shown in Table 3.
RESULTS
Recovery probability for all fully liberated particles (95%
surface liberation) of quartz, orthoclase, illite, muscovite,
and biotite at different hydrodynamic conditions are shown
in Figure 8. The highest recovery probability among the
gangue minerals was found to be that of biotite in most
cases while the least for illite. Increasing the Jg resulted in
an increase in recovery probability of all the indicated min-
erals in both cases, although the difference is small. Increase
in Jg also contributed towards the entrainment of coarser
particles, which is evident from the particle size at which
the recovery probability approaches zero. This size is higher
at higher Jg values. Increasing Vt also increased the recov-
ery probability of all gangue minerals, and promoted the
entrainment of coarser particles.
Figure 9 shows the recovery probability of the major
gangue minerals at a particular process conditions (Jg
=0.5 cms–1, Vt =5.5 ms–1), divided into different size
Table 2. Process conditions for the copper ore experimental campaign
Variable Minimum Center-Point Maximum
Superficial gas velocity (Jg) 0.40 cms–1 0.45 cms–1 0.50 cms–1
Impeller tip speed (Vt) 4.20 ms–1 4.85 ms–1 5.50 ms–1
Table 3. Settling velocities of fully liberated particles for
various gangue minerals at size 10μm and 30μm
Mineral
Particle Settling Velocity (ms–1)
@10 μm @30 μm
Illite 0.0168 0.0291
Muscovite 0.0168 0.0291
Biotite 0.0183 0.0316
Orthoclase 0.0208 0.0361
Quartz 0.0214 0.0371