2716 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
into a 10 µm sintered stainless steel sparger enclosed by
a chamber of compressed air. The feed slurry flows past
the internal sparger surface, shearing fine bubbles to be
entrained with the fluid. This creates a turbulent bubbly
mix to flow through the downcomer and into the middle
of the vertical section. The bubbles that rise to the top of
the rig are counter-currently washed by fluidization water
that enter through holes at each of the vertical faces of the
column. The bubbly flow that reaches the top goes through
a reduced opening that connects to a flexible tube, flowing
into a collection tank. At the end of the inclined channels
is an inverted pyramid section connected to a pump, which
draws the separated fluid out of the system into a collec-
tion tank separated from the feed. The feed, underflow and
wash water all operate using peristaltic pumps. Under high
fluxes, the bed will expand down into the incline channels
and reach a steady-state position.
The downcomer has an external diameter of 2.64 cm,
creating a cross-sectional area for above the downcomer exit
is ~74.5 cm2. The inclined channels, holding three plates,
are spaced approximately 2 cm apart creating 4 parallel
channels.
A differential pressure transducer (Dwyer Instruments
Series 645–1 differential pressure transmitter with an
accuracy of ± 0.25% full scale, equivalent to a maximum
inaccuracy of ± 35 Pa) is placed on the side panel collect-
ing the pressure data at a sampling rate of 100 Hz. The
pressure data were logged using a National Instruments
NI cDAQ-9171 chassis and NI 9203 module combina-
tion for recording and post-analysis of the data. They were
used to determine the gas holdup in the vertical section,
between the washing zone and the downcomer exit (Wu
et al., 2020). On the opposite side of the vertical section, a
camera equipped with a 10× zoom lens was setup to record
slow motion video of bubble rise within the same zone. A
strong light source with a diffuse pane was used to clarify
the image of the bubbles.
Feed Preparation
To overcome the challenge of feed variability a feed
preparation and delivery method, outlined in the work
of Crompton, Islam, &Galvin (2022), was utilized. The
methodology was adjusted to allow the use of a smaller
mixing tank.
Each experiment used approximately 23.5 kg of silica,
suspended in ~242 L of water dosed with MIBC and/or
NaCl making an ~8.7 wt. %slurry feed. The feed was
distributed into fifty 20 L buckets, each filled to approxi-
mately 5 L each. Subsequently, the tank was refilled and
dosed, without extra silica, to further dilute the contents
of each bucket by an additional 10 L. This process ensured
that each volume reached the desired concentration of 3
wt.%. The buckets were then arranged into a randomized
order. The wash water tank was filled and dosed to the same
MIBC and NaCl concentration as the feed.
Experimental Process
The mixing tank was first filled up to 220 L, using the
buckets in the randomised order. Each bucket was resus-
pended with a bucket mixer before adding to the tank. The
gas was switched on before any fluid enters the rig. The
RFC was filled with wash water, and pressure probes were
checked and calibrated. The wash water flowrates were cali-
brated and recorded, by taking a 60 s sample and recording
its mass. The RFC was then drained. The feed flowrates
were calibrated and recorded in the same way. The RFC was
then filled with the feed, from which the tailings rate was
calibrated and recorded. Each flowrate recorded was made
in triplicate. While solids were in the rig, the wash water
remained off, such that any mass taken out of the rig could
be resuspended into the tank to maintain pulp consistency.
Once rates were calibrated, the run began with all the
outputs (tailings and product) placed into the waste hold-
ing tank. Buckets of feed were then continually fed into the
mixing tank to maintain the tank level. After a few min-
utes, underflow and product measurements were taken to
ensure correct rates. Adjustments were made when needed.
Sample measurements were taken once steady state was
achieved. Steady state was determined by a steady pressure
drop reading and consistent product rates. The pressure
drop was recorded for approximately 5 minutes. During
this time, bubble behaviour was recorded for 1 minute,
along with a calibration image. Then mass samples of the
product and tailings were taken every 5 minutes, along
with a smaller untimed jar sample for each output. Feed
samples were taken at the end of each condition to prevent
disruption to the experiment during the steady-state sam-
pling. The pH and temperature of the feed were maintained
at 7.33 ± 0.23 and 22.1 ± 1.6°C.
Data Analysis
The timed mass samples were weighed and dried to deter-
mine the pulp density, recovery, and calculated flowrates.
The small sample jars were used to check pH, conductivity,
and particle size distribution. The solution conductivity was
measured to determine the salt content in the dried mass
samples. The particle sizing was processed using a Malvern
Mastersizer 3000. Frames were extracted from the slow-
motion video, using a python script. Frames were spaced by
200 ms to avoid duplicate measurements. An open-access
into a 10 µm sintered stainless steel sparger enclosed by
a chamber of compressed air. The feed slurry flows past
the internal sparger surface, shearing fine bubbles to be
entrained with the fluid. This creates a turbulent bubbly
mix to flow through the downcomer and into the middle
of the vertical section. The bubbles that rise to the top of
the rig are counter-currently washed by fluidization water
that enter through holes at each of the vertical faces of the
column. The bubbly flow that reaches the top goes through
a reduced opening that connects to a flexible tube, flowing
into a collection tank. At the end of the inclined channels
is an inverted pyramid section connected to a pump, which
draws the separated fluid out of the system into a collec-
tion tank separated from the feed. The feed, underflow and
wash water all operate using peristaltic pumps. Under high
fluxes, the bed will expand down into the incline channels
and reach a steady-state position.
The downcomer has an external diameter of 2.64 cm,
creating a cross-sectional area for above the downcomer exit
is ~74.5 cm2. The inclined channels, holding three plates,
are spaced approximately 2 cm apart creating 4 parallel
channels.
A differential pressure transducer (Dwyer Instruments
Series 645–1 differential pressure transmitter with an
accuracy of ± 0.25% full scale, equivalent to a maximum
inaccuracy of ± 35 Pa) is placed on the side panel collect-
ing the pressure data at a sampling rate of 100 Hz. The
pressure data were logged using a National Instruments
NI cDAQ-9171 chassis and NI 9203 module combina-
tion for recording and post-analysis of the data. They were
used to determine the gas holdup in the vertical section,
between the washing zone and the downcomer exit (Wu
et al., 2020). On the opposite side of the vertical section, a
camera equipped with a 10× zoom lens was setup to record
slow motion video of bubble rise within the same zone. A
strong light source with a diffuse pane was used to clarify
the image of the bubbles.
Feed Preparation
To overcome the challenge of feed variability a feed
preparation and delivery method, outlined in the work
of Crompton, Islam, &Galvin (2022), was utilized. The
methodology was adjusted to allow the use of a smaller
mixing tank.
Each experiment used approximately 23.5 kg of silica,
suspended in ~242 L of water dosed with MIBC and/or
NaCl making an ~8.7 wt. %slurry feed. The feed was
distributed into fifty 20 L buckets, each filled to approxi-
mately 5 L each. Subsequently, the tank was refilled and
dosed, without extra silica, to further dilute the contents
of each bucket by an additional 10 L. This process ensured
that each volume reached the desired concentration of 3
wt.%. The buckets were then arranged into a randomized
order. The wash water tank was filled and dosed to the same
MIBC and NaCl concentration as the feed.
Experimental Process
The mixing tank was first filled up to 220 L, using the
buckets in the randomised order. Each bucket was resus-
pended with a bucket mixer before adding to the tank. The
gas was switched on before any fluid enters the rig. The
RFC was filled with wash water, and pressure probes were
checked and calibrated. The wash water flowrates were cali-
brated and recorded, by taking a 60 s sample and recording
its mass. The RFC was then drained. The feed flowrates
were calibrated and recorded in the same way. The RFC was
then filled with the feed, from which the tailings rate was
calibrated and recorded. Each flowrate recorded was made
in triplicate. While solids were in the rig, the wash water
remained off, such that any mass taken out of the rig could
be resuspended into the tank to maintain pulp consistency.
Once rates were calibrated, the run began with all the
outputs (tailings and product) placed into the waste hold-
ing tank. Buckets of feed were then continually fed into the
mixing tank to maintain the tank level. After a few min-
utes, underflow and product measurements were taken to
ensure correct rates. Adjustments were made when needed.
Sample measurements were taken once steady state was
achieved. Steady state was determined by a steady pressure
drop reading and consistent product rates. The pressure
drop was recorded for approximately 5 minutes. During
this time, bubble behaviour was recorded for 1 minute,
along with a calibration image. Then mass samples of the
product and tailings were taken every 5 minutes, along
with a smaller untimed jar sample for each output. Feed
samples were taken at the end of each condition to prevent
disruption to the experiment during the steady-state sam-
pling. The pH and temperature of the feed were maintained
at 7.33 ± 0.23 and 22.1 ± 1.6°C.
Data Analysis
The timed mass samples were weighed and dried to deter-
mine the pulp density, recovery, and calculated flowrates.
The small sample jars were used to check pH, conductivity,
and particle size distribution. The solution conductivity was
measured to determine the salt content in the dried mass
samples. The particle sizing was processed using a Malvern
Mastersizer 3000. Frames were extracted from the slow-
motion video, using a python script. Frames were spaced by
200 ms to avoid duplicate measurements. An open-access