Flow
visualization around a particle bubble aggregate

normal"> 

normal">Keeping
in view of flotation recovery plummet for coarse particles as particle bubble
detachment becomes obvious in the turbulent flow field, however, the nature of
hydrodynamics? influence on this process remains mysterious. To explore
turbulence?s influence on particle bubble detachment process, flow
visualization was carried out where a bubble particle aggregate (BP) was
positioned in the centre of a pair of oscillating grids and turbulent liquid
motion was supplied. This process was captured by particle image velocimetry (PIV)
system to highlight turbulent liquid motion influence on BP detachment.

normal">In
tune with the aims, the experimental methodology adopted in this work consisted
of measurement of instantaneous velocity field around a particle bubble
aggregate using PIV and laser induced fluorescence (LIF). The liquid used was
Milli-Q water and atmospheric air was used to generate the bubble. The particle
used was a stainless steel ball with diameter of 3 mm.

justify;text-indent:36.0pt;line-height:normal">Turbulence
generator consisted of a rectangular Perspex tank with a pair of vertically
orientated grids moving horizontally in and out together. The two grids
were connected via connecting rods and linear bearings to variable speed
stepper motors. A capillary assembly used for bubble generation was mounted
vertically in the centre of the tank with the top of the nozzle being 50 mm
above the bottom of the grids. A font-family:"Times New Roman","serif"'> stainless steel ball with diameter of 3
mm
was mounted on the tip of the capillary tube. The experimental procedure for
the detached bubble experiments involved operating the micro syringe pump to
generate a bubble at the top of the particle. At this stage the grids were not
operating and the liquid was quiescent. The grids were set in motion and the
oscillation frequency was increased and remained at 4 Hz.

justify;text-indent:36.0pt;line-height:normal"> font-family:"Times New Roman","serif"'> 

justify;text-indent:36.0pt;line-height:normal"> font-family:"Times New Roman","serif"'>Velocity field measurements in the
centre region of the tank in the x-y plane were performed using a PIV system with
LIF
as shown in Fig.1. The system comprised: (1) Litron LDY 300 laser capable of
generating 30 mJ/pulse energy at 1000Hz: (2) Optics used to produce 2-D light
sheet at 512 nm wavelength: (3) Dantec Hisense camera with 1600x1600 pixels and
8-bit resolution.

justify;text-indent:36.0pt;line-height:normal"> font-family:"Times New Roman","serif"'> 

justify;text-indent:36.0pt;line-height:normal"> font-family:"Times New Roman","serif"'>PIV is an optical experimental
technique, where seeding particles? movements are traced from the reflected
laser light (green light with wavelength of 527nm). Bubble reflects stronger
laser light than seeding particles due to its large size, which not only
causing velocity contamination around the bubble but also damaging camera?s sensor.
Instead of general seeding particles, fluorescent particles were used, which
emitted orange light under the activation of laser light. An optical long wave
pass filter was placed in front of the camera to block the intense flaring of
the green laser light at the bubble interface. The filter had a transmission
edge at 570nm±5nm, which allowed almost all of the light emitted from the
fluorescent tracer particles, with emission wavelength of 555-585nm and peak at
566 nm, to be recorded by the camera.

justify;text-indent:36.0pt;line-height:normal"> font-family:"Times New Roman","serif"'> 

justify;text-indent:36.0pt;line-height:normal"> font-family:"Times New Roman","serif"'>The camera and laser, and image capture
software were synchronized through a Dantec hub in double frame mode. Time gap
between two frames was modulated by changing capturing frequency in such a way
that seeding particles? movement in the image pairs were conspicuous enough to
calculate trustworthy velocity field. 12.0pt;font-family:"Times New Roman","serif"'>The laser intensity was adjusted
to allow appropriate intensity of leakage light from the filter reflected from
the bubble, while ensuring that fluorescent tracer particles were recorded by
the camera as a sharp image.

justify;line-height:normal"> font-family:"Times New Roman","serif"'> 

Fig.1 A
schematic representation of PIV system

justify;line-height:normal"> 

An image processing program, written in
MATLAB R2011a, was used to perform this operation. A sequence of image
processing filters was applied to mark the boundary of the bubble. Firstly, the
raw PIV image is considered as is shown in Fig.2 a. The light intensity of the
fluorescing particle was always greater than that of the bubble. On a light
intensity scale of 0-255 the particle intensity is usually higher than 120;
whilst the intensity of visible bubble-particle aggregate boundary is at around
70 and in the same range as the intensity fluctuations from the background. A
filter was applied to remove light intensities higher than a threshold value,
leaving only the background noise and bubble-particle aggregate boundary. A
median filter was applied to remove the background noise. Finally, bubble was
filled with ?imfill? function in MATLAB to produce the bubble-particle
aggregate boundary as in Fig.2 b. More seeding particles appear at bubble
periphery than bulk liquid in the raw PIV image, as the fluorescent particles
are hydrophobic and have higher tendency of attachment to the bubble. Image
processing technique treated the fluorescent particles in the close vicinity of
the bubble as a part of the bubble particle aggregate, making bubble shape
spiky. Then, the mask was superimposed with the vector field extract the liquid
velocity field surrounding the bubble during detachment. Camera sensor size was
1600´1600 pixels, while physical region of image
capture was 50´50 mm. Cross correlation was used to
calculate velocity vectors from movement of particles in raw PIV images. PIV
interrogation area was set at 32 pixels with 50 percent overlap giving 99´99 vectors. The resolution of vectors is
0.5 mm. Note that the image resolution is 0.03 mm/pixel, while vector
resolution is 0.5 mm/(vector) pixel. The mask generated from image of
bubble-particle aggregate was used to remove the spurious velocity vectors
inside the bubble-particle aggregate. Snapshot of flow field around detaching
particle bubble aggregate at critical state just before detachment is shown in
Fig.2. Mask contour was interpolated over vector region, with approximately 5
vectors over a length equivalent to a bubble diameter. It is noted in the
vector field that a jet of liquid from upper right is pushing the bubble to the
left side. Simultaneously, liquid motion on the left side of the bubble is
pushing the bubble to the right side. Under the action of the liquid motion,
the bubble is squeezed. The bubble detached under such squeezing effect of
liquid motion. Bubble detachment occurs when liquid velocity is high near the
bubble surface.

z-index:251659264;left:-7px;top:6px;width:625px;height:310px"> a. Raw PIV image b. Bubble mask c. Vector field

Fig.2 flow field superimposed with the mask of particle bubble aggregate
"> 

normal"> 

normal">A
time series of particle bubble detachment is shown in Fig.3. It is observed
that when liquid motion was weak bubble seated vertically on the top of
particle. When strong liquid motion developed on the left side of aggregate,
bubble was observed to be entrained and carried away by the eddy. Particle
bubble detachment process took 120 milliseconds.

normal">Discrete
wavelet transform (DWT) was used to decompose velocities from a snapshot of
velocity field around the bubble particle aggregate as DWT was proved to be
advantageous over other methods for the reason that no priori cut-off frequency
needs to be defined and flow structures can be decomposed scale to scale. DWT was
proved to be a sufficient method to decompose velocity field and to achieve
flow structure visualization, which helps to understand eddies? influence on
particle bubble aggregate detachment.

normal">Instantaneous
velocity shown in Fig.2 (c) was analysed using DWT. Fig.4 shows instantaneous
velocity field around the detaching bubble particle aggregate alongside with
the scalewised velocity field. Velocity inside particle bubble aggregate was
masked out. Interesting coherent flow structures are identified in scalewised velocity
field of scale 4, 5 and 6. A large eddy of two size of particle bubble
aggregate rotating above the bubble in clockwise direction pushing the bubble
to the left side. In scale 5, an eddy of similar size to bubble particle
aggregate rotating anticlockwise on the left side of bubble pushes bubble?s
boundary on left side. Simultaneously, two smaller eddies rotating on bubble?s
root of both sides in scale 6. With combined effect from all the acting eddies,
bubble boundary is pushed and squeezed which leads to its detachment.

normal"> 

normal">

normal">Fig.3
Time series of particle bubble detachment process

           

normal">In
this study, instantaneous velocity field was captured for the process of
particle bubble detachment. DWT method was applied to achieve flow structure
visualization in analysing snapshot of critical flow field just before particle
bubble detachment. For the first time, velocity information was obtained for a
three phase system where a particle bubble aggregate detach in the turbulent
liquid flow field. Flow structure visualization was beneficial in understanding
turbulence?s influence on particle bubble detachment which is fundamental in
understanding recovery drop off for coarse particle in froth flotation
industry.

absolute;left:52px;top:-68px;width:552px;height:937px"> 

Fig.4 Flow structure visualization around the detaching
bubble