JJMIE Volume 4, Number 3, June, 2010
ISSN 1995-6665
Pages 412 - 417 Jordan Journal of Mechanical and Industrial Engineering
Simulation and Modeling of Bubble Motion in an Electrolytic
Bath of Soderberg Pot
C. Karuppannan, T.Kannadasan
CIT Sandwich Polytechnic College, Coimbatore-641 014, Tamilnadu, India
Abstract:
In Aluminium manufacturing the Soderberg pot is widely used. The study explains about the bubble motion in cryolite liquid in a Soderberg pot anode face using comprehensive VOF multiphase model in FLUENT, under the action of gravitation. The distribution of bubbles introduced under the anodes of an aluminum reduction cell at the initial time has fixed size. 2D simulation results are reported for four cases with different initial conditions for the bubbles, physics and geometries.
? 2010 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved Keywords: Bubble analysis; Aluminium manufacturing;numerical simulation;VOF model.
1.Introduction
In aluminium manufacturing, formation of bubbles is a
major problem. Soderberg pots are used for the
manufacture of aluminium. In Soderberg pot, the bubbles
are formed in the interpolar distance and deposited on the
anode face due to the thermal and chemical reactions in the
electrolytic bath. It reduces the input power and increases
the energy consumption and reduces the current efficiency.
Reduction in the formation of bubbles will increase the
current efficiency of the pot and output of aluminium.
From the previous studies, the temperature distribution
is
clearly studied and proved. It has been reported that
formation of bubbles is from the anode surface. Hence, a
model of the anode surface is taken for the analysis. Initial
distributions of bubbles and diameters of the bubbles are
taken at random. Bubbles considered are having a non-
zero density and deformable body; therefore having an
appropriate mass (depending on the volume), hence
providing a more accurate mathematical model. With the
aid of comprehensive Volume Of Fluid(VOF) model in
FLUENT under the action of gravitation, the formed
bubbles are coalescence and collapsed. The result is
compared with the mathematical model. The above work
proves that the bubbles are collapsed with the increase in
pressure in the feeding of Alumina. In this we can achieve
the result accurately and enhance the current efficiency
and quality of Aluminium production.
https://www.wendangku.net/doc/4b15401533.html,erning Equations
To solve the given cryolite and bubble phase flow, field
employs the conservation equations for mass and
momentum for incompressible fluid.
(1)
(2)
Momentum equations are solved for both cryolite and
fluid. Continuum surface model (CSF) is used to describe
the interfacial surface tension.
(3)
Equation (3) represents the VOF model governing
equation for the cryolite-bubble phase model, where
density and viscosity are defined
and
Primary volume fraction will be computed based on
(4)
3.Numerical Solution
Properties of cryolite and bubble used for the analysis
are shown in Table 1. Model and meshing are done in
GAMBIT.
Table 1 Properties of Cryolite and Bubble
Property Cryolite Air (Bubble)
Density2567 kg/m3 1.225 kg/m3
Viscosity 1.57 cP0.017894 cP
Surface Tension0.05 N/m
(5)
For interpolation of cryolite-bubble interface cell
region, Geometric Reconstruction Scheme is used. The
geometric reconstruction scheme represents the interface
between fluids using a piecewise-linear approach. When
the cell is near the interface between two phases, the
geometric reconstruction scheme is used to obtain the face
fluxes. It assumes that the interface between two fluids has
a linear slope within each cell, and uses this linear shape
? 2010 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved – Volume4, Number 3 (ISSN 1995-6665) 413
for calculation of the advection of fluid through the cell faces. The first step in this reconstruction scheme is calculating the position of the linear interface relative to the center of each partially-filled cell, based on information about the volume fraction and its derivatives in the cell. The second step is calculating the amount of fluid through each face normally using the computed linear interface representation and information about the normal and tangential velocity distribution on the face. The third step is calculating the volume fraction in each cell using the balance of fluxes calculated during the previous step.
Body force weighted scheme for differencing, Explicit scheme for temporal and PISO for pressure velocity coupling is used for solving. 4.Results & Discussions
4.1.Case 1: Coalescence with Gravity: Small Bubble and Big Bubble Below
Domain dimensions are 20 mm x 10 mm surface. Geometry with three fluid zones is modelled viz., Cryolite, Bigger bubble and smaller bubble. Bubble walls are marked as interior. Results at time steps 0s, 0.03s, 0.033s, 0.036s are shown.
a) initial configuration at t = 0s b) t = 0.03s
t = 0.033s t = 0.036s
Figure 4-1. Coalescence of two rigid circular bubbles. The contours of the bubbles as well as the fluid velocity vectors are shown.
The above results show that the bubbles move through
the cryolite medium towards upwards under the action of buoyancy since the system is in the gravitational field. The bubbles having a lower density than the surrounding liquid medium displace the liquid, volume equivalent to their own volume and this displaced liquid wants to push the bubble upwards so as to occupy its volume. This is in accordance with the Archimedes' principle. Also, bigger bubble moves faster than the smaller bubble due to different buoyant force exerted depending on their volume. This phenomenon therefore contributes to the different velocities of the bubbles.
The coalescence occurs already at time 0.03 seconds as opposed to the results depicted in the paper (3). This can be just due to different densities of the continuous fluid (in our case cryolite) which causes different speeds being attained by the bubble, hence decreasing the time to coalesce. The animation also shows the bigger bubble attaining a protruded shape in the beginning owing to being in the wake of the smaller bubble due to which it is in a low pressure region and gets specially pulled into the smaller bubble before coalescence occurs. These findings
can be verified by the paper (3).
Figure 4-2. Typical C-shape morphology attained by the smaller bubble. Protruding shape of the bigger bubble,
caught in the wake of the rising smaller bubble
4.2.Case 2: Surface Tension Effect: Single Elliptical
Bubble with Different Surface Tension Values of the Fluid
Domain dimensions are 20 mm x 12 mm. Normal
bubble rising problem with constant surface tension
specification (0.05 N/m). Results at time steps t = 0.1s, t =
0.1s, t = 0.05s are shown.
? 2010 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved – Volume4, Number 3 (ISSN 1995-6665)
414
a) initial configuration, t=0s
b) bubble configuration at t = 0.1s, αs
= 0.05 N/m
c) t = 0.1s, αs
= 0.005 N/m
d) time 0.05s, αs = 0.0005 N/m
Figure 4-3. Influence of the surface tension coefficient αs
The above results match with the results of the publication qualitatively, although the more intricate details like bubble break-up owing to lower surface tension (figure 2.1(c)) have not been captured in the paper. Besides, the C-shape morphology attained due to the low pressure region in the wake of the rising bubble has been captured much better in the FLUENT simulation. The velocity vector plots overlaid on the bubble images for the four situations explain these phenomena .
Bubble configurations at a) t = 0.1s, αs
= 0.05 N/m
b) t = 0.1s, αs
= 0.005 N/m;
c) time 0.05s, αs = 0.0005 N/m
Figure 4-4. Rising bubbles attaining different shapes owing to the recirculation zone below them
These results are in perfect harmony with the simulation results of the independent group which was mentioned earlier (3).
4.3. Case 3: Rigid Ellipsoidal Bubble Rising Around an Angular Wall
Domain dimensions are 2 x 2 cm with two angled walls Normal bubble rising problem with constant surface tension specification (0.05 N/m). Surface tension was included to come closer to the case description. Due to
surface tension, the bubble tends to attain minimum surface area as the liquid has a phobia towards it. As the surface tension value is quite high in our case, the bubble should not undergo easy break-up.
? 2010 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved – Volume4, Number 3 (ISSN 1995-6665)
415
(a) t = 0 s
(b) t = 0.09 s
(c) t = 0.18 s
(d) t = 0.27 s
(e) t = 0.36 s (f) t = 0.45 s
Figure 4-5. Air bubble position at different times under stronger surface tension effects The bubble is at different positions compared to the
literature values. This must be due to different densities
being taken for the liquid phase (our liquid is cryolite). As
the surface tension effects are considerable, the bubble
tends to avoid contact with the liquid (and rather prefers
contact with the wall) while moving up under the action of
buoyancy. The following images show the bubble
behaviour for weaker surface tension effects of the liquid,
where the bubble does not necessarily oppose contact with
the liquid while moving up due to buoyancy
Figure 4-6. Air bubble position at different times under weaker surface tension effects
4.4. Case 4: Around 20 Bubbles with Random Distribution Rising in a Column
Domain dimensions are 8 cm x 4 cm. 20 numbers of bubbles are taken for the analysis. Analysis is done for 27 time steps
(a) t = 0 s
(b) t = 0.15 s
(c) t = 0.6 s
(d) t = 0.9 s
Figure 4-7 .Twenty air bubbles coalescing and rising under strong surface tension and buoyancy effect
The results show that the buoyant force acts on all of the bubbles and pushes them upwards. It is also visible that
the bubbles below are caught up in the wake of the ones
above them and are pulled towards such low pressure areas
due to which they end up moving even in a zigzag fashion. The surface tension value being high, results in the bubbles
coalescing as the liquid would prefer to minimize the surface contact area with the bubbles. As a result, the number of bubbles reduces drastically and ultimately there is an air column formed in the top region of the column. Some of the small bubbles are caught up in the eddies and tend to remain for relatively longer period within the liquid as they easily give-in to the fluid behavior due to their own lower inertia 5. Conclusion: Since the results obtained by the VOF multiphase model of FLUENT demonstrate the typical C-shape morphology of the bubbles as they rise up the column (as evident from the animation), the FLUENT results are much more accurate compared to the simplified mathematical model applied in the publication. Multi bubble analysis shows the real effect of bubble in the bottom of the anode plate in soderberg pot. The formed bubbles are collapsed with the aid of increasing the pressure in the feeding of Alumina and prevent the bubbles deposited on the anode surface. Hence, we can obtain the minimum energy consumption and enhance the current efficiency in the soderberg pot. References
[1] A New Study On Bubble Behavior On Carbon Anode In
Aluminum Electrolysis. Gao, Bingliang, et al. [ed.] Halvor Kvande. s.l. : TMS (The Minerals, Metals & Materials Society), Light Metals 2005. [2] Numerical Simulation Of Bubble Coalescence Using A
Volume Of Fluid (Vof) Model. Delnoij, E, Kuipers, J.A.M. and Van Swaaij, W. P. Lyon : s.n., 1998. Third International Conference on Multiphase Flow, ICMF’98. [3] A New Modelling For Simulating Bubble Motions In A
Smelter. Michel, V. Romerio, Alexei, Lozinski and Jacques, Rappaz. [ed.] Halvor Kvande. s.l. : The Minerals, Metals &
Materials Society), Light Metals 2005.
(a) t = 0.09 s
(b) t = 0.12 s
(
c) t = 0.18 s
(d) t = 0.21 s
? 2010 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved – Volume4, Number 3 (ISSN 1995-6665) 416
[4] B. Glorieux et al., “Density of Superheated and Undercooled
Liquid Alumina by a Cont actless Method”, Int. J.
Thermophys., Vol. 20, No. 4, 1999, 1085 – 1094. [5]T.J.Chung, Computational Fluid Dynamics, Cambridge, UK,
Cambridge University Press, 2002.
[6]M.Dupuis, “Computation of Aluminum reduction Cell
Energy Balance Using ANSYS? Finite Elem ent Models”, TMS Light Metals, 1998, 409 – 417.
[7]M. Dupuis and al., “Cathode Shell Stress Modelling”, TMS
Light Metals, 1991, 427 – 430.
[8]M. Dupuis, “ Computation of Accurate Horizonta l Current
Density on Metal Pad Using a Full Quarter Cell Thermo-electric Model”, CIM Light Metals, 2001, 3-11.
[9]M. Dupuis and I. Tabsh, “Thermo-Electro-Magnetic
Modeling of a Hall-Heroult Cell”, Proceeding of the ANSYS ? Magnetic Symposium, 1994, 9.3 – 9.13.
[10]M. Segatz and D.Vogelsang, “Effect of Steel Parts on
Magnetic Fields in Aluminum Reduc tion Cells”, TMS Light Metals, 1991, 393 – 398.
[11]D.Richard and al., “Thermo-electro-mechanical Modelling of
the Contact between Steel and Carbon Cylinders using the Finite Elem ent Method”, TMS Light Metals, 2000, 523-528. [12] I.Eick and D.V ogelsang, “Dimensioning of Cooling Fins for
High –Amperage Reduc tion Cells”, TMS Light Metals, 1999, 339 – 345.
[13]C.Vanvoren and al., “AP 50: The Pechiney 500 kA cell”,
TMS Light Metals, 2001, 221 -226.
[14]H.Kvande and W.Haupin, “Inert Anodes for AI Smelters:
Energy Balance and Environmental Impact” JOM, Vol. 53, No. 5, 2001, 29-33.
[15]Bruggeman and D.J.Danka, “Two-Dimensional Thermal
Modelling of the Hall-Heroult Cell”, Light Metals, 1990, 203-209.
[16]Schmidt-Hatting and al., “Heat Losses of Different Pots”.
Light Metals, 1985, 609-624.
[17]Antille and al., “Effects of Current Increase of Aluminium
Reduction Cells”, Light Metals, 1995, 315-321.