7/24/2019 dupuis modelos de celdas.ppt
1/42
Modeling Aluminum Reduction
Cell since 1980 and Beyond
Marc Dupuis
7/24/2019 dupuis modelos de celdas.ppt
2/42
Plan of the presentation Introduction
Past developments1980, 2D in-house potroom ventilation model
1984, 3D ANSYS based thermo-electric half anode model
1986, 3D ANSYS based thermo-electric cathode side slice and cathode corner model
1988, 3D ANSYS based cathode potshell plastic deformation mechanical model
1992, 3D ANSYS based thermo-electric quarter cathode model
1992, 3D ANSYS based thermo-electric pseudo full cell and external busbars model
1992, 3D ANSYS based potshell plastic deformation and lining swelling mechanical model
1993, 3D ANSYS based electro-magnetic full cell model
1993, 3D ANSYS based transient thermo-electric full quarter cell preheat model
1993, 2D CFDS-Flow3D based potroom ventilation model
1994, in-house lump parameters dynamic cell simulator
1998, 3D ANSYS based thermo-electric full cell slice model
1998, 2D+ ANSYS based thermo-electric full cell slice model
1999, 2D+ ANSYS based transient thermo-electric full cell slice model
2000, 3D ANSYS based thermo-electric full quarter cell model
2000, 3D ANSYS based thermo-electric cathode slice erosion model
7/24/2019 dupuis modelos de celdas.ppt
3/42
Plan of the presentation
Past developments
2001, 3D CFX-4 based potroom ventilation model
2002, 3D ANSYS based thermo-electric half cathode and external busbar model
2003, 3D ANSYS based thermo-electric full cathode and external busbar model
2004, 3D ANSYS based thermo-electric full cell and external busbar model
2004, 3D ANSYS based full cell and external busbar erosion model
Future developments Weakly coupled 3D thermo-electric full cell and external busbar and MHD model
3D fully coupled thermo-electro-magneto-hydro-dynamic full cell and external busbar model
3D fully coupled thermo-electro-magneto-mechanico-hydro-dynamic full cell and externalbusbar model
3D fully coupled thermo-electro-magneto-mechanico-hydro-dynamic full cell and externalbusbar model weakly coupled with a 3D potroom ventilation model
Conclusions
7/24/2019 dupuis modelos de celdas.ppt
4/42
Introduction
Aluminum
reduction cells arevery complex to
model because it is
a truly multi-
physics modeling
application
involving, to be
rigorous, a fusionof thermo-electro-
mechanic and
magneto-hydro-
dynamic modeling
capabilities in a
complex 3Dgeometry.
7/24/2019 dupuis modelos de celdas.ppt
5/42
1980, 2D potroom ventilation model
Experimental results Best model results
The best results of my Ph.D. work: the 2D finite difference vorticity-streamfunction formulated model could not reproduces well the observed air flow
regardless of the turbulence model used.
7/24/2019 dupuis modelos de celdas.ppt
6/42
1984, 3D thermo-electrichalf anode model
That model was
developed onANSYS 4.1
installed on a
shaded VAX 780
platform.
That very first
3D half anodemodel of around
4000 Solid 69
thermo-electric
elements took 2
weeks elapse
time to compute
on the VAX.
7/24/2019 dupuis modelos de celdas.ppt
7/42
1986, 3D thermo-electric cathodeside slice and cathode corner model
The next step was
the development
of a 3D cathode
side slice thermo-
electric model
that included the
calculation of the
thickness of the
solid electrolytephase on the cell
side wall .
Despite the very
serious
limitations on the
size of the mesh, afull cathode
corner was built
next .
7/24/2019 dupuis modelos de celdas.ppt
8/42
Design of 2 high amperage cell cathodes
1987: Apex 4 1989: A310
Comparison of the predicted versus measured behavior was within 5%in both cases, demonstrating the value of the numerical tools developed.
7/24/2019 dupuis modelos de celdas.ppt
9/42
1988, 3D cathode potshell plasticdeformation mechanical model
The new model
type addresses adifferent aspect
of the physic of
an aluminum
reduction cell,
namely the
mechanical
deformation ofthe cathode steel
potshell under
its thermal load
and more
importantly its
internal
pressure load .
7/24/2019 dupuis modelos de celdas.ppt
10/42
1988, 3D cathode potshell plasticdeformation mechanical model
7/24/2019 dupuis modelos de celdas.ppt
11/42
1992, 3D thermo-electricquarter cathode model
With the upgrade of
the P-IRIS to 4D/35processor, and the
option to run on a
CRAY XMP
supercomputer, the
severe limitations on
the CPU usage were
finally partiallylifted.
This opened the door
to the possibility to
develop a full 3D
thermo-electric
quarter cathodemodel.
7/24/2019 dupuis modelos de celdas.ppt
12/42
1992, 3D thermo-electric pseudo full celand external busbars model
As a first step towar
the development of first thermo-electro-
magnetic model, a 3
thermo-electric
pseudo full cell an
external busbars
model was develope
That model was real
at the limit of what
could be built and
solved on the
available hardware
the time both in term
of RAM memory anddisk space storage.
7/24/2019 dupuis modelos de celdas.ppt
13/42
1992, 3D cathode potshell plastic deformationand lining swelling mechanical model
The empty quarter
potshell mechanicalmodel was extended to
take into account the
coupled mechanical
response of the swelling
lining and the restrainin
potshell structure.
As the carbon lining
swelling due to sodium
intercalation is somewh
similar to material
creeping, different mod
that represented that
behavior were develope
7/24/2019 dupuis modelos de celdas.ppt
14/42
1992, 3D cathode potshell plastic deformationand lining swelling mechanical model
That coupling was
important to conside
as a stiffer, morerestraining potshell
will face more
internal pressure
from the swelling
lining material.
Obviously, thatadditional load
needed to be
considered in order
truly design a potsh
structure that will n
suffer extensiveplastic deformation.
7/24/2019 dupuis modelos de celdas.ppt
15/42
1993, 3D electro-magnetic full cell modelThe development o
a finite element
based aluminum
reduction cellmagnetic model
clearly represente
a third front of
model developmen
Because of the
presence of theferro-magnetic
shielding structur
the solution of the
magnetic problem
cannot be reduced
to a simple Biot-
Savard integration
scheme.
19933D i h l i
7/24/2019 dupuis modelos de celdas.ppt
16/42
1993, 3D transient thermo-electricfull quarter cell preheat model
The cathode quarter
thermo-electric model
was extended into afull quarter cell
geometry in preheat
configuration and ran
in transient mode in
order to analyze the
cell preheat process .The need was urgent,
but due to its huge
computing resources
requirements, the
model was not ready in
time to be used to solvthe plant problem at
the time.
19933Dt i tth l ti
7/24/2019 dupuis modelos de celdas.ppt
17/42
1993, 3D transient thermo-electricfull quarter cell preheat model
In 2000, solving
a 30 hourspreheat using
36 load steps
required 171.5
CPU hours on
a Pentium III
800 MHz
computer.
The results file
containing all
the 36 load
steps results
required 3.711
GB of diskspace.
19932DCFDSFl 3D
7/24/2019 dupuis modelos de celdas.ppt
18/42
1993, 2D CFDS-Flow3Dpotroom ventilation model
2D Reynolds flux model results vs. physical model results
1994l
7/24/2019 dupuis modelos de celdas.ppt
19/42
1994, lump parametersdynamic cell simulator
Originally commercialized under the name ARC/DYNAMIC,the upgraded simulator is now available under the name DYNA/MARC
970
980
990
1000
0 4 8 12 16 20
Measured Calculated
Time (Hrs)
19983Dthermo-electric
7/24/2019 dupuis modelos de celdas.ppt
20/42
1998, 3D thermo-electricfull cell slice model
As described previously,
the 3D half anode model
and the 3D cathode side
slice model have been
developed in sequence,
and each separately
required a fair amount of
computer resources.
Merging them togetherwas clearly not an option
at the time, yet it would
have been a natural thing
to do. Many years later,
the hardware limitation
no longer existed so theywere finally merged.
1998,2D+thermo-electric
7/24/2019 dupuis modelos de celdas.ppt
21/42
1998, 2D+ thermoelectricfull cell slice model
2D+ version of the same
full cell slice model was
developed. Solving atruly three dimensional
cell slice geometry using
a 2D model may sound
like a step in the wrong
direction, but depending
on the objective of thesimulation, sometimes it
is not so.
The 2D+ model uses beam
elements to represent
geometric features lying
in the third dimension(the + in the 2D+ model).
19992D+transientthermo-electric
7/24/2019 dupuis modelos de celdas.ppt
22/42
1999, 2D+ transient thermo-electricfull cell slice model
An interesting feature of
that model is the
extensive APDL codingthat computes other
aspects of the process
related to the different
mass balances like the
alumina dissolution, the
metal production etc.
As that type of model
has to compute the
dynamic evolution of the
ledge thickness, there is
a lot more involved than
simply activating the
ANSYS transient modeoption.
19992D+transientthermo-electric
7/24/2019 dupuis modelos de celdas.ppt
23/42
1999, 2D+ transient thermo-electricfull cell slice model
20003Dthermo-electric
7/24/2019 dupuis modelos de celdas.ppt
24/42
2000, 3D thermo-electricfull quarter cell model
The continuous
increase of the
computer power now
allows not only to
merge the anode to the
cathode in a cell slice
model but also in a full
quarter cell model .
The liquid zone can
even be included if the
computation of the
current density in that
zone is required for
MHD analysis.
20003Dthermo-electric
7/24/2019 dupuis modelos de celdas.ppt
25/42
2000, 3D thermoelectricfull quarter cell model
2000,3Dthermo-electric
7/24/2019 dupuis modelos de celdas.ppt
26/42
2000, 3D thermoelectriccathode slice erosion model
Cathode erosion rate is
proportional to the
cathode surface currendensity and that the
initial surface current
density is not uniform,
the erosion profile will
not be uniform.
Furthermore, that initiaerosion profile will
promote further local
concentration of the
surface current density
that in turn will promot
a further intensification
of the non-uniformity o
the erosion rate.
2000,3Dthermo-electric
7/24/2019 dupuis modelos de celdas.ppt
27/42
2000, 3D thermoelectriccathode slice erosion model
20013DCFX4potroomventilationmode
7/24/2019 dupuis modelos de celdas.ppt
28/42
2001, 3D CFX-4 potroom ventilation mode
Streak lines
20013DCFX4potroomventilationmode
7/24/2019 dupuis modelos de celdas.ppt
29/42
2001, 3D CFX-4 potroom ventilation mode
Temperature
fringe plot
20023Dthermoelectrichalfcathode
7/24/2019 dupuis modelos de celdas.ppt
30/42
2002, 3D thermo-electric half cathodeand external busbar model
Temperature BusbarsVoltage Drop
20023Dth l tihlf thd
7/24/2019 dupuis modelos de celdas.ppt
31/42
2002, 3D thermo-electric half cathodeand external busbar model
Voltage Drop
in Metal
2003,3Dthermo-electricfullcathode
7/24/2019 dupuis modelos de celdas.ppt
32/42
2003, 3D thermoelectric full cathodeand external busbar model
A P4 3.2 GHz computer
took 16.97 CPU hours tobuild and solve that model
2003, 3D thermo-electric full cathode
7/24/2019 dupuis modelos de celdas.ppt
33/42
003,3 te oeect cu catodeand external busbar model
Voltage drop in
metal pad: 5 mV Current density in metal pad
2004, 3D thermo-electric full cell
7/24/2019 dupuis modelos de celdas.ppt
34/42
,and external busbar model
A P4 3.2 GHz
computer took 26.1
CPU hours to buildand solve that model
2004, 3D thermo-electric full cell
7/24/2019 dupuis modelos de celdas.ppt
35/42
,and external busbar erosion model
Once the geometry of
the ledge is
converged, a new
iteration loop start,
this time to simulate
the erosion of thecathode block as
function of the
surface current
density
F t d l t
7/24/2019 dupuis modelos de celdas.ppt
36/42
Cell
Design
Stress models which aregenerally associated withcell shell deformation and
cathode heaving issues.
Magneto-hydro-dynamic
(MHD) models which aregenerally associated with
the problem of cell stability.
Thermal-electric models
which are generallyassociated with theproblem of cell heat
balance.
Currently, we can fit Hall-Hroult mathematical models
into three broad categories:
Future developments
-0.01
0
0.01
0.02
H
01
23
45
6 0
1
2
Y
X
Z
LevelDH:
1-0.008
2-0.007
3-0.005
4-0.003
5-0.002
6-0.000
70.002
80.003
90.005
100.007
Interface
7/24/2019 dupuis modelos de celdas.ppt
37/42
MHD is affected by the ledge profile, mostly dictated by the cell heat
balance design.
local ledge profile is affected by the metal recirculation pattern mostly
dictated by the busbars MHD design .
shell deformation is strongly influenced by the shell thermal gradient
controlled by the cell heat balance design.
steel shell structural elements like cradles and stiffeners influence the
MHD design through their magnetic shielding property.
global shell deformation affects the local metal pad height, which inturn affects both the cell heat balance and cell stability
Future developments
Yet, to be rigorous, a fusion of those three types of model into a fully
coupled multi-physics finite element model is required because:
Weaklycoupled3Dthermo-electricfullcel
7/24/2019 dupuis modelos de celdas.ppt
38/42
Weakly coupled 3D thermoelectric full celand external busbar and MHD model
MHD model:
gives the liquids/ledge interface
boundary conditions based on
the average flow solution
Thermo-electric model:
gives the position of the ledge profilebased on the heat balance solution
- -magneto-hydro-dynamicfullcell
7/24/2019 dupuis modelos de celdas.ppt
39/42
magnetohydrodynamic full cell
and external busbar model
3D fully coupled thermo-electro-
7/24/2019 dupuis modelos de celdas.ppt
40/42
y pmechanico-magneto-hydro-dynamic full
cell and external busbar model
3D fully coupled thermo-electro-mechanico
7/24/2019 dupuis modelos de celdas.ppt
41/42
3 u ycoupedte oeecto eca comagneto-hydro-dynamic full cell and
external busbar model weakly coupled wit
a 3D potroom ventilation model
Conclusions
7/24/2019 dupuis modelos de celdas.ppt
42/42
Only a truly multi-physics modeling application could be
used as a design tool in order to fully take into accountsall of the complex interactions taking place in a H.H.
cell.
On the other hand, such a model even if it could be
available today, could not be used as a practical design
tool as it would require far too much computerresources to have a manageable turn around time and
operating cost.
As for past developments, the author believes that the
rate of future model development will be mainlydictated by the Moore law.
Conclusions
Top Related