CA1254855A - Low energy aluminum reduction cell with induced bath flow - Google Patents

Abstract

Abstract In an electrolytic cell for the production of alu-minum from alumina, having an anode-to-cathode gap of from 1 to 5 cm, the flow of electrolyte to the inter-electrode gap is induced by positioning the cathode at an inclination of from about 2 to about 15 degrees from the horizontal, and operating at an anode current density of from about 0.5 to about 3.0 A/cm2. The cathode surface is a Refractory Hard Material which is wetted by aluminum.

Description

55ii L~W E~IERGY ALUMINIJM REI~IJCTIO~I
CELL WITH INOlJCED BATH FLnW

~ackground of the Invention _ _ _ This invention relates to a novel electrolytic cell for the manufacture of aluminum from alumina, and to the opera-tion of such a cell. More particularly, the invention relates to an aluminum producing electrolysis cell using bath electrolyte - based on sodium cryolite, wherein the problems resulting ln from reduced anode to cathode gap distances achieved in previous drained cathode cells are demonstrated to have ~heen overcome by inducing a particular manner of bath flow in the A~ gap, thus facilitating alumina feeding, removal of~gaseous products, and enhanced drainage of product metal.
1~ nperation of a test electrolysis cell has demonstrated the abillty to provide a plentiful supply of dissolved alumina to the electrolysis zone even at very narrow anode to cathode ~spacings.
A commonly utilized electrolytic cell for the manufac ~n ture~of aluminum is of the classic ~all-Heroult design, util~izing carbon anodes and a substantially flat carbon-lined bottom which functions as part of the cathode system.
~An electrolyte is used in the production of aluminum by electrolytic reduction of alumina, which electrolyte con~

2~ sists primarily of ~olten cryolite with dissolved alumlna9 and which may contain other materials such as fluorspar9 al;u~inum flIloride, and other metal fluoride salts. rlolten aluminum resulting from the reduction of alumina is most ~ frequently permitted to accumulate in the bottom of the 3n: receptacle formin~ the electrolytic cell, as a molten metal pdd or~pool over the carbon-lined bottom, thus forming a .
~'q,?~

': :

~::
: ~, ~$f~

liquid metal cathode. Carbon anodes extending into the receptacle, and contactlng the molten electrolyte, are ad~justed relative to the liquid metal cathode. Current collector bars, such as steel, are frequently embedded in the carbon-lined cell bottom, and complete the connection to the cathodic system.
While the design and sizes of l~all-Heroult electrolytic cells vary, all have a relatively low energy efficiency, ranging from about 3~ to 45 percent, dependent upon cell ln geometry and mode of Dperation. Thus, while the theoretical power requirement to produce one pound of aluminum is about 2.85 kilowatt hours (kWh)~ in practice power usage ranges from 6 to ~.5 kWh/lb, with an industry average of about 7.5 kWh/lb. A large proportion of this discrepancy from theore-l~ tical energy consumption is the result of the voltagedrop of the electrolyte between the anode and cathocle. As a result, much study has gone into reduction of the anode-cathode distance (AC~ However, because the molten aluminum pad which serves as the cell cathode can becorne irregular 2n and variable in thickness due to electromagnetic effects and bath circulation, past practice has required that the AC~ be kept at a safe 3.5 to h cm to ensure relatively high current efficiencies and to prevent direct shorting between the anode and the metal pad. Such gap distances result in ~5 voltage ~rops from l.4 to 2.7 volts, which is in addition to the enerqy required for the electrochemical reaction itself (2.1 volts, based upon enthalpy and free energy calculations).
Accordingly, much effort has been directed to developing a more stable aluminllm pad, so as to reduce the Acn to less ~n than ~.~ cm, with attendant ener~y savings.
Refractnry hard materials (RHM), such as titanium dihor-ide, have heen under study for quite some time for use as cathnde surfaces in the form oF tiles, but until recently, adherent R~ tiles or surface coatings have no~ been available.
~S Titanium dihoride is l~no~n to be conductive, as well as .,. . ~
.

$~

possessing the unique characteristic of being wet-ted by molten aluminum, thus permitting formation of very thin aluminum films. The use of a very thin aluminum film draining down an inclined ca-thode covered with an RHM surface~ to replace the unstable molten aluminum pad of the prior art, has heen suggested as a means to reduce the ACn, thus improv-ing efficiency, and reducing voltage drop. However, attempts to achieve such goals in the past have failed due to the inadequacy of available RH~l surfaces, and the inability to 1 n overcome the difficulty of providing a sufficient supply of dissolved alumina to the narrowed ACn (as small as 1.~ cm).
Thus, problems of alumina starvation occur at minimal ACD, including excessive and ~pe~ t- anode effects. Overfeed-in~ alumina to prevent -these problems has resulted in deposits 1~ of sludge (muckin~, which can clog the cell and restrain its operation.
An alternative approach to reducing energy consumption has been to smelt aluminuln from aluminum chloride rather than alumina. This process requires 3n to 40 per cent less 2n electrical power to produce aluminum than conventional electrolysis. In this process, the conventional ~ayer method is utilized to convert bauxite to alumina, which is converte~ to alumillum chloride in a chemical plant, then smelted in an electro1ytic cell. In the cell, the aluminul11 2~ chloride breaks down into aluminum, which is drawn off, and chlorine, which may be recycled back to the chemical plant for the pro-luction of more aluminum chloride. Sucn techni-ques utilize a ~low-through reactor, having non-consumable anodes. However, the aluminum-producing cells of this ~n process are incompatible with the Hall-Heroult type cell, and cannot be retrofitted to an existing aluminum plant.
Thus, the chloride process requires the capital expense of an entirely new installation. To a greater degree, the present lnvention permits appreciable energy savings in a ~S retrofitted plant, and obviates the need for completely new facllities.

It is against this hackground that the present inven-tion was developed.

Summary of the Invention It is an object of this invention to provide an improved aluminum electrolysis cell, utilizing a bath electrolyte based on sodium cryolite, and having improved electrical efficiencies~ It is thus a purpose of this invention to provide a basic cell for the production of aluminum metal from alumina which may be used in a single cell system, or 1~ in a multiple cell system, to achieve construction and operating economies~
It is an object of the present invention to provide a cell design which provides a narrower ACn than the prior art, and hence improved current efFiciencies and voltage 1~ drops, and which may be retrofitted to existing Hall cell installations. Accordingly, an object of the present invention is to provide an electrolysis cell with an ACD
of less than 3 cm, ~herein an adequate alumina supply to the anode-cathode gap is assured, ~ithout requiring overfeeding.
Additi~onal ob~jects of the invention are to provide an aluminum redllction cell design providing a flow-through configuration, an~l a controlled atmosphere.
These, as well as other objects, which will become more apparent in the disclosure which follows, are achieved by , 2~ ~providing a cell in which individual process steps are separ-ated to the greatest extent possible. A sloped solid cathode is ~Itilized to shape the anode, which may be either Soderberg or pre-baked, which in concert with the proper choice of other parameters defined herein induces such a 3n bath flow through the anode-cathode gap as to provide the proper alumina supply to the anode, ~ithout adverse impact on the drainage of aluminum product into a collection sump.
In one configuration, gas pumping action under the anode causes the hath to flow through the cell, into a cell feeding .

': ,' , .

: ' `:

, chamber, out of the cell feecling chamber, and back to -the opposite side of -the cell. Mu]tiple individual cells of this nature may be combined in numerous configurations to struc-ture the system for the physical and economic restrictions of any plant site.
Thus, the present invention relates to an electroly-tic cell in which the cathode is so arranged and inclined that the anode gases induce a p]entiful and sufficient flow of en-riched materials between the anode and cathode, without impeding the drainage of molten aluminum to a collection sump, or disrupting the surface of the draining aluminum so as to cause lower current efficiencies. Thus, the bath is induced to flow frorn an enriched bath zone to a zone where the bath is somewhat depleted in alumina content. The bath may be supplied with an automatic feed of alumina in a controlled amount, to preven-t either mucking or anode effects.
Therefore, in accordance with the present invention there is provided a drained cathode cell, for the electrolytic reduotion of alumina to aluminum in a cryolite-based bath con-taining alumina, comprising: a shell havlng inner surfaceslined with refractory and carbon makerials to define a cathode cavity; a cathode having at least an upper surface containing , an aluminum wettable refractory hard material; at least two ~J~"~t anodes depending into the cathode cavity and each having a lower surface spaced from the upper surface of the cathode by an anode-to-cathode distance of about 1 cm to about 5 cm de~ining an anode-cathode-displacement (ACD) and a flow path between the~anodes and the cathode, the upper cathode surface and the lower anode surfaces being sloped Erom the hori~ontal ch/ Ih _ 5 _ "~;.;~
~s ,~., .

, ~ . .

by from about 2 to about 15 thereby providing a lower end and an upper end of -the :Elow path between the anodes and -the cathode; means in the cavity for providing a minimum bath flow Q about at least one substantially horizonta.:L bath C.iXCU~
lation loop to ensure a supply of alumina for the elec-trolysis reaction wherein

3 (0.008)(Total cell current(amps)) Q(cell)(cm /sec)= ~ wt.~A12O3 by providing at least one return flow channel to complete the loop, the loop comprising:

(i) the flow path between the anodes and the cathode;
(ii) at least one lower channel in fluid communi-cation with lower end of the flow path between the anodes and the cathode;
(iii) at least one upper channel in fluid communi-catIon wIth the upper end of flow path between the anodes and the cathode;
(iv) at least one return flow channel in fluid : communication with the upper channel and the lower channel;

at least one return Elow channel, located between adjacent anodes, having dimensions h, w and L wherein h is the bath depth i.n cm at any given point, w is the width of the return channel in cm, and L is the length of the return channel in cm and wherein h, w and L are determined by the relationship "~ R~(cell channel)=( h 3 3)L

ch/~ Sa -~' : ~

6~.93~~36 and wherein the maximum value of RE(cell channel) is cleterm.ined by the relationships ~ n [ ~
[Rf(cell)]l/2 i=l [R~(cell channel)~l/2 and R~(cell) = (X)(8g0 cm /sec) Q (cell)2 wherein Rf is a flow resistance geometry term, Rf(cell) is said term as applied to the cell overall, Rf(cell channel) is said ~erm as applied to each cell channel, and wherein (X) is a geometric resistance factor having a value in a range of from 0.2 X 10 3/cm4 BRIEF~DESCRIP~ION OF THE DRA~INGS
~ ~Figure 1 constitutes an end vie~ of an elect.rolytic cell employing an inverted V cathode, or dual slope cathode con-figurations.
Figure 2 constitutes an end view of a cell concep-t of : the present lnvention, utilizing a single slope cathode~
Figure 3 constitutes an end view of a single-slope cathode~t ~low-through aluminum reduction cell having external ; replenishment.
~ Figure 4 constitutes an end view of a V-shaped drained cathode cell.
Figure 5 illustrates a plan view o~ a single slope dralned cathode cell, utilizing a continuous aluminum tapping s~stem and a separate replenishment and mixing zone~

- 5b -.
.

:

Figure 6 represents a p:Lan view of a multiple ccll system, utilizing a plural.ity of single sloped drained ca~hode cells in tandem.

:: :

- 5c -: ~ :

: ~, Figure 7 represents a plan view of a sing1e slope drained cathode cell employing partial harriers to restrict bath flow velocity.
Figure ~ represents a plan view of d cell having two ; ~ oppositely sloped cathodes.
Figure 9(a) illustrates a side elevation view of a hydraulic analog model designed to simulate gas evolution in a sloping cathode cellO
Figure 9(b) illustrates normal bath flow in the ACD.
ln Figure ~(c) illustrates a region of reverse bath flow in the ACn, adjacent to the draining aluminum metal.
Figure 9td) illustrates a region of turbulent and dis~
rupted bath flow, whereby no net flow of bath material through the ACD occursO
1~ Figure ln constitutes a graphic representation of bath flow versus cathode slope in the cell simulation model under high flow resistance conditions.
Figure 11 constitutes a graphic representation oF bath flow versus cathode slope in the cell simulation model under 2n medium flow resistance conditions~
; Figure 12 constitutes a graphic representation of bath ~flow versus cathode slope in the cell simulation model under lo~ flow resistance conditions.
Figure 13 is a conceptual pump efficiency-flow resis-tance diagram for a drained cathode cell.
Fi~ure 14 graphically illustrates required flow resis-tahce in the return channel versus ca-thode slnpe~
Figure 1~ constitutes a cell design parameter diagram.
Figure l h i S a cross section of a sloping cathode electrolysis test cell.
Figure 17 is a graphic representation of cell operational stability achieved in three test cells.
Fi~ure 1~ illus~rates the relationship between anode-cathode polariza-tion and cathode slope at t~o current den-3~ sities.

.~

.
~, 61.~93-.136 Descr~E~ of PreEerred Embodirnents The present invention cons:i.sts o~ util.izin~
sloped, drained cathode, having a s~rface of titanium diboride or other simi].ar refractory hard mater:ial, in which the slope or inclination from the horizontal is selected so as to contribute to a controlled bath motion, which bath motion is hydrodynamically dependent upon specifically identified critical parameters. The bath motion is so controlled as to induce sufficient alumina-rich bath to flow into the interelectrode gap in the spaceimm~diately beneath the anode face to avoid any anode effects, while not in~erfering with the drainage of aluminum. Thus, it is possible to overcome problerns of alumlna feeding, such as muck formation resulting from - excessive bath addition, persistant anode effects resulting from insufficient bath addition, and ledging, which interferes with bath circulation, which occur at the very low values of anode-cathode distance (ACD) made possible by~the development of refractory hard metal surface : 20 cathodes operating without a conventional metal pad.
Success of a low energy drained cathode aluminum ; ~ reductlon cell concept depends on two critical components:
I~ a durable R~IM cathode material, and ~) the ability to contro~l and optimize the bath flow through the narrow anode-cathode gap ~ACD). Ongoing tests in production cells indlcate that a material has been developed which satisfies the f:irst requirementO Copending Canadian Patent Applicatlon ~32~49~

,, ~ .

:

:

~L~

assigned to the ass:ignee of this appli.cation, relates -to Refractory Hard Material coa-ting compositions, applicat:ion rnethocls, and eathode structures prov.iding durable, aclherent cathode surface.s exhibiting aluminum wettability and improved electrical cha:racter-isties.

-- 7a -~s~

Studies usincJ a water model to simulate hydrodynamic condltions in the ~CD of a slopong cathode cell have now generated the cell design criteria necessary to satisfy the other requirement. Resul-ts from the water model have been verified in a laboratory scale electrolytic cell. A number of cell designs, such as in the Kaiser-DOE sloping TiB2 cathode reported tests and as used in other published reports and patents, fall outside our design re~uirements, and fail to address crucial hydrodynamic issues, the criticality of which has now been established. These design criteria include slope or inclination of the cathode surface, the ACD spacing, the resistance to bath flow in the re-turn path, and anode current density. Only the first two of these criteria have been considered in previously existing designs, reports, and patents, while the latter two criteria, and their criticality, are newly established herein.
It is well established in aluminum electrolysis that the anode face is configured by the electrolysis process to be essentially parallel to the cathode. Thus, an inclined ca-thode causes the anode to be similarly inclined. As is also well known, there exists a plentiful gas evolution at the anode face due to the electrolysis process.
It ls also well known for aluminum reduction cells of conventional design, with hori~ontal anode and cathode faces, tha-t gas evolved at the anode face travels sideways (laterally) outwards underneath the anode face from the center of the cell to the nearest vertically oriented ven-t, said vents generally comprising the ~48~i -8a- 61293~136 vertical side faces of the anode. A sloped anode may either enhance or inhibit this gas motion, depending on the direction and magnitude of the slope. A suitable cathode, and therefore anode, inclination can therefore be found to cause the anode gases to move in a desired direction. In turn, these anode gases will drive, by a well known action also utilized in the design of gas lifts and gas jet pumps, as well as bubble columns, the bath near the anode surface in the same desired direction~
This drive may be supplemented and enhanced (if necessary) by other pumping action, e.g. a suitably designed pump.
It is well known in hydrodynamics that the flow of a fluid system is established by a balance between the fluid drive and the resis-tance to flow within the components of the system, and that, dependent upon the configuration, the velocity within local regions flow may be in the same direc-ln tion but may sometimes be in the direction opposite to thedirection of the fluid drive. It is a principle of the present invention to so arrange the slope, or slopes, so as to achieve that balance between buoyancy-generated bubble forces from the inclination and those forces which drive hubbles sideways beneath horizontal anodes on the one hand, and the flow resistance on the other hand, to give a net motion of the bath to provide the re~uired alumina supply.
The local bath velocity near the anode surface is in the desired direction, i.e. the direction in which the driving %n gas hubbles move. At the same time, the configuration is so positioned to provide hath velocities near the cathode surface in the same direction as those near the anode, yet not so large as to~interfere with the drainage of aluminum at the cathode, which aluminum drainage is opposite in direction to the Flow of the gas bubbles driving the bath. With the source of alumina-rich bath located on a particular side of the cell, a bath flow into the space immediately beneath the anode face is achieved when this side of the anode is low with an upward slope away from this side. This slope must 3n be sufficiently large to overcome, and reYerse, the flow of gas which would otherwise be toward this anode edge from the inner parts of the anode. The precise configuration and arrangement of inclinations may vary with the location of the alumina rich ba-th supply. This may be on one side~ on 3~ both sides, or in the center of the cell, dependlng upon the ~.rl -l o-size of the cell and the type of anodes used (prebaked or Soderberg).
Alternative embodilnents of the invention are envisioned in which the cathode (and therefore anode) inclination is - 5 uniform over the width of the cell, or in which the slope is variable, in a variety of formsO These may include, for example, unequal but constant inclinations in the same direc-tions in the two halves of a transverse cross-section of a cell, and also equal but opposite inclinations in these two ln halves (e.g. a double slope, inverted V configuration). The concept of this invention can be applied to cells of both prebaked and/or Soderberg anode design. For example, while the inverted ~I configuration may be most suitable for cells whose width is covered by two prebaked anode blocks, 1~ such a configuration can also be applied to a vertical stud Soderberg anode, vented in the center. An inverted V confi-guration would be applicable to feeding from both sides, while a V configuration would be suitable for center feeding.
Similarly, a design with the slope in one direction only 2n (constant or variahle) may be preferentially applied, but the invention is not restricted to a single monolithic unvented Soderberg anode fed from one side.
The choice of cathode slope for a particular applica-tion must be made compatible with other governing parameters ~S tn ach;eve the ~esired hydrodynamic characteristics. These parameters are -the Acn gap (i.e. the ACD spacing), anode current density, (current and anode face dimensions), and bath return resistance (i.e. the bath return channel or passageway length, depth, width, and wall material). The ~n controlled hath flow ensures a sufficient supply of alumina-rich electrolyte to the anode face for the prevention of excess-ive anode effects. Additional ~enefits are the avoid-ance of excessive gas accurnulation in the ~cn, and reduction of anode and cathode overvol~ages, and minimizing of disrup-tion of flow of aluminllm product.

., , ~

$

An end view of a vertical stud Soderberg reduction cell with side feed, utilizing an inverted V cathode desi~n with center gas exit, and collector bar, is shown in Figure 1.
The cathode surface, fi, is sloped upwarclly from the cell S sides toward the center of the anode, 1, where anode gas is vented through the gas vent, 2, to a collection system (not shown). Collector bars, 4~ are illustrated as closely parallel to the cathode surface, 6, and being offset to clear aluminum drain sump 5, 1 n The angle of the cathode surface 6 from the horizontal should be large enough to induce the desired bath flow, but not so large as to causP excessive deformation of the bubbles, or bath turbulence, in the anode-cathode space, 7, and in accordance with the limitations disclosed herein-after. It has now been observed, for example, that the angle of inclination has a pronounced, and previously unreport-ed, effect upon bubble configuration and travel. When the anode surface is tilted from the horizontal, the bubbles exhibit a more pronounced tendency towards an elongated oval ~n shape, the long axis of which is perpendicular to the direc-tion of travel of the bubble up the anode slope. The leading ed~e of the buhble forrns d brow, or increased thickness, as the relative speed between buhble and liquid increases.
Thus, the increasingly thick leading edge may result in decreased bubble driving action, due to increased resistance resulting from the effects o-f interfacial distortion and friction, over sorne range of inclination. For example, it was observe~ that large bubhles on the anode surface are subject to considerable distortion and resistance at an 3n anode inclination of ahout 15 from the horizontal. Suit-able angles of inclination have been found to range from ahout 2 to ahout 15 degrees, although slightly larger or smaller angles may be acceptable for given cell conditions.
A preferred slope is from about 5 degrees to ahout 1~ degrees from the horizontal, with a more preferred slope of from ~;~5~

about 6 to 8 degrees. The most preferred cathode slope has heen found to be about 8 degrees.
The cathode surface, 6, may be covered with an electrically conducting and aluminum wetted material, such as TiB2, to facilitate the formation of a thin film of aluminum (or aluminum alloy) on the cathode surface. While a titanium diboride containing surface is to be preferred, the use of other aluminum-wetted refractory hard materials (RHM), such as ti-tanium carbide, zirconium carbide, zirconium diboride, or mix-tures thereof, is also contemplated. Suitable cathode surfaces, and coating compositions for providing such cathode surfaces, are set forth in copending Canadian Paten-t Application 432,~98 assigned to the assignee oE this application.
The aluminum drains from the sloping surface of the cathode into a side collection drain pump, 5, thus minimizing the likelihood of back reactionr which would reduce current efficiency, and expediting tappingprocedures, and making them independent of other cell operations. This thin film of draining aluminum is insensitive to induced magnetic fields t a major improvement over the use o~ conventional molten aluminum pads. Further, the controlled bath motion, as described, does no-t in-terfere with its drainage~
The gas bubbles, 8, moving up along the lower inclined edge of the anode, are vented as re~uired, such that the bubbles do not get excessively large, i.e. do .

,~;,, , not cover a siglliElcant portion oE the anode Eace, or ex-tend excessively into the anode to cathode space.
However, over-ven-tiny should be avoided in order to not seriously diminish the desired drive Eor bath flow in the narrow anode to cathode space. Venting may be clone through a slot or slots or a pattern or vent -tubes, channels, holes, etc., in the anode.

cw/)~ - 12a -~, ~

~ c~3~3 Venting holes may be made by inserting pipes just through the paste or plastic zone of a Soderberg anode, or baked into a prebaked carbon anode. It is noted that a non-uni-form distribution of vents in the anode will redistribute the bath flow in the cell to compensate for bath flow non-uniformities in the cell.
Alumina ~ay he added to the bath at hopper 19, to re-plenish the depleted alumina bath. ~f course, other hath make-Ilp materials may also be added simultaneously. The ln arrows, ln, indicate f10w of the alllmina rich ba-th through the narrow anode to cathode space, 7, induced by gas bubbles ~, while arrow 13 indicates the flow of aluminum metal, as a thin film, to drain sump 5.
Figure 2 represents an aluminunl reduction cell utiliz-1~ ing a single sloped cathode, in end view. The single slopecathode surface, ~, enables -the use oF a straight collector bar, 4, permitting maintenance of a constant distance hetween the upper cathode surface and the collector bar.
While there are some voltage reduction advantages if the 2n collector bar extends from both sides of the cell, 22, this is not necessary if a deeper aluminunl drain sump, 5, is more advantageous. The configuration of this cell permits the use of more-or-less conventional construction and installation techniques for collector har placement. The inclination of the cathode surface, 61 must be sufficient to efficiently move gas bubbles up along the shaped anode face, and to allow aluminum to drain do~n the cathode surface. As illustrated, excess anode gas buhbles, ~, which might otherwise cause excessive distortion of the metal film flowing 3n down the cathode face, ~ay be vented through the gas vents in the anode, 2, to a gas collection system 3. IJse of a mechan~cal bath cover, 1~, ensures complete control of bath atmosphere and reduces emissions to the atmosphere. Alumina is added to the bath at the interelectrode gap o(ltlet, 21, by means of hopper 19. An automatic control, 2n, may be ' ::,,;

utilized to accurately replenish the bath. Addition of the alumina, and other bakh make-up materials, in this fashion has several advantages. First, after exiting the electrolysis zone, the bath is partially depleted of dissolved alumina, 5 and thus prov1des the most ideal environment for rapid alumina dissolution. If some alumina does settle out on the bottom of the side reservoir, 2~, the absence of a metal pad or metal sump to cover the settled alumina muck will facilitate its dissolution into the flowing bath. Further, the ln return path for the bath provides extra dissolution time and acts as a trap for any undissolved alumina in order to ensure that only muck-free bath is circulated into the AC~
gap. Whereas addition of alumina to the metal drain sump side, 5, of the cell could result in the formation of muck lS beneath the metal in the sump, thus causing a sump overflow and a possible direct metal electrical short between anode and cathode, such is avoided in the present arrangement.
An end view of a flow-through single-slope cathode aluminum reduction cell is given in figure 3. The cathode 2n angle, as illustrated, is sufficient to efficiently move ; ~anode gas buhbles, ~, up along the face of the shaped anode, 1, an~;to allow aluminum to drain down the cathode surface, ~, to the aluminum drain sump, ~, as indicated by arrow 1~.
~as vents, 2, which may be required, are illustrated in the anode, and may he a slot or a pattern of ~ent holes. The anode itself may be of the Soderberg or prebake carbon variety.
The aluminum drain su~p, 5, preferably leads to a continuous tapping system, which minimizes disturbances to the cell, such as variations in pad thickness. The gas vents, 2, and 3n unoccupied bath cavities, may advantageously be connected to a gas co11ection system, ~. In the flow-through cell, as illustrated, the electrolysis bath is replenished externally, and flo~s through bath inlet 9, in the path shown by arrows ln, through the anode-to-cathode space 7, to the hath outlet 11, at which point the electrolyte has been :

. ; ~ .

partlally depleted in alumina by the electrolysis, In Figure 4, a dual sloped drained cathode, or ~I-shapetl cathode, is illustrated. In this configura-tion, replenishment is provided by point feeders 27, which are shown as being located at the higher end of the cathode slope, although replenishment may also occur between the illustrated prebaked anodes9 1. Flow of the aluminum metal, 13, is down the cathode surface h to the drain sump 5, advantageously connected to a continuous tapping system. Anode gas bubbles 3 induce ln bath material flow 1~ along the bottom ~ace of the anodes.
Figure ~ illustrates a plan view of a single cell, in-corporating a cell feeding and mi~ing zone. In Figure 5, the anode 1 is illustrated as having a slot gas vent, 2.
The cathode surface, 6, is illustrated with the aluminu~
1~ drain sump, 5, feeding a continuous rnetal tapping syste~, 14. The direction of net flow of alumillum metal is shown by arrow 13, while the direction of net flow of the cryolite he bath is illustrated by arrows 1~. After passing between anode 1, and the cathode surface, h, the partially depleted 2n hath flows from outlet 11 to a cell feeding and mixing area or zone, lh, wherein alumina is replenished, 12. After re-plenishment with alumina, the cryolite bath is recirculated in the e~lectrolytic cell in the direction indicated by arrows ln. The drawing illustrates the use of an optional 2~ magnetic bath circulation pump, 17, adjacent to bath inlet 9, although such a pump is not considered necessary under most conditions. A barrier, 1~, may be positioned in return channels 2P, to prevent bath backflow and disr~lption of circulation in designs incorporating a cell feeding and 3n: replenishment zone (lh).
Figure 6 illustrates a plan view of a multiple cell system, cl7~prising four individual electrolytic cells oper-ating in tandem. This system as illustrated incorporates a continuous aluminum tapping system, 14, providing a constant 3~ aluminum level, an~l thus resulting in a more steady heat ::L2~

-l6-balance and hath level. ~onventional discontinuous metal tapping could be used, but provision would have to be made to compensate for resulting changes in bath level. Such changes would necessarily include a bath surye tank, a lower S level in the aluminum feeding/mixing tank, and heat balance adjustment means. As shown in Figure 5, alumina and other make-up materials may be added, 12, to a cell feeding and ~ixing zone, 16, and circulated through the anode~cathode gap in the direction indicated by arrows lO. The use of a ln magnetic circulation pump, l7, is optional. The aluminum formed drains into the aluminum drain sumps, 5, and passes in the direction indicated by arrows 13 to the continous metal tapping system l4. Rack flow of depleted bath is prevented by the presence of barriers l5.
l~ An alumina feeding and dissolution system may be read-ily incorporated into either a single cell syste~ or a more economical multiple cell system. A feature of the present system which enables it to be run under a controlled atmos-phere is a sealed cell top, thus reducing reactor materials 2n problems, e~ission control problems, and heat losses, although a seale-l cell top is not required. The addition of a non-con-sumable anode surface to this design results in a self con-tained, unattended, multiple electrolysis unit in a single vessel.
2s Figure 7 represents an aluminum reduction cell, utili~-ing a~single slope cathode. As illustrated, the figure shows d continous aluminu~ tapping system 14, to remove aluminum from the cell by way of drain sump ~. This acts -to provide a c~nstant aluminum level in the cell and also results in a ~n more steady heat balance and bath level in the cell. An adjustable partial barrier, 25, is used to restrict, if required, the overall bath circulation velocity within the cell in return channel ~R. The desired restriction can be achieved ~y ~eans of harrier plates composed of a non-cor-3~ rodable insulator, such as, for instance, silicon nitride.

",J

~ecessary replenishment means, while not illustrated, ~aybe of any suitahle type as previously discussed~
As previously indicated, a dual sloping cathode cell could alternatively be utilized within the scope of the present inven-tion, having, for example, an inverted V con figuration, as shown in Figure l, or a ~/ shape, as shown in Figure ~, whereby meta1 aluminum would flow downwardly to aluminum drain sumps. However, the single sloped cathode, inclined upward towards the alumina feeding side of the ln cell, as illustrated in Figure 2, may be considered a pre-ferred embodiment, since this arrangement makes cleaning and other pot room servicing of the cell much simpler, as well as cell construction per se.
Figure ~, illustrating a plan view of an aluminum reduc-tion cell utilizing dual single slope cathodes, illustratesanother alternative form of the present invention. The flow of electrolyte is illustrated by arrows 10, indicating a circulation between the anode l, and the cathode surface, h, through a zone where replenishment, 12, takes place, and ~n through the A~D of the adjacent anode-cathode pair, sloping in the opposite direction. The center line of the cell, 26, is conveniently the demarcation point of the two oppositely sloped cathodes. To ensure smooth and continuous circulation, barriers 15 and partial barriers 25 are present to pre~ent bath backflow, and to control net bath circulation flow rate, respectively. All other elements of the drawing are as previously discussed. It is to be noted that this confi-guration represents an alternative to the previously suggested ~ shape and inverted ~I configurations, and could ~n suitably be used with prebaked anodes rather than Soderberg.
Still~another configuration, particularly suitable for prebaked anodes, has the angle of inclination for adjacent cathodes (an-l anode surfaces) in the same direction although at ~ifferent values.

., ~ .

~l2S~ S~

1~ -It is to he noted that although the figures fail to illustrate the presence of side leclges and/or crusts of frozen bath materials, the present invention applies as well to cells having such.
The pumping action of gas bubbles, directed upwards beneath a slightly inclined surface, with the liquid con-fined to a thin layer between the upper surface and a parallel lower one, was first demonstrated in a hydraulic analog experiment. This experiment was performed in room temperature 1 n water, whose kinematic viscosity closely approximates that of molten cryolite. With a surface inclination of 2.5 degrees, and a gap between upper and lower surfaces of 2.2 centimeters, bubbles were generated on a porous upper surface by pressuri-zation with air at a rate of gas evolution matched to 1~ that of a typical high cu,-rent aluminum electrolysis cell, resulting in net or average liquid velocities in the gap between the simulated anode and simulated cathode measured in the range of from 5 to lO centimeters per second. These ~ velocities were calculated to be more than sufficient to : 2n ~supply alumina at a rate required for proper cell operation at~;normal~alumina concentrations in cryolite. While in this experiment the surface inclination did not have to overcome the oppositely directed bubble forces, due to magne-ic ~influences, for example, which would exist in an operating 2~ aluminu~ electrolysis cell with a horizontal anode, the ~principle of gas-driven bath circulation was nevertheless clearly demonstrated The resulting bath circulation was found to be con-trolled by the balance between the pumping efficiency of the ~0 ~as bubbles in the ACD gap and the back pressure, or flow résistance, to bath circulation through the return channels.
Further results and data were obtained from a 1:20 scale aluminum electrolysis cell and a hydraulic analog ~odel.
The hydraulic andlog ~odel was constructed to simulate a ~as evolving anode surface (about l.ll square meters) lo-, :

~5~~SS

cated above a sloplng solid cathode surface. This arrange-ment simulated a typical "drained cathode" aluminum reduc-tion cell desi~n in which the working anode is located above a sloping~ drained TiB2 cathode surface. The water model studies were performed in room temperature water. The flow patterns and velocities observed in the water model were similar to those anticipated in a full scale cell, since the observed flows had Reynolds numbers in the turbulent regime (> ~n~0). In this regime the flow is primarily controlled l~ by the physical dimensions of the flow channels and not the fluid properties (e.g. viscosity). An air pressurized chamber with a bottom constructed from Alundum porous plates (about 2~ micron pore diameter) was used to simulate a working anode evolving gas (e.g. C~. A gas velocity of 0.1778 l~ cm/second through the porolls anode plate was used to simulate an anode current density of ~.fi~ amps/cm2 (the gas velocity WdS; corrected for differences in temperature and hydrostatic pressure). ~imulated currents up to about 1.4 amps/cm2 were~tested~in the model.
~ ~n~ ~ ~ The model design, a side elevation view of which is ; ~ illustrated by Figure 9(a), simulated one half of the cell ;shown~in Fi~ure 1. In plan view, the model simulated the cell shown in Figure 7. The end wall at the "upper" end of ~t~he cathode corresponded to a vertical plane passing through the~center slot or gas vent, 2. The figure illustrates the relationsh~ip of such parameters as ACD, ~FL, ~, h, ho, ; lower channel width, and upper channel width (hereinafter ;defined),~to the anode (l), the cathode surface (6), and ~ ~ the ~bath t29).
: ; 3n ~: ~; The following definitions are used to describe the p~arameters of the water model and commercial scale aluminum reduction cells, relative to the desired bubble flow upward under the sloping anode face:
~FL - the anode face dimension in the directior of ; 3~ hubble flow (~FL equals 122 cm in most tests) ~ 7rc~Je ~qa ~k ~ ., :`
.

~s~

-2n RFW - the anode face dimension perpendicular to the direction of the buhble flow (BFW equals 61 cm ln most model tests) ACn - the perpendicular distance between the sloping anode and cathode surfaces, corresponding to numeral 7 in Figure 2 (the AC~ was varied from 1 - to 5 cm in the model tests) Cathode slope (~1 - the angle between the upward sloping cathode surface and a horizontal plane ln ~cathode slope was varied from O to 15 degrees in the model tests) ~h - the vertical anode immersion depth from the liquid I surface to the anode face (h varies along the BFL
~ dimension of the anode according to the degree of 1~ ~ cathode slope) ~ ho - the minimum vertical anode immersion depth, i.e.
- the anode immersion depth at the higher end of the cathode (ho equals ln cm in most of the model tests) n The~desired hath flow typically passes through four different types of channels or passageways, namely:
The A~n gap between the anode and cathode, where the introduction and flow of anode gas generates the force required to maintain the desired bath 2S~ circulation in the cell.
) The bath flow exits the ACD gap into an upper channel, corresponding to reservoir 24 in Figure 2, where the flow is directed to either or both .: :
sides of the anoden ~ost of the anode gas is n ~ expelled from the bath in this channel to ~rovide a turbulent bath action ideal for dissolution of the alumina feed to the cell. This channel can le along 'he side of the cell as shown in Figures 2 and 7, or in the center of the cell as shown in 35 1 ~ ~ Figure 1. The dimensions of the upper channel are defined as:

~:~

-:
.
.: ~ :: .
.; , :~ ., :' :

~ i 4 ~3 depth = depth of bath in the channel (ho ~ the vertical component of the ACD in the water model, which at low anqles is essentially equal to ACD) width = the horizontal dimension in the same direction as the bubble flow in the ACD
~ gap - length - the horizontal dimension across the BFW
: of the anode plus the width of the return 1 n channel 3) The return channel or channels convey the bath : flow from the upper channel to a corresponding lower channel located along the anode edge where : : the bath enters the ACD gap. Examples of the return channel are designated 28 in Figures 5 through 7. In the simplest form, and the type ~ ~odeled in this study, the return channel is as :~ : shown in Figure 7, where the return channel depth or ~idth may be varied to produce a variable 2n ~ ; flow resistance (similar to that ascribed to the partial barrier 25). In the water model and : cells of a similar design,the return channel imensions are defined as:
::
depth = the depth of bath in the channel (if the 25~ width of the channel is small and a working anode forms one or both sides of : the channel, the effective channel depth is the actual depth less the ACD, since : the upward flow in the adjacent ACD gap 30:~ ; tends to stagnate a layer of similar : : :: ~: : thickness in the bottom of the return channel, This reduced return channel depth was used in the ~ater model studies).
~;
ue to the sloplng cathode, the depth of : 3~ : the return channel increases as the :: : :
': ~ :, : ~: :
`'`
,..
:
:~ : : : :

: :, . , :

: : ,. .
.:
`

bath flows from the upper channel to the lower channel.
width = the horizontal width of the channel per-pendicular to the flow in the channel.
length = the horizontal distance between the upper and lower channels (in the water model this was approximately equal to the BFL).

4) A lower channel completes the bath circulation loop by conveying the bath from the return chan-ln nel channel or channels and distributlng the bath along the anode edge where the bath enters the ; AC~ gap. The dimensions for this channel are defined similarly to those for the upper channel~
In most cell designs and the water model, the l~ lower channel has a well, or trough, in the bot-tom of the channel to collect the aluminum metal as it drains off the sloping cathode. Such a well is shown in Figures l and 9a.
In all cases, the bath flow rate, Q, for the model and n full-scale cells is defined as the total volumetric rate of bath flow entering the ACO gap from the lower channel.
To simulate an actual electrolysis cell and to evaluate the ~fect of changing specified dimensions or operating ~m -~ ~ ~ conditions, the water model was capable of being altered in ?5 ~ ~;various ways. Provision was made in the model to simulate the adjustable partial barriers (number 25 ln Figure 7), The slope of the cathode and anode in the model was varied from n;to~l5 degrees to determine the effect of cathode ~ slope~on~bath flow in the AC~ gap. The ACD gap was varied n ~:: 3n ~rom 1 ko S~cm in the water model studies. Gas flow was variable;to simulate different current densities. Fluid low in~the~odel was observed and measured, using injec-tions o~ colored dye in the ACn gap and in the return channel.
Anode shaplng, ar conformance, to the underlying sloping : ' ~: :
:: : ~

.. :

.
.

~5~

cathode has been demonstrated in both laboratory electrolysis test cells and in reports from Kaiser Aluminum and Chemical Corporation to the Department of Energy (final report on D0 Contract No. Ey-76-c-n3-l257~ "Energy Savings Through the Use of an Improved Aluminu~ Reduction Cell Cathode," 30 November 1977).
Figure 9(b) illustrates the desired flow of gas bubbles, ~, and direction of bath flow, ln, in the ACD. It is clear from this figure that too large a velocity of bath at the cathode surface, in the same direction as at the anode surface, ln may interfere with drainage of metal on the cathode. Thus, while unidirectional flow is preferable, excessive flow velocity at the cathode surface is to be avoidedn As demon-strated in the water model, the water model studies revealed that under certain design conditions three different undesired ~l~ phenomena can occur in the ACD; specifically:
o Reverse flow occurred when the hath flowing upward ____ along the bottom of the anode reversed its flow direction and flowed downward along the cathode ~ ; surface, as shown in Figure 9(c). Under severe reverse ~flow conditions, as shown in Figure 9(d), the bath forms ~ultiple small eddies in the ACD and there is no, or very little, net fresh hath entering the AC~, Persistent anode effects result fro~ the insufficient supply of fresh ~ath in an actual electrolytic cell, ~2S ~ o Air lock occurred when the removal of the anode ~ases was too slow and the ACD hecame saturated with large stagnant gas bubbles. An inadequate bath flow results under this condition. In an operating cell ~ ~ under these conditions, such gas bubhles disrupt the 3~ ~ local electrolysis reaction, increase the anode current density in the af-fected areas of the anode, and eventually lead to increased anode polari2ation voltage losses, which can lead to the onset of an anoie effect.

"
::
;; . . :
::
' ' ~ ' , .

-2~-o Excessive bubble thickness occurred under certain _ _ conditions, extending through the bath and contact-ing the cathode surface. Under this conditiorl in an operating cell, rapid loss of current efficiency will result from the rapid back reaction between `5 the C02 bubbles and molten aluminum metal on the cathode surface.
The above phenomena and the net bath flow in the ACD
gap and the return flow channel were studied as a function of:~cathode slope, ACD, flow resistance in the return channel, ~n ~and simulated anode current density. While all of the ob-s~erved flow properties are consistent with and controlled by the general hydrodynamic principles referred to previously, the1r quantitative delineation has heretofore not been esta-blishedO
l~ The~following general cell specifications were used in the water~model and subsequent examples to illustrate the typical~application of this invention to the design of an improv~ed~aluminum reduction cell with reduced energy con-sumpt~ion~.: In practice, these general cell specifications 2n: would:be chosen to fit the act.ual application and then the teachings of this invention ~ould be used to provide the crltica1~cell design specifications~ Thermal balance and return:on;investment calculations have been performed, and indicate~that "drained cathode" cells should preferably be 25;~ ~ope~rqted~at anode current densities higher than those typic-al~l~y~:employed in present industrial practice.
;: ~ ::

:: :
: ~ :

: ~:
:
`:

"
:
, ~: :

~254BSS

H O ~ I ~1 ~ O
1--l ' . ~ --I
a) ~i ~ ~ o _l a~

X H
u~ ~ o co f~ r o ~ ~ O ~
O ~t ~1 O r-l 1~ ~ O
~ C~
~ V
~ ~ ~
~ ~ a) o .,, a~
t) N O V
a~ r~ ~ (a In ~ l o ~ U~ ~ O ~9 U~ S~ ~ . O
,-1 ~J 5 Ll~ O
,1 a~
O C~
^
a) ~ ~: v ~:: ~ ~1 3 0 C~
U~
C
~ v ~ a ~ 55 rl O C~ r~ ., ~ U~
a x ~ ~
~: ~ O ^
aJ _ o~ ~
~, a~ 3 X
a~
5J S~ ~
^ O ~ Q

v ~) -- a.~ 5 O ~ C Q
~: O ~
,~ r l~ 5 0 ~ m z ~

-26~

The design parameters used in the water model studies are included in Tahle I for comparisons to real cells. The general cell design considered is that shown in Figure 1, with return channels connecting the single center upper channel with the two outer lower channels at each end of the split VSS anode (each half of the cell is similar to that shown in Figure 7).
Figures 1~, 11 and 1~ illustrate the effect of cathode slope, ACn, and the flow resistance in the return channel, 1n Rf, on the net bath flow in the ACn gap at a simulated anode current density of 0.58 amp/cm2. In these figures, the observation of reversed flow at the cathode surface is indi cated by dashes. In all cases this is associated with signi-ficantly reduced net bath flow, becoming more severe as the 1~ cathode slope is reduced.
Sirnilar results have been obtained for other simulated current densities. A cell parameter design limit diagram, constructed from this data is shown in Figure 1~. Detailed descript~ions of the use of such a diagram for the design of 2n: a cell having controlled bath flow are given in the sub-sequent examples.
The~general features of the diagram are as follows.
Flow resistance is related to return channel wldth, for particular operating conditions~ For each return channel 25 ~ width, a cross-plot, at a constant flow rate, of data from figures~;correspondîng to Figures ln, 119 and 12 leads to a relationship between cathode slope and ACn which must be s~atisfied for a given anode current density in order to ach~ieve a flow rate selected to be adequate to supply the ~ ~ 3n ~cell, as calculated by methods defined hereinafter. This reldtionship between cathode slope and ACD is represented by the curved lines in Figure 15. For each return channel width~9 therefore, these curved lines represent a condition a~ which adequate alumina supply will be achieved. These lines are limite~l by boundary conditions defined by undesir-:
` ~ :
:
.~
.
~' ~

.

~2$~iS

ahle hydrodynamic conditions, as previously described.
Thus, a region forbidden by reason of excessive bubble thickness hounds the operating region at lower ACD ciimen-sions, i.e. at the leFt of the parameter diagram (Figure 15).
Fxcessive anode (and cathode) slope, leading to excessive anode immersion depth, forms an upper boundary limit at a slope of ahout 15. A preferred operating region is shown in Figure 1~ as one somewhat above the appropriate curved line representative oF the selected return channel width, In so as to minimize the ACn while being compatible with the region marked "bubble thickness restriction". Operating within this region simultaneously assures an adequate alumina supply, through a sufficiently large Q value, and avoids an excessively large velocity of bath at the cathode surface lS (and thus interference with aluminum drainage), as well as excessive bubble thickness and attendant losses in current efficiency. The following examples will illustrate the construction and use of such a parameter diagram.

Exarnple 1_- Choice of return channels dimensions : 2n The bath flow rate, Q, (cm3/second) must be sufficient to supply the alumina required to maintain the electrolysis reaction in order to prevent anode eflects ln an operating cell. At a maximum of 10~ current etficiency, the required ~inimum flow rate, n, for a working aluminum reduction cell 2~ is given by the equation I. ~(cm3/second) = (n.nO~[total cell current (amps)]
~ wt.~, Al2Q3 where ~ wt.~. A12n3 is the diFFerence in the wt.~o A12n3 in the hath entering and exiting tne AC~ gap, or, ~ alumina.
3n The n.n~ constant is derived Frnm the Faraday equation:

Il. n.nn~ = (Molec!~lar weight of Al2n~ nn~
(Faraday/mole)(Faraday Constant)(Bath density) . ' ' .

-2~-For convenience, Q may also be expressed in terms of bath volume per secon-J per uni-t of anode areaJ' Thus, assurning a minimum anode current density oF 0O5 A/cm~, and a maximum depletion of the bath per passage through the ACD, i.e.
wt.~o A12n3 = 5, one may calculate from equation I, a minimum acceptable value of ~ - 8 x 10-4 cm3/second/cm2 anode area.
Cell performance and ther~al stability are enhanced by maintaining a uniform A12n3 concentration throughout the hath. For the cells defined above, the calculated minimum 1n bath flow rates are 202, 4noo and 1200n cm3/second for the water model cell and example cells I and II, respectively.
While the minimum Q bath flows are theoretically sufficient, in practice such low values should be exceeded to prevent operating problems (e.g. 9 excessive anode effects, high effective bath resistance and overvoltages due to excessive bubble volume in the ACD gap). Cell operating conditions will modify the bath flow to a degree (e.g., due to ledging, crusting, etc.~ and hence could lead to less than theore-t;cal bath flow rates~ For these reasons, a design factor ~n of 4 to ~ has been applied to the minimum bath flow values to give preferred bath flows of ~90, 20,000 and 60,000 cm3/second for the water model and example cells I and II, respectively. The water ~odel data demonstrated a more reliable and stable bath circulation at these preferred Q
values th~n at the minimu~ theoretical Q values, wh11e values of about 4~ o,nno, and 3n,~0n cm3/second are considered suitable.
The water model data demonstra-te that the flow resis-tance properties of the return channel are a critical compon-3n ~nt of this invention. As the return channel becomes morerestrictive (a result in attempting to maximize the anode area in the cell) the bath flow becomes increasingly sensitive to changes in the flow resistance properties of the return channel. Since there are many cell designs with equivalent 3~ ~ffective flow resistance properties in the return channel, .

~l 2~5 48~S ~

it is beneficial to construct a simplifiecl hydraulic model to provide a generali~ed design criteria for the re-turn channel. The hydraulic head loss9 hQ, due to the flow resis-tance in the return channel is given by the well-known equa-tion:
III. h~ = 2f~ L vZ
neq 9 where ff = Fanning friction factor v = velocity n L = length (approximately BFL in the water model) Deq = equivalent hydraulic diameter g = gravitational constant.
The friction may be a composite value to reflect difrer-ences in the bottom and side surfaces of the channel. For a single open channel IV. neq = 4(cross se : wetted parameter /. neq = 2 hw 2h + w ~n where w - width of return channel - : h = depth of water at any point along the channel (less the ACD correction in the water model) h = ho ~ x sin ~
~2~ ho - h at upper end of the anode (i.e., minimum anode : im~ersion depth) x - the distarlce along the return channel length, from the upper channel ~ = cathode slope.
3~ Since the o~served velocity in the return channel varies with the changing water depth, it is preferred to use volume bath flow, which is independent of changing channel dimensions.

.

.

-3n The volume bath flnw, ~, is given by velocity times the channel cross section area, or VI. Q = v h W9 hence VII. v = Q
hw The R value used in these equations represents the net effec-tive average Q for the channel as determined by timing the period required for injected, highly colored dye to be carried through the channel.
In The hydraulic head loss can be written as:
VIII. h ~ Kf Q2 Rf where Rf represents a flow resistance geometry term, depen-dent upon the physical dimensions oF the return channel, and Kf is a fluid/materials properties coefficient which is lS less dependent upon physical scale Up and can be determined in practice:
IX. Rf = (2h ~ w)L, and h3 w3 X. Kf =
2n ~ ~ 29 Since the value of h varies in an inclined return channel, Rf is actually calculated as an integral function of h over the c~annel length L, with the width w genera1ly remaining constant.
Four of the different return channel flow resistances used 2~ in the water model studies are given in Table 2.
The cell design in this invention is analogous to a pumped fluid loop where the gas hubbles in the ACD gap per-form the pumping action to drive the bath around the ACD gap ~ return channel circuit. An analogous pump efficiency diagram 3n as shown in Figure 13 is used -to describe the circulation properties of the "drained cathode" cell. Pump efficiency increases through curves 1, 2, and 3, reflecting increasing cathode slope and/or decreasing ACD. Flow resistance increases through slopes a, b, and c, reflecting increased length 3~ and/or decreased cross sectiona1 area of the return channe1.

:

., ~ .
.

The observed Flo~ rate through the ACD is determined by the intersection o-f the appropriate pump efficiency and flow resistance curves. Hence, by superimposing a vertical line on the pump diagram at the minimum flow rate required to satisfy the Al2n3 feeding requirement to the electrolysis reaction in the ACn, it is possible to portray the pumping/
flow conditions. lhe abrupt change in slope in some of the pump efficiency curves at the lower flow rates is associa-ted with reverse flow in the ACn. Table 2 illustrates the effect 1n of the flow resistance in the return channel on the ACD flow rate at a constant cathode slope and AC~.
The low flow rate (~) illustrates that a resistance factor of 2.7 x ln-l cm~4 is outside of limits of this invention under the sta-ted conditions. This is confirmed by 1~ the observed dominant reverse flow condition in the ACD.
For the same stated conditions, an Rf value oF 2.7 x 1~-2 cm~4 is within the limits of this invention, hut is too large to achieve the preFerred flow of ~90 cm3/second for the water model, resulting in some (slight) reverse flow. I-t is ~n to be noted tha-t slight reverse flow may, in some instances, be advantageous since it tends to improve, rather than hin-der, drainage of aluminum down the cathode surface.
Scale up from the water model to a commercial aluminum reduction cell presents a nurnber of design alternatives 2~ within the scope of this inven~ion. The hasic requirements that: must ~e attained in production cells are:
lj ~ufFicient hath flow, Q, entering the AC~ to support the electrolysis reaction. The minimum required : bath flow is given by equation I:
t. ~min (c~3/second~ = n.n~ [total cell current (amps)]
~ wt.~, Al23 Thus, for a lnn ~ amp cell with a ~ wt.,' Al 23 oF
n.2, the minimurn hath flow is 4nn~ cm3/second.
2) The hydraulic head For the gas-induced pumping .

, ic~ ~
0 1~ ~ ~ ~
m ~ c r~ ~ ~ v ~
a~ . ~ c c o ~ C ~
~ ~ E
11 ~ OO ~ O
c ~ ~ z z ~n a o 0 a) o 0 o c a) s ~ ~ ~
c a~ o 0 JJ ~ U~ O O U~
~ 0 a SJ ` ~ ~ I`
~ a) a 3~ ~ ~
J ~ O
~; 0 P~
~ C 11 ~7 9 ,1 O
,) ~r ~1 :4 ~ X X X X
0 ~ o O ~ ~

a~
~; ~
~ ~ ~:
o3 ~) C .
U~ j~ 0 ,_ a) IJ C :~
~ ~ ~ O o .~ ~ aJ
c)ta ~ ~ E~ ~n ~ ~ ~ U~
~_ 3 a~ ~ a) O ~; 3 ~ ~ o o a) ~1 x S Cl ~ l :~ :

, . ,~

: ` .

`: ~ :
, :- `

action (less the flow restriction losses in the ~) is equal to that for the combined h of the return channel, ~pper channel and lower channel.
For a given slope, ACn, current density and anode length, the h developed in the water model and the full-scale cell should be similar.
3) Single or multiple return channel designs with the appropriate climensions can be used such that the desired effective flow resistance, Rf, term can be achieved. The Rf term for the cell is estimated from that for the model using the equation XI. Rf (cell) = Rf (model) x ~f (model~ Q (model~ 2 ~f (cell ~ ~ (cell ~
where the change in friction factors and total cell 1~ current are taken into account, If the dimensions of the upper and lower channels are equal to or small compared to those of the return channel, the variable mass flow rate in the upper and lower channels must be included in the calculation. In 2n most anticipated cell designs, and the water model, the dimensions of the upper and lower channels are ; suFficien-t that their small hp losses can be neg-lected compared to that for the more restrictive return channel. The effect of the flow resistance ~25 ~ in the ACn is included in the reported net pumping efficiency ~ata. Hence~ the (Rf) cell term provi~es the~necessary design criteria for the return channel or somhined effect of multiple return channels `in a full-scale cell. Since a "drained cathode" cell 3~ design could encompass multiple return channels, n, the effective value for each return channel is given by XII. Rf (cell channel) - Rf (model) ~ ~mode ~ ~Q (mode~ 2 ~ ~cell ~ ~ (cell 3~ where the respective ~'s are defined above.

.~4-~ low, one may calculate d value of R~ (cell) from a summation of all (n~ return channels, as given by X I I I . 1 = ~ D
[Rf(cell )~1/2 ~ ~ [Rf(cell channel~ )]1/2 J

5 and for n equivalent channels, XIV. Rf(cell channel) = n2Rf~cell), wherein Rf (cell) may be calculated from the equation XV. Rf~cell) = Rf(model)(h.25 x ln8) ~ wt.~,!,A120 ~ 2 L~ell curre~
n In Equation XV, Rf(model) is a geometric resistance factor having a value of from about 23n x ln~3/cm4 to about n. ~ x l0-3/c1n4.
For example, if the friction factors for the water model and the example cells (as set forth in Table l5 1) are assumed to be equal, the preferred Rf term for each return channel is given by equation XII, and the width thereof may be calculated from the appropriate equations.
Cell Example I
2() RF (cell channel~ = Rf (model)C4(890)/~Onnn]2 Rf (cell channel) = 0.032 Rf (model) w (ce11 channel) - 20 cm Cell Example II
Rf (cell channel) - Rf (model)~4(~90)/60000]2 Rf (cell channel) = n~oo3~ Rf (model) w (cell channel) ~ 52 cm The return channel widths given above are based on a design factor of 5 in the preferred bath flow (Q), and a preferred Rf (model) = 6.2 x ln-3 value. If the cell oper-~n ations for Cell Example II were ~judged stable enough to re-duce :the bath flow design factor to 3, then the calculated w (cell channel) value would be reduced to 33 cm. An al-ternative method to reduce the calculated w (cell channel) value to 33 cm, withollt sacrificing the bath flow design factor, is to increase the anode immersion depth by ~ crn (eOg-, ho = ln cm). These alternatives demonstrate the usefulness of the teachings of this invention in the design-ing of a "drained cathode" cell.
Figure l4 presents the Rf (model) values (as a function of cathode slope and ACD) required to obtain the preferred Q
value of ~90 cm3/second for the model. Rf values for other Q
values can be extracted from the water model data such as shown in Figures ln, ll and 12.

1n Example 2 - Choice of ACn Cell energy efficiency is improved hy decreasing the ACn (and thus bath voltage loss) without a significant loss in current efficiency. At the larger ACD values ~e.g., 4 cm or greater) there is a strong tendency for reverse flow condition to occur within the ACn gap. If this reversal becomes exces-sive, the net Q value will be reduced too much. Reducing the ACn values promotes the desired ba-th flow through the AC~ gap and higher flow resistances in the return channel can ~e tolerated. However, at very small A~D values (about ~n l cm) a secondary effect between the bubbles and the proximity of a solid cathode surface tends to slow down the desired bath flow (e.g., a result of the increased flow resistance in the ACn gap). Figures ln, ll and l~ show that the maximum hath flow does not always occur at the minimu~ ACn as might ~5 he expected.
The measureli maxinlum buhble thickness decreased -From ahout l.n cm to about n.~ cm as the cathode slope increased fro~ ~ to ~, with a corresponding increase in net liquid flow rate. As the slope was increased beyond 5, the bubble ~n thickness was observed to remain fairly constant, and the huhhles were observed to assume the characteristic shape previously described, and to move more slowly. Changing the AC~ was determined to have little effect on the observed :

_~fi_ bubble thickness. I~ the bubble protrudes across a major portion of the ACn, the current efficiency will be serlously degraded by the back reaction of the C~2 with the alurninum metal film on the cathode and the bath electrical resistivity will be greatly increased. For these reasons and practical considerations, the pre-ferred ACn is from about 2 to about ~
cm, with a more pre~erred range of from about 2 to about 3 cm.
Example 3 - Choice of Cathode Slope . .
In all cases except when the return flow channel is 1n highly restricted (Figure ln) or the ACD is 4 cm or greater, the hath flow, n, exceeds the preferred ~90 cm3/second value at cathode slopes greater than or equal to 5 degrees. Increas-ing the cathode slope helps overcome excessive reverse flow and air locking (excessivély large stagnant gas bubbles) 1~ problems in the ACn gap. nn the other hand, too steep an angle may disrupt the flow of molten metal on the cathode surface. ~ack pressure due to restrictions in the return channel can also be offset by increasing the cathode slope.
Large cathode slopes, however, could produce excessive up-slope flow of bath in the ACD gap, which could interfere with the draininq of aluminum from the cathode surface and cause practical problems with the large variation in depth of anode irnmersion in the hath. When the changes in bubble thickness descrihed in Example 2 are also considered, the 2~ preferred cathode slope is in the range of from about 5 to abcut 11, with abou~ ~ heing the most preferred slope.

Example ~ - Cell Parameter nesign Limit Diagram The construction and use of the previously described cell parameter diagram will now be set forth in detail.
3~ The parameter restrictions used to construct this diagram were as set for-th in Tahle 3O

C ~ o .r I O
c o a r~
r~ C ~ U~
o a~
3 ~ r _ O U
V ~ ~ r~
41 ~ r; ~J O ~
w o a) ~ c .n s ~ c ~ o ~ ~ O ~ 0 3 c ~ . u~ o ,~ a) ~
o ~ ~ 41 ~ O
tn ~1 ~) 5J S~ r Q~ U~
~ a~ 0 a~ ~ ~ ~ O U ~1 C U~
X o 5 ~ ~ C) ~ a ~ a) r~ 5 O ~
O ~ C aJ O
S ~ ~ C
U~ 3 ~ ~:1 41 ~ o C O
o ~~ ~
,~ o ~
o ~ Sr~ O
u a~ c 3 ~ ~'~q ~ r~
,~ N aJ U U~ ~ U ~ '~ Q~ X
,~ O X a u~ ~ 3 ~ r t~
U~
P;
a) r~ J
a~
J- r~
O
~ U
~ o a ~ a ~
~ .~, \ V ~ X ::
~ ~ ~ e ,., u ~ e ~ e , ~ u u o e C~ ~ ~ OP X O U
07 ~ C5 x ~ ~ o ~ e ~ u-:

3 ~ 3 o U~ O
C
a) u~
~ U~
a~
J~ ~ ~ ~ Y
Q ~ t~ o ~ c ~ ~ s ~n c s~ a~ ~ u U
C
o 4~ ~ Q s :~
o a) o ~ ~J a .., ~

~n For converlience, and to help visualize the cell design, in Figure 15 the flow resistance in the return channel, R-F, has heen recorded in terms of the corresponding channel width (assumin~ a channel length of 121 cm, an anode current density of n~hB A/cm2, a bath flow rate of at least ~9n cm3/second, and a minimum channel height of ho = 10-~ cm)O
The most preferred set of cell design parameters for the water model under the conditions stated are:
cathode slope = 7.5 n ACn = 2~5 cm 2f (return channel) = ~.2 x 10~3/cm4 bath flow, ~ = 89n cm3/second The fol10wing ranges for the critical cell hydrodynamic flow design parameters are hased on: 1) measured data and observations made in the water model simulation studies; 2) the data analysis presented; and 3) practicality considera-tions for scale-up to a commercial scale cell. A set of parameter ranges is given in Table 4, for the water model, which has a ~Fl of 122 c~, and a BFW of hl cm. Parameters for a corresponding commercial scale aluminum reduction cell are given in Table 5.
The criteria of this invention may be applied to several types of cells, such as those shown in Figures 1 to 7. In particular, the cell may consist of ar, anode which traverses ~5 the width of the cell as one continuous mass. An example of this anode is shown in Figure 2 with the center vent (2) being absent and hence, the Bubble Flow Length (BFL) is approximately equal to the total anode width. This type of cell design would result in a narrow cell (about 122 cm wide 3n anode for the most preferred case).
Another type of cell would be as shown in Figure 2 where the center vent (2) is included in the design. The center vent exhausts the bubbles accumulated under the first hal~ o~ the ~node traversed by the bath flow. In this case, :.' OJ~
~ CO ~ h ~`i ~D ~) P~
e O
a) ~ [ S~ ~
h I I I O ~ aJ
U~ aJ o 1` o U~ ~l o Ul o . o w . . ~ o al .
~_1 ~: O ~1 o :~ h ~`i O
SJ ~-1 0 1~ ~1 Ul O ~
o o ~ ~ o o ~; o aJ o S~ o -~ ~ l ~1 ~ e~
h ~ a h I I I O CJ C) ~ o o aJ u~ ~ O ~ o aJ u~ ~ o e ~ ~: ~ ~ G IJ :~ h ~ o ~ ~ ~1 ~ U;
_I h ~ C
aJ a E~ ~ ~
: ,,,, o O ~ O O
r~ ~ ~ U~ Ln . aJ aJ U~
O ~) ~ ; r-l O h ~ I --I o h C o I I ~ (-~ O
h la ~i O :~ 1~ . ~ o ~: a) ~;
v a~
~ ~ :rO O 'o ~_ .
C _ ~i o o C O O
E4 ~ ~1 U ~ ~ ~i a~ ,_ r~ Q~ P a) I
al ~ o a~ ~ o E --~ O ~1 E~ --' O ~1 r~ ~ _, ~ s --~ a ~ ~ a ~
~ C~ ~ ~ ~ C~ ~ ~
P~ ~ ~ ~
:

.
, ~
.

IL2548S.~

t~
~ ~ t~
o V t ~ CO
:~
S~ C~

I) o U~ O
~ ~ ~ c~
V o r` o ~
,~1 o t~ ~
P;
o o o ~ SJ In O
U~ ~ V o I I I
s~ a t ~ ~ o E~ ~
V
aJ ~ o ~ In ~ t~ ty~ tr~
o ~
_ o t~
t, aJ ~ O
V ~
v ~ ~ O
,~ ~ o a a s . .

~f~3~t~

~1-the BFL is deFined as half of the total anode width. Vent ing of the anode gas is necessary to prevent excessive bubble volume accuniulation in the ACD gap which can increase the voltage losses and decrease the curent efficiency. In this type of cell, the total working anode width for the most pre ferred case would be about 244 cm, which is a more practical cell width.
In a commercial cell, it is desirahle to operate at the highest anode current density possible without a loss of n current efficiency (in order to minimize capital and labor costs per pound of aluminum produced). At the low ACD values used in a "drained cathode" cell, these same high current densities are advantageous to help maintain the cell heat balance. However, at high anode current densitiesg gas 1~ bubbles will accumulate faster and grow larger in the ACD
qap than at lower current densities. Hence, to counter the detrimental effect of bubble accumulation in the ACD, the pre~errei RFL (anode dimension in the direction o~ the bubble flow? Is decreased proportionally to the increase in anode 2n current density. For example, if the anode current density is increased from l.n to 2.~ amps/cm2, the preferred BFL
should he decreased from 122 to 61 cm. For an inverted V
cathode cell, the preferred cell design for the lower anode current densities is shown in Figure 1.
~5 To increase the preferred total ~orking anode width for higher anode current densities (greater than about 1.3 A/cm2), the concept of a vent to release the anode gas may be expanded to multiple vents to further increase the width of the anode.
The vent ~iay be present as a slot between the adiacent anode ~n masses or a row of suitably spaced holes through -the anode mass.
Still another approach is to use a cell design typifiecl by that shown in Figure 1. In this case the center channe`l serves ~ultiple purposes, such as gas venting, and as the 3~ upper channel to convey the bath exiting the ACD gap to the return channel or channels, In this case, the ~FL is de-.

~$~

-~2-fined as half the total working anode width, which yields a more practical size commercial cell. It is also understood that the cathode slopes shown in Figure 1 could be reversed with a central metal collection trough or well. In this latter case, the upper channels would be located along the exterior sides of the cell, as illustrated in Figure 4.
- It is also possible to combine these cell design con-cepts into a multiple component cell comprising of many cells being interconnected by the bath and/or metal flows or a single cell cavity containing multiple adjacent units described above~
For all such cell designs, the dimensions of the return channel or channels would be calculated according to the teachings given in Example 1. The initial step is to select 1~ the desired bath flow, Q, based on the scale of the cell being designed, and then use the developed relationships to determine the appropriate flow resistance term, Rf. The hydraulic head loss equations may then be applied under the conditions stated to calculate a set of preferred return 2~ channel design options. Heat balance and other cell design and operation criteria are then used to select one of the hydraulically equivalent ret~lrn channel designs. This final selection of return channel design can now be done with regard to impact on the critical bath f1OW.
2~ Fxample ~ ~ Drained Cathode Test Cell _. . ,,, .. _ . _ Test data from a laboratory scale aluminum reduction cell with a sloping Ti~2 cathode suhstantiates the use oF d water ~odel to simlJlate the hydrodynamics of a commercial aluminum reduction cell. A 61 cm x 122 cm x about ~6 cm 3n deep graphite box, Filled with approximately 600 pounds of cryolite bath, was hea-ted in an enclosed electric furnace.
nncr the bath was molten, test cathodes and anodes ~ere lowered thro~gh a top-loading port and positinned in the cryolite bath as sho~n in Fi~ure lh. Automatic A12~3 feed -~3-and anode lowering equipment permitted continuous electrolysis tests for up to ~ weeks without any interruption for anode replacement, tJnder all electrolysis conditions, the starting slope (usually horizontal) of the lower anode surface was converted after l to 5 days of electrolysis to that of the cathode surface immediately below. The unevenness of the anode face shown in Figure 16 is consistent with the bubble behavior observed in the water model. At the starting edge of the anode face in the water model (corresponding to region ln A in Figure 16) the bubbles tend to have a very slow and -- irregular movement resulting in a very non-uniform coverage of the anode surface. This would produce an uneven consump-tion of the anode carbon. As the bubbles move upward alorg the anode face, they gain speed and create a more uniform 1~ flow along the anode face. Electrolysis under this flow condition results in a more uniform consumption of the anode carhon as shown in reg10n C of Figure 16. Visual obser~ations of the flow patterns in the working test electrolysis cell are consistent with those observed in the water model.

2n Example h - Electrolysis Test nata Cell voltage data were measured as a function of cur-ren~ ~ensity, ACn value and cathode slope in the electroly-sis test cell described in Example S. ReFerence voltage data~for a conventional Inetal pad cathode were obtained by replacing the sloping Ti~2 cathode with a 3 cm deep aluminum pad in the electrolysis test cell. Figure 17 illustrates the dramatic reduction in cell voltage noise achieved by replacing an unstahle horizontal alulninun pad with a sloping TiR~ cathode in an aluminum reduction cell. In addition 3n to the reduced cell voltage noise (e~g., improved cell sta-bility~, the drained cathode test cell demonstrated -the following properties.
l) The test cell voltage trace of the drained cathode cell e~hihited an ahsence of spikes, `': .

~2~

-~4-as illustrated in Figure 17, caused either by anode effects resulting from insufficient bath flow, or anode cathode shorts, resulting from turbu-lence in the metal pad (in turn a result of bubble thickness and excessive reverse flow). At low ACD values (e.g. less than 2 cm), a conventional cell voltage control system (through the raising an-l lowering of the anode), required significant operator intervention to operate a metal pad cathode 1n test, while with the sloping Ti~2 cathode tests, little or no operator intervention was required.
2) Stable sloping cathode test runs were made at ACD
values less than 1 cm.
3) The ce11 response to changes in the current density or ACD was rapid and decisive. The new steady-state cell voltage was achieved within about 1 minute, com-pared to a 15-30 minute restabilization period re-~ quired for the rnetal pad cathode, This enables one ; ~ ; to use simpler and more reliable cell autornation 20 ~ techniques.
4)~ Bubble venting from the anode face was predominant-ly along the uppermost edge of the sloping anode face. When an anode with a horizontal face was placed in the cell with a sloping cathode, the 2~ initial venting of the anode gas bubbles was random-ly dispersed around all four edges of the anode face. ~ith continued electrolysis, the anode face ; ~ began to conform to the cathode slope and gas bubble ~ ventiny became concentrated along the uppermost 3n ~ edge of the sloping anode face. nnly random gas bubble venting from the anode face was observed ~uring tests using a rnetal pad cathode.
In addition to improved cell stability, the electroly-sis test demonstrated that a significant reduction in the 3~ anode-cathode polarization voltage could be achieved through ~ the use of a "drained cathode". Figure 18 shows the rneasured :
: :

.

.

:~$~

_4~-anode-cathode polarization voltage as a function of the TiB2 cathode slope in the -test electrolysis cell. At the preferrerl cathode slope of ~ degrees, the test data indicates an approxi-mate ~50' reduction in the anode-cathode polarization voltage, ~ or an approximate n.~2 volt saving in the cell voltage in -- addition to that achieved by reducing the ACD. To date, this aivanta~e of a "drained cathode" cell has not been reported in the literature.
The ohserved reduction in concentration polarization as 1~ a function of increasing bath flow rate (see Figure 1~, and also Figure 1~ as an example of the flow rate as a function of cathode slope) is consistent with electrochemical theory in stirred or flowing systems. In the laminar region, the concentration polarization voltage is proportional to the 1~ square root of bath velocity. The rneasured voltages in Figure 1~ do not decrease with increasing cathode slope as rapidl~ as predicted for laminar flow. This indicate~ the bath conditions in the ACn are either in the transition region or in the turbulent region where the ci~ange in concen-2n~ tration polarization voltage is less sensitive to bath flowrate. The water rnodel data confirm this hypothesis.
The electrolysis test data for a sloping Ti82 cathode cell compared to a horizont.al metal pad cathode cell have demonstrated i~proved cell stahility (e.g., improved current efficiency and reduced anode effects), at reduced AC~ values (i.e., to reduce cell voltage), improved cell response to changes in AC~ and current (e.g., ability to use s-i~pler and nlore reliable cell control autolnation), more localized anode gas ~enting (e.g~, simpler fume control system), and reduced 3n anod~-cathode polarization voltages. The asymptotic portion o~ the anode-cathode polarization curve at cathode slopes greater than ~ degrees in ~igure 1~ is consistent with the pre~erred cathode slope range indicated in the water model studies.
3~ It is understood that the roregolng description of the present invention is susceptible to various modifications, .

:~2~ 51S''~

-4~-changes, and adaptations by those skilled in the art, and that the same are intended to be considered to be within the scope of the present invention, set forth by the appended claims.

: ~:

Claims (26)

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:

1. A drained cathode cell, for the electrolytic reduc-tion of alumina to aluminum in a cryolite-based bath containing alumina, comprising:
a shell having inner surfaces lined with refractory and carbon materials to define a cathode cavity;
a cathode having at least an upper surface containing an aluminum wettable refractory hard material;
at least two adjacent anodes depending into said cathode cavity and each having a lower surface spaced from said upper surface of said cathode by an anode-to-cathode distance of about 1 cm to about 5 cm defining an anode-cathode-displacement (ACD) and a flow path between said anodes and said cathode, said upper cathode surface and said lower anode surfaces being sloped from the horizontal by from about 2° to about 15° thereby providing a lower end and an upper end of said flow path between said anodes and said cathode;
means in said cavity for providing a minimum bath flow Q about at least one substantially horizontal bath circu-lation loop to ensure a supply of alumina for the electrolysis reaction wherein Q(cell)(cm3/sec)= by providing at least one return flow channel to complete said loop, said loop comprising:
(i.) said flow path between said anodes and said cathode;
(ii) at least one lower channel in fluid communi-cation with lower end of said flow path between said anodes and said cathode;
(iii) at least one upper channel in fluid communication with said upper end of flow path between said anodes and said cathode;
(iv) at least one return flow channel in fluid communi-cation with said upper channel and said lower channel;
said at least one return flow channel, located between adjacent anodes, having dimensions h, w and L wherein h is the bath depth in cm at any given point, w is the width of the return channel in cm, and L is the length of the return channel in cm and wherein h, w and L are determined by the relationship Rf(cell channel) = and wherein the maximum value of Rf(cell channel) is determined by the relationships and Rf(cell) =
wherein Rf is a flow resistance geometry term, Rf(cell) is said term as applied to the cell overall, Rf(cell channel) is said term as applied to each cell channel, and wherein (X) is a geometric resistance factor having a value in a range of from 0.2 X 10 3/cm to 230 X 10-3/cm4.

2. The drained cathode cell according to claim 1, wherein said ACD is in a range of from about 1.5 cm to about 4.0 cm.

3. The drained cathode cell according to claim 1, wherein said ACD is in a range of from about 2.0 cm to about 3.0 cm.

4. The drained cathode cell according to claim 1, wherein said ACD is about 2.5 cm.

5. The drained cathode cell according to claim 1, wherein said upper cathode surface and said lower anode surface are sloped from the horizontal by from about 5° to about 10°.

6. The drained cathode cell according to claim 1, wherein said upper cathode surface and said lower anode surface are sloped from the horizontal by from about 6° to about 8°.

7. The drained cathode cell according to claim 1, wherein said upper cathode surface and said lower anode surface are sloped from the horizontal by about 8°.

8. The drained cathode cell according to claim 1, wherein L(cm) is within a range of from about 15 to about 300.

9. The drained cathode cell according to claim 1, wherein L(cm) is within a range of from about 30 to about 250.

10. The drained cathode cell according to claim 1, wherein L(cm) is within a range of from about 60 to about 200.

11. The, drained cathode cell according to claim 1, where L(cm) is about 122.

12. The drained cathode cell according to claim 1, wherein X(10-3/cm4) is in a range of from about 1.0 to about 20.

13. The drained cathode cell according to claim 1, wherein X(10-3/cm4) is in a range of from about 5.0 to 10.

14. The drained cathode cell according to claim 1, wherein X(10-3/cm4) is about 6.2.

15. The drained cathode cell according to claim 1, wherein Q(cm3/sec/cm2) is in a range of from about 1.1 x 10-3 to about 6.0 x 10-2.

16. The drained cathode cell according to claim 1, wherein said cell comprises a plurality of return flow channels=n, and wherein Rf(cell channel)=n2Rf(cell).

17. The drained cathode cell according to claim 1, wherein said slope and said ACD are so related as to limit the thickness of bubbles generated in said electrolytic reaction to less than 50% of said ACD.

18. The drained cathode cell according to claim 1, wherein a return flow channel is located between an anode and an adjacent inner wall of said cathode cavity.

19. The drained cathode cell according to claim 1, wherein said aluminum wettable refractory hard material is selected from a group comprising titanium diboride, titanium carbide, zirconium diboride, zirconium carbide, and mixtures thereof.

20. A process for the production of aluminum which comprises subjecting alumina to electrolytic reduction in a cryolite based bath in a drained cathode cell as claimed in claim 1.

21. A process according to claim 20 wherein the depletion of alumina concentration in a supply of bath from the lower channel.
entering said flow path between said anodes and said cathode and emerging into said upper channel is in a range of from 0.2%
to about 0.5%.

22. A process according to claim 20, wherein a net total bath flow rate per unit anode area in said flow path between said anodes and said cathode is in a range of from about 1.2 x 10-1 to about 8 x 10-4 cm3/sec./cm2.

23. A process according to claim 20, wherein said cell has an anode current density (A/cm2) in a range of from about 0.5 to about 3Ø

24. A process according to claim 20, wherein said cell has an anode current density (A/cm2) in a range of from about 0.6 to about 2Ø

25. A process according to claim 20, wherein said cell has an anode current density (A/cm2) in a range of from about 0.7 to about 1.5.

26. A process according to claim 20, wherein said cell has an anode current density (A/cm2) in a range of from about 0.8 to about 1.2