Method for early detection of certain abnormal operating conditions in hall-heroult electrolysis cells

Abstract

A method of operating an electrolytic cell suitable for the Hall-Heroult electrolysis process, said method comprising: - monitoring a parameter V representative of the cell voltage, and determining its variation with time; identifying an abnormal increase of said parameter V; moving the anode assemblies downwards by a certain value D1; monitoring the parameter V and memorizing its base value Vb, calculating the expected value Vexp(1) of said parameter V due to the downwards movement of the anode assemblies; monitoring further said parameter V after a certain time interval At after the end of said downwards movement, to obtain Vm(1); calculating the rate of change R(1) of said parameter V; comparing R(1) to preset minimum Rmin and maximum values Rmax; if the condition Rmin < R(1) < Rmax is not fulfilled then continuing monitoring the parameter V and calculating the expected value Yexp(2) of said parameter V in case of any anode beam movement; monitoring further said parameter V and obtaining Vm(2), calculating R(2) and comparing it to Rmin and Rmax, and as long as the condition Rmin < R(i) < Rmax is not fulfilled, continuing monitoring the parameter V and calculating the expected value Yexp(i), and repeating the previous step; if the condition Rmin < R(i) < Rmax is fulfilled, identifying this situation as tapping and moving the anode assemblies downwards such as to catch up with the increasing anode - cathode distance due to the decrease of in metal level.

Description

1 Method for early detection of certain abnormal operating conditions in Hall-Héroult electrolysis cells Technical field of the invention The invention relates to the field of fused salt electrolysis and more precisely to the Hall- Héroult process for making aluminium by fused salt electrolysis. In particular, the invention relates to the control of the pot voltage. More precisely, the invention is about a method allowing to detect a rapid increase of the pot voltage, to determine the cause of this rapid increase, and to initiate appropriate corrective actions, in particular to avoid a possible open circuit. Such rapid increase in pot voltage can arise accidentally during tapping of liquid metal, and should be identified early in order to avoid potentially dangerous situations. Prior art The Hall-Héroult process is the only continuous industrial process for producing metallic aluminium from aluminium oxide. Aluminium oxide (Al2O3) is dissolved in molten cryolite (Na3AIF6), and the resulting mixture (typically at a temperature comprised between 940 °C and 970 °C) acts as a liquid electrolyte in an electrolytic cell. This liquid electrolyte is also called "liquid bath". An electrolytic cell (also called "pot") used for the Hall-Héroult process typically comprises a steel shell (so-called pot shell), a lining (comprising refractory bricks protecting said steel shell against heat, and cathode blocks usually made from graphite, anthracite or a mixture of both), and a plurality of anodes (usually made from carbon) that plunge into the liquid electrolyte contained in the volume defined by the cathode bottom and a side lining made from carbonaceous material. Anodes and cathodes are connected to external busbars. Typical Hall-Héroult cells comprise tens of individual anode assemblies, each anode assembly comprising one (or two) anodes connected to an anode rod, said anode rod being mounted on the anode busbar (so-called "anode beam"). An electrical current is passed through the cell (typically at a voltage between 3.5 V and 5 V) which electrochemically reduces the aluminium oxide, split in the electrolyte into aluminium and oxygen ions, into aluminium at the cathode and oxygen at the anode; said oxygen then reacts with the carbon of the anode to form carbon dioxide. The resulting metallic aluminium is not miscible with the liquid electrolyte, has a higher density than the liquid electrolyte and will thus accumulate as a liquid metal pad on the cathode surface below the electrolyte from where it needs to be removed from time to time, usually by suction into a crucible. This operation of metal suction from the electrolytic cell into a crucible is called metal tapping. It is usually carried out at fixed intervals, for instance 24, 32 or 48 hours. 2 The electrical currents in most modern electrolytic cells using the Hall-Héroult process exceed 200 kA and can reach 400 kA, 450 kA or even more. Most newly installed pots operate at a current comprised between about 350 kA and 600 kA, and more often in the order of 400 kA to 500 kA. The production of aluminium of an electrolytic cell is proportional to the electrical current (amperage) according to Faraday's law and therefore the daily cell production of aluminium exceeds 1 500 kg for 200 kA, can reach up to 4 600 kg for 600 kA cells, and more often is comprised between 3 000 to 3 800 kg for cells operating at 400 kA to 500 kA. A metal pad of 15 to 25 cm height is usually kept in the cell for ensuring thermal balance and stability. This corresponds, depending on the size and amperage of the electrolytic cell, to a metal inventory of approximately 5 to 20 tons, corresponding to 5 to 8 days of metal production. During the metal tapping operation which lasts only a few minutes, the height of the metal pad decreases rapidly by 2 to 4 cm. This leads to an increase in the interpolar distance between the bottom of the anodes surface and the top of metal pad, where electrolytic decomposition takes place (this parameter is called the anode-cathode distance, "ACD"). It is usually comprised between 2 and 5 cm. Any increase in the ACD will lead to an increase in cell voltage, and this is observed immediately when metal tapping starts. The ACD must be maintained during the metal tapping operation; this is usually done by lowering the anode frame, which is a means for lowering all the anodes at the same time by the same height. If the ACD is not maintained, there is a risk of liquid bath height dropping significantly to take the place of tapped aluminium. This is because bath height, from the top of the metal surface to the top of the bath between anodes, is normally comprised between 15 cm to 25 cm for a normal ACD. But if the ACD increases suddenly, the liquid bath located in the central channel, in the side channels and in the channels between anodes will flow below the anodes into the increased ACD trying to fill up a much larger surface. The liquid bath between in the channels will drop suddenly and the total bath height will drop by a factor 5 to 10 times (depending how much the anodes are filling up the cell in a given technology) the increase of ACD: This is called the piston effect. The risk is great of not having enough liquid bath to maintain the bottom of the anodes in the bath to ensure continuity of the electrical current flow. In that case an open circuit may occur in the pot, resulting in a high energy DC electric arc; this is a very serious incident that may injure operators and cause damage to equipment (see the article "Potline open circuit protection" by D. Duval et al., Light Metals 2012, p. 913 – 916). In order to avoid open circuits most modern cell control systems automatically adjust the ACD during metal tapping by lowering the anode frame to which all anodes are mechanically connected; microprocessor-controlled mechanical systems are used to this end. An automatic adjustment routine (so-called "tapping logic") is triggered by the operator prior to starting the tapping operation. According to the state of the art this triggering of the tapping logic is a manual operation. In fact, in the absence of lowering of the anode beam, the pot voltage behaviour during metal taping is very similar to that occurring during a so-called anode effect. Anode effects, which are known to person skilled in the art, are caused by depletion of alumina in the liquid bath and lead to the formation of an insulating layer of gas underneath the anodes, thereby dramatically increasing the electrical resistance (and the voltage) of the cell. This resistance can be lowered by lowering the anode beam (this is described for instance in US 3,539,461, DE 29 44 518, US 6 609 119), but this is often not sufficient for terminating anode effects. Most modern cell control systems do include an automatic adjustment routine for anode effects. WO 99 / 57336 describes an automatic method for predicting anode effects from statistical analysis of cell voltage data. However, existing methods cannot identify whether the increase in cell voltage is caused by anode effects or by metal tapping. For this reason, a specific tapping logic must be triggered manually by the operator. In the framework of this tapping logic, the pot control system, informed by the tapping operator that metal tapping takes place, will monitor the fast increase of pot voltage due to syphoning of metal (the latter leading to a decrease of metal level and an increase of ACD). It is configured to correct the measured voltage increase (or the calculated pot resistance) by lowering the anodic plane with down orders to the anode beams every few seconds. To start this rapid adjustment of the anodic plane, the pot control system needs to have received the information from the operator that the tapping operation is going to take place, since without this information the pot control system will continue to control the pot in normal operation and adjust the anodic plane only every few minutes. It sometimes happens that the operator forgets to activate the tapping logic of the pot to be tapped, or activates by mistake the tapping logic of another pot (most frequently of the adjacent pot), and the normal operation pot voltage adjustment will not be able to correct the increase of pot voltage due to fast increase of ACD. Activating the tapping logic of the wrong pot is a mistake that can happen in particular when a given input / output terminal in the potroom is used for controlling more than one pot; for example, two adjacent pots can have a common terminal located in the passageway next to the potline. As a consequence of such an input error the anodes might come out of the bath, leading to arcing and possible open circuit. 4 The goal of the present invention is to avoid this situation and to provide a pot control system that automatically detects metal tapping taking place, and takes appropriate measures to adjust the ACD. Objects of the invention The inventors have discovered that the difference between a real anode effect and the effect of metal tapping on the cell voltage can be recognized after a down movement of the anode beam. More precisely, in case of a real anode effect, lowering of the anode beam (i.e. decreasing the ACD) leads to massive voltage fluctuations, while in case of metal tapping, lowering of the anode beam leads to a quiet increase of cell voltage since the metal level keeps decreasing. According to the invention, the problem is solved by automatic activation of a specific method when the pot control system detects a fast increase of pot voltage (pot resistance) and is able to assign this fast increase to a tapping operation that has started without appropriate activation of the tapping control process, the purpose of which is to compensate for the rapid increase in ACD during tapping. Said specific method is based on a comparison of the normal voltage response after beam movements and the actual voltage response during the Near Anode Effect Quenching logic as explained further on. The object of the present invention is therefore a method of operating an electrolytic cell suitable for the Hall-Héroult electrolysis process, said electrolytic cell comprising _ a cathode forming the bottom of said electrolytic cell, a lateral lining defining together with the cathode a volume containing the liquid electrolyte and the liquid metal resulting from the Hall-Héroult electrolysis process, _ a plurality of anode assemblies suspended on said cathode, each anode assembly comprising a carbon anode and a metallic anode rod, - a superstructure comprising anode holders for holding said anode assemblies, said method comprising: (i) periodically or permanently monitoring a parameter V representative of the cell voltage, and determining its variation with time (so-called time slope), and optionally the variation of the time slope with time; (ii) identifying an abnormal increase of said parameter V; (iii) moving the anode assemblies downwards by a certain value D1, after identification of said abnormal increase; (iv) memorizing its substantially constant base value Vb, (v) calculating the expected value <semantics>Vexp(1)<annotation encoding="application / x-tex">V_{exp(1)}< / annotation>< / semantics> of said parameter V due to the downwards movement of the anode assemblies by D1, (vi) monitoring further said parameter V after a certain time interval ∆t after the end of said downwards movement of the anode assemblies, to obtain <semantics>Vm(1)<annotation encoding="application / x-tex">Vm(1)< / annotation>< / semantics>, (vii) calculating the rate of change R(1) of said parameter V, (viii) compare <semantics>R(1)<annotation encoding="application / x-tex">R(1)< / annotation>< / semantics> to preset minimum <semantics>Rmin<annotation encoding="application / x-tex">R_{min}< / annotation>< / semantics> and maximum values <semantics>Rmax<annotation encoding="application / x-tex">R_{max}< / annotation>< / semantics>, (ix) if the condition <semantics>Rmin<R(1)<Rmax<annotation encoding="application / x-tex">R_{min} < R(1) < R_{max}< / annotation>< / semantics> is not fulfilled then continuing monitoring said parameter V and calculating the expected value <semantics>Vexp(2)<annotation encoding="application / x-tex">V_{exp(2)}< / annotation>< / semantics> of said parameter V in case of any anode beam movement by D2; (x) monitoring further said parameter V and obtaining Vm(2) as in step (v), calculating R(2) as in step (vi) and comparing it to <semantics>Rmin<annotation encoding="application / x-tex">R_{min}< / annotation>< / semantics> and <semantics>Rmax<annotation encoding="application / x-tex">R_{max}< / annotation>< / semantics> as in step (vii), and as long as the condition <semantics>Rmin<R(i)<Rmax<annotation encoding="application / x-tex">R_{min} < R(i) < R_{max}< / annotation>< / semantics> is not fulfilled, continuing monitoring the parameter V and calculating the expected value <semantics>Vexp(i)<annotation encoding="application / x-tex">V_{exp(i)}< / annotation>< / semantics> as in step (viii), and repeating step (ix); (xi) if the condition <semantics>Rmin<annotation encoding="application / x-tex">R_{min}< / annotation>< / semantics> < <semantics>R(i)<annotation encoding="application / x-tex">R(i)< / annotation>< / semantics> < <semantics>Rmax<annotation encoding="application / x-tex">R_{max}< / annotation>< / semantics> is fulfilled, identifying this situation as tapping and moving the anode assemblies downwards in one or more steps such as to catch up with the increasing anode – cathode distance due to the decrease of metal level. Vm(1) and Vm(2) are measured values. Said parameter representing the cell voltage is typically the cell voltage or the cell resistance. If the condition <semantics>Rmin<R(i)<Rmax<annotation encoding="application / x-tex">R_{min} < R(i) < R_{max}< / annotation>< / semantics> is fulfilled in step (xi) the abnormal increase of parameter V is very likely to be due to metal tapping: detecting this event is the purpose of the process. In step (ii) said increase is considered abnormal when said parameter P, and / or its time slope, exceeds a given threshold value. In steps (x) and (xi), the anode beam may be moved up and then down by a distance D2, D3, Di to test the reaction of the cell voltage to these movements. Said value D1 is typically in the range of 5 mm to 15 mm while D2 (and any further movement Di at step (ix)) are typically in the range of 1 mm to 10 mm. During tapping the rate of change R is typically comprised between 4 mV / s and 30 mV / s. In an embodiment of this method, said calculating of said rate of change R(1), R(2), R(i) comprises calculating an expected value of said parameter <semantics>Vexpected=Vb+K×D<annotation encoding="application / x-tex">V_{\text{expected}} = V_b + K \times D< / annotation>< / semantics> wherein <semantics>Vb<annotation encoding="application / x-tex">V_b< / annotation>< / semantics> is said base value measured, D is the vertical distance by which the anode assemblies have moved, and K is a constant factor. 6 In an embodiment of the method in step (x) after each of said movements of anodes the parameter V is measured from which resistance is calculated and compared to target resistance. Said method can further comprise a step (xii) in which an acoustical and or visual alarm ("Auto Tapping Start") is emitted to alert the potroom operator and / or potroom supervisor, that the automatic tapping has started. Brief description of figures Figures 1 and 2 represent prior art, while figures 3 and 4 illustrate physical phenomena, and figure 5 shows an embodiment of the invention. Reference numbers with four digits refer to process steps. Figures 1 and 2 show schematic views of a typical Hall-Héroult cell of a known type. Figure 2 schematically shows a vertical cross section of a cell; it illustrates the current flow from the anode through the electrolyte to the cathode. Figure 1 show as perspective view. Figure 3 schematically shows a vertical cross section of a cell, illustrating in four steps the decrease of the liquid metal level during tapping. Figure 4 shows a time-dependent plot of the resistance (curve 1) and anode beam position (curve 2) of a typical Hall-Héroult cell operating at am amperage of 450 kA during a tapping operation without manual activation of metal tapping logic using a cell control method according to prior art. Figure 5 shows a flow chart of a typical embodiment of the automatic detection of tapping method according to the invention. Figure 6 shows typical variation of the cell resistance (curve B) during tapping when the tapping logic is activated; curve A represents the actual target cell resistance at the onset of the tapping, the so called "control resistance setpoint (CRSP)". Figure 7 shows an example of the application of automatic detection of tapping according to the invention. Detailed description of the invention As the Hall-Héroult process is known as such, it is sufficient to explain (as done here in particular in relation with Figures 1 and 2), that the pot 1 typically comprises a potshell 2 usually made from steel, and a lining comprising a carbonaceous cathode formed from individual, parallel cathode blocks 3 and a side lining 4, said lining defining a volume for the liquid electrolyte 5 and the liquid metal pad 6 produced by the electrolysis. Current is fed into the anode busbar 16 (also called "anode beam"), flows from the anode beam 16 to the anode rod 7 (said anode rod 7 being connected to the carbon anode 8 by means of anode yoke 20) and to the carbon anode 8 in contact with the liquid electrolyte 5 where the electrolytic reaction takes place, crosses the liquid metal pad 6 resulting from the electrolysis process and eventually will be collected at the cathode block 3. As cathode blocks are symmetric and have collector bar 9 ends coming out on each side, in side by side arrangements of electrolytic cells half of the current collected by the collector bars 9 of the cathode blocks 3 will flow directly to the downstream longitudinal part 10 of the cathode busbar system, while the other half flows to the upstream longitudinal part 11. Flexible connectors 12 are used to connect the ends of the cathode collector bars 9 to the cathode busbar 10, 11. Conductor 13 carries the current collected at the upstream part 11 of the cathodic busbar system to the anode risers 18 of the downstream pot. The current collected at the downstream part 10 of the cathodic busbar system directly feeds the anode risers 18. A Hall-Héroult cell further comprises an alumina feeding system (usually located inside the carcass of the superstructure 19) through which alumina powder is dumped from time to time into the cell volume. The air space above the cell is closed by a set of covers or hoods 15 that can be removed for maintenance and anode change; the anode rods 7 are adjustably fixed to the anode beam 16 using anode clamps 17 that allow to adjust the anode heights in order to keep the inter-electrode spacing constant as the anode is consumed. Figure 3 shows the decrease of the liquid metal level during tapping. The anode-cathode distance A is defined as the difference in height between the lower surface d of the anodes 8 and the upper surface b of the liquid metal pad 6. The liquid metal pad 6 has a thickness M defined by the upper level a of the cathodes 6 and the upper level b of the liquid metal pad. Figure 3(a) shows a starting level at which tapping occurs. Upon tapping liquid metal is removed and the thickness M of the liquid metal pad 6 decreases. As a consequence its upper level b decreases (figure 3(b)). This leads to an increase in the anode-cathode distance A, which leads to an increase in the cell voltage. If tapping (i.e. the removal of liquid metal from the liquid metal pad 6) continues, the metal level b will further decrease, and at a certain point the lower surface d of the anode 8 will no longer be in contact with the liquid electrolyte 5 (figure 3(c)). This is an open-circuit situation, which must be avoided because it is extremely dangerous and destructive. It can be avoided by lowering the anodes 8 (figure 3(d)) so that the lower surface d of the anodes 8 is below the upper surface c of the liquid electrolyte 5. During the pot operation, several cases of voltage increase may occur, which are called here "abnormal conditions". They all lead to an increase in cell voltage. The method according to the invention is capable of detecting and identifying these abnormal conditions at an early stage. This is essential in order to be able to distinguish metal tapping (which obviously is an abnormal condition in the sense of the above definition) from other abnormal conditions, in order to select an appropriate reaction or remedy. The first abnormal condition is the so-called anode effect which is known to a person skilled in the art. It occurs when the alumina content in the electrolyte reaches very low values (e.g., below 2 %) and during which another, unwanted electrochemical reaction takes place with the electrolyte being decomposed and various fluorinated gases being formed (e.g., perfluorocarbons, also called PFCs). These gases form an insulating layer underneath the anodes, which leads to a voltage increase. If this voltage increase reaches a given threshold (generally 7 to 8 volts), modern pot control systems according to prior art declare the pot on anode effect and automatically activate the anode effect quenching logic in which several down and up movements are given to the anode beam, along with additional alumina dumps in order to terminate the anode effect. In modern pot control technologies, suitable alumina feeding automatic control limits the occurrence of the anode effect on each cell to only a few times per month, thus avoiding unnecessary energy consumption and emission of perfluorocarbons that are greenhouse gases. The second abnormal condition is a so-called "near anode effect" or "near anode effect" event". This event has been described in the publication "DUBAL's Experience of Low Voltage PCF Emissions" by Abdalla Al Zarouni and Ali Al Zarouni, published in the Proceedings of the 10th Australasian Aluminium Smelting Technology Conference (Launceston, Australia, 9-14 October 2011, Paper 4a3). In the framework of the present invention, the inventors have found that an early detection of an approaching anode effect is possible, and this is the second abnormal condition that can be identified and treated by the method according to the present invention. More precisely, before an anode effect occurs, the cell condition can indicate the imminent occurrence of the anode effect and allows the pot control system to carry out a specific treatment or remedies in order to prevent the anode effect to take place. This specific condition prior to the anode effect is called here the "near anode effect", and methods for its detection and appropriate treatment to avoid the coming anode effect are called here the "near anode effect logic". The detection of the near anode effect is based on four possible criteria: the voltage change from a reference base voltage or the slope of the voltage curve with time or the variation of said slope with time or a combination of these three. If any one of these criteria is met, the near anode effect logic is triggered. A third case of abnormal voltage increase is represented by the tapping operation without manual activation of metal tapping logic. In pot control system according to prior art, the metal tapping logic must be activated manually by the operator prior to starting metal tapping. Within the tapping logic, the anode beam height will be adjusted automatically by the pot control program. If the tapping operation is carried out without manual activation of the metal tapping logic, there will be no appropriate adjustment of the anode beam height. While tapping as such is a normal operation, even if it leads to a perturbation of the various chemical and thermal equilibria of the cell, tapping without proper adjustment of the anode beam (i.e. tapping without manual activation of the tapping logic) leads to an abnormal condition in the sense of the present invention. This is illustrated on figure 4 which shows the cell voltage (curve A) and anode beam height (curve B) as a function of time, measured in a cell with a pot process control system according to prior art. Metal tapping starts around t = 150 seconds; the operator had omitted to activate the metal tapping logic. During tapping the height of the liquid metal in the pot decreases, which leads to an increase of ACD, and this in turn leads to an increase in cell voltage (curve A). The pot process control system detects an increase in cell voltage but interprets it as a near anode effect (point 1): the pot process control system responds by lowering the anode beam (using a microprocessor-controlled mechanical system, which is known as such), as shown on curve B, which is a well-known means to decrease the cell voltage. After having lowered the anode beam, the pot process control systems observes at t = 250 seconds a further increase in cell voltage; this increase is the consequence of the ongoing metal tapping operation, but is interpreted by the pot process control system in a different way. Various adjustments of the anode beam height are tried out (points 2,3,4,5,6) but do not turn out to be an efficient remedy against the overall tendency of the cell voltage to increase. This eventually leads to a critically unstable cell condition (point 7): the pot process control system has lost control over the pot; the cell is heading to an open circuit as the metal level continues to decrease due to tapping but the down beam movements do not follow because the tapping has not been activated. As this loss of control will eventually result in the visible emission of fumes, the operator may regain control over the pot by a manual intervention if it is very fast. However, this so-called near anode effect is a potentially dangerous situation that should be avoided; the operator intervention is often too late for quickly decreasing anode immersion in the bath and an imminent open circuit. In state of the existing (prior art) pot process control systems the metal tapping logic needs to be triggered manually, and this may lead to human errors (such as: failure to trigger the metal tapping logic, or input errors in the case in which the input terminal serves two or more pots at the same time) or may be done too late. As metal tapping is a discontinuous intervention carried out by operators, the present invention does not aim at replacing their manual triggering of the metal tapping logic in the pot process control system, but aims at automatically detecting their failure to said manual triggering. According to the invention, the pot process control system continuously monitors the pot voltage and potline amperage, and continuously calculates the pot resistance, the variation of pot resistance with time (slope) and the variation of slope with time. Using these parameters the pot process control system is able to identify the cause of a rapid resistance and voltage increase: an anode effect, or a near anode effect, or, according to the invention, the start of a metal tapping operation without prior manual activation of the tapping logic. In fact, the pot resistance and voltage behavior during metal tapping without manual activation of the metal tapping control logic is very similar to an increase of resistance and voltage before an anode effect and during a near anode effect detection. (It should be borne in mind that on a running pot the cell resistance is not directly measured but calculated from the measured voltage). According to the invention, the difference between a real anode effect or a near anode effect and metal tapping operation without metal tapping logic activation can be detected after an anode beam down movement; this down movement will be triggered automatically by the pot process control system as a reaction to the observed resistance and voltage increase. Indeed, the pot process control system will initially interpret the increase of resistance and voltage as an anode effect or a near anode effect and initiate the corresponding treatment which includes down movement of the anode beam. However, the reaction of the pot to this down movement will be different if the pot is being tapped (i.e., liquid metal is removed from the pot by suction) without prior activation of the metal tapping logic. In that case, the pot voltage will continue to increase because the down movement of the anode beam ordered by the anode effect or near anode effect treatment logic is not sufficient to compensate the increase of ACD due to metal tapping. The inventors have observed that a real anode effect gives massive voltage fluctuations with less ACD, but in case of tapping the voltage is quiet and it continues increasing gradually. As explained above, said prior activation of the metal tapping logic is usually to be done manually, and it may accidentally be omitted; inappropriate remedies (as those that will be tried by the process control systems according to prior art) can then lead to arcing and open circuit condition of the cell. From these differences in pot behavior, the pot control system according to the invention is capable to detect that the pot is neither having a near anode effect nor an anode effect, and will trigger automatically the metal tapping logic. Figure 5 shows a flow diagram of a method called here the "Automatic Tapping Detection" Method" which represents a typical embodiment of the present invention. The process is started (step 1000). At step 1010 the cell voltage is measured. If a voltage rise is detected and this rise is above a preset value, then this condition is identified as a 11 special condition called "near anode effect" (step 1020); otherwise the process is stopped at step 1016; it may then be restarted at step 1000. Let us note Vt (or threshold voltage) the preset value of voltage, and Vx (or excess voltage) the observed voltage rise. At step 1030 the system checks whether the quenching of the near-anode-effect is enabled (it could be disabled automatically or manually in special pre-defined cases); if this is not the case then the process is stopped (step 1036). If the quenching of the near anode effect is enabled, the system memorizes the original anode beam position (step 1040) and moves down the anodes (step 1050) by a preset value as shown in the beam position in figure 4. After a preset waiting time (step 1060), which confirms the stability of the voltage, the voltage is recorded as <semantics>Vb<annotation encoding="application / x-tex">V_b< / annotation>< / semantics> (or base voltage) (step 1070). If at this point the voltage is changing extremely rapidly and surpasses the limit value defining an anode effect, an anode effect is declared (step 1100) and the automatic tapping detection is ended (step 1106). In this case a specific anode effect quenching logic is activated (not shown on the figure and not part of the present invention); this includes a certain number of anode beam movements according to procedures known to a person skilled in the art which may eventually end with the beam position at original level (step 1110); the automatic tapping detection is stopped (step 1116), but the voltage continues to be monitored. If there is no anode effect and the anode is at rest (anode not moving, step 1120), the expected cell voltage is calculated from any anode beam movement (up or down), shown in figure 4. This calculation is typically done according to a formula of the type (step 1130) [Image disponible dans le document PDF, Image available in the PDF document] wherein Vb is the base voltage measured at step 1070, D1 is the above vertical distance by which the anode has moved down or up, and K (in millivolts / mm) is a constant factor that converts said vertical distance D1 (typically expressed in millimetres) into a voltage change (typically expressed in millivolts). Instead of distance D1; time length (s) of anode beam movement can be used and in that case, the movement constant would be expressed as mV / s. There is no point in calculating cell voltages as long as the anode is moving (step 1120). At step 1140 the rate of change of the cell voltage, which might be due to the liquid metal tapping (knowing that any decrease in the liquid metal level due to tapping will lead to an increase in ACD and thus to an increase in cell voltage), is calculated according to the formula [Image disponible dans le document PDF, Image available in the PDF document] wherein R is the rate of change of cell voltage in volts per second, for this first move along D1, <semantics>Vexpected(1)<annotation encoding="application / x-tex">V_{\text{expected(1)}}< / annotation>< / semantics> is as above defined and Vm(1) is the measured voltage after this first move. At step 1150 this calculated value for <semantics>R(1)<annotation encoding="application / x-tex">R(1)< / annotation>< / semantics> is then compared to preset minimum (<semantics>Rmin<annotation encoding="application / x-tex">R_{min}< / annotation>< / semantics>) and maximum (<semantics>Rmax<annotation encoding="application / x-tex">R_{max}< / annotation>< / semantics>) values. If at step 1150 the condition <semantics>Rmin<R(1)<Rmax<annotation encoding="application / x-tex">R_{min} < R(1) < R_{max}< / annotation>< / semantics> is not fulfilled (step 1156) then the process is repeated as shown on the figure; <semantics>Vexpected(2)<annotation encoding="application / x-tex">V_{\text{expected(2)}}< / annotation>< / semantics> and R(2) are calculated, the same way as <semantics>Vexpected(1)<annotation encoding="application / x-tex">V_{\text{expected(1)}}< / annotation>< / semantics> and R(1). As long as R(i) does not meet the above condition <semantics>Rmin<R(i)<Rmax<annotation encoding="application / x-tex">R_{min} < R(i) < R_{max}< / annotation>< / semantics> the process is repeated. Indeed, if R < <semantics>Rmin<annotation encoding="application / x-tex">R_{min}< / annotation>< / semantics> there could be a spurious voltage fluctuation, and if R > <semantics>Rmax<annotation encoding="application / x-tex">R_{max}< / annotation>< / semantics> this extremely rapid voltage increase could be indicative for an anode effect (in this case the specific anode effect quenching logic is activated, as explained above). As soon as the condition <semantics>Rmin<R(i)<Rmax<annotation encoding="application / x-tex">R_{min} < R(i) < R_{max}< / annotation>< / semantics> is fulfilled, then a timer is triggered at step 1160 for a preset period of time (for example four seconds). After the preset time, which confirms the value and thus avoids false triggering, the tapping logic is started at step 1200 and the automatic tapping detection logic terminates (step 1210). The goal of the invention has been achieved: the initial voltage decrease has been correctly analyzed as being due to tapping, and the tapping logic is activated and will decrease the cell voltage in rapid steps (by decreasing the ACD) to catch up with the increasing ACD due to the tapping of the metal. Figure 6 shows the cell resistance as a function of time before, during, and after tapping for the process when the tapping logic is activated manually before tapping begins. The target during tapping is a constant cell voltage (curve A) which was active at the moment when tapping begins; this is the so called "control resistance setpoint" (CRSP). The purpose of the resistance control during tapping is to keep the cell resistance within the tapping deadband and not allow the resistance increase much above the upper control limit. The decrease in metal level during tapping increases the ACD, which tends to increase the cell voltage beyond the tapping deadband rapidly; this increase is counterbalanced by an incremental downwards movement of the anode beam compensating the decrease in metal level: Each voltage decrease is the consequence of a downwards movement of the anode beam. As the system reacts to the voltage increase by lowering the anode beam in small increments, a series of spikes is observed (curve B). The upper control limit resistance represents the resistance that should not be exceeded during tapping, in order to avoid to operate the cell under conditions too far from a steady state, and in order to avoid eventually the risk of open circuit. At the end of tapping operation, the cell resistance may fall below the target value since the last down move of the anodes may be very close to the moment at which the suction of the metal is stopped. The end of tapping operation is given to the control system manually by the operator and at this moment the normal resistance control resumes to keep the resistance within the resistance control deadband. Example The test has been carried out in an industrial electrolytic cell representing the so-called DX+ Ultra technology developed by the applicant. In this cell, the method according to the invention was implemented. In the case shown, the Automatic Tapping Detection Logic was enabled. The cell voltage and anode beam movement are shown in figure 7. According to the evolution of the cell voltage curve, the following sequence can be observed (point numbers refer to figure 5): At 0 s tapping was started (point 1000) and the voltage started to increase from the base of 4.2 V. At 69 s, the voltage rise was identified as a near anode effect at the voltage of 4.565 V (point 1020). At this point the anode beam was moved down to 122 mm according to near anode effect logic (point 1050). This is the base anode beam position. As a consequence the cell voltage decreased to 4.314 V. This is the base voltage, Vb. The voltage continues to increase in spite of the beam position staying at the base of 122 mm. Expected voltage in millivolts (mV) is calculated as Vexpected = 4314 mV+60 mV / mm*(122 - 122) mm = 4314 mV (point 1130). 60 is the rate of change of cell voltage (in mV / mm) with the beam position movement (in mm). The anode beam is raised by a small amount at 82 s. The cell voltage continues to increase. The lowering of the anode beam at 92 s leads to a temporary decrease in cell voltage. If the voltage increase was really due to an anode effect the voltage would not increase again immediately after. Tapping rate between 69 and 110 seconds is evaluated as <semantics>R=(4807−4314) / (110<annotation encoding="application / x-tex">R = (4807 - 4314) / (110< / annotation>< / semantics> - 69) = 12 mV / s (point 1140). The preset minimum suction rate is 5 mV / s and maximum preset tapping rate for the automatic tapping is 30 mV / s. The condition Rmin < R < Rmax is fulfilled and the Automatic Tapping Detection activates the tapping logic and the alarm "Auto Tapping Start" is announced. A rapid sequence of anode beam down movements follow, decreasing the voltage at suction rate. This continues after the tapping is stopped until the voltage decreases to the normal control deadband, where the initial voltage was, too.

Claims

<pat:ClaimStatement>CLAIMS< / pat:ClaimStatement> <pat:Claims com:id="claims"> <pat:Claim com:id="CLM-00001"> <pat:ClaimNumber>1< / pat:ClaimNumber> <pat:ClaimText>1. A method of operating an electrolytic cell suitable for a Hall-Héroult electrolysis process, said electrolytic cell comprising a cathode forming a bottom of said electrolytic cell, - a lateral lining defining together with the cathode a volume containing a liquid electrolyte and a liquid metal resulting from the Hall-Héroult electrolysis process, - a plurality of anode assemblies (3a, 5a, 3b, 5b, 3n, 5n) suspended on said cathode, each anode assembly comprising a carbon anode and a metallic anode rod, - a superstructure (1; 101; 201) comprising anode holders for holding said anode assemblies, said method comprising: (i) periodically or permanently monitoring a parameter V representative of a cell voltage, and determining a time slope of the parameter V; (ii) identifying an abnormal increase of said parameter V; (iii) moving the anode assemblies downwards by a first distance D1, after identification of said abnormal increase; (iv) monitoring the parameter V representative of the cell voltage and memorizing a substantially constant base value Vb for the parameter V, (v) calculating a first expected value <semantics>Vexp(1)<annotation encoding="application / x-tex">V_{exp(1)}< / annotation>< / semantics> of said parameter V due to the moving of the anode assemblies downwards by the distance D1, (vi) monitoring further said parameter V after a certain time interval ∆t after said moving the anode assemblies downwards by the vertical distance D1, to obtain Vm(1), (vii) calculating a first rate of change R(1) of said parameter V, (viii) comparing the first rate of change <semantics>R(1)<annotation encoding="application / x-tex">R(1)< / annotation>< / semantics> to a preset minimum value <semantics>Rmin<annotation encoding="application / x-tex">R_{min}< / annotation>< / semantics> and a preset maximum value Rmax, (ix) if a condition <semantics>Rmin<R(1)<Rmax<annotation encoding="application / x-tex">R_{min} < R(1) < R_{max}< / annotation>< / semantics> is not fulfilled, then continuing monitoring the parameter V and calculating a second expected value <semantics>Vexp(2)<annotation encoding="application / x-tex">V_{exp(2)}< / annotation>< / semantics> of said parameter V in case of any anode beam movement by a second distance D2; (x) monitoring further said parameter V and obtaining Vm(2) as in step (vi), calculating a second rate of change R(2) as in step (vii) and comparing the second rate of change <semantics>R(2)<annotation encoding="application / x-tex">R(2)< / annotation>< / semantics> to the preset minimum value <semantics>Rmin<annotation encoding="application / x-tex">R_{min}< / annotation>< / semantics> and the preset maximum value <semantics>Rmax<annotation encoding="application / x-tex">R_{max}< / annotation>< / semantics> as in step (viii), and if a condition <semantics>Rmin<R(2)<Rmax<annotation encoding="application / x-tex">R_{min} < R(2) < R_{max}< / annotation>< / semantics> is not fulfilled, continuing monitoring the parameter V and calculating a further expected value <semantics>Vexp(i)<annotation encoding="application / x-tex">V_{exp(i)}< / annotation>< / semantics> as in step (ix); (xi) calculating a further rate of change R(i) as in step (vii), comparing the further rate of change R(i) to the preset minimum value Rmin and the preset maximum value <semantics>Rmax<annotation encoding="application / x-tex">R_{max}< / annotation>< / semantics> as in step (viii), and if a condition <semantics>Rmin<R(i)<Rmax<annotation encoding="application / x-tex">R_{min} < R(i) < R_{max}< / annotation>< / semantics> is fulfilled, identifying this situation as tapping and moving the anode assemblies downwards in one or more steps such as to catch up with an increasing anode – cathode distance due to a decrease of a metal level. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00002"> <pat:ClaimNumber>2< / pat:ClaimNumber> <pat:ClaimText>2. A method according to claim 1, wherein at step (ii) said increase is considered abnormal when at least one of said parameter V and the time slope of the parameter V exceeds a given threshold value. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00003"> <pat:ClaimNumber>3< / pat:ClaimNumber> <pat:ClaimText>3. A method according to claim 1 or 2, wherein said first distance D1 is in the range of 5 mm to 15 mm. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00004"> <pat:ClaimNumber>4< / pat:ClaimNumber> <pat:ClaimText>4. A method according to any one of claims 1 to 3, wherein said second distance D2 is comprised between 1 mm and 10 mm. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00005"> <pat:ClaimNumber>5< / pat:ClaimNumber> <pat:ClaimText>5. A method according to any one of claims 1 to 4, wherein the first rate of change R(1), the second rate of change R(2), and the further rate of change R(i) are comprised between 4 mV / s and 30 mV / s. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00006"> <pat:ClaimNumber>6< / pat:ClaimNumber> <pat:ClaimText>6. A method according to any one of claims 1 to 5, wherein said calculating the first rate of change R(1), the second rate of change R(2), and the further rate of change R(i)comprises calculating an expected value of said parameter <semantics>Vexpected=Vb+K×D<annotation encoding="application / x-tex">V_{\text{expected}} = V_b + K \times D< / annotation>< / semantics>, wherein Vb is said constant base value measured, D is a given distance by which the anode assemblies have moved, and K is a constant factor. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00007"> <pat:ClaimNumber>7< / pat:ClaimNumber> <pat:ClaimText>7. A method according to any one of claims 1 to 6, wherein said parameter V representing the cell voltage is the cell voltage or a cell resistance. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00008"> <pat:ClaimNumber>8< / pat:ClaimNumber> <pat:ClaimText>8. A method according to any one of claims 1 to 7, wherein in step (x) after each of said movements of anodes, the parameter V is measured from which resistance is calculated and compared to a target resistance. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00009"> <pat:ClaimNumber>9< / pat:ClaimNumber> <pat:ClaimText>9. A method according to any one of claims 1 to 8, further comprising a step (xii) in which an acoustical and or visual alarm is emitted to alert at least one of a potroom operator and a potroom supervisor, that an automatic tapping has started. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00010"> <pat:ClaimNumber>10< / pat:ClaimNumber> <pat:ClaimText>10. A method according to any one of claims 1 to 9, further comprising determining a variation of the time slope of the parameter V with time. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00011"> <pat:ClaimNumber>11< / pat:ClaimNumber> <pat:ClaimText>11. A method according to claim 10, further comprising analyzing the variation of the time slope of the parameter V with time. < / pat:ClaimText> < / pat:Claim> <pat:Claim com:id="CLM-00012"> <pat:ClaimNumber>12< / pat:ClaimNumber> <pat:ClaimText>12. A method according to claim 10 or 11, further comprising detecting an increase in the variation of the time slope of the parameter V with time. < / pat:ClaimText> < / pat:Claim> < / pat:Claims>