EP0201438B1 - Process for accurately regulating the low alumina content of an igneous electrolysis cell for aluminium production - Google Patents

Abstract

A process is disclosed for accurately maintaining a low alumina content of between 1 and 4.5% in a cell for the production of aluminum by electrolysis in the Hall-Heroult process. According to the invention, a control parameter P=-1/D(dR1/dt), is determined, wherein D is the variation in the alumina content of the electrolytic bath in % weight per hour, R1 is the internal resistance of the cell, and t is the time. A series of operations is then carried out in a repeated cycle, starting with the cell being fed alumina at a nominal rate which is substantially equal to the quantity consumed by electrolysis. At periodic intervals, an over-supply of alumina is added in order to enrich the bath, and the over-supply is continued for a preset time during which dR1dt is negative. The feed rate is then reduced to less than the nominal feed rate, during which time dR1dt passes through zero to become positive and the regulation parameter P, the value of which tends to rise, is measured often. The successive values of P are compared with a required preset value Po. As soon as P equals Po, the feed rate is returned to the nominal feed rate and a new cycle is recommenced.

Description

Subject of the invention

The present invention relates to a process for the precise regulation of a low alumina content in an igneous electrolysis tank for the production of aluminum according to the Hall-Héroult process, this regulation also having the aim of maintaining the Faraday yield at a level high, at least equal to 94%.

Exhibits prior art

In recent years, the operation of aluminum production tanks has been progressively automated, both to improve the energy balance and regularity of operation, as well as to limit human intervention and improve the capture efficiency of fluorinated effluents.

One of the essential factors, to ensure the regularity of operation of an aluminum production tank by electrolysis of alumina dissolved in the molten cryolite, is the rate of introduction of the alumina into the bath. An alumina defect causes the appearance of the "anodic effect", or "packaging" which results in a sudden increase in the voltage across the terminals of the tank, which can go from 4 to 30 or 40 volts, and which affects the whole series.

An excess of alumina creates a risk of soiling the bottom of the tank by alumina deposits which can transform into hard plates electrically insulating part of the cathode. This induces strong local horizontal currents in the liquid aluminum sheet which, by interaction with the magnetic fields, smooth the liquid aluminum sheet and cause instability of the bread-metal interface, and various other drawbacks well known in the art. skilled in the art.

This defect is particularly troublesome when it is sought to lower the operating temperature of the tank, which is very favorable for its lifespan and for the Faraday yield by adopting so-called very "acid" baths (with high AIF _{3 content} ). and / or comprising various additives, such as chlorides, lithium or magnesium salts. However, these baths have a significantly reduced capacity and speed for dissolving alumina, and their use implies that the alumina content is very precisely regulated, at relatively low concentrations and between two close extreme limits.

Although it is possible to directly measure the alumina content of baths by analysis of electrolysis samples, it has been chosen for many years to carry out an indirect evaluation of the alumina contents by following an electrical parameter reflecting the alumina concentration of said electrolyte.

This parameter is generally the variation of the internal resistance, or, more exactly, of the internal pseudo-resistance which is equal to:

Eo being an image of the counter-electromotive force of the tank, the value of which is generally assumed to be 1.65 volts, U the voltage across the terminals of the tank and J the intensity passing through it.

By calibration, one can draw a variation curve R _i as a function of the alumina content, and by measuring R _i at a frequency determined according to currently well known methods, one can estimate at any time the concentration, symbolized by [A1 ₂ 0 ₃₁ .

Attempts have been made for many years to introduce alumina into the bath with a certain regularity so as to maintain its relatively stable concentration around a predetermined value.

The processes for automatic supply of alumina, more or less strictly controlled by its concentration in the bath, have been described in particular in the following patents: French patent FR-1,457,746 to REYNOLDS, without which the variation in internal resistance of the tank is used as a parameter reflecting the concentration of alumina, the introduction of which into the bath is carried out by a distributor combined with a means of piercing in the frozen electrolysis crust; French patent FR 1 506 463 from V.A.W. which is based on the measurement of the time which elapses between the stopping of the supply of alumina and the appearance of the anodic effect; US patent US-3,400,062 to ALCOA, which implements a "pilot anode" to obtain an early detection of the tendency to runaway and to control the rate of introduction of alumina, which is distributed from '' a hopper fitted with a frozen electrolyte crust piercing device.

More recently, regulatory methods based on the control of the alumina content have been described in particular in Japanese patent application JA-5,228,417 / 77 to Showa Denko, and in US patent US-4126 525. from Mitsubishi.

In the first of these patents, the alumina concentration is fixed in the range of 2 to 8%. We measure the AV variation, as a function of time t, of the voltage at the terminals of each tank, we compare it with a predetermined value and we modify the rate of supply of alumina to bring the A V / t to the set value . The disadvantage of this process is that its sensitivity varies with the alumina content, which is precisely minimal in the interval used, from 3 to 5% of A1 ₂ 0 ₃ (table p. 8).

In the second of these patents, the alumina content is also fixed in the range of 2 to 8% and, preferably, 4 to 6%. The tank is fed for a predetermined time t with an amount of alumina higher than its theoretical consumption, until a predetermined alumina concentration is obtained (for example up to 7%), then the supply is switched to a rate equal to the theoretical consumption for a predetermined time t ₂ , then the feeding is stopped until the appearance of the first symptoms of anode effect ("packaging"), and the feeding cycle is resumed at a rate which is greater than the theoretical consumption. In this process, the alumina concentration varies, during the cycle, from 4.9 to 8% (example 1) or from 4.0 to 7% (example 2).

Finally, in our French patent FR-2 487 386 (Aluminum Péchiney), to which correspond the patents EP-44 794 and US-4 431 491, we have described a process for precise regulation of the alumina content, between 1 and 3, 5% by weight, process according to which the rate of introduction of aluine is modulated according to variations in the internal resistance of the tank during predetermined time intervals, by alternating cycles of equal duration of introduction of alumina at a slower rate and at a faster rate than the rate corresponding to the consumption of the tank.

This process, known as "slope calculation", is based on successive measurements of the internal resistance R;, at equal time intervals, on the evaluation of the slope dRi / dt of variation of R _i as a function of time, and the comparison of R _i on the one hand and of dRi / dt on the other hand, to set values, and on the modification of the rate of introduction of alumina, so as to reduce dRi / dt and R _i at setpoint values.

The search for the optimal operating mode, that is to say the search for the operating parameters of the electrolytic cells giving the best cost price, or the biggest profit margin for a given investment, has always been a permanent concern. for those skilled in the art.

In particular, research into the influence of various market parameters on current efficiency - also called Faraday efficiency - has been the subject of numerous publications, the most significant of which are cited in the work by K. Grjotheim and co- authors, entitled "Aluminum Electrolysis", the second most recent edition of which was published in 1982 by Aluminum Verlag (Dûsseldorf, FRG).

In this work, page 339, figure 9.11, it can be seen that all the cited authors agree to confirm that an increase in bath temperature is harmful for the current yield. On the other hand, the phase diagram of the cryolite-alumina system, represented on page 29, FIG. 2.3, of the same work, shows that the liquidus temperature of the bath is higher the higher the alumina content of this bath. low.

It would therefore be logical for the Faraday yield to be higher the higher the alumina content of the bath. This is, in fact, what many authors have thought to observe, on industrial tanks, as shown in Figure 9.20, page 356 of the aforementioned work.

Statement of the problem

At present, the economic and technical conditions of the production of aluminum by the Hal-Héroult process require that the operator constantly seeks to optimize the various factors which determine the cost price of the metal; among these factors, the Faraday yield is one of the most important, and also one of the most fragile, since small disturbances can significantly degrade it. It is therefore desirable to research all the factors which influence Faraday's yield, so as to maintain it at a high and stable value, at the current price of aluminum at LME ($ 1,200 per tonne at the end of April 1985) 0.1 Faraday point on a production of 500,000 tonnes / year corresponds to a gain of nearly $ 380,000 / year.

Subject of the invention

The object of the invention is an improvement of the process for the precise regulation of a low alumina content in the electrolysis tank, making it possible to significantly improve the Faraday yield. By observing the implementation of the method of regulation by slope calculation, object of our aforementioned patent, on our modern electrolysis cells operating at 175,000 or 280,000 amperes, with a so-called "acid" bath composition, this is that is to say a bath comprising more than 8% by weight of aluminum fluoride AIF ₃ in excess relative to the neutral cryolite of formula Na _3AIF ,, we have found, contrary to the general opinion of the specialists that we indicated previously , that, despite the increase in temperature of the electrolysis bath, the current yield increased rapidly when the alumina content of the bath decreased.

We discovered that this phenomenon had a previously unsuspected amplitude since, by reducing the alumina content from 2.5% by weight in the bath, to 1.5% by weight, this drop of 1 point in the alumina content allowed to increase the current yield from 94% to 95.7%, an increase of 1.7% in the electrolysis yield. However, due to the increase in the operating temperature of the electrolysis bath which had gone from 946 ° C to 951 ° C at the same time, a decrease of 1% in this yield should logically have been observed.

However, this increase in yield is accompanied by an increase in the electrolysis voltage, the more rapid the lower the alumina content.

Energy consumption per tonne of aluminum produced can depend on the yield Faraday, F, and the voltage across a tank, U, in the form:

On the other hand, at a fixed electrolysis intensity J, the production of a cell is proportional. to its yield F, that is to say that the incidence of "fixed" costs (depreciation, financial costs, and a large part of labor and maintenance costs) is all the lower that the Faraday yield is better.

Given the discovery that we have made of the very strong impact of the alumina content of the bath on the Faraday yield, it is understandable that there is every interest in adjusting the alumina content of the bath to a low value, but sufficient however , to prevent the energy cost due to the increase in the voltage at the terminals of the tank from exceeding the gains hoped for by improving Faraday efficiency.

In general, and for normal economic conditions, this optimal alumina content is very close to the minimum content below which appears the "anode effect", also called "packaging" or "polarization" , which results in a very sharp rise in the voltage across the cell and in the temperature of the electrolysis bath, and in the release in significant quantities of fluorinated products from the decomposition of the electrolysis bath.

To avoid such a phenomenon, disastrous both for energy performance and for the environment, while getting as close as possible to the alumina content giving the best economic performance, we understand that it is extremely important to have a process making it possible to control and very finely regulate the alumina content of the electrolysis bath in the range of low contents, for example between 1% and 3% and preferably between 1% and 2.5%.

The object of the invention is therefore to provide such a process for regulating the alumina content of the bath in the low content range, by the use of a synthetic parameter P which can be calculated simply from conventional measurements made on a tank. electrolysis, namely: the voltage at the terminals of each cell, the intensity traversing the file of cells and the rate of supply of alumina (in kg / hour for example).

More precisely, the subject of the invention is a process for the precise regulation of a low alumina content, of between 1 and 3%, in a tank for the production of aluminum by electrolysis according to the Hall-Héroult process. making it possible to obtain a Faraday yield at least equal to 94% by implementing alternating supply periods of the alumina electrolysis tank at a nominal rate CN, at a reduced rate C- and at a higher rate C + at nominal rate, this process consisting first of all in determining a regulation parameter P = -1 / D. (dRi / dt) expressed in microohms per second and per% by weight per hour, Ri being the internal resistance of the cell and t time, D being the derivative of the content of the alumina electrolysis bath expressed in% by weight per hour, according to the relation

CN being the nominal rate of supply of alumina and C- the rate of undernourishment, counted in kg of alumina per unit of time, Q (AI ₂ 0 ₃ ) being the quantity of alumina consumed by the tank in the same time unit and

• a) on alimente la cuve à la cadence nominale CN (telle que la quantité d'alumine introduite dans le bain soit sensiblement égale à la quantité consommée par l'électrolyse);
• b) on déclenche périodiquement une suralimentation en alumine à une cadence C+ supérieure à la cadence nominale CN, de façon à enrichir le bain en alumine, et pendant une durée t+ prédéterminée. Pendant cette période dRi/dt est négative;
• c) on passe en sous-alimentation, c'est-à-dire à une cadence C- inférieure à CN. La pente dRi/dt s'annule puis devient positive. On mesure, de façon fréquente, le paramètre de régulation P, dont la valeur tend à augmenter;
• d) on compare les valeurs successives de P à une valeur de consigne Po prédéterminée. Dès que P = Po on repasse en cadence d'alimentation nominale CN et on recommence un nouveau cycle en (a).

Q (bl) being the quantity of liquid electrolysis bath contained in the tank, the process then consisting in repetitively carrying out the following operations:

• a) the tank is supplied at the nominal rate CN (such that the quantity of alumina introduced into the bath is substantially equal to the quantity consumed by the electrolysis);
• b) a periodic supercharging of alumina is initiated at a rate C + greater than the nominal rate CN, so as to enrich the bath with alumina, and for a predetermined period t +. During this period dRi / dt is negative;
• c) we go into undernourishment, that is to say at a rate C- less than CN. The slope dRi / dt is canceled out then becomes positive. The regulation parameter P is frequently measured, the value of which tends to increase;
• d) the successive values of P are compared with a predetermined setpoint Po. As soon as P = Po, we return to the nominal supply rate CN and start a new cycle in (a).

Definition of process parameters and steps

The parameter P is evaluated from the internal pseudo-resistance of the tank, R;, itself defined by:

where U is the voltage across the tank (in volts)

Eo is a standard value, in volts, of the dynamic counter-electromotive force of the tank, generally between 1.5 and 2.0 volts, and most often of the order of 1.65 to 1.75 volts

J is the electrolysis intensity expressed in kiloamperes (= 103 amperes) R _i is then expressed in microohms;

(its derivative dR; / dt is generally expressed in microohms per second).

(P étant exprimé en microohms par seconde et par % poids par heure).More precisely, if D is the alumina content drift of the electrolysis bath, expressed in weight percent per hour, P is expressed by the formula:

(P being expressed in microohms per second and per% weight per hour).

• a) pour cela, on prévoit d'alimenter régulièrement la cuve à une cadence, dite cadence nominale CN, telle que la quantité d'alumine introduite dans le bain soit sensiblement égale à la quantité d'alumine consommée par électrolyse. Pendant ces périodes à cadence nominale, on peut ajuter sans difficulté la distance interpolaire en se basant sur la valeur de la pseudo-résistance qui est alors mesurée pour une teneur en alumine du bain sensiblement constante;
• b) puis, partant de cette situation stable, on déclenche, à des moments choisis, une suralimentation c'est-à-dire une alimentation en alumine à une cadence C+ supérieure à la cadence nominale CN. Dans ces conditions s'enrichit progressivement en alumine, à un rythme d'autant plus rapide que la cadence de suralimentation est plus grande.

The regulation of the tank according to the invention consists in staying as long as possible in an alumina content zone, not necessarily known in a precise manner, but such that P is as close as possible to a Po value that one s 'is fixed beforehand.

• a) for this, provision is made to regularly supply the tank at a rate, called the nominal rate CN, such that the amount of alumina introduced into the bath is substantially equal to the amount of alumina consumed by electrolysis. During these periods at nominal rate, the interpolar distance can be adjusted without difficulty based on the value of the pseudo-resistance which is then measured for a substantially constant alumina content in the bath;
• b) then, starting from this stable situation, a supercharging is triggered at selected times, that is to say a supply of alumina at a rate C + greater than the nominal rate CN. Under these conditions is gradually enriched with alumina, at a rate all the more rapid as the rate of overeating is greater.

The duration t + of this supercharging is fixed so as to cause an enrichment in alumina of the electrolysis bath. Note that it is not necessary to measure or calculate the exact value of this enrichment We can, during this period of overeating, follow the evolution of the pseudo-resistance of the tank (= dR; / dt). however, a risk that all the alumina introduced does not dissolve instantaneously in the bath, this risk being all the greater the faster the rate of supercharging.

The values of P that are measured are then tainted with a non-zero risk of error, and, in general, they will only be used to detect serious feeding anomalies.

c) After this overfeed of fixed duration t +, we then go into underfeeding, that is to say that the tank is fed at a rate C- slower than the nominal rate corresponding to the consumption of alumina by electrolysis. It is generally noted, at the beginning of undernourishment, that the slope (dR; / dt), normally negative during overeating, takes a certain time to cancel out and then to take larger and larger positive values.

This initial period, which generally lasts only a few minutes, corresponds to the end of dissolution of the excess alumina introduced during the overeating period and not immediately assimilated by the bath.

We can easily neutralize this initial period when the alumina content of the bath does not change in accordance with the rate of introduction of this alumina: indeed, by frequent measurements, we have found that the duration of this initial period was l '' order of 2 to 3 times the duration separating the launching of the period of under-timing and the moment when the slope of dR; / dt calculated passed by the value 0.

Another method is to add a period of a few minutes at nominal rate after overeating before going on to undereating.

After this initial period, the alumina content of the bath decreases all the faster the slower the feed rate, and, in parallel, the measured slope dR; / dt increases.

The alumina content drift, D, counted in% weight per hour, is then proportional to:

where C- is the rate of undernourishment counted in kg of A1 ₂ 0 ₃ introduced per second and C _N is the nominal rate of feed (counted in the same units). Any other coherent system of units can of course be used, for example the inverse of the time separating the introduction of 2 consecutive doses of alumina.

. Q (Al ₂ O ₃ ) is the weight of alumina consumed per unit of time, by electrolysis.

. Q (bl) is the weight of liquid bath, capable of dissolving alumina, contained in the crucible of the tank (for information, if the weight of liquid bath is measured in kg, of the order of 30 J where J is the electrolysis intensity counted in kA): note that the time constant for the melting or solidification of the bath at the slope is very large (generally of the order of several hours), this quantity only varies very slowly over time.

For example, for a 280 kA tank containing 8000 kg of liquid bath and consuming 170 kg of alumina per hour, and a rate C- = 0.7 CN, it comes to D = -0.64% per hour.

• d) Au fur et à mesure que la sous-alimentation se prolonge, la valeur de P, initialement inférieure à la valeur visée Po, augmente et finit par atteindre cette valeur visée. Cet évènement se produit au bout d'un temps t de sous-cadencement que l'on ne peut prévoir de façon certaine et généralement différent du temps t⁺ de surcadencement.
• e) On repasse alors en cadence nominale CN, c'est-à-dire à une cadence d'alimentation égale au rythme de consommation de l'alumine par électrolyse, pendant un temps t_N au bout duquel le cycle de mesure et ajustement redémarre en (a).

We can then reliably and frequently measure the synthetic parameter P - - 1 / D. (dR _{; /} dt).

• d) As the undernourishment continues, the value of P, initially lower than the target value Po, increases and eventually reaches this target value. This event occurs after a time t of under-timing that cannot be predicted with certainty and generally different from the time t ⁺ of over-timing.
• e) We then return to the nominal rate CN, that is to say to a feed rate equal to the rate of consumption of the alumina by electrolysis, for a time t _N at the end of which the measurement and adjustment cycle restarts. in (a).

The targeted alumina content being close to the limit content triggering the appearance of a polarization of the tank, it is essential that after operation at nominal rate, the readjustment is done by preceding the search phase of the operating point (characterized by Po), during an under-timing, by a period of over-timing which allows s '' move away from this limit content before starting the search.

Of course, the regulation method according to the invention can be used only during part of the operating time of the tank, and preferably when the tank is stable.

Certain operations indeed disturb normal operation, and this is particularly the case of the operations of changing anodes and casting the metal produced.

It is obvious to those skilled in the art that it will be possible to adopt particular regulation algorithms during and after the course of these disturbing operations, until the tank has regained sufficient operating stability.

où: Po est exprimé en microohms par seconde et par pour-cent poids par heure,

• K_I est un coefficient "économique" synthétisant les conditions économiques du moment (en particulier coût de l'énergie comparé aux autres postes du prix de revient, hors alumine),
• K₂ est un coefficient "technique" synthétisant les caractéristiques technologiques et physicochimiques de la cuve (K₂ est sensiblement indépendant de K₁,

It is also obvious to a person skilled in the art that it will be possible to interpose, between the supercharging step (3) and the controlled undernourishment step (4), an additional step of feeding at nominal rate - or slight over- or under-feeding - without this appreciably disturbing the process according to the invention, that is to say it does not prevent the search for the operating point such that P = -1 / D (dR _; / dt) is close to Po. As regards the estimation of the value of Po corresponding to the operation of the tank closest to the economic optimum, it appeared to us that Po could be described by a very simplified equation:

where: Po is expressed in microohms per second and per weight percent per hour,

• K _I is an "economic" coefficient synthesizing the economic conditions of the moment (in particular cost of energy compared to the other cost items, excluding alumina),
• K ₂ is a "technical" coefficient synthesizing the technological and physicochemical characteristics of the tank (K ₂ is substantially independent of K ₁ ,

J is the operating intensity of the tank, expressed in kilo-amperes (= 103 Amperes).

Preferably, this parameter Po will be maintained between the limit values of 2/100 J and 10/100 J.

The estimation of K _I and K ₂ can be done as follows:

The economic coefficient K _I summarizes the economic conditions of the moment. It is substantially equal to the ratio of the sum of the fixed transformation costs (excluding alumina), including in particular the cost of energy and consumable carbon products, labor and depreciation, including financial costs, at cost Energy.

• A = coût de l'alumine et matières premières diverses (hors carbone)
• C = coût des matières premières carbonées
• E = coût de l'énergie (électroyse et captation)
• PR = autres coûts de production (essentiellement main d'oeuvre et frais d'entretien)
• AFF = amortissements et frais financiers

By way of illustrative and nonlimiting example, a good approximation of this coefficient K _I can be obtained by breaking down the production costs of a tonne of aluminum as follows.

• A = cost of alumina and various raw materials (excluding carbon)
• C = cost of carbon raw materials
• E = cost of energy (electroyse and capture)
• PR = other production costs (mainly labor and maintenance costs)
• AFF = depreciation and finance charges

(une estimation de K₁ à ± 20 % est largement suffisante pour s'approcher suffisamment de l'optimum économique).We then write that K _I is approximately equal to:

(an estimate of K ₁ at ± 20% is more than enough to get close enough to the economic optimum).

• A = 4000 F/tonne
• C = 1000 F/tonne
• E = 2000 F/tonne
• PR = 2000 F/tonne
• AFF = 1200 F/tonne
• Il vient: K₁ = 1000 + 2000 + 2000 + 1200/2000 = 6200/2000 = 3,10

For example, for an aluminum production cost broken down into:

• A = 4000 F / tonne
• C = 1000 F / tonne
• E = 2000 F / tonne
• PR = 2000 F / ton
• AFF = 1200 F / tonne
• It comes: K ₁ = 1000 + 2000 + 2000 + 1200/2000 = 6200/2000 = 3.10

The "technical" coefficient K ₂ summarizes the technological and physicochemical characteristics of the tank and can be evaluated as follows:

où: U est la tension aux bornes de la cuve, comptée en volts, généralement comprise entre 3,8 et 5,5 volts pour des cuves correctement conduites par l'homme de l'art,we find experimentally as a first approximation (generally sufficient to determine a sufficiently optimized cell line):

where: U is the voltage at the terminals of the tank, counted in volts, generally between 3.8 and 5.5 volts for tanks correctly operated by those skilled in the art,

F is the Faraday yield of the tank, generally between 0.88 and 0.96 for these same properly conducted tanks dF / d (AI ₂ 0 ₃ ) is the algebraic drift of the Faraday yield relative to the alumina content of the bath , counted as a% of Faraday per% of alumina, in the region of alumina contents between 1% and 4%, and preferably in the region of A1 ₂ 0 ₃ contents between 1.5% and 3%.

This factor dF / d (Al ₂ O ₃ ) depends on many factors such as the composition of the bath (acidity = excess of AIF ₃ ), its overheating (i.e. the difference between the effective temperature of the bath and its starting solidification temperature).

magnetic balance (and in particular the agitation and deformation of the bath / metal interface).

In general, this factor dF / d (Al ₂ O ₃ ) must be determined experimentally for each type of tank and for the various types of bath used (low acid baths, with less than 8% excess of AIF ₃ , or very acid baths, with more than 8% excess of AIF ₃ or with sub-additives such as LiF and MgF ₂ ). Once determined, it no longer depends, as a first approximation, on economic conditions.

(c'est-à-dire que le rendement Faraday augmente de 1,5 % quand la teneur en alumine baisse de 1 %). Pour cette même cuve, dans les mêmes conditions de fonctionnement, on a mesuré

• F = 0,95 (soit 95 %)
• V - 4,10 volts

By way of nonlimiting example, for a tank of nominal intensity 280 KA, operating with a bath at 13% excess AIF ₃ and less than 1% LiF, with a bath temperature of approximately 950 ° C and an alumina content of between 1.7% and 2.5%, we have found:

(i.e. the Faraday yield increases by 1.5% when the alumina content drops by 1%). For this same tank, under the same operating conditions, we measured

• F = 0.95 (i.e. 95%)
• V - 4.10 volts

We deduce the technical coefficient K ₂ for this type of tank working in acid bath:

Example of implementation

The invention was applied to a series of electrolysis cells operating at an intensity of 280 KA, at a voltage of 4.10 volts per cell and giving a Faraday yield of 95.0% for an average alumina content in the bath. electrolysis equal to 2.3%, previously regulated according to the method of our patent FR-2 487 386 already cited (process called "slope calculation").

The daily production of the series per tank was 2.145 kg / day, for an energy consumption of 12.860 kWh / ton.

• K1 la valeur de: 3,10
• K2 la valeur de: + 1,8/100
• (J nominal étant égal à 280 KA)
  

Ayant choisi une cadence de sous alimentation C- égale à 70 % de la cadence nominale C_N, correspondant à une dérive de teneur en alumine D = - 0,64 % par heure, on s'est alors placé selon la méthode de recherche précédente, au point de fonctionnement tel que dRi/dt = Po x D = + 130.10-⁶ microohm/seconde au moment où l'on enclenchait la cadence nominale C_N.The parameter Po was determined by taking for:

• K1 the value of: 3.10
• K2 the value of: + 1.8 / 100
• (Nominal J being equal to 280 KA)
  

Having chosen a rate of undernourishment C- equal to 70% of the nominal rate C _N , corresponding to a drift in alumina content D = - 0.64% per hour, we then placed ourselves according to the previous research method , at the operating point such that dRi / dt = Po x D = + 130.10- ⁶ microohm / second when the nominal rate C _N was engaged.

The following results were obtained:

The gain on the cost price (depreciation and financial costs included) was 20 F per tonne of aluminum produced.

Claims (8)

1. A process for precisely regulating a low alumina content of between 1 and 3 % in a cell for the production of aluminium by electrolysis using the Hall-Heroult process, which makes it possible to obtain a Faraday efficiency which is at least equal to 94 %, by using an alternation of periods of feeding the electrolysis cell with alumina at a nominal rate CN, at a reduced rate C- and at a rate C+ which is higher than the nominal rate, the process firstly comprising determning a regulation parameter P = -1/D. (dRi/dt) expressed in micro-ohms per second and per % by weight per hour, Ri being the internal resistance of the cell and t being time, D being the derived function of the alumina content of the electrolysis bath expressed in % by weight per hour, in accordance with the relationship

CN being the nominal alumina feed rate and C- being the under-feed rate, reckoned in kg of alumina per unit of time, Q(AI₂0₃) being the amount of alumina consumed by the cell in the same unit of time and Q(b.l.) being the amount of liquid electrolysis bath contained in the cell, the process then comprising repetitively effecting the following operations:

a) the cell is fed at the nominal rate CN (such that the amount of alumina introduced into the bath is substantially equal to the amount consumed by electrolysis);

b) an over-feed of alumina at a rate C⁺ which is higher than the nominal rate CN is periodically triggered off so as to enrich the bath with alumina, for a predetermined time t^+. During that period dRi/dt is negative;

c) the feed rate goes to an under-feed rate, that is to say a rate C- which is less than CN. The curve dRi/dt passes through zero and then becomes positive. The regulation parameter P, the value of which tends to rise, is measured frequently; and

d) the successive values of P are compared to a predetermined reference value Po. As soon as P = Po, the feed rate goes back to the nominal feed rate CN and a fresh cycle is begun again at (a).

2. A regulating process according to claim 1 characterised in that after over-feed stage (b), the feed rate goes to the normal rate CN for a few minutes before going to an under-feed rate.

3. A regulating process according to claim 1 characterised in that after the over-feed stage (b) the feed rate goes to a rate which is little different from CN for a few minutes.

4. A regulating process according to claim 1 characterised in that the reference value Po in respect of the regulation parameter P is detetmined on the basis of the current strength J in kA of the electrolysis current and two coefficients K1 which is related to production costs and K2 which is related to the physico-chemical characteristics of the cell, in accordance with the relationship Po = K1 K2/J.

5. A regulating process according to claim 4 characterised in that the coefficient K1 is substantially equal to the ratio of the sum of the fixed transformation costs (energy, consumable carbonaceous products, labour and depreciation) to the cost of the electrical energy.

6. A regulating process according to claim 4 characterised in that the coefficient K2 is substantially equal to:

7. A regulating process according to any one of claims 1 to 4 characterised in that the reference value Po of the regulation parameter P expressed in micro-ohms per second and per percent by weight per hour is fixed between 2/100 - J and 10/100 - J, the strength J of the electrolysis current being expressed in kA.