EP0841112B1 - Process for casting between cylinders - Google Patents

Abstract

A detection method of continuous casting between rolls of steel products comprises measuring continuously the roll separation strain (RSF) and a signal representative of the variations in this strain (RSF) generated as a function of time. The signal is used to act upon the roll separation to compensate for any untrue roundness of the rolls. During these measurements the signal is separated into different harmonic components and the resulting comparison of the harmonic components thus obtained with some reference harmonics is representative of the state of defects in the rolls. This state of defects in the rolls allows the definition of different rules for conducting the casting process as a function of the gravity of the roll defects measured.

Description

The present invention relates to continuous casting between cylinders of thin metal products, in particular in steel.

According to this known technique, the product produced, for example a thin steel strip of a few millimeters thick, is obtained by pouring the molten metal in a casting space defined between two cylinders with parallel axes, cooled and driven rotating in opposite directions. In contact with the walls cold cylinders, called ferrules, the metal solidifies and the metal skins solidified, entrained by the rotation of the cylinders, meet at the neck between the cylinders, to form said band, pulled down.

The operation of the casting process between cylinders is subject to various constraints relating to both poured product only at the installation of the installation casting.

The casting strip must in particular have a section corresponding, in shape and dimensions, to the desired section, the actual section of the strip being directly function of the space, called air gap, between the cylinders at the neck.

For this, we know a method of regulating continuous casting between cylinders, described in patent application FR-A-2728817, according to which we measure the cylinder spacing force (RSF) and we act by consequence on the relative position of said cylinders. This process makes it possible to act on the relative position of cylinders to spread them in the event of excessive effort or bring them together in the case of too little effort, so especially to avoid liquid metal breakthroughs or even a break in the casting strip, and also to avoid damage to cylinders in the event of over-solidification cast metal.

Furthermore, we know that a false round of cylinders cannot be totally avoided on the one hand for mechanical reasons and on the other hand because of thermal deformations undergone by the shell, during first contact of molten metal at start-up of the casting, and also later during the rotation cylinders. It is already known to compensate for this false round, which will be called thereafter "normal runout" (or still "mechanical out of round" although partly of thermal origin), by acting automatically on the position of the bearings of at least one of the cylinders in function of the angular position of the cylinders, so as to keep the air gap as constant as possible. Given the practical impossibility of measuring directly the air gap, it has already been proposed to use as a parameter representative of the runout a signal provided by the means of measuring the spacing force cylinders, the runout compensation system then being combined with a regulatory system such as that described in the document FR-A-2728817 already cited.

The implementation of these methods does not allow however not detect in real time certain faults likely to disrupt the flow or even lead to stopping it, or permanently damaging the cylinders.

We already know methods of detecting defects, visual or otherwise, making it possible to detect defects related to the casting process, to the thermo-hydraulic molten metal, or those known under the name of "shiny bands". This last type of default corresponds to a localized decrease in the surface roughness of the cylinders, which causes variations in strip cooling which can be detected by temperature measurements on the casting strip. But the observation of these faults does not can be done only a posteriori, on the strip already formed, and therefore belatedly after their appearance. These defects may damage the surface condition of the cylinders, and all the more so since they are perceived late, these damage can then become irreparable.

Certain faults could a priori be detected from direct observation of the representative signal the effort of spacing the cylinders. But the variations of this signal represent both variations of effort due to normal runout and variations due to other parameters where events that may occur in casting course. Direct observation of the signal effort therefore does not make it possible to determine the share of each of these causes in signal variations.

The object of the present invention is to resolve the problems mentioned above and aims to allow, from of the measurement of the spacing force of the cylinders (RSF), real-time detection of faults, before that an amplification of these defects causes damage irremediable in particular to cylinders. The invention has also intended to allow monitoring of the evolution of these faults, in order to be able to offer the operator corrective actions or interruption of the casting in depending on the severity of said defects.

With these objectives in view the object of the invention is a continuous casting process between product cylinders thin metals, especially steel, according to which continuous measurement, during the casting, the effort cylinder spacing, and a signal is generated representative of variations in spreading force (RSF) as a function of time, and we act, in particular by function of said signal, on the spacing of the cylinders to compensate for the runout of the cylinders, this process being characterized in that, for the purpose of detecting faults other than the runout of the cylinders, we breaks down said signal into different components harmonics, and we compare these said components harmonics to rank reference harmonics corresponding, the results of said comparison being representative of a fault state of the casting process, and we define, according to the results of said comparison, rules of conduct for the casting process.

The inventors were able to establish, following numerous tests carried out on an industrial scale, there is a certain relationship between variations in signals representative of the spreading force and the appearance of defects during casting. For example, the appearance on a cylinder of the defect called strip brilliant is characterized by the presence of a disturbance on the spreading force signal measured. This disturbance is cyclical and manifests itself at each revolution of the cylinder. It reflects an over-solidification of the product when it goes to the collar and translated by variations in effort which are clearly faster than those that can be generated by example by variations in product thickness solidified.

The inventors then imagined making a decomposition into harmonics of said signals so to differentiate in these signals the part that can be attributed to normal runout and that from other causes. They thus verified, by comparison of harmonic components noted during various flows, that although representative signals from the spacing effort varies in particular depending even if it is offset by a compensation system, variations of some harmonic components corresponded to the appearance of defects during casting. It therefore appeared that a analysis, carried out continuously during casting, of these harmonic components could allow, by comparison with a reference obtained experimentally when casting is considered to be faultless, to detect almost real-time deviations revealing such casting faults, much faster than by known methods.

An explanatory hypothesis of the existing relationship between the variations of the harmonic components and the presence of casting faults is that the normal runout causes variations in the signal representative of the cylinder spacing force (RSF) which are mostly slow and gentle, in other words that said signal emerge, due to the said normal false round, essentially a harmonic component of low rank, of frequency equal to the frequency of rotation of cylinders. On the other hand, real defects, such as shiny bands mentioned above, generate mainly abrupt variations of said signal and therefore higher order harmonics. Typically, the spectrum of the signal representative of the spreading force of cylinders and resulting from the only normal runout is characterized by a harmonic component of rank 0 significant (for example 70% of the total amplitude of the signal) and higher order harmonics in rapid decay (20% for the 1st harmonic, 10% for the 2nd harmonic). We rarely note the presence of higher order harmonics. On the other hand, in the case of bright bands, the distribution of harmonics is different from the above case, the presence of an over-solidification front at level of the shiny strip generating more high harmonics.

It is specified that here and thereafter, the term harmonic of rank i denotes the component of the signal at a frequency F _i = 2 ⁱ F ₀ , F ₀ being the fundamental frequency corresponding to the speed of rotation of the cylinders. Similarly, the amplitude of the harmonic components of rank i will be designated subsequently by h _i , and by H _i a value representative of the harmonics of rank i considered over a predetermined number of revolutions of the cylinders.

According to a particular provision of the invention when an air gap regulation system, such as described above, is set up, we can use as a signal representative of variations in effort cylinder spacing (RSF), derived from the measurement of said effort, an associated signal used as a setpoint displacement of the bearings of at least one cylinder. It's at say that the signal which is then broken down into different harmonic components is directly related to said displacement setpoint developed by a module offset compensation, and therefore reflecting the variations in the spreading force.

To decompose the signal into its different harmonic components, we can in particular use a fast Fourier transform applied to signal representative of the spreading force of the cylinders (RSF), this signal therefore being either directly the spreading force measurement signal, i.e. a signal correspondent developed by the said compensation module out of round.

In a preferred arrangement of the invention, the value H _i representative of each harmonic of rank i is calculated as being an average value of the amplitudes h _i of each harmonic, determined over a given number of revolutions of the cylinders. The value H _i representative of each harmonic being calculated as being an average over a given number of turns of the amplitudes measured, this makes it possible to attenuate the effect of random and localized defects in time and space, non-repetitive over several turns cylinder. Thus, if the defect is caused by a lasting problem on a cylinder, the system will completely integrate this data after said number of turns, while the effect of harmonics only appearing on a low number of turns, significantly less than said number of turns given, will be considerably reduced.

The comparison of the measured signal with a signal of a flow deemed good can be done in different ways. One can simply compare term by term the values H _i representative of each harmonic of the signal measured with reference values H _ir coming from measurements carried out during flows considered good, and verify that the sum of the differences of the values H _i representative of each harmonic with the reference H _ir values is not too high. We can also compare the proportion of each harmonic compared to a proportional reference distribution. However, preferably, the comparison will be carried out on the basis of a barycenter of the harmonics, this barycenter being calculated by weighting each harmonic by a predetermined coefficient, so as to give the different harmonics relative importance by unequally weighting the latter. This method of calculation is justified by experimental observations: during a casting considered good, the first harmonic is the most important, the importance of the different harmonics being decreasing according to the increasing rank of the harmonics considered. By weighting the harmonics of higher rank by an adapted coefficient, the variations of these harmonics of higher rank will be somehow amplified, making their appearance or increase more easily perceptible in the result of the barycenter calculation.

We can for example calculate a frequency barycenter B _f by assigning to each frequency of harmonics a coefficient representing the amplitude of the harmonic considered: B f (Hz) = Σ H i * F i / Σ H i and normalize this barycenter by the fundamental frequency to obtain a ratio R = B _f / F ₀ which can be compared with a predetermined reference value R ₀ , so as to overcome any differences in fundamental frequency, and therefore speed effective of the cylinders, between the casting considered and the reference.

We can also calculate the derivative dR / dt and also compare the result to a second threshold predetermined, thus allowing to follow the evolution of the ratio R over time, a rapid evolution of R being somehow the sign of a rapid worsening of a default.

A représentant l'amplitude totale des variations : A = Σ Hi

R représentatif de la part ou de l'importance des défauts dans le signal,

et E = dR/dt

   on peut établir un tableau de décision, comme on le verra par la suite, qui pourra être utilisé pour proposer en temps réel à l'opérateur des actions correctives sur certains paramètres de coulée, dans le but visé de remédier à des défauts le plus rapidement possible après leur apparition.With the values of the different parameters:

A representing the total amplitude of the variations: A = Σ H i

R representative of the share or the importance of the defects in the signal,

and E = dR / dt

we can establish a decision table, as we will see later, which can be used to propose in real time to the operator corrective actions on certain casting parameters, with the aim of remedying faults as quickly as possible possible after their appearance.

• la figure 1 est une vue schématique d'un dispositif de coulée entre cylindres avec un système de régulation de type connu en soi, mais utilisant une décomposition en harmonique du signal de compensation de faux rond,
• la figure 2 représente un tableau de décision permettant de définir la démarche à suivre au cours de la coulée en fonction des valeurs des différents paramètres fournis par le procédé selon l'invention,
• les figures 3a, 3b, 3c et 3d présentent, sous forme de tracés représentant les variations des différents paramètres mesurés ou calculés, les résultats obtenus lors d'une coulée jugée bonne avec procédé de compensation de faux rond,
• les figures 4a, 4b, 4c et 4d présentent les tracés correspondants obtenus lors d'une coulée jugée mauvaise.

Other advantages and particularities will appear on reading the detailed description which will follow of embodiments of the invention, given by way of indication and in no way limitative, to be read in conjunction with the appended drawings among which:

• FIG. 1 is a schematic view of a device for casting between cylinders with a regulation system of the type known per se, but using a harmonic decomposition of the runout compensation signal,
• FIG. 2 represents a decision table making it possible to define the approach to be followed during casting as a function of the values of the various parameters provided by the method according to the invention,
• FIGS. 3a, 3b, 3c and 3d show, in the form of plots representing the variations of the various parameters measured or calculated, the results obtained during a casting judged to be good with a method of compensation for runout,
• Figures 4a, 4b, 4c and 4d show the corresponding plots obtained during a casting deemed bad.

The casting installation, shown only partially in Figure 1, includes so conventional, and known per se, two cylinders 1, 2, of axes parallel, spaced from each other by a named distance air gap. It corresponds to the desired thickness of the casting strip, minus the crushing due to the RSF. Both cylinders 1, 2 are rotated in direction opposite, at the same speed. They are carried by bearings 3, 4, schematically represented, of two supports 5, 6 mounted on a chassis 7. Support 5, and therefore the axis of the corresponding cylinder 1 is fixed by relative to the chassis 7. The other support 6 is movable in translation on the chassis 7. Its position is adjustable and determined by thrust cylinders 9 acting from so as to bring the supports 5, 6, one apart the other. Means for measuring the spreading force cylinders (RSF), such as load cells 8, are arranged between the fixed support 5 and the chassis 7. Sensors 10 measure the position of the mobile support 6, and therefore the position variations with respect to a predetermined set position according to the desired thickness of the strip.

During casting, the molten metal is spilled between the cylinders, and begins to solidify at contact of their cooled walls by forming skins solidified which are driven by the rotation of cylinders and meet substantially at the neck 11 between the cylinders to form the solidified strip pulled down. In doing so, the metal exerts on the cylinders a spreading force (RSF), measured by weigh 8, this effort being variable in particular depending the degree of solidification of the metal.

To regulate this effort, and guarantee the continuity of the casting, the casting installation includes a regulation system. In this regulation system, the difference between the effort setpoint signal and the effort signal measured by the effort sensor 8 is calculated by a first comparator 12. The signal for this difference is introduced into a regulator force 13 which determines a position setpoint signal introduced into a second comparator 14. The force signal measured by the force sensor 8 is also introduced into a runout compensation system 15 which performs a decomposition into harmonics of the force signal and generates signals H ₁ , H ₂ , H ₃ for compensation of each of said harmonics. These signals H ₁ , H ₂ and H ₃ are added to an adder 16 which generates a position correction setpoint signal which is transmitted to the second comparator 14. The output signal from the second comparator 14 is introduced into a third comparator 17 as well as a position signal from the position sensor 10. The output signal from the third comparator 17 is introduced into the position regulator 18 which controls the jacks 9.

The rotation of cylinders 1 and 2 is ensured respectively by motors 19 and 20 controlled by a cruise control 21. This cruise control 21 receives a signal from a thickness regulator 22 receiving itself a thickness reference signal, the signal force emitted by the force sensor 8 and the signal position emitted by the position sensor 10.

An action on the cylinders 9 is executed automatically by this regulation system, which allows for example to act on the cylinders 9 in the direction leading to a spacing of the cylinders to reduce the spreading force (RSF), or vice versa in the direction a bringing together of the cylinders to increase the effort. Similarly, this system allows for a at least partial compensation for the normal false round, i.e. to compensate for any existing misalignment between the axis of the ferrule and its axis of rotation as well as irregularities in the shape of a cylinder, that these are of mechanical or thermal origin. The system of regulation then takes these shape defects into account and of coaxiality to give a displacement instruction to thrust cylinders 9 controlling the air gap of the cylinders in order to keep this air gap as constant as possible during the rotation of the cylinders.

We will now describe a preferred method of determination of the different parameters A, R and E which will be used to warn the operator of the presence defects and the seriousness of these.

In this method we use a decomposition into harmonics of the signal representative of the force cylinder spacing, this decomposition being performed in the runout compensation module 15 using a Fourier transform. We could everything as well perform the same operation not using of a Fourier transform but with a transform of Laplace or any other mathematical or signal processing such as for example the use filters to achieve the same result, know a breakdown of a signal into different harmonic components.

The values H _{i are} then calculated as indicated above, that is to say by performing an average of the amplitudes h _i over a predetermined number of revolutions of the cylinders, for example over the last ten revolutions. It will be noted that the preceding method for calculating the coefficients H _i is given only by way of example and is not at all limiting. The values H _i representative of each harmonic of rank i can also be calculated as being the effective value of the amplitude h _i of the harmonics or any other calculated value characterizing these said harmonics, this calculation being able to be made according to an arithmetic mean, according to the least squares method or any other method.

Whatever the calculation method, the values H _i are representative of the relative amplitude of each harmonic of rank i and of frequency F _i .

The criterion B _{f is} then calculated as being a frequency barycenter of the different harmonics. That is to say that the barycenter of the frequencies of the harmonics under consideration is calculated, each value F _i being assigned a weight constituted by the corresponding value H _i , that is: B f = Σ H i x F i / Σ H i .

We will generally only use harmonics of rank 0, 1 and 2. However it would obviously be possible to take into account more harmonics.

In order to be able to make valid comparisons at different speeds of rotation of the cylinders, the ratio R _f = B _F / F ₀ , F ₀ corresponding to the frequency of rotation of the cylinders is then calculated.

• amplitude globale des variation du signal : A=H1+H2+H3,
• barycentre normé : Rf = (F1xH1+F2xH2+F3xH3)/((H1+H2+H3)xF0)
• évolution de R_f dans le temps : E=dR_f/dt.

In the case taken as an example where only the first three harmonics are taken into account, the following three criteria are obtained:

• overall amplitude of the signal variation: A = H 1 + H 2 + H 3 ,
• standardized barycenter: R f = (F 1 xH 1 + F 2 xH 2 + F 3 xH 3 ) / ((H 1 + H 2 + H 3 ) xF 0 )
• evolution of R _f over time: E = dR _f / dt.

A comparison of these different calculated criteria during casting with a predetermined threshold allows then to detect for the casting in progress if such fault appears.

H₀ = 700 µm, H₁ = 200 µm, H₂ = 100 µm, avec

F₀ = 0,2 Hz, F₁ = 0,4 Hz et F₂ = 0,8 Hz ,

alors B_f = 0,3 Hz et R_f = 1,5 .

By way of example, in a case where the signal representative of the spacing force of the cylinders is the signal from the runout compensation module, that is to say expressed in displacement value of the movable cylinder, and in presence of the only normal runout, we can have:

H ₀ = 700 µm, H ₁ = 200 µm, H ₂ = 100 µm, with

F ₀ = 0.2 Hz, F ₁ = 0.4 Hz and F ₂ = 0.8 Hz,

then B _f = 0.3 Hz and R _f = 1.5.

If a bright band appears, these values respectively become 350 µm, 350 µm and 300 µm for H ₀ , H ₁ , H ₂ , and then R _f = 2.25.

It is thus seen that by simply fixing an adequate threshold on R _f , for example R _fseuil = 1.6, the passage of R _f above this threshold can activate an alarm signaling a fault.

A better appreciation of the severity of faults can be obtained by taking into account of the three criteria mentioned above.

For this, we could for example use a decision table as shown in figure 2 which directly indicates to the operator the defective state of casting, i.e. gives it an indication of the presence, importance, and evolution of faults and signals the need to take corrective actions, such as changes to certain parameters of casting to try to remedy the defects which have appeared, or at worst the need to stop pouring to avoid irreparable damage to the casting installation.

• A "petit" est le signe de faibles variations de l'effort d'écartement des cylindres, la coulée se déroule dans de bonnes conditions,
• quand A est "moyen",
• si R et E sont "petit", ce qui signifie pas ou peu de défauts,, la coulée se déroule encore dans de bonnes conditions,
• si R est "petit" et E "grand", cela peut signifier que, bien qu'il n'y ait pas de présence réelle de défauts, le point de fonctionnement de l'installation est instable, pour des raisons liées essentiellement au faux rond "normal", et une alarme du procédé de coulée est déclenchée pour avertir l'opérateur d'un besoin de modifier par exemple les conditions thermiques de la virole (température ou débit de l'eau de refroidissement),
• si R est "grand" et E "petit", ce qui signale la présence de défauts, sans tendance notable à une éventuelle aggravation de ceux-ci, une alarme du procédé de coulée est déclenchée,
• si R et E sont "grand", signalant la présence de défauts et l'aggravation de ceux-ci, un arrêt de la coulée est demandé,
• quand A est "grand",
• si R et E sont "petit", aucun défaut latent n'est signalé, le faux rond normal est correctement compensé, mais l'amplitude des déplacements du cylindre mobile pour réaliser cette compensation est importante, ce qui n'est pas grave pour la coulée elle-même, mais peut révéler des problèmes de géométrie des cylindres,
• si R est "grand" et E "petit", ce qui signale en plus la présence de défauts, mais sans aggravation notable, une alarme du procédé de coulée est déclenchée,
• si E est "grand", quelque soit la valeur de R, une forte aggravation des défauts est signalée et un arrêt rapide de la coulée est demandé.

This table presents for example the procedure to be followed according to the relative values of the coefficients A, R _f and E:

• A "small" is the sign of small variations in the spacing force of the cylinders, the casting takes place in good conditions,
• when A is "average",
• if R and E are "small", which means little or no defects, the casting is still taking place in good conditions,
• if R is "small" and E "large", this can mean that, although there is no real presence of faults, the operating point of the installation is unstable, for reasons essentially related to the false round "normal", and an alarm of the casting process is triggered to warn the operator of a need to modify for example the thermal conditions of the shell (temperature or flow of cooling water),
• if R is "large" and E "small", which signals the presence of faults, without any noticeable tendency for them to worsen, an alarm of the casting process is triggered,
• if R and E are "large", signaling the presence of faults and their worsening, a stop of the casting is requested,
• when A is "big",
• if R and E are "small", no latent fault is reported, the normal runout is correctly compensated, but the amplitude of the displacements of the movable cylinder to achieve this compensation is important, which is not serious for the casting itself, but can reveal cylinder geometry problems,
• if R is "large" and E "small", which additionally signals the presence of faults, but without noticeable deterioration, an alarm of the casting process is triggered,
• if E is "large", whatever the value of R, a strong aggravation of the defects is signaled and a rapid stopping of the casting is requested.

Note that the characters "small", "medium" and "large" of the different criteria are appreciated by comparison with experimental data acquired during previous flows.

As an illustration of the possibilities of detection of faults by the method according to the invention, can refer to Figures 3a, 3b, 3c and 3d which show variations of different parameters measured and calculated during a casting process with compensation for runout deemed good, and in Figures 4a, 4b, 4c and 4d which show comparatively the curves obtained during casting with defects in shiny bands.

Figures 3a and 4a show the variations of the spacing force of the cylinders expressed in percentage of allowable RSF, measured for 40 minutes from the start of casting.

Figures 3b and 4b show the evolution during this time of parameter A, i.e. the average amplitude over 10 turns, in µm, of the displacement of the bearings of the mobile cylinder controlled by the compensation module of the false round.

Figures 3c and 4c show in correspondence evolution of the parameter R.

Figures 3d and 4d show in correspondence on the same graph the changes in the values H ₀ , H ₁ and H ₂ , representative of the amplitudes of the harmonics of rank 0, 1 and 2, the first (H ₀ ) being represented at the bottom of the diagram. , the second (H ₁ ) in the middle and the third (H ₂ ) at the top.

It is noted that, in the case of the casting judged good, the growth of A for approximately the first 20 minutes corresponds to a similar growth of H ₀ and essentially reflects the evolution of the compensation for runout, until a stability is obtained. of A in the vicinity of 50 µm, signaling an almost perfect out-of-round compensation. There is also a stability of the parameter R after ten minutes, after an excursion of R towards higher values, corresponding to a relatively large amplitude of H ₂ during the same period of start of casting.

By comparison, the plots in FIGS. 4b, 4c and 4d, relating to a flow the course of which has been greatly disturbed, show large amplitudes of H ₁ and H ₂ for approximately 40 minutes, with a high value of a during the same period. , and above all a high value of R.

It will be readily understood from these readings that a comparison, carried out in real time during casting, of the values of A and especially of R with predetermined thresholds would have made it possible to quickly detect the defects corresponding to the large amplitudes of the harmonics H ₁ and H ₂ , and to act immediately on the casting parameters to prevent them from getting worse.

The invention is not limited to the calculation methods of the various parameters indicated above only at as an example.

In particular, by always using the same values H _i representative of the amplitude of each harmonic, it will be possible to calculate another barycenter B of the harmonic spectrum of the value representative of the force of spacing of the cylinders, for example by then assigning to each value H _i a judiciously chosen weighting coefficient making it possible to accentuate in the calculated value of this barycenter the influence of the harmonics of higher rank, which are significant of defects. Whatever the type of barycenter calculation used, we will use values representative of the different harmonics and weighting coefficients relating to each harmonic such that we can easily follow the evolution of the value of the barycenter and compare it with experimental values with a view to defining a defective level in real time by comparison with the defective state (trouble-free casting, disturbed casting, bad casting having led to a stop or damage to the cylinders, etc.) of the previous castings.

To compare the harmonics, we can also define a reference distribution of the amplitudes of the harmonics, as a percentage of each harmonic relative to the total signal, for example by assuming a priori that the first harmonic represents 66% of this signal, the second 17% and the third also 17%. We can then follow the evolution of this distribution during each casting and, by comparison with the reference distribution, easily assess any deviations. This comparison could for example be made by calculating a sum R _d of the differences between the proportion H _i / A of each harmonic component in the measured signal representative of the spacing force and the reference proportion α _i : R _d = pos (α ₀ -H ₀ / A) + pos (H ₁ / A-α ₁ ) + ... + pos (H _i / A-α _i ), (i.e., each element of this sum n ' is counted only if it is positive). In this way, if the proportion of the harmonic of rank 0 is greater than the reference proportion or if the proportion of a harmonic of rank greater than or equal to 1 is less than the reference proportion, the difference relative to the harmonic considered is not taken into account. For example, if the first harmonic represents for example 98% of A, the second 2% and the third 0%, which would correspond to an almost total absence of harmonics of rank greater than 0 and therefore to an absence of faults, R _d = 0.

In the case where the continuous casting installation between cylinders would not include a system for regulating the air gap as a function of the runout, it would obviously be possible to apply the method according to the invention described above by taking directly as a signal subjected to a harmonic decomposition the direct measurement of the variations in the force of spacing of the cylinders (RSF), the use of the values H _i coming from the module of out-of-roundness remaining however particularly practical when such a module of compensation already exists in the installation and already performs, as part of its normal operation, the required harmonic decomposition.

Claims (10)

1. Process for the continuous twin-roll casting of thin metal products, especially steel products, in which the roll separation forge (RSF) is continuously measured during the casting run and a signal representative of the variations in the separation force (RSF) as a function of time is generated, and the roll separation is varied, especially according to the said signal, in order to compensate for the out-of-roundness of the rolls, characterized in that, for the purpose of detecting defects other than the out-of-roundness of the rolls, the said signal is decomposed into various harmonic components and these said harmonic components are compared with reference harmonics of corresponding rank, the results of the said comparison being representative of an improper state of the casting process, and, depending on the results of the said comparison, rules for carrying out the casting process are defined.
2. Process according to Claim 1, characterized in that the said representative signal, coming from the measurement of the variations in the roll separation force (RSF), is an associated signal used as a setpoint for displacement of the bearings of one roll in a loop for controlling the gap between the said rolls.
3. Process according to any one of the preceding claims, characterized in that a Fourier transform is used so as to decompose the said signal representative of the roll separation force (RSF) into various harmonic components.
4. Process according to any one of the preceding claims, characterized in that, in order to make the comparison, what is used as value representative of each harmonic of rank i is a value H_i corresponding to the mean of the amplitudes h_i of the harmonics of this rank which are measured over a given number of revolutions.
5. Process according to any one of the preceding claims, characterized in that a centroid of the harmonics is used to make the comparison, this centroid being calculated by weighting a value representative of each harmonic by a predetermined coefficient.
6. Process according to Claim 5, characterized in that a frequency centroid (B_f) = (Σ(H_i×F_i))/(ΣH_i) is calculated in which the value representative of each harmonic is its frequency F_i and the weighting coefficient H_i represents the amplitude of the harmonic in question.
7. Process according to Claim 6, characterized in that the comparison is made on the basis of a ratio R_f = B_f/F₀, where F₀ is the frequency corresponding to the speed of rotation of the rolls.
8. Process according to Claim 1, characterized in that the comparison is made by using, as comparison criterion, the proportion H_i/A of each harmonic component with respect to the signal representative of the separation force, H_i representing the amplitude of the harmonic of rank i and A = ΣH_i.
9. Process according to Claim 8, characterized in that the result of the comparison is represented by the sum R_d = pos(α₀-H₀/A)+pos(H₁/A-α₁)+...+pos(H_i/A-α_i).
10. Process according to either of Claims 7 and 9, characterized in that a decision table is used for determining the operating procedure to be followed for the casting run, according to the values of the criteria:

    A = ΣH_i;

    R (R_f or R_d);

    E = dR/dt.

Images