Computer implemented method for providing control data for a continuous casting facility

Abstract

The invention relates to a computer-implemented method for providing control data for a continuous casting facility for casting a metal, in particular steel, in a continuous casting process, the continuous casting facility comprising a mould, wherein the mould comprises a plurality of temperature sensors distributed across at least one mould wall, the method comprising the following steps: (a) providing temperature data from the plurality of temperature sensors; (b) optionally providing casting parameters comprising settings and / or measurement data of the casting process; (c) providing at least one control model configured to provide control data for the continuous casting facility based on the temperature data and optionally the casting parameters, wherein the control model is configured to compute at least one key parameter indicator characterizing the behaviour of the mould powder during the casting process; and (d) providing control data for the continuous casting facility utilizing the control model, wherein the control data is based on the at least one key parameter indicator.

Description

[0001] Computer implemented method for providing control data for a continuous casting facility

[0002] FIELD OF THE INVENTION

[0003] The invention relates to a computer-implemented method for providing control data for a continuous casting facility for casting a metal, in particular steel, in a continuous casting process.

[0004] BACKGROUND TO THE INVENTION

[0005] In the state of the art, the continuous casting process of steel is controlled by means of a few thermocouples for measuring temperature positioned inside the copper mould plates. In recent years, Fibre Bragg Gratings (FBG) have been used as temperature sensors in the mould, which greatly increased the number of possible measurement points in the mould.

[0006] Various methods of using the temperature data have so far been proposed. For example, A. Krasilnikov et al. in “Local Heat Transfer Through Mould Flux Film and Optimal Narrow Face Taper Adjustment”, AISTech 2015 Proceedings, propose to perform taper changes in response to temperature measurements in the mould by means of fibre-optical sensors.

[0007] Patent document JP-2003-181609 discloses a method for controlling a flow pattern of molten steel in the mould during continuous casting. The method includes the steps of: (a) continuously casting molten steel delivered from an immersion nozzle; (b) measuring the temperature of copper plate of the mould in a plurality of points in the width direction of the a long-side of the mould; (c) detecting a flow pattern of the molten steel within the mould based on a change in copper plate temperature in each measuring point with the elapse of time; and (d) controlling the flow pattern based on the results of detection so as to provide a predetermined flow pattern. The flow pattern may be controlled by applying a magnetic field to the molten steel.

[0008] Patent document JP-2017-087235 A discloses a continuous casting method capable of reducing shrinkage cavities without causing any quality defect in a cast slab. Mould powder is added to end parts of a cast slab after the supply of molten steel to a casting mould is completed; and microwaves generated by a microwave oscillator are radiated to an upper space ofthe supplied mould powder to heat the mould powder.

[0009] However, so far, a comprehensive control model allowing to evaluate the behaviour of the mould powders and its influence on the solidification quality and stability during casting has not yet been proposed. In the state of the art, a mould powder composition is chosen based upon experience and expertise . However, when a new steel grade is being cast, the composition of the mould powder must be adapted mainly based on trial and error, which is a slow and resource consuming method.

[0010] OBJECT OF THE INVENTION

[0011] It is thus an object of the invention to find a method of characterising the mould powder behaviour during the casting process to thereby generate control data, which allows to control the casting process based on the characterisation. It is a further object of the invention to propose suitable control models configured for analysing sensor data and providing useful control data for a continuous casting facility.

[0012] DESCRIPTION OF THE INVENTION

[0013] These objects are met or exceeded by a computer implemented method for providing control data for a continuous casting facility according to independent claim 1, a system for providing control data for a continuous casting facility of claim 13, a computer program element according to claim 14 and a use according to claim 15.

[0014] According to a first aspect of the present invention, a computer-implemented method for providing control data for a continuous casting facility for casting a metal, in particular steel, in a continuous casting process, is provided, the continuous casting facility comprising a mould, wherein the mould comprises a plurality of temperature sensors distributed across at least one mould wall, wherein the continuous casting process comprises feeding liquid metal and a mould powder to the mould and extracting a strand of metal from the base of the mould at a casting speed, the method comprising the following steps:

[0015] (a) providing temperature data from the plurality of temperature sensors, the temperature data indicating a temperature distribution in the mould wall during the casting process;

[0016] (b) providing casting parameters comprising settings and / or measurement data of the casting process;

[0017] (c) providing at least one control model configured to provide control data for the continuous casting facility based on the temperature data and optionally the casting parameters, wherein the control model is configured to compute at least one key parameter indicator characterizing the behaviour of the mould powder during the casting process; and

[0018] (d) providing control data for the continuous casting facility utilizing the control model, wherein the control data is based on the at least one key parameter indicator.

[0019] A further aspect ofthe present invention relates to a system for providing control data for a continuous casting facility for casting a metal, in particular steel, in a continuous casting process, the continuous casting facility comprising a mould, wherein the mould comprises a plurality of temperature sensors distributed across at least one mould wall, comprising: a processing circuitry; a storage medium; and a data interface; wherein the storage medium comprises a computer program that comprises instructions which, when the program is executed, cause the processing circuitry to carry out the method according to an embodiment of the invention; wherein the data interface is configured to receive input data for the control model including the temperature data from the plurality of temperature sensors, and optionally casting parameters. The data interface may further be configured to output control data for the casting facility. The system may be comprised in a control computer for a continuous casting facility. It may also be a separate system. The processing circuitry may be comprised in a processing unit such as a CPU or GPU. It may be part of a computer, cloud computer, laptop, PC or mobile device such as a mobile phone or tablet computer. Optionally, the system may also comprise a plurality of temperature sensors distributed across at least one mould wall, in particular ofthe mould ofthe continuous casting facility. The mould may also be part of the system. The temperature sensors may be FBGs or thermocouples. Further, the invention is also directed to a continuous casting facility comprising such a system.

[0020] A further aspect of the present invention relates to a computer program comprising instructions, which, when executed on a computing device of a computing environment, is configured to carry out the steps ofthe computer-implemented method according to an embodiment of the invention. In particular, the computer program comprises instructions to cause the claimed system to execute the steps ofthe method ofthe invention. Further aspects of the present invention relate to the use of temperature data from a plurality of temperature sensors distributed across at least one mould wall, the use of casting parameters, and / or the use of a control model in a computer- implemented method according to an embodiment of the invention and / or in a system as described herein. The invention in particular relates to the use of temperature data from a plurality of temperature sensors in a method or a system according to an embodiment of the invention.

[0021] A further aspect ofthe present invention relates to a computer readable storage medium, in particular a non-transient storage medium, comprising instructions which, when executed by a computer, causes the computer to carry out the computer- implemented method for providing control data as described herein. The computer- readable storage medium may be any digital data storage medium, for example an optical, magnetic or solid -state storage medium, such as a disc, hard disc, SSD-card, SD-card, cloud computer, etc.

[0022] A further aspect of the present invention relates to a method for controlling a continuous casting facility comprising the following steps: providing temperature data from a plurality of temperature sensors, the temperature data indicating a temperature distribution in a mould wall during the casting process; controlling the casting process using control data provided according to the computer-implemented method for providing control data as disclosed herein.

[0023] All advantages and embodiments described herein relate to the methods, the system, the computer program element and the computer-readable storage medium alike . Advantageously, the benefits provided by any ofthe aspects, embodiments and examples equally apply to all other aspects, embodiments and examples, and vice versa.

[0024] The present invention allows to evaluate the temperature data from a plurality of temperature sensors in the mould wall(s) and thereby describe and control the mould powder’s behaviour during casting. The method may be used for controlling a continuous casting facility for metals, in particular for steel or non-ferrous metals and alloys. In a continuous casting facility, molten metal is continuously fed through a Submerged Entry Nozzle (SEN) into a water-cooled mould. The mould is typically ringshaped with an open base . The mould is typically made of copper. The mould’s horizontal cross-section may have any shape, wherein moulds for casting billets often have a round or square of dog-bone cross-section. Moulds for casting rolling slabs typically have a rectangular cross-section. Thus, the at least one mould wall may be one mould wall (e .g. in case of a round cross-section of the mould) or several mould walls. For example, moulds with a rectangular cross-section typically have four mould walls, also referred to as mould faces or sides, wherein each mould wall typically has an essentially rectangular shape . When reference is made to “mould walls” in the following, this is also meant to cover the case where there is only one mould wall, and vice versa. The mould wall(s) is / are cooled, typically by means of cooling water flowing through ducts in the mould wall(s). The molten metals thus start to solidify at the mould walls, so that a solidified steel shell is formed and is extracted at the base ofthe mould. The strand of solidifying metal, having a solidifying shell and a liquid centre, is continuously extracted from the mould by means of support rolls and further heat is extracted by means of spray cooling. During the casting process, the growth of the solidifying shell must be closely monitored, because an insufficient shell-thickness at the mould exit can result in a number of problems like bulging or cracks in the strand, or even to a catastrophic breakout which is also a major safety issue . Therefore, such events must be avoided by closely monitoring the heat-extraction from the mould.

[0025] To better monitor the temperature distribution in a mould wall, Fibre Bragg Gratings (FBG) are preferably used as the temperature sensors. FBG sensors are optical fibres that are usually made of silica. These fibres are configured for temperature measurements by introducing, at intervals along its length, segments having a periodic variation in the refractive index, so that the segments reflect particular wave lengths of light and transmit all others. When the optical fibre warms up, the periodic variation changes its wavelength due to thermal expansion, thus allowing to detect a temperature change by a change in wavelength. It is possible to have several reflective segments along the length, each having different reflective spectres, thereby, it is possible to include several temperature measurement points along one optical fibre, each being referred to as a “temperature sensor” herein. The plurality of temperature sensors in the mould therefore preferably comprises 200-5000 temperature sensors, more preferred 1000-4000, most preferred 1500-3000 temperature sensors. Preferably the mould comprises a grid of temperature sensors, preferably formed by FBG sensors. For example, the FBG sensors may be integrated into the mould wall at regular intervals ofbetween 10 and 100 mm.

[0026] Preferably, the plurality of temperature sensors is arranged on a grid, in particular a grid with the grid points spaced at intervals between 10 and 100 mm. The grid is preferably rectangular grid. This fine grid makes it possible to construct accurate heat maps to directly examine the in-mould behaviour. In other words, the temperature sensors are distributed across the mould wall in two dimensions (feed direction and perpendicular thereto, or vertical and horizontal). Preferably, the temperature sensors provide a two-dimensional (2D) temperature distribution of the mould wall, in particular having between 200 and 5000, preferably between 1000 and 4000 points at which the temperature is measured.

[0027] In order to prevent sticking of the solidified shell to the mould wall, mould powders are used. Mould powders are generally composed of various minerals. Mould powders are fed continuously to the top of the mould, where they melt. The mould powder first forms a sintered layer and then completely melts to form a fully liquid layer on top of the liquid steel, wherein the liquid mould powder is called mould slag. The mould slag penetrates the gap between the mould and the solidifying metal shell in order to lubricate the process and allow the solidifying shell to slide along the mould wall. The mould slag also controls the heat transfer through the mould wall. Since each steel grade has its own desired cooling rate, one mould powder is not suitable for every casting. For example, low carbon steels can be cast at fast cooling rates, while peritectic steels cannot. This is because peritectic steels have a higher thermal expansion coefficient, therefore as a result of fast cooling, they tend to crack. For peritectic steels, mould powders with particular specifications are required. For example, peritectic steels often require a mould powder that - when molten to form mould slag - controls and evens the local differences in heat transfer. In other words, the casting of peritectic steels requires a homogeneous lubrication and heat transfer, and this can be aided by a careful selection of mould powder. Thus, different steel grades (or groups of grades with similar composition) are cast with different mould powder chemistries. For example, there are special mould powder compositions for ultra-low carbon steels (ULCs), different compositions for peritectic grades etc. Thus, each steel grade has its own mould powder chemistry, but within such chemistry, changes in the composition can be done to improve its performance . Embodiments of the present invention allow to make such changes on the basis of the temperature measurements during the casting by generating corresponding control data.

[0028] The main components of mould powder usually are SiO2and CaO. A mould powder composition is characterised by its basicity. The basicity may be calculated by the ratio in weight percent of CaO to SiO2. Other possible components are : A12O2, Na2O, MgO, B2O2, Fe2O3, TiO2, and ICO. Further components may be network formers and network breakers, which introduce or prohibit the formation of Silica networks. The amount and type of the Silica networks influence the liquid mould slag viscosity and its tendencies to crystallise .

[0029] The invention has recognised that mould powders play a crucial role in the stability, quality and safety of a casting process. Therefore, the computer-implemented method of the invention provides at least one control model configured to provide control data for the continuous casting facility based on the temperature sensor data, wherein the controlled model is configured to compute at least one key parameter indicator, indicating the behaviour ofthe mould powder and / or the mould slag which is derived from the mould powder during the casting process. This key parameter indicator (KPI) preferably describes the mould powder’s in-mould behaviour, especially its thermal behaviour. Preferably, the KPI characterises the interaction of the mould slag with the mould wall and / or the solidifying metal inside the mould during the casting process. This knowledge allows inter aha to better evaluate whether the mould powder used is the best choice for the metal alloy, in particular steel grade, which is being cast. This allows to adjust the mould powder composition for a particular metal chemical composition even during a casting process. In the prior art, usually a change in the cast steel grade required a few weeks of trial and error experiments to find the right mould powder.

[0030] According to the invention, the temperature data from the mould wall is preferably used to assess and monitor the thermophysical state of the mould slag, in particular the state ofthe mould slag in the gap between the mould and the solidifying metal shell. The mould powder should be designed to perform at its optimum when is in the molten phase and is distributed in the gap between the growing steel shell and the mould. At that point it is called mould slag film and has two main tasks: (1) it controls the heat transfer from steel to mould, (2) it lubricates the steel being cast and extracted. If one of those tasks is not well done, it happens the following: for (1). If task (1) is not well done, i.e . the heat is not evenly / homogeneously extracted, the steel shell will grow unevenly, and this will lead to local stresses and consequently the formation of defects / cracks. Regarding task (2), if the lubrication is not good, then the steel shell sticks to the mould and cause a rupture that may end up in a breakout. Therefore, having control of these tasks through the KPIs is highly advantageous to ensure a stable casting process.

[0031] According to an embodiment, at least one key parameter index is computed by analysing the temperature distribution in the at least one mould wall in the feed direction and a direction perpendicular thereto and / or by analysing the distribution of the extracted heat over the at least one mould wall. The invention makes use of the two-dimensional nature ofthe temperature distribution which can be provided by novel temperature sensors, wherein e .g. between 200 and 5000, in particular more than 1000 sensors, may be distributed across a mould wall. The temperature sensors are in particular an FBG sensor system. The invention provides novel KPIs which provide insight into the behaviour of the mould powder / slag and can therefore be used to provide novel control data, in particular control data relating to the mould powder, such as a recommended change in mould powder. However, also novel control data relating more generally to the stability ofthe casting process can be provided.

[0032] According to an embodiment, the at least one key parameter indicator comprises one or more of:

[0033] (a) a COG indicator, the COG indicator being indicative of a shift in a centre of gravity of a heat transfer distribution across the at least one mould wall (BFF, BFL, NFS, NFN) from a target position,

[0034] (b) a heat transfer variation value, which is indicative of a variation of the total transferred heat along a direction perpendicular to the feed direction, wherein the total transferred heat describes the total amount of heat extracted from a point on the strand of metal, and / or (c) evaluations of breaking point positions, which are the positions at which a layer of resolidified mould powder breaks, in particular an upper breaking point position with respect to a meniscus level of the mould (4), an amplitude and / or a frequency of the breaking point positions, and / or a measure of regularity of the oscillations in the breaking point positions.

[0035] On the basis of (a), the COG indicator, in particular a shift of the centre of gravity of the heat transfer distribution in the feed direction, control data may be generated / provided which comprise a recommended change in the mould powder composition and / or a compatibility indicator which is indicative of the compatibility of the compositions of the cast metal and the mould powder. Further, on the basis of the COG indicator, in particular a value related to a shift in the centre of gravity in the horizontal direction, control data may be provided which comprise settings for an electromagnetic brake assembly of the mould and / or an Argon flow into the mould. This is explained herein below in more detail.

[0036] On the basis of (b), a heat transfer variation value, control data may be provided which comprise settings for an electromagnetic brake assembly of the mould and / or an Argon flow into the mould. How to compute the heat transfer variation value is explained herein below.

[0037] On the basis of (c), evaluations of breaking point positions, control data may be provided which comprises a compatibility indicator, which is indicative of the compatibility of the compositions of the cast metal and the mould powder, and / or a parameter indicating a recommended change in the mould powder composition. How to compute and evaluate the breaking point positions, i.e . the positions at which a layer of resolidified mould powder breaks, are explained herein below.

[0038] The continuous casting process of the invention preferably does not involve heating the mold powder by radiating microwaves generated by a microwave oscillator to an upper space of the supplied mould powder.

[0039] The KPI provided by the control model may be used to provide control data for the continuous casting facility. The term “control data” as used herein is understood broadly in the present disclosure and comprises any data or information that is suitable or useful in controlling a continuous casting facility. It may be a piece of information which is provided to a user and allows the user to manually inf ience / control the continuous casting facility, for example by adjusting a mould powder composition. The control data may also comprise data indicating whether the casting process is stable or unstable . Such data may be used to trigger a warning signal to the user, for example an acoustic or visual signal, when the casting process is unstable . The control data may also comprise data which may directly be used to control the continuous casting process, for example by adjusting one or more settings of the casting process, as described herein below.

[0040] The term “data” as used herein is to be understood broadly and represents any kind of information or digital data. Data may be single numbers / numerical values, a plurality of numbers / numerical values, a plurality of numbers / numerical values being arranged within a list, a two-dimensional or a three-dimensional array, but are not limited thereto. Data may also comprise words or control signals, such as acoustic or visual signals.

[0041] The term “control model” as used herein is to be understood broadly and preferably comprises an algorithm for processing the temperature sensor data and optionally the casting parameters as input data. The output data ofthe control model is control data for the continuous casting facility, which is computed based on at least one KPI. The control model is configured to analyse the temperature sensor data and optionally the casting parameters, wherein one or several KPI is calculated. The KPIs described herein are useful in characterizing the casting process. The control model may further be configured to visualize the temperature data, and / or the analysed temperature data. The visualisation may be presented to a user e .g. on a computer screen, and may contain the control data.

[0042] The algorithm may comprise decision trees, various computations such as algebra, differentiation, neural networks, lineal regression, but not limited thereto. The algorithm may be a trained algorithm, in particular a machine learning algorithm. Such algorithm may be trained using records of training data. A record of training data comprises training input data, namely temperature sensor data and optionally casting parameters, and training output data, which is in particular at least one KPI which has been determined from the input data under user-supervision. The deviation between this expected result and the actual result produced by the algorithm is observed and rated by means of a loss function. The loss function may be used as feedback for adjusting the parameters of the internal processing chain or neural network of the trained algorithm. For example, the parameters may be adjusted with the goal of minimising the values of the loss function. The control model may comprise several different controls routines of sub-models, in particular for providing more than one KPI.

[0043] The term “key parameter indicator” as used herein is to understood broadly and comprises any data that is suitable for characterising the behaviour of the mould powder or mould slag, or other characteristics of the casting, during the casting process. The data may be a single number or numerical value or a plurality of numbers / num eric al values.

[0044] The computer-implemented method is preferably configured for providing the control data in “real time” during a casting process, wherein the term “real time” as used herein is to be understood broadly. It means that the control data is to be provided continuously during the casting process and at a time interval after the temperature has been acquired, which still allows timely intervention to the casting process, for example in case of it being instable. The temperature data from the plurality of temperature sensors is preferably acquired at a rate of 0.2-20s-1, preferably 0.2-ls1. Preferably, the control model is configured to provide control data at the same rate . The control model is preferably configured to provide the control data within 5s, preferably within Is, more preferred within 0.5s of receiving the temperature data.

[0045] The term “casting parameters” is to be understood broadly in the present case and comprises any settings and measurement data relating to the casting process. In particular, the casting parameters may comprise the casting speed, e .g. the speed at which the metal strand is extracted from the base of the mould, the casting width, e .g. the width of the cast strand, the dimension of the mould, in particular the dimensions of its faces, the mould cooling water temperature before and after flowing through the mould, the cooling water flow rate, the cooling water pressure . Some ofthe parameters are measured in real time during the casting process and may be used as input data to the control model.

[0046] According to an embodiment, the casting parameters and / or the control data comprise a parameter relating to the chemical composition of the mould powder used in the casting process, and optionally a parameter relating to the physical properties of the mould powder or mould slag. Preferably, the parameter is or comprises the basicity of the mould powder. The parameter may also comprise several parameters, in particular further data relating to the composition of the mould powder and / or its interaction with the chemistry of the steel through solidification of the metal. For example, it may relate to the minerals contained therein in and / or to the network formers and / or network breakers. The parameter relating to the physical properties of the mould powder may be or may be related to the density or the viscosity of the mould powder, which may influence the interaction between steel and mould slag, and thereby influence the steel solidification behaviour. Preferably, the control data comprises such parameter(s), and preferably this allows to adjust the mould powders composition during the casting process. Since the mould powder directly influences the heat extraction process in the mould, it is advantageous to adjust the mould powder composition to the composition of the cast metal, in particular the steel grade . This is because different steel grades require faster or slower heat extraction. However, with the inventive method, it is also possible to adjust the mould powder composition to other casting parameter than the grade, for example the casting speed. Thus, the control data may comprise a parameter indicating a target composition, for example a target basicity of the mould powder. It may also comprise a parameter indicating whether the basicity should be lowered or increased.

[0047] According to an embodiment, the control model configured to compute a KPI characterising the behaviour ofthe mould powder also receives a parameter relating to the chemical position and optionally the physical properties of the mould powder or mould slag as input, as part of the casting parameters. Thereby, the control model is able to relate the temperature data and any analysis thereof to the properties of the mould powder which is currently in use . According to an embodiment, the casting parameters comprise at least one of the following:

[0048] • a parameter relating to the composition ofthe cast metal;

[0049] • a parameter relating to the casting speed ofthe casting process;

[0050] • at least one parameter relating to the dimensions ofthe mould, in particular the mould width;

[0051] • a parameter relating to a metal flow within the mould, in particular a parameter relating to an electromagnetic brake assembly of the mould and / or a parameter relating to an Argon flow into the mould;

[0052] • a parameter relating to a taper setting ofthe mould;

[0053] • measurement data relating to the cooling water flow in the mould; and

[0054] • a parameter relating to the chemical composition and / or the physical properties ofthe mould powder.

[0055] The parameter relating to the composition ofthe cast metal preferably comprises data relating to the carbon equivalent of a cast steel, in particular the carbon equivalent calculated according to the article by Kenneth Blazek et al. “Calculation of the Peritectic Range for steel alloys” presented on AlSTech 2007. This reference presents formulae based on regression analysis to calculate the carbon equivalent of a steel from its composition, in particular the content of various alloying elements including Al, Mn, Si, Ni, Mo, V, Cr and W. The amount of carbon in iron or steel can affect its strength and brittleness, as well as the way it is processed and welded. However, carbon is not the only alloying element; other elements also contribute to the material properties. Therefore, the carbon equivalent concept is used to "convert" all alloying elements to carbon equivalent percentages. The carbon equivalent of steel is the conversion ofthe content of alloying elements in steel, with carbon included, which has an effect on hardening, cold cracking and embrittlement. The parameter relating to the composition of the cast metal may also comprise several parameters describing the content of various alloying elements in the metal, e .g. in wt.%. In further embodiments, the parameter relating to the composition ofthe cast metal may comprise a parameter related to the phase transformations occurring during solidification of the cast steel (from ferrite to austenite, etc). The parameter relating to the casting speed is preferably the extraction speed of the metal strand from the base of the mould, given for example in m / min (metre per minute) or ipm (inch per minute).

[0056] The parameter relating to the dimensions of the mould preferably comprises the mould width, which indicates the distance between the two broad faces of a rectangular mould. The parameter may further comprise the mould length, which is the distance of the two narrow faces of a rectangular mould. A mould often comprises four faces being arranged in an at least approximately rectangular shape . The two large faces are herein referred to as broad face loose (BFL) and broad face fixed (BFF), because one of them is loose and may be adjusted in order to adjust the mould width. The two small faces are herein referred to as narrow face south (NFS) and narrow face north (NFN). Herein, the terms “mould wall” and “mould face” and “mould side” are used interchangeably. Preferably, at least the broad faces each comprise a plurality of temperature sensors distributed across the respective mould face . More preferred, each ofthe mould faces forming the mould comprise a plurality oftemperature sensors distributed across the mould face . Thereby, it is possible to analyse the heat transfer across each mould face separately. Further, by collecting high -re solution temperature data from the complete circumference of the mould, it is possible to gain information also about the flow of molten metal and the behaviour ofthe mould slag.

[0057] The parameter relating to a taper setting ofthe mould is preferably a parameter describing an inclination of at least one mould wall. Since the metal shrinks as it solidifies, it is possible to incline the mould faces, in particular the narrow faces, so that the shrinking metal strand does not lose contact with the mould walls. Since the mould walls are being actively cooled, it is essential that the strand of metal is in good thermal contact with the mould wall over the entire length of the mould, since otherwise heat cannot be extracted and the above difficulties relating to an insufficiently solidified metal shell may result. The parameter relating to a taper setting may for example be an inclination angle .

[0058] A parameter relating to a metal flow within the mould is in particular a parameter describing a setting of the equipment regulating the metal flow. Preferably, this is a parameter relating to an electromagnetic brake assembly of the mould and / or a parameter relating to an Argon flow into the mould. Electromagnetic brakes are typically installed at the broad faces of the mould. By generating a strong magnetic field inside the mould, the turbulent flow of liquid metal therein is slowed down and thereby controlled to an acceptable level. The parameter may also relate to an Argon flow into the mould. Argon is introduced at the SEN together with the liquid metal, in order to “blow out” any inclusions. By adjusting the Argon flow, the flow behaviour of the liquid metal inside the mould can also be regulated.

[0059] The casting parameters may further comprise measurement data relating to the cooling water flow in the mould, in particular the mould cooling water temperature at the mould inlet and outlet. From the difference, the extracted heat can be inferred. Further casting parameters comprise settings of measurement data relating to the cooling water flow in the mould, such as the water flow rate and the water pressure . These parameters may be used as input to the control model.

[0060] The above parameters may be used as input to the control model. Some or all of them may also be the output of the control model, thus they may be comprised in the control data. In particular, the control data preferably comprises at least one of the following:

[0061] • a parameter indicating whether the casting process is stable;

[0062] • a parameter relating to the casting speed ofthe casting process;

[0063] • a parameter relating to the casting width;

[0064] • a parameter relating to a metal flow within the mould, in particular a parameter relating to an electromagnetic brake assembly of the mould and / or a parameter relating to an Argon flow into the mould;

[0065] • a parameter relating to a taper setting ofthe mould;

[0066] • a parameter relating to the chemical composition and / or the physical properties of the mould powder; and

[0067] • a compatibility indicator.

[0068] In particular, the control data may comprise one or more of the casting parameters, for example a target setting of a casting parameter, or an amount by which a casting parameter needs to be changed, or the direction of change, i.e . whether a casting parameter needs to be increased or decreased. For example, the control data may comprise information that the casting speed should be increased or decreased, and optionally by how much.

[0069] The control data may further comprise a parameter relating to the metal flow within the mould. The motion steel flow in the mould is very important for the casting process, and thus it is important to characterise it, e .g. by means of KPIs as described herein. The flow of molten steel may be dynamically controlled by Argon, magnetic brakes and / or the casting speed or throughput through the mould. Thus, the control data may comprise a parameter indicating that the intensity of the electromagnetic brake should be increased or decreased, in particular by increasing or decreasing the current through the electromagnetic brake, and optionally by how much. It may also comprise a parameter indicating that Argon flow should be changed, an optionally by how much. Further, the metal flow is affected by the shape / design of the submerged- entry nozzle (SEN). Thus, the control data may in some embodiments comprise a parameter relating to the shape ofthe SEN, and may thus include recommendations to change the SEN.

[0070] The control data may also comprise a parameter indicating a target taper setting, or an indication that the inclination angle of the taper setting should be increased or decreased, and optionally by how much.

[0071] Moreover, the control data may comprise a parameter indicating whether the casting process is stable . This is for example a value indicating the degree of stability. It may have only two values, thus either indicating stability or non-stability. The parameter may have one of more than two values, thereby indicating several degrees of stability or instability. The parameter may also comprise or trigger a warning signal, alerting the user that the process is unstable .

[0072] According to an embodiment, the control data comprises a compatibility indicator, which is indicative ofthe compatibility ofthe compositions ofthe cast metal and the mould powder. Optionally, the control data further comprises a parameter indicating a recommended change in the mould powder composition, wherein this parameter may be the compatibility indicator. As will be described herein below, the invention provides methods of analysing the temperature sensor data so as to infer whether the heat transfer in the mould is compatible with the requirements of the steel grade which is being cast. In particular, the evaluations ofthe breaking point positions, which are the positions at which the resolidified slag layer brakes, allow to infer valuable information on the homogeneity and stability of the solidification process inside the mould. These are processes which are otherwise completely invisible and very hard to observe and control. Thereby, the invention provides useful methods to analyse the high-resolution temperature data from the mould and thereby control the casting process.

[0073] According to an embodiment, the control model is configured to compute at least one key parameter indicator which characterises the stability of the casting process. Example of such KPIs will be given below.

[0074] According to an embodiment, the at least one key parameter indicator comprises a COG indicator, the COG indicator being indicative of a shift in a centre of gravity of a heat transfer distribution across the at least one mould wall from a target position, wherein the COG indicator may be indicative of a shift ofthe centre of gravity (COG) in a feed direction ofthe metal and / or in a direction perpendicular thereto. The COG shift is indicative not only of a shift on the feeding, but also of a shift in the homogeneity of the heat extraction. This means that a shift in the COG of heat transfer is due to the feeding, and / or the electromagnetic brakes, and / or Argon, but could be also inherent to the mould powder / slag properties i.e . viscosity, crystallinity ratio, etc. In other words, the COG indicator may be indicative of a shift in the heat transfer, which is influenced by feed direction or by mould flow e .g. magnetic brakes, Argon and / or symmetry.

[0075] The COG indicator may be calculated for one or several mould walls separately. The COG indicator is calculated by dividing the mould wall into sub-parts, wherein there are preferably as many sub-parts as temperature sensors. It is assumed that the temperature data from that sensor measurement is representative of its sub-area’s temperature . Further, the cooling water temperature Tcwin each sub-part is estimated, in particular from the measurements ofthe cooling water temperature at the entry and the exit from the mould. For example, the water temperatures at each position may be calculated by a linear interpolation from the entry and exit temperatures. To compute the COG indicator, the total extracted heat Qtota! for each mould face is calculated from the cooling water temperatures at the inlet and outlet, TWlCl t;and TWiinand the mass flow rate mi of the cooling water according to: where cpis the water’s specific heat at the given temperature . The heat extracted by the individual sub-areas is proportional to their areas A; and to the driving force for heat transfer (AT), therefore : where Qj is the individual part’s extracted heat, A is the area and Ti is where TFBG is the temperature measured by the FBG sensor, Tcwis the cooling water temperature at the sub-part I, and dcuis the thickness ofthe copper mould.

[0076] It is thereby possible to calculate the heat extracted by each individual sub-area in one or several or all mould walls. By multiplying each individual heat extraction value qi with its distance vector from the centre and summing up, the centre ofgravity (COG) of the heat transfer is calculated. It is a position value which shows by how much the middle point of the heat transfer has shifted compared to an ideal scenario, where the heat transfer is uniform across the whole mould. Thus, the target position(s) of the COG indicator(s) indicate the ideal and most homogenous situation for the particular steel, with a compatible mould powder and for certain casting parameters.

[0077] The COG indicator preferably comprises several parameters, in particular one parameter indicating the shift in the feed direction, which is also referred to as vertical direction, and one parameter indicating a shift in a direction perpendicular thereto, which is also referred to as the horizontal direction. It may further be computed for each mould side . Preferably, it is computed at least for the broad sides. If the temperature profile is moving and thus the COG indicator shifts, the casting process may be going out of balance and thus deviating towards an unstable situation. For example, if the COG has shifted towards the top part of the mould, this means that the mould slag extracts more heat in the top part of the mould than in the bottom. In this case, the mould powder is suitable for low carbon steels, which can handle high thermal stresses and therefore high cooling rates without cracking. On the other hand, for casting peritectic steels, it is required that the COG of the heat transfer is further in the bottom of the mould, since peritectic steels require a more even heat extraction to avoid cracking and other quality issues. Therefore, the target position of the centre of gravity will be further downwards. Further, allowed variations of the centre of gravity of the heat transfer distribution may be smaller than for other steel grades. Thus, control data which may be generated on the basis of the COG indicator, in particular on the basis of a shift of the centre of gravity of the heat transfer distribution in the feed direction, comprise a recommended change in the mould powder composition and / or a compatibility indicator which is indicative of the compatibility of the compositions of the cast metal and the mould powder.

[0078] According to an embodiment, the control data comprises a symmetry value which is indicative of the symmetry of the flow of molten metal inside the mould, which is computed based on the at least one COG indicator. Optionally, the control data further comprises settings for an electromagnetic brake assembly of the mould and / or an Argon flow into the mould. It has been found that a shift in the centre of gravity, in particular in the horizontal direction, indicates a non-symmetric fluid flow in the mould. This can be controlled by controlling the settings for the electromagnetic (EM) brake and optionally the Argon flow into the mould. Further, according to an embodiment, the at least one KPI comprises a COG indicator which is indicative of the symmetry of the flow of molten metal inside the mould. In particular, the COG indicator may be or may comprise a value related to a shift in the centre of gravity in the horizontal direction. On the basis of this KPI, control data may be provided which comprise settings for an electromagnetic brake assembly ofthe mould and / or an Argon flow into the mould.

[0079] It has further been found that the vertical COG shift in the different mould faces may give a valuable indication of the behaviour of the mould powder. In particular, it has been found in experiments that the COG shift in the vertical direction may differ between the narrow faces and the broad faces. The vertical shift generally behaves in the same way on the same-sized mould walls. However, there is a growing difference between the broad faces and the narrow faces as the carbon content ofthe steel grade is increased. This is due to the increasing thermal expansion coefficient. The steel is continuously cooling down as it descends in the mould, thus it is shrinking. The shrinking is compensated by the narrow face ’s taper setting, but as the thermal expansion coefficient grows, it is increasingly difficult to accurately tune the taper setting. With growing taper error, an air gap also grows at the lower parts ofthe narrow faces, thus leading to a lower heat extraction in this region. As a result of lower heat extraction in the lower regions ofthe mould, the centre of gravity ofthe heat extraction shift upwards, because relatively more heat is extracted in the upper regions of the mould. Thus, the COG indicator may comprise a parameter relating to the taper setting. By observing the differences in the vertical COG shift between broad and narrow faces, the taper setting may be adjusted very finely. In particular, the control data which is based on the COG indicator, in particular the difference in COG vertical shift between broad and narrow faces, may be a taper setting.

[0080] Moreover, the vertical COG shift is also indicative for the cooling power of the mould slag. Thus, when comparing the average vertical shifts in the centre of gravity between different mould powders, an overall shift towards lower values means that each powder is cooling the steel slower than the previous one . Thereby, the COG indicator can be used to compute control data relating to the composition ofthe mould powder. For example, if the centre of gravity of heat transfer shifts to higher regions in the mould, the control data may comprise a suggestion to change the composition of the mould powder. In particular, the mould slag should be cooling the steel slower, which may be achieved by increasing the mould powder basicity.

[0081] According to an embodiment, the control model is configured to compute the total transferred heat at a plurality of positions along a line perpendicular to the feed direction, wherein the at least one KPI optionally comprises a QT variation value which is indicative of a variation ofthe total transferred heat along a direction perpendicular to the feed direction. Alternatively or additionally, the at least one KPI comprises a QT symmetry value, which is indicative ofthe symmetry ofthe total transferred heat along a direction perpendicular to the feed direction. The total transferred heat of a point on metal strand is calculated by following the point as it descends down the mould, and summing the extracted heat along the corresponding vertical line, as explained below. The QT variation value is useful because it represents the evenness or stability of the process: If the QT values are the same or similar in the horizontal direction, it means that the heat extracted along the perimeter is even or at least within similar values, which will lead to an even / homogeneous growth of the steel shell. By contrast, if the steel shell does grow uneven, this will lead to local stresses and may lead the the formation of cracks (defects). Therefore, control data may be provided on the basis of this KPI which aims at correcting the unevenness. For example, the provided control data which comprise settings for an electromagnetic brake assembly of the mould and / or an Argon flow into the mould.

[0082] While the COG indicator describes the mould slag’s behaviour from the mould’s point of view, the total transferred heat describes the total amount of heat extracted from each point on the steel shell. The total transferred heat of a point or sub-area on the cast steel slab is calculated by following the point or sub-area as is descends down the mould, and summing the extracted heat along the corresponding line in feed direction. The extracted heat may be calculated at each point or sub-area according to formulas (2) and (3). The total transferred heat (QT) is calculated by following a horizontal line as it descends in the mould with the casting speed, for each point along the line, calculating the overall heat extraction it experiences in that time period. In a first step, the vertical positions of the horizontal line at certain time points is calculated. Preferably, a time step equivalent to the data collection rate of the temperature sensor system is chosen, for example 0.1 -Is . The temperature at each line position is either directly given from the temperature sensor data, or it may be computed from the nearest neighbour temperature sensors via interpolation, for example linear interpolation. The extracted heat at each given point or sub-area is calculated in the same way as described above, summing these values along a vertical line . The total heat that got extracted through the mould wall is calculated at each point of a given horizontal line on the steel shell’s surface .

[0083] According to an embodiment, the temperature data acquired above the meniscus line are excluded from the calculations. However, for the QT variation and QT symmetry, this is not so relevant because the temperature is relatively low in these areas. Thus, according to an alternative embodiment, the temperature data acquired above the meniscus line are not excluded from the calculations

[0084] When looking at the horizontal lines of heat transfer, there may be a considerable variation in between individual values along the horizontal line . The variation may for example be calculated as the standard deviation, but can also be the variance or any other value characterizing the amount of deviations from a mean value . A high standard deviation indicates that the mould powder’s liquid layer (mould slag) is more likely to locally thicken or thin out. Thus, the QT variation value may be used to provide a control data relating to the mould powder composition. In particular, it has been found that mould powders with a fast cooling powder (low basicity) have a higher QT variation than a mild cooling powder or slow cooling powder having a higher basicity value . Thus, the KPI may comprise a QT variation value . On the basis of this KPI, control data relating to the mould powder composition may be provided.

[0085] According to an embodiment, the at least one key parameter indicator comprises the overall total transferred heat, summed over all positions along a line perpendicular to the feed direction, wherein the overall total transferred heat is in particular the basis of control data which comprise settings of one or more casting parameters, in particular casting speed, and / or mould powder composition. When looking at the overall total transferred heat QT, for example the average total transferred heat along each line, it has been found that the total extracted heat decreases as the carbon content of the steel increases. Thus, a target total extracted heat value can be defined to avoid thermal stresses which would cause cracking. The total transferred heat, summed over all positions along a line perpendicular to the feed direction, may thus be the basis of a control data which comprises settings of one or more casting parameters, in particular casting speed, mould powder composition, etc.. Thus, the KPI may comprise the overall total transferred heat QT, either for one or several mould faces, or the average total transferred heat in the mould.

[0086] According to an embodiment, the at least one control model is configured to compute a breaking point position, which is indicative of a position where a layer of solidified mould slag close to the mould wall breaks, and wherein the control model is preferably configured to compute, from a time series of breaking point positions, an upper breaking point position with respect to a meniscus level of the mould, an amplitude and / or a frequency of the breaking point positions, and / or a measure of regularity of the oscillations in the breaking point positions. Thus, the at least one KPI may comprise one or all of these values (upper breaking point position, amplitude, frequency or measure of regularity).

[0087] The breaking point position is computed by looking at the temperature distribution in the feed direction, in particular a vertical direction. Preferably, in order to make further evaluations, the breaking point position is compared, assessed and evaluated together with the points in the surroundings, thus also in the horizontal direction.

[0088] Up to now, the breaking point position could only be estimated through simulations or very rough estimations based on the “dips” in the thermocouple readings. However, these estimations are very coarse due to the delay in the thermocouple signals and the fact that only very few thermocouples are available . The invention now allows to detect the breaking point positions with much higher accuracy from the temperature measurements. Generally, the slag resolidifies at the mould wall and forms a slag layer. The composition of the resolidified slag is not necessarily the same as that of the mould powder, since some components leave the system during melting, like carbon that burns and forms CO / CO2, or fluor that evaporates (sublimates), or in some cases the liquid flux / slag reacts with the steel alloys and picks up some ofthe elements, i.e . Al from the steel can go to the flux and form AL2O3. The solidified slag layer regularly breaks and the broken-off part moves downwards. This will influence the heat transfer at the breaking point, because first, there is a vacuum or air gap, which is then filled up by liquid slag. Therefore, the local heat transfer at the breaking point will differ from the surrounding points’ heat transfer because of the change in slag layer thickness and composition (first air then liquid slag). The liquid slag has higher heat transfer coefficients than solidified slag, and will thus result in a local heat transfer increase . As this liquid layer resolidifies, its local heat transfer approaches the surrounding points again.

[0089] In order to find the breaking points, the temperature distribution along the feed direction (vertical direction) has been analysed. According to an embodiment, the control model is configured to compute the breaking point position by analysing a temperature distribution in the mould wall in the feed direction and detecting inflection points in the respective temperature curve . The inflection points indicate a change in local heat transfer. In an embodiment, the temperature curve is analysed by taking the first and second derivative with respect to the vertical direction, to see how the heat transfer rate changes. In other words, the temperature curves are differentiated with respect to the vertical direction (height). The derivatives may be calculated along every FBG sensor line by using a numerical approximation. The inflection points may be identified as zero crossings from positive to negative in the second derivative . However, there may be several points which satisfy this criterium. Therefore, in a useful embodiment, two criteria are used to label a point as a possible breaking point. Firstly, only points that are well below the meniscus line are considered, wherein a predetermined threshold value is preferably used. Secondly, there needs to be a change in the local heat transfer, which could be a local maximum or a plateau in the first derivative . For example, this may be detected by a zero crossing of the second derivative . Plateau points are included because, although the breaking will increase the local heat transfer, resulting in a local maximum, this increase might not be enough to compensate for other effects like the heat transfer decrease due to the growing steel shell thickness or the growing air gap.

[0090] The mathematical criteria for the i-th position to be labelled as a plateau point is: where Tj is the temperature change at the i-th position and C&Tis a correction constant. This correction is preferred in order to take into account the noise and reduce the number of possible breaking point positions. In the example, the value used was between 1.2 and 2.5, in particular 1.8. After calculating these possible breaking point positions, they are saved over time and the time series is further analysed. It is preferable to calculate the relative distance of the breaking point to the meniscus line, since the meniscus oscillates up and down. It has been found that several parameters of the time series of breaking point positions are of interest and characterize the stability of the casting and / or the mould slag film. Each ofthese may be comprised in the KPI according to the invention:

[0091] First of all, the upper breaking point position with respect to a meniscus level of the mould indicates the start ofthe breaking positions. This position is indicative ofthe casting speed and the mould powder composition. It has in particular been shown that the upper breaking point position is different for different mould powders.

[0092] Further, the amplitude ofthe oscillations in the breaking point positions varies also with the casting speed. The amplitude is the position difference between the upper breaking point position and the lowermost breaking point position of a single breaking event. With increasing casting speed, the mould slag layer thickness decreases. A thinner layer solidifies faster, therefore it breaks again sooner, leading to a narrower breaking region. Preferably, an average ofthe upper breaking point position as well as the amplitude may be calculated, for example the average ofthe past 0.5-10 minutes.

[0093] A further parameter which may be comprised in the KPI is the number of breaking points detected at the same time . In some castings, two or more breaking points are detected at the same time . This may be the result of many mould phenomena happening in the same time in an unstable way. Thus, the detection of two or more breaking point positions at the same time may result in control data indicating an unstable casting process.

[0094] A further parameter which may be comprised in the KPI is the frequency of oscillation ofthe breaking point positions. This is indicative both ofthe casting speed, but also the interaction of mould slag and steel grade . The frequency is influenced by the casting speed because, at higher speeds, the broken-off mould slag piece slides more on the steel surface, since the static coefficient of friction is larger than the kinetic coefficient of friction.

[0095] Thus, the breaking point calculations show that casting parameters like casting speed and material properties like mould powder composition, effect the breaking phenomena. Therefore, they may be used to control the casting process, and in particular to control the mould powder composition. Therefore, the upper breaking point position, an amplitude and / or a frequency of the breaking point positions may be comprised in the KPI characterising the behaviour of the mould powder.

[0096] The invention shall now be described by means of embodiments with respect to the attached drawings. In the drawings:

[0097] Fig. 1 shows a schematic cross-section through a continuous casting facility;

[0098] Fig. 2 shows a schematic longitudinal section through a mould;

[0099] Fig. 3 schematically shows a mould;

[0100] Fig. 4 schematically illustrates a single mould face and a grid ofFBG sensors;

[0101] Fig. 5 shows measurement results ofthe COG shift in the vertical direction;

[0102] Fig. 6 shows measurement results of the COG shift in the horizontal direction for various steel grade / mould powder combinations;

[0103] Fig. 7 shows the COG shifts on the four mould faces in the vertical and horizontal directions for various steel grade / mould powder combinations;

[0104] Fig. 8 shows the total extracted heat calculated for the broad faces and narrow faces;

[0105] Fig. 9 shows a graph ofthe standard deviation of QT for various steel grade / mould powder combinations;

[0106] Fig. 10 shows the total extracted heat on a broad face for various steel grades;

[0107] Fig. 11 shows temperature curves along the vertical direction (left) as well as the first and second derivative thereof (centre) and the points detected therefrom for the meniscus, the face change, and the slag layer breaking points;

[0108] Fig. 12 shows temperature curves along the vertical direction (left) as well as the first and second derivative thereof (centre) and the points detected therefrom for the meniscus, the face change, and the slag layer breaking points;

[0109] Fig. 13 shows breaking point positions for the same steel grade / mould powder combination, but at different casting speeds;

[0110] Fig. 14 shows a time series of breaking point positions for various steel grade / mould powder combinations; Fig. 15 shows a flow diagram of a method according to an embodiment of the invention;

[0111] Fig. 16 illustrates a system according to an embodiment ofthe invention.

[0112] The same elements are designated with the same reference signs in the drawings.

[0113] Fig. 1 illustrates the general layout of a continuous casting facility 1, also referred to as caster, which may be controlled by control data generated according to an embodiment of the inventive method. The casting facility 1 comprises a tundish 3 which is fed with molten steel 10 from a ladle 2. From the tundish 3 the molten steel 10 is fed into the mould 4 through a submerged entry nozzle (SEN) 5. The vertical or feeding direction of the steel in the mould 4 is indicated at by arrow z. Both in the tundish 3 and in the mould 4, the top ofthe liquid steel is covered by a casting powder 20, which is continuously replenished. The walls of the mould 4 are water cooled, so that the steel starts solidifying below the meniscus level 6 of the mould, thereby forming a solidifying shell 12. The mould oscillates vertically during the casting process because this prevents the solidifying shell 12 sticking to the mould, and also helps the liquid mould slag 24 (mould powder) to penetrate into the shell-mould gap. This is essential for the process, because the mould slag provides lubrication and controls the heat transfer between the shell 12 and the mould 4. At the base of the mould, the strand 14 of metal, still having a liquid core 11, is extracted and bent into the horizontal by means of support rolls 7. At the same time, the strand 14 is further cooled by means of spray cooling 8. At the metallurgical length 15, the strand is completely solidified. At the torch cut-off point 16, the strand is cut into slabs 18, which may then be further processed, for example by hot-rolling.

[0114] Fig. 2 is a longitudinal section through one half of the mould 4 and shows the solidifying shell 12 ofthe steel and the mould powder 20 and mould slag 21, 24, 25 in more detail. The liquid steel enters through the SEN 5. In some casters, the liquid metals flow enters the mould through two slot in the SEN, only one of which is shown, and usually splits into four streams 32, forming a double-roll pattern. On each side of the SEN, one stream 32 is going upwards, the other stream 32 is going downwards. These flows can entrap droplets of the mould slag, leading to inclusions and quality issues. The solidification ofthe steel shell 12 starts in the mould at the meniscus line 6. From this point down, the goal is to get a steadily growing steel shell 12. The behaviour ofthe mould powder is essential in this process. It is fed into the caster in the region on top ofthe mould, where it forms a mould powder layer 20. Below that, it starts to melt, forming a mushy zone or sinterized layer 21. The sinterized layer 21 is a result ofthe different melting rates of the different mould powder constituents. A slag ring 22 is formed on the upper part ofthe mould wall. The slag ring 22 helps in pushing the liquid mould powder 24 into the mould-shell gap during the mould’s vertical oscillations. The liquid mould slag 24 solidifies in the mould -she 11 gap as it descends, forming a layer 25 of solidified mould powder. This solidified slag layer 25 will break off regularly, for example at position 30. As the resolidified slag layer 25 breaks, the broken piece sticks to the steel, which moves down forming a gap in place ofthe resolidified slag layer 25. This gap is first filled by air or vacuum. It is then filled up by the liquid mould slag 24. The liquid mould slag has a good contact with both the mould 4 and the solidifying steel shell 12, and a higher heat transfer coefficient than air or vacuum, therefore, it will result in a local increased heat transfer, which can be detected according to the inventive method. With time, the liquid layer solidifies again, after which breaking takes place again. While the slag layer is solidifying, the first derivative ’s local maximum position climbs up, because in the lower regions ofthe mould, the temperature is lower and thus the solidification takes place faster, as can be observed by the vertical temperature distribution in the mould wall.

[0115] Fig. 3 illustrates the dimensions of a typical mould 4 having an essentially rectangular shape . The mould 4 has four faces, wherein the liquid steel is fed from the top and extracted from the bottom of the page . The mould has two broad faces, wherein the face 4.2 depicted on the front is the broad face fixed (BFF) and the face 4.1 on the back is the broad face loose (BFL). The narrow faces are the face 4.3 on the left, the narrow face north (NFS), and the face 4.4 on the right, which is the narrow face north (NFN). The faces’ coordinate systems are also indicated, wherein one can see that opposing faces, for example the two broad faces 4.1, 4.2, have opposite signs in the x-direction.

[0116] Fig. 4 illustrates a plurality oftemperature sensors 43 on a mould wall 4.2. This is schematic and illustrates an embodiment of a layout of a fine grid 44 of FBG temperature sensors 43. Similar grids 44 are preferably present at the other mould faces 4.1, 4.3, 4.4. In the mould wall, the optical fibres 40 are placed horizontally, wherein each optical fibre comprises a number of FBG temperature sensors 43. Thus, each optical fibre 40 can acquire temperature data from a number of points 43 spread across the mould face 4.2. preferably, the plurality of temperature sensors 43 are distributed across the mould face in a regular pattern, for example in a rectangular grid 44.

[0117] Fig. 15 illustrates a computer-implemented method according to an embodiment of the invention. In step 80, temperature data from the plurality of temperature sensors 43 is provided. In step 82, casting parameters comprising settings and / or measurement data of the casting process are provided. In step 84, a control model is provided which is configured to compute at least one KPI characterising the behaviour of the mould powder during the casting process. In step 86, control data is provided, which is based on the at least one KPI.

[0118] Fig. 16 schematically illustrates a system 90 according to an embodiment ofthe invention. The system 90 comprises a storage medium 88, a processing circuitry 92 and a data interface 94. The storage medium 88 comprises a computer program which comprises instructions which causes the processing circuitry 92 to carry out the method according to an embodiment of the invention. The data interface 94 is configured to receive the temperature data 80, further casting parameters 82, and to output control data 86.

[0119] EXAMPLES

[0120] The invention will now be further described by means of examples. In the examples, the KPIs according to embodiments of the invention were computed from temperature data collected during several example casting processes. The casting facility was equipped with an FBG temperature sensor system, collecting high- resolution temperature data from the mould. In particular, the FBG sensor system comprised 2350 temperature sensors arranged in a fine grid in all four mould sides.

[0121] Four different steel grades were cast and combined with three different mould powders, as described in Tables 1 and 2 below. The parameter indicative ofthe steel composition is the carbon equivalent. The parameter indicative of the mould powder composition is the basicity of each mould powder.

[0122] Table 1 : Steel grades

[0123] Table 2: Mould Powders

[0124] Thereby, the behaviour of five mould powder - steel grade combinations was analysed, namely the combinations ULCl-a, ULC2-a, ULCl-b, LC-b and PERI-c.

[0125] Further, each casting was performed with three different casting speeds (m / min), as summarised in Table 3 below.

[0126] Table 3 : Casting speeds The following KPIs were calculated and are presented in Figs. 5-8:

[0127] The shift in the centre of gravity of the heat transfer, in the horizontal and vertical directions for each mould face,

[0128] The total transferred heat QT for each mould side,

[0129] The standard deviation ofthe total extracted heat, for each mould face,

[0130] The standard deviation ofthe total heat transfer for the BFF sides, for different steel - mould powder combinations;

[0131] The total extracted heat for different steel grades - mould powder combinations,

[0132] The first and second derivatives of vertical temperature curves, The vertical position ofthe ferrite to austenite transformation, Calculated from the vertical temperature cures and / or their derivatives, a time series of slag layer breaking point positions;

[0133] Calculated therefrom, the upper breaking point positions with respect to a meniscus level of the mould, the amplitude and the frequency of the breaking point positions, for various steel - mould powder combinations and casting speeds.

[0134] The vertical COG shift for each mould face and the different castings is depicted in Fig. 5. It is given as the average of the individual measurements, the bar indicating the standard deviation. The horizontal shifts are displayed in Fig. 6 As evident from Fig. 5, the broad face loose (BFL) and the broad face fixed (BFF) sides ofthe mould have very similar values, thus the mould powders behave in the same way on the same sized mould walls. This is also true for the two narrow faces NFS and NFN. There is a growing difference between the broad and narrow faces as the carbon content ofthe steel is increased. This is due to the increasing thermal expansion coefficient. The steel is continuously cooling down as it descends in the mould thus it is shrinking. This shrinkage is compensated by the narrow face ’s taper setting, but as the thermal expansion coefficient grows it is increasingly difficult to accurately tune . With growing taper error, the airgap also grows at the lower parts ofthe narrow faces thus leading to a lower heat extraction in this region. As a result of lower heat extraction in the lower regions, the COG ofthe heat extraction shifts upwards in the narrow faces. Thus, the difference in vertical shift between broad and narrow faces may be used to control the taper setting. Comparing the average COG values ofthe three different powders (a~8.6 b~7.1 c~6.4) it is visible that they have a continuously decreasing value which means that each powder is cooling the steel slower than the previous one . This trend coincides with the materials’ need (milder cooling-rate for the more crack susceptible grades).

[0135] The horizontal COG shift values depicted in Fig. 6 show that the faces on the opposite sides have opposite signed values which suggests that there is a non- symmetric fluid flow in the mould. However, the horizontal COG shift values are relatively small across all grades and mould powders, indicating a stable process. However, ifthe horizontal COG shift were higher, this would indicate an unstable flow.

[0136] Fig. 7 summarises again the COG shift for each ofthe four faces, in both vertical and horizontal direction. One can see that on the broad faces, where the shrinkage and taper settings do not influence the results substantially, powder (a) has the largest vertical COG shift followed by powder (b) and then powder (c). This shift is influenced by both the steel grade and the mould powder composition. Thus, the vertical COG shift may be used to control the mould powder composition, in order to reach a target vertical shift on these faces. Moreover, the horizontal shift in the COG may serve to indicate instability in the casting process.

[0137] Looking at the QT calculations in Fig. 8, it is visible that the same sized faces behave similarly. As the mould position approaches the sides, the QT values start to decrease . At the comers of the strand, the heat extraction value decreases because from this region not only one but two of the mould faces extract heat, thus the decrease .

[0138] In Fig. 9, it is seen that powder (a) with a fast-cooling powder has a higher standard QT variation than the mild cooling powder (b) or the slow cooling powder (c). This means that this powder’s liquid layer is more likely to locally thicken or thin out.

[0139] In Fig. 10, the QT values are calculated for the BFF sides for various steel grades.

[0140] One can see that the total extracted heat decreases as the carbon content ofthe steel increases. Fig. 11 shows on the left side in graph 50 the temperature distribution along a vertical line of a broad mould face (BFF) for the ULCl-a combination. The first derivative 52 and the second derivative 54 with respect to the vertical height have been taken and analysed. The idea behind this analysis is that a change in curvature of the temperature distribution indicates a relevant change in the heat transfer of the mould slag, which is very hard to observe from the temperature curve 50 per se, and which is interpreted by the inventive method as follows, and as indicated in the right graph 55 in Fig. 11 :

[0141] The meniscus line is depicted by a horizontal line 56. As can be seen in graph 52, the first derivatives have several local maximum and plateau points, not just one . This is because the mould powder breaking is not the only phenomena to cause a local heat transfer increase . Local thinning of the mould layer, extra superheat from the fluid flow or a phase transformation in the steel could also result in a local heat transfer increase . The ULC grades usually have 2-4, the LC steel grades usually have 3 - 4 and the PERI usually have 4 - 5 local maximum points in the first derivative .

[0142] The first maximum 58 in the first derivative is relative stable position-wise and is at approximately 170 mms below the meniscus. The stab ility of this line suggests that it is a representation of a steel phenomenon. It has been confirmed by FEM simulations that this line 58 represents the ferrite to austenite transformation (5 a transition). The position of this phase transformation can be used as a KPI and, on the basis of this KPI; casting parameters can be adjusted to reach a target phase transformation position.

[0143] The second maximum 60 in the first derivative is identified as a breaking point. It is at the expected mould region, and its fluctuations over time also support this. As explained above, the break in the resolidified slag layer will move up and down the vertical direction, reflecting the breaking / filling with air and liquid / downward moving processes, which are happening at the breaking points. In Fig. 11, the oscillations in the breaking point position are relatively regular, indicating a stable casting process.

[0144] Fig. 12, by contrast, illustrates the same graphs for the ULC2-b combination. Therein, significantly more possible breaking points 60a, 60b are detected than for other combinations. Examining the temperature curve and the derivatives on the left, one can see a complex temperature profile . This is the result as many mould phenomena happen at the same time in an unstable way. Thus, this steel grade - mould powder combination resulted in an unstable casting process, as indicated by the strong fluctuations in high number of breaking point positions, and a great amplitude as well as strong fluctuations in the breaking point frequency.

[0145] Fig. 13 illustrates the influence of casting speed on the breaking points for the ULC2-a combination. One can see that the 5 - a transition line 58 is stable, as this is mostly governed by the steel composition which is unchanged. However, there is a change in the start of the breaking positions, e . g. the position 62 of the uppermost breaking points. More importantly, there is a great difference between the amplitude and frequency in the breaking point positions. The breaking point’s amplitude is the position difference between the uppermost breaking point of a single breaking event, wherein an average thereof over a pre -determined time interval is computed by the control model. Also, the variation in the breaking point amplitude may serve as KPI. The difference in amplitude between two casting speeds in Fig. 13 (casting speed 1 = 1.1 m / min, casting speed 2 = 1.4 m / min) may be explained as follows. With increasing casting speed, the mould powder layer thickness decreases. Therefore, the layer solidifies faster, therefore it breaks again sooner, leading to a narrower breaking region, as indicated by the amplitude of the breaking point. In peritectic steels, the casting speed changes not only the amplitude and the position ofthe breaking points, but the frequency, too. This is because at higher speeds, the broken off mould slag piece slides more on the steel surface, due to the fact that the static coefficient of friction is larger than the kinetic coefficient of friction.

[0146] Fig. 14 depicts an overview of different steel grade - mould powder behaviours. This overview shows that the breaking points’ pattern is different from powder to powder. The breaking point calculations are supported by the production practices that with increasing carbon content, and therefore crack sensitivity, slower cooling powders are used. The breaking point’s position, amplitude and frequency vary according to several parameters. Both process parameters (like casting speed) and material properties (like mould powder composition) affect the breaking phenomena.

[0147] The invention has discovered that it is possible to formulate new ways to evaluate the mould powder behaviour in a continuous casting facility. In particular, several numerical values or KPIs have been derived. These are simplified numerical values to characterise a certain aspect of the solidification and heat extraction in the continuous casting process. The KPIs serve as the basis to compute control data for controlling and thereby improving the steel manufacturing process. These KPIs are particularly useful during production, to characterise the behaviour of the given mould powder in combination with a steel grade . A first KPI is the shift in COG (centre of gravity), which describes the distribution of heat transfer in the mould. A second KPI, the extracted heat, describes the heat extraction experienced by the steel shell. A last group of KPIs describe the mould slag breaking phenomena inferring the breaking point position. Any reference sign in the claims should not be construed as limiting the scope ofthe appended claims.

Claims

1. - 36 -CLAIMS1. Computer-implemented method for providing control data for a continuous casting facility (1) for casting a metal (10), in particular steel, in a continuous casting process, the continuous casting facility comprising a mould (4), wherein the mould (4) comprises a plurality of temperature sensors (43) distributed across at least one mould wall (BFF, BFL, NFS, NFN), wherein the continuous casting process comprises feeding liquid metal and a mould powder (20, 21, 24, 25) to the mould (4) and extracting a strand of metal from the base of the mould (4) at a casting speed, the method comprising the following steps:(a) providing temperature data (80) from the plurality of temperature sensors (43), the temperature data indicating a temperature distribution in the mould wall (BFF, BFL, NFS, NFN) during the casting process;(b) optionally, providing casting parameters (82) comprising settings and / or measurement data ofthe casting process;(c) providing at least one control model (84) configured to provide control data for the continuous casting facility based on the temperature data and optionally the casting parameters, wherein the control model is configured to compute at least one key parameter indicator characterizing the behaviour of the mould powder (20, 21, 24, 25) during the casting process; and(d) providing control data (86) for the continuous casting facility utilizing the control model, wherein the control data is based on the at least one key parameter indicator.

2. Computer-implemented method according to claim 1, wherein the at least one key parameter index is computed by analysing the temperature distribution in the at least one mould wall in the feed direction and a direction perpendicular thereto and / or by analysing the distribution of the extracted heat over the at least one mould wall.

3. Computer-implemented method according to claim 1 or 2, wherein the at least one key parameter indicator comprises one or more of:- 37 - a COG indicator, the COG indicator being indicative of a shift in a centre of gravity of a heat transfer distribution across the at least one mould wall (BFF, BFL, NFS, NFN) from a target position; a heat transfer variation value, which is indicative of a variation of the total transferred heat along a direction perpendicular to the feed direction, wherein the total transferred heat describes the total amount of heat extracted from a point on the strand of metal; and / or evaluations of breaking point positions, which are the positions at which a layer of resolidified mould powder breaks, in particular an upper breaking point position with respect to a meniscus level of the mould (4), an amplitude and / or a frequency of the breaking point positions, and / or a measure of regularity ofthe oscillations in the breaking point positions.

4. Computer-implemented method according to any one ofthe preceding claims1 to 3, wherein the casting parameters comprise at least one ofthe following:• a parameter relating to the composition ofthe cast metal;• a parameter relating to the casting speed ofthe casting process;• at least one parameter relating to the dimensions of the mould (4), in particular the mould width;• a parameter relating to a metal flow within the mould (4), in particular a parameter relating to an electromagnetic brake assembly ofthe mould (4) and / or a parameter relating to an Argon flow into the mould (4);• a parameter relating to a taper setting ofthe mould (4);• measurement data relating to the cooling water flow in the mould (4); and• a parameter relating to the chemical composition and / or the physical properties ofthe mould powder.

5. Computer-implemented method according to any one ofthe preceding claims1 to 4, wherein the control data comprise at least one ofthe following:• a parameter indicating whether the casting process is stable;• a parameter relating to the casting speed ofthe casting process;• a parameter relating to the casting width;• a parameter relating to a metal flow within the mould (4), in particular a parameter relating to an electromagnetic brake assembly of the mould (4) and / or a parameter relating to an Argon flow into the mould (4);• a parameter relating to a taper setting of the mould (4);• a parameter relating to the chemical composition of a mould powder (20, 21, 24, 25) used in the casting process, and optionally a parameter relating to the physical properties ofthe mould powder.

6. Computer-implemented method according to any one ofthe preceding claims 1 to 5, wherein the control data (86) comprise a compatibility indicator which is indicative ofthe compatibility of the compositions ofthe cast metal and the mould powder (20, 21, 24, 25), and / or wherein the control data comprises a parameter indicating a recommended change in the mould powder (20, 21, 24, 25) composition.

7. Computer-implemented method according to any one ofthe preceding claims 1 to 6, wherein the at least one key parameter indicator comprises a COG indicator, the COG indicator being indicative of a shift in a centre of gravity of a heat transfer distribution across the at least one mould wall (BFF, BFL, NFS, NFN) from a target position, wherein the COG indicator may be indicative of a shift in a feed direction of the metal and / or in a direction perpendicular thereto.

8. Computer-implemented method according to claim 7, wherein the at least one key parameter indicator comprises a COG indicator which is indicative ofthe symmetry of the flow of molten metal inside the mould, wherein the COG indicator is in particular indicative of a shift in a centre of gravity of a heat transfer distribution across the at least one mould wall in a direction perpendicular to the feed direction ofthe metal, , and wherein the control data comprise settings for an electromagnetic brake assembly of the mould (4) and / or an Argon flow into the mould (4).

9. Computer-implemented method according to any one ofthe preceding claims 1 to 8, wherein the control model is configured to compute the totaltransferred heat at a plurality of positions along a line perpendicular to the feed direction, wherein the total transferred heat of a point on metal strand is calculated by following the point as it descends down the mould, and summing the extracted heat along the corresponding vertical line, and wherein the at least one key parameter indicator comprises a QT variation value which is indicative of a variation of the total transferred heat along a direction perpendicular to the feed direction.

10. Computer-implemented method according to claim 9, wherein the control model is configured to compute the total transferred heat at a plurality of positions along a line perpendicular to the feed direction, and wherein the at least one key parameter indicator comprises the overall total transferred heat, summed over all positions along a line perpendicular to the feed direction, wherein the overall total transferred heat is in particular the basis of control data which comprise settings of one or more casting parameters, in particular casting speed, and / or mould powder composition.

11. Computer-implemented method according to any one of the preceding claims 1 to 10, wherein the at least one control model is configured to compute a breaking point position, which is indicative of a position where a layer of resolidified mould powder (20, 21, 24, 25) close to the mould wall (BFF, BFL, NFS, NFN) breaks, and wherein the control model is preferably configured to compute, from a time series of breaking point positions, an upper breaking point position with respect to a meniscus level of the mould (4), an amplitude and / or a frequency of the breaking point positions, and / or a measure of regularity ofthe oscillations in the breaking point positions.

12. Computer-implemented method according to claim 11, wherein the control model is configured to compute the breaking point position by analysing a temperature distribution in the mould wall (BFF, BFL, NFS, NFN), and detecting inflection points in the temperature curve in a feed direction.

13. System (90) for providing control data for a continuous casting facility for casting a metal, in particular steel, in a continuous casting process, thecontinuous casting facility comprising a mould (4), wherein the mould (4) comprises a plurality of temperature sensors (43) distributed across at least one mould wall, comprising: a processing circuitry (92); a storage medium (88); and a data interface (86); wherein the storage medium (88) comprises a computer program that comprises instructions which when the program is executed, cause the processing circuitry to carry out the method according to any one of claims 1 to 12; wherein the data interface (86) is configured to receive input data for the control model including the temperature data from the plurality oftemperature sensors (43).

14. Computer program comprising instructions to cause the system of claim 13 to execute the steps of the computer-implemented method according to any one of the claims 1 to 12.

15. Use of temperature data from a plurality of temperature sensors (43) distributed across at least one mould wall (BFF, BFL, NFS, NFN), casting parameters, and / or control model (84) in a computer-implemented method according to any one of the claims 1 to 12 and / or in a system according to claim 13.

Images