EP1310573A2 - Process to produce a metal melt on the basis of a dynamic process model, including a correction model - Google Patents

Abstract

Process control for molten metal involves using a dynamic computer process model together with a correction model. <??>To produce a molten metal in a metallurgical plant, a process model is used with the plant computer control. The process model has the behavior for at least one variable process parameter between an actual process value, a setting value, and a final process value. The process model takes the actual process value data to a given time point, as the temperature and/or the chemical composition where a computer simulation gives the direct time point where the actual process value gives the final process value at a later point in time. On deviations of the simulated process value from a nominal value, corrections are computed and the actual process value is altered accordingly. The actual process value is repeated from the data taken to the further time point. The process module has at least one module with intelligent self-optimizing e.g. with a neural network.

Description

The invention relates to a method for producing a molten metal in a metallurgical plant, especially for freshening a molten metal, preferably for the production of steel, e.g. of alloyed, stainless steel or stainless steel, which Process on a running according to a process model and the metallurgical plant controlling computing technology is based.

There has always been a need for one for the industrial production of steel Process that enables optimal and inexpensive process control. The The invention therefore relates in particular to a method for producing steel by Fresh a predetermined amount of molten pig iron, in addition to the usual Elements possibly necessary for the production of different qualities Contains alloying elements such as chrome and nickel.

The decarburization of a molten metal, such as a pig iron melt with oxygen for a balance between metal, carbon and oxygen at a given Temperature and at a certain pressure. This balance determines that The extent to which carbon can be removed from the melt without that metallic components, e.g. Chromium as well as iron are oxidized. With this freshening process, the thermodynamic activity within the Metal melt pool and that between the elements contained in the melt pool and the evolving gas atmosphere adjusting balance by the mixture of Oxygen can be influenced with inert gas (as a diluent gas).

AT 339 938 B is a program for optimizing the decarburization of a mass of molten metal, which is based on knowledge of the initial temperature, chemical Initial composition of the premelt and the weight is based, with target values can be obtained under economically favorable conditions. As a diluent any gas that is inert during decarburization is selected. Under Using the initial information, the program uses several coefficients, which are the thermodynamic activities of each element present in the weld pool define as a function of the bath composition, calculated. Using this The program calculates the equilibrium carbon monoxide partial pressure coefficients with carbon and the various metallic elements and oxides. The disadvantage is here that during decarburization, the values of the temperature, the chemical composition etc. are not taken into account, so that continuously changing actual values can have no influence on the end product.

• ein erstes neuronales Netzwerk, um Eingangs- und Ausgangsdaten aus Daten zu analysieren, zu denen die Zusammensetzung, das Gewicht und die Temperatur des Bades am Anfang jeder Prozeßperiode, das zu verwendende Gasverhältnis Sauerstoff zu Verdünnungsgas während jeder Prozeßperiode und die am Abschluß jeder Prozeßperiode erreichte Endtemperatur gehören, bis das neuronale Netz in der Lage ist, einen wesentlichen genauen Ausganswert zu liefern, welcher die Sauerstoffzählwerte darstellt, die in das Bad eingeblasen werden müssen, um zu bewirken daß die Temperatur des Bades auf einen bestimmten Soll-Temperaturwert ansteigt,
• ein zweites neuronales Netzwerk, um Eingangs- und Ausgangsdaten aus Daten zu analysieren, zu denen die Zusammensetzung, das Gewicht und die Temperatur des Bades am Anfang jeder Prozeßperiode und die am Abschluß jeder Prozeßperiode erreichte Endtemperatur gehören, bis das neuronale Netz in der Lage ist, einen im wesentlichen genauen Ausgangswert von Sauerstoffzählwerten zu liefern, um den Kohlenstoffpegel entsprechend einer vorgewählten Aufstellung von Verhältnissen von Sauerstoff zu Verdünnungsgas auf den vorgewählten Soll-Pegel zu senken,
• Zuschlagstoffe werden in drei weiteren neuronalen Netzwerken berücksichtigt, um den Kohlenstoffgehalt, die Temperatur und die Endzusammensetzung des Bades am Ende des Einblasens von Sauerstoff zu berechnen.

Patent EP 0 545 379 B1 describes a method for refining steel by controlling the decarburization of a predetermined molten metal bath with a known chemical composition and temperature. The process has the following process steps:

• a first neural network to analyze input and output data from data including the composition, weight and temperature of the bath at the beginning of each process period, the oxygen to diluent gas ratio to be used during each process period and the final temperature reached at the end of each process period belong until the neural network is able to provide a substantially accurate output value which represents the oxygen counts that must be injected into the bath to cause the bath temperature to rise to a certain target temperature value,
• a second neural network to analyze input and output data from data including the composition, weight and temperature of the bath at the beginning of each process period and the final temperature reached at the end of each process period until the neural network is able to provide a substantially accurate baseline of oxygen counts to reduce the carbon level to the preselected target level according to a preselected set of oxygen to diluent ratios,
• Aggregates are taken into account in three other neural networks to calculate the carbon content, the temperature and the final composition of the bath at the end of the oxygen injection.

The disadvantage here is that there is no continuous recording over the entire treatment time and the plausibility of the calculated temperature and analysis values is checked. The Process optimization is difficult in this way if it is feasible, especially if the conditions in the melting unit (change in refractory temperature, change in Refractory strength, change in reaction volume, ...) are also taken into account.

The invention according to EP 0 857 222 B1 relates to a process for decarburizing a molten steel for the production of high-chromium-containing steels, in which the decarburization rate is measured continuously and the amount of oxygen to be blown in is set as a function of the measured values, the decarburization rate being determined from the CO and CO ₂ content in the exhaust gas and the exhaust gas flow is determined. The measurement of the exhaust gas compositions listed above is possible, but relatively inaccurate. The position of the measuring probe in the exhaust gas flow must be positioned in the vicinity of the mouth of the crucible, on the one hand to obtain the exhaust gas information relatively quickly and on the other hand to minimize / avoid the falsification of the exhaust gas composition by fresh air entering the crucible mouth area. This method is less suitable for the production of alloyed steels, since metal oxidation is not taken into account and cannot be determined with this method either.

According to DE 33 11 232 C2, a computer-controlled refining of metal melts takes place with oxygen and a diluent gas, the gas flow rates through Calculate the extent of metal oxidation using calculated values be fixed.

The method described in DE 33 11 232 C2 is for decarburizing Metal melting is suitable, but this method is due to the model used not suitable, exactly the time of reaching the transition point from the Decarburization reaction to determine metal oxidation. The consequence is an increased Chromium erosion and therefore additional quantities of reducing agents required (Ferrosilicon, lime) and a reduced durability of the converter.

• Bei Verwendung einer Sublanze muß der Prozeß für die Messung nicht unterbrochen werden (Temperaturangaben liegen unmittelbar nach Eintauchen der Meßsonde in die Stahlschmelze vor; bei einer Probenahme muß auf die Analyseergebnisse vom Labor gewartet werden (etwa 3 - 6 Minuten)).
• Wenn keine Sublanze verwendet wird, muß für eine Messung der AOD-Prozeß unterbrochen werden. Bei einer Handmessung liegt der Temperaturwert ebenfalls unmittelbar nach Eintauchen der Sonde in die Stahlschmelze vor. Bei einer Probenahme muß auf die Analyseergebnisse wie oben angeführt ebenfalls ca. 3 bis 6 Minuten gewartet werden.

The common practice in the AOD steelmaking process is as follows:
Measurements (temperature, sampling for chemical analysis) are carried out during a batch (especially for stainless steel production).

• When using a sublance, the process for the measurement does not have to be interrupted (temperature information is available immediately after immersing the measuring probe in the molten steel; when taking a sample, the analysis results have to be waited for by the laboratory (about 3 - 6 minutes)).
• If no sublance is used, the AOD process must be interrupted for a measurement. With a manual measurement, the temperature value is also available immediately after immersing the probe in the molten steel. When taking a sample, the analysis results as described above must also be waited for about 3 to 6 minutes.

The disadvantage of both measurement methods is that the information about the steel melt (Temperature, chemical analysis) only done selectively. It comes with hand measurement also with each measurement for a process interruption (the converter must be Measurement can be switched). This requires an increase in tap to tap time are higher refractory wear, lowering the temperature of the steel melt (caused by flipping the converter), etc.

Corrective action at too high or too low temperature can only be done after performed measurement.

The object of the invention is, especially for conventionally difficult to regulate Production processes, e.g. in the AOD process to specify a method with which a Production increase, energy saving, shortening the tap to tap time (ttt time), Aggregate optimization and a higher durability of the in the production process for Use of refractory materials can be achieved. In particular, the Invention also in the production of alloyed steels, e.g. Cr-Nr alloyed stainless steels be applicable.

• mit dem Prozeßmodell wird mit zu einer bestimmten Zeit (t_i) erhobenen Daten einer Ist-Prozeßgröße, wie der Temperatur der Schmelze und/oder der chemischen Zusammensetzung der Schmelze, durch Simulation mit Rechentechnik unmittelbar zum Zeitpunkt der Erhebung der Ist-Prozeßgröße eine Prozeßgröße für einen späteren Zeitpunkt (t_i + dt), vorzugsweise eine Prozeßendgröße, ermittelt und
• bei Abweichungen der simulierten Prozeßgröße von einem gewünschten Soll-Wert werden mittels des Prozeßmodells mit Rechentechnik Korrekturmaßnahmen zur Änderung der Ist-Prozeßgröße errechnet und die Ist-Prozeßgröße entsprechend geändert
• worauf zu einem späteren Zeitpunkt (t_i + dt) mit weiters erhobenen Daten der Ist-Prozeßgröße das Verfahren wiederholt wird.

According to the invention, this object is achieved by the following method steps:

• with the process model, data of an actual process variable, such as the temperature of the melt and / or the chemical composition of the melt, is collected at a specific time (t _i ) by simulation with computer technology immediately at the time of the collection of the actual process variable a later point in time (t _i + dt), preferably a final process variable, is determined and
• in the event of deviations of the simulated process variable from a desired target value, corrective measures for changing the actual process variable are calculated using the process model with computing technology and the actual process variable is changed accordingly
• whereupon the process is repeated at a later point in time (t _i + dt) with further data of the actual process variable.

According to the invention, a measurement on the melt is thus carried out at a specific point in time carried out, etc. at least the temperature or at least the chemical Composition, based on a process model with the measured value Value for a later time is calculated (preferably at the end time), and it if the value thus extrapolated from the measured value differs from one a corrective measure for the process variable is taken in the desired predetermined value, this can of course be done immediately or at a later date. If e.g. should point out that the temperature at the end of a batch is too high corrective action can be taken by adding refrigerated scrap what either immediately at the time of measurement or, if for the process, for metallurgical reasons cheaper, can also be done a little later.

The implementation of the method according to the invention can be carried out all the better if the faster the measurement data are available after the measurement has been carried out. As well as temperature measurement as well as for chemical analysis are known measuring methods (WO 97/22859 and WO 02/48661) with which the almost immediately after the measurement measured data are available. So are the temperature values of the melt after ms and a temperature value averaged from several measurements after about one s available. Chemical analysis values are e.g. after 0.05 to 0.1 s and one over about 100 measurements averaged value available after 5 to 10 s.

The process model expediently checks the data of an actual process variable at a specific time (t _i ) for plausibility and only makes plausible data available for simulating the process variable, and non-plausible data is discarded, the simulation being based on the last one in the latter case certain plausible data is continued. In this way, despite incorrect data, for example caused by measurement errors, etc., the process can be ended without delay and the process objective is achieved.

As a result of malfunctions, incorrect measurements can occur. Such disorders are for example, slag splashes on sensors or an opening in the opening metallurgical vessel through which the melt is observed by means of the sensor. in the in the latter case, a past state would be continuously measured. Such Permanent measured values are only permissible for a certain period of time, which time period from the course the procedure, the measured value and the procedure already carried out is dependent. If the permanent measured value is recognized as inadmissible, i.e. as not plausible, you can go straight to fixing the problem, for example the burn overgrown opening through which the melt is to be observed, if necessary to clean a window, etc.

A plausibility check can be used to measure values outside a predetermined Bandwidth of one at a certain time according to an ideal course of the process predetermined value are eliminated. This range is in the Order of magnitude ± 10% of the specified ideal value, preferably ± 5%. The definition the ideal course of the process or an ideal value at a specific time is based on from empirical values or calculated values. If there is a new measured value within the Bandwidth, i.e. it is accepted as plausible, in turn becomes one from this last value Ideal course of the process based on empirical values or calculated values accepted and the simulation continues due to this new process continued so that ultimately measurement values are accepted that are outside the bandwidth one of the previous ideal courses of the process that have since been abandoned lying are recognized as plausible. This makes the system self-learning.

A separate process model module is advantageously found for each variable process variable Application. Under the process model module, a self-contained part of the Understand the process model that is responsible for a certain functional sequence, such as for decarburization, desulfurization, temperature prediction and alloying with alloying elements. The process model modules are, however, as far as the individual Functional processes influence each other, taking these influences into account linked with each other.

The process model itself or at least one module of the process model is based on one preferred embodiment equipped with intelligent self-optimization. The The object of the invention is achieved here by an intelligently designed system, the independently based on the given prior knowledge, appropriate instructions for safe and optimal process control. The independently given Instructions are continuously compared with the actual data and for plausibility checked.

The previously entered knowledge (process knowledge) is automatically and continuously carried out on the process Knowledge gained during production is improved. This self-generating Process knowledge is adopted from a data store as new prior knowledge, i.e. it comes to a constantly improved basis for further adaptation or optimization of the process.

The method is expediently carried out in such a way that an existing mode of operation (obtained, for example, based on many years of experience; know how) for different grades is optimized in such a way that changes which are favorable for the process (such as reduction of O ₂ , optimization of the process temperature curve are reduced to lower ones) Values; reducing / optimizing gas quantities; shortening process times; ...) for the respective quality can be adopted and saved. As a result, the process optimization steps are tightened. This process is accomplished using neural networks, among other things.

The optimal execution of the method is at least part of the process time the time difference between two successive surveys of an actual process variable kept very small, preferably the time difference is infinitesimal, i.e. the actual process variable is collected continuously.

A particularly simple process model with which very good end results can be achieved let, is characterized in that the actual process variables the chemical Composition of the exhaust gas and the chemical composition of charged Material and possibly the chemical composition of the slag collected become.

The chemical composition of the Exhaust gas and the chemical composition of charged material as well if necessary, the chemical composition of the slag is collected.

The final process variables, which represent quality parameters for the finished molten metal, if possible to match the specified ideal values to be sought, are advantageous, adjustable process variables such as flow, pressure, temperature and Composition of gases and / or quantities, compositions and Charging speeds of additives optimized on the process model.

The method according to the invention is based on knowing the actual situation to be determined Points in time during the procedure, whereby, as stated above, permanent knowledge the actual situation is optimal for at least the most important process steps method according to the invention results. Based on the knowledge of the actual state, i.e. the actual process variables, the process parameters, such as oxygen flow, inert gas flow, Time of addition, chemical composition and quantity of additives, optimized become. Knowing the current situation allows the target points to be approached exactly, e.g. can already with knowledge of the chemical composition and the temperature of the Metal melt the missing and still to be added materials and gas quantities in time precisely introduced in terms of quantity and analysis.

The melting process for producing a steel melt specifically for the AOD process would proceed as follows:

With the model, the static becomes according to the specification of the respective quality product Process behavior for a process variable y = f (manipulated variables, process variables that cannot be influenced) simulated. The final process variable y * calculated from this is a typical quality parameter of the product to be manufactured and deviates from the actual size y more or less.

The process model is improved with a model adaptation, so that the model behavior matches the actual process behavior as closely as possible.

By means of an optimization process carried out with a process optimizer Control variables found that lead to good process behavior. For example, at too high steel melt temperature either reduces the oxygen supply or the Time of addition for refrigerated scrap or alloying elements can be changed.

Because the process model is based to a certain percentage on uncertain knowledge the process model is adapted and changed based on the process data obtained. This adaptation takes place with the process model adaptation, which is based on data from past process states. The process results are checked by plausibility queries.

The basic structure of the method according to the invention is simplified in FIG. 1 and FIG. 2, each in block diagram form. Processes according to the invention decarburization (DeC process), reduction, addition of alloys and cooling illustrated in block diagram form in Figs. 3, 4, 5 and 6. 7 and 8 show Temperature curve and chemical analysis values on the process flow of the following Example.

Metallurgische Gleichungen

Algorithmen, Parameter

Festlegung interner Modellschnittstellen

Struktur, interner Ablauf

The following functions are provided as process model functions:

Metallurgical equations

Algorithms, parameters

Definition of internal model interfaces

Structure, internal process

Before starting the process model, knowledge of the state of the premelt and Slag necessary.

The initial analysis, analysis of the aggregates and the quantities of gas introduced are can be determined without difficulty and are available at any time. The knowledge the temperature of the molten steel, the exhaust gas composition and the analysis of the Melting steel is based on snapshots, realized by a continuous Working temperature measuring and analysis system for the molten metal, such as e.g. according to WO 97/22859 or WO 02/48661.

This knowledge only makes it possible to optimize and supplement existing models. The exemplary AOD model designed according to the invention takes this into account and became so built that after knowing the actual state, the process model parameters so be adjusted / changed so that the desired values are actually achieved. With This procedure ensures that the FF consumption reduces the flow rates optimized and thus reduced and the ttt time can be shortened. In addition, a targeted application strategy optimizes the amount of aggregates.

Detailed model description:

According to an AOD converter 1 shown in FIG. 1, oxygen and a Dilution gas (inert gas) both via a top lance 2 and via nozzles 3 which are provided below the molten steel bath level 4 on the converter 1, in the interior thereof 5 introduced. The converter 1 is, as is customary in the case of steelworks converters, tiltably mounted, which is not shown in more detail. The converter 1 is also equipped with a device 6 for Temperature measurement of the molten steel 7 and a device 8 for determining the chemical analysis of the molten steel 7 equipped. You can also use Measuring devices 9 and 10 the flow rates through the top lance 2 and the nozzles 3rd determine the gases introduced. The exhaust gas emerging from the converter 1 is also by means of a device 11 with regard to its chemical composition analyzes which analysis device 11 in the exhaust gas chimney, which is not shown in detail, is provided.

Before starting the process, the molten steel charged in the converter 1 7 Weight, chemical analysis and temperature determined. Also from the on the Premelt resting slag 12 are weight, chemical composition and Temperature determined. Of the planned aggregates to be introduced are also Weight, chemical composition and temperature known.

Information about steel qualities and tapping condition include all limit and target values of Steel elements at the time of tapping as well as the tapping temperature.

1. Düsen 3: Durchflußmengen für Ar, N₂, O₂, Luft, CH₄ oder C_nH_m (Nm³/h) = f (t)

2. Toplanze 2: Durchflußmengen für Ar, N₂, O₂ = f (t)

3. Position der Toplanze 2 = f (t)

4. Materialzugabestoffe: Art, chemische Zusammensetzung, Geschwindigkeit, Gewicht, Temperatur = f (t)

5. Chemische Analyse Stahlschmelze 7 = Funktion (t)

6. Temperatur Stahlschmelze 7 = Funktion (t)

7. Konverterposition (Grad der Neigung) = f (t)

8. Abgas: Chemische Zusammensetzung, Temperatur, Menge = f (t)

The following data are advantageously collected as actual process variables for the model calculation:

1. Nozzles 3: flow rates for Ar, N ₂ , O ₂ , air, CH ₄ or C _n H _m (Nm ³ / h) = f (t)

2. Top lance 2: flow rates for Ar, N ₂ , O ₂ = f (t)

3. Position of the top lance 2 = f (t)

4. Material additives: type, chemical composition, speed, weight, temperature = f (t)

5. Chemical analysis of molten steel 7 = function (t)

6. Temperature of molten steel 7 = function (t)

7. Converter position (degree of inclination) = f (t)

8. Exhaust gas: chemical composition, temperature, amount = f (t)

The temperature and the concentration of an element in the molten steel 7 are determined using the following definitions: Concentration (target, element x in the molten steel) [%; t] = concentration (is, element x in the molten steel) [%, t-Dt] + f (amount of gas (Nl / min), type of gas, operating weight Molten steel (t 0 ), Weight of aggregates, spreading, temp; t-Dt). concentration (is, element x in the aggregate) [%]

f (..,t), g (...,t)

zeitabhängige Funktionen

t

Zeit

soll

Soll-Wert

ist

Ist-Wert

Zuschlagstoffe

Legierungselemente, Kühlschrott, Schlackebildner, .....

Here mean:

f (.., t), g (..., t)

time-dependent functions

t

time

should

Target value

is

Actual value

aggregates

Alloy elements, cooling scrap, slag formers, .....

1. Concentration of an element x in the molten steel:

Total concentration (target, element x) = 100% Concentration (element x) (%; t) = f (amount of gas (Nl / min), type of gas, operating weight Molten steel (t 0 ), Weight of aggregates, spreading, temp; t-Dt). Concentration (is, Element x aggregate) [%]

2. Determination of the steel melt temperature:

Tempsoll (t) = Tempist (t-Dt) + g (oxidation, gas quantity (Nl / min), heat losses, exhaust gas losses; t-Dt) t-Dt ........ the measurement is available at time t-Dt, where Dt is the time step between two calculation steps. The smaller Dt is selected, the more precisely the target curve can be achieved.
For short time intervals it is necessary that the measurement data describe the current state in the molten steel.

During the model calculation, the values of the elements (%) and the temperature (° C) for steel, slag, exhaust gas (components, amount) and the heat balance at time i calculated cyclically after model start.

The gas phase, the steel and the slag are balanced.

A total (kg or mole) for each element / substance is used for the balancing set up at any time.

x_k

Anteil von x im Zugabestoff k

Ausbringen_k

Ausbringen des Zuschlagstoffes k

Gewicht x_Oxidation

Gewicht des Stoffes x, der im Zeitabschnitt Dt oxidiert

Gewicht x_vap

Verdampfungsverlust im Zeitabschnitt Dt

The balance equation for element x is: Weight x i + 1 = Weight x i + ΣAdditive substances k * x k * Spread k - weight x oxidation - Weight x vap

x _k

Share of x in the additive k

Spread _k

Application of the aggregate k

Weight x _oxidation

Weight of the substance x that oxidizes in the period Dt

Weight x _vap

Evaporation loss in the period Dt

The oxidation and reduction products are in the slag and gas phase consider.

Important reactions (these are stored in a database) are: ½ O ₂ → [O] [C] + [O] → CO _gas CO _gas + 1 / 2O _2gas → CO ₂ Fe + [O] → (FeO) [Mn] + [O] → (MnO) [Mo] +2 [O] → (MoO ₂ ) 2 [Al] + 3 [O] → (Al ₂ O ₃ ) 2 [P] + 5 [O] → (P ₂ O ₅ )

The sequence of the iteration and calculation process is shown in FIGS. 3 and 4 using a logical plan, with FIG. 3 being decisive for the decarburization process and FIG. 4 for the reduction process. These processes are based on the following forms of calculation. CrO ₃ + 2Al → 2 [Cr] + Al ₂ O ₃ Cr ₂ O ₃ + 3Si → 2 [Cr] + 3/2 SiO ₂ 2FeO + Si → 2 [Fe] + SiO ₂ 3FeO + 2Al → 3 [Fe] + Al ₂ O ₃ 2MnO + Si → 2 [Mn] + SiO2 3MnO + 2 Al → 3 [Mn] + Al ₂ O ₃

Alloy and cooling calculation 1. Alloys

a) Einstellung der Analyse

b) Erreichung des Abstichgewichtes

The alloy substance calculation has two main objectives:

a) Stop the analysis

b) reaching the tapping weight

Based on the condition of the premelt (chemical analysis of the molten steel, weight, Temperature), all necessary additions are calculated based on the targeted racking analysis.

• Kostenoptimierung, d.h. Kostenminimierung für jedes Legierungselement
• Zugabe von C>2% sind nur in den steps 1, 2a, 2b gestattet
• Cr-Zugabe: in step 1 und 2a (50%), 2b (50%), Reduktion (Feineinstellung)
• Ni-Zugabe: in step 2c (90%), restliche Zugabe in step 3 (30%), step 4 (20%), Reduktion (20%)
• MN-Zugabe:  in step 2b (100% HCMn) oder    in step 2b (50%) und Rest mit SiMn in Reduktionsphasen (wenn verfügbar)
• Zum Masseaufbau wird eine zusätzliche neutrale Zugabe mit Abstichanalyse berechnet.

The choice of alloying materials for preliminary calculation is based on the following criteria:

• Cost optimization, ie cost minimization for each alloy element
• Addition of C> 2% is only allowed in steps 1, 2a, 2b
• Cr addition: in step 1 and 2a (50%), 2b (50%), reduction (fine adjustment)
• Ni addition: in step 2c (90%), remaining addition in step 3 (30%), step 4 (20%), reduction (20%)
• MN addition: in step 2b (100% HCMn) or in step 2b (50%) and the rest with SiMn in reduction phases (if available)
• An additional neutral addition with tapping analysis is calculated for mass build-up.

A shift in the timing of alloying between the different ones steps is done to maintain the desired temperature.

The logical plan for alloy calculation can be seen in FIG. 5.

2. Cooling

FeNi or Ni, lime and / or scrap are primarily used for cooling.

Lime / Dolo consumption

The total consumption of lime and dolo depends on the Si entry and SiO ₂ input during the entire process: SiO ₂ (kg / t) = Σ SiO _2j (%) + j _addition + 2.14 * (Σ Si% * j _addition * + Si _Red )

The logical plan for calculating the lime / dolo consumption is shown in FIG. 6.

• Badspiegelberechnung
• Chargiermodell
• Thermische Modellberechnung
• Analyseberechnung
• Stickstoff Modell
• DeC-Modell
• DeC-Modell (Feinabstimmung Abstich)
• Reduktionsmodell
• DeS-Modell
• Neutrales Zuschlagstoff-Modell
• Legierungsmodell
• Abstich-Modell
• Temperatur-Modell
• T-Kontroll-Modellfunktion
• Reblow-Modell

The following model functions are preferably used for the process model:

• Badspiegelberechnung
• Chargiermodell
• Thermal model calculation
• analysis calculation
• Nitrogen model
• DeC model
• DeC model (fine-tuning racking)
• reduction model
• THe model
• Neutral aggregate model
• alloy model
• Racking model
• Temperature model
• T-control model function
• Reblow model

The results of the model calculation are continuously updated with the actual values of the Melting steel temperature, slag temperature, analysis of the molten steel and the Exhaust gas measurement compared.

Example:

The following process flow is used to manufacture the quality AISI 304 illustrated from the raw steel used to the finished product. 7 and 8 show the course of the temperature change and the course of the change in chemical analysis (for C, Mn, Cr, Ni) in this process.

AOD process model in combination with a continuous temperature and analysis measurement Process flow for AISI 304:

1. Charge (no addition of slag formers)

2. Temperature measurement (manually or with a continuous T-measuring system)

3. 1st bubble level with top lance

4. Continuous measurement of the temperature and throughout the first blowing stage the steel bath analysis

5. 2nd bubble level to 4th bubble level without top lance

6. Continuous measurement of the entire 2nd bladder stage to 4th bladder stage Temperature and the steel bath analysis

7. 5th bubble level

8. Continuous measurement of temperature and throughout the 5th blowing stage the steel bath analysis

9. Reduction and desulphurization phase

10. Continuous throughout the reduction and desulfurization phase Measurement of temperature and steel bath analysis

11. upon reaching the target analysis and target temp. Racking with possible Si fine correction by adding alloying agents to the pan during tapping

Blowing scheme AISI 304 O ₂ lance
[Nm ³ / min] O ₂ nozzles
[Nm ³ / min] N ₂ lance
[Nm ³ / min] Ar-lance
[Nm ³ / min] Target salary C
[%] target temperature
[° C] Step 1 100 30 15 - 0.6 1710 Step 2 - 60 20 - 0.4 1740 Step 3 - 45 45 - 0.2 1760 Step 4 - 20 60 - 0.08 1750 Step 5 - 12 48 - 0.04 1750 Reduction and desulfurization - - - 45 - 1665

Carbon removal efficiency (CRE):

CRE step 1:

average 78%

CRE total:

68% on average

1.Charging (step 1):

Einsatzgewicht:

Stahl 86,5 t
Schlacke: 0,6 t

Analyse C 1,8 Si 0,14 Mn 0,59 P 0,023 S 0,024 Cr 19,03 Ni 7,14 Temperatur zur Zeit t=O:  1538°C
Medienverbräuche: siehe Tabelle 1

Zugabe:

Dolomit 1500 kg
Kalk 2000 kg
FeNi 1500 kg

Temperatur nach step 1: 1709°C
Behandlungszeit step 1: 11,25 Minuten

Operating weight:

Steel 86.5 t
Slag: 0.6 t

analysis C 1.8 Si 0.14 Mn 0.59 P 0.023 S 0.024 Cr 19,03 Ni 7.14 Temperature at time t = O: 1538 ° C
Media consumption: see table 1

addition:

Dolomite 1500 kg
Lime 2000 kg
FeNi 1500 kg

Temperature after step 1: 1709 ° C
Treatment time step 1: 11.25 minutes

2. Decarburization stage 2 (step 2)

Zugaben:

Kalk 1000 kg
FeNi 1000 kg

Temperatur nach step 2: 1743°C
Behandlungszeit step 2: 4,83 Minuten Analysis after step 1 C 0.507 Si 0.0 Mn 0.49 P 0.023 S 0,022 Cr 18.39 Ni 7.91 Media consumption: see table 1

Add-ons:

Lime 1000 kg
FeNi 1000 kg

Temperature after step 2: 1743 ° C
Treatment time step 2: 4.83 minutes

3. Decarburization level 3 (step 3)

Zugabe:

Kalk 1500 kg

Temperatur nach step 3: 1760°C
Behandlungszeit step 3: 5,66 Minuten Analysis after step 2 C 0,347 Si 0 Mn 0,475 P 0.023 S 0,022 Cr 17.77 Ni 8.34 Media consumption: see table 1

addition:

Lime 1500 kg

Temperature after step 3: 1760 ° C
Treatment time step 3: 5.66 minutes

4. Decarburization level 4 (step 4)

Zugabe:

Kalk 480 kg
Shredder 1000 kg

Temperatur nach step 4: 1749°C
Behandlungszeit step 4: 7, 92 Minuten Analysis after step 3 C 0,199 Si 0 Mn 0.457 P 0.023 S 0,021 Cr 17.426 Ni 8,395 Media consumption: see table 1

addition:

Lime 480 kg
Shredder 1000 kg

Temperature after step 4: 1749 ° C
Treatment time step 4: 7, 92 minutes

5. Decarburization level 5 (step 5)

Analysis after step 4 C 0,119 Si 0 Mn 0.462 P 0.0238 S 0.0209 Cr 17.197 Ni 8,431 Media consumption: see table 1
addition:
Temperature after step 5: 1752 ° C
Treatment time step 5: 16.08 minutes

6. Reduction and desulfurization phase

Zugabe:

CaF2 710 kg
SiMn 900 kg
FeSi 1602 kg
Nimet 90 kg
Kühlmittel ss 500 kg

Temperatur nach Reduktions- und Entschwefelungsphase: 1660°C Analysis after step 5 C 0.0397 Si 0 Mn 0.453 P 0.024 S 0.0208 Cr 16,853 Ni 8.479 Media consumption: see table 1

addition:

CaF2 710 kg
SiMn 900 kg
FeSi 1602 kg
Nimet 90 kg
Coolant ss 500 kg

Temperature after the reduction and desulphurization phase: 1660 ° C

7th racking

analysis C 0.0558 Si .5933 Mn 1.2516 P 0.0251 S 0.0003 Cr 18.485 Ni 8.0788 Tapping temperature 1660 ° C
Tapping weight: 90.5 t
Slag: 9.478 t

The curves of the diagrams shown in FIGS. 7 and 9 show that there is almost complete agreement with the target curves. This means that the actual process flow follows the predetermined ideal process flow and the Target values with only slight deviations, which - if at all - only through Measurement inaccuracies can be caused. This is made possible by the Immediate intervention in the process that is provided according to the invention should be based on the measurement results and the simulation based on them, i.e. Precalculation of the expected results, deviations from the target values can be expected.

In addition to this, there is already the learning effect for this dynamic model carried out processes.

Claims (9)

A process for the production of a molten metal in a metallurgical plant, in particular for the refining of a molten metal, preferably for the production of steel, such as alloyed, stainless steel or stainless steel, which process is based on a computer technology which operates according to a process model and controls the metallurgical plant, the Process model describes the behavior for at least one variable process parameter between an actual process variable, a manipulated variable and a final process variable, and the method comprises the following steps: with the process model, data of an actual process variable, such as the temperature of the melt and / or the chemical composition of the melt, is collected at a specific time (t _i ) by simulation with computer technology immediately at the time of the collection of the actual process variable a later point in time (t _i + dt), preferably a final process variable, is determined and If the simulated process variable deviates from a desired target value, corrective measures for changing the actual process variable are calculated using the process model with computing technology and the actual process variable is changed accordingly, whereupon the process is repeated at a later point in time (t _i + dt) with further data of the actual process variable. Method according to Claim 1, characterized in that the process model is used to check the data of an actual process variable at a specific time (t _i ) for plausibility and only to provide plausible data for simulating the process variable and to reject non-plausible data in the latter case, the simulation is continued on the basis of the plausible data determined last. Method according to Claim 2, characterized in that data are used as plausible for the simulation of the process variable, which lie within a bandwidth of ± 10%, preferably ± 5%, of an ideal course of the manufacturing process, which is based on empirical values or on calculated values. Method according to Claim 2 or 3, characterized in that , based on a value which is recognized as plausible but which differs from the last valid ideal course of the manufacturing process, a new ideal course is created and the further process is used as a basis. Method according to one or more of claims 1 to 4, characterized in that a separate process model module is used for each variable process variable. Method according to one or more of claims 1 to 5, characterized in that the process model or at least one module of the process model is or are equipped with intelligent self-optimization, for example using a neural network. Method according to one or more of claims 1 to 6, characterized in that the time difference between two successive surveys of an actual process variable is very small, preferably infinitesimal, over at least part of the process duration. Method according to one or more of claims 1 to 7, characterized in that the actual process variables are the chemical composition of the exhaust gas and the chemical composition of the charged material and optionally the chemical composition of the slag. Method according to one or more of claims 1 to 8, characterized in that adjustable process variables, such as flow, pressure, temperature and composition of gases and / or quantities, compositions and charging speeds of additives, are optimized on the process model.

Images