WO2025190825A1 - Procédé de commande et d'optimisation d'un procédé de raffinage de cuivre, le procédé de raffinage de cuivre comprenant au moins les étapes de fusion, d'oxydation et de réduction - Google Patents

Procédé de commande et d'optimisation d'un procédé de raffinage de cuivre, le procédé de raffinage de cuivre comprenant au moins les étapes de fusion, d'oxydation et de réduction

Info

Publication number
WO2025190825A1
WO2025190825A1 PCT/EP2025/056367 EP2025056367W WO2025190825A1 WO 2025190825 A1 WO2025190825 A1 WO 2025190825A1 EP 2025056367 W EP2025056367 W EP 2025056367W WO 2025190825 A1 WO2025190825 A1 WO 2025190825A1
Authority
WO
WIPO (PCT)
Prior art keywords
refining process
copper
copper refining
optimizing
controlling
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/EP2025/056367
Other languages
English (en)
Inventor
Nikolaus Peter Kurt Borowski
Markus Andreas Reuter
Ali Akouch
Rolf Degel
Christoph Kirmse
Sabrine KHADHRAOUI
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
SMS Group GmbH
Original Assignee
SMS Group GmbH
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by SMS Group GmbH filed Critical SMS Group GmbH
Publication of WO2025190825A1 publication Critical patent/WO2025190825A1/fr
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • C—CHEMISTRY; METALLURGY
    • C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
    • C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
    • C22B15/00—Obtaining copper
    • C22B15/0026—Pyrometallurgy
    • C22B15/0028—Smelting or converting
    • C—CHEMISTRY; METALLURGY
    • C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
    • C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
    • C22B15/00—Obtaining copper
    • C22B15/0026—Pyrometallurgy
    • C22B15/0056—Scrap treating
    • C—CHEMISTRY; METALLURGY
    • C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
    • C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
    • C22B15/00—Obtaining copper
    • C22B15/0026—Pyrometallurgy
    • C22B15/006—Pyrometallurgy working up of molten copper, e.g. refining
    • C—CHEMISTRY; METALLURGY
    • C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
    • C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
    • C22B15/00—Obtaining copper
    • C22B15/0095—Process control or regulation methods
    • C—CHEMISTRY; METALLURGY
    • C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
    • C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
    • C22B5/00—General methods of reducing to metals
    • C22B5/02—Dry methods smelting of sulfides or formation of mattes
    • C22B5/18—Reducing step-by-step

Definitions

  • Method for controlling and optimizing a copper refining process wherein the copper refining process comprises at least the steps of smelting, oxidation and reduction
  • the invention relates to a method for controlling and optimizing a copper refining process, wherein the copper refining process comprises at least the steps of smelting, oxidation and reduction.
  • the present invention particularly refers to controlling and optimizing a full copper refining process comprising the three major steps of smelting, oxidation and reduction.
  • the first step melts the copper scrap, then one or more oxidation steps will remove impurities such as Al, Fe, Ni, Pb, Sn, Zn, and so on from the copper alloy into the slag by oxidation of especially the ignoble metals to their oxides.
  • the surplus oxygen present in the copper alloy will be removed during the reduction step, yielding a pure copper anode.
  • the previously described process can take up to 24 hours.
  • Document CN116258087A provides a matte grade soft measurement method and device, electronic equipment and a storage medium, and relates to the technical field of data processing.
  • the copper concentrate is smelted in the molten pool to obtain matte as an intermediate product, and then blown to produce blister copper, in which matte is not only the product of molten pool smelting, but also the raw material for blowing to produce blister copper.
  • the grade of matte determines whether it meets the requirements of blister copper blowing in the next step, and the measurement of matte grade is of great significance for the control and parameter adjustment of the entire process.
  • Basic historical data are obtained based on the historical database, and the basic historical data comprise a plurality of original characteristic variables and historical measurement grades; performing abnormal value processing on the plurality of original feature variables to obtain a plurality of corrected sample input vectors; performing time sequence matching and sample weighting on the corrected sample input vector and the historical measurement grade to obtain corrected historical data; constructing a modeling data set according to the corrected historical data; obtaining a current working condition point, and determining a similar sample set according to the current working condition point; based on the similar sample set, establishing a local soft measurement model through Gaussian process regression; and obtaining the predicted grade of the local soft measurement model for the current working condition point.
  • real-time soft measurement of the matte grade can be realized, the measurement precision is relatively high, and the method can be widely applied to the molten pool smelting process flow.
  • This document only refers to the first step of a complete copper refining process. Since it bases on processing of historical data, it cannot be applied to the complete copper refining process, which takes up to 24 hours, due to the large amount of data and the complex combined process.
  • Document WO2021203912A1 discloses an online prediction method for parameters in a copper converting process based on an oxygen bottom blowing furnace.
  • the method comprises: establishing a bottom blowing converting furnace mechanism model according to a raw material input condition and on the basis of a material balance model, an energy balance model, and a multiphase balance model; establishing a bottom blowing converting furnace data driving model according to actual production data and on the basis of a crude copper grade neural network model, a slag ferrosilicon ratio neural network model, and a slag temperature neural network model between a target parameter and an input parameter; integrating the mechanism model and the data driving model by using an intelligent coordinator to obtain a bottom blowing converting furnace mixing model relating to a crude copper grade prediction value, a slag ferrosilicon ratio prediction value, and a slag temperature prediction value; and outputting final prediction values of the crude copper grade, the slag ferrosilicon ratio, and the slag temperature in a copper bottom blowing converting process by using the mixing model
  • the prediction method can effectively solve the problem that the existing prediction models and methods are poor in adaptability and not satisfactory in actual operation effect and can significantly improve the accuracy of a prediction result.
  • the disclosed method only refers to one particular step of the copper refining process, namely a smelting step executed in an oxygen bottom blowing furnace.
  • the used models are also data driven and hence cannot be applied to the complete copper refining process, which takes up to 24 hours, due to the large amount of data and the complex combined process.
  • document CN1100001 OA also discloses an oxygen bottom blowing copper smelting process feedback control method and device, electronic equipment and a storage medium.
  • the method comprises the steps: acquiring an operation variable of the bottom blowing smelting process and a preset value of the operation variable, and acquiring actual production data of the operation variable to obtain predicted values of important parameters in the smelting process; obtaining target values of the important parameters, and calculating deviation values between a predicted value and the target values; when the deviation values exceeds a preset specified range, calculating a feedback compensation value of the operation variable according to the deviation values; and generating a set value of the operation variable according to the preset value of the operation variable and the feedback compensation value.
  • errors caused by measurement delay and irregularity of an actual detection value can be reduced.
  • Document CN111554353A again discloses an online prediction method for copper smelting process parameters of an oxygen bottom blowing furnace.
  • the method comprises the following steps: establishing a bottom blowing smelting furnace mechanism model based on a material balance model, an energy balance model and a multiphase balance model according to raw material input conditions; establishing a bottom blowing smelting furnace data driving model according to the actual production data and based on the copper matte grade neural network model, the silicon iron ratio neural network model and the slag temperature neural network model between the target parameters and the input parameters; and integrating the mechanism model and the data driving model by using an intelligent coordinator to obtain a bottom blowing smelting furnace mixing model about the copper matte grade prediction value, the silicon iron ratio prediction value and the slag temperature prediction value, and outputting final prediction values of the copper matte grade, the silicon iron ratio and the slag temperature in the copper bottom blowing smelting process by using the mixing model.
  • the prediction method can solve the problems that an existing prediction method is poor in adaptive capacity and not ideal in
  • Document CN112083694A discloses an oxygen bottom blowing copper blowing process feedback control method and device, an electronic device and a storage medium.
  • the method comprises the following steps: acquiring an operation variable of a bottom blowing refining process and a preset value thereof, and acquiring actual production data of the operation variable; inputting the actual production data of the operation variable into a pre-established blowing process neural network prediction model to obtain a prediction value for the production index parameter; obtaining a target value for the production index parameter, and calculating a deviation value between the predicted value and the target value; when the deviation value exceeds a preset specified range, calculating a feedback compensation value of the operation variable according to the deviation value; and generating a set value of the operation variable according to the preset value of the operation variable and the feedback compensation value.
  • errors caused by measurement delay and irregularity of an actual detection value can be reduced.
  • this document again refers to the smelting step only and is data driven.
  • Document CN102560143A discloses a flash smelting method and system for copper and relates to a copper matter production technology.
  • the flash smelting method and system are used for realizing higher control accuracy in comparison with the prior art, effectively improving the operation rate of a flash furnace and the production stability, stabilizing the grade of copper matte and leading the standard derivation of the copper matte to be reduced as well as increasing the comprehensive recovery rate of smelting.
  • the flash smelting method comprises a feed-forward data processing step, a feedback data processing step and a technology parameter adjustment step, wherein in the feed-forward data processing step, an initial value of the technology parameter needed for charging amount is worked out according to a planned value of a target parameter; in the feedback data processing step, a measured value of the target parameter of the flash furnace is collector, the measured value and the planned value are calculated and processed, and the compensation dosage of the technology parameter is worked out; in the technology parameter adjustment step, the initial value and the compensation dosage are combined, correction computation is carried out on the technology parameter to obtain a corrected value of the technology parameter; and a control system of the flash furnace implements a charging technology parameter according to the corrected value.
  • the flash smelting method is mainly used for producing and purifying the copper matte, i.e. an intermediate product in copper refining. This document also refers to the smelting process and is data driven.
  • Document CN114783540A discloses a multi-component alloy performance prediction method based on a particle swarm optimization backpropagation (BP) neural network.
  • the method comprises the steps that 1 , a database is established according to material component-performance historical data; 2, establishing and training a particle swarm optimization BP neural network component-performance prediction model, and taking a decision coefficient as an evaluation criterion of a prediction effect; 3, substituting an operation set obtained by the prediction model trained in the step 2 into the genetic optimization BP neural network, and taking a decision coefficient as an evaluation standard of a prediction effect; 4, if the decision coefficient calculated in the step 3 meets the judgment standard, substituting the to-be-predicted multi-component alloy components and the microalloying element array into the particle swarm optimization BP neural network obtained in the step 2 to complete performance prediction; otherwise, repeating the steps 2 and 3.
  • BP particle swarm optimization backpropagation
  • the prediction method is suitable for performance prediction of multi-component alloy and microalloyed alloy, is suitable for the research and development process of new material components, and can predict performance change influenced by smelting burning loss in industry.
  • This document refers to a data driven multi-component alloy prediction method and not a complete copper refining process.
  • Document CN113033704B refers to an intelligent judgment method for the end point of the copper converter converting period based on pattern recognition.
  • the obtained blister copper image is preprocessed, and then the VGG16 network is used to construct a copper period blowing state recognition model to predict the blister copper image in the later stage of converter converting.
  • extract the feature information of the blister copper image in the late converter converting stage as the input of the support vector machine prediction model and build a support vector machine prediction model based on particle swarm optimization to predict the end point of the copper converter converting period.
  • the invention can avoid errors in manually judging the end point of the copper converter blowing and copper making period, and effectively improves the hit rate of the copper converter smelting end point, thereby improving production efficiency and reducing costs.
  • Document CN116822380A discloses a collaborative optimization method for tail gas recycling in the copper smelting process based on digital twinning and belongs to the field of intelligent control of metal smelting tail gas recycling.
  • the method comprises the following steps: firstly, collecting state parameters and process parameters of a sulfuric acid reaction tower to obtain complete data information of the whole production process, secondly, constructing a digital twinborn model of the desulfurization acid production process flow according to the data, and continuously iterating by utilizing the constructed digital twinborn model of the production process flow through a BPNLP neural network to obtain optimal operation parameters.
  • the data is analyzed and compared with real-time data of entity equipment, and the acquired data is uploaded to a human-computer interaction interface through a communication protocol, so that virtual-real linkage between the whole industrial production and a client is realized; and finally, performing real-time optimization control on operation parameters of an actual production flow process through a process parameter control system according to an optimal parameter result, so as to realize 'controlling real by virtual', and ensure the desulfurization efficiency and the maximization of the sulfuric acid preparation speed for a long time.
  • This document refers to recycling tail gas in a copper smelting process, primarily during copper production and not in a copper refining process. Furthermore, the disclosed method bases real data and is hence data driven.
  • Document CN116705211 B provides an online prediction method and system for the copper loss rate of an oxygen-rich copper melt pool based on digital twins.
  • the method of the present invention can realize data sharing between physical equipment and virtual equipment in the smelting process, thereby realizing virtual and real linkage between physical equipment and virtual equipment. In this way, the operating parameters are continuously optimized, and finally the real-time prediction of the copper loss rate during the smelting process is achieved and kept within the target range.
  • This document again targets a primary copper smelting process only and iis data driven.
  • Document CN110362044B relates to the technical field of copper ore flotation, in particular to a limestone addition amount prediction control system and method of a copper ore flotation device. This document hence refers to a data driven primary copper production process.
  • Document CN104328285A relates to a hybrid-model-based on-line estimation method of oxygen-enriched bottom blowing copper smelting process parameters.
  • the method comprises the following steps: collecting key parameters of a realtime production process of an oxygen-enriched bottom blowing copper smelting process, inputting off-line detection key process parameters by man-machine interaction, storing the parameters in a data storage platform, determining the structure of an on-line estimation model, estimating a scaling factor based on the history data of the actual production process and the actual operation experience, determining to-be-determined parameters in the actual production process and building the on-line estimation model with hybrid mechanism and experience; operating the on-line estimation model, comparing the key process parameter for solving the model with the actual production data and detecting a result of the on- line estimation model.
  • the hybrid-model-based on-line estimation method of the oxygen-enriched bottom blowing copper smelting process parameters is assisted for mastering the production condition in a furnace and guiding the production operation and is also assisted for predicting products of the melting process, discharging slag and easily warning the copper matte discharging operation; the industrial automation level is improved; the intelligent determination of the bottom blowing cooper making process is achieved.
  • This document refers to a oxygen- enriched bottom blowing furnace smelting process of a data driven primary copper production process.
  • Document CN101353729A relates to an intelligent integrated modeling method based on working condition judgment, which establishes a matte grade prediction model based on the working condition judgment during the flash smelting process.
  • the invention selects the ingredient, blast volume and oxygen demand of material sent into a furnace as the input of a model after analyzing copper flash smelting production technology and relevant factors influencing matte grade to predicate the matte grade.
  • a mechanism model of the matte grade is established based on material balance
  • a fuzzy neural network model of the matte grade is established by utilizing historical data
  • the integrated prediction model of the matte grade is established by adopting an intelligent coordination strategy on the basis of the judgment of the working condition stability situation.
  • this object is solved by the method according to claim 1 , the system according to claim 18 and the computer program according to claim 19.
  • the invention refers to a method for controlling and optimizing a copper refining process, wherein the copper refining process comprises at least the steps of smelting, oxidation and reduction, wherein the method comprises the steps of: receiving at least the following input information regarding the copper refining process: copper scrap weight or molten copper weight, composition analysis of the copper scrap or molten copper, available fluxes, oxidation and reduction gas rates, feed temperature and time schedule regarding the copper refining process; defining or receiving the desired output of the copper refining process, which bases on the intended use of the copper and the impurities of the copper; creating a separate physics-based surrogate model for each copper refining process step, wherein each physics-based surrogate model represents the respective real-world copper refining process step and bases on thermochemical and solution chemistry calculations and a process model of a reactor that performs the respective copper refining process step; simulating the copper refining process using the created separate physics-based surrogate models by feeding the input information and the
  • the method according to the present invention preferably relates to a complete refining process comprising at least the steps of smelting, oxidation and reduction and not only to a first copper production process.
  • the input information regarding the copper refining process like for example copper scrap weight or molten copper weight, composition analysis of the copper scrap or molten copper, available fluxes, oxidation and reduction gas rates, feed temperature and time schedule regarding the copper refining process, is provided from previous processes, a corresponding process control, production control or similar processes or control instances.
  • the desired output depends on customer orders and is provided for example by a production planning system, which manages incoming customer orders and takes care of a timely production and delivery of the respective orders.
  • a surrogate model is a mathematical model of the outcome of the respective process step based on certain input variables.
  • the surrogate model is a simplified representation of the underlying complex process. It approximates the behavior of the original process using a computationally more efficient model than a physical process model.
  • the surrogate model is a deep learning model, which has been trained based on physical and thermochemical relationships of the available elements and features.
  • Each physics-based surrogate model represents the respective real-world copper refining process step. It is physics-based because it bases on thermochemical and solution chemistry calculations and a process model of a reactor that performs the respective copper refining process step, i.e. it relies on the physics of the respective process step and not on evaluating past data from real processes. Due to the physics-based surrogate models, it is possible to control and optimize the complete secondary copper refining process. Since such a copper refining process lasts up to 24 hours and depends on many input variables, different process steps and hence multiple internal process step states, data driven process models would require excessive computing power and memory, if they can handle such long and complex processes at all.
  • the created surrogate models are physics-based because they each base on thermochemical and solution chemistry calculations and a process model of a reactor that performs the respective copper refining process step.
  • the Software FactSageTM is used for the thermochemical and solution chemistry calculations.
  • the complete copper refining process is simulated by using the created separate physics-based surrogate models by feeding the input information and the desired output according to the received time schedule to the created separate physics-based surrogate models, wherein the output of the separate physics-based surrogate models is used for mapping the real copper refining process into rapid real-time calculations.
  • the simulation hence provides a real-time output for controlling and optimizing the complete copper refining process for certain input and output variables.
  • the separate physics-based surrogate models can be interconnected with each other by using the output of a physics-based surrogate model referring to a copper refining step as a further input in a physics-based surrogate model referring to a subsequently following copper refining step.
  • the method comprises the step of optimizing the copper refining process to achieve the desired output, particularly a desired copper quality, based on the simulations by evaluating for each process step an optimal flux combination, optimal process time, minimal costs and/or an optimal oxidation and/or reduction gas rate for the copper refining process based on the separate physics-based surrogate models.
  • the copper refining process further comprises a copper scrap optimizing step and/or a burner optimizing step.
  • the copper scrap optimizing step is located before the smelting step and the burner optimizing step is located after the reduction step.
  • the sequence of process steps is: scrap optimizing, smelting, oxidation, reduction and burner optimization.
  • a separate physics-based surrogate model is created for the copper scrap optimizing step and/or the burner optimizing step.
  • these steps can be included in the simulation and hence are part of the control and optimization of the complete copper refining process.
  • each process model of the corresponding reactor that perform the respective copper refining process step comprises multiple calculation zones.
  • each physicsbased surrogate model bases on thermochemical and solution chemistry calculations and a process model of a reactor that performs the respective copper refining process step.
  • This process model of the respective reactor advantageously comprises multiple calculation zones, i.e. the process in the reactor is divided into several zones, which interact with each other. In this way, the complex process in the reactor can be modeled more accurately, i.e. closer to reality, and efficiently.
  • the interconnection of the multiple calculation zones is optimized by an appropriate optimization algorithm. This increases the accuracy of process model of the corresponding reactor and hence also of the corresponding physics-based surrogate model.
  • the separate surrogate models for each copper refining process step are interconnected with each other by interlinking the process models of the corresponding reactors with each other that perform the copper refining process steps.
  • the interlinking of the process models is optimized by an appropriate optimization algorithm.
  • This optimization algorithm can be based on machine learning, particularly a neural network.
  • the separate surrogate models are implemented as neural networks, particularly based on optimization algorithms.
  • the simulations additionally provide information of the slag weight and composition, the off-gas composition, the solid weights and/or further outputs of the copper refining process. This will prevent over oxidation, over reduction and reducing carbon dioxide footprint. In other words, it helps in having a more efficient refining process and more efficient energy consumption.
  • the output of the simulations relates to the amounts or volumes of the copper refining process, particularly the resulting copper mass, the output composition for different stages of the copper refining process, particularly liquid, slag, gas and solid stages.
  • the separate physics-based surrogate models are trained using data generated based on the underlying physics and/or the underlying reactor technology, particularly zone based theoretical thermochemical and solution chemistry calculations. This makes sure that the process know-how has been included, like for example zone stream exchanges, such as off gas stream exchange between zones. It further closes the mass and energy balance in an accurate way and prevents the surrogate models from being a data cruncher only, as it will retain the rigor of thermodynamics in this way.
  • the separate models are continuously trained during the execution of the method based on the inputs and results of the real copper refining process.
  • the method further comprises the step of closing the mass balance of the copper refining process based on the statistical calculations of data reconciliation by considering measurement errors and/or model errors.
  • the method according to the present invention relies on the input data, which is at least partially captured by sampling and/or measurements. Hence, the input data is not completely free of errors.
  • the separate physics-based surrogate models introduce certain errors into the control and optimization method. Due to these errors the elemental mass balance is not closed. Thus, a statistical technique is used to close the mass balance by accounting for errors.
  • the step of closing the mass balance of the copper refining process uses a machine-learning model, which is continuously trained for identifying gross measurement errors in copper scrap or liquid phase compositions.
  • the method is executed in real-time.
  • the method uses dynamic simulations and/or optimizations, which provide intermediate results for each step of the copper refining process.
  • dynamic simulations the input variables are varied to optimize the output variables, particularly to achieve the desired output with an optimal input, reduced additives (fluxes), minimal energy and/or other optimization objectives.
  • the step of optimizing the copper refining process further uses an artificial intelligence (Al) learning algorithm for detecting a gross error of the copper refining process and particularly an interconnection between different process steps of the copper refining process.
  • This Al learning algorithm is learning in real-time and constantly improving.
  • the Al learning algorithm understands the gross error of the whole copper refining process and also the details of this gross error.
  • the Al learning algorithm has information about connections between zones in the reactor models and possible inefficiencies thereof and interlinks this information to scrap variability or other aspects of the copper refining process.
  • the object is further solved by a system for controlling a copper refining process, wherein the copper refining process comprises at least the steps of smelting, oxidation and reduction, wherein the system is configured to implement the method according to the invention.
  • the object is solved by a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method according to the invention.
  • the method according to the invention can be adapted to control and optimize complete refining processes of other metals from the 8 th to 14 th group of the table of periodic elements, particularly aluminum, lead or tin. Similarly, the method could also be used with respect to ferroalloys.
  • Fig. 1 a schematic view of a first embodiment of a method for controlling and optimizing a copper refining process according to the invention
  • Fig. 2 a schematic view of a second embodiment of a method for controlling and optimizing a copper refining process according to the invention
  • Fig. 3 a schematic view of a process model of a reactor that performs a copper refining process step.
  • Fig. 1 shows a schematic view of a first embodiment of a method 1 for controlling and optimizing a copper refining process according to the invention.
  • the copper refining process shown in Fig. 1 comprises the steps of smelting 2, oxidation 3 and reduction 4. As shown in Fig. 1 , the oxidation step 3 can be repeated multiple times.
  • the method is advantageously executed in real-time.
  • the copper refining process uses the following input information 5: copper scrap weight or molten copper weight (metal weight), slag weight, composition analysis of the copper scrap or molten copper (metal composition), available fluxes (flux ranges), oxidation and reduction gas rates, feed temperature and time schedule regarding the copper refining process (not shown in Fig. 1 ).
  • the input information regarding the copper refining process like for example copper scrap weight or molten copper weight, slag weight, composition analysis of the copper scrap or molten copper, available fluxes, oxidation and reduction gas rates, feed temperature and time schedule regarding the copper refining process, is provided from previous processes, a corresponding process control, production control or similar processes or control instances.
  • the metal weight, slag weight and metal composition are dependent variables between the process steps, i.e. the output of the first process step is used as input in the second process step, and so on.
  • the variables temperature, oxidation rate, reduction rate and flux ranges are process specific and do not depend on the same variables of previous of following process steps.
  • the desired output 6 is defined or received from outside, like a production planning or process control.
  • the desired output 6 bases at least on intended use of the copper and the impurities of the copper, and particularly on customer orders and is provided for example by a production planning system, which manages incoming customer orders and takes care of a timely production and delivery of the respective orders.
  • Each physics-based surrogate model 7, 8, 9 represents the respective real-world copper refining process step 2, 3, 4 and bases on thermochemical and solution chemistry calculations and a process model 10, 11 , 12 of a reactor that performs the respective copper refining process step 2, 3, 4.
  • a corresponding process model 10, 11 , 12 of a reactor that performs the respective copper refining process step 2, 3, 4 is created.
  • a surrogate model 7, 8, 9 is a mathematical model of the outcome of the respective process step 2, 3, 4 based on certain input variables 5.
  • the surrogate model 7 ,8 ,9 is a simplified representation of the underlying complex process. It approximates the behavior of the original process using a computationally more efficient model than a physical process model.
  • the surrogate model 7, 8, 9 is a deep learning model, which has been trained based on physical and thermochemical relationships of the available elements and features.
  • Each physics-based surrogate model 7, 8, 9 represents the respective real-world copper refining process step 2, 3, 4. It is physics-based because it bases on thermochemical and solution chemistry calculations and a process model 10, 11 , 12 of a reactor that performs the respective copper refining process step 2, 3, 4, i.e. it relies on the physics of the respective process step 2, 3, 4 and not on evaluating past data from real processes. Due to the physics-based surrogate models 7, 8, 9, it is possible to control and optimize the complete copper refining process.
  • the created surrogate models 7, 8, 9 are physics-based because they each base on thermochemical and solution chemistry calculations and a process model 10 ,11 , 12 of a reactor that performs the respective copper refining process step 2, 3, 4.
  • the Software FactSageTM is used for the thermochemical and solution chemistry calculations.
  • each process model 10, 11 , 12 of the corresponding reactors that perform the respective copper refining process step 2, 3, 4 comprises multiple calculation zones 17.
  • the interconnection of the multiple calculation zones 17 is for example optimized by an appropriate optimization algorithm.
  • the separate surrogate models 7, 8, 9 for each copper refining process step 2, 3, 4 are preferably interconnected with each other by interlinking the process models 10, 11 , 12 of the corresponding reactors with each other that perform the copper refining process steps 2, 3, 4.
  • the interlinking of the process models 10, 11 , 12 can be optimized by another appropriate optimization algorithm.
  • the method 1 simulates the complete copper refining process using the created separate physics-based surrogate models 7, 8, 9 by feeding the input information 5 and the desired output 6 according to the received time schedule to the created separate physics-based surrogate models 7, 8, 9.
  • the output of the separate physics-based surrogate models 7, 8, 9 is used for mapping the real copper refining process into rapid realtime calculations.
  • the simulations additionally provide information of the slag weight and composition, the off-gas composition, the solid weights and/or further outputs of the copper refining process.
  • the output of the simulations relates to the amounts or volumes of the copper refining process, particularly the resulting copper mass, the output composition for different stages of the copper refining process, particularly liquid, slag, gas and solid stages.
  • the complete copper refining process is optimized, to achieve the desired output 6, particularly a desired copper quality.
  • the optimization bases based on the simulations by evaluating for each process step 2, 3, 4 an optimal flux combination, optimal process time, minimal costs and/or an optimal oxidation and/or reduction gas rate for the copper refining process based on the separate physics-based surrogate models 7, 8, 9.
  • the separate surrogate models 7, 8, 9 are implemented as neural networks, particularly based on optimization algorithms.
  • the method 1 uses dynamic simulations and/or optimizations, which provide intermediate results for each step 2, 3, 4 of the copper refining process.
  • the separate physics-based surrogate models 7, 8, 9 are trained using data generated based on the underlying physics and/or the underlying reactor technology, particularly zone based theoretical thermochemical and solution chemistry calculations.
  • the separate physics-based surrogate models 7, 8, 9 are continuously trained during the execution of the method 1 based on the inputs 5 and results of the real copper refining process.
  • Fig. 2 shows a schematic view of a second embodiment of a method 1 for controlling and optimizing a copper refining process according to the invention.
  • This second embodiment differs from the first embodiment of Fig. 1 in that the copper refining process further comprises a copper scrap optimizing step 13 and a burner optimizing step 14.
  • the copper scrap optimizing step 13 is arranged before the smelting step2 and the burner optimizing step 14 is arranged after the reduction step 4.
  • the input 5 of the scrap optimizing step 13 relates for example to the market price of copper scrap, to copper scrap weight and/or to composition analysis of the copper scrap.
  • a separate physics-based surrogate model 15 is created for the copper scrap optimizing step 13 and a separate physics-based surrogate model 16 is created for the burner optimizing step 14.
  • these steps can be included in the simulation and hence are part of the control and optimization of the complete copper refining process.
  • the second embodiment of Fig. 2 further comprises the step of closing the mass balance of the copper refining process based on the statistical calculations of data reconciliation by considering measurement errors and/or model errors.
  • the method 1 according to the present invention relies on the input data 5, which is at least partially captured by sampling and/or measurements. Hence, the input data 5 is not completely free of errors. Furthermore, the separate physics-based surrogate models 7, 8, 9, 15, 16 introduce certain errors into the control and optimization method 1. Due to these errors the elemental mass balance is not closed. Thus, a statistical technique is used to close the mass balance by accounting for errors.
  • Fig. 3 shows a schematic view of a process model 10, 11 , 12 of a reactor that performs a copper refining process step 2, 3, 4.
  • the process model 10, 11 , 12 of Fig. 3 refers to a tilting refining furnace (TRF) as reactor.
  • TRF tilting refining furnace
  • the shown concept can be generalized for other reactor (furnace) types.
  • the terms reactor and furnace are used interchangeably.
  • the process model 10, 11 , 12 of the corresponding reactor that perform the respective copper refining process step 2, 3, 4 comprises multiple calculation zones 17, as shown in Fig. 1 and 2.
  • a zone is defined based on a possible chemical reaction that can occur in the reactor during corresponding the copper refining process step 2, 3, 4.
  • the zones 17 and their interconnection depend on the reactor (furnace) type, the copper refining process step 2, 3, 4 executed in the reactor and the internal processes in the reactor. Therefore, the number of zones and the interconnection of the zones 17 can vary for different reactors and/or copper refining steps 2, 3, 4.
  • the shown process model 10, 11 , 12 comprises a hearth zone 19, a burner zone 20 and a free-board zone 21 .
  • the hearth zone 19 is a zone of the furnace where metal collects, reacts and is subject to high temperature. Depending on the furnace type, the hearth zone 19 can comprise multiple zones. In the hearth zone 19 the reaction between gasses, fluxes and metal takes place.
  • the input of the hearth zone 19 is for example: copper scrap or molten copper, fluxes and Tuyere-gas.
  • the output of the hearth zone is for example: slag, refined copper and hearth-offgas.
  • the burner zone 20 is a zone in the furnace where combustion reaction takes place. In other words, the burner zone 20 provides the required heat for various stages of the copper refining process.
  • the input to the burner zone 20 is for example burner gas and the output of the burner zone is burner-offgas.
  • the free-board zone 21 is a zone where the gas phase reactions take place.
  • the free-board zone 21 refers to an interaction between the hearth zone 19 and the burner zone 20.
  • the input of the free-board zone 21 is for example hearth-offgas and burner-offgas and the output is offgas in general.
  • the tilting refining furnace (TRF) of Fig. 3 can be used for smelting 2, oxidation and reduction 4 process step.

Landscapes

  • Engineering & Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Manufacturing & Machinery (AREA)
  • Materials Engineering (AREA)
  • Mechanical Engineering (AREA)
  • Metallurgy (AREA)
  • Organic Chemistry (AREA)
  • Automation & Control Theory (AREA)
  • Manufacture And Refinement Of Metals (AREA)

Abstract

L'invention concerne un procédé (1) de commande et d'optimisation d'un procédé de raffinage de cuivre comprenant les étapes de fusion (2), d'oxydation (3) et de réduction (4), le procédé comprenant les étapes consistant à : recevoir des informations d'entrée (5) concernant le processus de raffinage de cuivre ; définir ou recevoir la sortie souhaitée (6) du procédé de raffinage de cuivre ; créer un modèle de substitution basé sur la physique séparé (7, 8, 9) pour chaque étape de traitement de raffinage de cuivre (2, 3, 4), chaque modèle de substitution basé sur la physique (7, 8, 9) représentant l'étape de traitement de raffinage de cuivre du monde réel respective (2, 3, 4) et des bases sur des calculs de chimie thermochimique et de solution et un modèle de traitement (10, 11, 12) d'un réacteur qui effectue l'étape de traitement de raffinage de cuivre respective (2, 3, 4) ; simuler le processus de raffinage de cuivre à l'aide des modèles de substitution basés sur la physique séparés créés (7, 8, 9) en fournissant les informations d'entrée (5) et la sortie souhaitée (6) en fonction du calendrier reçu aux modèles de substitution basés sur la physique séparés créés (7, 8, 9) ; et optimiser le processus de raffinage de cuivre pour obtenir la sortie souhaitée (6) sur la base des simulations en évaluant pour chaque étape de traitement (2, 3, 4) une combinaison de flux optimale, un temps de traitement optimal, des coûts minimaux et/ou un taux de gaz d'oxydation et/ou de réduction optimal pour le processus de raffinage de cuivre sur la base des modèles de substitution basés sur la physique séparés (7, 8, 9).
PCT/EP2025/056367 2024-03-11 2025-03-10 Procédé de commande et d'optimisation d'un procédé de raffinage de cuivre, le procédé de raffinage de cuivre comprenant au moins les étapes de fusion, d'oxydation et de réduction Pending WO2025190825A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP24162750.4A EP4617387A1 (fr) 2024-03-11 2024-03-11 Procédé de commande et d'optimisation d'un processus de raffinage du cuivre, le processus de raffinage du cuivre comprenant au moins les étapes de fusion, d'oxydation et de réduction
EP24162750.4 2024-03-11

Publications (1)

Publication Number Publication Date
WO2025190825A1 true WO2025190825A1 (fr) 2025-09-18

Family

ID=90364623

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2025/056367 Pending WO2025190825A1 (fr) 2024-03-11 2025-03-10 Procédé de commande et d'optimisation d'un procédé de raffinage de cuivre, le procédé de raffinage de cuivre comprenant au moins les étapes de fusion, d'oxydation et de réduction

Country Status (2)

Country Link
EP (1) EP4617387A1 (fr)
WO (1) WO2025190825A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN121707078A (zh) * 2026-02-12 2026-03-20 福建万聚福冰雪体育科技有限公司 面向动态环境参数的生态冰制备工艺自优化方法及系统

Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2200090A1 (fr) * 1994-07-18 1996-02-01 David B. George Procede et appareil pour produire du cuivre noir
US20050218001A1 (en) * 2004-03-17 2005-10-06 You Eugene Y Monitoring electrolytic cell currents
CN101353729A (zh) 2008-07-18 2009-01-28 中南大学 一种基于工况判断的智能集成建模方法
CN102560143A (zh) 2010-12-30 2012-07-11 中国瑞林工程技术有限公司 铜闪速熔炼的方法以及系统
CN104328285A (zh) 2014-10-29 2015-02-04 中国科学院沈阳自动化研究所 基于混合模型的富氧底吹炼铜过程参数在线估计方法
US20180081339A1 (en) * 2016-09-21 2018-03-22 International Business Machines Corporation System and predictive modeling method for smelting process control based on multi-source information with heterogeneous relatedness
CN110000100A (zh) 2019-05-07 2019-07-12 中山市镭通激光科技有限公司 一种ftu自动插拔检测设备
CN111554353A (zh) 2020-04-10 2020-08-18 中国恩菲工程技术有限公司 氧气底吹炉铜熔炼过程参数在线预测方法
CN112000010A (zh) * 2020-07-30 2020-11-27 中国恩菲工程技术有限公司 氧气底吹铜熔炼过程反馈控制方法、装置及电子设备
CN112083694A (zh) 2020-07-30 2020-12-15 中国恩菲工程技术有限公司 氧气底吹铜吹炼过程反馈控制方法、装置及电子设备
WO2021203912A1 (fr) 2020-04-10 2021-10-14 中国恩菲工程技术有限公司 Procédé de prédiction en ligne de paramètres dans un processus de conversion de cuivre basé sur un four de soufflage à fond d'oxygène
CN114783540A (zh) 2022-03-17 2022-07-22 中国科学院金属研究所 一种基于粒子群优化bp神经网络的多元合金性能预测方法
CN114822723A (zh) * 2022-03-14 2022-07-29 中国恩菲工程技术有限公司 精炼阳极铜反应终点预测方法及装置
CN110362044B (zh) 2019-06-03 2022-09-06 齐鲁工业大学 一种铜矿浮选装置的石灰石添加量预测控制系统及方法
CN116258087A (zh) 2023-05-15 2023-06-13 矿冶科技集团有限公司 冰铜品位软测量方法、装置、电子设备及存储介质
CN116822380A (zh) 2023-08-29 2023-09-29 昆明理工大学 一种基于数字孪生在铜冶炼过程中尾气回收利用的协同优化方法
CN116705211B (zh) 2023-08-04 2023-10-10 昆明理工大学 基于数字孪生的富氧铜熔池铜损失率在线预测方法及系统
CN113033704B (zh) 2021-04-22 2023-11-07 江西理工大学 基于模式识别的铜转炉吹炼造铜期终点智能判断方法及系统

Patent Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2200090A1 (fr) * 1994-07-18 1996-02-01 David B. George Procede et appareil pour produire du cuivre noir
US20050218001A1 (en) * 2004-03-17 2005-10-06 You Eugene Y Monitoring electrolytic cell currents
CN101353729A (zh) 2008-07-18 2009-01-28 中南大学 一种基于工况判断的智能集成建模方法
CN102560143A (zh) 2010-12-30 2012-07-11 中国瑞林工程技术有限公司 铜闪速熔炼的方法以及系统
CN104328285A (zh) 2014-10-29 2015-02-04 中国科学院沈阳自动化研究所 基于混合模型的富氧底吹炼铜过程参数在线估计方法
US20180081339A1 (en) * 2016-09-21 2018-03-22 International Business Machines Corporation System and predictive modeling method for smelting process control based on multi-source information with heterogeneous relatedness
CN110000100A (zh) 2019-05-07 2019-07-12 中山市镭通激光科技有限公司 一种ftu自动插拔检测设备
CN110362044B (zh) 2019-06-03 2022-09-06 齐鲁工业大学 一种铜矿浮选装置的石灰石添加量预测控制系统及方法
WO2021203912A1 (fr) 2020-04-10 2021-10-14 中国恩菲工程技术有限公司 Procédé de prédiction en ligne de paramètres dans un processus de conversion de cuivre basé sur un four de soufflage à fond d'oxygène
CN111554353A (zh) 2020-04-10 2020-08-18 中国恩菲工程技术有限公司 氧气底吹炉铜熔炼过程参数在线预测方法
CN112000010A (zh) * 2020-07-30 2020-11-27 中国恩菲工程技术有限公司 氧气底吹铜熔炼过程反馈控制方法、装置及电子设备
CN112083694A (zh) 2020-07-30 2020-12-15 中国恩菲工程技术有限公司 氧气底吹铜吹炼过程反馈控制方法、装置及电子设备
CN113033704B (zh) 2021-04-22 2023-11-07 江西理工大学 基于模式识别的铜转炉吹炼造铜期终点智能判断方法及系统
CN114822723A (zh) * 2022-03-14 2022-07-29 中国恩菲工程技术有限公司 精炼阳极铜反应终点预测方法及装置
CN114783540A (zh) 2022-03-17 2022-07-22 中国科学院金属研究所 一种基于粒子群优化bp神经网络的多元合金性能预测方法
CN116258087A (zh) 2023-05-15 2023-06-13 矿冶科技集团有限公司 冰铜品位软测量方法、装置、电子设备及存储介质
CN116705211B (zh) 2023-08-04 2023-10-10 昆明理工大学 基于数字孪生的富氧铜熔池铜损失率在线预测方法及系统
CN116822380A (zh) 2023-08-29 2023-09-29 昆明理工大学 一种基于数字孪生在铜冶炼过程中尾气回收利用的协同优化方法

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
ANDERSON ALEXANDRA ET AL: "A Review of Computational Capabilities and Requirements in High-Resolution Simulation of Nonferrous Pyrometallurgical Furnaces", JOM: JOURNAL OF METALS, SPRINGER NEW YORK LLC, UNITED STATES, vol. 74, no. 4, 22 February 2022 (2022-02-22), pages 1543 - 1567, XP037752159, ISSN: 1047-4838, [retrieved on 20220222], DOI: 10.1007/S11837-022-05169-4 *
JAROSZ PIOTR ET AL: "Metamodeling and Optimization of a Blister Copper Two-Stage Production Process", JOM: JOURNAL OF METALS, SPRINGER NEW YORK LLC, UNITED STATES, vol. 68, no. 6, 12 April 2016 (2016-04-12), pages 1535 - 1540, XP035775906, ISSN: 1047-4838, [retrieved on 20160412], DOI: 10.1007/S11837-016-1916-Z *
MOSKALYK R R ET AL: "Review of copper pyrometallurgical practice: today and tomorrow", MINERALS ENGINEERING, ELSEVIER, AMSTERDAM, NL, vol. 16, no. 10, 1 October 2003 (2003-10-01), pages 893 - 919, XP002743501, ISSN: 0892-6875, [retrieved on 20030917], DOI: 10.1016/J.MINENG.2003.08.002 *
WANG YANCHAO ET AL: "Simulation of flows of hazardous elements in copper smelting process based on Bayesian network", JOURNAL OF CLEANER PRODUCTION, 10 November 2022 (2022-11-10), XP093197355, Retrieved from the Internet <URL:https://doi.org/10.1016/j.jclepro.2022.135137> DOI: 10.1016/j.jclepro.2022.135137 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN121707078A (zh) * 2026-02-12 2026-03-20 福建万聚福冰雪体育科技有限公司 面向动态环境参数的生态冰制备工艺自优化方法及系统

Also Published As

Publication number Publication date
EP4617387A1 (fr) 2025-09-17

Similar Documents

Publication Publication Date Title
CN108676955B (zh) 一种转炉炼钢终点碳含量和温度控制方法
US8002871B2 (en) Methods and apparatus for an oxygen furnace quality control system
CN109447346B (zh) 基于灰色预测与神经网络组合模型转炉氧耗量预测方法
CN115456264B (zh) 一种中小型转炉的终点碳含量和终点温度预测方法
CN106096788B (zh) 基于pso_elm神经网络的转炉炼钢工艺成本控制方法及系统
CN113919206A (zh) 一种炼钢合金化窄成分控制方法
CN114021913A (zh) 一种基于差分进化算法的高炉配料优化方法、电子设备以及存储介质
WO2025190825A1 (fr) Procédé de commande et d&#39;optimisation d&#39;un procédé de raffinage de cuivre, le procédé de raffinage de cuivre comprenant au moins les étapes de fusion, d&#39;oxydation et de réduction
CN118016203B (zh) 一种基于机理模型和XGBoost算法的LF精炼温度预报方法
Liu et al. Establishment and application of steel composition prediction model based on t-distributed stochastic neighbor embedding (t-SNE) dimensionality reduction algorithm
CN116384254A (zh) 一种用于lf精炼炉钢水温度预测的误差修正方法及系统
Tang et al. An estimation of distribution algorithm with resampling and local improvement for an operation optimization problem in steelmaking process
CN115081305B (zh) 多工况级联冶金反应过程氧化还原电位优化设定方法
Niyayesh et al. Predicting endpoint parameters of electric arc furnace–based steelmaking using artificial neural network
Liu et al. XGBoost-based model for predicting hydrogen content in electroslag remelting: YX Liu et al.
Zhang et al. Prediction of lime utilization ratio of dephosphorization in BOF steelmaking based on online sequential extreme learning machine with forgetting mechanism
CN118070645A (zh) 一种基于BiLSTM-CNN的电炉终点温度预测方法
CN119956017A (zh) 一种面向低碳低成本冶炼的转炉炼钢工艺智能在线优化方法
CN119940656B (zh) 一种钢铁行业炼钢工艺优化方法、设备及存储介质
CN110059940B (zh) 一种炼钢-连铸界面衔接节能方法及系统
Zhang et al. Steel Scrap Yield Prediction in Basic Oxygen Steelmaking Based on Random Forest and Neural Networks
CN114418464B (zh) 一种基于案例推理的二步法冶金流程能源评价方法及系统
CN106971079B (zh) 沉铁过程出口亚铁离子浓度的协调优化设定方法及装置
CN105242545B (zh) 一种基于工艺批次模型的转炉用氧流量的预测方法
WANG et al. Research progress on intelligent control and decision-making models for the ladle furnace refining process

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 25711458

Country of ref document: EP

Kind code of ref document: A1