A high-oxygen-enriched double-side blown smelting process optimization system and method based on digital twinning
By constructing a multi-scale coupling mechanism digital twin module and an online calibration module, the high-oxygen-enriched double-sided blowing smelting process was optimized, solving the problems of copper loss in slag and furnace lining erosion, and achieving efficient production optimization and safety improvement.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHIFENG YUNTONG NON FERROUS METAL CO LTD
- Filing Date
- 2026-01-19
- Publication Date
- 2026-06-05
AI Technical Summary
Existing technologies struggle to effectively control slag properties and furnace lining safety within the molten pool under high oxygen-enriched conditions, leading to increased copper loss in the slag and exacerbated furnace lining erosion. The lack of deep coupling between multi-scale, multi-physical fields and key chemical reaction kinetics makes it difficult to achieve root cause analysis and forward-looking optimization.
A multi-scale coupled mechanism digital twin module was constructed, including microscopic chemical reaction kinetics, mesoscopic slag-gold transport and separation, and dynamic slag layer sub-modules. Combined with a multi-objective optimization module and an online calibration module, the oxygen concentration, slag-forming agent ratio, and blasting parameters were optimized through digital twin simulation to achieve synergistic optimization of copper loss in slag and slag layer thickness.
It has achieved a reduction in copper loss in slag and stable furnace lining safety, shortened the optimization cycle, enhanced the flexibility and robustness of the production process, supported rapid process decision-making and risk warning, and improved production intensity and safety.
Smart Images
Figure CN122154391A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent control technology for copper smelting, and in particular to a simulation and optimization system and method for the entire process of double-sided blowing pool smelting based on digital twins. Background Technology
[0002] Double-sided blowing smelting is a mainstream technology for efficient smelting of non-ferrous metals such as copper and nickel. By blowing oxygen-enriched air into the molten pool, this technology can significantly improve reaction intensity, processing capacity, and matte grade. However, when the oxygen concentration is increased to a high level to achieve intensified smelting, localized over-oxidation can easily occur inside the molten pool, causing the iron oxidation product to transform from low-melting-point ferrous oxide (FeO) to high-melting-point magnetite (Fe3O4). This triggers a series of chain reactions: increased slag viscosity and density severely hinder the settling and separation of matte droplets in the slag phase, leading to increased copper loss from the slag; simultaneously, the deteriorated slag properties make it difficult to form a stable and uniform slag protective layer on the furnace lining surface, exacerbating the erosion of refractory materials by the high-temperature melt and directly threatening the safety and lifespan of the furnace. Therefore, how to accurately control and mitigate the negative impacts on slag properties and furnace lining safety while enjoying the efficiency gains brought by high oxygen enrichment has become the core technical bottleneck for the optimization and intelligent upgrading of this process.
[0003] To achieve understanding and control of the smelting process, existing technologies have explored various aspects. For example, patent CN118171458A uses cold-state physical simulation and image recognition technology based on a water model to study the flow field characteristics and macroscopic material migration laws within the molten pool. This method is valuable for understanding flow characteristics, but it completely ignores the chemical reaction kinetics at high temperatures, the dynamic changes in slag properties, and the formation mechanism of the slag layer. Its conclusions are difficult to directly guide real smelting processes involving complex physicochemical changes. With the improvement of computing power, methods based on numerical simulation and data-driven approaches have been developed. For example, patent CN119720847B discloses a deep learning-based intelligent prediction system for side-blown furnaces, which predicts the furnace state through sensor data, CFD simulation, and graph neural networks. These methods represent progress towards data-driven intelligence, but their core models still focus on prediction through data correlation and macroscopic physical field simulation, lacking deep embedding and dynamic calculation of the multi-scale strongly coupled mechanism of "microscopic chemical reaction generating Fe3O4—affecting mesoscopic slag viscosity—determining macroscopic slag-gold separation efficiency and slag-hanging behavior." Furthermore, while traditional thermodynamic equilibrium calculation software (such as FactSage) can predict the phase composition under equilibrium conditions, it cannot describe the non-equilibrium kinetic processes that determine reaction rates and transport efficiency in actual molten pools.
[0004] In general, existing technologies either overemphasize physical simulation while neglecting chemistry, or rely on data correlation but lack interpretable mechanistic depth. They have not yet been able to construct a high-fidelity digital model that deeply couples multi-scale, multi-physics fields with key chemical reaction kinetics and can be continuously calibrated through online data. Therefore, it is difficult to fundamentally analyze and proactively optimize the core contradictions under high-oxygen-enriched conditions. In view of this, a new technical solution is urgently needed to address these problems. Summary of the Invention
[0005] To address the aforementioned problems in the existing technology, this invention provides a digital twin optimization system for a high-oxygen-enriched double-sided blowing pool smelting process, comprising:
[0006] A multi-scale coupling mechanism digital twin module, configured to simulate the smelting process, includes: a microscopic chemical reaction kinetics submodule, used to dynamically calculate the Fe3O4 content in the slag based on the oxygen potential and temperature field within the molten pool; a mesoscopic slag-gold transport and separation submodule, used to calculate the dynamic slag viscosity based on the Fe3O4 content and predict the copper loss in the slag based on the dynamic slag viscosity; and a dynamic slag layer submodule, used to predict the thickness of the slag layer on the furnace lining based on the dynamic slag viscosity and the heat flow conditions of the furnace lining wall.
[0007] The multi-objective optimization module is communicatively connected to the multi-scale coupling mechanism digital twin module. It is configured to minimize the copper loss in the slag and / or stabilize the thickness of the slag layer as optimization objectives. Using the simulation results of the digital twin module, it performs synergistic optimization on at least two of the following: oxygen concentration, slag-forming agent ratio, and blasting parameters, and outputs one or more sets of optimized operating parameters.
[0008] Furthermore, the mesoscopic slag-gold transport and separation submodule is a population equilibrium model, in which the settling rate of matte droplets in the slag phase is modeled as a function of the dynamic slag viscosity.
[0009] Furthermore, the prediction of the dynamic slag layer submodule is also based on the solidification temperature of the slag, the melt flow rate near the furnace lining wall, and the temperature gradient.
[0010] Furthermore, it also includes an online calibration module, which is communicatively connected to the multi-scale coupling mechanism digital twin module and is configured with a data interface; wherein, the online calibration module acquires the production data of the physical melting furnace in real time through the data interface, compares the production data with the predicted data output by the multi-scale coupling mechanism digital twin module, and generates adjustment instructions for the parameters in the multi-scale coupling mechanism digital twin module based on the comparison results.
[0011] Furthermore, the online calibration module uses Bayesian inference or ensemble Kalman filtering algorithms for parameter adjustment.
[0012] Furthermore, the multi-objective optimization module employs a multi-objective evolutionary algorithm and conducts virtual experiments based on the multi-scale coupling mechanism digital twin module to obtain a Pareto optimal solution set as the optimized operating parameters.
[0013] This invention also provides a method for optimizing a high-oxygen-enriched double-sided blowing pool smelting process based on digital twins, comprising the following steps:
[0014] S1. Construct and run a multi-scale coupled mechanism digital twin model to simulate the smelting process, wherein:
[0015] The content of Fe3O4 in the slag is dynamically calculated based on the oxygen potential and temperature field in the molten pool.
[0016] The dynamic slag viscosity is calculated based on the Fe3O4 content, and the copper loss in the slag is predicted based on the dynamic slag viscosity.
[0017] Based on the dynamic slag viscosity, solidification temperature of the molten slag, melt flow rate and temperature gradient near the furnace lining wall, the thickness of the slag layer on the furnace lining is predicted.
[0018] S2. With the optimization objective of minimizing the copper loss in the slag and / or stabilizing the thickness of the slag layer, at least two of the following parameters—oxygen concentration, slag-forming agent ratio, and blasting parameters—are optimized collaboratively using the simulation results of the multi-scale coupling mechanism digital twin model.
[0019] S3. Output one or more sets of optimized operation parameters obtained from the optimization process.
[0020] Furthermore, in step S1, a group equilibrium model is used to simulate slag-gold transport and separation, wherein the settling rate of matte droplets in the slag phase is set as a function of the dynamic slag viscosity.
[0021] Furthermore, it also includes an online calibration step: acquiring production data of the physical smelting furnace in real time; comparing the prediction results of the multi-scale coupling mechanism digital twin model with the corresponding actual production data; and, based on the comparison results, adjusting the parameters used in the multi-scale coupling mechanism digital twin model for calculating Fe3O4 content and / or slag-gold separation in reverse to reduce prediction errors.
[0022] Compared with the prior art, the technical solution provided by the present invention can achieve the following significant beneficial effects:
[0023] (1) It achieves synergistic optimization of the core contradictory objectives of "reducing copper loss" and "protecting the furnace lining". By coupling Fe3O4 formation kinetics, slag viscosity prediction and slag layer growth model, the impact of changes in operating parameters on the two key indicators of "copper content in slag" and "furnace lining safety" can be directly quantified and evaluated. Based on this, multi-objective optimization can effectively reduce metal loss while ensuring the stability of the slag layer on the furnace lining, thus fundamentally solving the core bottleneck of high oxygen enrichment enhancement.
[0024] (2) Supporting rapid process decision-making and safe improvement of production intensity based on virtual simulation. Using a high-fidelity digital twin as a "virtual melting furnace", hundreds of process schemes can be simulated, tested and evaluated within hours, greatly shortening the optimization cycle that traditionally relies on on-site trial and error. This can not only be used for daily parameter fine-tuning, but also safely explore production limits, realizing a scientific and safe improvement of production intensity.
[0025] (3) It has formed an adaptive optimization and risk warning capability to cope with complex operating conditions. By integrating an online calibration module, the system can dynamically correct model parameters using real-time production data, so that the twin continuously approximates the real furnace conditions, ensuring the reliability of long-term predictions. On this basis, the system can perform forward-looking "hypothesis analysis" on unknown operating conditions such as fluctuations in raw material composition, and immediately recommend optimization schemes for adjusting the slagging agent ratio, guiding production to make proactive adjustments before entering the furnace, successfully avoiding the deterioration of indicators and operational failures, and significantly enhancing the flexibility and robustness of the production process in dealing with fluctuations. Attached Figure Description
[0026] Figure 1 This is a schematic diagram of the system architecture for optimizing the high-oxygen-enriched double-sided blown pool smelting process based on digital twins, according to an embodiment of the present invention.
[0027] Figure labeling: 100, Physical melting furnace; 200, Online calibration module; 300, Multi-scale coupling mechanism digital twin module; 310, Microscopic chemical reaction kinetics sub-module; 320, Mesoscopic slag-gold transport and separation sub-module; 330, Dynamic slag layer sub-module; 400, Multi-objective optimization module. Detailed Implementation
[0028] The present invention will be further described below with reference to embodiments, but the present invention is not limited to the following embodiments. Those skilled in the art will understand that various changes and modifications can be made to the present invention without departing from the spirit and scope thereof.
[0029] This invention provides a digital twin-based optimization system and method for high-oxygen-enriched double-sided blowing smelting processes, aiming to solve the contradictory problems of increased copper content in slag and intensified furnace lining erosion caused by molten pool over-oxidation under high-oxygen-enriched intensified smelting conditions. The system constructs a high-fidelity digital twin that interacts in real-time with the physical smelting furnace 100 and evolves dynamically. Through a closed loop of "mechanism modeling - online calibration - virtual simulation - intelligent optimization," it achieves in-depth process insight and forward-looking collaborative optimization.
[0030] The core of the system lies in its multi-scale coupling mechanism digital twin module 300. Specifically, its microscopic chemical reaction kinetics submodule 310 dynamically calculates the amount of Fe3O4 generated, based on the oxygen potential and temperature field obtained from computational fluid dynamics (CFD) simulation within the molten pool. This process incorporates experimentally verified FeO... x A thermodynamic database of the SiO2-CaO-MgO-Al2O3 multi-element slag system is used, and a FeS selective oxidation kinetic model is employed to accurately describe the formation rate of Fe3O4 under high oxygen potential. Next, a mesoscopic slag-gold transport and separation submodule 320 receives Fe3O4 content data and calculates the dynamically changing slag viscosity using a strong correlation model between Fe3O4 and slag viscosity (e.g., based on the Urbain model or experimental correlations). Based on this viscosity, this submodule further predicts the copper loss in the slag. In a preferred embodiment, this process uses a population equilibrium model (PBM) to simulate the behavior of matte droplets in the slag phase, where the settling rate of the matte droplets is set as a function of the current dynamic slag viscosity, thus realistically reproducing the mechanism by which slag viscosity hinders copper droplet settling and increases loss. Finally, the dynamic slag layer submodule 330 integrates dynamic slag viscosity, furnace lining wall heat flux density (which can be obtained through thermocouple measurement and heat transfer inverse problem calculation), slag solidification temperature, and melt flow velocity and temperature gradient near the wall to predict the growth and erosion dynamics of the protective slag layer on the furnace lining surface and quantitatively assess the safety status of the furnace lining.
[0031] The digital twin module is communicatively connected to a multi-objective optimization module 400. The optimization module focuses on minimizing copper loss in the slag and / or stabilizing the slag layer thickness as its core optimization objectives, utilizing the simulation results from the twin module as the basis for performance evaluation. It collaboratively optimizes at least two key process parameters, including oxygen concentration, slag-forming agent ratio (e.g., SiO2 / CaO / Al2O3 ratio), and blast parameters (e.g., air volume and air pressure). The optimization process typically employs a multi-objective evolutionary algorithm (e.g., NSGA-II), using the digital twin module as a "virtual smelting furnace" for rapid virtual testing, ultimately outputting a set of Pareto optimal solutions, providing operators with multiple optimized operating schemes that balance economy and safety.
[0032] To ensure the long-term consistency between the digital twin and the physical entity, the system also includes an online calibration module 200. This module acquires production data (such as raw material composition, actual slag matte composition, temperature, etc.) in real time through a data interface and compares it with the predicted values of the twin module. Using data assimilation algorithms such as Bayesian inference or ensemble Kalman filtering, it reverse-calibrates uncertain parameters such as the reaction rate constant and mass transfer coefficient in the mechanistic model, continuously improving the prediction accuracy of the twin and forming a self-learning intelligent system.
[0033] Example 1: Synergistic Optimization of Peroxidation Risk in High-Speed Rail Raw Materials
[0034] This embodiment focuses on a double-blown smelting furnace used in a factory to process complex copper concentrate containing high-iron (Fe content ~30%). Operating at a baseline oxygen enrichment concentration of 72%, it faces the risk of excessive Fe3O4 formation.
[0035] First, a 1:1 three-dimensional CFD model was constructed based on the factory drawings, integrating the aforementioned multi-scale mechanism model to form an initial digital twin. Using one month of historical data, the kinetic parameters were calibrated through an ensemble Kalman filter algorithm in the online calibration module, bringing the average prediction error of the model for key indicators (such as the copper content in the final slag) to within 8%.
[0036] After the system went online, it detected that the iron content of the raw material entering the furnace was too high, and automatically triggered the optimization process. The multi-objective optimization module set the following objectives: Objective 1, minimize the copper content in the slag; Objective 2, maximize the thickness of the slag layer on the sidewall. The decision variables were the oxygen concentration (70-75%) and the SiO2 / CaO ratio (3.1-3.5). The optimization module drove the digital twin to conduct more than 500 sets of virtual experiments.
[0037] Simulation results show that, under baseline parameters, the predicted copper content in the slag is 0.65%, and due to the increase in slag viscosity caused by Fe3O4 formation, the slag layer on the furnace wall opposite the spray gun is relatively thin. After optimization, the system recommends scheme A: slightly reducing the oxygen enrichment concentration to 71.5%, while increasing the SiO2 / CaO ratio from 3.2 to 3.4. Digital twin predictions indicate that under this scheme, the slag viscosity will decrease by approximately 15%, the Fe3O4 content will decrease, and the predicted copper content in the slag can be reduced to 0.58%. Simultaneously, due to improved slag fluidity and fine-tuning of heat flux, the predicted slag layer thickness in key areas can be increased by approximately 3 mm, and the heat flux density will decrease by 15%.
[0038] The factory adopted Option A and conducted a trial for one shift (8 hours). Actual production results showed that the average copper content in the slag during this shift was 0.60%, and the furnace wall infrared thermal imager showed a more uniform temperature distribution, indicating an improved slag layer condition. The online calibration module then used this actual data to fine-tune the model, making the model's predictions for this type of raw material more "personalized."
[0039] Example 2: Multi-constraint global optimization with the core objective of increasing processing capacity
[0040] Based on the calibrated digital twin system established in Example 1, this embodiment aims to maximize the copper concentrate processing capacity while ensuring product quality and furnace lining safety.
[0041] The optimization objective is set as maximizing the charging rate. Constraints include: slag copper content ≤ 1.5%, matte grade maintained at 58%-62%, and furnace top temperature ≤ 1350℃. Decision variables are the total charging rate, total blast volume, and oxygen enrichment concentration.
[0042] The system, through digital twin simulation, discovered that simply increasing the feeding rate leads to incomplete reaction, resulting in excessive copper content in the slag and SO2 concentration in the flue gas. A multi-objective optimization module, combined with surrogate modeling technology, seeks the optimal solution within a complex constraint space. Ultimately, the system provides a Pareto optimal solution set. The operator selects the robust option B: increasing the feeding rate from 85 tons / hour to 93.5 tons / hour (a 10% increase), while simultaneously increasing the total air volume by 8% and the oxygen enrichment concentration from 78% to 79.5%.
[0043] The digital twin predicted the following results for Scheme B: slag copper content 1.4%, matte grade 60.5%, furnace top temperature 1335℃, all within the constraints. After one week of actual operation, statistics showed that the average throughput reached expectations, and all technical indicators matched the predictions well, achieving a safe increase in production intensity.
[0044] Example 3: Adaptive optimization and forward-looking regulation to cope with fluctuations in high Al2O3 feedstock
[0045] This example demonstrates the system's "hypothesis analysis" and adaptive capabilities when raw material composition fluctuates drastically. In one batch of concentrate fed into the furnace, the Al2O3 content abnormally increased to 8% (normal is 4%).
[0046] The operator inputs the new raw material composition into the system's operational simulation interface. The digital twin completes a rapid simulation within 30 minutes, and the prediction report issues a warning: if the current operating parameters are used, the high Al2O3 will overlap with the high Fe3O4, causing the slag viscosity to rise sharply. It is predicted that the copper content in the slag will exceed 0.8%, and the deterioration of slag fluidity may lead to operational difficulties.
[0047] The system immediately triggered emergency optimization. The optimization objective was adjusted to control the copper content in the slag below 0.6% while maintaining normal slag fluidity. The decision variable focused primarily on the slag-forming agent ratio. Digital twin virtual experiments showed that adjusting the CaO / SiO2 ratio could effectively control the performance of the high-alumina slag system. The system recommended the following solution: while maintaining the oxygen enrichment concentration, adjust the CaO / SiO2 ratio from the conventional 0.8 to 0.7, and appropriately increase the MgO content.
[0048] The factory adjusted the batching according to the recommended scheme before the new material was fed into the furnace. During the actual production process, the smelting was stable, and the final slag copper content was 0.58%. This successfully avoided a potential deterioration in production indicators and process failures due to a sudden change in raw materials, fully demonstrating the forward-looking decision support value of the digital twin system.
[0049] In summary, the embodiments of the present invention, by constructing a digital twin system integrating multi-scale mechanism models, online calibration, and intelligent optimization, realize the visualization, predictability, and optimization of the complex process of high oxygen-enriched double-sided blowing pool smelting. This effectively addresses the contradiction between "quality improvement and consumption reduction" and "furnace safety," significantly enhancing the technical and economic indicators and the level of intelligence in production.
[0050] The above description is a preferred embodiment of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.
Claims
1. A digital twin optimization system for a high-oxygen-enriched double-sided blown pool smelting process, characterized in that, include: A multi-scale coupling mechanism digital twin module, configured to simulate the smelting process, includes: a microscopic chemical reaction kinetics submodule, used to dynamically calculate the Fe3O4 content in the slag based on the oxygen potential and temperature field within the molten pool; a mesoscopic slag-gold transport and separation submodule, used to calculate the dynamic slag viscosity based on the Fe3O4 content and predict the copper loss in the slag based on the dynamic slag viscosity; and a dynamic slag layer submodule, used to predict the thickness of the slag layer on the furnace lining based on the dynamic slag viscosity and the heat flow conditions of the furnace lining wall. The multi-objective optimization module is communicatively connected to the multi-scale coupling mechanism digital twin module. It is configured to minimize the copper loss in the slag and / or stabilize the thickness of the slag layer as optimization objectives. Using the simulation results of the digital twin module, it performs synergistic optimization on at least two of the following: oxygen concentration, slag-forming agent ratio, and blasting parameters, and outputs one or more sets of optimized operating parameters.
2. The digital twin optimization system for the high-oxygen-enriched double-sided blowing pool smelting process according to claim 1, characterized in that, The mesoscopic slag-gold transport and separation submodule is a population equilibrium model, in which the settling rate of matte droplets in the slag phase is modeled as a function of the dynamic slag viscosity.
3. The digital twin optimization system for the high-oxygen-enriched double-sided blowing pool smelting process according to claim 1 or 2, characterized in that, The predictions of the dynamic slag layer submodule are also based on the solidification temperature of the slag, the melt flow rate near the furnace lining wall, and the temperature gradient.
4. The digital twin optimization system for the high-oxygen-enriched double-sided blowing pool smelting process according to claim 1, characterized in that, It also includes an online calibration module, which is communicatively connected to the multi-scale coupling mechanism digital twin module and is configured with a data interface; wherein, the online calibration module acquires the production data of the physical melting furnace in real time through the data interface, compares the production data with the predicted data output by the multi-scale coupling mechanism digital twin module, and generates adjustment instructions for the parameters in the multi-scale coupling mechanism digital twin module based on the comparison results.
5. The digital twin optimization system for the high-oxygen-enriched double-sided blowing pool smelting process according to claim 4, characterized in that, The online calibration module uses Bayesian inference or ensemble Kalman filtering algorithms to adjust parameters.
6. The digital twin optimization system for the high-oxygen-enriched double-sided blowing pool smelting process according to claim 1, characterized in that, The multi-objective optimization module employs a multi-objective evolutionary algorithm and conducts virtual experiments based on the multi-scale coupling mechanism digital twin module to obtain a Pareto optimal solution set as the optimized operating parameters.
7. A method for optimizing a high-oxygen-enriched double-sided blowing pool smelting process based on digital twins, characterized in that, Includes the following steps: S1. Construct and run a multi-scale coupled mechanism digital twin model to simulate the smelting process, wherein: The content of Fe3O4 in the slag is dynamically calculated based on the oxygen potential and temperature field in the molten pool. The dynamic slag viscosity is calculated based on the Fe3O4 content, and the copper loss in the slag is predicted based on the dynamic slag viscosity. Based on the dynamic slag viscosity, solidification temperature of the molten slag, melt flow rate and temperature gradient near the furnace lining wall, the thickness of the slag layer on the furnace lining is predicted. S2. With the optimization objective of minimizing the copper loss in the slag and / or stabilizing the thickness of the slag layer, at least two of the following parameters—oxygen concentration, slag-forming agent ratio, and blasting parameters—are optimized collaboratively using the simulation results of the multi-scale coupling mechanism digital twin model. S3. Output one or more sets of optimized operation parameters obtained from the optimization process.
8. The method for optimizing the high-oxygen-enriched double-sided blowing pool smelting process based on digital twins according to claim 7, characterized in that, In step S1, a group equilibrium model is used to simulate slag-gold transport and separation, wherein the settling rate of matte droplets in the slag phase is set as a function of the dynamic slag viscosity.
9. The method for optimizing the high-oxygen-enriched double-sided blowing pool smelting process based on digital twins according to claim 7, characterized in that, It also includes an online calibration step: Real-time acquisition of production data from physical smelting furnaces; The prediction results of the multi-scale coupling mechanism digital twin model are compared with the corresponding actual production data; Based on the comparison results, the parameters used to calculate Fe3O4 content and / or slag-gold separation in the digital twin model of the multi-scale coupling mechanism are adjusted in reverse to reduce prediction errors.