Investigation On Hybrid-control Of Metallurgical Mechanism And Data-driven Technology Over Intelligent Steelmaking Slag System

The ironmaking and steelmaking industry is a representative of the traditional process-oriented manufacturing industry,and it also is a pillar industry of the Chinese national economy.High standard of steel quality requires much strict operation of steelmaking processes.Improving the proportion of automatic operation and reducing unnecessary human intervention in the process are an effective way to achieve strict operation.Therefore,realizing automation or achieving intelligence in steelmaking processes is an inevitable development trend for steel plants in the future.With the strong impetus brought by ’Made in China 2025’,the intelligent steelmaking technology has been developing fast with features like automation,digitalization and networking.The traditional dynamic model that can explain the changes in the critical process parameters based on events taking place in the steelmaking operation is a must-have tool for steelmaking operators.But the continual adjustment of the complex dynamic model coefficients was the main reason that is not conducive to the popularization in practical production.Benefiting from the great progress of data mining technology in recent years,this study established an intelligent slag-composition predicting model which combined steel metallurgical mechanism and data mining technology together.The intelligent model could realize the prediction of slag composition during total steelmaking process(converter-converter/LF stage-LF).Furthermore,the informationization,intelligence and standardization of the ironmaking and steelmaking industry would be enhanced.In the present work,a mathematical model employing multi-zone reaction kinetics has been developed to predict the compositional evolution of hot metal and slag during a 150t BOF steelmaking process.The three primary refining zones,namely:jet impact,slag-bulk metal and slag-metal-gas emulsion zone were considered in the simulation of the overall hot metal refining.The fundamental process understanding has been applied to mathematically describe the transient rate parameters.A micro and macroscopic rate calculation methodology(micro-kinetics and macro-kinetics)were proposed to estimate the total refining by the recirculating metal droplets through the slag-metal emulsion zone.And the real-time change of(Fe2O3)in the converter slag was researched especially.Furthermore,the accurate slag composition prediction of the converter blowing process was realized.Converter carryover slag during tapping has been a trouble for improvement of steel quality.We constructed a weight forecasting model of the converter carryover slag by combining the infrared video visual learning technology with historical production data mining technology.Hence this slag weight predicting model realized the threshold value of the infrared detector adjusting dynamically.In order to give an insight into the action mechanism of carryover slag on steel during tapping,a serious of molten steel-slag reaction experiments was conducted to elucidate the reaction mechanism between[Al]and CaO-SiO2-MgO-Al2O3-FeO type slag at 1600℃ as well as its resulting effect on cleanliness under varying amount of converter carryover slag was investigated in laboratory.Composition evolutions in molten steel and slag during the reaction and microstructure on the steel-slag interface were investigated by employing different experimental techniques.2[Al]+(FeO)=2[Fe]+(Al2O3)and 4[Al]+3(SiO2)=3[Si]+2(Al2O3)were confirmed as a major reaction in the system.The initial(Fe2O3)content in experiments increased from 3%to 15%.The results indicated the metal-slag reaction tended to stable after about ten minutes,the mass transfer of[Al]in molten steel was the rate-controlling step.[Als]was more prone to react with(SiO2)for slags with initial(FeO)less than 4.0%,but it reacted with both(FeO)and(SiO2)when(FeO)content was more.It was therefore suitable addition of Al alloy should take(SiO2)into consideration in addition to(FeO)and dissolved[O].Increase of carryover slag amount represented by increase of initial(FeO)decreased the sulfur distribution but promoted the formation of MgO·Al2O3 inclusion.To reveal the dissolution mechanism of inclusion in slag during converter tapping process,some comparison experiments were conducted by adding isolated spherical alumina balls as inclusions in CaO-SiO2-Al2O3-Fe2O3-MgO slag,and FetO content was contained in slag up to 10%.The results showed that there was a dissolution path of alumina balls in slag under different dissolution time,and the dissolution path was CaO·2Al2O3-CaO·Al2O3-2CaO·Al2O3-(3CaO·Al2O3).At the end of the dissolution process,the interface composition of alumina ball changed to 2CaO·Al2O3.Whether 3CaO·Al2O3 would appear is dependent on the content of mass ratio(CaO)/(Al2O3)in the slag.If the mass ratio(CaO)/(Al2O3)>3,3CaO·Al2O3 would appear.In addition,the upgraded Diffusion-distance-controlled dissolution model(DDD-Model),which FetO content was introduced and applied in the study.The results illustrated that Al2O3 inclusion apparent dissolution rate was improved by high(FetO)content,increasing(CaO)/(SiO2)and raising temperature as soon as possible at early stage of the converter tapping process.It is not necessary to increase(FetO)content in the slag to enhance the dissolution rate of(Al2O3)inclusion at the last tapping stage.The complete dissolution time of spherical Al2O3 inclusion with 0-100 μm in diameter based on the upgraded DDD-Model could be predicted.Through the prediction of converter endpoint slag composition,the strictly control of converter carryover slag weight,the research of slag-steel reaction mechanism in the secondary refining process and the process operation optimization in the whole steelmaking process,finally the real-time prediction of the slag composition was realized.This study broke the data barrier between different equipment like converter and LF furnace,and it built a steelmaking process operation control database to improve the slag composition prediction accuracy.As a result,this study found a way to make metallurgical mechanism data-driven techniques work effectively together during steelmaking process,and an intelligent data-driven technology and stable mechanism hybrid-control for steelmaking slag system was established and applied widely.