Thread Content
For a long time, the coking industry has relied heavily on the traditional production model based on the experience of veteran workers, and it faces challenges such as data silos and difficulties in understanding cost structures. At the 8th Sixth Executive Director (Expanded) Meeting of the China Coking Industry Association and the Annual Exchange Meeting for Its Professional Committees, held recently, several experts present outlined the path forward: to shift the coking industry from \"experience-based coking\" to \"scientific coking\", it is necessary to rely on a framework that combines \"industry big data, industry mechanisms, industry-specific large models, and general-purpose large models\". It is essential to build a solid data foundation and establish a digital management system that covers all aspects of the process, including procurement, production, and management. The pain points in the industry’s intelligent transformation need to be addressed. As a traditional industry, the coking sector faces numerous challenges in its journey toward intelligence. Wang Yan, director of the Coking Research Institute at the Coal Science and Technology Research Institute Co., Ltd., pointed out that raw coal is the cornerstone of cost, quality, and sustainability for coking enterprises, yet there are significant shortcomings in the management of coking coal. In terms of cost control, \"high prices\" and \"black boxes\" coexist, leading to inaccurate cost management. Wang Yan analyzed that, against the backdrop of generally rising prices for raw coal, coking enterprises lack real-time and transparent data throughout the entire process, often finding themselves in a situation of \"blind purchasing,\" which leads to soaring hidden costs. At the same time, the data from processes such as procurement, production, and inventory is fragmented, leading to an increase in the phenomenon of \"islands.\" This results in a lack of coordination and optimization of overall supply chain costs, and the quality losses caused by internal inefficiencies end up being borne by the coking plants, further increasing the company’s financial burdens. Problems are also prominent in the quality inspection stage. Wang Yan explained that the traditional \"sampling–sample preparation–testing\" approach has delays, which prevent real-time guidance for production adjustments; as a result, the production process becomes somewhat \"out of control,\" affecting the accuracy of quality testing for coking coal. In the error composition of coking coal quality testing, sampling error accounts for 70% of the total error, sample preparation error accounts for 20%, and testing error accounts for 10%. These issues ultimately led to the coking industry’s excessive reliance on “experience”. “Data fragmentation has become the primary obstacle to the industry’s intelligent transformation. ”Wang Yan emphasized that due to the inability to integrate data from various departments, coking enterprises find it difficult to systematically explore \"coal blending\" as a key opportunity for cost reduction. From delays in incoming material inspection and inaccurate inventory counts, to rough models used in the coal blending process as well as a disconnect between theoretical calculations and actual operations, all these factors ultimately affect the stability of coke quality and the accuracy of cost control. To establish an intelligent control system across the entire value chain and address the challenges associated with the industry’s transformation toward intelligence, Shi Yanfeng, president of the China Coking Industry Association, proposed that efforts should be accelerated to develop and improve a centralized management platform that integrates distributed control systems (DCS), manufacturing execution systems (MES), and enterprise resource planning (ERP) systems, thereby enabling real-time and accurate collection, analysis, and sharing of data on all relevant factors. At the same time, by integrating real-time production data with expert systems related to application and development processes, intelligent optimization of individual process units is achieved, paving the way toward systematic optimization of the entire coking process. In addition, it is necessary to fully implement an Energy Management System (EMS) to conduct real-time measurement, online analysis, and dynamic balancing of key energy sources such as gas, steam, electricity, and water resources, in order to identify areas of leakage and energy waste within the system. Shi Yanfeng divides the advancement of digital and intelligent management in the coking industry into three phases. The first phase is the one for infrastructure development and data integration, during which a unified IoT platform and data hub are established to enable the digitization of key business processes such as material intake, testing, and storage, as well as automatic data collection. Visual dashboards are also developed to provide an initial display of critical data. Next is the stage of deeper integration of intelligent applications: digital twins are introduced to enable 3D visualization of the stockyard, an intelligent coal blending optimization module is developed and put into use, and a preliminary cost accounting and early warning system is established. Ultimately, it comes to the advanced stage of intelligent decision-making, where AI applications are further enhanced, such as for demand forecasting and predictive equipment maintenance. The decision support center is improved to offer advanced data mining and simulation analysis capabilities, thereby fostering an ecosystem for industrial collaboration. Wang Yan suggested that the \"Digital and Intelligent Management System for the Whole Life Cycle of Metallurgical Raw Coal\" developed by the institute covers aspects such as intelligent procurement and supplier management, intelligent coal blending and cost optimization, whole-life-cycle cost accounting, precise management of coal arrival and storage, quality control and tracking of coke, as well as monitoring and feedback during the production process. This system enables the creation of intelligent management systems for storage and stacking, digital coal plants, and intelligent equipment for mining, processing, and transportation; it helps to address issues such as difficulties in inventory counting and the lack of transparency in logistics, ensuring that there are no disruptions or distortions in the storage and distribution processes, thereby reducing the cost per ton of coke. Intelligent applications are showing tangible results; how can coking enterprises achieve intelligence? Zheng Hulei, deputy chief engineer at Huatai Yongchuang (Beijing) Technology Co., Ltd., said that the automation of coking enterprises relies on industry mechanisms and data from production equipment; it is necessary to adopt an approach that combines \"industry big data + industry mechanisms + industry large models + general large models\" in order to develop software products that are platform-based, capable of iteration, and easy to use. Taking the dry quenching process as an example, Zheng Hulei introduced the practical applications of AI in addressing boiler tube failures and desulfurization issues. For new projects, issues such as boiler tube bursts can be fundamentally resolved through technologies like cyclone dust removal ； For projects that are already in operation, this issue is effectively addressed through AI-based forecasting and in-furnace desulfurization technologies. The HTZW-01 intelligent desulfurization device for dry quenching furnaces developed by this company enables automatic, uniform, and stable distribution of the desulfurizing agent by continuously monitoring the SO2 content in the gases emitted from the dry quenching furnace. This device can be installed online without interrupting production, resulting in annual cost savings of around 3 million yuan (based on a coking capacity of 1 million tons per year). It reduces the SO2 content in the gas circulating inside the furnace by 35%~55%, and decreases the amount of desulfurizing agent required outside the dry quenching coke ovens by 25%~45%. Xu Xiuli, head of the Coal Resources Department at Anshan Thermal Energy Research Institute Co., Ltd. under Sinosteel Group, believes that the key for coking enterprises to reduce costs lies in \"expanding the sources of coking coal, gaining a deeper understanding of coal quality, and optimizing the coal blending structure.\" She pointed out, “The prerequisite for achieving intelligent coal blending is data governance.” Sometimes, we need to spend 3 months or even longer on data governance for a company, removing invalid and distorted data to lay a solid foundation for the models. ”Xu Xiuli explained that the smart coal blending software developed by the company has established a \"resource identity\" system for coal and coke, enabling full-track data tracing. Through mechanism research and a \"re-evaluation\" of coal quality, this software enables the optimization of coal blending predictions; for example, by using low-quality coal and high-quality coal in place of coking coal and lean coal, or coking coal in place of expensive high-quality coal and lean coal, the cost of coal blending can be reduced by about 40 yuan per ton. “Coal blending should not be limited to final proportion adjustment, but should cover the entire process from coal source selection, procurement, arrival at the plant, storage to blending. ”Wang Yan believes that by establishing an intelligent coal blending system that achieves optimal costs and quality, enterprises can address challenges such as large fluctuations in coal quality, high coal blending costs, inaccurate predictions of coke quality, and untapped data value, thereby truly transitioning from \"experiential coke production\" to \"scientific coke production\".