The steelmaking process is being revolutionized through technology. At the finals of CITIC Group's third "Blooming Cup" Digital Application Competition, CITIC Pacific Special Steel Group Co., Ltd. received second prize for its "Integrated Innovation and Application of Artificial Intelligence Technology Across the Entire Special Seamless Steel Pipe Manufacturing Process."
The "Blooming Cup" competition serves as a key component of CITIC Group's digital development initiative, showcasing outstanding digital and intelligent innovations from across the conglomerate. How did CITIC Pacific Special Steel distinguish itself in such a competitive field? The answer lies in a technological revolution reshaping special steel manufacturing.
Each steel pipe is now assigned a unique digital identity. Special steel forms the foundation of the modern industrial system, and special seamless steel pipes are critical materials for national infrastructure, defense, aerospace, marine energy, and high-end equipment due to their high reliability under extreme conditions.
Historically, purchasing special steel pipes was akin to a blind box purchase for customers, who relied on batch-average parameter reports without certainty about individual pipe quality. Production was similarly imprecise. At CITIC Pacific Special Steel's Daye Special Steel, these issues have been fundamentally resolved. From hot rolling to heat treatment, every single pipe among thousands in continuous production is now accurately tracked and identified.
The company's AI system addresses three major industry challenges: low R&D efficiency for high-end products, difficulties in flexible manufacturing, and complex global optimization decision-making. It has developed 34 AI applications, creating a matrix that covers the entire seamless pipe manufacturing process for intelligent, coordinated decision-making.
A standout feature is material tracking. By applying AI technologies such as image recognition of heat numbers, individual tracking models for hot rolling and heat treatment, AI-based identification, multi-target tracking, and intelligent pipe positioning, the system has developed multi-scenario perception fusion technology. This solves decision-making problems caused by multi-source information coupling loss and uncertainty, enabling individual pipe tracking.
"Eight key node cameras are installed at the precision cutting station and its upstream/downstream areas. Through a video AI analysis system, these cameras can identify each pipe's number and flow direction in real-time. Combined with position prediction algorithms and an automatic mark correction mechanism, recognition accuracy and efficiency are significantly improved, resolving issues like signal lag and material confusion," explained Ma Jinhui, Head of Quality Assurance at Daye Special Steel. The system now achieves a multi-target, cross-camera tracking accuracy rate of ≥96.8% and an automatic insertion warning rate of ≥96.5%.
This individual traceability, combined with full-process AI, means each pipe has its own identity and traceability information. Data such as the originating furnace, rolling temperature curves, and heat treatment duration can all be tracked and reviewed. Details including the exact casting time, furnace temperature fluctuations during production, and the responsible operator are all recorded.
"The individual pipe tracking system overcomes the long-standing industry challenge of full-process traceability for single pipes, providing unprecedented precision for quality management," stated Bai Xiansong, Head of Intelligence and Informatization at CITIC Pacific Special Steel Group. "Any quality issue with a pipe can be traced back to its complete logistics trajectory, including the billet used, processing equipment, operating team, and process parameters. Furthermore, it provides an accurate data foundation for intelligent scheduling and cost accounting, making it possible to plan and collect costs per single pipe. It also breaks down information silos within the factory, creating a precise mapping between physical pipes and digital data flows, which is a crucial step toward building a digital factory and advancing to higher-level intelligent manufacturing."
"If material tracking is not precise, analysis and online control cannot be accurate," Bai Xiansong emphasized, noting that this has been a persistent industry problem that only CITIC Pacific Special Steel has currently solved.
These special seamless pipes, though not encountered in daily life, are essential components in major national projects and critical equipment. For instance, the company's pipes are used in extreme conditions such as 10,000-meter ultra-deep wells, the world's largest 3300-ton-meter wind power tower crane, the global biggest ring rail crane, and high-pressure pipes for petrochemical projects.
Described as supporting projects that "reach for the skies and delve into the earth," these pipes must perform reliably in unseen locations for long periods. How is the quality of every single pipe guaranteed?
According to Ma Jinhui, the full-process AI technology has greatly enhanced the efficiency and quality of the inspection phase. It enables precise measurement of key quality characteristics like steel grade, surface defects, and dimensional contours, replacing numerous tedious and error-prone manual inspections. This allows for multi-dimensional, accurate, online dynamic detection of pipe quality. After AI implementation, billet information recognition accuracy reached 99.5%, pipe surface defect detection rate hit 99.2%, and full-size inspection speed increased threefold.
Addressing R&D inefficiencies, another major industry pain point, Ma Jinhui noted that by integrating a special steel vertical large model, the company built a digital material R&D system. This intelligently transforms data into knowledge for process R&D, reducing high-end product development time by 57% and aiding the development of five strategic new products, including the TP140V steel pipe for deep-sea oil casings.
Additionally, for complex scheduling challenges under the "multi-variety, small-batch" production model, the company established a full-process intelligent scheduling system for seamless pipes. This improved overall production scheduling efficiency by 30%, achieved over 95% accuracy in automatic production plan generation, and significantly enhanced on-time order delivery rates.
"Some technologies already exist, but they cannot be applied rigidly without deep integration with specific scenarios and business operations. If technology is not deeply bundled with business, its value is hard to realize," Bai Xiansong summarized, reflecting on the key lesson from their AI implementation journey.
CITIC Pacific Special Steel has redefined quality through AI—quality is no longer just about inspection or even manufacturing alone, but a synergistic combination of "design + prediction + control + traceability." This technological system has earned authoritative recognitions, including the MIIT "Excellent-Level Smart Factory" designation and being the first "Lighthouse Factory" in the global special steel industry. It has been featured as a benchmark case in national initiatives for "AI + Steel."
By integrating AI technology throughout the entire production chain, CITIC Pacific Special Steel is leading a revolutionary shift in the special steel production paradigm.
In decision optimization scenarios, AI addresses the difficulty of global optimization and real-time dynamic control across multiple business functions in the production process. The company built an industrial internet platform, developed a model for decomposing factory-level KPI indicators layer by layer, and created a digital evaluation model for "golden standard" pipes. This establishes lean data element windows for production, quality, and energy, supporting coordinated control across all business operations and processes.
This breaks down business barriers and enables smart multi-business collaboration, upgrading the steel industry from an "experience-dependent" production model to an "AI-driven" one—a transformative change.
Technological innovation has translated into tangible benefits. Ma Jinhui reported that after implementing the full-process AI integration, the factory's hot rolling area saw an annual output increase of 11.35%, energy cost per ton of steel reduced by 10.42%, and metal loss decreased by 2.13%. The heat treatment area achieved a 2.35% annual output increase and a 10.1% reduction in energy cost per ton of steel, generating total annualized benefits of approximately 110 million yuan.
In 2025, CITIC Pacific Special Steel Group Co., Ltd. received awards including the first prize of the Metallurgical Science and Technology Progress Award, the MIIT Excellent-Level Smart Factory recognition, the Hubei Provincial Science and Technology Award, and the Hubei Province "Data Factor X" Intelligent Manufacturing Track award.
Looking ahead, the company aims to actively implement CITIC Group's strategic deployment for technological innovation and digital transformation. It plans to create a digital-intelligent paradigm through integrated innovation, replicate and promote AI experience, empower transformation and efficiency gains across the entire chain, and consolidate its industry leadership. Starting with special seamless pipes as an entry point, it will deepen the implementation of intelligent models in various scenarios, extend AI to all processes in special steel manufacturing, integrate multi-source data to build an AI ecosystem platform, and achieve a closed-loop for intelligent decision-making across the entire chain, empowering digital transformation, quality improvement, and cost reduction. With "AI+" and "Data Factor X" as dual core engines, the goal is to build a globally leading smart special steel ecosystem by 2030 and establish an AI-deeply-empowered benchmark smart special steel enterprise by 2035.
Through three consecutive "Blooming Cup" competitions, CITIC Group has identified and nurtured a group of digital-intelligent applications, like the one from CITIC Pacific Special Steel, which demonstrate strong exemplary effects, comprehensive benefits, and significant development potential across different business areas. These provide valuable references and leadership for the transformation and upgrading of traditional industries.
From explorations in cutting-edge fields like quantum technology and AI for Science to leading practices in emerging industries such as intelligent manufacturing, embodied intelligence, and the low-altitude economy, along with initiatives in tech finance empowering the AI industry and building an international innovation ecosystem, CITIC Group has firmly implemented the central government's strategic plans for building a strong nation in science and technology, cyberspace, and digitalization. It has executed the "Bedrock" science and technology project, constructed a high-standard CITIC-characteristic sci-tech innovation system, and focused on key core technology breakthroughs and industrial applications in critical areas like intelligent manufacturing, robotics, new materials, biological breeding, and salt lake resources. By advancing the "AI+" action, creating a group-level common capability base around technology, computing power, and data, and promoting the deep integration of AI with scientific research, industrial development, and management decision-making, the group has achieved a number of industry-leading application results.
In 2025, CITIC Group won 106 national, provincial/ministerial, and renowned industry science and technology awards. Related projects have served as excellent demonstrative examples within the industry. These outstanding applications are creating a vibrant and diversified landscape, gradually transforming into core corporate capabilities and market competitiveness, thereby accumulating momentum and empowering the group's high-quality development.