Sinopec's Zhenhai Refining and Chemical Company, the largest integrated refining and chemical enterprise under Sinopec, was recently the only refining company selected for a national list of 15 leading smart factories. In recent years, the company has driven business transformation with digital technology, continuously upgraded its smart factory, and achieved lean management from engineering construction to production operations.
How is a smart factory built and what does it look like? A recent visit to Zhenhai Refining and Chemical provided a firsthand look at its intelligent production line.
System Automation Standing before the large data screen in the Zhenhai Refining and Chemical Equipment Health Management Center, one can see numerous equipment nodes individually marked. Different colors represent different health indices, and alerts from the past 24 hours are listed, highlighting potential operational risks. The facility has over 800,000 pieces of equipment, each undergoing three rounds of daily "health checks." Previously, isolated systems meant that when an abnormality occurred, technicians relied on manual experience to troubleshoot, a process that was slow and prone to oversight. Now, with the health management platform, the system can score each piece of equipment in real-time, issue automatic warnings, and dynamically generate equipment and unit health indices.
The development of this platform was a key step in advancing from traditional to digital and intelligent equipment management. Among nearly 10,000 pumps, more than 4,200 have been fitted with sensors that upload real-time data such as vibration and temperature. The platform uses expert models for intelligent diagnosis, helping technicians quickly identify key pumps and resolve issues efficiently instead of searching blindly. Structured in a three-tier "company-operations department-unit" framework, the platform provides personalized data displays. Each equipment manager logs in to see a data view relevant to their role and specific units, with trends clearly visible. The platform also automatically generates weekly "health reports," serving as an "equipment doctor" in production management meetings and significantly improving problem analysis and maintenance scheduling. This system has transformed the work approach from reactive problem-solving to proactive maintenance and predictive intervention, allowing equipment to effectively "speak for itself."
Above Jingshi South Road within the plant, a complex network of pipelines spans six levels, with over 100 pipes of varying diameters crisscrossing like a neural network. The entire plant features 55 kilometers of system pipe racks and 3,500 kilometers of various pipelines. To manage this dense network, a 3D model allows precise location of key valves with a simple click. In hazardous chemical pipeline management, rapid response is critical, requiring quick identification of faulty sections and access to comprehensive pipeline information for repairs. Previously, this relied on experience and archived records, which was time-consuming. Now, nearly 10,000 pipe supports are equipped with QR code "identity tags." Field personnel can scan these with mobile devices to instantly pull up 3D cross-sectional views of the pipeline section, displaying material, medium, maintenance records, and other information for intuitive and efficient operation.
This capability stems from the full implementation of a digital twin factory. A digital twin creates a highly accurate 3D virtual replica of the physical plant, synchronizing equipment status and production processes in real-time. The system is used not only for emergency management but is deeply integrated into daily production. For example, the ethylene unit, which involves extremely complex physical and chemical reactions under high temperature and pressure, uses the digital twin model to simulate and optimize production parameters in a virtual environment first, effectively improving product qualification rates and planning execution efficiency. The digital twin acts as an intelligent practice field, making production more predictable and controllable.
Unmanned Inspection In the spherical tank area, a cable-suspended robot travels along two steel wires only 7 millimeters thick, conducting aerial patrols. This compact robot is equipped with 36 wireless sensors, enabling round-the-clock monitoring of key areas like pipeline bands and valves, achieving inspection without blind spots. The spherical tank area, built in the 1990s, consists of twelve 2,000-cubic-meter tanks and is a critical node for liquefied gas storage and transportation. It is a high-frequency operation area with significant stored energy and long operation times, classified as a major hazard monitoring zone. Just one year ago, this 12,000-square-meter area relied entirely on manual inspections, with workers using handheld detectors to check each point. Now, the aerial robot not only avoids ground obstacles but also reaches corners inaccessible to humans. Fitted with panoramic cameras, gas sensors, and infrared thermal imagers, the robot can detect gas leaks within a 50-meter range in two seconds, capture subtle abnormal sounds, and navigate autonomously while avoiding obstacles. Upon detecting anomalies, data is transmitted in real-time to a command center for analysis by AI algorithms, effectively providing the tank area with a full-time "intelligent manager."
Scenarios of "unmanned or reduced personnel" are ubiquitous at Zhenhai Refining and Chemical. Through real-time monitoring of electrical equipment, online analysis of power systems, and real-time fault diagnosis upgrades, substations no longer require on-site personnel, enabling remote operation confirmation and automated robot inspections. After 3D scanning detects coke distribution, grab buckets are automatically controlled via wireless networks for decoking, stacking, and discharging, eliminating the need for manual direction. To handle numerous repetitive, rule-based tasks prone to human error, software robots are deployed to simulate manual operations across application systems, achieving process automation.
Digital Management After completing a reservation and driver training, a driver arrives at the loading area. The vehicle license plate is automatically recognized, the barrier gate lifts, and the weighbridge automatically records the weight. Within minutes at the loading island, the system completes precise connection and automatic loading. Once loading is finished, the weight is verified, an electronic weighbill is generated, and the driver confirms the pickup information on a mobile phone before driving away, all without leaving the cab. Functions like pickup reservations, driver training, and vehicle access are consolidated on a platform where carriers can view schedules and arrange transport capacity online. The company can schedule off-peak loading based on system planning, with a clear digital workflow from order receipt to product dispatch. The implementation of intelligent inbound/outbound logistics management and automated warehouses has increased product dispatch efficiency by 40%, reducing customer vehicle wait times from one day to under one hour.
This is just one example of remote controllability. The large screen in the production command center displays real-time operational dynamics and personnel activities across the 23.1-square-kilometer Zhenhai base. Personnel at all levels can remotely control and manage all on-site equipment through automated and intelligent means. The catalytic cracking unit in the second-phase project enables high-value utilization of inferior heavy oil and low-cost oil conversion. On the control room display for the second phase, curves for temperature, pressure, and load are clearly visible, with the status of over 200 control loops一目了然. The project team used self-tuning algorithms to model and assess 271 loops individually, with some key loops now achieving online automatic tuning.
Intelligent factory concepts are no longer abstract but are experienced daily. Functions like "automatic throughput ramping" and "adaptive liquid level disturbance resistance" demonstrate how intelligence stabilizes fluctuations, reduces energy consumption, and increases product value added. Guided by the "molecular refining" concept, the smart factory achieves efficient resource conversion, ensures coordinated operation through standardized scheduling, and enhances unit profitability via real-time optimization. Through its smart factory construction, Zhenhai Refining and Chemical has significantly boosted production levels and labor efficiency, generating over 200 million yuan in annual benefits from intelligent production and increasing average labor productivity by more than 50%. Looking ahead, the company will leverage the "Great Wall Large Model" and the "Sinopec Intelligence Cloud" platform to make production operations more efficient, precise, and safe, exploring new models for AI-empowered petrochemical production.