China's leading cement producer faced mounting pressure to reduce coal consumption and greenhouse gas emissions across its extensive plant network. The Yiyang facility alone consumed approximately 165,000 tons of coal annually at a cost of US$16.5M per year, producing 5,000 tons of clinker per day. Conventional PID loop controls could only manage a limited number of variables in a process where kilns must sustain temperatures above 1,200°C and free lime content must remain within a narrow 1–2% tolerance. Operators were spending 50–70% of their time manually monitoring thermal cameras and adjusting fuel flows, yet thermal efficiency did not improve — leaving significant cost and emissions reduction potential unrealized.
Rockwell Automation deployed FactoryTalk Analytics PavilionX MPC — a modular model predictive control platform built on supervised machine learning — starting with a pilot at the Yiyang facility before expanding to additional plants. The software integrated with the plant's pre-existing control infrastructure without requiring a full system replacement. By ingesting historical production data, real-time process variables, and laboratory free lime measurements, the MPC models predict how disturbances such as variable coal heat value and recirculated hot air from clinker coolers will affect calciner and kiln temperatures. The system then autonomously adjusts coal additions and setpoints across both kiln and mill operations. Following the successful Yiyang pilot, the solution was extended to at least five additional plants, with five more planned.
The Yiyang plant reduced kiln coal and energy consumption by up to 2%, translating to fuel savings of up to US$330K per year against a prior annual coal spend of $16.5M. Beyond the headline metric:
Have a similar implementation?
Share your customer's AI results and link it to your vendor profile.
Submit a case study →