A leading Chinese polymer producer operating one of the country's largest petrochemical complexes faced persistent operational instability across its polyethylene and polypropylene production lines. The facility ran on advanced process technologies — Hostalen ACP and Spherizone — that set high technical thresholds few operators could consistently meet. Inexperienced staff caused frequent quality fluctuations, while manual grade transitions required complex personnel coordination across multiple parameters: melt index, density, temperature, and ethylene partial pressure. Each transition introduced uncertainty in production volume and equipment adaptation, making cost reduction and throughput improvement difficult to sustain at scale.
Rockwell Automation deployed FactoryTalk Analytics PavilionX, a process modeling and advanced process control (APC) platform built on multivariable predictive ML — specifically a mixed nonlinear model combining mechanism-based and statistical approaches, including Extrapolated Gain Constrained Neural Networks (EGCNN) for accuracy beyond normal operating conditions. The software integrated via a service-oriented architecture (SOA) into the existing plant infrastructure across two polymer production units, requiring no full DCS replacement. A web-based graphical interface surfaced real-time trend data to operators at all experience levels. The system's automatic grade transition module allowed operators to initiate transitions with a single menu selection, with the controller autonomously managing all parameter changes and recovery sequences previously handled manually.
After APC retrofitting on two polymer production units, the facility achieved measurable improvements across quality, efficiency, and labor utilization. Automatic grade transitions eliminated manual parameter tuning, delivering consistent transition profiles and freeing operators from high-coordination tasks. Rockwell Automation's broader polymer APC deployment data — spanning 200+ polymerization reactors — establishes the performance envelope this implementation targets:
Supplemental material and energy consumption also declined, and the customer initiated plans to extend the solution to additional polymer and chemical devices.
Have a similar implementation?
Share your customer's AI results and link it to your vendor profile.
Submit a case study →