Statistical Process Monitoring of Cryogenic Air Separation Unit Startups*

The startup phase of cryogenic air separation units (ASUs) includes complex, non-routine events characterized by dynamic, nonlinear, and nonstationary conditions. Developing an effective process monitoring framework, which is crucial for ensuring safe and efficient ASU startups, is a challenging task. This study introduces a statistical process monitoring system based on variable-wise multi-way principal component analysis (MPCA) for monitoring the startup of cryogenic ASUs. By leveraging the multi-stage multi-unit nature of startups, the proposed framework addresses the unique complexities of the system, enabling the prompt detection of slow or faulty startups. The effectiveness of the proposed monitoring approach is demonstrated in an industrial case study, which successfully identifies faulty startups.