Control Performance Monitoring: a Data-driven Approach“, Proc. of Monitoraggio delle Prestazioni dei Processi e dei Sistemi di Controllo

Plant-wide disturbance; control loop performance; principal components; diagnosis; non-linearity; multivariate analysis The paper aims at showing how process data and basic process understanding can be used to properly address, prioritize and solve some of the most challenging problems which have heavy impact on maintenance and control engineers ’ everyday operation. While monitoring single control loop performances is a valuable procedure to quickly and efficiently spot suddenly arising faults and/or slowly developing degradation, it still misses the crucial feature to recognize, group and analyze similar patterns which might be the signature of a single major cause simultaneously affecting multiple loops. The proposed methodology combines the bottom-up approach of single loop performance assessment with an innovative top-down procedure for plant-wide disturbance analysis. The combination of the two techniques, able to maximize plant asset utilization without requiring major investments, is achieved through the use of a specific, multi-technology software package, which is briefly introduced in the second section. The last section is devoted to a case study where the methodology has been applied to an actual petrochemical process