Synergy between Data Reconciliation and Principal Component Analysis in Energy Monitoring

Abstract

Monitoring of energy consumption is central importance for the energy-efficient operation of chemical processes. Fault detection and process monitoring systems can reduce the environmental impact and enhance safety and energy efficiency of chemical processes. These solutions are based on the analysis of process data. Data reconciliation is a model-based technique that checks the consistence of measurements and balance equations. Principal component analysis is a similar multivariate model based technique, but it utilises a data-driven statistical model. We investigate how information can be transferred between these models to get a more sensitive tool for energy monitoring. To illustrate the capability of the proposed method in energy monitoring, we provide a case study for heat balance analysis in the well-known Tennessee Eastman benchmark problem. The results demonstrate how balance equations can improve energy management of complex process technologies.