On-line first principle models - State estimation

This paper describes the application of rigorous first principle process models corrected using real time on-line process data to match current plant response. Marrying detailed and rigorous first principle models, embodying converged heat and material balances, with real-time on-line process data, using State Estimation technology for correction, provides the ideal means of obtaining on-line real-time inferential data over the broadest range of process operation. This provides the ability to infer the value of process variables throughout the plant in a single application. This can be used for: Reconciliation of the plant data with the model predictions to produce best estimates for measurements; Infer values which cannot, and could never be, measured on the plant; Bridge the time intervals in existing measurements which are only available infrequently; Predict future values of plant variables. The additional information results in better operation through, for example, improved product qualities and reduced transition times. This paper outlines the technology developed and describes some of the application work completed by AspenTec(R). The use of state estimation techniques for 'real world' process applications of significant size is believed to be unique, and the developments described here allow a new generation of applications to be considered. The technology has been developed to be applicable to a range of processes.