FROM PROCESS SIMULATION TO GENERAL ESTIMATION

The ability to use a simulation program with the aim of making mass an d energy balances in a plant, starting from just raw data and process knowledge, is an ambitious and attractive goal. We propose to address a general formulation for the estimation problem in the framework of a simultaneous modular approach. As raw data are corrected to satisfy m ass and energy balances and some parameters are computed to give a uni t operation performance diagnosis, the estimation consists both of dat a reconciliation and parameter identification. It follows that the num erical problem is a minimization of a least square objective function. Constrained minimization is performed with an infeasible path strateg y which solves the optimization problem and process balance convergenc e at the same level, with a nonlinear programming code. The confidence intervals computation, for parameters and corrected values of measure ments, is also discussed. A simple flash loop flowsheet example gives a better understanding of the thee etical aspects. A more complex flow sheet has been chosen as a second illustrative example. Copyright (C) 1996 Elsevier Science Ltd