Performance limitation improvement for a highly-integrated chemical process

Abstract

Control systems can have inherent performance limitations that are independent of the controller design ( Seron et al., 1997). However, recent work ( Cao, Rossiter and Owens, 1998a) shows that such performance limitation can be improved by introducing several appropriately selected secondary control loops. Here, selection of secondary manipulated variables and measurements for thesesecondary controlloops iscritical to the performance improvements. In this work, a generic selection method is presented. This method represents the selection as a Linear Quadratic Gaussian problem and uses a global optimization technique, the Branch-and-Bound method to select the best control structure without exhaustive evaluation of all possible secondary control input/output combinations. A case study for a highly integrated chemical plant — the HDA process shows that by closing two temperature controlloops in the benzene column, the performance of the primary outputs: benzene production rate and purity of the process is dramatically improved.