A subtask in integration of design and control of chemical processes is the
selection of a control structure. Automating the selection of the control
structure enables sequential integration of process and control design. As
soon as the process is specified or computed, a structure for decentralized
control is determined automatically, and the resulting decentralized contr
ol structure is automatically tuned using standard techniques. Dynamic simu
lation of the resulting process system gives immediate feedback to the proc
ess design engineer regarding practical operability of the process. The con
trol structure selection problem is formulated as a special MILP employing
cost coefficients which are computed using Parseval's theorem combined with
RGA and IMC concepts. This approach enables selection and tuning of large-
scale plant-wide decentralized controllers through efficient combination of
model formulation and mathematical programming. (C) 2000 Elsevier Science
Ltd. All rights reserved.