(372b) The Automated Generation of Constrained Models from Feasibility Analysis for Reactive Distillation

Feasibility analysis in reactive distillation allows for the rapid screening of potential design alternatives. Based on difference points, feasible regions may be constructed for column sections, which in turn can determine overall column feasibility [1,2]. Feasible region analysis is only a portion of a larger design paradigm. After a column is determined to be feasible it must be modeled and then optimized. Previously feasibility analysis serves only as a check of whether further modeling activities are worthwhile. We will show how insights gained during feasibility analysis can be used to constrain the optimization model. In addition we will demonstrate how to automatically generate constrained models from feasibility analysis.

Reactive distillation models are non-convex NLP's, leading to multiple locally optimal solutions. We have previously shown that for reactive distillation column sections governed by saddle pinches there is a known limit for the extent of reaction on each stage that will shift the column profile towards potentially superior solutions [3]. It is from this knowledge that we can construct explicit constraints for each stage on the extent of reaction. These constraints are able to cut our decision space and can potentially remove inferior locally-optimal solutions. We will demonstrate the use of this method in the production of TAME and the production of acetic acid from acetic anhydride and water.

[1] Hauan,S. and Ciric, A.R. and Westerberg, A.W. and Lien, K.M. "Difference points in extractive and reactive cascades. I - Basic properties and analysis." Chem.Eng.Sci. 59(17): 3671-3693, 1999.

[2] Hoffmaster, W.R. and Hauan, S. "Difference points in extractive and reactive cascades. III - Properties of column section profiles with arbitrary reaction distribution." Chem.Eng.Sci. 55(16): 3145-1075, 2004.

[3] Turnberg, S.L. and Hoffmaster, W.R. and Hauan, S. "Design of a TAME reactive distillation process using feasible regions." Presented at the AIChE annual meeting, paper 472b, November 2005.