Abstract

A novel simulation strategy for dynamic distillation of complex mixtures, such as wine, is proposed and evaluated in terms of computing efficiency and accuracy. The model developed describes wine distillation as a multicomponent reactive batch distillation process. The simulation approach transforms the system of differential algebraic equations (DAE) into a set of ordinary differential equations, by pre-solving the algebraic equations and replacing them with artificial neural networks.This new simulation strategy for wine distillation is 40% faster than the rigorous solution of the DAE system, compared at the same level of accuracy. The model can be applied to the distillation of other spirits or complex mixtures, as well as in other separation processes in which the recovery of aromas is essential. © 2003 Elsevier Ltd. All rights reserved.