Reduced distillation models via stage aggregation

A method for deriving computationally efficient reduced nonlinear distillation models is proposed, which extends the aggregated modeling method of Lévine and Rouchon (1991) to complex models. The column dynamics are approximated by a low number of slow dynamic aggregation stages connected by blocks of steady-state stages. This is achieved by simple manipulation of the left-hand sides of the differential equations. The algebraic equations resulting from the reduction procedure are replaced by interpolation in tables or polynomial approximations. The resulting reduced model approximates the original dynamic model very accurately, and for a realistic case study increases the simulation speed several times. This makes the reduced models interesting for real-time applications. The numerical properties of the models and possible improvements are discussed.