Development and Validation of a Reduced-Index Dynamic Model of an Industrial High-Purity Column

Abstract

A depropanizer column is a high purity distillation unit, whose modeling is challenging due to its high nonlinear behavior and intricate dynamics. Also, depending on the assumptions, a differential algebraic equation (DAE) system of high index can arise in the modeling of the column. Such systems create significant difficulties to numerical simulation. Phenomenological hydraulic relationships can be used to avoid the index problems. However, these relationships may rely on unknown and arbitrarily assumed parameters and project details. This paper presents an alternative modeling approach called dynamic design using proportional action (DDPA), which is used to simplify the hydraulic and pressure drop relation for a distillation unit. The DDPA approach generates a reduced index dynamic model by adding equations similar to a proportional controller with arbitrarily large gain. A depropanizer column was modeled using DDPA, and the model was validated with real process data obtained in an industrial-scale depropenizer column located at Refinaria de Paulínia, Paulínia, São Paulo, Brazil, and owned by Petrobras S.A. The results show that the model with DDPA is able to predict the actual column dynamic behavior properly.

Real-time optimiztion with persistent parameter adaptation using online parameter estimation.

“…These large values generate a stiff system that tends to an index-2 system when K tends to infinity. For further information regarding DDPA numerical solution difficulties, the impact of different K values on the solution and details about the column modeling, please refer to Matias, Ebrahimpour and Le Roux (2018).…”