Dynamic and steady state modeling of VRD column in equation-oriented environment.

• Ano de defesa: 2015
• Autor(a) principal: Ebrahimpour, Misagh
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Ambiente orientado a equações; Destilação (Modelagem; Simulação); Equation-oriented environment; Formulação de índice reduzido; Modeling; Processos químicos; Reduced index formulation; Simulation; Vapor recompression distillation
• Link de acesso: http://www.teses.usp.br/teses/disponiveis/3/3137/tde-18072016-083635/

Dynamic and steady state modeling and simulation validation of an industrial depropanizer owned by Petrobras are carried out in the Equation-Oriented environment using EMSO (Environment for Modeling, Simulation and Optimization). The depropanizer is a high purity distillation column with high nonlinear behavior because of the strong interactions due to the vapor recompression. Furthermore, the difference between internal and external material/energy flows causes a complex multitime-scale dynamics. Modeling such process is a challenging problem due to these characteristics. Initially, a steady state model, robust, fast and precise, able to provide steady state predictions that are necessary for effective implementation of Real Time Optimization (RTO) was developed in EMSO. In addition, the modeling of dynamic equilibrium processes often results in higher index DAE systems. Usually, phenomenological relationships are used to solve the index problem, but this approach gives rise to errors as a result of unknown parameters and project details that are assumed. Considering that the column\'s response to composition changes, in general, takes place over a timescale one or two order of magnitude slower than those of flow rate changes, an approach similar to a proportional loop with arbitrarily large gain is used as an alternative to solve the index problem. The dynamic model structure is based on the steady state model and contains more than nine thousand equations. Validation simulation results from comparison between real plant data and dynamic model have shown that the proposed approach is able to predict the dynamic behavior of the column properly.