Modelagem e simulação de regeneradores industriais

AUTOR(ES)
DATA DE PUBLICAÇÃO

1994

RESUMO

In this work, it is developed a software to simulate a regeneration system of a catalytic cracking industrial unit. Firstly, deterministic models are developed, and latter on, it is used backpropagation neural network and a hybrid model with neural network. The regeneration system is formed by a sequence of reactors with different hydrodynamic regime. A fluidized bed reactor is followed by a riser and freeboard. Initially, the solid recirculation inside of system is not taken into account, and latter on, this recirculation is considered. In this latter case, an iterative procedure is proposed to break the information cicle. The material and energy balance equations form a system of nonlinear integrodifferencial equations. Firstly, these have been solved using Runge-Kutta method and Trapezoidal Rule. An alternative procedure is proposed, where the Orthogonal Collocation Method is used. When this is the case, differencial equations are discretized and integrals are solved by Quadrature. The regeneration system is also simulated by use of Backpropagation Neural Network. Process and Project Neural Network are developed. Parameters like compounds concentration profiles and temperature profiles are obtained. A hybrid model with neural network is proposed, where the material balance is solved by deterministic model, whereas the temperature profiles are obtained by neural network

ASSUNTO(S)

modelos matematicos simulação (computadores) reatores fluidizados

ACESSO AO ARTIGO

Documentos Relacionados