Hybrid artificial neural network—First principle model formulation for the unsteady state simulation and analysis of a packed bed reactor for CO2 hydrogenation to methanol

Abstract

Abstract Carbon dioxide emission is well recognized as the main source of global warming. The catalytic hydrogenation of carbon dioxide to methanol represents an effective method for preventing this side effect. The objective of this paper is to present a hybrid neural network model (NNM) for the simulation of a differential catalytic hydrogenation reactor of carbon dioxide to methanol. The hybrid model consists of two parts: a mechanistic model and a neural model. The mechanistic model employs heat transfer, mass transfer and pressure drop equations and calculates the effluent temperature of the reactor by taking outlet mole fractions from the output of a neural network model. The prepared hybrid model was used to simulate and identify an existing industrial methanol reactor. The bed of the reactor was assimilated to a pile of layers, each corresponding to a neural network (NN) model that can predict outlet composition of each layer as a function of time. The model was successfully tested with plant experimental data. The insights of this research indicate a very fast responding model in comparison to traditional models to demonstrate CO2 reduction as a function of time and reactor length. Variation of temperature and other compositions with time and bed height are also investigated in this article.