Liquid‐phase chemical reactors: Development of 3D hybrid model based on CFD‐adaptive network‐based fuzzy inference system

Abstract

AbstractThe Euler‐Euler method plus an intelligent algorithm was used to predict bubbly flow in a reactor as a function of column height. The combination of computational fluid dynamics (CFD) and the adaptive network‐based fuzzy inference system (ANFIS) method was used for a chemical bubble column reactor to understand the complex behaviour of fluids in a multiphase reactor. Air fraction as one of the main factors in the scale‐up of reactors was selected as an output parameter for the prediction tool (ANFIS method) at different positions in the reactor. To train and test the prediction ability of this method, a 3D position of one CFD element was selected and, based on that position, a training algorithm was started. After an appropriate learning step, the method was used to simulate gas at different locations of the reactor. The different structures of the ANFIS algorithm were designed to obtain a correct predictive tool for fluid behaviour inside the bubble column. The ANFIS approach shows that it can simulate the liquid behaviour and the CFD results and ANFIS output correlate with one another.