Fuzzy neural network for mapping CFD data of air-water flow in chemical reactors: Computational simulation

Abstract

ANFIS (Adaptive Neuro-Fuzzy Inference Systems) artificial intelligence is used to enhance the computational fluid dynamic (CFD) modeling of an air-water bubble column reactor for the first time. Air temperature and velocity are 500 K and 0.006 m/s, respective in the process. The water temperature is considered at 295 K. This turbulent kinetic energy prediction is performed to measure the mixing flow within the reactor. The ANFIS projection is examined to see what impact the number of membership functions and input has. The findings showed that the most intelligent output is calculated by inputting two numbers and using a membership function with a value of 10. With respect to that, under this condition the ANFIS is capable of estimating the turbulent kinetic energy with more nodes on the domain while skipping mesh refinement in the CFD domain. The hybrid model was indicated to be robust in predicting the performance of liquid-phase reactor.