Prediction of fluid interface between dispersed and matrix phases by Lattice Boltzmann-adaptive network-based fuzzy inference system

Abstract

Listen

Translate article

ABSTRACT The integration of Lattice Boltzmann (LB) and adaptive-network-based fuzzy inference system was utilised to simulate mesoscale fluid interface in a multiphase fluid system. This method uses data generated by LB and based on the local population and density data. The interface between dispersed and matrix phases on the neural points was simulated. The neural mesh of interface was created by the ANFIS method and locally was compared by the CFD results. The results showed that there is a great agreement between ANFIS and numerical data when a particularly more number of rules are used in the learning step of simulations. Since this new overview of modelling can predict fluid flow based on existing data, less number of rules cause over prediction for the ANFIS method; however, by enhancing the number of rules, ANFIS can solve this over-prediction far from the curvature of the dispersed phase. This algorithm can be a promising tool for simulation of macroscopic parameters in the large-scale multi-phase reactors.