Drag coefficient prediction for non-spherical particles in dense gas–solid two-phase flow using artificial neural network

Abstract

Artificial neural network (ANN) was adopted to predict and analyze the relationship of drag coefficient, Reynolds number, and sphericity for non-spherical particles in dense gas–solid two-phase flow. We first employed Back Propagation Neural Network (BPNN) and Radial Basis Function Neural Network (RBFNN) to predict drag coefficients based on experimental results (Pettyjohn, 1948; Yow et al., 2005). Comparisons between simulation and experimental results indicate that RBFNN efficiently predicts the drag coefficients with as high precision as BPNN. Furthermore, we made predictions and analyses of drag coefficients under different sphericities employing Radial basis function neural network. Results reveal that artificial neural network is applicable in predicting and investigating drag coefficient in gas–solid non-spherical particulate systems. Based on the predicted results of drag coefficients, we conducted a curve fitting of drag coefficient, Reynolds number, and particle sphericity, obtaining a correlation on drag coefficient. Incorporating the drag coefficient correlation with Syamlal-O'Brien and Gidaspow-blend model, simulation on gas–solid two-phase flow by Eulerian–Eulerian model was carried out from fixed beds to bubbling fluidized beds. Simulation pressure drop are compared with experimental results obtaining a good agreement with each other, which indicates that the drag force of non-spherical particles in a gas–solid system could be predicted reasonably by an artificial neural network method. This work provides a reference for predicting drag coefficients of particles with complex shape in gas–solid two-phase system.