Experimental and numerical investigations of a fixed-bed distributor for obtaining the outlet fluid velocity profile

Abstract

In this study, computational fluid dynamics (CFD) was used to simulate a fixed-bed distributor, investigating spherical particles in body-centered cubic (BCC) and hexagonally close-packed (HCP) structures. The bed-to-particle diameter ratio (D/dp) varied in the range of 4.158–16.65, while the range for the ratio of bed height to particle diameter (h/dp) was 5–13. The simulations were carried out for Reynolds number (Rep) in the range of 4–589, including laminar and turbulent flow regimes. To conduct validation, the numerical results were compared with our experimental data as well as seven empirical equations, where perfect match was found for both laminar and turbulent flows. Then simulations were conducted to generate the required data for an artificial neural network (ANN) to predict the velocity profile at the distributor outlet in order to save the computational CPU time. The R2, MAE and RMSE values of the neural network for predicting the fluid outlet velocity were 0.972, 0.0274 and 0.0512, respectively. The function obtained from the neural network is an efficient tool for the optimum design of fixed-bed distributors. This function could be directly used in three-dimensional fixed-bed distributor models. © 2021, The Author(s), under exclusive licence to Società Italiana di Fisica and Springer-Verlag GmbH Germany, part of Springer Nature.