Search

The effect of fluid turbulence models on the coarse-grid simulation for the mesoscale gas-solid flow system is not as clear as that for gas-liquid/single flows. In this study, the effect of different turbulence models on predictions of riser-only, turbulent and bubbling flow hydrodynamics is quantified. Then the selected turbulence models are examined by coarse-grid simulations in a full-loop circulating fluidized bed (CFB) with a proposed hybrid drag model which integrates two SGMs and Hulin-Gidaspow model. Results show that the standard k -epsilon and k -omega models fail to capture a correct fluidization pattern in riser. Meanwhile, the RNG and realizable k -epsilon models and SST k -omega model enable satisfactory predictions which are highly comparable to discrete model predictions and experiments over different flow regimes. However, the realizable k -epsilon model produces more pronounced flow fluctuations causing computational instability. Using superior turbulence closures, full-loop simulations with the proposed hybrid model can predict desirable hydrodynamics. Unlabelled Image • A hybrid drag model is proposed for feasible coarse-grid full-loop CFB simulations. • Applicability of different turbulence models over various fluidization regimes. • Validated against the discrete particle model and experiments. • The RNG k -epsilon and SST k -omega are optimal for gas-solid flows. [ABSTRACT FROM AUTHOR]

Machine learning (ML) is experiencing an immensely fascinating resurgence in a wide variety of fields. However, applying such powerful ML to construct subgrid interphase closures has been rarely reported. To this end, we develop two data‐driven ML strategies (i.e., artificial neural networks and eXtreme gradient boosting) to accurately predict filtered subgrid drag corrections using big data from highly resolved simulations of gas‐particle fluidization. Quantitative assessments of effects of various subgrid input markers on training prediction outputs are performed and three‐marker choice is demonstrated to be the optimal one for predicting the unseen test set. We then develop a parallel data loader to integrate this predictive ML model into a computational fluid dynamic (CFD) framework. Subsequent coarse‐grid simulations agree fairly well with experiments regarding the underlying hydrodynamics in bubbling and turbulent fluidized beds. The present ML approach provides easily extended ways to facilitate the development of predictive models for multiphase flows. [ABSTRACT FROM AUTHOR]

In this work, a three-dimensional (3-D) filtered two-fluid model (TFM) was developed to describe the gas–solid flow behavior in a large-scale methanol-to-olefins (MTO) fluidized bed reactor (FBR). The cold-model flow behaviors were characterized successfully via the filtered TFM with a coarse grid. A coarse-grid sensitivity test was first carried out, and the filtered model was testified using predictions from classical models and the experimental data. Moreover, four drag models have been incorporated into the TFM for evaluating the effectiveness of these models at the same coarse-grid condition. Subsequently, the effects of some important model parameters including solid stresses, wall corrections and filter size on the flow behaviors were also investigated numerically. Finally, the filtered model was applied to predict the effects of the operating gas velocity, distributor shape and solid particle size. The results suggested the effectiveness of the sub-grid models for simulating large-scale MTO FBRs at coarse-grid conditions. This study further confirmed that the filtered drag coefficient correlation plays a significant role in capturing flow behaviors and the filter size is nearly independent on the grid size when the filter size is larger than or equal to twice the grid size. The simulation results by coarse-grid also show that the catalyst particles are easier to be fluidized with the increase of the operating gas velocity. It is also found that a triangle-shaped distributor strengthens the mixing behaviors and weakens the clustering of the near-wall regions. Additionally, our study indicates that the clustering phenomena near the wall regions are more obvious with the decrease of the catalyst particle size. [ABSTRACT FROM AUTHOR]