Introducing an artificial neural network energy minimization multi-scale drag scheme for fluidized particles

Nikolopoulos, A.; Samlis, Christos; Zeneli, Myrto; Νikolopoulos, Nikos; Karellas, Sotiriοs; Grammelis, Panagiotis

doi:10.1016/j.ces.2020.116013

Cited by 28 publications

(16 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Later, a revised cluster size model extended the applicability of the EMMS drag to both Geldart A and Geldart B particles . The bubble-based EMMS model extended its applicability to bubbling and turbulent fluidization regimes. − Further research includes the extension to binary systems , and the coupling with Euler–Lagrangian simulations. − Recently, a generic EMMS drag model was developed using an Artificial Neural Network (ANN) to cover a wide range of material properties and operating conditions. , …”

Section: Introductionmentioning

confidence: 99%

Development of a Filtered CFD-DEM Drag Model with Multiscale Markers Using an Artificial Neural Network and Nonlinear Regression

Gao

Dietiker

et al. 2021

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

The accuracy of coarse-grained Euler−Lagrangian simulations of fluidized beds heavily depends on the mesoscale drag models to account for the influences of the unresolved subgrid structures. Traditional filtered drag models are regressed with mesoscale markers such as voidage and slip velocities. In this research, a filtered drag was regressed with both mesoscale and macroscale markers using fine-grid Computational Fluid Dynamics−Discrete Element Method (CFD-DEM) simulations. The traditional nonlinear regression method was compared with machine learning regression using an Artificial Neural Network (ANN) implemented in PyTorch and coupled with MFiX. The new drag showed higher accuracy than the Wen−Yu drag and another filtered drag derived from the two-fluid model. The nonlinear regression shows slightly better results than ANN regression in cases with similar R 2 values. The utilization of the gas inlet velocity as an additional macroscale marker reduced the errors by up to 55.3% in the tested cases.

show abstract

Section: Introductionmentioning

confidence: 99%

Development of a Filtered CFD-DEM Drag Model with Multiscale Markers Using an Artificial Neural Network and Nonlinear Regression

Gao

Dietiker

et al. 2021

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

show abstract

“…In recent years, with the enhancement of computers' processing capability, the big data technology is developing rapidly and has been widely used in many branches of science and technology. Machine learning (ML) methods such as eXtreme gradient boosting (Xgboost), 38 support vector machine (SVM), 39 random forest, 40 and artificial neural network (ANN) 41‐46 have been increasingly used in the field of chemical engineering. It was around since the 1940s, 47 but it was not until the last two decades that the ANN was applied to engineering (e.g., the filtered drag development using ANN 32,41,46 ).…”

Section: Introductionmentioning

confidence: 99%

Data‐driven modeling of mesoscale solids stress closures for filtered two‐fluid model in gas–particle flows

2021

View full text Add to dashboard Cite

This study performs data‐driven modeling of mesoscale solids stress closures for filtered two‐fluid model (fTFM) in gas–particle flows via an artificial neural network (ANN) based machine learning method. The data used for developing the ANN‐based predictive data‐driven modeling framework is systematically filtered from fine‐grid simulations. The loss function optimization result reveals that coupling two loss functions promotes more accurate predictions of the mesoscale solids stresses than using a single loss function. Further comprehensive assessments of closure markers demonstrate a systematic dependence of the mesoscale solids stresses on the filtered particle velocity and its gradient as additional anisotropic markers, instead of using the conventional isotropic filtered rate of solid phase deformation as a closure marker. An optimal three‐marker mesoscale closure is thus proposed. Comparative analysis of the conventional filtered model and present three‐marker model shows that the data‐driven model can substantially enhance the prediction accuracy.

show abstract

“…It is important to select the appropriate drag models for the concerning problem since they considerably affect the computational results 32,33 . In this study, the Huilin‐Gidaspow drag model 34 integrated with an additional smooth switch in case of α s < 0.2 was adopted to characterize the momentum exchange coefficients between the air and sand phases 30…”

Section: Materials and Methodologymentioning

confidence: 99%

“…30 It is important to select the appropriate drag models for the concerning problem since they considerably affect the computational results. 32,33 In this study, the Huilin-Gidaspow drag model 34 integrated with an additional smooth switch in case of α s < 0.2 was adopted to characterize the momentum exchange coefficients between the air and sand phases. 30 Given substantial modifications in the formulations of turbulent viscosity, dissipation rate ε, and Reynolds stresses, the realizable k-ε model combined with dispersed turbulence method and standard wall functions was preferred to perform well the air-sand turbulent flow behaviors in the considering system.…”

Section: Mathematical Modelmentioning

confidence: 99%

Effect of sand particle size on the hydrodynamics of a dual fluidized bed cold flow system applied for waste‐to‐energy gasifiers

Dinh

Hsiau

et al. 2022

Intl J of Energy Research

View full text Add to dashboard Cite

Summary Since the dual fluidized bed (DFB) system has been increasingly adopted for waste‐to‐energy gasification, a three‐dimensional computational fluid dynamics full‐loop model of a DFB cold flow system has been developed to predict the pressure and sand circulation rate (SCR) under various operating conditions. The pressure data were recorded at 18 specified points along the system height. In addition, the SCRs were determined at the two crucial positions connecting the riser and the gasifier and vice versa. The simulation results were then validated against the experimental data, taking into account the effects of the sand particle size (ds), the air inlet velocity in the riser (v0,riser), and the initial gasifier bed height (h0,gasifier). It could be observed in the riser that the flow structures of the smaller ds(s) were almost scattering patterns, while those of the bigger ones were mainly clusters. Compared with bigger ds(s), more homogeneous pressure distributions in the riser and higher bed expansion in the gasifier were obtained with smaller ones. A compromise between the system performance and pre‐treating cost for the feedstock particle sizes should be considered further to achieve the best system performance. On the other hand, the increases of the v0,riser and the h0,gasifier enhanced the gasifier bed pressure, facilitating the downward sand flow back to the riser via the LLS. Although some minor discrepancies existed between the simulation and experimental data, the predicted tendencies agreed well with the measured ones. It was also found that a system operating with ds = 500 μm, at v0,riser = 4.0 m/s and h0,gasifier = 0.25 m yielded stable flow characteristics and optimal global SCR. Moreover, some unexpected phenomena predicted and verified with the experimental observations could be avoided or minimized using suitable particle sizes of the feedstock and operating parameters. In sum, the better validation results with lower errors than our previous study, the more reasonable air‐sand flow patterns, and the substantial predictions of undesirable problems are the significant improvements of this study, providing valuable information for effectively designing and operating the practical DFB hot flow systems. Novelty Statement This study developed a 3D computational fluid dynamics full‐loop model of a DFB cold flow system for predicting the pressure and sand circulation rates under various operating conditions of particle size, inlet velocity, and initial bed height. The better validation results against the experimental data, the more reasonable air‐sand flow patterns, and the substantial predictions of undesirable problems are the significant improvements of this study, providing helpful information for designing and operating the DFB hot flow gasifiers.

show abstract

Introducing an artificial neural network energy minimization multi-scale drag scheme for fluidized particles

Cited by 28 publications

References 49 publications

Development of a Filtered CFD-DEM Drag Model with Multiscale Markers Using an Artificial Neural Network and Nonlinear Regression

Development of a Filtered CFD-DEM Drag Model with Multiscale Markers Using an Artificial Neural Network and Nonlinear Regression

Data‐driven modeling of mesoscale solids stress closures for filtered two‐fluid model in gas–particle flows

Effect of sand particle size on the hydrodynamics of a dual fluidized bed cold flow system applied for waste‐to‐energy gasifiers

Contact Info

Product

Resources

About