2020
DOI: 10.1002/aic.16973
|View full text |Cite
|
Sign up to set email alerts
|

Machine learning to assist filtered two‐fluid model development for dense gas–particle flows

Abstract: 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 in… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
2

Citation Types

1
62
0

Year Published

2020
2020
2022
2022

Publication Types

Select...
5

Relationship

4
1

Authors

Journals

citations
Cited by 106 publications
(65 citation statements)
references
References 59 publications
1
62
0
Order By: Relevance
“…In addition, we have developed a data loader program to transfer mesoscale solids stress data between the CFD solver and the ANN computing process in order to assess the prediction reliability and accuracy of the ANN‐based mesoscale solids stress model in realistic coarse‐grid simulations. The detailed coupling scheme can be found in our previous work 32 . Here, an example of the coarse‐grid results predicted by the coupled CFD‐DDM simulations is presented in Supporting Information.…”
Section: Assessment Of Model Developmentmentioning
confidence: 99%
See 2 more Smart Citations
“…In addition, we have developed a data loader program to transfer mesoscale solids stress data between the CFD solver and the ANN computing process in order to assess the prediction reliability and accuracy of the ANN‐based mesoscale solids stress model in realistic coarse‐grid simulations. The detailed coupling scheme can be found in our previous work 32 . Here, an example of the coarse‐grid results predicted by the coupled CFD‐DDM simulations is presented in Supporting Information.…”
Section: Assessment Of Model Developmentmentioning
confidence: 99%
“…In the establishment of the fTFM, correlations for mesoscale drag and mesoscale stresses are fitted as a function of some filtered quantities, the so‐called markers. Most of the previous studies have been focusing on heterogeneous drag closures, in some of which reasonable predictions are made by correcting the homogeneous drag model alone 13,15,16,18,20,28‐32 . But other studies 18,33 revealed that the mesoscale stresses, which are at least one order of magnitude larger than the microscale stresses, should not be ignored.…”
Section: Introductionmentioning
confidence: 99%
See 1 more Smart Citation
“…The filtering procedure adopted for microscopic TFM and CFD–DEM formulations is analogous as those in the literature 19,35,46 . The definition of the filtered particle phase volume fraction is as follows ϕfalse¯s()boldx,t=ϕs()boldr,tG()boldrboldxdboldr0.25em …”
Section: Cfd–dem Model Developmentmentioning
confidence: 99%
“…However, the DDM does not require an understanding of the specific function relationship between the feature inputs and output, thus is more flexible in predicting the target variable 15 . Recently, the DDM has been demonstrated to be an effective way to model the mesoscale drag correction, 16–20 turbulence closures, 21 and flow parameters (e.g., solid hold‐up, solid flux) 22–24 . For instance, Jiang et al 16 applied the ANN to estimate the filtered HDC (fHDC) correction in gas–particle fluidization.…”
Section: Introductionmentioning
confidence: 99%