Toward accelerated data-driven Rayleigh–Bénard convection simulations

Toward accelerated data-driven Rayleigh–Bénard convection simulations. Alieva, A., Hoyer, S., Brenner, M., Iaccarino, G., & Norgaard, P. The European Physical Journal E, 46(7):64, July, 2023.

Paper doi abstract bibtex

A hybrid data-driven/finite volume method for 2D and 3D thermal convective flows is introduced. The approach relies on a single-step loss, convolutional neural network that is active only in the near-wall region of the flow. We demonstrate that the method significantly reduces errors in the prediction of the heat flux over the long-time horizon and increases pointwise accuracy in coarse simulations, when compared to direct computations on the same grids with and without a traditional subgrid model. We trace the success of our machine learning model to the choice of the training procedure, incorporating both the temporal flow development and distributional bias.

@article{alieva_toward_2023,
	title = {Toward accelerated data-driven {Rayleigh}–{Bénard} convection simulations},
	volume = {46},
	issn = {1292-895X},
	url = {https://doi.org/10.1140/epje/s10189-023-00302-w},
	doi = {10.1140/epje/s10189-023-00302-w},
	abstract = {A hybrid data-driven/finite volume method for 2D and 3D thermal convective flows is introduced. The approach relies on a single-step loss, convolutional neural network that is active only in the near-wall region of the flow. We demonstrate that the method significantly reduces errors in the prediction of the heat flux over the long-time horizon and increases pointwise accuracy in coarse simulations, when compared to direct computations on the same grids with and without a traditional subgrid model. We trace the success of our machine learning model to the choice of the training procedure, incorporating both the temporal flow development and distributional bias.},
	language = {en},
	number = {7},
	urldate = {2024-02-22},
	journal = {The European Physical Journal E},
	author = {Alieva, Ayya and Hoyer, Stephan and Brenner, Michael and Iaccarino, Gianluca and Norgaard, Peter},
	month = jul,
	year = {2023},
	keywords = {Precourt, SOE, Sustainability},
	pages = {64},
}

Downloads: 0

{"_id":"3EB2nY2PqyGjWpwNh","bibbaseid":"alieva-hoyer-brenner-iaccarino-norgaard-towardaccelerateddatadrivenrayleighbnardconvectionsimulations-2023","author_short":["Alieva, A.","Hoyer, S.","Brenner, M.","Iaccarino, G.","Norgaard, P."],"bibdata":{"bibtype":"article","type":"article","title":"Toward accelerated data-driven Rayleigh–Bénard convection simulations","volume":"46","issn":"1292-895X","url":"https://doi.org/10.1140/epje/s10189-023-00302-w","doi":"10.1140/epje/s10189-023-00302-w","abstract":"A hybrid data-driven/finite volume method for 2D and 3D thermal convective flows is introduced. The approach relies on a single-step loss, convolutional neural network that is active only in the near-wall region of the flow. We demonstrate that the method significantly reduces errors in the prediction of the heat flux over the long-time horizon and increases pointwise accuracy in coarse simulations, when compared to direct computations on the same grids with and without a traditional subgrid model. We trace the success of our machine learning model to the choice of the training procedure, incorporating both the temporal flow development and distributional bias.","language":"en","number":"7","urldate":"2024-02-22","journal":"The European Physical Journal E","author":[{"propositions":[],"lastnames":["Alieva"],"firstnames":["Ayya"],"suffixes":[]},{"propositions":[],"lastnames":["Hoyer"],"firstnames":["Stephan"],"suffixes":[]},{"propositions":[],"lastnames":["Brenner"],"firstnames":["Michael"],"suffixes":[]},{"propositions":[],"lastnames":["Iaccarino"],"firstnames":["Gianluca"],"suffixes":[]},{"propositions":[],"lastnames":["Norgaard"],"firstnames":["Peter"],"suffixes":[]}],"month":"July","year":"2023","keywords":"Precourt, SOE, Sustainability","pages":"64","bibtex":"@article{alieva_toward_2023,\n\ttitle = {Toward accelerated data-driven {Rayleigh}–{Bénard} convection simulations},\n\tvolume = {46},\n\tissn = {1292-895X},\n\turl = {https://doi.org/10.1140/epje/s10189-023-00302-w},\n\tdoi = {10.1140/epje/s10189-023-00302-w},\n\tabstract = {A hybrid data-driven/finite volume method for 2D and 3D thermal convective flows is introduced. The approach relies on a single-step loss, convolutional neural network that is active only in the near-wall region of the flow. We demonstrate that the method significantly reduces errors in the prediction of the heat flux over the long-time horizon and increases pointwise accuracy in coarse simulations, when compared to direct computations on the same grids with and without a traditional subgrid model. We trace the success of our machine learning model to the choice of the training procedure, incorporating both the temporal flow development and distributional bias.},\n\tlanguage = {en},\n\tnumber = {7},\n\turldate = {2024-02-22},\n\tjournal = {The European Physical Journal E},\n\tauthor = {Alieva, Ayya and Hoyer, Stephan and Brenner, Michael and Iaccarino, Gianluca and Norgaard, Peter},\n\tmonth = jul,\n\tyear = {2023},\n\tkeywords = {Precourt, SOE, Sustainability},\n\tpages = {64},\n}\n\n\n\n","author_short":["Alieva, A.","Hoyer, S.","Brenner, M.","Iaccarino, G.","Norgaard, P."],"key":"alieva_toward_2023","id":"alieva_toward_2023","bibbaseid":"alieva-hoyer-brenner-iaccarino-norgaard-towardaccelerateddatadrivenrayleighbnardconvectionsimulations-2023","role":"author","urls":{"Paper":"https://doi.org/10.1140/epje/s10189-023-00302-w"},"keyword":["Precourt","SOE","Sustainability"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"article","biburl":"https://bibbase.org/zotero-group/madison.pobis/5118222","dataSources":["AJrnt9YgYmPoCrDbx"],"keywords":["precourt","soe","sustainability"],"search_terms":["toward","accelerated","data","driven","rayleigh","nard","convection","simulations","alieva","hoyer","brenner","iaccarino","norgaard"],"title":"Toward accelerated data-driven Rayleigh–Bénard convection simulations","year":2023}