Machine learning for nonintrusive model order reduction of the parametric inviscid transonic flow past an airfoil
Fluid flow in the transonic regime finds relevance in aerospace engineering, particularly in the design of commercial air transportation vehicles. Computational fluid dynamics models of transonic flow for aerospace applications are computationally expensive to solve because of the high degrees of fr...
Gespeichert in:
Veröffentlicht in: | Physics of fluids (1994) 2020-04, Vol.32 (4) |
---|---|
Hauptverfasser: | , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | |
---|---|
container_issue | 4 |
container_start_page | |
container_title | Physics of fluids (1994) |
container_volume | 32 |
creator | Renganathan, S. Ashwin Maulik, Romit Rao, Vishwas |
description | Fluid flow in the transonic regime finds relevance in aerospace engineering, particularly in the design of commercial air transportation vehicles. Computational fluid dynamics models of transonic flow for aerospace applications are computationally expensive to solve because of the high degrees of freedom as well as the coupled nature of the conservation laws. While these issues pose a bottleneck for the use of such models in aerospace design, computational costs can be significantly minimized by constructing special, structure-preserving surrogate models called reduced-order models. In this work, we propose a machine learning method to construct reduced-order models via deep neural networks and we demonstrate its ability to preserve accuracy with a significantly lower computational cost. In addition, our machine learning methodology is physics-informed and constrained through the utilization of an interpretable encoding by way of proper orthogonal decomposition. Application to the inviscid transonic flow past the RAE2822 airfoil under varying freestream Mach numbers and angles of attack, as well as airfoil shape parameters with a deforming mesh, shows that the proposed approach adapts to high-dimensional parameter variation well. Notably, the proposed framework precludes the knowledge of numerical operators utilized in the data generation phase, thereby demonstrating its potential utility in the fast exploration of design space for diverse engineering applications. Comparison against a projection-based nonintrusive model order reduction method demonstrates that the proposed approach produces comparable accuracy and yet is orders of magnitude computationally cheap to evaluate, despite being agnostic to the physics of the problem. |
format | Article |
fullrecord | <record><control><sourceid>osti</sourceid><recordid>TN_cdi_osti_scitechconnect_1632325</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1632325</sourcerecordid><originalsourceid>FETCH-osti_scitechconnect_16323253</originalsourceid><addsrcrecordid>eNqNjMFKAzEQQENRsNb-w-B9IbvB1J6l0os37yVkJ92RdMZOpvX33YIf4Om9w-Mt3LL3r9tuE2O8u_nGdzGG_sE9tvblvQ_bIS7d-SPliRihYlImPkIRBZZZTS-NrggnGbGC6IgKiuMlGwmDFLAJ4TtpOqEpZSC-Uss0gmniNh8ylCo_c9IMEkMiLUL1yd2XVBuu_7hyz--7z7d9J83oMA8M85SFGbMd-hiGMLyEf0W_Z3dNZA</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Machine learning for nonintrusive model order reduction of the parametric inviscid transonic flow past an airfoil</title><source>AIP Journals Complete</source><source>Alma/SFX Local Collection</source><creator>Renganathan, S. Ashwin ; Maulik, Romit ; Rao, Vishwas</creator><creatorcontrib>Renganathan, S. Ashwin ; Maulik, Romit ; Rao, Vishwas ; Argonne National Lab. (ANL), Argonne, IL (United States)</creatorcontrib><description>Fluid flow in the transonic regime finds relevance in aerospace engineering, particularly in the design of commercial air transportation vehicles. Computational fluid dynamics models of transonic flow for aerospace applications are computationally expensive to solve because of the high degrees of freedom as well as the coupled nature of the conservation laws. While these issues pose a bottleneck for the use of such models in aerospace design, computational costs can be significantly minimized by constructing special, structure-preserving surrogate models called reduced-order models. In this work, we propose a machine learning method to construct reduced-order models via deep neural networks and we demonstrate its ability to preserve accuracy with a significantly lower computational cost. In addition, our machine learning methodology is physics-informed and constrained through the utilization of an interpretable encoding by way of proper orthogonal decomposition. Application to the inviscid transonic flow past the RAE2822 airfoil under varying freestream Mach numbers and angles of attack, as well as airfoil shape parameters with a deforming mesh, shows that the proposed approach adapts to high-dimensional parameter variation well. Notably, the proposed framework precludes the knowledge of numerical operators utilized in the data generation phase, thereby demonstrating its potential utility in the fast exploration of design space for diverse engineering applications. Comparison against a projection-based nonintrusive model order reduction method demonstrates that the proposed approach produces comparable accuracy and yet is orders of magnitude computationally cheap to evaluate, despite being agnostic to the physics of the problem.</description><identifier>ISSN: 1070-6631</identifier><identifier>EISSN: 1089-7666</identifier><language>eng</language><publisher>United States: American Institute of Physics (AIP)</publisher><subject>Aerodynamics ; Aerospace engineering ; Artificial neural networks ; Computational fluid dynamics ; Fluid mechanics ; Functions and mappings ; Machine learning ; MATHEMATICS AND COMPUTING ; Numerical linear algebra ; Transonic flows</subject><ispartof>Physics of fluids (1994), 2020-04, Vol.32 (4)</ispartof><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><orcidid>0000000197318936 ; 0000000169486932 ; 0000000243956075</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>230,314,776,780,881</link.rule.ids><backlink>$$Uhttps://www.osti.gov/servlets/purl/1632325$$D View this record in Osti.gov$$Hfree_for_read</backlink></links><search><creatorcontrib>Renganathan, S. Ashwin</creatorcontrib><creatorcontrib>Maulik, Romit</creatorcontrib><creatorcontrib>Rao, Vishwas</creatorcontrib><creatorcontrib>Argonne National Lab. (ANL), Argonne, IL (United States)</creatorcontrib><title>Machine learning for nonintrusive model order reduction of the parametric inviscid transonic flow past an airfoil</title><title>Physics of fluids (1994)</title><description>Fluid flow in the transonic regime finds relevance in aerospace engineering, particularly in the design of commercial air transportation vehicles. Computational fluid dynamics models of transonic flow for aerospace applications are computationally expensive to solve because of the high degrees of freedom as well as the coupled nature of the conservation laws. While these issues pose a bottleneck for the use of such models in aerospace design, computational costs can be significantly minimized by constructing special, structure-preserving surrogate models called reduced-order models. In this work, we propose a machine learning method to construct reduced-order models via deep neural networks and we demonstrate its ability to preserve accuracy with a significantly lower computational cost. In addition, our machine learning methodology is physics-informed and constrained through the utilization of an interpretable encoding by way of proper orthogonal decomposition. Application to the inviscid transonic flow past the RAE2822 airfoil under varying freestream Mach numbers and angles of attack, as well as airfoil shape parameters with a deforming mesh, shows that the proposed approach adapts to high-dimensional parameter variation well. Notably, the proposed framework precludes the knowledge of numerical operators utilized in the data generation phase, thereby demonstrating its potential utility in the fast exploration of design space for diverse engineering applications. Comparison against a projection-based nonintrusive model order reduction method demonstrates that the proposed approach produces comparable accuracy and yet is orders of magnitude computationally cheap to evaluate, despite being agnostic to the physics of the problem.</description><subject>Aerodynamics</subject><subject>Aerospace engineering</subject><subject>Artificial neural networks</subject><subject>Computational fluid dynamics</subject><subject>Fluid mechanics</subject><subject>Functions and mappings</subject><subject>Machine learning</subject><subject>MATHEMATICS AND COMPUTING</subject><subject>Numerical linear algebra</subject><subject>Transonic flows</subject><issn>1070-6631</issn><issn>1089-7666</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><recordid>eNqNjMFKAzEQQENRsNb-w-B9IbvB1J6l0os37yVkJ92RdMZOpvX33YIf4Om9w-Mt3LL3r9tuE2O8u_nGdzGG_sE9tvblvQ_bIS7d-SPliRihYlImPkIRBZZZTS-NrggnGbGC6IgKiuMlGwmDFLAJ4TtpOqEpZSC-Uss0gmniNh8ylCo_c9IMEkMiLUL1yd2XVBuu_7hyz--7z7d9J83oMA8M85SFGbMd-hiGMLyEf0W_Z3dNZA</recordid><startdate>20200423</startdate><enddate>20200423</enddate><creator>Renganathan, S. Ashwin</creator><creator>Maulik, Romit</creator><creator>Rao, Vishwas</creator><general>American Institute of Physics (AIP)</general><scope>OIOZB</scope><scope>OTOTI</scope><orcidid>https://orcid.org/0000000197318936</orcidid><orcidid>https://orcid.org/0000000169486932</orcidid><orcidid>https://orcid.org/0000000243956075</orcidid></search><sort><creationdate>20200423</creationdate><title>Machine learning for nonintrusive model order reduction of the parametric inviscid transonic flow past an airfoil</title><author>Renganathan, S. Ashwin ; Maulik, Romit ; Rao, Vishwas</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-osti_scitechconnect_16323253</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Aerodynamics</topic><topic>Aerospace engineering</topic><topic>Artificial neural networks</topic><topic>Computational fluid dynamics</topic><topic>Fluid mechanics</topic><topic>Functions and mappings</topic><topic>Machine learning</topic><topic>MATHEMATICS AND COMPUTING</topic><topic>Numerical linear algebra</topic><topic>Transonic flows</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Renganathan, S. Ashwin</creatorcontrib><creatorcontrib>Maulik, Romit</creatorcontrib><creatorcontrib>Rao, Vishwas</creatorcontrib><creatorcontrib>Argonne National Lab. (ANL), Argonne, IL (United States)</creatorcontrib><collection>OSTI.GOV - Hybrid</collection><collection>OSTI.GOV</collection><jtitle>Physics of fluids (1994)</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Renganathan, S. Ashwin</au><au>Maulik, Romit</au><au>Rao, Vishwas</au><aucorp>Argonne National Lab. (ANL), Argonne, IL (United States)</aucorp><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Machine learning for nonintrusive model order reduction of the parametric inviscid transonic flow past an airfoil</atitle><jtitle>Physics of fluids (1994)</jtitle><date>2020-04-23</date><risdate>2020</risdate><volume>32</volume><issue>4</issue><issn>1070-6631</issn><eissn>1089-7666</eissn><abstract>Fluid flow in the transonic regime finds relevance in aerospace engineering, particularly in the design of commercial air transportation vehicles. Computational fluid dynamics models of transonic flow for aerospace applications are computationally expensive to solve because of the high degrees of freedom as well as the coupled nature of the conservation laws. While these issues pose a bottleneck for the use of such models in aerospace design, computational costs can be significantly minimized by constructing special, structure-preserving surrogate models called reduced-order models. In this work, we propose a machine learning method to construct reduced-order models via deep neural networks and we demonstrate its ability to preserve accuracy with a significantly lower computational cost. In addition, our machine learning methodology is physics-informed and constrained through the utilization of an interpretable encoding by way of proper orthogonal decomposition. Application to the inviscid transonic flow past the RAE2822 airfoil under varying freestream Mach numbers and angles of attack, as well as airfoil shape parameters with a deforming mesh, shows that the proposed approach adapts to high-dimensional parameter variation well. Notably, the proposed framework precludes the knowledge of numerical operators utilized in the data generation phase, thereby demonstrating its potential utility in the fast exploration of design space for diverse engineering applications. Comparison against a projection-based nonintrusive model order reduction method demonstrates that the proposed approach produces comparable accuracy and yet is orders of magnitude computationally cheap to evaluate, despite being agnostic to the physics of the problem.</abstract><cop>United States</cop><pub>American Institute of Physics (AIP)</pub><orcidid>https://orcid.org/0000000197318936</orcidid><orcidid>https://orcid.org/0000000169486932</orcidid><orcidid>https://orcid.org/0000000243956075</orcidid><oa>free_for_read</oa></addata></record> |
fulltext | fulltext |
identifier | ISSN: 1070-6631 |
ispartof | Physics of fluids (1994), 2020-04, Vol.32 (4) |
issn | 1070-6631 1089-7666 |
language | eng |
recordid | cdi_osti_scitechconnect_1632325 |
source | AIP Journals Complete; Alma/SFX Local Collection |
subjects | Aerodynamics Aerospace engineering Artificial neural networks Computational fluid dynamics Fluid mechanics Functions and mappings Machine learning MATHEMATICS AND COMPUTING Numerical linear algebra Transonic flows |
title | Machine learning for nonintrusive model order reduction of the parametric inviscid transonic flow past an airfoil |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-03T08%3A01%3A32IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-osti&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Machine%20learning%20for%20nonintrusive%20model%20order%20reduction%20of%20the%20parametric%20inviscid%20transonic%20flow%20past%20an%20airfoil&rft.jtitle=Physics%20of%20fluids%20(1994)&rft.au=Renganathan,%20S.%20Ashwin&rft.aucorp=Argonne%20National%20Lab.%20(ANL),%20Argonne,%20IL%20(United%20States)&rft.date=2020-04-23&rft.volume=32&rft.issue=4&rft.issn=1070-6631&rft.eissn=1089-7666&rft_id=info:doi/&rft_dat=%3Costi%3E1632325%3C/osti%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |