Machine Learning and Its Application to Reacting Flows. ML and Combustion PDF
Preview Machine Learning and Its Application to Reacting Flows. ML and Combustion
Lecture Notes in Energy 44 Nedunchezhian Swaminathan Alessandro Parente Editors Machine Learning and Its Application to Reacting Flows ML and Combustion Lecture Notes in Energy Volume 44 LectureNotesinEnergy(LNE)isaseriesthatreportsonnewdevelopmentsinthe studyofenergy:fromscienceandengineeringtotheanalysisofenergypolicy.The series’scopeincludesbutisnotlimitedto,renewableandgreenenergy,nuclear,fossil fuels and carbon capture, energy systems, energy storage and harvesting, batteries andfuelcells,powersystems,energyefficiency,energyinbuildings,energypolicy,as wellasenergy-relatedtopicsineconomics,managementandtransportation.Books published in LNE are original and timely and bridge between advanced textbooks andtheforefrontofresearch.ReadersofLNEincludepostgraduatestudentsandnon- specialistresearcherswishingtogainanaccessibleintroductiontoafieldofresearch aswellasprofessionalsandresearcherswithaneedforanup-to-datereferencebook onawell-definedtopic.Theseriespublishessingle-andmulti-authoredvolumesas wellasadvancedtextbooks. **Indexed in Scopus and EI Compendex** The Springer Energy board welcomes yourbookproposal.PleasegetintouchwiththeseriesviaAnthonyDoyle,Executive Editor,Springer([email protected]) · Nedunchezhian Swaminathan Alessandro Parente Editors Machine Learning and Its Application to Reacting Flows ML and Combustion Editors NedunchezhianSwaminathan AlessandroParente DepartmentofEngineering Aero-Thermo-MechanicsLaboratory UniversityofCambridge ÉcolepolytechniquedeBruxelles Cambridge,UK UniversitéLibredeBruxelles Brussels,Belgium BrusselsInstituteforThermal-fluid Systems,Brussels(BRITE) UniversitéLibredeBruxellesandVrije UniversiteitBrussel Brussels,Belgium ISSN 2195-1284 ISSN 2195-1292 (electronic) LectureNotesinEnergy ISBN 978-3-031-16247-3 ISBN 978-3-031-16248-0 (eBook) https://doi.org/10.1007/978-3-031-16248-0 ©TheEditor(s)(ifapplicable)andTheAuthor(s)2023.Thisbookisanopenaccesspublication. OpenAccessThisbookislicensedunderthetermsoftheCreativeCommonsAttribution4.0International License(http://creativecommons.org/licenses/by/4.0/),whichpermitsuse,sharing,adaptation,distribu- tionandreproductioninanymediumorformat,aslongasyougiveappropriatecredittotheoriginal author(s)andthesource,providealinktotheCreativeCommonslicenseandindicateifchangeswere made. Theimagesorotherthirdpartymaterialinthisbookareincludedinthebook’sCreativeCommonslicense, unlessindicatedotherwiseinacreditlinetothematerial.Ifmaterialisnotincludedinthebook’sCreative Commonslicenseandyourintendeduseisnotpermittedbystatutoryregulationorexceedsthepermitted use,youwillneedtoobtainpermissiondirectlyfromthecopyrightholder. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthors,andtheeditorsaresafetoassumethattheadviceandinformationinthisbook arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor theeditorsgiveawarranty,expressedorimplied,withrespecttothematerialcontainedhereinorforany errorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictional claimsinpublishedmapsandinstitutionalaffiliations. ThisSpringerimprintispublishedbytheregisteredcompanySpringerNatureSwitzerlandAG Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland Preface Machine learning (ML) has been around for many decades and has been explored inthepastformanypracticalapplications.Currently,MLisinterpretedinabroader context and finding its way into a number of sectors such as Engineering, Health Care,TransportincludingTrafficPredictionandControl,driverlesscar,Information Technology,BigDataAnalysisandProcessing,Agriculture,Agronomy,etc.Ithas founditswayalsointoourdailylife,forexample,temperatureandlightingcontrols andinformationsearchesontheinternet.Inanutshell,MLisnothingbutstatistical interferenceusingdatacollectedorknowledgegainedthroughpasttargetedstudies or real-life experiences. The sophistication level of ML depends on the intended application and the advanced nature of the algorithms used for statistical learning and inference. This area has attracted huge interest recently because of the advent of the computational power, technology and algorithms required for data training, verificationandvalidation,andthereadinessandavailabilityofthesealgorithmsfor applicationtoawiderangeoffieldsandpracticalsystems.Hence,itisverytimely to overview the various ML techniques or algorithms for big data analyses with a specificapplicationtocombustionscienceandtechnology. Thisparticulartopicischosenbecauseoftheimportantroleofcombustionsystems andtechnologiescoveringmorethan90%oftheworld’stotalprimaryenergysupply (TPES). Although alternative renewable energy technologies are coming up, their sharesfortheTPESarelessthan5%currentlyandoneneedsacompleteparadigm shift to replace combustion sources. Whether this is practical or not is entirely a differentquestionandananswertothisquestionislikelytodependontherespon- dent. However, a pragmatic analysis suggests that the combustion share to TPES islikelytobemorethan70%evenby2070asdiscussedinthechapter“Introduc- tion”ofthisbook.Hence,itwillbeprudenttotakeadvantageofMLtechniquesto improvecombustionsciencesandtechnologiestobettercombustionsystemdesign anddevelopmentsothattheemissionofgreenhousegasescanbecurtailedalongwith improvingoverallefficiencies.ThelevelofinterestinapplyingMLtocombustionis clearlyevidentfromtherecentsurgeinresearchactivitiesonthistopic.Hence,the aimofthisvolumeistobringthisknowledgetogetherandmakeitreadilyaccessible forresearchersandgraduatestudentsinterestedinthismulti-andcross-disciplinary v vi Preface topic. We attempted to keep the discussion accessible to students and researchers interestedinturbulentcombustion,MLtechniques,anditsapplicationtoturbulence andcombustiononasimplephysicalbasiswhilehighlightingtheneedforML. Chapter“Introduction”givesanintroductiontotheroleofcombustiontechnolo- giesinthefuturepurelybasedonthecurrentpracticalandscientificevidence.This chapteralsoidentifiestheopportunitiestouseMLalgorithms(MLA)whileinvesti- gatingturbulentcombustion.Thechapter“MachineLearningTechniquesinReactive AtomisticSimulations”surveysvariousMLtechniquesanddiscussestheirapplica- tionforestimatingatomicpotentialenergies,requiredforchemicalkinetics,through moleculardynamicssimulationasanexample.Thechapter“ANovelInSituMachine LearningFrameworkforIntelligentDataCaptureandEventDetection”introduces in situ training for MLA which is a useful idea as it can save considerable efforts required in the training phase while using MLA. The chapter “Machine-Learning for Stress Tensor Modelling in Large Eddy Simulation” discusses the use of ML to estimate subgrid scale stresses and fluxes needed for large eddy simulation of turbulentcombustion.TheapplicationofMLforcombustionchemistryisdiscussed in the chapter “Machine Learning for Combustion Chemistry”. The turbulence- chemistry interaction is a highly nonlinear stochastic problem ideally suited for ML and chapters “Deep Convolutional Neural Networks for Subgrid-Scale Flame WrinklingModeling–AISuper-Resolution:ApplicationtoTurbulenceandCombus- tion” give different perspectives on the use of ML for estimating filtered reaction rate. Data-driven approaches can also be leveraged for reduced-order modeling of turbulentcombustionandthisisdiscussedinthechapter“Reduced-OrderModeling of Reacting Flows Using Data-Driven Approaches”. The use of ML for thermoa- cousticsisdescribedinchapter“MachineLearningforThermoacoustics”.Someof thesechaptersarewritteninatutorialfashionandalsoprovidehyperlinkstoaccess the associated computer codes. The concluding remarks and future directions are summarisedinthefinalchapter.Eachofthechaptersprovidesamplereferencesfor furtherreadingbycuriousreaders. Theideaforthisbookcameduringacollaborativeproject,ALCHEMY(mAchine LearningforComplExMultiphYsicsproblems),betweenCambridgeUniversityand ULBfundedbyFondationWiener-Anspach,ULB,Brussels.Thefundingfromthis foundationisgratefullyacknowledged.Wecannotunderstatethededicationofthe contributorstothisvolumeandwethankthemfortheircontributions. Cambridge,UK NedunchezhianSwaminathan Brussels,Belgium AlessandroParente May2022 Contents Introduction ...................................................... 1 N.SwaminathanandA.Parente MachineLearningTechniquesinReactiveAtomisticSimulations ...... 15 H.Aktulga,V.Ravindra,A.Grama,andS.Pandit ANovelInSituMachineLearningFrameworkforIntelligentData CaptureandEventDetection ....................................... 53 T.M.Shead, I.K.Tezaur, W.L.DavisIV, M.L.Carlson, D.M.Dunlavy, E.J.Parish, P.J.Blonigan, J.Tencer, F.Rizzi, andH.Kolla Machine-LearningforStressTensorModellinginLargeEddy Simulation ........................................................ 89 Z.M.Nikolaou,Y.Minamoto,C.Chrysostomou,andL.Vervisch MachineLearningforCombustionChemistry ........................ 117 T.Echekki,A.Farooq,M.Ihme,andS.M.Sarathy DeepConvolutionalNeuralNetworksforSubgrid-ScaleFlame WrinklingModeling ............................................... 149 V.XingandC.J.Lapeyre MachineLearningStrategyforSubgridModelingofTurbulent CombustionUsingLinearEddyMixingBasedTabulation ............. 175 R.Ranjan,A.Panchal,S.Karpe,andS.Menon OntheUseofMachineLearningforSubgridScaleFilteredDensity FunctionModellinginLargeEddySimulationsofCombustion Systems ........................................................... 209 S.Iavarone,H.Yang,Z.Li,Z.X.Chen,andN.Swaminathan Reduced-OrderModelingofReactingFlowsUsingData-Driven Approaches ....................................................... 245 K.Zdybał,M.R.Malik,A.Coussement,J.C.Sutherland,andA.Parente vii viii Contents AISuper-Resolution:ApplicationtoTurbulenceandCombustion ...... 279 M.Bode MachineLearningforThermoacoustics ............................. 307 MatthewP.Juniper Summary ......................................................... 339 Index ............................................................. 341
The list of books you might like
