ExperimentsinFluids23(1997)262—2Springer-Verlag1997
Anoteonsecondaryflowinbendsandbendcombinations
H.E.Fiedler
262
AbstractInthispaperanexplanationfortheformationofswirlpastthree-dimensionalbendcombinationsinpipeflowsisgiven.Supportiveandcomparativeswirlmeasurementspastadoublebendcombinationandpastacombinationofanasymmetricdiaphragmandafollowinganglebendarepresentedanddiscussed.
1
Introduction
Aflowthroughathree-dimensionalcombinationoftwo
90°-bendswheretheexitplaneofthesecondoneisnotparalleltotheintakeplaneofthefirstoneexperiencesaswirl.Thisphenomenonhasoftenbeendescribed(e.g.Miller1978);asketchoftheconfigurationisgiveninFig.2).
Whilethecharacteristicdoublevortexdevelopingintheflowthroughasinglebendiswellknownandequallywellunder-stood(seee.g.Schlichting(1958)andFig.1a),nosimpleexplanationappearstobeathandforthecaseofadoublebendandindeedtheoftensuggestedsuperpositionofthesuccessivepairsofsecondaryvorticesisnotonlyquestionablebutdoesnotprovideanexplanationeither.Thereis,however,asimpleexplanationforwhichthesecondaryvorticesproducedbythefirstbendareofnoeffect:
Ofbasicimportancefortheflowpastabendistheveloc-itydistributionoftheflowinitsentranceplane.Thus,thesecondarytwin-vorticesareaconsequenceoftheprimaryvelocitydistributionbeinginhomogeneousbutsymmetricalaboutthex—y-plane.Asymmetriesofthevelocityprofileintheentranceaboutthex—y-planewillbynecessityleadtoasym-metriesintheensuingsecondaryflow,i.e.tounequalstrengthsofthecounter-rotatingvortices.Avelocityprofilepastafirstbendisasymmetricinthex—y-plane.Forasecondbend,turnedbyanangle0°::180°againstthefirstone,thisprofileisasymmetricinitsy—z-planewhichcausesasymmetryintheensuingsecondaryflow.Thisasymmetryisinmostcasesstrongenoughtosuppresstheformationofthesecond,smaller,vortex,leadingalmostimmediatelytoasinglevortex,
Received:21August1996/Accepted:12February1997H.E.FiedlerHermann-Fottinger-InstitutfurStromungsmechanik,TechnischeUniversitatBerlin,D-10623Berlin,GermanyThemeasurementsweredonebyD.Obermann.AdditionalhelpwasprovidedbyDipl.-Ing.R.Nagel.Theircontributionisgratefullyacknowledged.
thesenseofrotationofwhichisobviousfromFig.1b:Theswirlvectoroftheflowpastthesecondbendpointsinthedirectionofthevectorproductoftheprofileexcentricityzandthe
maximumvelocitycofthevelocityprofileinthemy—z-planein
theentranceofthembend(z;c).Thus,asymmetryofthe
primaryvelocityprofiletowardsmmpositivez-value(asinFig.1b)causesformationofaclockwiseswirlintheexitarea(y—z-plane)andviceversa—aconsequenceofthefactthatavorticityvectorretainsitsdirectionwhengoingthroughabend.
2
Experiments
Experimentswerecarriedoutinsupportofthisexplanation.ForthispurposeacommercialplasticpipeofinnerdiameterD:70mmandlengthL:700mmwasfittedwitharoundedinletmountedinsideapressurechamber.Commercialbendswereconnectedtothepipeexit.TheradiusofcurvatureofthecenterlineofthebendsusedinthisexperimentwasR+0.8D:56mm(seeFig.2).Alldataweretakenmean!*
withvelocities(averagedoverthepipecross-section)ofcm+10m/s,correspondingtoRe:cNodetailedvelocitymeasurementsm
D/+50000.weredoneaswewereonlyinterestedinbulkeffects.ForthispurposeabulkswirlmeterwithopticalRPMcounterwasinsertedinthepipeimmediatelydownstreamofsecondbend.Itconsistedofapaddle-wheeloftworectangularplatesalignedparalleltothepipeaxis,whicharemountedonalightlyrotableshaftinthepipeaxis.TheRPM-valuesS*measuredwiththisinstrumentprovidesonlyacomparativemeasureforthestrengthoftheswirlinthepipe,whichinthiscontextshouldsuffice.InparallelflowS*:0.ThisbulkquantityS*isrelatedtotherealswirlnumberSoftheflow.Anexactrelationshipwas,forlackofdetailedvelocitymeasurements,notevaluated.
FirstweinvestigatedtheflowpasttwosuccessivebendsfordifferentdistancesHbetweenbends.Figure2givesasketchwithdefinitions.ThemeasuredswirlnumberS*versus
angleisplottedinFig.3.Thisdistributionhasasine-likeappearancewithmaximumvaluesbetween:<70°and:<90°dependingonH.
Thefirstbendwasthenreplacedbyanasymmetricallyinsertedpartialdiaphragm,whichcreatesanasymmetricvelocityprofilesimilartothatpastasinglebendwithoutproducingsecondaryvortices.ThearrangementissketchedanddefinedinFig.4.Figures5and6givedistributionsofS*versusanglefordifferentrelativedistancesH/Dandrelativeinsertiondepthsb/Dofthediaphragm.Someofthose
Fig.1a,b.Secondaryflowinabendasaconsequenceofvelocityprofileintheentranceplane.Situationsforasingleandbdoublebend.Coordinatesforsecondbendare(x,y,z);forfirstbend(x,y,z)
Fig.2.Doublebendcombinationanddefinitions
Fig.3.BulkswirlS*pastdoublebendversusanglefordifferentdistanceH/D
measurementswerethenrepeatedwithaperforateddia-phragmof90%solidity,wheretheswirldistributionwasfoundtobemorelikethatinFig.3(Fig.7).Whenusedwitharotablesleevethisdiaphragmprovidesasimpleandflexiblemeanstoproduceswirlinapipeofvariablestrengthandsignwithoutnecessitationofgeometricalchangesaswouldbeneededinadoublebendarrangement.
Fig.4.Replacementoffirstbendbypartial(asymmetric)diaphragm;definitionsofdiaphragm-bendcombination
Fig.5.BulkswirlS*asfunctionofangleandrelativeinsertionb/D;H/D:1
Fig.6.BulkswirlS*asfunctionofangleandrelativeinsertionb/D;H/D:2
3
Summary
Theinvestigationhasshownthatasimpleexplanationfortheswirlpastadoublebendcanbegiven.Furtherobservationsare:
—Swirlofcomparablestrengthaswithadoublebendisalsoachievedwhenthefirstbendissubstitutedbyasidewayspartiallyinserteddiaphragm
263
2
Fig.7.BulkswirlS*asfunctionofangleandvariableb/DandH/D;perforateddiaphragm
—Perforateddiaphragmscausesmootherswirldistributions,moreakintothedoublebendcase
—Forthepurposeofcreatingswirlofvariablestrengthandsignthediaphragm-bendcombinationoffersmoreflexibilitythanthedoublebendcombination
—Partialsuppressionofswirlasfoundbehindacombinationoftwobendsispossiblebypositioningapartiallyinserteddiaphragmbetweenthetwobends.
References
SchlichtingH(1958)Grenzschichttheorie.Karlsruhe:G.Braun
MillerDS(1978)InternalFlowSystems.BedfordUK:BritishHydro-mechanicsResearchAssociation,pp50—51
.