Figure4cillustratesthetransientpartafterkmax,1andkmax,2aresettoone-halfoftheirpreviousvalue(comparewithFigure4b).Becausetheflowratek01islargerthanthenewlysetparameterkmax,1,k01decreasesevenforhigh[I-]inthereactor.Afterseveralbursts,theflowratek01dropsbelowkmax,1.Theflowratek01cannotexceedtheparameterkmax,1butmustdecreaseiftheinitialvalueofk01isabovekmax,1.Theratiokmax,1/kmax,2playsamoresignificantroleinthedynamicsthantheabsolutevaluesofkmax,1andkmax,2.Whenwechangekmax,1andkmax,2whilemaintainingtheirratio,theburstingpersistswithonlyslightchanges(seeFigure4b,c).NumericalSection
Model.ThemodeloftheCDIreactionproposedbyLengyeletal.15isbasedontwooverallstoichiometricreactions:
ClO-fClO-1
2+I2+2
2
R1)k1XY
ClO-+2+4I-+4HfCl-+2I2+2H2O
R2)k2aZYH+
k2bZPYu+Y
2
(3)
HereX)[ClO2],Y)[I-],Z)[ClO2-],P)[I2],H)[H+],andRirepresenttheratelaws.Therateconstantsandparametersusedinthesimulationsarek1)6000M-1s-1,k2a)460M-2s-1,k2b)2.55×10-3s-1,u)1×10-14M2.ForaCSTRwithaniodide-dependentflowrateofallinputreagents(setupA,Figure1a)thesystemofdifferentialequationsis
J.Phys.Chem.A,Vol.101,No.28,19975151
dX
dt
)-R1+k0(X0-X)dY
dt
)-R1-4R2+k0(Y0-Y)dZ
dt
)R1-R2-k0Z(4)
τdk0
dt)kmax-k1+(Y/I0T)n
-Y
P)
Y02
Herek0isthedynamicflowrate,whichvariesaccordingtotheconcentrationofiodideinthereactor,andX0andY0aretheinputconcentrationsofXandY.ThefeedbackmechanismcontrolsthespeedoftheinputreagentsflowintotheCSTRwhilekeepingtheinputconcentrationsconstant.BecausetheinputconcentrationofH+ishighincomparisonwiththatoftheremainingspecies,wecanneglectprotonconsumptionineq3andassumethat[H+]insidethereactorisfixedatitsinitialconcentration.
InsetupB(Figure1b)theinputconcentrationsofallspeciesvarywiththeratioofthefixedflowratek01tothedynamicalflowratek02,andwemustconsider[H+]tobeadynamicalvariableaswell.TheequationsforsetupBare
dX
dt
)-R1+k01Xs-k0XdY
dt
)-R1-4R2+k02Ys-k0YdZ
dt
)R1-R2-k0ZdH
dt
)-4R2+k01Hs-k0H(5)
τdk02dt)kmax,2-k1+(Y/I02T)n
P)
k02Ys-k0Y
2k0
k0)k01+k02
Heresubscriptsindicatesconcentrationofthestocksolutions,k01isthefixedflowrateofClO2andH2SO4,andk02isthedynamicalflowrateofiodide.
ThemathematicalmodelofsetupCisdescribedbytheequationsforsetupB(eq5)supplementedwiththeequationforthepositivefeedbackofflowratek01:
τdk01
kmax,1dt)-k1+(I/Y)n
01(6)
T
Toanalyzeeqs4,5,and6,weusedtheCONTnumerical
bifurcationandcontinuationpackage.18Periodicsolutionswereobtainedbynumericallyintegratingthesystemofordinarydifferentialequations.Theintegrationusedasemiimplicit
5152J.Phys.Chem.A,Vol.101,No.28,1997Figure4.DynamicalbehaviorinsetupCexperiments:iodideandflowratetimeseries.(a)Maximumflowrateskmax,1)kmax,2)0.00833s-1,targetiodideconcentrationIT)1×10-6M.(b)kmax,1)0.00833s-1,kmax,2)0.01666s-1,IT)1×10-5M(firsthalf),1×10-6(secondhalf).(c)kmax,1)0.00416s-1,kmax,2)0.00833s-1,IT)1×10-6
M.
fourth-orderRunge-Kuttamethodwithautomaticcontrolofthestepsize.
ResultsofSimulations
ThemodeloftheCDIreactioninaCSTRwithoutfeedbackcontrolshowsgoodagreementwithexperiments.8,9,16Previoussimulations14withthemodelofsetupArevealedthepresenceofburstingandchaos.TheseresultsinspiredtheexperimentalinvestigationoftheCDIreactionwithfeedbackcontrol.WeperformedadditionalnumericalsimulationstoinvestigatethedynamicsofsetupsA,B,andC.Modelsofthesesetupsshowburstinginsimulations(Figure5).Shownherearetheiodideconcentration(pI)-log10[I-])andthedynamicallychangingflowrates.Figure5adisplaysburstingwithfivespikesperburstinsetupA.Duringspiking,theflowrateincreasessharply,andthemaximumflowratereachesnearly3timesitsminimumvalue.Theflowrateincreaseisfollowedbyaquiescentperiodduringwhichtheflowratedecreaseisnotassharpasthepreviousincrease.BurstinginsetupBshowsmuchsmallervariationsoftheflowrate(Figure5b).InsetupC(Figure5c)theoppositesignsoftheflowratevariationsresultinanalmostconstanttotalflowrate.ThespikinginsetupsBandChashigherfrequencythaninsetupA,asobservedinourexperi-ments.
Wehavestudiedindetailthedynamicsofallsetupsasafunctionofthetargetconcentrationandthecontrolparameterkmax.Figure6displaysthebifurcationdiagramofdynamicregimesforsetupAasafunctionofthetargetconcentrationITwhenkmaxisfixedat0.05s-1.ThediagramisobtainedfromaPoincare′map,withtheflowratek0evaluatedat[I-])1×10-7M,fordecreasing[I-].Figure6ashowsafulldiagramwithprevailingburstingbehavior.Thenumberofoscillationsperburstdecreaseswithincreasingtargetconcentration.Simplelow-frequencyperiod-oneoscillationsemergejustbeforethe
Dolniketal.
Figure5.Burstingoscillations,simulations.(a)SetupA:targetconcentrationIT)1×10-5M,maximumflowratekmax)0.05s-1.(b)SetupB:IT)1×10-5M,kmax)0.0166s-1,fixedflowratek01)0.0025s-1.(c)SetupC:IT)1×10-5M,kmax,1)0.0083s-1,kmax,2)0.0166s-1.
highvalueofthetargetconcentrationstabilizesthesystemattheHIsteadystate.WefindsimilarscenariosforothervaluesofkmaxandalsoformodelsofsetupsBandC.Inmostsimulationsthechaoticoscillationsarerestrictedtoarelativelynarrowparametricregion,whileregularburstingandsimpleperiodicoscillationsarewidelypresent.Figure6bshowsablowupofthediagramwiththeperioddoublingsequencesleadingtochaos.Thechaoticregioncontainsnarrowperiodicwindows,includingperiod-threeandperiod-fivewindowswiththeirownperioddoublingsequences.Burstingchaoticoscil-lationsemergeatIT≈9×10-7MwithincreasingIT.AsITincreasesfurther,theburstingoscillationsbecomeperiodic.WesummarizetheresultsofsimulationsforsetupsA,B,andCinFigures7and8.Inthesefigures,thesolidlinesconfinetheentireoscillatoryregionsandthedashedlinesrepresentsubcriticalHopfbifurcationlinesobtainedfrombifurcationanalysis.Outsidethesolidlinesthecontrolmechanismalwaysstabilizesthesteadystate.Thesteadystatecanalsobestabilizedforspecificinitialconditionsoutsidethedashedlines.Shadedareasrepresentbursting;greylevelsillustratethenumberofspikesperburst.Theblackregionisthedomainofchaoticoscillations.Figure7showsthatburstingisfoundonlyforintermediatevaluesofkmaxinallsetups.Thedynamicsinthesesetupsisqualitativelythesame;fromaquantitativepointofviewthereisbetteragreementinthelocationoftheburstingregionbetweensetupsBandC(Figure7b,c).Theshapeoftheshadedareassuggeststhatthedynamicsisalmostinde-pendentoftargetconcentrationwhenIT∈(1×10-7,1×10-5)s-1,whilethestrongdependenceonkmaxisapparent.Figure8demonstratesthecloseresemblancebetweensetupsBandC.Thefixedflowratek01insetupBplaysnearlythesameroleasthemaximumflowratekmax,1insetupC.Thevariations
of
OscillatoryChemicalReactionFigure6.One-parameterbifurcationdiagramssimulationsforsetup
A.Pointsinthediagramcorrespondtovaluesofflowratek0onPoincare
′surface:Y)1×10-7M,kmax)0.05s-1
.Smallnumbersofpointscorrespondtoperiodicoscillations,whilemanypointsindicatechaos.(a)Fulldiagramwithprevailingburstingbehavior,(b)detailofdiagramwithchaotic
oscillations.
thesetwoparametersaffectthecontrolleddynamicsoftheCDIreactioninessentiallythesamefashion.Discussion
BurstingandchaosemergeintheCDIreactionasaresultofafeedbackcontrolmechanism,whichregulatestheflowrateand/ortheratioofinputreagents.WeattributethisemergencetothespecificdynamicfeaturesoftheCDIreaction.Asmentionedabove,theCDIreactioninaCSTRwithoutfeedbackcontrolexhibitstwodifferentsteadystates,LIandHI,bistabilityofLIandHI,period-oneoscillations,andcoexistenceofperiod-oneoscillationswithHI(seeFigure9).AdynamicalphasediagramsimilartoFigure9ahasbeenobtainedfromexperi-mentaldataontheCDIreaction.16WesuggestthatthecoexistenceofoscillationswiththeHIsteadystateisresponsibleforburstinginthecontrolledsystem.
WeillustratetheemergenceofburstinginsetupsA,B,andCinFigure9a.Weassumethattheoscillatorystateofthesystem,justbeforethecontrolisimposed,ischaracterizedbypointSandthatthetargetconcentrationITissettobehigherthantheaverageconcentrationduringanoscillatorycyclebutlowerthaninanyHIsteadystate.Figure9ashowshowthestateofthesystemchangesduringonecycleofburstinginsetupA.TheoverallincreaseoftheflowratefromtheinitialpointSisillustratedbythearrowmarkedA+.BecauseITishigherthan[I-],theflowrateincreasesuntilitsvalueisequaltothatatthesaddle-node(limit)point.Duringthisincreaseink0,theoscillationsremainstable.WhenthevalueofthedynamicflowrateexceedstheHopfandlimitpointvalues,theoscillationsbecomeunstableandthesystemmonotonicallyapproachesthe
bbs.99jianzhu.com内容:建筑图纸、PDF/word 流程,表格,案例,最新,施工方案、工程书籍、建筑论文、合同表格、标准规范、CAD图纸等内容。