工业熔炉的流程控制

SecondYearProcessControlProjectControlofanindustrialfurnacePair:07Name:CHINGUOZHENG00549782CHENCHAO00548719AbstractTheprojectaimistodesignanefficientsystemstrategytocontrolandmaintainoutlettemperatureofcrudeoilandoxygencompositionoffluegasatsetpointvaluethroughsimulationtestsofanindustrialfurnace.Byfirstlyidentifyingasystemmodeltodescribethesystemdynamics,anrelativegainarray(RGA)analysiswascarriedouttodeterminethebestwaytocontrolsystemoutputs.Fromtheresultsitwasfoundthatthemosteffectivewaytocontroltemperatureandoxygencompositionwastoutilisethegasandairvalvesrespectively.Toimproveonthecontrol,feedbackwithPandPIcontrollerswereimplementedseparatelyforcomparison.TheCohen-CoonandZeiglerNicholstuningmethodswereusedasfirstapproximation.Subsequently,manualtuningwasdoneoncontrollergainandintegraltimetofindthecombinationwiththeshortestsettlingtimeandleastoffset.Detuninganddecouplingmethods,aswellasfeedforwardcontrollerswereinvestigatedandintroducedintothesystemtofurtherregulateandimprovethedynamics.Thefinalstrategyinvolvedtwofeedbackcontrolloopsonoxygencompositionandoutlettemperaturewithtwo-waydetuningandone-waydecoupling.Thefeedforwardcontrollerwasalsointroducedinthestrategywhichimprovedtheprocessbehaviourinthefaceofdisturbances.Finally,thefinalsystemstrategywastestedtovalidateperformance.KeywordsFurnace,processcontrol,PIDcontroller,feedback,feedforward,detuning,decouplingIntroductionItisvitalintheindustrytodayforsystematiccontrolofplanttoensuresafetyandsecurity.Withadvancesinengineering,modernplantsarerequiredtocarryouthighlycomplexprocessesthataresubjectedtoriskfactorssuchasdisturbances.Assuch,deviceswhichcanregulatebehaviourofthesystemdynamicsareimplementedinplants.Controlsystemsaimtokeeptheoverallimpactsofdisturbanceonthesystemataminimumbymaintainingcontrolledvariablesataconstant.Theaimofthisprojectistointroduceafullyfunctionalcontrolsystemtothefurnaceshowninfigure1.Thefurnaceisusedtoheatacrudeprocessstreambyburningfueloiland/orgasintheair.Theflowratesoffueloil,gasandaircanbedefinedasthemanipulatedvariables.Disturbancestothesystemarisefrominlettemperatureandcompositionfromtheupstreamandareuncontrollable.Controlvariablesherearetheoutlettemperatureandcompositionofthefluegas.Theprocessstreamshouldbekeptconstanttemperatureforthedownstreamprocess,andtheoxygenconcentrationofthefuelgasshouldbekeptatacertainlevelinordertoensuretheeffectivecombustionofthefuelgasandoil.Thusthemainobjectiveistomaintaintheoutlettemperatureinthevicinityof560Kandtheoxygencompositioninthevicinityof5%.ThecontrolsystemstrategywasbuiltinParagonwhichalsorunssimulationfortestsonthesetpointanddisturbanceschanges.TransferfunctionsmodelswerefirstbuiltafterrunningpreliminarysimulationsonParagon.Feedbackcontrollers(PandPI)usingFigure1:SchematicrepresentationofthefurnaceprocesstheCohen-CoonandZieglerNicholstuningwereinvestigated.Detuninganddecouplingmethodsaswellasafeedforwardstrategywasalsotestedtocreateanoptimalsystem.Beforeprogressingfurther,massandenergybalancesneedtobeperformedonthesystemtounderstandthedynamics.Thefurnacewasassumedtobeadiabaticandthegaswasassumedtobehaveideally.Also,thecombustionoffueloiland/orgaswasassumedtobecompletewithconstantheatcapacities.MassBalance:𝑑𝑀𝑑𝑡=𝐹𝑜𝑖𝑙+𝐹𝑎𝑖𝑟+𝐹𝑔𝑎𝑠−𝐹𝑓𝑙𝑢𝑒+(𝐹𝐶𝑖𝑛−𝐹𝐶𝑜𝑢𝑡)EnergyBalance(heatingstream):𝐹𝐶𝑝𝑑𝑑𝑡𝑣𝑇−𝑇𝑟𝑒𝑓=𝐹𝑜𝑇𝑜−𝑇𝑟𝑒𝑓+𝐹𝑎𝑇𝑎−𝑇𝑟𝑒𝑓+𝐹𝑔𝑇𝑔−𝑇𝑟𝑒𝑓−𝐹𝑓𝑙𝑢𝑒𝑇𝑓−𝑇𝑟𝑒𝑓−𝛥𝐻𝑅EnergyBalance(processstream):𝐹𝐶𝑝𝑑𝑑𝑡𝑣𝑇−𝑇𝑟𝑒𝑓=𝐹𝐶𝑖𝑛𝑇𝐶𝑖𝑛−𝑇𝑟𝑒𝑓−𝐹𝐶𝑖𝑛𝑇𝐶𝑜𝑢𝑡−𝑇𝑟𝑒𝑓+𝛥𝐻𝑅TheoryandmethodologyTransferFunctionModelDevelopmentThetransferfunctionswereapproximatedtobethefirstorderplustimedelay(FOPTD)andtheywerederivedfromaseriesofexperimentsonthefurnaceusingParagonsimulations.Thetransferfunctionsdescribedtheprocessdynamicsbyrelatingthechangesinthecrudeexittemperatureandthefluegasoxygenconcentrationstothechangesintheflowratesoffueloil,gasandair.Inaddition,thedisturbances,namelytheflowrateandtemperatureofthecrudesupply,alsogaveamanualstepchangeinthesimulationinordertofindthecorrespondingtransferfunctionsrelatingthecontrolvariablestothedisturbances.Inthiscontrolproject,thesensorsh(s)wereassumedtobeinstantaneousandwellcalibrated,i.e.h(s)=1.TheFOPTDmodelcanberepresentedinthefollowingform,𝑔(𝑠)=𝐾𝑒−𝑇𝑑𝑠𝜏𝑠+1whereKisthegainoftheprocess,𝜏isthetimeconstantandTdisthetimedelay.IntheParagonsimulation,aftermakingasetchangeofeachmanipulatedvariablesandthedisturbances,thesimulationdatawerecollectfromtheParagonandthetrendofeachcontrolvariableswasplottedagainsttime.TheparametersoftheFOPTDmodelcanbedeterminedgraphicallyasshowinthefigure2.ThetangenttotheΔ𝑝𝑣wasdrawnattheinflectionpointwherethegradientisFigure2:TheFOPTDapproximation[Ref.1]maximum.Thevalueofthegainwasdeterminedusingtheequationbelow.𝐾𝑝=Δ𝑝𝑣ssΔ𝑚ThestepresponseofFOPTDwasplottedforcomparisonwiththesimulationdata.ThiswasdonebyapplyingtheinverseLaplacetransformbyconvertingthefunctionfromthesdomaintothetimedomain.Theequationwasgivenasfollowing.∆𝑝𝑣(𝑡)=𝐾𝑝×1+𝑒−𝑡−𝑇𝑑𝜏×∆𝑚𝑣𝑠𝑠,𝑡≥�