FLuent换热器的相变模拟计算.pptx

UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.1 New Initiatives at Fluent Inc.Phase Change in Heat ExchangersBrian Bell,Fluent Inc.UGM UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.2 MotivationDemonstrate the use of Fluent to model phase change in heat exchangersProcesses of interestuCondensationuEvaporationuBoiling Illustrate how to model one such process through use of a detailed exampleShell-and-tube condenserProvide motivation for users to begin developing models of their own UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.3 OutlineProblem DescriptionShell-and-tube condenseruPure vapor condensationuNon-condensable gasesModeling ApproachPorous mediumHeat and mass transfer modelingModel ImplementationUser-Defined Functions and User-Defined MemoryResultsSteam condenser with non-condensable gasesCommercial chiller UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.4 Description of ProblemShell-and-tube condenser UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.5 Goals of CFD ModelingCondenser performance characterized by heat and mass transfer rateCFD allows evaluation of factors affecting heat and mass transfer in condenseruTube bundle configurationtTube arrangementtNumber of passestLocation of inlet portstBafflesuPressure dropuVelocity fielduNon-condensablestLocation and configuration of purge systemResults allow identification of potential design UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.6 Film Condensation ProcessDriving potential for condensation is the temperature difference between vapor and cooling waterDriving potential variation caused by Pressure dropRise of cooling water temperatureNon-condensablesTPH2OPairairCondensate layerTube wallCooling UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.7 CFD Modeling TheoryPorous medium approachTube bundle treated as porous mediumEnables computationally efficient modeling of entire condenserComparison with detailed modeling approachuIn 2-D,O(100)-O(1000)control volumes per tube versus more than one tube per control volumeHeat and mass transfer modelsCondensation rate calculationuCondensation rate determined from local flow field and cooling water temperatureuLiquid film flow rate tracked in bundle from top to bottomuCooling water temperature tracked from inlet to UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.8 Porous Medium ApproachRepresentation of tube bundle as porous mediumPorosity is only required parameterPorosity defined as ratio of fluid volume to total volume PduExample:staggered tube bundle with equilateral triangular layoutPorosity,b,expressed as: UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.9 Transport EquationsGeneric transport equation for porous medium approachconvectiondiffusiondistributed resistanceEqn.continuity1x-mom.uy-mom.vspeciesw w Distributed resistance takes form of source terms that model details of the flow that are not resolved by the gridPorosity in convection and diffusion terms not modeled in FluentDistributed resistance terms most significant in tube bundle UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.10 Evaluation of Modeling ApproachAdvantagesComputationally efficientuDoes an alternate,tractable approach exist?Approach demonstrated to give meaningful data by several authorsDisadvantagesLoss of some flow details due to averagingCan be overcome by detailed modeling of small regions of UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.11 Heat Transfer ProcessFilm condensation on horizontal tubeCooling WaterTube WallCondensate FilmLiquid-vapor InterfaceRefrigerantVaporLatent heat released at liquid-vapor interface transferred to cooling UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.12 Heat Transfer ModelHeat transfer is modeled by coupling of thermal resistance network with CFD codeTcwTt,iTt,oTiRcwRtubeRcondCooling WaterCFD code provides interface temperature,Ti Cooling water and tube thermal resistances are generally well-knownFilm heat transfer coefficient is required for R UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.13 Film Heat Transfer CoefficientCritical component of heat transfer modelObtain from experimentOr obtain from literatureSteam condensation on smooth tubesFigure courtesy of Kansas State University,Professor Steve Eckles,and Duane L.R UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.14 Modeling AssumptionsEffect of liquid on flow field is neglectedApproach can also be implemented in Eulerian-Eulerian multiphase frameworkuSatisfactory model for liquid phase representation not currently availableuPublished results of this type of model do not appear to show significant advantageVapor is assumed to be saturatedNo superheatingVapor temperature determined from pressure field calculated by CFD UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.15 Implementation of Model with UDFsUDFs are required for:Source terms required by porous medium approachuCondensation rateuPressure drop in porous regionRepresentation of tube bundleuPorosityuCondensate film flow rate accountinguCooling water temperature calculation with multiple tube UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.16 Cooling water temperature calculation for each segmentuEvery iteration,condensation rate is summed over each segmentuInlet cooling water temperature=outlet temperature from previous segmentuSegment outlet cooling water temperature calculated by energy balance.uLog-mean temperature for each segment calculated based on vapor temperature and cooling water inlet and outlet temperaturesTube Bundle RepresentationBundle consists of N passes and M segmentsEach segment defined as unique cell zoneExample:2 Pass bundleN=2,M= UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.17 Tube Bundle Grid StructureStructured,cartesian grid used in tube bundleEach control volume has unique i,j,k indexi=1j=1k=1i=1j=2k=1i=1j=3k=1i=1j=3k=2i=1j=2k=2i=1j=1k=2i=1j=1k=3i=1j=2k=3i=1j=3k=3Grid structure created with UDFsuGrid generator,solver do NOT utilize structureUsed to track condensate film flow UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.18 Source TermsAlgorithm for source term in continuity equationObtain pressure,velocity and species mass fraction(if necessary)from current solution valuesObtain film Reynolds number and cooling water temperature from User-Defined MemoryCalculate heat flux based on current value of solution variables Translate heat flux into volumetric mass source termUnder-relax source termuSi+1=Si+a(So Si)uRequired for solution stability.Alpha typically 0.01 0.10uValue of source term from previous iteration,So,stored in User-Defined MemorySource term in momentum equations Calculated using empirical correlations with tube bundle porosity and current UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.19 Define_On_Demand FunctionsDefine_On_Demand functions executed once per iterationUpdate condensate film mass flow rateUpdate cooling water temperatureuAssume uniform temperature for each bundle segmentNew values stored in User-defined memoryAutomatic Define_On_Demand execution possibleExample: UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.20 Solution AlgorithmInitialize Solution:Assign porosity,tube bundle orientationUpdate cooling water temperature and liquid condensate mass flow rateCalculate source termsSolve flow equationsYesNoSolution Converged?S UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.21 ExamplesSteam condensation with non-condensable gasesMcAllister Condenserfrom:Bush et al.,1990,Proc.Int.Symp.On Condensers and C UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.22 McAllister Condenser GeometryBoundary conditions and model inputsShell Dimensions 1.02 m X 1.22 m X 0.78 mCooling water flow directionInlet temperature:17.8 CInlet velocity:1.19 m/sTube BundleSingle pass,4 segmentsOuter Diameter:.0254 mInner Diameter:.0242 mPitch:.0349 mPorosity:0.52PurgeMass flow rate:.011 kg/sInletPressure:27670 PaAir mass fraction: UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.23 Condenser Grid15,000 Control VolumesSimple geometry allows structured grid throughout domainGrid profile in x-z UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.24 ResultsCondensation RateInlet mass flow rateCFD:2.124 kg/sExp.:2.032 kg/sError:4.5%Cooling water temperature contoursVolumetric condensation rate UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.25 McAllister Condenser Flow FieldVelocity MagnitudeMax:34.4 m/sMin:0.02 m/sPressureMax:27,663 PaMin:27,530 PaAir Mass FractionMax:.534Min:.00122Condensation Rate*Max:6.1 kg/smMin:0.0 kg/sm*Minimum condensation rate in tube bundle is 0.18 kg/ UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.26 Effect of Air on Condensation RateVolumetric condensation rate contours without airVolumetric condensation rate contours with airComparison of modeling results with and without non-condensable UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.27 Effect of Inundationon Condensation RateTracking condensate film flow rate from upper tubes to lower tubes allows use of inundation correction factorContours of condensate film mass flow rateContours of volumetric condensation UGM 2001Company Confidential Copyright 2001 Fluent Inc.All rights reserved.28 ConclusionsModeling phase change processes in heat exchangers is possible in Fluent 5 through the use of UDFs This approach is well-suited for falling film condensation or evaporation processesHeat and mass transfer models must be provided by the userCFD modeling of McAllister condenserDescription of model development process intended to serve as a reference for users who wish to develop similar modelsAccurate results demonstrate the potential of this approach