Abstract :
InLagrangianparticledispersionmodeling,theassumptionthatturbulenceisisotropiceverywhereyieldserroneouspredictionsofparticledepositionratesonwalls,eveninsimplegeometries.Inthisinvestigation,thestochasticparticletrackingmodelinFluent6.2ismodifiedtoincludeabettertreatmentofparticle–turbulenceinteractionsclosetowallswhereanisotropiceffectsaresignificant.ThefluidrmsvelocitiesintheboundarylayerarecomputedusingfitsofDNSdataobtainedinchannelflow.Thenewmodelistestedagainstcorrelationsforparticleremovalratesinturbulentpipeflowand90°bends.Comparisonwithexperimentaldataismuchbetterthanwiththedefaultmodel.Themodelisalsoassessedagainstdataofparticleremovalinthehumanmouth–throatgeometrywheretheflowisdecidedlythree-dimensional.Here,theagreementwiththedataisreasonable,especiallyinviewofthefactthattheDNSfitsusedarethoseofchannelflows,forlackofbetteralternatives.TheCFDBestPracticeGuidelinesarefollowedtoalargeextent,inparticularbyusingmultiplegridresolutionsandatleastsecondorderdiscretizationschemes.
