简介：Themulti-scalestructuresofcomplexflowsinchemicalengineeringhavebeengreatchallengestothedesignandscalingofsuchsystems,andmulti-scalemodelingisthenaturalwayinresponse.Particlemethods(PMs)areidealconstituentsandpowerfultoolsofmulti-scalemodels,owingtotheirphysicalfidelityandcomputationalsimplicity.Especially,pseudo-particlemodeling(PPM,Ge&Li,1996;Ge&Li,2003)ismostsuitableformolecularscaleflowpredictionandexplorationoftheoriginofmulti-scalestructures;macro-scalePPM(MaPPM,Ge&Li,2001)andsimilarmodelsareadvantageousformeso-scalesimulationsofflowswithcomplexanddynamicdiscontinuity,whilethelatticeBoltzmannmodelismorecompetentforhomogeneousmediaincomplexgeometries;andmeso-scalemethodssuchasdissipativeparticledynamicsareuniquetoolsforcomplexfluidsofuncertainpropertiesorflowswithstrongthermalfluctuations.Allthesemethodsarefavorableforseamlessinterconnectionofmodelsfordifferentscales.However,asPMsarenotoriginallydesignedaseithertoolsforcomplexityorconstituentsofmulti-scalemodels,furtherimprovementsareexpected.PPMisproposedformicroscopicsimulationofparticle-fluidsystemsasacombinationofmoleculardynamics(MD)anddirectsimulationMonte-Carlo(DSMC).ThecollisiondynamicsinPPMisidenticaltothatofhard-sphereMD,sothatmass,momentumandenergyareconservedtomachineaccuracy.However,thecollisiondetectionprocedure,whichismosttime-consuminganddifficulttobeparallelizedforhard-sphereMD,hasbeengreatlysimplifiedtoaprocedureidenticaltothatofsoft-sphereMD.Actually,thephysicalmodelbehindsuchatreatmentisessentiallydifferentfromMDandismoresimilartoDSMC,butanintrinsicdifferenceisthatinDSMCthecollisionsfollowdesignedstatisticalrulesthatarereflectionoftherealphysicalprocessesonlyinverylimitedcasessuchasdilutegas.PPMisidealforexploringthemechanismofcomplexflowsa