Reduced-orderdiscreteelementmethodmodeling

[محمد بهبهانی]

a b s t r a c t
Reduced-ordermodeling(ROM)techniquesareplayingaverysignifi cantroleintherecentliterature
bridgingthegapbetweencomputationallyexpensivesimul atorsandtheirapplicationinoptimization
and controlofdistributedparametersystems.Inmodelingofs olid-basedprocesses,DiscreteElement
Method(DEM)isaverypopularhigh-fidelitymodelingapproachwhichenablesthetrackingofd iscrete
entitiesastheyareprocessedinsidecomplexgeometriesi nteractingwitheachotherandwithprocess
boundaries.However,thecomputationalcostofDEMmodels limitstheirapplicabilitytosimulation
of specificandlimitednumberofcases.Inthiswork,wedevel optheDE–ROM(DiscreteElement–
Reducer-ordermodel)approachwhichreducesthecomputationalcos t—while maintainingmodel
accuracy—of expensiveDEMmodels.Theproposedmethodologyisbasedon PrincipalComponent
Analysis(PCA),combinedwithsurrogate-modelmappinganddataimputationforthecaseswhere
there ismissinginformation.Theresultsoftheapplicationoft hedevelopedmethodtothepowder
mixing process,demonstratethattheDE–ROMmodelproducesacc urateprofilesofdistributedparticle
propertiesinrealtime.Finally,thedevelopedmodelisus edfortheidentificationoftheoptimaldesign
and operatingconditionsofthemixingprocess

a r t i c l e info
Article history:
Received 3August2012
Received inrevisedform
17 January2013
Accepted 28January2013
Available online5March2013
Keywords:
Model reduction
Particulate processes
Optimization
Mixing
Discrete elementmethod
Principal componentanalysis