ApplicationofaaofApplicationaNeuralandAlgorithmGeneticObjectiveMultiNetworktotheoptimisationoffoundryprocesses.foundryofoptimisationthetoNetworkG.Meneghetti*,V.V.*,G.Meneghetti**PoloniC.Pediroda**,*TradingSoftEnginItalySrl,**DipartientodiEnergetica,UniversitàdiItalyTrieste,AbstractAimAimoftheworkwastheanalysisandtheoptimisationofaductileironcastingusingtheFrontiersoftware.software.Fivegeometricalandtechnologicalvariableswerechoseninordertomaximisethreedesigndesignobjectives.ThecalculationswereperformedusingthesoftwareMAGMASOFT,devotedtothethesimulationoffoundryprocessesbasedonfluid-dynamics,thermalandmetallurgicaltheoreticalapproaches.ResultsarecriticallydiscussedbyandarediscussedtheResultscomparingthetraditionalcriticallysolution.approaches.optimised1.IntroductionAAverypromisingfieldforcomputersimulationtechniquesiscertainlygivenbythefoundryindustry.industry.Thepossibilityofreliablyestimatingboththefluid-dynamics,thermalandmicrostructuralevolutionevolutionofcastings(fromthepouringofthemoltenalloyintothemouldtillthecompletesolidification)solidification)andthefinalpropertiesareveryinteresting.Infactifthefinalmicrostructureandthenthenthemechanicalpropertiesofacastingcanbepredictedbynumericalsimulation,thea-priorioptimisationoptimisationoftheprocessparameters(whosenumberisusuallyhigh)canbecarriedoutbyexploringexploringdifferenttechnologicalsolutionswithsignificantimprovementsinthequalityoftheproduct,managingofmanagingandproduct,humanresourcesandeconomicaltime-savings.ThisThisapproachisextremelynewinfoundryandinthisworkanexploratoryprojectaimedattheprocessoptimisationofanironoptimisationprocessductilewillcastingindustrialbepresented.2.Thesimulationoffoundryprocessesand2.ofprocessesThefoundrysimulationfoundamentalequationsFromFromatheoreticalpointofview,afoundryprocesscanbeconsideredasthesequenceofvariousevents[1-4]:-thefillingofacavitybymeansofamoltenalloy,asdescribedbyfluid-dynamicslaws(equation),Navier-Stokes-thesolidificationandcoolin