L'alliage de première génération (AM1) est un superalliage base-nickel utilisé par la Snecma pour élaborer des aubes de turbine monocristallines de géométrie complexe capables de résister aux conditions extrêmes de température et de pression. Cependant, lors du procédé de solidification, le contrôle du flux de chaleur est difficile pour des raisons de complexité de la géométrie, de la mise en grappe des pièces ou de la cinétique de solidification de l'alliage. Par conséquent, le risque de germination parasite peut avoir lieu dans le liquide surfondu. Pour comprendre l'origine de la germination parasite, il est important d'identifier avec précision la variation des isothermes dans la pièce au cours de la solidification pour localiser les zones de surfusion critiques. Pour ce faire, nous avons prédit par simulation numérique la germination et la croissance des grains au cours de la solidification. Nous avons implémenté dans le module CAFE du code Procast un algorithme de couplage permettant de tenir compte de la transformation liquide→solide dans la résolution du problème thermique. Ce couplage a permis de prédire le phénomène de recalescence qui témoigne de la germination parasite. La conversion enthalpie→température dans ce couplage tient compte de plusieurs chemins de solidification qui dépendent du taux de refroidissement à chaque nœud du maillage éléments-finis. Ces chemins de solidification sont tabulés grâce à un nouveau modèle de microségrégation conçu pour les alliages multicomposés en solidification dendritique colonnaire et équiaxe. Ce modèle est basé sur les équations de conservation de la masse totale et de la masse des solutés moyennées sur un volume représentatif. Dans ce modèle, la diffusion des espèces chimiques est contrôlée dans toutes les phases. Le modèle est aussi couplé avec un logiciel de calcul thermodynamique et un modèle de cinétique de croissance dendritique conçu pour les alliages multicomposés. Les prédictions du modèle retrouvent d'expériences menées sur l'AM1. / AM1 is a nickel-based superalloy that Snecma relies on to elaborate single crystal turbine blades having complex geometry and high resistance to extreme conditions of temperature and pressure. However, controlling heat flux during solidification process is difficult because of many reasons such as the complex geometry, the way of clustering parts and the superalloy solidification kinetics. Consequently, stray grain nucleation can occur in the undercooling liquid. Therefore, it is important to precisely identify critical undercooled zones during solidification. To do this, a new coupling algorithm is integrated in Procast software through its CAFE module. This coupling considers liquid→solid transformation in solving thermal problem. Thus, predicted recalescence during stray grain nucleation can be observed. Enthalpy→temperature conversion is based on tabulated solidification paths depending on cooling rate computed at each Finite Element node. Solidification paths are calculated using a new microsegregation model based on total mass and solute mass conservation equations over a representative volume element. It includes both finite diffusion in phases and growth kinetics for multicomponent alloys. It is also coupled with a thermodynamic software for equilibrium computation. The microsegregation model fits experimental data provided by quenching tests on AM1 superalloy.

A phase transformation from liquid to solid phase takes place when the material solidifies. Limited diffusion during this phase transition causes microsegregation, which is aphenomenon during solidification that leads to the formation of secondary phases andconcentration gradients. This affects the properties of the material and how it is to be treatedin further processing steps. Due to the complexity of the solidification process, a modelingapproach has been used for investigation of microsegregation. The Scheil-Gulliver model assumes negligible diffusion in the solid phase but an exchange ofsolute during back-diffusion can have an important influence on solidification. While the Brody-Flemings model considers this, it is questionable in its assumptions. A new proposedmodel by John Ågren that considers multicomponent diffusion effects have been implementedusing the Matlab-Toolbox for Thermo-Calc. The model is used to perform solidification simulations for the binary Al-2.1Cu and theternary Al-2.1Cu-1Si system (at%) and the results are compared to the Scheil-Gulliver modeland DICTRA simulations. The Ågren model gives excellent results for the binary system athigh cooling rates. It does however show deviations from the DICTRA results at lowercooling rates and for the ternary system. / När ett material stelnar sker en fasomvandling från flytande till fast fas. Begränsad diffusionunder denna fasövergång orsakar mikrosegring, vilket är ett fenomen under stelningsförloppetsom skapar sekundära faser och koncentrationsgradienter. Detta påverkar egenskaperna hosmaterialet and hur den ska behandlas under senare processteg. På grund av komplexiteten avstelningsprocessen har modellering använts för att undersöka mikrosegring. Scheil-Gulliver modellen antar att det inte sker någon diffusion i fast fas men ett utbyte avinlösta ämnen kan ha ett viktigt inflytande på stelningsförloppet. Fastän Brody-Flemingsmodellen tar hänsyn till detta så är den tveksam i sina antaganden. En ny föreslagen modell avJohn Ågren betraktar diffusionseffekter för ett multikomponent system och denna harimplementerats med hjälp av Matlab-Toolbox för Thermo-Calc. Modellen har använts för att simulera stelning för ett binärt Al-2.1Cu och ett ternärt Al-2.1-1Si system (at%) och resultaten har jämförts med Scheil-Gulliver modellen och DICTRAsimuleringar. Ågren modellen ger utmärkta resultat för det binära systemet vid högakylhastigheter. Modellen visar däremot avvikelser från DICTRA vid låga kylhastigheter samtför det ternära systemet.

Cette étude s’inscrit dans le cadre du projet européen IMPRESS, "Intermetallic Materials Processing in Relation to Earth and Space Solidification". Elle porte sur la compréhension des mécanismes fondamentaux qui contrôlent la solidification de l’alliage Ti-46Al-8Nb envisagés pour le développement d’aubes de turbines. La première partie de cette étude caractérise l’influence de l’oxygène sur le comportement à la solidification d’alliages de base TiAl coulés et contenant de 40 a 48 at.% d’aluminium. L’addition d’oxygène augmente la fraction volumique de phase [alpha] formée pendant la solidification péritectique et conduit au changement de la phase primaire de solidification de la phase [bêta] à la phase [alpha] dans les alliages ternaires Ti-44,3Al-1,5O, Ti-47,7Al-0,8O et Ti-47,3Al-1,5O (at.%). Les coefficients de partage pour l’aluminium et l’oxygène kAl[alpha]/l= 0,9 et kO[alpha]/l = 1,29, ont été déterminés pour l’alliage ayant [alpha] comme phase primaire de solidification. Dans un deuxième temps, l’étude du comportement à la solidification de l’alliage Ti-46Al-8Nb, est décrite en portant une attention particulière sur les effets de la contamination et des conditions de croissance sur la formation des microstructures et des microségrégations. Deux comportements différents ont été mis en évidence : dans la phase primaire de solidification [alpha] rencontrée pour des hauts niveaux de contamination en oxygène, une très faible rétrodiffusion est observée. Pour de faibles quantités d’oxygène, la phase primaire de solidification est la phase [bêta]. Dans ce cas la présence d’oxygène induit une augmentation de l’amplitude de ségrégation malgré une forte rétrodiffusion dans la structure cubique centrée / This study was performed in the framework of the IMPRESS, "Intermetallic Materials Processing in Relation to Earth and Space Solidification", and is dedicated to the understanding of the fundamental mechanisms that control solidification in a Ti-46Al-8Nb alloy selected for TiAl-based alloy turbine blade development. The first part of this work is devoted to study the influence of oxygen on the solidification behaviour of cast TiAl-based alloys containing from 40 to 48 at.% of Al. Increasing the oxygen content affects significantly the macrostructure of the as-cast ingots, increases volume fraction of the phase formed during the peritectic solidification and leads to a change of the [bêta] primary solidification phase to the [alpha] phase in the ternary Ti-44.2Al-1.4O, Ti-47.3Al-0.9O and Ti-47.2Al-1.5O (at.%) alloys. When [alpha] is the primary solidification phase, the partition coefficients has been determined as kAl[alpha]/l= 0,9 and kO(alpha]/l = 1,29. In a second part, the solidification behaviour of Ti-46Al-8Nb alloy is investigated, with a particular attention to the contamination and growth conditions effects on microstructures and microsegregation formations. For high contamination levels, the solidification phase is [alpha] where oxygen reduces solute mobility, leading to limited backdiffusion. In low oxygen containing alloys, [bêta] is the primary solidification phase, and oxygen leads to an extent of the segregation amplitude despite high level of back-diffusion in the bcc structure

Cette étude financée par SNECMA et le CNES a pour objectif de mieux comprendre et caractériser le résultats de la solidification de l'inconel 718. Cet objectif est motivé par le besoin grandissant de fiabilité sur les pièces brutes de fonderies utilisées dans l'assemblage des moteurs de fusée. En effet, lors de la solidification, la ségrégation chimique provoque la formation de phase fragile. C'est notamment le cas des phases de Laves. Ainsi, un four de fonderie sous vide et un moule instrumenté ont été mise au point afin de couler au laboratoire des plaques de géométrie variable. Le but étant d'obtenir plusieurs échantillons obtenu sous différente conditions de solidifications. Une méthode d'analyse quantitative au MEB a par la suite été mise au point pour analyser les 4 coulées exploitables. Elle a permis de quantifier la quantité d'intermétalliques en fonction de la vitesse de refroidissement. Mais aussi leur répartition vis à vis des joints de grains d'une part et l'impact de la désorientation entre chaque grain d'autre part. La ségrégation chimique a aussi pu être quantifiée selon ces mêmes paramètres.De travaux numériques ont été entrepris en parallèles avec tout d'abord la simulation macroscopique des coulées avec le logiciel Thercast. Ces simulations ont permis de connaître les conditions de solidification de l'ensemble des échantillons coulés et ont fourni des données d'entré pour les autres aspects numériques. Par exemple, avec un modèle élément fini et automate cellulaire, la formation de la structure granulaire a pu être modélisée. Enfin, la simulation de la microségrégation a permis de valider notre compréhension des phénomènes prédominant lors de la solidification de l'inconel 718. Ces simulations numériques apportent à la fois un éclairage sur les résultats expérimentaux mais aussi démontrent leurs possibilités dans l'optique de mettre en place un modèle global de solidification recouvrant ces trois échelles pour des pièces industrielles / The metallic alloy “Inconel 718” is frequently found in aerospace and aeronautic components. In this study, it is used in the case of the main engine turbopump of the Ariane 5 launcher. This case is obtained by investment casting. Thereby, a great focus is turned on as cast defect like detrimental interdendritic phase. For example, the Laves phase, about 1%, has poor mechanicals characteristics. Several samples was cast under vacuum at MINES ParisTech Cemef. Various superheat and cooling rate are tested. After specific preparation, the microstructure was observed by scanning electron microscopy (SEM). On a surface representative of solidification phenomenon, the NbC phase, the Laves phase and the Delta phase were quantified and chemical segregation measurements were performed. The results give a tendency of the detrimental phases to form at the grain boundaries. This phenomenon is stronger as the cooling speed is lower. Furthermore, the grains boundaries disorientations as an impact. This experimental information is used to model the sample solidification at three different scales. At the macroscopic scale with Thercast software, at an intermediate scale whit the simulation of the grain structure. These simulations were performed whit a model developed at Mines Paristech, CEMEF. The last scale is the simulation of the microsegregation. This model takes into account thermodynamic equilibrium, cooling rate and microstructure geometries. This numerical work improved the comprehension of the solidification phenomenon. A global solidification simulation of industrial piece is the aim of Cnes (French space agency) and Snecma who support this work.

Os processos de fundição que visam à obtenção de produtos fundidos com formatos e propriedades mais próximos dos necessários ao produto final demandam um conhecimento detalhado das relações entre processo e microestrutura e entre microestrutura e propriedades. Não há dúvida de que aspectos microestruturais como o tamanho, a morfologia e o nível de microssegregação da estrutura dendrítica são importantes em definir as suas propriedades. Por outro lado, as variáveis de processo que afetam a transferência de calor no sistema estão entre as mais importantes em definir as características de microestrutura bruta de fundição. O presente trabalho tem como objetivo principal estudar os efeitos de algumas variáveis do processo de fundição na microssegregação de ligas fundidas em moldes metálicos. Para este fim, escolheu-se a liga Cu-8%Sn, que foi obtida na forma de amostras solidificadas unidirecionalmente. Os efeitos de diferentes sistemas de extração de calor, diferentes temperaturas de vazamento e da adição de inoculante foram investigados. Dois tipos de sistema de extração de calor foram empregados, um utilizando uma base de cobre refrigerada a água, outro uma base maciça de cobre revestida superficialmente com uma tinta isolante térmica. As temperaturas de vazamento escolhidas foram \'1110 GRAUS\'C e \'1270 GRAUS\'C e o inoculante adicionado foi a liga Cu-50%Zr. Modelos matemáticos para a transferência de calor e massa no sistema foram desenvolvidos para auxiliarem noentendimento dos fenômenos importantes envolvidos na solidificação das amostras. As temperaturas em diversos pontos do metal foram medidas durante a sua solidificação através de um sistema digital de coleta de dados. Os resultados foram utilizados para o cálculo de diversos parâmetros, como o coeficiente de transferência de calor entre o metal e a base de cobre. O nível de microssegregação das amostras foi calculado a partir de microanálises conduzidas ) através de uma microssonda acoplada a um microscópio eletrônico de varredura. Imagens de microscopia ótica e de varredura auxiliaram nas medidas de espaçamento entre braços de dendrita e na interpretação da morfologia dendrítica. A partir dos resultados, observou-se que é possível modelar matematicamente a transferência de calor no sistema sem a consideração detalhada de efeitos macro e microestruturais. Notou-se, também, que há um decréscimo no nível de microssegregação próximo à superfície das amostras, efeito este que só pode ser modelado matematicamente quando as curvas experimentais de temperatura durante a solidificação forem conhecidas. Verificou-se que a estrutura colunar apresentou um menor nível de microssegregação do que a estrutura equiaxial, o que parece estar relacionado com a maior facilidade de homogeneização da estrutura colunar durante a solidificação / Casting processes which aim at products of near-net-shape and near-net-properties need a detailed konwledge of the relationship between processing variables and microstructure, and between microstructure and properties. It is well known that microstructural features, such as dendrite size, morphology and microsegregation severity have a tremendous effect on casting product properties. In addition to it, the processing variables on which heat transfer in the system depends arise as the most important ones ti define the characteristics of as-cast microstructures. The main objective of this work is to study the effects of some processing variables on the microsegregation of permanent mould cast alloys. Directionally solidified samples of a Cu-8%Sn alloy were chosen to be studied, and the effects of inoculation, different heat extraction devices and different pouring temperatures were investigated. The pouring temperatures were 1110 ºC and 1270 ºC, and a CU-50%Zr alloy was added for inoculation. Mathematical models of heat and mass transfer for the system were developed to help understand the important phenomena underlying the solidification of the samples. During solidification, temperatures were measured at some points within the samples by means of thermocouples connected to a data acquisition system. The measured temperatures were used to calculate some parameters, such as the heat transfer coefficient between the alloy and a copper chill. Severity of microsegregation was assessed through microanalysis carried out with the microprobe of a scanning electron microscope. Dendrite arm spacing measurements and the examination of dendrite morphology were on optical and scanning electron microscope images. Results have shown that proper mathematical models of heat transfer does not need to account for detailed effects of macro and microstructural features. Severity of microsegregation was noticed to decrease along samples, towards their surfaces. This trend cannot be modelled unless temperature measurements during solidification are provided. Columnar dendrites have shown less microsegregation than equiaxed ones, which seems likely to be effects of a higher homogenization degree of the columnar structures .

Os processos de fundição que visam à obtenção de produtos fundidos com formatos e propriedades mais próximos dos necessários ao produto final demandam um conhecimento detalhado das relações entre processo e microestrutura e entre microestrutura e propriedades. Não há dúvida de que aspectos microestruturais como o tamanho, a morfologia e o nível de microssegregação da estrutura dendrítica são importantes em definir as suas propriedades. Por outro lado, as variáveis de processo que afetam a transferência de calor no sistema estão entre as mais importantes em definir as características de microestrutura bruta de fundição. O presente trabalho tem como objetivo principal estudar os efeitos de algumas variáveis do processo de fundição na microssegregação de ligas fundidas em moldes metálicos. Para este fim, escolheu-se a liga Cu-8%Sn, que foi obtida na forma de amostras solidificadas unidirecionalmente. Os efeitos de diferentes sistemas de extração de calor, diferentes temperaturas de vazamento e da adição de inoculante foram investigados. Dois tipos de sistema de extração de calor foram empregados, um utilizando uma base de cobre refrigerada a água, outro uma base maciça de cobre revestida superficialmente com uma tinta isolante térmica. As temperaturas de vazamento escolhidas foram \'1110 GRAUS\'C e \'1270 GRAUS\'C e o inoculante adicionado foi a liga Cu-50%Zr. Modelos matemáticos para a transferência de calor e massa no sistema foram desenvolvidos para auxiliarem noentendimento dos fenômenos importantes envolvidos na solidificação das amostras. As temperaturas em diversos pontos do metal foram medidas durante a sua solidificação através de um sistema digital de coleta de dados. Os resultados foram utilizados para o cálculo de diversos parâmetros, como o coeficiente de transferência de calor entre o metal e a base de cobre. O nível de microssegregação das amostras foi calculado a partir de microanálises conduzidas ) através de uma microssonda acoplada a um microscópio eletrônico de varredura. Imagens de microscopia ótica e de varredura auxiliaram nas medidas de espaçamento entre braços de dendrita e na interpretação da morfologia dendrítica. A partir dos resultados, observou-se que é possível modelar matematicamente a transferência de calor no sistema sem a consideração detalhada de efeitos macro e microestruturais. Notou-se, também, que há um decréscimo no nível de microssegregação próximo à superfície das amostras, efeito este que só pode ser modelado matematicamente quando as curvas experimentais de temperatura durante a solidificação forem conhecidas. Verificou-se que a estrutura colunar apresentou um menor nível de microssegregação do que a estrutura equiaxial, o que parece estar relacionado com a maior facilidade de homogeneização da estrutura colunar durante a solidificação / Casting processes which aim at products of near-net-shape and near-net-properties need a detailed konwledge of the relationship between processing variables and microstructure, and between microstructure and properties. It is well known that microstructural features, such as dendrite size, morphology and microsegregation severity have a tremendous effect on casting product properties. In addition to it, the processing variables on which heat transfer in the system depends arise as the most important ones ti define the characteristics of as-cast microstructures. The main objective of this work is to study the effects of some processing variables on the microsegregation of permanent mould cast alloys. Directionally solidified samples of a Cu-8%Sn alloy were chosen to be studied, and the effects of inoculation, different heat extraction devices and different pouring temperatures were investigated. The pouring temperatures were 1110 ºC and 1270 ºC, and a CU-50%Zr alloy was added for inoculation. Mathematical models of heat and mass transfer for the system were developed to help understand the important phenomena underlying the solidification of the samples. During solidification, temperatures were measured at some points within the samples by means of thermocouples connected to a data acquisition system. The measured temperatures were used to calculate some parameters, such as the heat transfer coefficient between the alloy and a copper chill. Severity of microsegregation was assessed through microanalysis carried out with the microprobe of a scanning electron microscope. Dendrite arm spacing measurements and the examination of dendrite morphology were on optical and scanning electron microscope images. Results have shown that proper mathematical models of heat transfer does not need to account for detailed effects of macro and microstructural features. Severity of microsegregation was noticed to decrease along samples, towards their surfaces. This trend cannot be modelled unless temperature measurements during solidification are provided. Columnar dendrites have shown less microsegregation than equiaxed ones, which seems likely to be effects of a higher homogenization degree of the columnar structures .

Made available in DSpace on 2015-04-01T12:17:41Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 2728849 bytes, checksum: 473329771da433558aad326b78ebbb89 (MD5) Previous issue date: 2013-09-30 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / A model was used to simulate solute distribution during rapid solidification of a system Ag-15%massCu with initial undercooling of 5 K, 10 K and 30 K. Boettinger et al (1987) measured the microsegregation profile in Ag-Cu alloys superficially treated with a focused electron beam (resolidification). In that work, they found an irregular variation for Cu concentration in cross sections of the resolidified microstructures. In some cases, the profile of concentration versus radius presented a W shape. Current models of microsegregation predict a monotonic variation the concentration depending on the cell radius, which are apparently not in accordance with the acquired measurements. The present study used a numeric model to evaluate the solidification of Ag-Cu alloy under same conditions to the Boettinger´s experiment. Numerical results are for a metallic spherical drop, initially liquid, cooled by convection. Solidification starts in the center of the droplet in a given undercooling. The algorithm used here is based on a model in volumetric average of the quantities and the balance equations in a representative elementary volume for treating a solid/liquid mixture. Inside the droplet transport of energy and mass are repurely diffusive processes of average quantities. The algorithm used permitted a global analysis of the phenomena occurring during the solidification process, as well as a local analysis of the quantities involved, which has been shown to be a prerequisite for the prediction of non-monotonic profiles of microsegregation, such as those measured in the experiment mentioned above. / Foi utilizado um modelo para simular a distribuição de soluto durante a solidificação rápida de um sistema Ag-15%Cu com superresfriamentos iniciais de 5 K, 10 K e 30 K. Boettinger et al (1987) mediram o perfil de microssegregação em ligas Ag-Cu tratadas superficialmente com um feixe de laser (resolidificação). Nesse trabalho, os autores constataram uma variação irregular para a concentração de cobre em seções transversais de células nas microestruturas resolidificadas. Em alguns casos um perfil de concentração ao longo do raio em forma de W foi medido. Os modelos de microssegregação atuais, em geral, preveem uma variação monotônica da concentração em função do raio da célula, o que aparentemente não está de acordo com as medições obtidas. No presente trabalho, foi usado um modelo numérico para avaliar a solidificação de uma liga Ag-Cu em condições análogas às do experimento de Boettinger. Os resultados numéricos são para uma gota metálica esférica, inicialmente líquida, resfriada por convecção. A solidificação tem início no centro da gota em um superresfriamento dado. O algoritmo usado foi desenvolvido previamente para um modelo baseado na tomada de médias volumétricas das grandezas e das equações de balanço descritivas em um volume elementar representativo para tratar com uma mistura sólido/líquido. No interior da gota, os transportes de energia e massa são processos puramente difusivos de grandezas médias. O algoritmo usado permite não apenas uma análise global dos fenômenos que ocorrem durante o processo de solidificação, mas também uma análise local das grandezas envolvidas, o que demonstrou ser condição necessária para a previsão dos perfis não monotônicos de microssegregação, tais como os medidos no experimento mencionado acima.

This diploma thesis deals with segregation cycles in austenitic CrNi steels. In first section deals with theory of macrosegregation and microsegregation cycles. The second part is focused on this cycles in experimental casting. Macrosegregation was measured by spectrometric analyser and by MAGMA simulation program. Microsegregation was measured by electron microscope. The aim of this diploma thesis is verify the influence of setting conditions on heterogenity chemical composition in austenitic CrNi steel. For experimental casting with a wall thickness of about 500 mm will be verified effect of the setting time, residence time between solid and liquid phases and others parameters for macrosegregation and microsegregation in selected elements.

Cette thèse est le volet expérimental d’un projet de recherche qui vise à développer un code de calcul qui prédit finement la micro ségrégation des aciers présentant une réaction péritectique. La caractérisation expérimentale de la micro ségrégation a été menée grâce à des essais de solidification dirigée et des essais de trempe en cours de solidification dirigée ainsi que sur des échantillons de lingots industriels. Les alliages sélectionnés font partie des systèmes Fe-C-Ni et Fe-C-Cr. Nous avons d’abord étudié sur des aciers solidifiés l’effet de la composition chimique et l’effet du premier solide ferrite / austénite sur la micro ségrégation. Par la suite nous avons caractérisé la formation de la micro ségrégation avec et sans réaction péritectique. On a montré aussi les effets de la taille de la structure, de la microstructure (dendritique / cellulaire)et de la texture du grain sur la micro ségrégation (colonnaire / équiaxe) sur la micro ségrégation. Enfin, nous avons comparé les résultats expérimentaux avec les résultats de la modélisation. / This PhD work is an experimental part of a global R&D project witech aim to study the microsegregation in peritectic steels. The experimental investigations were made by a Directional Solidification device (DS), Quenched Directional Solidification (QDS) and from parts of industrials ingots. The alloys selected were taken from the Fe-C-Ni and Fe-C-Cr systems. On the solidified steels we investigate the effect of the chemical composition and the first solid (ferrite/austenite) on the microsegregation. The next step was the characterization of the microsegregation during the solidification with and without a peritectic reaction. Afterwards we study the effects of the length of the structure, the microstructure (dendritic/cellular) and the texture of the grains (columnar/equiaxed) on the microsegregation. At the end we have do a comparison bethween the experimental results and those from the numerical simulations

A good understanding of the solidification process of a cast material is essential to understand how the combination of alloy composition and the casting process variables combines into the solid cast component and its performance. The wrong combination may result in poor performance or casting defects. Spheroidal graphite has been well researched in ductile irons (SGI) where it is predominant. Spheroidal graphite is also present in smaller amounts in compacted graphite irons (CGI), but its nature and role in this material is less understood. Recent associations of spheroidal graphite in CGI with shrinkage porosity problems highlights the need for better understanding in this area. The importance of the dendritic austenite structure to the properties and solidification behaviour of cast irons has been highlighted in recent research. However, progress is to a degree limited by lack of practical means to characterize the structure. In the present work, the transition of a cast iron from SGI to CGI though remelting was studied. As the fraction of SG dropped, the tips of the compacted graphite tended to lose contact with the melt at a later stage. After this occurred, solidification appeared to be dominated by spheroidal graphite. Compacted and spheroidal graphite was found to solidify mostly segregated, and the increased recallescence induced by a higher fraction of compacted graphite displayed small influence on the size distribution of spheroidal graphite apart from the total number and fraction. The partitioning of Si, Mn and Cu in SGI and CGI was found to agree well with each other, as well as with theoretical predictions under the assumptions of zero diffusion of the elements in the solid. This shows that the proportions of spheroidal and compacted graphite has small or no influence on the evolution of these elements in the melt during solidification. A method for characterization of the dendritic austenite in quenched cast irons was introduced and evaluated. The method includes a technique for producing a visual contrast between the ledeburite matrix and the dendritic austenite, and a scheme for producing binary images from the resulting micrographs which are suitable for automatic image analysis. Measurements of the volume fraction and surface area per unit volume of the dendritic austenite structure using the introduced method was found to agree reasonably with traditional point counting and line intercept techniques. The difficulty in finding the exact boundary was proposed to be the major source of systematic disagreement. / En god förståelse för stelningsprocessen av ett gjutet material är väsentligt för att förstå hur kombinationen av legeringens kemiska sammansättning och gjutprocessens variabler resulterar i den stelnade komponenten och dess prestanda. Fel kombination kan resultera i sänkt prestanda eller gjutdefekter. Sfärisk grafit är väl studerad i segjärn (SGI) där grafitmorfologin är dominant. Mindre mängder sfärisk grafit är dock närvarande även i kompaktgrafit, där dess karaktär och roll är mindre känd. Samband mellan sfärisk grafit i kompaktgrafitjärn och krympporositet understryker behovet av bättre förståelse inom detta område. Dessutom har betydelsen av den tidiga dendritiska austenitstrukturen för senare delen av stelningen uppmärksammats. Utveckling inom detta område är dessvärre till en grad begränsad av bristen på kunskap om praktiska metoder för att karaktärisera dess struktur. I detta arbete studerades övergången från segjärn till kompaktgrafit genom omsmältning. Vid sänkt fraktion sfärisk grafit visade sig kompaktgrafiten tappa kontakten med smältan vid senare stadie av stelningen. Efter detta inträffade, dominerades stelningen till synes av tillväxt av sfärisk grafit. Kompaktgrafit och sfärisk grafit bildades i huvudsak segregerade. Ökad rekallesens till följd av ökad fraktion kompaktgrafit visade sig ha låg inverkan på storleksfördelningen av eutektisk sfärisk grafit bortsett från dess totala antal och fraktion. Omfördelningen av Si, Mn och Cu mellan stelnad matris och smälta i segjärn och kompaktgrafitjärn fanns stämma bra överens med varandra, samt med teoretiska värden med antagande om försumbar diffusion i stelnad matris. Detta visar att proportionerna av sfärisk och kompaktgrafit har liten eller ingen inverkan på halten av dessa ämnen i smältan under stelningen. En metod för karaktärisering av den dendritiska austenitstrukturen i släckt gjutjärn introducerades och utvärderades. Metoden inkluderar en teknik för att åstadkomma kontrast mellan ledeburitmatrisen och den dendritiska austeniten, och en teknik för att producera binära bilder från resulterande mikroskopbilder som är lämpliga för automatisk bildanalys. Mätningar av volymfraktion och yta per volymenhet av dendritstrukturen genom tillämpning av den introducerade metoden visade rimlig överensstämmelse med traditionella punktfraktion‐ och linjetekniker. Svårigheten att hitta den exakta gränslinjen mellan dendritisk struktur och ledeburit föreslogs vara den huvudsakliga källan till systematisk oöverensstämmelse.