5th International Conference on Advances in Solidification Processes (ICASP-5) 5th International Symposium on Cutting Edge of Computer Simulation of Solidification, Casting and Refining (CSSCR-5)

The aim of this talk is to present the evolution of the important concepts that led to the modelling of solidification, before the landmark paper by Tiller, Jackson, Rutter and Chalmers in 1953 on constitutional supercooling, and how they influenced our present-day models. Before the relevant research on solidification could start it required the development of essential basic concepts. Important theoretical contributions during the 19th century were; on contact angle, capillary pressure, heat and solute diffusion, statistical mechanics, equilibrium thermodynamics, and solidification as a free boundary value problem. The most important experimental developments were; the optical microscope and observations of solidification of transparent organics, followed by the methods of directional solidification, the discovery of X-rays and their diffraction in crystals, the electron microscope, and the microprobe. On the ground of these developments some 25 scientists, mainly during the first half of the 20th century, created the foundations of solidification science, a field that will be presented in six topics. **Nucleation** theories began with condensation of a supersaturated vapour and were later applied to nucleation in metals. **Crystal growth** was analysed and the energetics of attachment evaluated. Non-equilibrium solute distribution (**segregation**) was modelled. **Morphological stability** was described, first qualitatively and later the concept of “diffusional undercooling” developed. Observations of growth of transparent salt **dendrites** led to the conclusion of a parabolic tip with side branches that ripen during solidification. The diffusion equation for the tip was solved in parabolic coordinates. **Eutectic** growth forms were examined and growth rates of transparent organics as a function of undercooling determined. A solution of the diffusion equation for coupled eutectoid growth was given. Finally Zener presented in his highly original 1946 paper a first complete theory for single-phase (plates) and two-phase (eutectoid) structure formation using a new criterion of growth at extremum in combination with approximate solutions of the diffusion problems. Zener’s extremum criterion has been used during the following three decades before it was replaced by marginal stability. - Scientific research is done by people. The most influential authors in the field of solidification science before 1953 were, in order of the year of their contributions; Gibbs, Tammann, Volmer, Papapetrou, Scheil, Zener, Ivantsov, Turnbull. They all influenced in one or other way our modelling capabilities - the theories of Gibbs, Ivantsov and Turnbull being used still today.

Thanks to the recent advance in high-performance computing, the range of application of atomistic simulation is rapidly expanding. We have performed large-scale molecular dynamics (MD) simulations of solidification process including nucleation from undercooled melt and grain growth in the system [1], and discussed heterogeneity in homogeneous nucleation [1] and decrease of the grain boundary mobility during the grain growth [2]. The space scale of the large-scale MD simulation in our approach already reaches that of the phase-field method (PFM), which creates a new possibility for bridging the gap between MD and PFM [3]. For example, MD-generated microstructure is converted into interfacial profile of PFM to perform a direct comparison between MD and PFM at the same spatiotemporal scale [4]. Moreover, the direct mapping of atomistic configuration into interfacial profiles of the phase-field model creates further new concept of the on-the-fly use of information combined with the data-driven technique. In the presentation, state-of-art collection of large-scale MD simulation of solidification will be introduced. Moreover, how MD simulation of solidification links to PFM will be discussed. [1] Y. Shibuta, et al., Nature Comm. 8 (2017) 10. [2] S. Okita, et al., Acta Mater. 153 (2018) 108. [3] Y. Shibuta, et al., Adv. Theor. Simul. 1 (2018) 1800065. [4] E. Miyoshi et al., Comp. Mater. Sci. 152 (2018) 118.

The remelting behavior of the tempering steel 50CrMo4, was investigated with several experimental melts on a lab-scale ESR-plant. The investigated parameters included a variation of the slag compositions and the use of a protective nitrogen atmosphere. Variations of the slag composition comprised slags with different contents of CaF2, CaO and Al2O3 as well as a variation of the SiO2-content. The remelted ingots were forged and analyzed regarding their chemical composition as well as their distribution and composition of the non-metallic inclusions (NMI) by automated SEM-EDX method. The chemical composition of the slag after remelting was analyzed as well. The results clearly show a relationship mainly of Si and Al in the steel with the process parameters. NMI changed in their total amount, type and size distribution. The protective atmosphere reduced the Si-losses during remelting. The majority of the NMI were of the Al2O3- & MnS-type. In general, remelting lead to an almost complete removal of sulfides, a reduction of oxisulfides and a shift towards more oxides. The total amount of NMI was most strongly reduced by the high CaF2-containing remelting slag.

This paper summarizes the development of a computational fluid dynamics (CFD) model based on the commercial software package ANSYS-CFX, which has been used to examine the kinetics of dissolution of a single stationary Ti-N particle surrounded by moving liquid titanium. The model accounts for diffusional mass transport in the solid particle, the formation of a two-phase solid/liquid boundary layer and both advective and diffusional transport to the bulk liquid. The results have been used to estimate an effective mass transfer coefficient that may be applied in models that only consider solid-state mass transport of nitrogen. The results show that a correction is needed to the conventional Ranz-Marshall correlation to more accurately calculate the mass transfer coefficient during the Ti-N particle dissolution process.

Secondary metallurgical processes such as Vacuum arc remelting (VAR) are extensively used for production of metal ingots with superior quality. The complex VAR process involves a wide range of physical phenomena including heat transfer with phase change (solidification/melting), Plasma arc, and the interplay between flow and electromagnetic field known as magnetohydrodynamics (MHD). The spatial and temporal distribution of the arc has a significant influence on the energy and electric current supplied to the ingot top surface. Thus, the evolution of the ingot is to a great extent controlled by the behavior of the arc. During the process, a notable fraction of current (~ 50%) in the vacuum region is transferred directly between the electrode and mold without flowing through the ingot known as “side-arcing”. The latter influences the electric current path in the whole system that in turn determines the hydrodynamics in the molten pool. Of note, the flow in the molten pool is driven by buoyancy, and self-induced electromagnetic force. Therefore, the amount of current which flows through the molten pool impacts the strength of the electromagnetic force that in turn determines the quality of the final ingot. Most often, the quality of the ingot is characterized by the shape of melt pool, i.e., the depth and thickness of mushy zone. The desired outcome of the VAR is a shallow melt pool that promotes unidirectional (upwards) solidification of the ingot and consequently formation of segregation-minimal alloy. The degree of macrosegregation is dependent on the slope of the solidus/ liquidus isotherms that in turn is related to the mushy zone depth. In fact, deep mushy zone results in a severe macrosegregation and subsequently inadequate mechanical properties, cleanliness, and yield. Therefore, the melt pool profile is often used as an indicator of the internal quality of the ingot. Generally, experimental analysis and measurements are quite difficult during the operation at the elevated temperature (~ 2000 K) of the process. Therefore, simulation tools can be applied to obtain a deeper knowledge of the VAR process. In this paper, a numerical study is performed to investigate the effect of “side arcs” on the pool profile of the VAR ingot. A series of simulations are performed considering different amount of “side arcs” (e.g. 20%, 50%, and 70%). Details of the analysis of the electric current, flow, and thermal/solidification fields are presented.

This paper aims to obtain fundamental information on chemical homogenisation process of liquid steel with alloy additions in the bloom tundish. For alloy feeding to liquid steel pulse step alloying method was applied. Author checked the effect of hydrodynamic conditions occurs in the internal working space of tundish on the process of alloy mixing with liquid steel. Within the work basic and proposed tundish equipments were considered. Numerical modelling technique was employed to demonstrate the process of alloy addition mixing during continuous casting process. Ansys-Fluent computer program was used for numerical simulation. For particular continuous casting strands time mixing was calculated.

Band segregation has been found in the H13 die steel produced by the electroslag remelting (ESR) technology. Chemical and metallographic studies have been carried out on a one ton ESR ingot of H13 die steel, so as to understand the formation mechanism of the band segregation. The results indicate that the T.O content and S content decreased because of cleanliness improvement of ESR process. Transverse macrosegregation of S content decreased after ESR. The overall removal ratio of the inclusion is around 65.8%. The original complex inclusions would be modified to the CaO∙Al2O3 inclusions. Al2O3 and MnS inclusions can be found after ESR. Both of Al2O3 and MnS inclusions were found to be the core of primary carbides. The net like structure in ESR ingot and banded structure in the forged steel were observed. V, Mo, Cr and S are rich in the segregation areas of ESR ingot. Besides, black and white segregation bands can be observed on the forged steel samples after etching. Uneven distribution of carbides rich in V, Mo and Cr was observed in banded structure.

Recent Advances in Understanding the Early Stages of Solidification Z Fan BCAST, Brunel University London, Uxbridge, Middlesex, UB8 3PH, UK (zhongyun.fan@brunel.ac.uk) Solidification of single phase alloys occurs in a number of distinctive stages during the cooling process. For isothermal solidification these may include prenucleation (atomic ordering in the liquid adjacent to the liquid/substrate interface), heterogeneous nucleation, grain initiation, spherical growth, morphological instability, dendritic growth and grain impingement. Here, we define the early stages of solidification as the process between prenucleation and the point of morphological instability. Since the majority of historic research has concentrated on dendritic growth, our current understanding on the early stages of solidification has been very limited despite it contributing predominantly to the formation of the solidified microstructure. In recent years, we have been devoting our research effort to understanding the early stages of solidification and have made good progress. In this contribution we present an overview of the recent advances in understanding the early stages solidification, focusing on heterogeneous nucleation, grain initiation and their effect on grain refinement. We will demonstrate both theoretically and experimentally the following key conclusions: • The current grain refining approach (take Al-5Ti-1B grain refiner as an example) has reached a saturation point; it is difficult to achieve any further grain refinement; • More significant grain refinement can be achieved by impeding heterogeneous nucleation using less potent nucleant particles (larger nucleation undercooling), which is in contrast to the conventional approach to grain refinement by enhancing heterogeneous nucleation on the most potent nucleant particles. • Al- and Mg-alloys do not need grain refiners since they contain sufficient native solid particles to achieve grain refinement once they are made available by appropriate melt treatment, for example, by intensive melt shearing.

Grain refinement of aluminium alloys has always been desirable in foundries. It is a common practice to add grain refiners into the alloy melt to promote heterogeneous nucleation and hence achieve effective grain refinement. However, the efficiency of commercial grain refiners is extremely low as less than approximately 1% of the added particles initiate aluminium grains during solidification while the rest remain inactive and stay in the final castings. These particles become inclusions when the castings are recycled and the effect accumulates with repeated recycling. Recently, it has been demonstrated that the native oxide particles can be utilised for effective grain refinement in aluminium alloys. This holds great promise to achieve grain refinement without the addition of grain refiners and provides a solution to produce closed-loop recyclable aluminium alloys. Nevertheless, further investigation is still needed for commercial aluminium alloys, which normally contain certain levels of alloying and impurity elements. One of the major reasons is that it is unclear how these solute elements affect heterogeneous nucleation, particularly the potency of nucleation substrates. Most recently, it has been found that the adsorption of solute elements onto the surface of nucleant particles plays a significant role in affecting the effectiveness of these particles acting as nucleation sites for aluminium grains. Therefore, it is intriguing to study the potential elemental adsorption on the surface of native oxide particles in aluminium alloys. In the present work, the potential element adsorption onto the surface of native oxide particles has been investigated by analytical transmission electron microscopy on samples prepared using a melt filtration technique. Using selected area diffraction and high-resolution imaging, the crystal structure and the habit plane of the native oxide particles has been firstly determined. Furthermore, the potential elemental adsorption at the interface between oxide particles and Al grain matrix has been studied by using super scanning transmission electron microscopy combined with electron energy loss spectroscopy. Based on the experimental results, the nature of the elemental adsorption in terms of composition and structure has been analysed using a simple lattice matching model. Finally, the effect of elemental adsorption on the nucleation potency of oxide particles and hence on their role in solidification will be discussed.

A homogeneous microstructure of as-cast magnesium alloys is desired to improve the formability during their subsequent thermomechanical processing. Owing to its similar crystal structure to Mg, the part of Zr formed by peritectic reaction during solidification was considered to be the most effective nucleants for alpha-Mg. However, regarding the Al-containing magnesium alloys, up to now no suitable and effective external nucleants were found for them. Recently, it was demonstrated that the additions of SiC worked in refining both the Mg-Al and Mg-Zn alloys. The SiC particles acted as nucleants in magnesium alloys are attracting more attentions. The present work investigated and compared the effects of external SiC particle additions on the grain refinements of Mg-Al and Al-free Mg-Zn (Mn) alloys. Their microstructures were characterized using XRD, SEM and TEM. It was found that the additions of SiC particles could refine the grains of both Mg-Al and Mg-Zn alloys. The SiC particles cannot act as a direct heterogeneous nucleant for the nucleation of alpha-Mg in both Mg-Al and Mg-Zn (Mn) alloys. The responsible mechanisms for their grain refinements are different. Regarding for Mg-Al alloys, the grain refinement caused by the addition of SiC particles is attributed to the formation of a ternary intermetallics Al2MgC2, which has a very similar crystal structure to that of Mg. As for Mg-Zn (Mn) alloys, the grain refinement is attributed to the formation of a (Mn, Si)-enriched intermetallics by the interactions between SiC and impurity Mn in alloys.

Effective grain refinement strongly depends on the interplay between nucleation potency of nucleant particles (measured by nucleation undercooling), grain initiation behaviour (affected by size, size distribution, spatial distribution and number density of nucleant particles) and solidification conditions (specified by alloy concentrations and cooling rate). Recently, we have made good progress in understanding grain initiation behaviour, and developed the concepts of progressive grain initiation and explosive grain initiation. We found that the most effective grain refinement can be achieved by impeding nucleation using nucleant particles of least potency, which is in direct contrast to the traditional approach to grain refinement, in which the most potent nucleant particles are used to enhance heterogeneous nucleation. In this contribution we present our latest understanding of effective grain refinement though analytical and numerical modelling of solidification processes. We will show that grain initiation behaviour can be best represented by a grain initiation map, which can be used as a practical guide for effective grain refinement.

The fragmentation of dendrites immediately following the recalescence phase of growth during the solidification of undercooled melts has been invoked to explain various rapid solidification microstructures. Despite this, little direct evidence of such a fragmentation process usually survives in the as-solidified material. We report on the rapid solidification of the single phase, congruently melting intermetallic -Ni3Ge. During equilibrium solidification this material solidifies to the chemically ordered L12 crystal structure. Conversely, during rapid solidification, disorder trapping results in solidification to a random fcc solid solution, thereby providing a means to distinguish the rapidly solidified structures. We present results which show a range of microstructures in which the dendrite fragmentation process has been captured in progress. Results from EBSD Euler mapping reveal that dendrite fragmentation is a potential, but not particularly efficient, route to grain refinement.

Experiments on the International Space Station (ISS) lead to the idea of a special growth mechanism in Al-rich Al-Ni melts. Previous analyses of dendrite growth velocity as a function of undercooling in these alloys showed anomalous behaviour, particularly a decreasing velocity for increasing undercoolings [1]. The experiments on the ISS reproduced the results obtained on earth but with an unprecedented clarity. During rapid solidification, the visible growth front does not consist of dendrite tips that grow along the surface of the sample, but is formed by a sequence of nucleation events propagating perpendicularly to the actual dendrite growth direction. The front progression is thus a superposition of growth of existing surface nuclei and new nucleation events in the vicinity of the existing nuclei. Microstructure analysis strongly supports the theory of this growth mechanism. It shows dendrites growing towards the sample centre and provides clear evidence of nucleation at the sample surface. However, the question concerning the anomalous behaviour (slower apparent growth with higher undercoolings) is not yet fully answered by the nucleation mechanism. For this, a time resolved analysis of nucleation and surface growth is presented to investigate the relation between the nucleation rate, front propagation and undercooling. Acknowledgements ---------------- This work was financially supported by the DFG project HE 1601/26-3 and ESA-MAP project AO-2009-0829. The authors thank the team from the Microgravity User Support Center at DLR, Cologne, and ESA for support with the experiments in the ISS-EML. Reference --------- [1] R. Lengsdorf, D. Holland-Moritz, D. M. Herlach, Scripta Materialia 62 (2010) 365-367

Under a NASA (National Aeronautics and Space Agency)-ESA (European Space Agency) collaborative research project, MICAST (Microstructure formation in casting of technical alloys under a diffusive and magnetically controlled convection conditions), three Al-7wt% Si samples (MICAST-6, MICAST-7 and MICAST2-12) were directionally solidified at growth speeds varying from 10 to 50 μm s-1 aboard the International Space Station to determine the effect of mitigating convection on the primary dendrite array. The observed primary dendrite trunk diameters during steady-state growth of MICAST samples show a good agreement with predictions from a coarsening based model developed by the authors. The trunk diameters in the terrestrial-grown equivalent samples were larger than those predicted from the model. This suggest that thermosolutal convection increases the trunk diameter of primary dendrites, perhaps by increasing their tip radius due to compositional changes.

Feathery grains have attracted considerable research interest since they were first observed in semi-continuous casting of industrial aluminum alloys about 70 years ago. Different from the regular columnar or equiaxed morphologies observed in alloy castings, this special crystal structure exhibits the alternant twinned and untwinned lamellae morphology. Previous studies focus on the analysis and characterization of the twinned dendrite growth appearing in direct-chill (DC) or similar semi-continuous casting processes. And the twinned dendrite growth during these solidification processes is believed to occur when the following solidification conditions are met: (i) a relatively high thermal gradient (typically 100 K/cm); (ii) a large growth speed (at least 1 mm/s); (iii) the presence of convection in the melt. In this work, we first successfully produced twinned dendrites in Al-4.5 wt.% Cu alloys during Bridgman solidification. Based on the twin planes already formed at 3000μm/s, by introducing an abrupt decrease in growth speed, the twinned growth could be produced till the end of solidification after abruptly decreasing the growth speed to 100, 50, 20, 10, 5, 3, 1 and 0.3 μm/s, respectively. The twinned patterns in aluminum alloys are richer than previously expected during a wide range of solidification processes. The corresponding pattern selection map was further revealed. Besides, an interesting orientation rotation phenomenon was verified by EBSD on either side of the coherent twin boundary during twinned cell growth. In high-resolution transmission electron microscopy (HRTEM) views, successive stacking faults (SFs) occurred nearby the twin boundaries (TBs) at 1μm/s, while the distortion around the TBs at 3000μm/s was quite slight. Importantly, distinct SF structures directly connecting two adjacent TBs, with a spacing of about 221nm, provides a direct experimental evidence for the close correlation between the SFs and the twinned dendrite growth in aluminum alloys.

Traditional aluminium casting alloys are based on binary eutectic systems with additions of minor alloying elements for improved castability and mechanical properties. The increasing demand for high strength aluminium casting alloys has led to several studies of ternary eutectic aluminium alloy systems, such as Al-Si-X (X = Sn, Ge, Cu) for the formation of bulk nano/ultrafine composite microstructures. Recent research at BCAST has led to the development of a hypoeutectic quaternary Al-xCu-2.2Si-1.1Mg (x = 5, 6.6, 10.6) (wt%) alloy for high pressure die casting (HPDC), with a good combination of strength (e.g. yield strength between 219 and 276 MPa) and ductility (e.g. elongation between 3 and 7%). The resultant as-cast microstructure of these multicomponent HPDC aluminium alloys, characterised by optical microscopy, scanning electron microscopy (SEM), transmission electron microscopy (TEM) and X-ray diffraction (XRD), consisted of a hierarchical composite of ductile, micron-sized, dendritic -Al phase and a hard nanocrystalline eutectic mixture. This contribution presents a detailed study of the solidification behaviour of Al-Cu-Si-Mg multicomponent hypoeutectic HPDC alloys, with an emphasis on the size/morphology/volume fraction/chemical composition of each phase as a function of Cu content.

The Arvedi ESP process and a variety of produced materials have been continuously developed since the opening of the Arvedi ESP plant in Cremona in 2009 to meet market demands for more sophisticated steel grades. The development of grades for more advanced applications such as advanced high strength steels and multiphase grades is of interest. Dual phase grades such as DP600 are already produced through an ESP line on an industrial scale; additional multi-phase grades such as TRIP are under development. High-strength steels for the automotive industry have especially high demands on material properties. In addition to the mechanical material properties, an excellent surface quality is required. The fundamental basis for such material properties on rolled coils needs to be provided from continuous casting. This paper deals with the classification of different – either Si- or Al-based – alloying concepts for TRIP steels with respect to their prospective behaviour in a thin slab caster.

Belt casting as a technology has been suggested as a potential casting method for steels for decades. However it has been hindered by technological capabilities and investment. The potential draw of > 3GJ/tonne reduction in energy this process has on conventional continuous casting has meant that this technology has always shown great potential. With recent improvements and the active mill in Salzgitter then the number mechanical limitations of this process are quickly being reduced, leaving metallurgical implications or potential areas of exploitation. The work presented here focuses on some of the implication belt casting has on the microstructure of the cast product. This includes, cast grain size, segregation and precipitation. Whilst also highlighting the further benefits of this techniques with regards to castability of advanced high strength steel, low density steels etc, as well as the opportunities it offers. Through a range of lab based experimentation, including high temperature confocal microscopy, small scale lab casts and full casting XRF and EBSD, a large amount of information has been gained about the implication and possible exploitations of belt casting.

Transverse cracking of continuously cast products has been encountered at almost every caster operation. Enormous efforts have been carried out in the past aiming at identifying the cause and reducing the problem, especially on steel grades with peritectic chemistries. So far, however, there is still no cost-effective solution with good trade-off for internal quality and productivity. In this study, a new cracking formation mechanism is proposed based on observations of equally spaced crack pattern and the undulation (shell thinning pattern) observed on the inside and/or outside (surface depression) of the breakout shell with similar spacing. This wavy solidification shell forms at initial solidification stage and induces the local shell thinning and reheating which in turn causes the local “Blown grains” inside the solidification shell. As observed on slab surfaces, these blown grains are closely related to transverse cracking problem.

Previous studies have shown that it is extremely problematic to add synthetic inclusions successfully into liquid steel for clean steel experiments. Small micro-particles encapsulated in a metallic parcel are difficult to pass through the melt-gas interface and inclusions tend to agglomerate then float up to the liquid surface. In this study, powder metallurgy is used to distribute cerium oxide particles of a known size from 1µm to 14µm within a small-scale steel ingot. Carbonyl iron powder has been mixed with 0.1 wt.% cerium oxide (CeO2) then sintered in an electrical furnace to produce sintered ingots of 500 grams. A series of induction furnace melting trials using a total of 400 grams of electrolytic iron and cerium oxide sintered ingot have been undertaken. The work has included extensive FE-SEM analysis using the INCA Feature® to characterise synthetic cerium oxide inclusions from both the sintered ingots and trial ingots. The INCA Feature® results showed that the size distribution and number of cerium oxide inclusions agreed well between the sintered ingot and trial ingot. The synthetic cerium oxide inclusions are homogeneously dispersed through the bottom, centre and top of the trial ingot with approximately 16 number counts/mm2. Analysis suggests that more than 95% of the cerium oxide from the sintered ingot has been distributed throughout the trial ingot. The work has also been upscaled from 400 grams up to 1.5 kg and has been successful. The method developed will be useful for further studies into steel cleanness of high value alloyed steels.

Maintaining competitiveness in steel manufacturing requires improving process efficiency and production volume whilst enhancing product quality and performance. This is particularly challenging for producing value-added advanced steel grades such as advanced high strength steels and electrical steels. These grades due to higher weight percentage of alloying elements cause difficulties in various stages of upstream and downstream processing, and this includes continuous casting, wherein high solute levels are critical towards macro-segregation. Interface growth direction in systems with more than one component is dictated by the solute profile ahead of the moving solidification front. Understanding the profile of growth direction with casting process parameters during the progress of casting will provide an important perspective towards reducing the macro-segregation in the cast product. In the present study, two steel slab samples from conventional slab caster under the influence of electromagnetic brake (EMBR) at Tata Steel in IJmuiden (The Netherlands) have been investigated for dendrite deflection measurements. The samples showed a transition zone where a change in the deflection behavior occurs. Also, the magnitude of the deflection angle decreases away from the slab surface. Correlating these experimental data with modeled fluid flow profile will help in improving the understanding of the dynamic nature of the solute advancement so that the casting parameters can be optimized to improve product quality.

The ongoing RFCS project SUPPORT-CAST is aimed at finding optimal casting parameters and cooling rates to improve product quality based on high-resolution visualization topography scanning in combination with advanced numerical modelling. Surface scanning is performed by a blue laser sensor adapted to withstand temperatures up to 1200°C and capable of resolutions up to 30 microns. This enables detection of defects such as depressions and oscillation marks as well as small corner cracks. Furthermore, the full slabs sections can be reconstructed from the scans to create a virtual representation of the products. At the same time, advanced numerical models have been developed to account for the flow generated by the nozzle/SEN as well as including extensive details of the mould geometry; steel grade, slag and casting conditions (e.g. mould oscillation, superheat, cooling water, etc.). The model developed comprises a ¼ section of an industrially produced slab with 2000 x 150 mm dimensions. A reference case including flow, heat transfer and solidification is undergoing validation in order to identify casting combinations that lead to formation of defects during shell growth in the mould. The final aim is to develop online-monitoring systems and numerical models able to identify defects as well as support decision making to formulate guidelines that improve the surface quality of cast products.

Frank in 1952 [1] already postulated that Icosahedral Short-Ranger Order (ISRO) in the liquid might explain the large undercoolings measured by Turnbull in fcc of hcp metals. ISRO was later confirmed by several observations such as the formation of quasicrystals or the atomic structure of their approximant intermetallic phases, neutron small angle scattering and atomistic simulations. In contrast to the explanation of Frank, Kurtuldu et al [2,3] have shown recently that ISRO can act as a precursor to what has been called "iQC-mediated nucleation" of fcc alloys. For two alloys based on Al and Au, it has been observed that minute additions of Cr and Ir, respectively, drastically change the final grain size and induce an abnormal fraction of twinned grain boundaries. The occurrence and geometrical configuration of multi-twinned nearest-neighbour grains can only be explained by heteroepitaxy relationships between 5-fold symmetry and fcc phases, where 111 and 112 directions of the fcc phase correspond to 3-fold and 2-fold symmetry axes of the icosahedron or interlocked icosahedron, respectively. The formation of these twins at the onset of solidification can explain the origin of twinned dendrites (or feathery grains), a morphology which has been observed in Al alloys since the mid-fourteens. But ISRO appears to also affect the dendrite growth directions via an attachment kinetics contribution, as first observed by Kurtuldu in Al-20wt%Zn with small additions of Cr [4] and very recently by Zollinger et al in Au-21.5wt%Cu-4.5wt%Ag with Ir traces [5]. This keynote will try to summarize these recent developments and outline future research directions that can confirm the influence of ISRO on nucleation and growth of metallic alloys.

The present contribution reviews the recent progress related to the influence of Icosahedral Short-Range Order (ISRO) and icosahedral Quasicrystals (i-QC) formation on the solidification of fcc alloys through minor solute element additions. From intensive crystallographic analysis of multi-twinned regions in as-cast Al-based and Au-based fcc alloys, Kurtuldu et al. have shown recently that a so-called “iQC-mediated” nucleation mechanism occurs when a few hundred ppm of Cr and Ir, respectively, are added to the melt [1] [2]. Similarly, it appears that the growth directions of dendrites in Al-Zn:Cr is also influenced by ISRO in the liquid, thus showing an attachment kinetics effect [3]. In a recent contribution, we have shown that iQC-mediated nucleation also occurs in pink gold alloys with Ir-additions, but two additional phenomena at high solidification speed [4]: (i) a spinodal-type decomposition of the liquid, leading to the formation of twinned Cu precipitates in addition to multi-twinned Au-rich grains; (ii) a change of the microstructure of the Au-rich grains, from 100 dendrites to 111 textured cells in the columnar zone. > [1] G. Kurtuldu, P. Jarry, and M. Rappaz, Influence of Cr on the > nucleation of primary Al and formation of twinned dendrites in > Al-Zn-Cr alloys: Can icosahedral solid clusters play a role? Acta > Mater. 61:7098–7108 (2013). > > [2] G. Kurtuldu, A. Sicco, and M. Rappaz, Icosahedral > quasicrystal-enhanced nucleation of the fcc phase in liquid gold > alloys. Acta Mater. 70:240–248 (2014). > > [3] G. Kurtuldu, Influence of Trace Elements on the Nucleation and > Solidification orphologies of fcc Alloys and Relationship with > Icosahedral Quasicrystal Formation. PhD thesis #6057, EPFL (2014). > > [4] J. Zollinger, B. Rouat, S. K. Pillai and M. Rappaz, Influence of > Ir additions and Icosahedral Short Ranger Order (ISRO) on nucleation > and growth kinetics in Au-20.5wt%Cu-4.5wt%Ag alloy. Acta Mater. > Submitted.

Inoculation with exogenous particles has been widely used in industry to grain refine various alloys during solidification. Among a series of Al-Ti-B based grain refiners developed for Al-alloys, the Al-5Ti-1B master alloy, which contains potent TiB2 particles for enhancing heterogeneous nucleation of α-Al grains, has been most widely used for decades. However, it has been reported that, in the presence of certain alloying or impurity elements in Al-alloys, the effectiveness of Al-Ti-B master alloys for grain refinement can be dramatically reduced, resulting in a coarse and columnar grain structure in some cases. This adverse effect on grain refinement is often referred to as “poisoning” in the literature. When its content exceeds 2-3 wt.%, Si is one such element that poisons the Al-Ti-B based grain refiner. So far, the exact mechanisms for Si poisoning have not been fully understood, although significant research effort on the subject has been made. In this work, state-of-the-art electron microscopy, including aberration (Cs)-corrected high resolution STEM, has been carried out focusing on the Al/TiB2 interface at the atomic scale, in order to reveal any possible modification of TiB2 substrates caused by interaction between Si and TiB2 at the interface. In this contribution, we present the mechanism underlying Si poisoning in the context of Si segregation at the Al/TiB2 interface and its consequence on the potency of TiB2 particles as substrates for heterogeneous nucleation of Al.

Nucleation of crystalline materials, the starting point for crystal formation from melts, has long been an important issue in condensed matter physics and materials science. Nucleation plays a key role in determining the microstructures and mechanical properties of crystalline materials; therefore, controlling nucleation is a very effective way for regulating the macrostructures of materials for specific applications. As nucleation occurs at atomic length scales and diffusional time scale, it is still hard to investigate such kind of multiple scale issue with regular methods. Classical nucleation theory (CNT) can descript nucleation issues well, however, it does not take the lattice structure transition during nucleation into consideration, which makes CNT can’t descript crystal nucleation quantitatively. Further, more and more investigations reported that nucleation often pass through some intermediate states, and the properties of the intermediate phases has significant influence on nucleation pathways. This kinds of nucleation often calls two-step nucleation (TS), however, the atomistic pathways of TS still unknown. A deeper understanding of nucleation process requires experimental or numerical work to provide atomistic visualizations and images of nucleation. In this dissertation, both the Phase field crystal model and Molecular dynamic method are employed to study the nucleation pathways and pathway selection mechanisms on atomistic scales for both single phase and binary alloy solidification.

Prenucleation refers to the phenomenon of atomic ordering in the liquid adjacent to the liquid/substrate interface at temperatures above the nucleation temperature, which becomes the precursor for the subsequent heterogeneous nucleation process through structural templating. Understanding of prenucleation is therefore of both scientific and technological importance. Based on recent investigations of the effects of lattice misfit and chemical interactions on prenucleation, in this work we study atomic ordering adjacent to the surfaces of native oxides (alumina, magnesia) in Al and Mg melts, using a parameter-free ab initio molecular dynamics simulation (MD) technique based on the density-functional theory (DFT). Our modelling reveals that in liquid, the oxide surfaces exhibit a rich variety of 2D ordered structures containing various structural defects, depending on the structural nature of the substrate and chemical interactions between the substrate and the liquid metal. Formation of atomically rough substrate surfaces was observed on the oxide particles in both Mg and Al melts during the ab initio MD simulations. This atomic roughness of the substrate surface weakens prenucleation and reduces significantly the potency of oxide particles as substrates for heterogeneous nucleation.

It has been recently realized that atomic ordering in the liquid adjacent to the substrate (i.e., prenucleation) has a significant implication on the subsequent heterogeneous nucleation process. In this paper, we report an atomistic mechanism of heterogeneous nucleation through structural templating. Using molecular dynamics (MD) simulation, we investigated the process of heterogeneous nucleation in a model system consisting of liquid Al and fcc substrates with <111> surface orientation and varied lattice misfit. We found that heterogeneous nucleation occurs at a critical undercooling, proceeds layer-by-layer through a structural templating mechanism and finishes by creating a template (i.e., a crystal plane), which permits further growth of the new solid phase. In most cases, Shockley partial dislocations with predominant screw component are generated in the second layer, leading to a twist of the new phase relative to the substrate. Our study indicates that the energy barrier is dependent on the density of the dislocation network, which is directly related to the lattice misfit. Further, we calculated interfacial energy of the liquid/substrate interface, which increases with increasing lattice misfit. We show that potency of a nucleant particle is directly relevant to the structural property, chemical nature and surface condition of the substrate.

X-ray radioscopic studies have been performed to improve the understanding of the complex interrelation between melt flow and the evolution of dendritic structures during solidification of Ga-In alloys. Natural convection is caused by density variations within the solidifying alloys. Forced convection was produced by electromagnetic stirring. Within this work special interest was focused on dendrite fragmentation and segregation phenomena. Melt convection alters the solutal field near the solidification front, leading to different microstructures or even to the formation of freckle defects. Essential process parameters such as flow patterns, solute concentration, the mushy zone morphology and permeability, dendrite growth velocities were quantified by image analysis. Particular attention is paid to the development of segregation structures and to the “self-healing” process of segregation zones. The observations indicate that if the local melt flow near the solidification front is destabilized, the Indium-rich melt flows inside the channel. The consequence is the “self-healing” process, i.e. the channel is filled in a short time by new dendrites and finally disappears. Eventual mechanisms that destabilize the channel formation are discussed so that the freckle defect can be eliminated by electromagnetic stirring on early stage of solidification. Moreover, our experiments demonstrate how the melt flow contributes to grain refinement, the CET (columnar to equiaxed transition) and dendrite fragments transport, which are discussed intensively in the literature.

Thermal diffusivity and individual solutal diffusivity of component play a significant role during natural convection dominated solidification of multicomponent alloys. In this study, in situ experimental observations of solidification of a ternary salt solution (water-potassium nitrate-ammonium chloride) are reported. The phase diagram of the ternary salt solution consists of three distinct regimes, distinguished based on the mode of heat transfer through the liquids. Each regime is divided into two subregimes on the basis of the primary solidifying component. In Regime II, the cotectic solidification leads to conduction dominated heat transfer through the liquid. The present study shows the real-time evolution of the convective flow in Regime II with different primary solidifying elements (ammonium chloride or potassium nitrate) using in situ optical techniques. A bottom-cooled experiment was performed using Mach-Zehnder interferometer (for thermal/solutal transportation) and particle image velocimetry (for flow field) in a rectangular cavity. The experiments show that the sub-regime II (with ammonium chloride as the primary phase) leads to the formation of double-diffusive convecting layers whereas the sub-regime-II leads to a more complicated convective flow with potassium nitrate as the primary solid. Furthermore, the mechanism of double diffusive convection in ternary mixtures is hypothesized.

Dendrites are ubiquitous structures that are central to setting material properties but the mechanism behind how dendrites grow is not fully understood. After x-ray synchrotron radiation experiments are performed, the dendritic growth of Al-12.6wt%Cu alloys is studied in 4-dimensions to track the evolution of the 3-dimensional structure as a function of time using novel x-ray tomography algorithms. Starting from a temperature above the liquidus, samples were cooled, at multiple cooling rates, to observe dendritic growth. These experiments provide information on the differences in dendrite tip growth rates, morphologies, branch spacing and interfacial curvatures due to different cooling rates. Two-point spatial correlations are used to investigate these quantities to provide a better insight into the process of dendritic growth.

The microstructure formed during the solidification processing from the melt is a critical issue for industrial applications. Most of the mechanical properties of alloys are a result of these microstructures. Understanding and predicting both the microstructure formation and evolution are crucial for the development of performant materials. Since microstructures are strongly influenced by the history of the solid-liquid interface evolution during the solidification process, in situ observation of the interface patterns is an important tool to gain knowledge on microstructure formation. To study the fundamental physical mechanisms in the dynamical formation of two-dimensional arrays under diffusive growth conditions, two experimental campaigns of directional solidification in a model transparent alloy - succinonitrile–camphor - have been conducted onboard the International Space Station using the Directional Solidification Insert (DSI) of DECLIC (Device for the Study of Critical Liquids and Crystallization) facility. This facility was developed by the French Space Agency (CNES) in collaboration with NASA. The solute concentration was changed between the two experimental campaigns so that both cellular and dendritic regimes could be explored up to highly-branched structures. The combination of in situ observation on transparent systems and microgravity environment which allows solidification experiments of unprecedented duration offered a very unique possibility to study solidification and the dynamics of patterns. This possibility puts us on the path to produce benchmark data needed to validate and develop theoretical and numerical models. Experimental observations, supported by 3D phase-field simulations, demonstrate the influence of subgrain-boundary, misorientations and macroscopic heterogeneities of the thermal field on the primary spacing selection and pattern organization. Some of the most striking results from both campaigns regarding the pattern selection will be presented.

The growth dynamics of multiple equiaxed dendrites in a thin metallic sample is studied using *in situ* synchrotron X-ray imaging of Al-4 wt.%Cu alloy solidification experiments and corresponding full-scale three-dimensional (3-D) phase-field (PF) simulations. The tip growth velocity *V*, tip radius *ρ*, the secondary dendritic arm spacing (SDAS) *λ*$_2$ and the total solid fraction *f*$_s$, are analyzed systematically in a relatively large range of cooling rate *R*$_c$. Through an in-depth comparison of the growth dynamics with experiments, including the tip velocity and the primary arm length, the nucleation undercooling for each grain is estimated. Quantitative agreements in *V* of several selected grains and *f*$_s$ between experiments and simulations have been achieved. The tip velocity, followed by free growth from the melt that is cooled continuously at constant *R*$_c$, decreases first at the very beginning of solidification in both experiments and simulations. Subsequently, *V* increases up to a peak value until the grains interact with each other by solutal effect. Moreover, an oscillation growth of the dendritic tip is observed during the soft-impinged growth stage in both experiments and simulations, featured by an acceleration of *V* after it goes down to a minimum from the peak value. With increasing *R*$_c$, *V* will be enlarged while *ρ* will be reduced, due to the narrowed solute boundary layer ahead of the tip. Furthermore, *ρ* and *λ*$_2$ show good agreements between simulations and dendrite growth theories, with *ρ* $\sim$ *V*$^{-1/2}$ and *λ*$_2$ $\sim$ *R*$_c^{-1/3}$. However, experimental measurements of *λ*$_2$ are generally larger than *λ*$_2$ in PF simulations, and this discrepancy could be attributed to liquid convection in experiments, despite the thin sample configuration. We incorporate liquid flow dynamics into the computationally less-demanding 2-D PF simulations. It has been found that with the gravity-driven convection the sidebranches are more developed and coarsened, thereby enlarging the spacing between the secondary arms.

In modern automotive structural components, metal sheets are combined with casting parts to enable a light and mechanical stable structure. In this study, the bonding between steel metal sheets and aluminum casting parts are achieved as part of the high pressure die casting (HPDC) process. In this way additional bonding steps like welding can be omitted. Gas enclosure at the interface between metal sheet and casting part may hinder a strong bonding. This effect is supposed to be enlarged by the rapid solidification at the interface between sheet and casting, where hot melt hits a cold metal sheet. Cooling rates are in the range of 1000 K/s. Such a rapid solidification leads to a stop of the melt flow already during the filling process and can trap gas enclosure at the interface. Coupled flow and solidification simulation is used to analyze the hybrid casting process with a focus on the wetting of the metal sheet by the melt and the formation of a strong bonding during solidification. Standard solidification models, based on a fraction solid curve, cannot describe essential phenomena of fast solidification, like undercooling of the melt. These phenomena may have a strong effect on the stopping of the melt flow and, hereby, the gas enclosure. Therefore, an advanced rate based solidification model is developed to address non-equilibrium aspects of the fast solidification and enable a more realistic simulation of the bonding process during solidification. The solidification model is calibrated by microstructure simulations using the multiphase-field method. The simulation results are validated against experiments performed on a cold-chamber HPDC machine (Buehler H630-SC) using simplified test geometries.

In Low Pressure casting (LPC), the filling stage is led by gradually increasing the gas pressure above the liquid metal, which is pushed through the pipe and into the mould cavity. In opposite to gravity casting, the LPC filling stage however does not only depend on the filling system and part geometries. By smartly choosing the pressure casting ramp, one could control the filling flow and thus avoid the filling induced defects. To avoid defects when considering the filling stage, it is necessary to simultaneously fill the mould cavity fast enough to avoid misrun and sufficient slowly to avoid oxides defects in the part. Indeed, when filling too fast, the oxides forming at the metal front are more susceptible to entrapment into the bulk [1], leading to poorer final mechanical properties. The relationship between the imposed gas pressure ramp, the system geometry and the induced metal filling dynamics needs to be investigated. Moreover, several oxides entrapment risk criteria were proposed in the literature [2][3][4] without reaching consensus. Filling flow criteria adapted to LPC should be defined. In this work, the filling dynamics induced by the geometry and LPC process parameters is investigated. An experimental setup developed at the semi-industrial scale permits to track the metal front during filling using electrical contacts. The experimental results are compared to commercial software fluid flow simulations and to a new proposed analytical model. Combining those three techniques, the link between process parameters, geometry and filling dynamics is quantitatively determined. Moreover, experiments with different process conditions are analysed in order to link the filling flow to the final mechanical properties of the parts. Eventually, new design rules adapted to LP can be proposed.

Ultrasonic processing (USP) during direct-chill (DC) casting of light metal alloys is typically applied in the sump of a billet. This approach, though successful for structure refinement and modification, has two main drawbacks: (a) mixture of mechanisms that rely heavily on dendrite fragmentation and (b) a limited volume that can be processed by a single ultrasonic source. We suggest moving the location of USP from the sump to the launder and applying it to the melt flow for continuous treatment. The apparent benefits include: (a) degassing of the melt volume, (b) grain refinement through activation of non-metallic inclusions, fragmentation of primary crystals, and deagglomeration of grain refining substrates, and (c) a possibility to use a single ultrasonic source for processing large melt volumes. To optimize this process with regard to the acoustic intensity and melt residence time in the active cavitation zone, flow modification with baffles as well as informed location of the ultrasonic source are required. In this paper, we demonstrate the results of experimental trials where the degassing degree and grain refinement have been the indicators of the USP efficiency for two aluminium alloys, i.e. LM25 and AA7050. The results are supported by acoustic measurements and computer simulations.

Piston Al-Si eutectic alloys are used to produce direct-chill cast billets for subsequent forging. Because of a very complex composition and multi-phase heterogeneous structure, it is necessary to control the formation of primary and eutectic compounds either through alloying or casting conditions (or both). In this study we used ultrasonic melt processing above or across the liquidus line to affect the occurrence and size distribution of primary Si as well as morphology of primary Al dendrites and high-temperature eutectic phases. The refinement of these particles has potential benefit for mechanical properties and formability during forging.

For directional solidification experiments with low withdrawal rates near and even below the constitutional undercooling limit of the corresponding alloy, a sample is needed which reveals a homogeneous concentration along the sample axis. As stirring in samples of a certain length is difficult, homogenization by rapid solidification is very common. Hereby, the liquid sample is rapidly moved from hot to cold and thus fine and long dendrites with segregated inter-dendritic liquid are formed. By subsequently melting and holding the sample in a resting position within a given temperature gradient, diffusion in the liquid and enhanced diffusion in the hot solid diminishes local concentration differences. In this paper, we report on homogenization and solidification experiences with transparent organic model compounds of TRIS-NPG. During one hour of homogenization we observed (i) formation of a coarse grain structure, (ii) liquid film and droplet migration by temperature gradient zone melting, and (iii) formation and eventually dis-appearance of liquid channels. During solidification experiments we occasionally found liquid channels, which revealed a connection to the solid/liquid interface; liquid that rose in the channel and thus formed micro-plumes at the planar solid/liquid interface. This upwards motion is ex-plained by solutal buoyancy of NPG-rich liquid and a possible feeding through an interconnected network of channels.

An in-situ study on the directional solidification of an inoculated Al-20 wt%Cu alloy under well controlled constant cooling rates and temperature gradients has been carried out using a microfocus X-radiography set-up. The influences of temperature gradient and cooling rate on the heterogeneous nucleation rate and growth kinetics of equiaxed grains have been studied quantitatively. It is shown that under the same cooling rate, the nucleation rate of grains decreases with increasing temperature gradient. A high temperature gradient also promotes preferential growth of dendrite arms along the temperature gradient direction, and therefore the formation of elongated grains. However, the temperature effects on nucleation and grain growth decrease with increasing cooling rate. It is revealed that the propagation velocity of the nucleation front in directional solidification castings is approximately equal to the ratio between cooling rate Ṫ and temperature gradient G. Based on the experimental observations, a grain size prediction model has been developed, in which the temperature gradient effect on the nucleation kinetics was rigorously treated by introducing two new concepts termed as ‘inhibited nucleation zone’ (INZ) and ‘active nucleation zone’ (ANZ). The model has been applied to simulate the present in-situ solidification experiments. A good agreement was achieved between the predicted grain number density and the experimental measurements. Furthermore, such a model can be used to predict the temperature gradient necessary for the transition from equiaxed to columnar grain growth.

The Interdependence model [1] predicted that nucleation would occur in waves of events with regions of no nucleation in between each wave. The waves continue to form until nucleation covers the sample. The cause of this phenomenon was attributed to the formation of a nucleation-free zone which incorporates solute suppressed nucleation and inhibited nucleation zones. Recent real-time synchrotron x-ray studies by Prasad et al [2], Liotti et al [3] and Xu et al [4] have confirmed this hypothesis showing nucleation occurs in a step-wise fashion with a number of events occurring followed by little or no nucleation for a short period before another set of events occurs. A microscale solidification model that predicts diffusion-controlled dendritic growth has successfully shown the effect of the developing constitutional supercooling on the selection of nucleation events. In this study, we use this model to predict the solidification behaviour under the conditions experienced during these real-time synchrotron studies. 1. StJohn et al., Acta Materialia, 2011; 59: 4907 2. Prasad et al., Journal of Crystal Growth, 2015; 430: 122 3. Liotti et al., Sci. Adv. 2018; 4:eaar4004 4. Yijiang Xu et al., Acta Materialia, 2018; 149: 312

For liquid metal, heterogeneous nucleation is main channel towards the formation of solid crystal. Fine tuning of nucleation means an optimized solidification structure, and consequently promises tailored properties for application. There are a few extensively used industrial practices for nucleation tuning, as AlTiB catalyzed grain refining in Al alloys, silicon modification for AlSi alloys, spheroidization of graphite in cast iron, and so on. In new energy storage area, the nucleation control is also important for high energy lithium-ion batteries. An unexpected nucleation of lithium leads to infinite relative volume expansion presenting low cycling efficiency and even safety hazards in batteries with lithium metal electrodes. The nucleation potential of a new crystal on substrate was believe to be a function of lattice misfit between new crystal and substrate as stated by conventional nucleation theory. But as we understood, solidification is a liquid/solid transition rather than solid/solid transition, thus the lattice misfit between two crystals are hardly to elaborate the real scenario of nucleation process. The state-of-the-art analytical techniques offer the opportunities tracking the atomic structural evolution adjacent to solid-liquid interface which makes the in situ observation on heterogeneous nucleation possible. Here we report up-to-date results on in situ observed heterogeneous nucleation of a few metals on different substrates. The results indicates that the nearest neighbor distance of liquid atoms and their coordination number can be tuned by the alloying element or substrate. The prenucleated ordering structures in the liquid have a tendency to be aligned in a preferred orientation, which is templated by the lattice structure of the substrate. This prenucleated ordering structure with preferred orientation will match the atomic structure of the substrate forming new crystals. That’s the reason why the lattice mismatching, a parameter describing the matching of new crystal with substrate, is still able to predict the nucleation potential of the substrate in a given liquid. The alloying element can be used to tune the lattice mismatching at the interface through interfacial structural tuning to enhance heterogeneous nucleation in the liquid. This findings will extend the understanding on heterogeneous nucleation leading to more effective nucleation tuning approaches in casting foundries, epitaxial growth of functional metal film or even more extensive application as energy storage area.

The nucleation and Growth behaviors of primary Si particles in hypereutectic Al-Si-Cu alloys solidified under near-isothermal melt and constant cooling rate condition, have been studied by in-situ X-radiography. The influences of cooling rate and addition level of P on the nucleation temperature, maximum nucleation undercooling, nucleation rate, growth velocity and morphology development of primary Si particles were quantitatively studied. In addition, thermal analysis and ex-situ characterization of the TP-1 type solidification samples by SEM were performed. The in-situ experimental results were compared with TP-1 type solidification samples. An EBSD study has been conducted on the as-solidified in-situ X-radiographic samples to reveal the Growth crystallogrphy of the Si particles. The effect of P on the nucleation and growth of primary Si has been discussed based on experimental results and modeling results.

Isomorphic inoculation has recently been introduced by the authors as a successful method to grain refine cast titanium aluminides [1]. Analyses of the cast grain size together with introduced particle size distributions revealed anomalously high grain refinement efficiency which was attributed to the particles breaking up during the holding stage prior to solidification [2]. In the present work, the microstructure of the inoculant powders is investigated in both the cryo-milled state as well as after simulated thermal cycles to reproduce their heating and holding in the melt. Results show that milling time does not impact the grain size in the particles, only their size distributions. Heat treatments between 1500 and 1600°C for short periods of time allowed the activation energy for grain growth and evaluation of the grain size evolution in the particles during the isomorphic inoculation process to be determined. Assuming that grain boundary melting is the predominant break up mechanism, a model to estimate dissolution of the powders is presented which includes diffusion and fluid flow. Despite its relative simplicity, the predicted number of particles remaining after heating and holding, which lead to grains in the as-cast structure, are in good agreement with the measured grain size. Finally, the paper summarizes the main features and mechanism making isomorphic inoculation a promising route for grain refining as-cast alloys. [1] J. R. Kennedy, D. Daloz, B. Rouat, E. Bouzy, J. Zollinger, Grain refinement in TiAl alloys by isomorphic self-inoculation, Intermetallics, 95 (2018) 89 – 93. [2] J. R. Kennedy, D. Daloz, B. Rouat, E. Bouzy, J. Zollinger, Effect of inoculant alloy selection on isomorphic inoculation of Ti-Al, Materials 11(5) (2018) 666.

Observations of melting crystallites in microgravity showed unusual shape changesas melting proceeded toward extinction. When re-analyzed in 2011, shape evolution data showedneedle-like crystallites becoming spheroids as they melted toward extinction, suggesting thatsome type of capillary phenomenon at solid-liquid interfaces was responsible for an energy releasecapable of spherodising particles on melting, and stimulating pattern formation during unstablecrystal growth. The presence of these previously undetected energy fields was recently uncoveredusing phase-field simulations that employ an entropy density functional. Simulations allowmeasurement of interfacial energy distributions on equilibrated solid-liquid interfaces configuredasstationarygrain boundary grooves (GBGs). Interfacial energy source fields—related togradients in the Gibbs-Thomson temperature—entail persistent cooling along GBG profiles,a new result that fully confirms earlier predictions based on sharp-interface thermodynamics.This study also provides new insights to improve microstructure control at reduced scales byexplaining the thermodynamic fields responsible for pattern formation in castings.

The quantitative prediction of dendrite fragmentation has remained a major issue in the modeling of solidification processes. A systematic approach to this issue is presented that consists of two parts. First, a thorough analysis is carried out of existing and new experimental data. A suite of different image processing techniques is developed that allows for obtaining robust statistics of the complex history of dendrite fragments, ranging from the growth conditions of sidearms through their detachment to the buoyant motion as free dendrites. Second, an analytical model is proposed that incorporates local solidification conditions of the sidebranches as well as the dynamics of the pinch-off process, which has been characterized in a previous numerical study. The predictive potential of this model is discussed in the context of the columnar-to-equiaxed transition and formation of single-crystal defects.

Molecular dynamic simulations, ab initio (DFT) calculations and experimental evidence suggests that there is a liquid-solid transition region which may be characterised by an order parameter. In this interface region the order parameter is not observed to be symmetrical, rather it tends to be steep on the solid side and exponentially decreasing on the liquid side. The order parameter in phase field computations is, to date, always assumed to give a symmetrical interface region. Hence, we ask how to extend the phase field model to give a profile that fits this data, and how such a model affects the simulation.

Modelling of non-equilibrium solidification is of singular importance in microstructure control, which however owing to the complex systems with complex additional constraints is still an open problem. In this work, the thermodynamic extremal principle was applied to solve the complex additional constraints self-consistently in thermodynamics for both binary and multi-component alloys. Consequently, short-range solute redistribution and long-range solute diffusion that share the same mobility are integrated naturally into the solute diffusion equations, thus avoiding the introduction of additional kinetic coefficients (e.g. interface permeability) to describe solute redistribution. Under non-equilibrium conditions, solute trapping as well as solute drag happens and at the critical interface velocity that is equal to the maximal solute diffusion velocity in liquid, abruptly concurrent occurrence of diffusionless solidification and absence of solute drag happens, thanks to the adoption of effective mobilities for non-equilibrium solute diffusion. Under equilibrium conditions, the interface and bulk contributions are completely decoupled, indicating that the current model might be preferred for simulations not only because of its simplest form of evolution equations but also its feasibility to increase the simulation efficiency by the “thin interface limit” analysis.

Kinetics of crystal growth in undercooled melts is analyzed by methods of theoretical modeling. Special attention is paid to rapid growth regimes occurring at deep undercoolings at which non-linearity in crystal velocity appears. A traveling wave solution of the phase eld model (PFM) derived from the fast transitions theory is used for a quantitative description of the crystal growth kinetics. The \velocity { undercooling" relationship predicted by the traveling wave solution is compared with the data of molecular dynamics simulation (MDS) which were obtained for the crystal-liquid interfaces growing in the h100i-direction in the Ni50Al50 alloy melt.

Due to the significant effects of the underlying microstructure on the material properties, accurate understanding of microstructure evolution in solidification processes has attracted lots of attentions in recent years. The low melting point of the ternary eutectic Bi-In-Sn alloy has lead to accurate in situ observations of its solidification process in experimental studies. Although these observations have deepened our understanding of microstructure evolution, due to complexity of the procedure there are lots of open questions of ongoing mechanisms. In the current study a thermodynamically consistent phase-field model based on the Grand potential approach is utilized to simulate the pattern formation in directional solidification of the system. As the first step of simulation studies, CALPHAD database is benefited to model the Gibbs energies and derive the chemical potentials of associated phases and the driving forces of the solidification process. Based on the reported system and process parameters like interfacial energies, solidification velocities and temperature gradient, two and three dimensional simulations are performed. The experimentally investigated 2D lamellar microstructure with ABCB repeat unit (A: $\gamma$-Sn , B: BiIn$_2$ , C: $\beta$-In) is achieved and the effects of material and simulation parameters on the phase arrangement is studied. Case studies of different instabilities like oscillations of phases boundaries, bifurcation and tilting of solidified phases are performed to investigate their stability ranges. The anisotropy of interfacial energies is considered as an effective parameter in pattern formation which is difficult to be controlled or measured in experimental studies. By means of different anisotropy models and the possibility of exact control of parameters,the effects of orientation and strength of the exposed anisotropy is determined. These effects can be summarized as changes in repeat unit and strength of the observed instabilities. Finally for each solidification velocity, the most stable conditions are introduced.

The industrial sectors of energy and transport require the development of equipment with high levels of safety and performance. These requests compel steelmakers to develop cast products with a high microstructural, chemical and mechanical homogeneity. As part of the SOFT-DEFIS project, collaboration between academic and industrials partners has been developed to optimize steel ingots quality in response to these expectations. The numerical modeling activity presented here is part of this project and aims at predicting the formation of solidification grain structures - columnar or equiaxed - and associated segregations. In addition, comparisons between simulation results and experimental analyzes of industrial ingots are planned as validation step. A finite element approach is proposed for the modeling of grain structures development [1]. A level set method is applied in order to track the columnar front interface and a dendritic growth kinetics model is used to estimate its velocity. Equiaxed grains then develop in the undercooled liquid domain. A solid, an intra-granular liquid and an extra-granular liquid phases are associated to each columnar or equiaxed microstructure with specific chemical compositions. This original approach make possible to model the evolution of solidification progress at the level of elementary volumes and the exchanges between the two types of grain structure. The conservation equations are solved using a splitting method [2]. Thus, their complexity is reduced by a sequential and time-decoupled resolution of the macroscopic transport and microscopic growth stages. Finally, at the scale of the ingot, the macrosegregation process is simulated considering the convective transport induced by buoyancy forces and the transport of equiaxed grains [3]. 1D and 2D simulation cases demonstrate the interest and efficiency of this approach. In addition, the importance to distinguish columnar and equiaxed grain structures is shown, in particular to predict the segregation processes and to improve the resolution steps. The model is applied, finally, to the scale of industrial foundry parts to predict the evolution of the developed structures and segregated zones, in comparison with the experimental analyzes of the project partners. [1] N. Leriche, Doctorat de l’Université de Lorraine, 2015 [2] M. Založnik, H. Combeau, An operator splitting scheme for coupling macroscopic transport and grain growth in a two-phase multiscale solidification model: Part I–Model and solution scheme, Comp. Mater. Sci. 48 (2010) 1-10 [3] T.T.M. Nguyen, Doctorat ParisTech, 2015

The properties of a material are mainly defined by their chemical composition and by the underlying microstructure. Depending on the process conditions during the directional solidification and the applied material system different microstructures evolve, which are suitable for the demands of specific applications. Instable growth conditions during the directional solidification can lead to an increase of microstructure regions with irregular phase arrangements. These areas can compromise the resulting properties of the material and hence their applicability. A result of instabilities is the growth of eutectic colonies in the microstructure. Eutectic colonies are mainly observed in ternary systems for alloy compositions in the vicinity of a binary eutectic reaction. The formation of colonies is driven by instabilities in the planar solidification front, which are caused by the ternary impurities diffusing from the two solidifying phases into the liquid. To investigate the formation process of eutectic colonies in their complex spatial arrangement, two- and three-dimensional large-scale phase-field simulations based on a Grand potential formalism are conducted. The formation of eutectic colonies is observed for two independent systems in different manners. In a first study, increasing amounts of silver are added to the isothermal undercooled binary eutectic system Al-Cu, to study the influence of the third component concentration on the colony formation. Due to the different compositions, colonies in different shapes and sizes form. In an additional study, the composition of the high-performance material system NiAl-34Cr is set fixed, but the solidification velocity is systematically varied beyond the stability range of the microstructure to initiate the growth of colonies. Based on this the stability of the evolved eutectic colonies is analyzed by a systematic variation of the applied temperature gradient. The focus of both studies is the investigation of the underlying mechanism of colony formation and of the correlation between the colony arrangement and the eutectic structure inside each colony. Furthermore the interactions in the contact zones between the colonies are studied as these are indicated to be the weak points for the applicability.

Magnetic fields have varieties of ubiquitous influence on nucleation and growth in solidification. As long as plenty of experimental phenomenon and results respectively have been revealed and retrieved, the mechanism of effect of magnetic fields on nucleation and growth in solidification keeps ambiguous. Not only thermodynamics mainly clarifies mechanism of growth qualitatively in solidification under magnetic field, but also it try to explain nucleation qualitatively in vain. Crystal-melt interfacial energy, a crucial parameter of dynamics, plays a key role in nucleation and growth in solidification. Crystal-melt interfacial free energy represents an important material constant influencing various aspects of microstructural evolution and thus materials properties decisively. Static magnetic field performs a notable effect on the undercooling of pure Al and Bi and the morphology of crystal-melt interface of Al-Cu alloys. Several studies have inferred that static magnetic field may affect the absolute value and anisotropy of crystal-melt interfacial free energy. Herein, experiment determination of crystal-melt interfacial free energy of opaque metals and alloys is always difficult and the examination precision is fundamental to the absolute value and anisotropy of solid/liquid interfacial free energy. Grain boundary groove method and dihedral angle method are very plausible to determination of the absolute value and the relative value of crystal-melt interfacial free energy respectively. Equilibrium shape method is very reliable to measurement of the anisotropy of crystal-melt interfacial free energy. The absolute value, relative value and anisotropy of crystal-melt interfacial free energy in the Al-Cu system have been retrieved with and without static magnetic field in this work. Key words: Crystal-melt interfacial free energy, Static magnetic fields, Solidification

Low pressure turbine blades (LPT) made by centrifugal casting from titanium aluminides require demanding process control, as to achieve desired solidification microstructures. The columnar-to-equiaxed transition (CET) and the related texture are of special interest. In the joint European Project “GRADECET” they were investigated in μg, 1g and hyper-g conditions. Solidification experiments of the alloy Ti-46Al-8Nb in EAS’s Large Diameter Centrifuge (LDC) show that with increasing angular velocity the equiaxed grain formation is promoted, while also depending on alloy’s composition and solidification pathway. Our numerical model handles thermal radiation and conduction in the entire furnace, the transient steps of melting and columnar solidification coupled with thermo-solutal flow. Furthermore, to understand the observed asymmetric CET event and distinct scenarios about the origin of equiaxed grains, motion of nuclei with flow under centrifugal condition was modelled by the Lagrangian approach. Taken together the simulation results reveal that the centrifugal and Coriolis forces affect the pattern and magnitude of the flow thus changing the growth conditions including undercooling, temperature gradient and nucleation ahead of the columnar front. When the angular velocity in the LDC increases, CET conditions based on Hunt’ model are met more early, which is in good agreement with experimental observations.

The columnar growth of Ti-48Al-2Nb-2Cr alloy was investigated using power down directional solidification experiments under various gravity conditions, ranging from micro-gravity to 20 g. The g-vector always pointed in direction opposite to the growth direction. The columnar dendrite morphology and primary spacing was analyzed in longitudinal and transverse sections and for selected experiments in 1g and 15g also by X-Ray tomography. The results show that the primary dendrite arm spacing decreases with increasing gravity from an average value of 790 µm at µg down to 370 µm at 20g. Furthermore, the spacing evolution along the entire solidification length depends not only on the temperature gradient and velocity but also on the magnitude of the applied gravity. The experimental analysis was complemented by numerical simulations using among other an OpenFOAM code to couple solidification and fluid flow. We will present and discuss the observed spacing refinement and the associated refinement mechanisms based on current models of dendrite growth under the action of upward flow in the mushy zone. Key observations form phase-field modeling are presented in a second contribution.

This second paper addresses the phase field modeling of columnar dendritic growth for different scenarios of fluid flow developing in response to gravitational forces. For this the experimental Ti-48Al-2Cr-2Nb alloy has been simplified to a binary alloy Ti-48 at.% Al. The 2D simulations confirm the decrease of the primary dendrite arm spacing with increasing gravity, nearly matching the experimental findings. In addition, the simulations show that above a critical magnitude of the gravity level dendrite tips display a characteristic splitting behavior, leading to highly branched microstructures. While tip-splitting can be periodic also without flow [1], the present results indicate that splitting events are closely coupled to the convection rolls between neighboring dendrites. The highly dynamic growth behavior however leads to an overall stable advancement of the average solidification front. The instabilities are discussed based on the imposed growth conditions and fluid flow velocities. The role of the anisotropy of the solid/liquid interface energy on the pattern dynamics will also be discussed. [1] Chen, Y., Billia, B., Li, D. Z., Nguyen-Thi, H., Xiao, N. M., & Bogno, A. A., Acta Mater. 66 (2014) 219-231

The modelling of solidification processes in combination with fluid flow is one application of ANSYS Fluent. The solidification is modelled with the enthalpy porosity technique. Therefore, the fluid flow is damped similar to a flow through a porous media of dendrites. For the numerical simulation of remelting processes like electro slag remelting or vacuum arc remelting ANSYS Fluent is used due to the fast solvers and large implementation possibilities. In case of materials with large solidification ranges, like alloy 718, the built-in adjustment possibilities of ANSYS Fluent are often not adequate. The program postulates a simple dependency between liquid fraction and solidification enthalpy. To improve the simulation, the solidification was implemented by a user defined function (UDF). The principal modelling of fluid flow is based on the theory of ANSYS Fluent, but it is now possible to adjust the solidification enthalpy for each temperature exactly.

The formation of eutectic microstructures, composed of two phases or more, is often encountered during the solidification of industrial alloys. Fraction and characteristics of this structure have consequences on the subsequent phase transformations, mechanical strength and final end-use properties of metallic pieces. Consequently, estimation of eutectic growth kinetics is required in order to predict its development and internal scale during solidification processes. However, most of existing eutectic growth models are within the frame of the primary Jackson-Hunt theory [1] and systematically use linear approximations for thermodynamic properties and solidification front curvature effects. These approximations limit their application to low growth rate. In addition, these models also neglect variations of densities between phases and advection of liquid phase induced by shrinkage effect. An original model of directional growth for lamellar eutectics is presented. This model is based on the analysis and computation of thermodynamic equilibrium at the solidification front. Contrary to previous models, this new approach does not make use of linearization hypotheses for the alloy phase diagram. In addition this model considers variations of densities within phases along with the influence of curvature on thermodynamic equilibrium at the solidification front. The numerical implementation is made possible through a coupling with a thermodynamic software based on the CALPHAD approach. This new numerical model is applied to the Al-Al2Cu eutectic structure and compared to the Jackson-Hunt theory. Large differences between results of the present model and this theory are observed at high growth rate. Validations with experimental observations developed in a large range of growth velocities and reported in literature are also proposed. Analytical developments are finally provided in order to explain and estimate evolutions observed in eutectic microstructure evolution. The application of this model to multicomponent alloys will be discussed. [1] K.A. Jackson, J.D. Hunt, Lamellar and rod eutectic growth, Trans. AIME, 236 (1966), 1129–1142

Silicon is widely used in the industry of solar panels. During the manufacturing process, the molten silicon is solidified into ingots. These ingots are cut into very thin layers which are assembled into solar cells. It turns out that the solar cells efficiency depends on the solidification process and the final microstructure of the ingots. Several experimental studies of solidification showed that the growth shapes exhibit facets along the {111} directions because the growth rate of facets being much lower than that of rough surfaces, it thus controls the process. There seems to be an overall agreement on the fact that the facets appear because of the anisotropy of surface energy, kinetic attachment, or both. However, the experimental results do not allow a definite conclusion about which phenomenon is the most relevant and this question is still a challenge for the numerical methods. In this work, we developed a 3D phase-field code incorporating the two phenomena and that takes into account the experimental conditions described in [1,2]. The results of the simulations are compared to the experimental ones and conclusions are drawn out [3]. [1] X. Yang, K. Fujiwara, K. Maeda, J. Nozawa, H. Koizumi and S. Uda, Crystal growth and equilibrium crystal shapes of silicon in the melt, Prog. Photovolt: Res. Appl. 2014,22:574-580. [2]A. Tandjaoui, N. Mangelinck-noel, G. Reinhart, B.Billia, T. Lafford, J. Baruchel, Investigation of grain boundary grooves at the solid-liquid interface during directional solidification of multi-crystalline silicon: in situ characterization by X-ray imaging, J.Cryst.Growth, 2013,377:203-211 [3] A. K. Boukellal and J-M. Debierre, the effect of the anisotropy of surface energy and kinetic attachment on silicon solidification (in progress).

Near eutectic Al-Cu droplets were rapidly solidified by Impulse Atomization. A wide range of microstructural scales was obtained at different cooling rates and undercoolings. The micrographs of the investigated samples revealed two distinct zones of different structural morphologies: An undulated eutectic morphology developed during recalescence following the single grain nucleation and a regular lamellar eutectic morphology resulting from the solidification of the remaining liquid post recalescence. The volume fraction of each zone was measured as a function of the droplet diameter, and the nucleation undercooling was deduced using the hypercooling limit equation. Scanning Electronic Microscopy imaging and microhardness measurements were used to evaluate the microstructural scale, and mechanical properties.

Microstructures, growth orientation and the room temperature mechanical properties of Mg-14.61Gd alloy were investigated by using directional solidification technology and CAFE simulation method under G=30K/mm and v=10-200μm/s. The experimental alloys were mainly consisted of primary α-Mg phase and (α-Mg+Mg5Gd) eutectic phase under all the growth conditions, which is consistent with the calculated results by Scheil model. It was found that α-Mg primary phase presented unidirectional dendritic morphologies on longitudinal cross-section, the growth interface appearance of α-Mg changed from the protruding forward growth to the flat growth gradually and the dendritic arm spacing decreased gradually with the increasing v. When v increased from 10 μm/s to 100 μm/s, the main growth orientation of α-Mg changed from <11-20> and <10-10> to <11-20> , and the deviation angle ( θ ) from solidification heat flow direction reduced from 11.070° to 5.711°, the reason for this lied mainly in the change of the heat flux. Further increasing v up to 200 μm/s, the main growth direction of α-Mg was still in <11-20>, but the value of θ increased to 10.620°, and the anisotropy of the crystal was the dominant factor then. It was proved that the CAFE numerical simulation model could predict the grain structure and growth orientation reasonably for Mg-alloy. In addition, the intrinsic mechanism between microstructures, growth orientation and mechanical properties was also discussed. KEY WORDS Mg-Gd alloy, directional solidification, CAFE model, growth orientation, mechanical properties

Semi-grand canonical Monte Carlo (SGCMC) simulations are performed to derive thermodynamic properties of binary alloy from atomistic-based simulations. Particularly, solidus and liquidus compositions are directly derived for Fe-Cr alloy described by two different EAM potentials. Although the SGCMC simulation can derive relationship between the free energy and composition at any temperature straightforwardly, partial phase diagram obtained from SGCMC simulations strongly depends on the choice of interatomic potential.

The preferred growth direction of crystal in solidification of metallic materials is mainly determined by anisotropy of solid-liquid interface energy, g(n), where n represent the crystallographic orientation normal to the solid-liquid interface. The anisotropy of g(n) is described by anisotropy parameters e1 and e2 that characterize <100> and <110> growth, respectively. Effect of e1 has been considered in most of phase-field simulations for dendritic growth in metallic alloys which have a cubic crystal such as fcc and bcc, because the preferred growth orientation in cubic crystals has been supposed to be <100>. However, it was reported that the preferred growth direction of Al-Zn alloy, which has fcc structure, changes from <100> to <110> with an increase in Zn concentration [T. Haxhimali, A. Karma, F. Gonzales and M. Rappaz, Nat. Mater., 2006(5), 660.]. This phenomenon implies that anisotropy parameters depend on solute concentration. The morphological change of dendritic structures with transition in the growth direction of Al-Zn alloy was investigated in detail by phase-field simulations [J.A. Dantzig, P.D. Napoli, J. Friedli and M. Rappaz, Metall. Mater. Trans. A, 2013(44), 5532.] and several important findings were reported in the early work. A further investigation should be aimed at investigating effects of solidification conditions and type of alloy systems on the morphological change. Therefore, in this study, we conducted quantitative phase-field simulations to clarify the morphology of dendrites for different sets of e1 and e2 systematically. Also, effects of several factors such as degree of undercooling and temperature gradient in different alloy systems were investigated.

Thermal-solutal convection necessary occurs during the terrestrial solidification of alloys. Thanks to the recent progress in the in-situ observation technique using X-ray radiography, the direct observation of dendrite growth under high temperature has been made possible. In the in-situ observation, however, it is difficult to see the liquid flow during the dendrite solidification. In order to clarify the physics occurred in the images obtained in the in-situ observation, it is essential to reproduce the same phenomenon by the numerical simulation. Although a phase-field method is the most powerful numerical tool to simulate the dendrite growth, the high computational cost is a drawback. Moreover, the computational cost of the phase-field simulation coupled with thermal diffusion, solute diffusion, and melt convection is quite expensive. Therefore, the most phase-field simulations of dendrite growth performed so far are limited to a single dendrite growth or dendrites growth in two-dimension. In this study, we develop a high performance computing scheme of the three-dimensional phase-field simulation for the dendrites growth with thermo-solutal convection. The computations are accelerated by a multi-GPU parallel computation in a GPU supercomputer, TSUBAME3.0 at Tokyo Institute of Technology. Then, we introduce the large-scale phase-field simulations of dendrite growth taking into account the thermal-solutal convection.

Microsegregation has a great influence on properties of materials such as mechanical and corrosive properties and it entirely depends on size and morphology of solidification microstructure. Although a number of efforts have been devoted to understanding and controlling of microsegregation in alloy systems, little has been clarified about microsegregation behavior at grain boundaries formed by growth of differently-oriented dendrites. Quantitative phase-field simulations have become one of the most powerful approaches for studying solidification phenomena and they have recently been employed in studies on competitive growth of dendrites in bi-crystal systems. Importantly, it is possible to analyze detail of microsegregation behavior using a quantitative phase-field model for two-sided asymmetric diffusion. In this study, therefore, we investigate microsegregation behavior at grain boundaries by means of two-dimensional quantitative phase-field simulations of directional solidification in bi-crystal systems. Our focus is placed on grain boundaries which are formed by two columnar dendrites; one is favorably-oriented (FO) dendrite and the other is unfavorably-orientated (UO) dendrites against the direction of temperature gradient. The microsegregation at the grain boundaries are investigated by changing the inclination angles of UO dendrite against the temperature gradient. The simulations are accelerated by using parallel computing on graphics processing unit (GPU) and by using moving frame scheme.

Computational simulations of solidification microstructures have contributed to understanding of time evolutions of complex morphologies of crystals in a variety of solidification processes. A phase-field method is an effective method in simulating the solidification microstructures such as a dendrite structure, a typical morphology of growing solid in solidification of metals. Importantly, accuracy of phase-field simulations is entirely determined by accuracy of input parameters for the simulations. In particular, accuracy of parameters for solid-liquid interfacial properties such as interfacial energy and its anisotropy strength are very important because such parameters directly affect the morphology of growing solids. However, it is not straightforward to determine interfacial properties with high accuracy by means of experimental methods. Methods such as capillary fluctuation method and cleaving technique have been developed to calculate the interfacial properties based on molecular dynamics simulations. Although these methods are very effective, these methods are applicable only to calculation of the equilibrium properties of solid-liquid interface. Since little has been clarified about non-equilibrium properties of interface, it is necessary to develop a method that can be applied to estimation of non-equilibrium interfacial properties as well as equilibrium one. In addition, it is highly desirable to develop a simple but accurate method for estimating the interfacial properties. In this study, data assimilation is applied to the estimation of interfacial properties at the solid/liquid interface during solidification of metals. The twin experiment was performed in order to investigate the effectiveness of data assimilation to the estimation of interfacial properties.

After solidification, grain growth occurs in the metallic materials, in general. It is very important to understand grain growth as well as solidification in order to control the mechanical properties of metallic materials. Abnormal grain growth(AGG), in which some grains preferentially grow to be extremely large, often occurs during cooling and isothermal holding at high temperatures after the solidification. AGG is related to non-uniformity of secondary phase particles and/or change in their distribution with time. Since AGG degrades mechanical properties, it is necessary to elucidate the mechanism and condition for occurrence of AGG. The precipitates in matrix and also remaining liquid in solidification serve as the second phases associated with AGG and the size of such pinning particles is often much smaller than the grain size of the matrix. Phase-field models have been utilized for simulating pinning effects on grain growth. However, early models cannot be applied to analysis of pinning effect and AGG when the pinning particle is very small compared with the grain size of matrix. This is because the pinning particles are explicitly described in the early models and hence the simulation of grain growth with very fine particles requires huge computational cost. Therefore, in this study, the pinning effects of particles are introduced in the phase-field model based on a mean-field approximation. The present model can reproduce the curvature driven motion of grain boundary with pinning force. Using this model, we investigate occurrence of AGG in carbon steels under various heat treatments.

Time-resolved in situ tomography of dendritic growth in Fe–0.45 mass% C carbon steel was performed using synchrotron radiation X-rays at SPring-8 synchrotron radiation facility (Japan) with improvement of the image quality using a physics-based filter. The voxel size of the reconstructed image was approximately 6.5 μm × 6.5 μm × 6.5 μm, and the time resolution (duration of 360° rotation) was 4 s (0.25 rps). Three-dimensional images of the dendrites were reconstructed even without image processing; however, the low contrast resolution in Fe–C alloys led to poor image quality. Consequently, it was impossible to precisely track the solid/liquid interface or evaluate the average curvature. To improve the image quality, a physics-based filter (a PF filter) was developed using a phase-field model. In the PF filter, images were retrieved in terms of interface curvature. The PF filter significantly improved the computed tomography image quality. As a result, dendritic growth was clearly observed even in Fe–C alloys. Moreover, the average curvature of the solid/liquid interface was evaluated as a function of solidification time (solid fraction). The ability to systematically characterize growing dendrites will be beneficial for modeling and simulation of solidification phenomena.

The growth of faceted crystals occurs often in nature and industry, involving always the presence of flow. Insulin, silicon, pyrite, quartz, gallium are only few examples of faceted crystals. The present paper presents a numerical model for the simulation of facetted crystal growth, taken into account the incidence of flow. The growth in facetted crystals is the results of interface kinetics and flow hydrodynamics. This model was applied to the Fe2Al5 faceted crystal, presenting a hexagonal crystal shape. These facetted crystals (Fe2Al5), so called top dross particles are forming during the production of Zn coated steel sheets (Galvanizing industry). In the galvanizing industry their occurrence is a problem and therefore it should be limited. Comparison was made between simulation and experimental observation of crystal shape. The growth was found to be the result of the coupling between the interfacial kinetic and the hydrodynamic induced kinetics.

Twin roll casting (TRC) is an energy efficient way of producing Mg alloy sheet for lightweight applications. However, unfavourable microstructural features, such as columnar grains and centreline segregation, are present in the as-cast alloy sheet. By modelling the effects of casting parameters, melt conditioning and material parameters on microstructure development, feedback can be provided to optimise the TRC process to improve the quality of TRC strip. A multiscale model is used to simulate the evolution of microstructure during TRC, in which the process of grain growth in fluid flow is modelled via phase field simulation coupled with a lattice Boltzmann model, and a Lagrangian macroscale model is applied for heat transfer analysis, which provides the boundary conditions for the microscale grain growth model. Results from the macroscale model can be used to deduce the upper bound casting speed, thus allowing manufacturing efficiency to be improved, and the deformation zone to be reduced, which is particularly favourable in TRC of Mg alloys.

The endogenous inclusion during solidification may generate from the combination of solute and non-metallic element, i.e, O, S, therefore the formation of such inclusion would take effect to global macrosegregation. To legitimately examine the interaction between endogenous inclusion and macrosgregation, an inclusion-combined macrosegregation model, which coupling the inclusion growth theory with the multicomponent solidification model, has been developed. Here solidification of a ternary alloy (Fe–0.45 wt.%C–1.06 wt.%Mn) is considered and MnS is the only generated inclusion. Spherical granular morphology is assumed for MnS which is treated as another continuous phase, apart from liquid, columnar and equiaxed crystals. Thermodynamic-controlled classical nucleation theory and diffusion-governed growth model are applied to calculate its precipitation process. The mixed solidification model principally takes some features into account as follows: growth of columnar dendrite trunks; nucleation, growth and sedimentation of dendrite equiaxed crystals; thermosolutal convection of the melt; solute transport by both convection and grains(equiaxed and inclusion) floatation; and the columnar-to-equiaxed transition (CET). This inclusion-combined macrosegregation model had been employed to study the formation of macrosegregation and inclusion in a 2.45-ton ingot. The final segregation pattern are in qualitative agreement with the reported experimental results. The formation mechanism of inclusion in such an ingot, and its influence in macrosegreation have been also discussed.

Permeability for flow of interdendritic liquid is a very important parameter in a macrosegregation prediction during alloy solidification. For systematical permeability prediction in various solidification conditions, a numerical simulation coupling a phase-field method and computational fluid dynamics would be the most promising approach. In our previous study, we have developed a permeability prediction method by large-scale simulations of phase-field and lattice Boltzmann methods [Takaki et al., (2018) submitted]. In the study, we have concluded that the permeability normal to a columnar dendrite in a regular hexagonal array can express one of the multiple columnar dendrites [Takaki et al., Acta Mater. 118 (2016) 230-243]. In addition, one advantage of the developed permeability prediction method is that we can use a periodic boundary condition for the single columnar dendrite. This means that we can apply the liquid flow in any direction to the single columnar dendrite. In this study, using the permeability prediction method developed in our previous study, we compute the permeability in various directions of a columnar dendrite. And, we investigate the direction dependent permeability of columnar dendrite in detail. Through the detail investigation, we attempt to develop a permeability tensor of the columnar dendrite.

Solidification cracking is an important issue during welding, casting and some of the additive manufacturing process. In order to illuminate the failure mechanisms, solidification cracking during arc welding of steel are investigated in situ with high-speed, high energy, synchrotron X-ray radiography approach. Analysis of the in situ radiography sequence revealed the solidification cracking initiates in the weld sub-surface trailing the welding electrode at relatively low true strain of about 3.1% in the form of micro-cavities. Although both material type and bending speed influence solidification cracking, cracks propagate from the core of the weld towards the free surface along the solidifying grain boundaries was found at a speed of between 1.7 -2.6 × 10−3 m s−1 for three different steels. In addition, a three-stage mechanistic model for solidification cracking during welding of steel is proposed.

Due to start-up and shut-down operations of engine, TiAl structural components usually undergo not only static but also cyclic mechanical loading. The crack propagation mechanisms of γ-TiAl under two types of loading are studied in this work to reveal the differences of the mechanisms under constant strain rate and cyclic loading. Since the crack prefers to nucleate at the interface, two types of loadings are applied to a γ-TiAl interface system with a pre-existing micro-crack at the interface by the means of classical molecular dynamics simulation, the loading direction is along [111] perpendicular to the interface. The evolution of crack tip and dislocation is observed in atomistic scale. The results show that, under both loading types, the crack propagates asymmetrically, Shockley dislocations emit on the (-1-11) slip plane from the right crack tip and slip along [-1-1-2] direction. The dislocations blunt the extension of crack while the left crack tip propagates in a brittle way. During the cyclic loading, different with constant strain rate condition, the crack advances and dislocations slip with increasing loads and retreat during unloading. In addition, the stress decreases and the crack length increases with the increase of the cyclic loading number.

The study is focused on the influence of solidification thermal parameters upon the evolution of the microstructure (either cells or dendrites) of an Al-3wt%Mg-1wt%Si ternary alloy. It is well known that the application properties of metallic alloys will greatly depend on the final morphology of the microstructure. As a consequence, various studies have been carried out in order to determine the ranges of cooling rates associated with dendritic-cellular transitions in multicomponent alloys. In the present research work, directional solidification experiments were conducted using either a Bridgman (steady-state) device or another device that allows the solidification under transient conditions (unsteady-state). Thus, a broad range of cooling rates ( ̇ ), varying from 0.003K/s to 40K/s could be achieved. This led to the identification of a complete series of cellular/dendritic/cellular transitions. For low cooling rate experiments, low cooling rate cells to dendrites transition happens. Moreover, at a high cooling rate, a novel transition from dendrites to high cooling rate cells could be observed for the Al- 3wt%Mg-1wt%Si alloy. Additionally, cell spacing λC and primary dendritic spacing λ1 are related to the cooling rate by power function growth laws characterized by the same exponent (-0.55) for both steady-state and unsteady-state solidification conditions.

All growth and casting processes, used for example in the superalloy and photovoltaic industrial sectors, face challenges linked to the grain structure and crystalline defects left during the solidification step when aiming at improving the final desired properties. The post-mortem studies of the solidified ingots provide limited information on the mechanisms occurring during solidification, on their symbiosis or competition and on their kinetics which makes it difficult to control. As a consequence, it is essential to improve the understanding of the mechanisms of the formation of the final crystalline grain structure, of segregation and of the density of structural defects, namely dislocations. Within this context, our contribution consists in characterizing the fundamental growth mechanisms of high melting temperature (up to 1800°C) materials using in situ X-ray imaging in a unique device named GaTSBI (Growth at high Temperature observed by X-ray Synchrotron Beam Imaging). Two imaging characterisation techniques are combined during solidification using X-ray synchrotron radiation at the ESRF (European Synchrotron Radiation Facility, Grenoble, France): X-ray radiography and X-ray Bragg diffraction. On the one hand, the X-ray radiography method brings information on the morphology and kinetics of the solid/liquid (S/L) interface as well as on solute segregation. On the other hand, the X-ray Bragg diffraction gives additional information about the evolution of the grain shape and structure, misorientation, defect formation and the local level of crystal lattice distortion during growth. During the presentation, results concerning the solidification of Ni-Al alloys and silicon for PV applications will be presented. These experiments demonstrate the wide range of fundamental growth mechanisms that can be addressed and better understood with X-ray imaging applied during solidification. In metallic alloys, the formation of a dendritic network is studied as well as its interaction with convective flow. As concerns silicon for PV applications, essential features of twinning, grain competition and the origin and interaction of dislocations with grain boundaries in crystalline silicon are characterized.

This talk will review recent progress to bridge the gap between phase-field modeling on the microstructure scale and grain structure modeling. While phase-field modeling has been used successfully to model quantitatively dendritic alloy microstructures, only volumes up to about a millimeter cube (or even smaller for concentrated alloys) can be simulated even on today’s massively parallel computer architectures [1]. On a much larger scale, Cellular Automata coupled with Finite Elements (CAFE) models have yielded impressive predictions of grain structures of castings in both two and three dimensions [2]. However, those models do not resolve dynamical interactions between branches of hierarchical dendritic networks, which can strongly influence both intra-grain microstructure selection and the growth competition of different grains [3]. This talk will discuss efforts to overcome this limitation by using phase-field simulations to inform the choice of CAFE model parameters [4], thereby improving their predictions, or by developing coarse-grained models that can simulate dendritic microstructures on much larger length and time scales than phase-field models and that can be integrated into CAFE models. Among such approaches, the ones most explored and tested to date include the dendritic needle network (DNN) model that tracks the dynamical evolution of the hierarchical dendritic network on scale larger than the dendrite tip radius [5], and a mesoscopic model that approximates the complex dendritic morphology by its envelope [6]. Examples will be given that illustrate how those approaches have been used to model columnar and equiaxed microstructures, and transitions between them. Advantages and limitations of those approaches will be discussed to highlight ongoing challenges in scale-bridging and to provide an outlook for future developments. 1. A. Karma and D. Tourret, Current Opinion in Solid State and Materials Science 20, 25-36 (2016). 2. T. Carozzani, H. Digonnet, and Ch.-A. Gandin, Modelling Simul. Mater. Sci. Eng. 20, 015010 (2012). 3. D. Tourret, Y. Song, A. J. Clarke, and A. Karma, Acta Materialia 122, 220-235 (2017). 4. A. Pineau, G. Guillemot, D. Tourret, A. Karma, and Ch.-A. Gandin, Acta Materiala 155, 286-301 (2018).
 5. D. Tourret and A. Karma, Acta Materialia 120, 240-254 (2016). 6. M. Zaloznik, A. Viardin, Y. Souhar, H. Combeau, M. Apel, Materials Science and Engineering 117, 012013 (2016).

Directional solidification of eutectics has been extensively studied in systems which display regular coupled growth with the solid-liquid interfaces being non-faceted. The solid-solid phase boundaries however may be locked or loosely attached to planes with a low energy configuration being associated with a characteristic crystallographic orientation relationship (OR). The phase boundary energy landscape and its anisotropy can have a major impact on eutectic growth and we shall give an account of the current state of knowledge and of the most recent observations gained from bulk Al-Al2Cu eutectics during (i) the maze-to-lamellar transition, (ii) the steady state growth and (iii) the formation of eutectic cells. We refer to Al-Al2Cu eutectics for which the phase boundary energy anisotropy was estimated from molecular dynamic computations (MD). We shall give an outlook on future research tasks and discuss the ability to understand and control the orientation and defect density of an arbitrary lamellar eutectic.

Directionally solidified binary eutectics are of great practical interest as self-organized composite materials with tunable microstructural features. In a first approach, eutectic microstructures can be classified into rod-like and lamellar ones. Their formation dynamics, which results from a complex diffusion controlled dynamics of coupled-growth front patterns, has been extensively studied by in situ experimentation and time-resolved numerical simulations. Important results have thus been obtained on the morphological stability of eutectic growth shapes and some pattern selection processes. However, the problem of the possible coexistence of rod-like, lamellar, and other more complex shapes in a given sample, in brief, the lamellar-to-rod transition, still remains poorly explored. We have developed an in-situ experiment method that permits to visualize optically the evolution of coupled-growth front patterns in real time in transparent eutectic alloys. In the succinonitrile-(D)camphor system, we identified and located the stability limits (rod elimination, and rod splitting, respectively) of the stability interval of hexagonal eutectic-rod patterns. In this system, we also discovered a transition from rods to lamellae in the presence of a finite-size effect in semi-thin samples. These conclusions could be drawn from laboratory experiments during which thermosolutal convection in the liquid was negligible. On ground, this narrows the range of explorable parameters (composition, thermal field) and prevents one to undertake a systematic study of the lamellar-to-rod transition in large samples. We undertook a science-in-microgravity project (ESA/NASA) called TRANSPARENT ALLOYS (TA). An apparatus for in situ directional solidification experiments in large samples has been installed on board the ISS/MSG (Dec. 2017). We will present the results of the first TA campaign (January-March 2018 ; SEBA program) in collaboration with U. Hecht and V. Witusiewicz (Access e.V., Aachen, Germany).

During the solidification of ternary eutectics, three solid phases form simultaneously from the melt. One system often employed to study this complex multi-phase solidification behavior is ternary eutectic $Al$-$Al_2Cu$-$Ag_2Al$. In this system the solubility of silver in the α-Al phase decreases by approximately 50% in a range of less than 20 K below the ternary eutectic temperature. This diffusion-controlled solubility shift is obvious from a comparison of microstructure in a normally processed sample with microstructure from directly behind a quenched interface, in which the solubility shift did not have time to occur. There are differences not only in the phase fractions of the three solid phases, but also in the shape of the phase regions and the pattern itself. In order to study the evolution of the microstructure caused by this solubility shift, quenched samples of directionally solidified ternary eutectic $Al$-$Al_2Cu$-$Ag_2Al$ were systematically annealed and the resulting post-solidification effects were studied with scanning electron microscopy for two different solidification patterns. The results show that the adjustment of the phase fractions takes place much faster than the rearrangement of the shape of the phases. This shape evolution is strongly influenced by the anisotropic interface energies, resulting in more straight solid-solid interface boundaries instead of the the competing curvature reduction. The outcomes of this work demonstrate that in order to obtain a more complete understanding of directionally solidified ternary eutectic $Al$-$Al_2Cu$-$Ag_2Al$, both the solidification and the subsequent post-solidification processes must be considered.

Our comprehensive understanding of binary eutectic solidification is largely due to years of extensive studies of many metallic and non-metallic systems. Recently, there has been increasing effort to understand ternary eutectic solidification due to the potential for refinement of microstructure resulting from increased number of co-existing crystalline phases during the solidification of multicomponent alloy systems. So far, our understanding of quaternary eutectic solidification is very limited. Recent research work at BCAST has revealed the existence of nanostructured anomalous eutectic regions within intercellular colonies of the lamellar eutectic mixture found in suction cast Al-28%Cu-6%Si-2.2%Mg (wt%) quaternary eutectic alloy, characterised by a combination of EDX, high resolution SEM and TEM techniques. This contribution presents detailed studies of the microstructure and chemistry of the cellular colonies of lamellar eutectic and nanostructured anomalous eutectic regions in the as-cast Al-28%Cu-6%Si-2.2%Mg (wt%) alloy, together with a description of the quaternary eutectic solidification behaviour in the Al-Cu-Si-Mg multicomponent alloy system.

Al-Si alloys, which contains eutectic, are used in large quantities. Those are used not only like cast alloys but for base material of brazing technology as well. It is well known that too long Si lamellae can reduce the joining force of brazing, therefor it is necessary to produce finer eutectic structure. During manufacturing of brazing materials, the use of modifying agent (Na, Sr) is not always permitted. In these cases, we can perform to stir the melt. Eutectic and hypoeutectic Al-Si samples were solidified unidirectionally in a rotating magnetic field of different intensity and the developing structure was investigated. The regular investigation method for these irregular eutectics is the measure of the average distance of eutectic Si lamellae. But this technique is not saying anything about the size of lamellae and about the direction of growing them. Because that we worked out a measuring method by use of mosaic pictures. During our research work it was found that stirring of melt can change the microstructure of the sample radically: primary silicon and primary aluminum appear near each other and the structure of eutectic is very different around the primary phases. The size and the direction of eutectic lamellae were modified by stirring as well. Our study shows interesting effects of magnetic stirring and generate some question about the causes.

The solidification of alloys shows a large variety of different microstructures depending on the material system and processing conditions. Since materials properties such as tensile strength are dependent on the microstructure, its prediction is a topic of high interest in order to produce materials with tailored properties. Whereas theory is capable of investigating simple geometries, simulations are utilized to ascertain the influence of processing conditions on complex evolving geometries. An example of this is the coupled growth of dendrites and eutectics, which typically evolve at different length scales. % irgendiwe ohne example schreiben In order to simulate this coupled growth, the phase-field method is chosen as it has been established as a versatile tool to investigate microstructural evolution. The phase-field model is based on a grand potential approach with parabolic free energies approximating thermodynamic CALPHAD data of the system Al-Cu. With this the coupled growth of coarse dendrites and fine eutectics during directional solidification is investigated in two as well as three dimensional simulations. Depending on the process parameters, observations include: Closely-spaced dendrites turning into cells, stable coupled dendritic-eutectic growth, nucleation of eutectics on dendritic sidebranches as well as transitions to a completely eutectic state. Based on these results a tentative microstructure map is established. Finally, the influence of the dendritic growth direction on the microstructure is investigated.

A two dimensional model for the simulation of dendritic growth in binary alloys is developed. Phase field model is used to derive the system of partial differential equations describing the temporal evolution of the solid-liquid interface and concentration field. Quantitative modeling is assured by the use of thin interface limit of the isothermal phase field model. Meshless local radial basis function collocation method and explicit Euler scheme are used for the spatial and temporal discretization of the phase field equations, respectively. An h-adaptive computational node arrangement is developed in order to assure the high density of computational nodes near the solid-liquid interface. The model is verified and assessed by comparison with the analytical models. The influence of the node arrangement to the dendritic morphology at different preferential growth directions is analyzed. The speed-up of the simulation due to an h-adaptive computational node arrangement is assessed. The originality of the work is in the novel, adaptive approach to the efficient and accurate solution of the phase field model. Advantages and shortcomings of the novel adaptive meshless method as well as further developments are discussed.

The 3-D morphology, orientation selection and behind mechanism of magnesium alloy dendrite were investigated by performing multiscale mathematical simulations. The results showed that the α-Mg dendrite prefers to grow along <11-20> basal direction and <11-23> nonbasal direction, and the resultant 3-D dendritic morphology exhibits a typical 18-primary branch pattern. Such dendritic microstructure formation was found to be associated with the surface energy anisotropy in light of the hexagonal-close-packed (hcp) lattice structure of the α-Mg dendrite. An anisotropic function model describing the α-Mg dendrite growth was developed by combining the observed 3-D growth pattern, the spherical harmonics, and the calculated anisotropic strength via DFT-based atomistic simulations. Accordingly, the dendritic growth behavior and morphological evolution were further investigated by coupling the anisotropy function with the phase-field model. The simulated results on the 3-D dendritic morphology agreed well with those reconstructed from X-ray tomography experiment. Our investigations provides a deep understanding on the growth behavior of magnesium alloy dendrite from phenomenological descriptive picture to a more intrinsic predictive way.

Experiments performed in the International Space Station have evidenced oscillatory modes for the solidification of a succinonitrile-based transparent alloy. These oscillations have been studied by phase-field numerical simulations that allowed a better understanding of their origin [1-2]. A recent re-examination of the experimental data showed that, depending on the misorientation of the crystal axes with respect to the temperature gradient direction, the oscillations may be triggered or inhibited [3]. We present here new results of phase-field simulations that show a similar sensisitivity to the cystal axes misorientation. These results are analyzed and compared with the conclusions drawn from the experimental data. Combining both, a quantitative criterion for the inhibition of the oscillations is obtained. [1] Bergeon et al, Spatiotemporal dynamics of oscillatory cellular patterns in three-dimensional directional solidification, Physical Review Letters, vol 110, p. 226102-1-5, 2013 [2] Tourret at al, Oscillatory cellular patterns in three-dimensional directional solidification , Phys. Rev. E , vol. 92 , p. 042401-1-16, 2015 [3] Pereda et al, Experimental observation of oscillatory cellular patterns in three-dimensional directional solidification, Phys. Rev. E, vol. 95, p. 012803-1-19, 2017

Grain growth, a competitive growth of crystal grains that occurs after or simultaneously with solidification, is one of the most important phenomena in controlling the microstructure of polycrystalline materials. The fundamental process underlying grain growth is the migration of grain boundaries; thus, the growth behavior is largely dominated by the properties (energy and mobility) of grain boundaries. In real materials, these properties usually exhibit strong anisotropies depending on the boundary structures. Over the past few decades, many researchers have attempted to elucidate the effects of the anisotropies of the boundary properties on grain growth by performing numerical simulations. However, conclusive knowledge is not yet established especially for three-dimensional systems, mainly due to the limitations in the computational accuracy of the grain growth models and computer resources that have been employed so far. To address the above issues, based on the multi-phase-field (MPF) grain growth model [I. Steinbach and F. Pezzolla, Physica D (1999) 385], this study proposes a novel numerical scheme to accurately handle the anisotropic grain boundary properties. Furthermore, the MPF simulations are drastically accelerated by parallelizing multiple graphics processing units (GPUs) on a GPU-rich supercomputer. Using these techniques, we perform a series of large-scale simulations on anisotropic grain growth, through which the effects of the anisotropic grain boundary properties on the growth behaviors are elucidated in detail.

Additive manufacturing by Laser Powder-Bed Fusion (LPBF) is accompanied by solidification under large thermal gradients and cooling rates. As a consequence, solidification morphologies between dendritic with only limited sidebranches or rather cellular pattern are commonly observed. The growth regime is characterized by Péclet numbers of order or larger than 1 (with cell tip radii of order or larger than the length scale of the diffusive pile-up in the melt), and by significant kinetic effects at the solid/liquid interface. We will present results from 2D and 3D phase-field simulations of primary solidification of the Ni based alloy IN718 for a range of solidification rates (1 ... 10 cm/s) and temperature gradients (10^5 K/cm ... 10^6 K/cm) typically seen in LPBF. The operating state of the growth front will be discussed in terms of cell tip shapes, growth undercooling, primary spacing and the effect of crystallographic orientation. A general observation is that, for such high Péclet numbers, usual mechanisms of spacing selection and accommodation are less effective. Cellular morphologies and spacing variations show good agreement with experimental data for IN718.

The need for efficient and clean solutions, due to the increasing current environmental regulations puts extra pressure on new combustion engine development, to compete in a market with alternative driving concepts. Downsizing and weight reduction can reduce the engine emission and efficiency, but require light alloys with superior thermo-mechanical properties for high temperature exposure to maintain the same engine performance. Cast Al-Cu could be alternative to standard Al- Si alloys for new engine generations due to their higher temperature strength, creep-resistance and long term stability of engine components. In Al-Si and Al-Cu cast alloys with heterogeneous microstructures a composite-like deformation behavior is responsible for superior high temperature properties. Stiff Si or Al2Cu particles, respectively reinforce a ductile -Al matrix to a composite with improved thermo-mechanical strength. However, different Young’s moduli and coefficients of thermal expansion are responsible for micro stress gradients and unpredictable micro crack formation under operation. These micro-mechanical deformation mechanisms in Al-Si and Al-Cu systems, responsible for crack initiation and growth, have been scarcely investigated so far. This manuscript describes an example of elasto-plastic deformation mechanisms in an AlCu7 alloy. Tensile testing shows anomalous macroscopic deformation behavior indicating unknown internal micro-mechanical processes. External loading until yield strength and beyond are applied under laboratory conditions and during in-situ neutron diffraction. The results of macroscopic deformation and micro strain evolution are compared and correlated with the heterogeneous micro structure. High resolution synchrotron computed tomography reveals conclusions on the micro-mechanic deformation mechanisms and their effects on the macroscopic damage initiation and material’s service performance.

Continuous casting is currently the main industrial process for steel production. Since long time, industries search for efficient simulation methods, by which macrosegregation and deformation induced cracks can be predicted. As a first step this requires achieving concurrent simulation of fluid flow and stress-strain. Therefore, a partitioned solution algorithm is developed for such simulation with application to continuous casting. Liquid flow induced by natural convection or filling step, solidification shrinkage as well as thermally induced deformation of solid phase are accounted for.

The grain boundary groove method has been successfully used to measure solid-liquid interfacial energies, σSL, experimentally for binary eutectic and peritectic systems, multi-component systems as well as pure materials and for opaque materials as well as transparent materials. It was shown that the grain boundary groove method can be use to obtain σSL for any alloy system provided that the prepared alloy sample can be held at the evaluated temperature for a long enough time with a very stable temperature gradient. In order to show the applicability of the groove method to any system, a part of the Al-Zn phase diagram was chosen. Equilibrated grain boundary groove shapes for solid Alα solution (Al-30wt%Zn) in equilibrium with AlZn liquid (Al-60wt%Zn) have been directly observed with a radial heat flow apparatus. The Gibbs-Thomson coefficient, Γ, was determined with a numerical method using observed groove shapes. The measured thermal conductivities of the solid Alα solution and AlZn liquid phases and the temperature gradient in the solid phase at the solid-liquid interface were used for the calculation of Γ and then σSL was determined using the Gibbs-Thomson equation. The grain boundary energy for the same system was also obtained from the observed groove shapes. The results of the work were compared with the results of the related experimental works.

The foundry engineering still needs new cast materials of improved properties. These can be achieved by elaborating completely new alloys and metal matrix composites or by elaborating the alloys/composites basing on the already very well-known matrixes. The good example of the latter solution are high-aluminium zinc (H-Al Zn) and high-zinc aluminium cast alloys (H-Zn Al). Both of these groups show good damping and strength properties but rather low ductility and insufficient structural (and dimensional) stability, caused by long term transformation of the Cu-containing bearing phases. The performed works were devoted, among others, to building composites of improved structural stability reinforced with ternary (Al,Zn)-Ti aluminides. The in-situ reinforcement was built by needle-shape ternary aluminides based on the DO22 TiAl3 binary phase introduced with AlTi-based master alloys or by compacted semi-globular ones based on the L12 Zn3Ti particles introduced with a ZnTi-based master alloy. The mentioned particles substitute partly or totally for Cu-based bearing phase and the influence of this substitution on the structural stability and tribological properties is also discussed in the paper.

In this study, the pores in die-cast AlSiMgMn alloys were inspected and reconstructed with high resolution three-dimensional (3D) X-ray micro computed tomograpgy (μ-CT) technique. Finite element (FE) meshes were built with consideration of the pore actural morphorloges from the CT inspection. Based on dutile damage model, the FE simulation of tensile fracture of the alloys was carried out. The simulation results were compared and verified with the tensile of in-situ scanning electron microscopy (SEM). The two results are agreement in the main crack path and pores on the fracture. With the pore-scale simulation, the effects of pore characteristics on the stress distributions and crack initiation and growth during the tensile were analyzed. It was found that the pores of lower sphericity and larger project area in tensile axis direction are prone to form microcracks and promote main crack deflection. The results also show that aggregation of brittle alpha-Fe intermetallics of the alloys also has important influence on the main crack propagation.

Lamellar eutectic two-phase growth is in principle well understood for alloys in which the solid-liquid and solid-solid interfaces are isotropic. However, there are numerous experimental observations that cannot be explained by theories and numerical models with isotropic interfaces. Examples are the occurrence of lamellar growth directions that markedly differ from the direction of the temperature gradient, or the emergence of large regions of perfectly aligned lamellae in large samples. We have developed a phase-field model in which the anisotropy of each interface can be controlled separately, and we have investigated the case of anisotropic solid-solid (interphase) boundaries. Results will be presented (i) on the selection of lamellar growth directions in thin-sample directional solidification, (ii) on the ordering of lamellae in bulk samples, and (iii) on the dynamics of spacing homogenization in inhomogeneous lamellar arrays. Simulation data will be compared to available theories and experimental data.

We present an in situ experimental study of crystallographic effects during lamellar- eutectic growth in thin Al-Al2Cu samples. The Al-Al2Cu alloy is known to form eutectic-grain dependent lamellar eutectic microstructures, and to present special, or heteroepitaxy orientation relationships (OR) between the eutectic crystals. The term crystallographic effects refers here to a marked dependence of the growth microstructure on the orientation of the two solid phases relative to each other, and the direction of growth. We prepared thin (≈10 µm thick) metallic films by plasma sputtering on sapphire plates. The thin-sample geometry allows one to observe lamellar-growth patterns in real time without convection in the liquid, by using a long distance optical microscope (reflected-light mode) focusing the contact surface of the metallic film with one of the transparent substrates. We could identify both « floating » grains with a regular dynamic behavior in steady-state, and locked grains, within which the lamellae grow inclined, parallel to a fixed direction. We combined our optical observations with ex situ crystallographic analyses usign two methods, namely, x-ray diffraction and EBSD in order to determine the ORs in the different eutectic grains. We will comment on our results in the light of a recent work on the effect of an anisotropy of the surface free energy of the interphase boundaries on lamellar eutectics.

Directional solidification is a favored process to manufacture homogeneous microstructures which lead to macroscopically unique properties for a material. The dependence of the spacing and type of the arising microstructure from the solidification velocity for constant velocities is extensively investigated. However the effect of changes in the solidification velocity on the resulting microstructure adjustment processes is still unclear. Therefore large-scale (3D+t) simulations of the ternary eutectic system Ag-Al-Cu with changing solidification velocities are conducted with a phase-field model based on the grand potential approach. To study the spatially complex rearrangement process during velocity changes in statistical representative volume elements, simulations with different velocity profiles are calculated in large-scale domains. The results show, that the evolving microstructure continuously rearranges by splitting and merging of the rods despite constant growth conditions. By increasing the velocity, the microstructure refines by splitting of the Al2Cu phase. Whereas by decreasing the velocity, the microstructure coarsens by overgrowing events of both intermetallic phases.

Eutectic alloys are self-organising composite materials with a wide variety of microstructural features. Size, shape, distribution, and orientation of these features can be modified by process parameters (temperature gradient, velocity of interface) as well as material parameters (volume fractions, anisotropy of interfacial energies, diffusivities, impurity nature and its percentage etc.). The objective of this work is to study the role of solid-solid and solid-liquid anisotropies in eutectic structure formation particularly the internal structure of colonies that arise due to a two-phase growth instability. We have chosen the Sn-Te eutectic system as the base binary system which has SnTe and Te phases, with Ag/Sb as impurity additions for triggering the colony formation. The binary and ternary (Ag/Sb addition) alloys are directionally solidified at different interfacial velocities to study the morphological evolution. In case of binary Sn-Te eutectic, broken labyrinth/ rod morphologies are observed (see left fig.). Upon addition of a third component, a diffusive instability (similar to a Mullins-Sekerka instability) leads to the formation of two-phase colonies that arise beyond a particular velocity. Critical velocities beyond which instabilities form are determined for each of the alloying additions. In our experiments we find colony structures having an internal sub-structure upon additions of both Ag/Sb. The internal structure due to Ag addition however, is different from Sb Addition (See figues). We explain the role of anisotropies in the free-energy of the interphase boundaries that lead to the formation of the different eutectic colony structures using results obtained from experimental characterization techniques (using SEM/TEM/EBSD).

We present an experimental investigation on the effects of the interphase energy anisotropy on the formation of three-phase growth microstructures during directional solidification (DS) of the β(In)–In2Bi–γ(Sn) ternary-eutectic system. Standard DS and rotating directional solidification (RDS) experiments were performed using thin alloy samples with real-time observation. We identified two main types of eutectic grains (EGs): (i) quasi-isotropic EGs within which the solidification dynamics do not exhibit any substantial anisotropy effect, and (ii) anisotropic EGs, within which RDS microstructures exhibit an alternation of locked and unlocked microstructures. EBSD analyses revealed (i) a strong tendency to an alignment of the In2Bi and γ(Sn) crystals (both hexagonal) with respect to the thin-sample walls, and (ii) the existence of special crystal orientation relationships (ORs) between the three solid phases in both quasi-isotropic and anisotropic EGs. We initiate a discussion on the dominating locking effect of the In2Bi–β(In) interphase boundary during quasi steady-state solidification, and the existence of strong crystal selection mechanisms during early nucleation and growth stages.

Spherical pre-alloyed Nb-20Si-24Ti-2Cr-2Al (at.%) powders were prepared by plasma rotating electrode processing (PREP). The diameters of the pre-alloyed powders ranged from 45 μm to 380 μm. The microstructure and surface oxides of Nb-Si based alloy powders were investigated by X-ray diffraction (XRD), scanning electron microscopy (SEM), X-ray photoelectron spectra (XPS) and Auger electron spectroscopy (AES). The main phases of Nb-Si based alloy powders were Nb solid solution (Nbss), βNb5Si3 and Nb3Si. Fine lamellar eutectic structure was formed during solidification, taking the central parts of eutectic branches. Anomalous eutectics were distributed at the edges of eutectic branches. The lamellar spacing (λ) of lamellar eutectics ranged from 0.12 μm to 0.38 μm, increasing with the increase of powder diameter. In addition to the fine lamellar eutectic colonies, large silicides (>15 μm) are occasionally observed in the sectional microstructures of some Nb-Si based alloy powders. The bulk oxygen content of the Nb-Si based alloy powders was as low as 0.06 wt.%. An oxygen-enriched layer with 5.14 nm in thickness, consisting of Nb2O5, SiO2, TiO2, Cr2O3 and Al2O3, was formed on the surface of pre-alloyed powders. The oxygen content was found to decrease exponentially with etch depth, and the amount in the oxygen-enriched layer accounts for 49.66 % of that in a whole single powder particle with a diameter of 120 μm. The outermost surface of Nb-Si based alloy powder was Nb-depleted and (Ti, Si)-enriched, resulting from the lower thermodynamic stability of Nb2O5 compared to the thermodynamic stability of TiO2 and SiO2.

A 3D meso-scale discrete-element model has been developed to simulate fluid flow during dendritic solidification of steel. The model domain is a representative volume element consisting of a set of equiaxed dendritic grain envelopes along with extra-dendritic liquid channels, where the final grain shape is given by a Voronoi tessellation. Solidification of each grain is simulated via a volume average approach. The output of the solidification simulation at a given solid fraction is used as the input mesh for the fluid flow simulation. A single domain Darcy-Brinkman model is used to calculate the pressure field within the liquid channels, with Poiseuille flow assumed to occur in the extra-dendritic region, and Darcy flow assumed to occur within the dendrite envelope. Mass conservation over each element is then used to derive a flow equation that is solved via the finite element method. The results of this new model are first compared with a previously-developed granular model [1] where fluid flow only occurs between the grains, and then compared with different forms of the Carman-Kozeny equation. It is shown that the intra-dendritic liquid fluid flow plays a major role in the semi-solid pressure field, and thus needs to be included when investigating hot tearing susceptibility in engineering alloys undergoing dendritic solidification.

In this talk, we will review recent developments using the multiscale Dendritic Needle Network (DNN) approach for dendritic growth, with a particular emphasis on its application to convective transport in the liquid phase. The DNN method was developed for the modeling of dendritic growth of hierarchical needle-like dendritic crystals, which typically form at low solute supersaturation or undercooling that are common to a host of solidification processes. It retains a good accuracy for a numerical space discretization about one order of magnitude larger than what is typically required in phase field simulations. By not tracking complex morphological details of the solid-liquid interface, the DNN method allows a scaling-up of phase-field with quantitative simulations orders of magnitude faster, while still tracking the transient growth competition of individual dendritic branches in each grain at the larger scale of heat and mass transport. After briefly summarizing key aspects of the method, we will illustrate validations of the model in 2D and 3D, including quantitative predictions of microstructural features measured in directional solidification experiments, and critical quantitative comparisons with other multiscale approaches for dendritic growth. We will demonstrate the applicability of the model for fluid flow in the liquid phase. We will show that quantitative predictions comparable to those from phase-field simulations can be achieved, hence opening the way to macro-scale simulations with experimental/processing relevant transport conditions, e.g. accounting for gravity-driven buoyancy. We will offer a critical assessment of the DNN approach, of the investigations it makes possible, of its current limitations, and of the resulting next developments in order to address more complex mechanisms in dendritic growth. We will present ongoing work, challenges, and perspectives on the effect of buoyancy in constrained directional solidification, and on further coupling with micro-mechanical modeling, i.e. crystal plasticity, in order to link processing, microstructures, and properties.

Mesoscopic envelope models rely on the description of the complex morphology of a dendritic grain by an envelope, which is a smooth surface connecting all of the dendrite tips. This simplification in the description of the grain shape enables to reduce requirements on spatial resolution and allows for modelling of larger number of grains or domains, when compared to more detailed microscopic approaches. Mesoscopic envelope modelling has already shown its capability of simulate columnar growth in transparent alloy systems during directional solidification, including misorientation between columnar grains [1], as well as equiaxed grains [2]. Here, we apply 3D mesoscopic envelope modelling to simulate columnar dendritic growth in the transparent alloy Neopentylglycol-(D)Camphor (NPG-DC) during directional solidification under microgravity conditions. Experimental results for comparison are taken from the « TRACE-3 » experiment [3], conducted on a TEXUS sounding rocket. Special emphasis is put on a comparison of characteristics of columnar growth and the conditions for columnar-to-equiaxed transition CET. [1] Viardin A., Založnik M., Souhar Y., Apel M., Combeau H., 2017, Acta Materialia 122, 386-399. [2] Souhar Y., De Felice V.F., Beckermann C., Combeau H. and Založnik M., 2015, Computational Materials Science 112A, 304–317. [3] Zimmermann G., Hamacher M., Sturz L., 2017, Proc. Of the 23 rd ESA Symposium on European Rocketr and Ballon Programmes and related research, A-035.

The last decade has witnessed the development of various solidification models based on volume averaging methods. To reduce the calculations cost, the averaging reduces information contained within the mushy zone to a single variable, the solid volume fraction. The solidification rate and the permeability are then calculated with the help of closure laws often based on some semi-empirical correlations. In the present paper it will be shown that beyond the need of sub-grid correlations, some essential mesoscopic features lakes these models. During solidification, natural convection arises from the temperature and concentration gradients. Concentration gradients are caused by the preferential incorporation or rejection of solute element at the solid-liquid interface. In order to catch the smallest solute plumes scale the solidification model simulates directly the envelope of the columnar dendrites with a cellular automaton model. The mushy interior of the dendrite is modelled with a volume averaging method. The dendrites tips are assumed to growth with a modified LGK model to account for the effect of the magnitude and the direction of the flow. Secondary arms are allowed to transform into a primary arm if the curvature of the envelop exceeds a critical value. The competition between grains is well reproduced by the variation between dendrite tips velocities. This model allows the study of the effect of both the grains, and the primary arm spacing on the hydrodynamics. The results given by this model (with primary arms) are compared with the ones predicted with a model using a full volume averaging method (primary arms invisible).

Multiple dendritic equiaxed grain formation is common to solidification in many technical alloys in industrial solidification processes. We investigate this type of grain formation and competition during directional solidification of grain-refined Al-3.5wt.-%Ni at various solidification conditions by comparison of experimental data and 3D dendrite needle network (DNN) modeling. For the experiments in-situ x-ray radiographic characterizations in thin samples reported in literature [1] are used. DNN modeling makes use of heterogeneous nucleation, branched dendritic growth and solutal interaction between branches and multiple equiaxed grains. Here, a first comparison using a 2D modeling approach valid at low Péclet numbers [2] is extended to account for thin samples in 3D, higher Péclet-number, as well as different grain orientations. [1] N. Mangelinck-Noël, J.E. Spinelli, Ch.-A. Gandin, G. Reinhart, H. Nguyen-Thi, B. Billia, IOP Conf. Ser.: Mater. Sci. Eng. 27 (2011) 012090. [2] L. Sturz, A. Theofilatos, Acta mat. 117 (2016) 356.

A 3D mesoscopic envelope model is used to numerically simulate the experimental X-ray observations of the equiaxed dendritic isothermal solidification of a thin sample of Al-20 wt%Cu alloy including the natural convection flow. Several four-grain simulations are run to investigate the effect of the convection, of the grain position, and of the grain rotation on the tip growth kinetics of one of the grains. We show that the effect of convection flow – consequence of the presence of gravity parallel to the sample thickness direction, z – on the growth kinetics of the reference grain depends significantly on the position of the grain along the sample thickness.

Comprehensive thermomechanical casting simulations offer the opportunity to increase the dimensional accuracy of cast parts. These simulations require accurate material properties, which are delicate to measure experimentally in the mushy zone. These properties are often specified only as function of the temperature, neglecting the impact of the non-equilibrium character of the evolving microstructure. Therefore, in order to derive a more precise description of effective properties of the mushy state, a mutiscale approach is presented here. It couples quantitative multi-phase field simulations of the microstructure evolution with a homogenization technique. Effective thermo-elastic properties of the predicted directional and equiaxial 3D microstructures are derived and their impact on the thermomechanical casting simulation of an axisymmetric A356 bowl outlined. Moreover, in this paper we want to develop for each semi-solid regime, the coherent and non-coherent one, a specific constitutive law and to ensure continuity of the stress state at traction coherency and at Tsol. Both constitutive laws depend on microstructural features like fraction solid and grain size, which are extracted from the coupled multi-phase field simulation. The nonlinear behaviour of the coherent semi-solid phase on the macro-scale is described by an original single surface viscoplastic flow potential that includes micro-structural parameters and takes the internal cohesion of the solid skeleton into account. The variation of the cohesion with the fraction solid is described by a simple law and compared to Ludwig’s expression [1]. The proposed viscoplastic potential takes the effect of isotropic hardening, pressure dependence of yielding and the strength difference in tension and compression of the coherent semi-solid state into account. For the non-coherent mushy state, presenting some shear resistance, a simplified viscoplastic potential is adopted, neglecting pressure effects on the solid dendrites and describes large plastic flow under shear loading. The stress continuity condition at mechanical coherence allows us the introduction of only one additional material parameter. The developed constitutive laws are implemented in the FE program Abaqus via a dedicated CREEP user routine. Simulations of uniaxial, isothermal tension, compression or pure shear experiments permits us to identify the model parameters of an A356 aluminium alloy. [1] O. Ludwig, J.-M. Drezet, Ch. Martin and M. Suéry: „Rheological behaviour of Al-Cu alloys during solidification. Constitutive modeling, experimental identification and numerical study”, Metallur. and Mat. Trans. A, vol. 36A, pp. 1525-1535, 2005.

In recent years, the automotive industry has been increasing the production of small, high-power gas engines as part of several strategies to achieve the new “Corporate Average Fuel Economy” (CAFE) standards. This trend requires an improvement in the thermal and mechanical fatigue durability of the aluminium alloys used in the production of the cylinder blocks in these engines. Conventionally, solid chills are employed in areas of these castings subject to high cyclic loading to enhance the mechanical performance of the cast material – i.e. in the main bearing bulkhead. A potential means of improving the efficacy of these chills is to incorporate water cooling. To assess the effectiveness of this method, a water-cooled chill was designed and installed in a bonded-sand engine block mould package (1/4 section). The moulds were instrumented with thermocouples, to measure the evolution of temperature at key locations in the casting, and “Linear Variable Displacement Transducers” (LVDTs), to measure the gap formation at the interface between the chill and the casting. This paper summarizes, at a high level, some of the findings of this work.

As it is known from literature, the metals tend to follow a viscoplastic law at high temperatures. Thereby, based on the authors’ previous developments to simulate the behavior of the equiaxed crystals packed bed, a viscoplastic stress model is applied to the thin slab casting process. The model is reduced to the single phase mixture formulation for faster and robust simulations. In this idea, the solidifying shell represents a ‘creeping solid’ and the Norton-Hoff type stress model is formulated with the model parameters obtained experimentally. A coupling procedure is established to converge the non-linear terms. The influence of the stress model parameters is investigated. Next the model is applied for the simulation of the thin slab casting to improve a previously developed technique: a viscoplastic rheology is applied to calculate the motion of the solidifying melt instead of imposing the velocities of the mush. The simulation results show that the most deformations happen at the funnel part of the mold, causing highest strain rates and the significant drop of the viscoplastic ‘apparent viscosity’ according to the Norton-Hoff law. The solid shell velocities are mostly uniform at the straight parts of the strand but a slight acceleration of the shell is observed along the funnel surface. Strong compression / expansion zones are detected at the funnel part, which could lead to defects formation. The solid shell thickness was successfully predicted as well and compared to the previous work by the authors.

The determination of pattern dimensions using simulation is an inefficient trial-anderror process that requires several design iterations. In this study, the finite element inverse elastoplastic analysis is utilized to calculate the pattern geometry in a single iteration for a plastically deformed body. A simplified casting system is simulated to demonstrate the feasibility of the inverse method. An inverse simulation is performed first to calculate the pattern shape. This configuration is then used as the input geometry for a forward simulation, which is shown to successfully recover the original as-cast shape used for the inverse analysis. Through this sequence, the inverse deformation method is shown to be a viable technique for the determination of pattern allowances in production castings.

Cast components generally show a heterogeneous distribution of material properties, caused by variations in the microstructure that forms during solidification. Variations caused by the casting process are not commonly considered in structural analyses, which might result in manufacturing of sub-optimised components with unexpected in-use behaviour. In this paper, we present a methodology which can be used to consider both thermomechanical and thermophysical variations using finite element analyses in cast components. The methodology is based on process simulations including microstructure modelling and correlations between microstructural features and material properties. Local material data are generated from the process simulation results, which are integrated into subsequent structural analyses. In order to demonstrate the methodology, it is applied to a cast iron cylinder head. The heterogeneous distribution of material properties in this component is investigated using experimental methods, demonstrating local variations in both mechanical and physical behaviour. In addition, the strength-differential effect on tensile and compressive behaviour of cast iron is considered in the modelling. The integrated simulation methodology presented in this work is relevant to both design engineers, production engineers as well as material scientists, in order to study and better understand how local variations in microstructure might influence the performance and behaviour of cast components under in-use conditions.

Chills are used in the production of metal castings as a thermal aid to promote directional solidification in casted sections. This study will review microstructure and thermal circumstances of the internal chill interface where the chill is intended to be incorporated into the structure of the cast section. The conditions for interface coherency of cast steel sections from 25 to 50mm will be shown. Solidification proceeds away from the interface chill growing to a maximum thickness and then melting back to the original chill geometry. Furthermore, it is shown that promoting section solidification prior to the complete melting of the internal chill can lead to the formation of interfaces will be coherent across the chill and cast sections for compatible steel alloys. A computer model of the heat transfer and interface evolution show the possibility of using the coherent interface of internal chills in the design of other cast components.

Additive manufacturing (AM) is inducing some kind of next industrial revolution. Components develop layer by layer in a powder bed by selective beam melting according to 3D model data. This technique allows manufacturing of highly complex components and is especially interesting for high performance materials that are difficult to process in conventional technologies. Nevertheless, AM is challenged by material quality issues such as porosity, binding faults, surface roughness, selective evaporation, texture, etc. In this contribution, the variety of physical phenomena important during powder bed based AM of metallic alloys is discussed based on mesoscopic simulation also taking into account the effect of individual powder particles. We show the influence of the powder properties, such as bulk density or size distribution, on the consolidation process and stochastic appearing faults. Examples show the influence of the processing parameters and the consolidation strategy on phenomena such as selective evaporation or texture evolution. In summary we show, how mesoscopic simulation improves process understanding and represents the basis for further process development.

A casting process usually starts with a liquid alloy that solidifies on cooling. Forced and/or natural convection in the melt as well as stresses and deformations that occur from uneven cooling of the just solidified part can be described quite accurately. The challenging range is when the amount of solid increases so that melt flow becomes more and more difficult while solid starts to build up a coherent network that becomes more and more rigid. In such regions, the solid network starts to transmit stress while the flow is dammed to zero. A physically sound description of these regions are crucial for understanding the formation of shrinkage porosity, hot tearing and even (deformation-induced) macrosegregation. In the present contribution we are presenting an overview about up-to-date experimental and numerical findings that are relevant for the aforementioned topics and discuss related issues that are still unresolved and thus need more research.

Chemical inhomogeneity and inclusions are the most severe defect types in bearing steel which have great influences on the ingot quality and final product properties. Based on a series of full-dissections of steel ingots varying with composition and weight, numerous experimental characterizations by SEM, EPMA, TEM, ASPEX and 3D Micro-CT techniques and multi-scale simulations, a new driving force, inclusion flotation, has been discovered to drive the formation of solutal macrosegregation, especially channel segregation. Sufficient populations of light oxide-based inclusions with appropriate size are able to alter the local flow pattern and destabilize the mushy zone, consequently dominating the channel segregation formation. Accordingly, a significant concept to control the macrosegregation in practice is the low-oxygen purifying treatment of the melt and inclusion modification, such as via Rare earth addition. Actually, even with a highly low proportional of RE’s additions in industrial-level 100Cr6 bearing steel ingots, inclusions can be effectively modified into smaller and much more compatible RE-O-S type with Fe matrix, and over 60% large-size inclusions beyond 5 μm disappear, avoiding the growth of large-size MnS/Al2O3. In addition, the small density contrast between RE-O-S inclusions and steels also leads to a slow flowing among inter-dendritic regions. These two factors reduce and even eliminate channel segregation, and also lower the global macrosegregation extent significantly. The magic of RE treatment on segregation control has been widely verified and applied in other special steels of mould, gear and nuclear power component. Besides, the appropriate casting process and optimized ingot design are also significant factors to reduce macrosegregation in bearing steel. For instance, the higher superheat degree above 40 oC induces the longer solidification time and stronger interdendritic natural convection; meanwhile the decrease of the local cooling rate coarsens the dendrite arm, consequently elevating the mushy zone permeability. Obviously, both the large density difference and coarse microstructure provoke the severity of macrosegregation. On the other hand, the larger height, narrower shoulder width and the smaller taper in the riser are beneficial to shifting the hot spot and the maximum segregation value from the ingot body towards the riser top. In summary, the current study highlights the importance to produce super-homogenized bearing steel by jointly purifying the steel melt, modifying the inclusions, lowering superheat degree and ideal hot-top design.

For a long time, the γ phase in metallic alloys has seemed to be produced through a peritectic reaction between the δ and liquid phases. However, direct observations have shown that a massive-like transformation, in which the δ phase transforms into the γ phase in the solid state, is dominant during or after solidification of the δ phase in carbon steels. To characterize such massive-like transformation, we use time-resolved tomography (4D-CT) to demonstrate the volume change during cooling from the melt and the crystallographic orientation relationship between the δ and γ phases. The volume changes from solidification and from the massive-like transformation from the δ to the γ phase were −3% and −0.5%, respectively. The transformation from the δ to the γ phase finished quickly, as demonstrated by the volume change. Fine γ grains were produced even in a single δ grain through the massive-like transformation. Also, the refined γ grains showed a wide crystallographic distribution.

We have used a concentric solidification technique to study events occurring in the early stages of solidification during the continuous casting of steel. We have found that the major factors contributing to crack susceptibility are the fraction of δ-phase present prior to the occurrence of the peritectic reaction; the rate at which the δ-to-γ interface propagates and the extent of undercooling. In addition, we have studied the δ-to-γ solid-state phase transition in low-carbon iron alloys of non-peritectic compositon and have shown that at cooling rates pertinent to continuous casting, a massive-type of δ-to-γ phase transformation can occur. We shall present and discuss important aspects of the concentric solidification technique and present selected studies in more detail. More recently, we have developed and will discuss, a differential thermal analysis technique, which combines synchronically in-situ observations with thermal analyses, thereby making it possible to correlate bulk behavior with in-situ observations.

We have used high temperature laser-scanning confocal microscopy by utilizing a concentric solidification technique to observe in-situ and in real time solidification events and high-temperature microstructural development, with a special emphasis on high temperature phase transformations. However, the solidification and subsequent solid-state transformation kinetics, on both heating and cooling, are largely determined by the axi-symmetric and dynamically changing temperature profile within the cylindrical specimen, thereby hindering quantitative analysis. We have experimentally determined the temperature distribution and created a simulation domain whereby we can replicate almost exactly the concentric configuration. This combination of thermal analysis with numerical simulations assistance with 3D modeling has enabled us to develop a powerful technique to convert in situ observations into quantitative analysis. We are now able to characterize by numeric modelling, the pertaining solute distribution at different stages of solidification incorporating grain boundary diffusion and maintaining the same solute boundary layer thickness as in the experimental set-up. In the present study, we have carefully defined the domain prior to the solidification, and then compared the numerically calculated solid/liquid interface velocities with the experimental determined values in Fe-0.18C and Fe-4.2Ni alloys at cooling rates that varied from 2K/min to 200K/min. Peritectic reaction kinetics are compared with experimental results for the developed solute profiles at the solid/liquid interface and optimized mobility coefficient between δ/γ . Because we have experimentally determined the pertaining thermal gradients, it is possible to assess thermodynamic parameters at the solid/liquid interface and at triple points using Thermo-Calc. The agreement we obtained between simulation and experimental results, provided us with confidence that the technique can be extended to the quantitative prediction (calculation) of the fundamental thermodynamic parameters pertaining to the concentric solidification platform.

Peritectic transitions - characterized by the formation of a secondary phase from decomposition of the primary phase and the melt - are very common in the solidification of metallic alloys. During the peritectic reaction, all three phases are in direct contact, thus forming a trijunction. Already in 1979, Hillert [1] emphasized the important role of the junction force balance in the peritectic reaction and pointed out that opposing capillarity forces are required to meet the local equilibrium conditions at the different boundaries in the vicinity of the junction. However, so far, little attention has been paid to the actual impact of the individual liquid-solid and solid-solid phase boundary energies, which cannot be expected to be equal in real materials. In this work, the multi-phase-field method was applied to study the impact of unequal phase boundary energies by the example of a peritectic TiAl alloy with negligible solid-state diffusion. Simulations were performed under isothermal conditions and systematic variation of the involved interfacial energies. Reaction rates, evaluated after reaching steady-state conditions, are discussed and correlated to the thickness of the peritectic layer. A major result of the case study is that two different growth modes can be distinguished a) when the triple junction is leading and rate-controlling and b) when the peritectic growth front is leading and its tip is rate-controlling. Which of these two growth modes is selected can be described by a critical ratio of the interfacial energies. [1] Hillert M. Solidification and Casting of Metals. London: The Metals Society; 1979.

Retained free energy drives the transformation from metastable to stable phase during double recalescence of steel alloys. Statistical methods were used to identify the mechanism controlling cluster growth based on the principle of microstructural reversibility. Application of the coefficient of determination during optimization procedures showed that an extrinsic mechanism controls nucleation and that negligible healing occurs such that retained free energy from primary phase undercooling and melt convection enhance secondary nucleation.

Formation process of equiaxed structure is quite complicated phenomenon. The most difficult point in the modeling of the equiaxed structure formation process would be a motion of dendrites. In our previous study, we succeeded in expressing the growth of a single dendrite accompanying the motion by coupling phase-field method, lattice Boltzmann method, and equations of motion [R. Rojas, T. Takaki, M. Ohno, J. Comp. Phys. 298 (2015) 29-40]. After that, we have extended the model to the polycrystalline solidification, where we expressed the multiple dendrite growth with motion, collision, and coalescence and subsequent grain growth occurred after the formation of grain boundaries [T. Takaki, R. Sato, R. Rojas, M. Ohno, Y. Shibuta, Comp. Mater. Sci. 147 (2018) 124–131]. In the model, we assumed the same energy and mobility for all solid-liquid interfaces and grain boundaries. In this study, we improve the accuracy of our previous polycrystalline solidification model by employing a multi-phase-field model [I. Steinbach, F. Pezzolla, Physica D 134 (1999) 385-393], and accelerate the simulation by employing the APT (active parameter tracking) algorism for multiple phase-field variables and the multi-GPU (graphical processing unit) parallel computation. Finally, we introduce some simulation examples, such as the formation processes of the sediment bed [C. Beckermann, C.Y. Wang, Metal. Mater. Trans. A 27 (1996) 2784-2795] and the equiaxed structure through dendrite fragmentation [L. Arnberg, R.H. Mathiesen, JOM 59 (2007) 20-26].

In this study, a numerical model was developed, and direct simulations were performed to predict solidification grain structures and macrosegregation based on a threedimensional cellular automaton finite difference (CAFD) method coupled with flow calculation of natural convection and shrinkage flow. First, to evaluate the model coupled with natural convention, simulations of unidirectional solidification for Al-10wt% Mg alloy were performed. Mg-rich plumes rising in the melt were seen due to subsequent upward flow, and Mg-rich channels forming in the mushy zone were observed. Columnar grains were then formed, and they became coarse afterwards. Next, to evaluate the model coupled with shrinkage flow, simulations of casting Al-10wt% Cu alloy in a unique mold, which can form macrosegregation in the central region of the small ingot, were performed. The bridging of columnar grains formed during solidification, and the positive segregation was generated in the region below the bridging. Thus, the main factor for this macrosegregation is the shrinkage flow with bridging. From the comparison of simulation results with and without the chill for the unique mold, it was established that the shrinkage flow and the bridging of solidification structures play an important role in macrosegregation.

The existence of oxide inclusions (e.g. Al2O3) in the steel melt is unavoidable. Most of these nonmetallic inclusions come from the steel-making process. During solidification, some of these inclusions remain in the melt and interact with melt flow and the growing crystals. The movement of solid oxide inclusions in steel melt is directly numerically simulated through coupling the solid particle moving dynamics with the Navier-Stokes equations, which are solved using a vector-valued method based on the adaptive finite element method in three dimensions. Prior to this simulation, the settling of a single spherical Nylon solid particle in water is firstly calculated and then benchmarked by the experimental measurements, which shows a very good agreement of the terminal velocity between simulations and experiments. Then settling of many particles is performed to verify the treatments of collision between particles and collision between a particle and walls. With these benchmarks and numerical tests, the liquid-particle interaction dynamics are incorporated into the phase-field model for the alloy dendritic crystal growth. The natural convection due to the solutal buoyancy is also taken into account. Two-dimensional numerical simulations of flotation of many light oxide inclusions in the interdendritic region during solidification are performed. Simulations show that when considering the inclusion flotation, the interdendritic melt convection becomes orders of magnitude stronger than that only considers the solutal buoyancy. The inclusion flotation during directional solidification not only changes the flow strength but also alters the selection process of primary dendrites. In the multiple equiaxed solidification, the large-size inclusions lead to the flow of inclusion and melt, and thereby generating a flow passage between dendritic crystals. Most of the particles flow to the top along this passage when the dendritic branches are not well developed. While the dendritic arms become sufficiently large, more and more inclusions are blocked by the dendrites near this passage.

Grain refinement of metallic alloys is of critical importance to minimize the risk of solidification defects and obtain the desired mechanical properties. The effectiveness of grain refinement depends strongly on the interplay between potency of nucleant particles for heterogeneous nucleation and their ability for grain initiation. For a given alloy, the former is an inherent property of the nucleant particle while the latter is strongly dependant on the size, size distribution and spatial distribution of the nucleant particles. Through analytical and numerical modelling, it has been shown recently that agglomeration of the nucleant particles has a strong influence on their grain initiation behaviour, and is largely responsible for the discrepancy in grain size between theoretical predictions and experimental measurements. In this work, a cellular automaton model is used to investigate the effect of nucleant particle agglomeration on the effectiveness of grain refinement. The spatial distribution of nucleant particles is modelled by a log-normal distribution and the mean particle spacing is used to represent different levels of particle agglomeration. Other variables include alloy composition, particle size, size distribution and number density.

The interaction between the capillarity and the thermal-solute-convection fields determines the microstructure evolution during solidification, which significantly influences the eventual mechanical properties of materials. However, to achieve the coupled thermal-solute-convection microstructure evolution with a realistic Lewis number (Le ~ 104) and Prandtl number (Pr ~ 10-2), the computing overhead is gigantic. In this work, the multi-physical microstructure evolution was simulated via a phase-filed lattice-Boltzmann approach, which enhanced the time step by 2-3 orders of magnitude in comparison with the explicit finite difference method. A parallel and adaptive mesh refinement algorithm was developed to further improve the computational efficiency. Accordingly, the fully coupled 3-D thermal-solute-convection dendritic and eutectic growth for Al-Cu alloys was first reproduced with a realistic Lewis number and Prandtl number. It was confirmed that the domain temperature and the presence of convection have significant influence on the final morphology. The simulated results agree well with the experimental findings.

Solidification phenomena during laser additive manufacturing (LAM) occur under rapid solidification conditions, with the entire melting and solidification process occurring in a few milli-seconds. New characterisation techniques are required to capture the powder-melting, weld pool formation, together with the subsequent solidification phenomena including defect formation. To enable this, a laser additive manufacturing process replicator (LAMPR) was designed and commissioned that performs full LAM builds in situ and operando on a synchrotron beamline. This enables imaging of the process in real and reciprocal space at ultra-fast speeds (up to 50,000 frames per second) and high resolutions (ca. 1 microns). Dozens of layers can be built with hatching in each layer, whilst capturing the laser-powder interactions, including melting and the formation of spatter caused by combined metal vaporisation and argon gas heating. During track formation, we observe the mechanisms by which contiguous track is formed, and when beading occurs and why. The different types of porosity formation are also observed and quantified, suggesting ways of avoiding their formation. Finally, the processing space is explored for an SS316 powder, and a mechanistic processing map is produced which not only quantifies the processing window, but also shows the mechanisms which limit operation to within this window, and hence how the window can be expanded to produce higher strength components more efficiently.

Niobium silicide-based composites, in the application of gas turbine blades, promise significant efficiency improvements compared to current Ni-based alloys. The higher temperature capability would allow the engine to run at a higher temperature than that of current alloys, increasing engine efficiency. Nb-Si based composites possess a lower density, due to the presence of ceramic phases such as Nb5Si3 and/or Nb3Si. This would reduce the weight of the rotating blades. However, improvements in certain properties, such as room temperature toughness and oxidation resistance are needed. This study focuses on the manufacturability aspect of the powder feeding laser additive manufacturing (LAM) process to engineering Nb-Si based alloy samples. LAM has the advantage of forming near-net shapes without the use of expensive cores and moulds for the reactive Nb-Si melt. Fine microstructure and even chemical composition distribution with reduced macro-segregation are obtained. With the use of power feeding system, new Nb-Si based alloys are LAMed with varying atomic composition. Microstructures of the LAMed alloys will be presented, and the relationship between the microstructure and the alloy chemistry will be reported.

Recently, Additive manufacturing (AM) has emerged as a disruptive technology to manufacture complex parts with greater scales of economy and also with a reduced need for post-machining. However, the influence of processing conditions and the alloy compositions on the strength, defect formation are not yet fully understood. A quantitative simulation of the microstructure morphologies and segregation patterns for a given alloy composition will help towards a better prediction of the physical and mechanical properties of the developed part. More particularly, properties such as the susceptibility to solidification cracking are directly related to the segregation behavior as well as the structure of dendritic network in the mushy zone. In this work, a Cellular Automata based three-dimensional (3D) model was developed to simulate dendritic growth for the case of AM. The Thermocalc database was used to obtain the equilibrium multi-component phase diagram. Finite Difference schemes were used to obtain the temporal and concentration fields in the computational domain. The interface velocity and concentrations were obtained by simultaneously solving the corresponding Stefan conditions for individual components along with the Gibbs-Duhem condition assuming local equilibrium at the interface. A height-function based curvature technique was used to accurately estimate the curvature of the sharp interface. In order to achieve a nearly grid-independent growth in any specified crystallographic direction, a decentered-cube algorithm was incorporated. The model validation was performed by comparing the velocity and tip radius selection with analytical relations based on the marginal stability criterion as well as simulations results obtained using the phase-field method. Further, the code was parallelised using MPI libraries, thus enabling it to be run on larger domains. Multi-dendrite simulations were carried out under directional solidification corresponding to AM conditions. The dendrite morphology, orientation and primary arm spacing selection were studied for different cooling rates and thermal gradients. This model can be used to simulate microstructure evolution for generic two-phase alloys under AM conditions.

A set of single track laser melting experiments was performed in a selective laser melting (SLM). The tracks were done on an Inconel 718 plate with various laser scan velocities at a constant laser power of 150 W. The geometries of the molten pool (MP), as well as the solidified dendrite structures, i.e., primary and secondary dendrite arm spacing (PDAS and SDAS), in the cross sections of the molten path were characterized to evaluate the effect of the laser scan velocity during SLM. Moreover, the local solidification thermal conditions (cooling rate R*, tip growth velocity V* and temperature gradient G*) at the MP bottom were deduced from the SDAS and the geometries of the molten pool. Finally, the mesoscopic envelope model was used to simulate the PDAS selection of the columnar dendrite growth in the molten pool. The simulated results were compared with the experimental data, and a good agreement was achieved under different laser scan velocities.

Additive manufacturing has been regarded as a highly potential method to produce metallic parts with complex geometry. Its basic process in fact is laser melting and the fast solidification afterward. Controlling such process or tailoring the solidification structure of the metallic samples fabricated via additive manufacturing is therefore the control of that fast solidification process. This is also the researchers desire to be realized. Several methods have been proposed such as rolling after several layers’ deposited, adding grain refiner and applying ultrasound. They need to contact the sample or at least add another elements. Applying magnetic fields may give a method to affect the fast solidification process during additive manufacturing without any contact or elements added. In this work, Al-10wt.%Si, Inconel 718 Nickel based and Ti6Al4V alloys have been fabricated by additive manufacturing under a static magnetic field. The results show that all of their solidification structure has been affected by the magnetic field. In details, the dendritic growth tendency of primary phase was enhanced and the depth of each melting pool was decrease. These observations can be attributed to the results of compaction between the damping and accelerating (the thermoelectric magnetic flows) effect of static magnetic field on the melts.

Al-Mn-Fe intermetallics are important for the corrosion resistance of cast Mg alloys but there have been few studies on their solidification mechanisms or microstructure development. Here EBSD, deep etching and FIB-tomography are combined to study the nucleation and equaixed growth crystallography of B2-Al(Mn,Fe), Al8Mn5, and Al11Mn4 in Mg-Al-Mn-based alloys. It is shown that twinning readily occurs during the solidification of all three compounds. In the case of rhombohedral Al8Mn5, the twinned particles consist of four orientations related by ~90° rotations around three common 〈1-102〉, which is discussed in terms of the pseudo-cubic <100> axes of the Al8Mn5 rhombohedral gamma brass. Under a wide range of solidification conditions, Al8Mn5 grew as equiaxed polyhedra that are explained by polyhedron models based on different combinations of {100}, {110} and {112} facets using a pseudo-cubic Al8Mn5 cell. It is shown that twinning also plays a role in the preferred orientation relationships that develop between the different Al-Mn-Fe compounds. Reproducible orientation relationships (ORs) and the lattice match between the phases are discussed in terms of the group-subgroup relationship between these phases. The role of solute rejection on the continuous nucleation of Al-Mn-Fe particles during cooling in a near-uniform thermal field and on the final number density of Al-Mn-Fe intermetallics is then discussed based on analysis of synchrotron radiogaraphy image sequences.

In aluminium alloys, iron is a common impurity as it is unavoidably picked up in practice. The excessive Fe is strongly prone to form various intermetallic phases. These Fe-rich intermetallics are generally brittle and act as stress raisers to weaken the coherence with Al matrix, therefore decreasing elongation. However, Fe addition in Al-Mg alloys may be beneficial because of the improvement in the yield strength with the scarification of ductility of die-cast aluminium alloys. The morphology of intermetallic phases has a vital effect on the properties of aluminium alloys. In the present work, the 3D morphology of Al6 (Fe, Mn) in diecast Al-Mg-Mn alloys with different levels of Fe contents were revealed. The formation of Al6 (Fe, Mn) was also studied through crystal features and solidification behaviours.

This paper reports on the development of Solidification Continuous Cooling Transformation diagrams that relate the solidification paths to the inherent microstructures of binary and ternary alloys. The methodology is based on the quantification of a solidified microstructure for its various phase fractions. This measured data is combined with well-established solidification models and phase diagrams to yield undercooling temperatures of individual phases. The thermal history and undercooling of different phases in the solidified alloy are estimated for a wide range of cooling rates (from 10-2 °Cs-1 to 105 °Cs-1). To describe the methodology, dedicated samples of Al-Cu, Al-Cu-Sc and Al-Si alloys were studied. With said alloys being solidified in a controlled manner, over a wide range of cooling rates and undercoolings, via Impulse Atomization, Electro-Magnetic Levitation, and Differential Scanning Calorimetry

The solidification path and the σ-phase precipitation mechanism of UNS S31254 alloy were studied on the basis of directional solidified experiments accompanied by scanning electron microscopy observations and energy dispersive X-ray a nalysis. The resulting temperatures of solidification paths and phase transformation were compared with Gulliver-Scheil and equilibrium calculations predicted using ThermoCalc© software. It was confirmed that the experimental solidification path was in agreement with the thermodynamic calculations. The complementarity of the results have made it possible to propose a solidification path and a σ-phase precipitation mechanism for the UNS31254 steel.

This study aims in combining the material properties and numerical modelling techniques through practical application to provide understanding of surface edge cracking caused by V(C,N) precipitation to enhance the yield of the continuous bloom caster at ArcelorMittal Duisburg. The investigation for this work is carried out on three different micro-alloyed steel grades; one of them being the most crack sensitive 20MnV6. A process model is used which calculates the solidification process of the strand to design an optimum cooling strategy. Therefore, two cooling patterns are employed for the steel grade 20MnV6. Reduction of area values of the steel grades 20MnV6, 27MnSiVS6 and 38MnSiVS5 evaluated from hot tensile tests using a Gleeble simulator have been used as an indication in assessing steel's cracking behaviour. Further precipitates have been analysed by SEM at ArcelorMittal. These laboratory results suggest that precipitation kinetics of V(C,N) influences the crack sensitivity of the micro-alloyed steel. The software MatCalc® is used to simulate precipitation for process parameters of continuously cast blooms at Arcelor Mittal Duisburg as well as the parameters of the Gleeble experiments. From these simulations the Zener pinning force (ZPF), resulting from V(C,N) particles on grain boundaries, is evaluated which can be used as a measure for the crack sensitivity. Using the values of the ZPF it is possible to identify the set of casting parameters of the steel grade 20MnV6 which lead to the minimum crack intensity on the surface of the blooms. Moreover it is possible to make a ranking list with respect to the ductility drop occurring in the Gleeble experiments for the analysed steel grades. The proposed statement is an issue for a continuing investigation of precipitation modelling.

Metal matrix composites (MMCs) can be tailored to produce various combinations of stiffness and strength and have found a wide range of applications where existing materials are not suitable for use in automotive and aerospace industries and many other areas. Stir casting is commonly used for fabricating large quantity of primary MMCs, in comparison to liquid metal infiltration and powder metallurgy, which are methods for making near net-shape MMC components or small quantity MMCs. Stir casting is cost-effective but suffers from severe particle agglomeration, particularly when the size of reinforced particles is down to nanoscale. The present work was carried out to investigate the effect of high shear treatment on the distribution of reinforcing particles in the fabrication of magnesium and aluminium alloy matrix particulate nanocomposites. An Mg-2Zn-0.5Ca alloy and commercially pure aluminium was selected as the matrix. Hydroxyapatite nanoparticles (~50nm, spherical, 1-5wt%) were added to the magnesium alloy as reinforcing elements and alumina nanoparticles (~30nm, spherical, 1-3wt%) to aluminium. The high shear treatment was employed after the admission of reinforcing particles by mechanical stir and performed with a rotor-stator device at a speed of ~5000rpm. Experimental results showed that the high shear treatment effectively reduced particle agglomeration for both Mg/HA and Al/Al2O3 nanocomposites, although the features of particle segregation are different, probably due to difference in wettability between the matrix and reinforcing elements in these two systems. The as-cast Mg/HA and Al/Al2O3 nanocomposites were deformed by hot extrusion and cold rolling respectively. The microstructure and mechanical properties of the material were characterized and the effect of processing parameters was studied.

The simultaneous prediction of macrosegregation and columnar to equiaxed transition (CET) in metal casting is still an important challenge. One of the open questions is the role of melt convection on the CET and the effect of the CET on macrosegregation. A three-phase Eulerian volume-averaged model for macrosegregation and CET is developed. The model uses our recently-developed accurate constitutive relations for equiaxed solidification. It accounts for columnar dendrite tip undercooling and nucleation of equiaxed grains ahead of the columnar tips. The model is used to perform macrosegregation and CET simulations for a numerical benchmark problem involving solidification of a lead-18 wt. pct. tin alloy in a side-cooled cavity. Simulations are performed for different values of the dendrite tip selection and nucleation parameters. As the value of the dendrite tip selection parameter is increased, the predicted number and depth of channel segregates decreases significantly and the grain structure transitions from mixed columnar/equiaxed to fully columnar.

Since last century, two and three-phase models are generally adopted for the meso-scale simulations of the growth of equiaxed grains in solidification of binary and ternary alloys [1]-[3]. Yet, as far as three-phase models are concerned, their implementation in the macro-scale model with the equiaxed grains transport can be realised in different ways. Indeed, since the solid fraction of a grain and its envelope are attached and transported together, they share the same velocity field. Furthermore, different implementation of the multiphase meso-scale models into macro-scale models implies different treatment of the fluid flow and the macroscale segregation. In the present work advantages and problems adherent to different models are discussed and some results obtained for macrosegregation are compared. Acknowledgements This work is a joint cooperation between SIMAP laboratory of Grenoble INP (France) and Key Laboratory of EPM of Northeastern University (P. R. China). The authors gratefully acknowledge financial support from China Scholarship Council (no. 201706080074), the Program of Introducing Talents of Discipline to Universities (The 111 Project of China, Grant No. B07015) and ESA-MAP project MICAST (contract number ESA-AO-99-031). References 1 M. Rappaz and Ph. Thevoz, Acta metall., 35 (1987), 1487-1497 2 C.Y.Wang and C.Beckermann, Mat. Sci. Engineering, A171 (1993), 199-211 3 M.Rappaz and W.J.Boettinger, Acta mater., 47 (1999), 3205-3219

A three-phase equiaxed solidification model where macroscale heat transfer and fluid flow are coupled with microscale nucleation and dendrite growth, is applied to the simulation of the macrosegregation in binary alloy solidification subjected to the electromagnetic stirring. The investigated experimental solidification case is conducted in a cavity which has a good control of the thermal boundary conditions. The proposed model uses a double time step scheme to accelerate the solution. Electromagnetic force is introduced as a source term into momentum equation in analytical form. To account for the friction from the side walls, a 2D½ flow model is applied to a three-dimensional experimental configuration. A comparison between the results of simulation and experimental ones is made. Acknowledgements ---------------- This work is a joint cooperation between SIMAP laboratory of Grenoble INP (France) and Key Laboratory of EPM of Northeastern University (P. R. China). The authors gratefully acknowledge financial support from China Scholarship Council (no. 201706080074), National Key R&D Program of China (Grant No. 2017YFE0107900) and National Nature Science Foundation of China (Grant No. U1760206). References ---------- 1. A.I. Ciobanas, Y. Fautrelle: J. Phys. D. Appl. Phys., 40(12) (2007) 3733–3762 2. C.Y. Wang, C. Beckermann: Metall. Mater. Trans. A, 27(9) (1996) 2754–2764

During casting, diffusion phenomena are often simplified by Lever-Rule or Gulliver-Scheil hypotheses, thus simplifying chemical diffusion to extreme configurations. The present work proposes an extension of the Tong-Beckermann microsegregation model to multi-component alloys while considering finite diffusion in liquid and solid phases plus the effect of tip undercooling of the columnar front. The behaviour of this model is studied according the solidification conditions (growth velocity and thermal gradient) and comparison with microprobe measurements is proposed for a seven components nickel-base superalloy.

An efficient model for the prediction of dendritic grain growth is developed coupling the lattice Boltzmann method for solving the transport of solute and a cellular automaton algorithm for determining the evolution of grains’ envelope and the release of solute during phase change. In contrast to solving equations from the field of continuum mechanics the new model is more related to particular occasions what is more similar to the behaviour of cellular automaton algorithms. The resulting dendritic grain growth shows qualitative correctness, although the consideration of solute conservation is still missing. It is shown that neglecting proposed conditions regarding the choice of time step size can destabilize the solid-liquid interface resulting in secondary and ternary dendrite arms.

A thermo-mechanical model of powder bed fusion by laser beam is developed by the finite element method at the macro-scale. This approach focuses on the part construction, in which a level-set framework is adopted to track first the interface between the constructed part and the non-exposed powder, and second, the interface between gas and the successive powder bed layers. The accounting for the non-irradiated powder bed in the model allows an improved heat transfer resolution. In order to reach reasonable simulation time for industrial parts, the energy input and the formation of the additive deposit are simplified by considering them at the scale of fractions of layer, an entire layer or a so-called “super-layer”, that is a set of several layers. In each simulation strategy, the three stages of heating, cooling and the dwell time between two successive layers are included in the process modeling. In addition, the mechanical analysis following the thermal analysis is performed in each time step during all the stages of the construction process. At the end of the construction, the residual stress and distortion of the constructed part are presented. In all three strategies, mesh adaptation is employed, offering a good compromise between accuracy and sustainable computation time. Quantitative comparison between the three approaches will be presented and discussed.

Interest has recently emerged for the manufacture of aeronautical parts by Laser Beam Melting (LBM) additive process. This energy efficient process can for instance be used to build complex geometries, which cannot be made with traditional processes. However, complex phenomena occur during powder melting and track development : vaporisation phenomena influence laser-matter interaction by creating metal vapours that are responsible for the reduction of absorbed energy. The recoil pressure generated by the vaporisation counteracts the surface tension between the melt pool and the inert gas, also inducing liquid instabilities. The study of laser-matter interaction and induced phenomena can help understand the origin of defects such as porosities or cracks. In this approach, a level-set modelling of the LBM process at a mesoscopic scale is proposed to follow melt pool evolution and track development during build. A volume heat source model is used for laser/powder interaction considering the material absorption coefficient. A surface heat source is used to take into account the high laser energy absorption by dense metal alloys. An energy solver is coupled with thermodynamic database and pre-determined solidification path. Shrinkage during consolidation from powder to liquid and compact medium is modelled by a compressible Newtonian constitutive law. An automatic remeshing adaptation is also used to save time and avoid high computational cost. In the future, the computation of multiple beads or the build of a wall in a context of lattice structures will have to be considered.

Selective Laser Melting (SLM), the most popular metal additive manufacturing (AM) process, is well suited for making complicated parts which are difficult to manufacture by conventional manufacturing techniques. Currently, the main bottlenecks inhibiting the usage of the Selective Laser Melting (SLM) parts include the problems, such as porosity, low resolution, low surface finish quality and low build rate. In order to overcome the aforesaid problems, latest SLM machines are now being equipped with laser having small spot radius for enhanced resolution and surface finish, and high power to increase the build rate. The combination of high power and small spot radius leads to high energy density, exceeding the threshold value, resulting in transition of melting mode in the SLM process from conduction mode to keyhole mode and a formation of porosity due to collapsing of keyhole. In this study, high fidelity particle scale model is developed using open-source codes LIGGGHTS and OpenFOAM to understand the formation of porosity and to describe the physical phenomena (convection, melting, evaporation and solidification), melt flow dynamics and melting mode transition occurring in the SLM process.

Powder bed fusion processes in additive manufacture (AM) of metals typically involve solidification in high thermal gradients, leading to unwanted anisotropic columnar growth and high residual stress in as-printed components. Equiaxed solidification can alleviate such problems but is difficult to achieve due to the thermal fields naturally evolving in typical AM processes. Pre-heating of the powder bed in electron beam-based processes can reduce the thermal gradient and, together with grain refinement of the alloys, this encourages equiaxed solidification. However there is a material-dependent possibility of sintering of loose powders if this approach is taken which makes it more difficult to recycle un-used powder after print runs. The process is quite complicated, experimental trials are expensive and can be carried out only within limited ranges of specific machine parameters, and in-process experimental measurements are difficult. Computer simulation and modelling is therefore a key tool in investigating possible ways of creating conditions suitable for equiaxed solidification. A multi-scale modelling approach is presented which will predict columnar and equiaxed solidification and the columnar-to-equiaxed transition, using front-tracking at the scale of grains to simulate the non-equilibrium growth of dendritic fronts and grain envelopes, with phase field formulations for the micro-scale physics of dendritic growth. Preliminary results are outlined.

Understanding the dynamic evolution of primary dendritic spacing in the laser melt pool is significant from a technological viewpoint because primary spacing is one of the foremost parameters that control the final mechanical properties of additive manufactured products. In this work, a multi-scale computational framework that couples FEM and a developed quantitative phase field method is employed to simulate the evolution of microstructure and primary spacing of a nickel-based superalloy during wire and laser additive manufacturing (WLAM) solidification. Transient conditions in the laser melt pool are considered in which both temperature gradient G and solidification speed VP are made time-dependent. Through the use of this model, the dendritic morphology, tip velocity and spacing evolution during the solidification are investigated to provide the relationship between the laser processing parameters and the final spacing. Moreover, we attempted to clarify the intrinsic mechanism of spacing adjustment under different laser processing parameters from a novel perspective. This work provides meaningful understanding of spacing evolution in nickel-based superalloy and demonstrates the potential of controlling the complex microstructure morphologies and final primary spacing during wire and laser additive manufacturing process.

Laser metal deposition is an additive manufacturing technique for fabricating complex metal components layer by layer. The complex thermal behavior during this process results in the complex microstructure evolution, which directly affects the final mechanical properties of the products. Numerical modeling offers a cost efficient way to better understand the related complex physics in laser metal deposition process. It helps to reveal the effects of processing parameters on the desired characteristics of deposition parts. In this work, a heat transfer finite element model is coupled with a multi-phase-field model to predict the thermal behavior and solidification microstructure evolution during laser metal deposition process. The thermal behavior during single-track and multi-track deposition process was simulated numerically by using a three-dimensional transient finite-element model, where deposition of material was modeled through activation of a new set of elements within each solution step. The deposition geometry was well predicted without assuming a prior shape. The influences of the scanning speed and laser power on the morphology and dimensions of molten pool were investigated. It is found that the molten pool height decreases with the increase in the scanning speed, while the increase of laser power results in the increase of molten pool size. The temperature history extracted from the macro simulation was then transferred to a micro region inside the mushy zone of the molten pool, where dendrite growth during solidification was simulated by the multi-phase-field model. The effects of several process parameters on the solidification microstructures were investigated. Directional dendritic growth from the bottom of the pool was observed with various dendrite arm spacing and orientation depending on the location in the pool. The microstructure and the value of dendrite arm spacing obtained in simulation agree well with previous experimental observation.

A mathematical model has been established to coupling predict the formation of inverse segregation and porosity for up-vertical unidirectional solidification of Al-Cu alloys. Based on the analysis on the redistribution behaviours of both gas element and the alloying elements in the mushy zone. The model first investigates the volume change during solidification intensively, including the specific mass variation and the solidification shrinkage both during the growth of primary phase and eutectic reaction, to derive the back flow needed. Then, the fraction of porosity is obtained by combing the effects of hydrogen segregation and the pressure depression associated with the feeding liquid flow through a porous media and being solved with Darcy’s equation. Consequently, both the feeding flow and the well known 'local solute redistribution equation' are modified with the presence of porosity. Finally, the solute distribution is obtained by coupling a finite difference solidification model with the modified segregation model. Numerical results show that the overall back-flow, which is needed to compensate the volume shrinkage, decreases with an increasing initially dissolved hydrogen concentration. As a result, the inverse segregation, being induced mainly by back-flow which is generally rich in solute, decreases with an increasing amount of porosity. The calculating results are also compared with the experimental results reported in literatures, which shows rather good agreement. The model can be used to predict inverse segregation with the presence of porosity in vertically unidirectionally solidified Al-Cu alloys.

The alloy casting process is one of the major manufacturing processes to produce near net shape components. The casing process is prone to a wide variety of defects, with hot tear being one of the most detrimental. The two main factors generally recognized as the primary cause for formation of hot tears are the mechanical response of the mush (which effects its permeability), and the solidification range (solidification time). The response of the mushy zone under deformation is mainly affected by the solid fraction, strain rate and grain morphology. Even though the science behind the formation of hot tear is understood, there is no general criterion to quantify the hot tear formation under varying casting conditions. The development of ultra-fast X-ray imaging has facilitated the means to quantify the effects of the critical parameters in-situ and develop better correlations for hot tear prediction. The in situ experiments will also provide insights into mush rheology, which has significant influence on hot tear formation. In this study, isothermal semi solid compression studies of Al-Si-Cu alloys were carried out using specially built thermo-mechanical rig. We studied the effects of the strain rate in the range of 2×10-4– 0.02/s and solid fraction (~0.6-0.9) on the mechanical response of the mushy zone. The sample were characterized before and after deformation using X-ray micro tomography. The data was subjected to an image processing routine and the amount of porosity and hot tear was quantified. The stress-strain curve of the semisolid alloys showed a characteristic strain softening behaviour for semi solid samples with ~0.6-0.7 solid fraction, irrespective of loading rates, whereas the behaviour at higher fractions were that of constant flow stress. Additionally, in situ compression experiments were carried out, wherein the liquid channel thickness at various strain values were measured. Isolated liquid channels were formed under loading, from where the hot tears were found to nucleate. Hot tear susceptibility was found to increase with increasing strain rate and rheology of the mush, which is dependent on solid fraction.

In the present study, effect of cooling rate on the formation of the porosity in the thick aluminum sand casting was investigated. Nowadays large scale thick aluminum casting replaces steel frame for vacuum chamber for semiconductor production, with the consideration of weight and cost reduction. Several thick aluminum castings were manufactured using chill with temperature measurements. The castings were inspected by using 3D computed tomography in order to quantify the porosity defect density in the castings. Effect of the thickness of the chill on the porosity defect density were discussed.

Correct prediction of composition heterogeneities and grain structure across a steel ingot is critical in optimizing the industrial processing parameters for enhanced performance. The columnar to equiaxed transtion (CET) is a microstructural transition which is strictly controlled as it affects the mechanical properties of the final product along with the macrosegregation patterns. Larger equiaxed regions are preferred for most industrial applications. CET is significantly affected by the number density of equiaxed grains and by the nucleation undercooling. 8 kg 42CrMo4 alloy steel ingots (240 mm x 60 mm x 60 mm) were cast. The cast structure was characterized by ASCOMETAL. The experiments were simulated with a process-scale model of solidification that incorporates a multiscale description of the microstructure formation. The goal of the present study is to show the capabilities of such a process-scale solidification model to explain the observed structure distributions (extent of the columnar and equiaxed zones, equiaxed-to-columnar transition).

Al-Mn alloys (3003) alloys are the preferred alloy series for brazed heat-exchangers for applications in the automotive industry. Cu, Si and other alloying elements are added in different amounts to improve alloy properties and achieve excellent thermal efficiency, high strength and corrosion resistance. Addition of Cu and Si results in longer solidification intervals which makes the alloys susceptible to hot-tearing and hard to cast. To investigate the influence of Cu and Si on the hot-tearing susceptibility, different amounts of these alloying elements were added to commercial alloys and a series of hot-tearing experiments were performed. Analysis of the experimental results were combined with theoretical analysis including hot-tearing indicators and microstructural simulations. An increase in the Cu and Si contents lead to increase in the hot-tearing tendency until a maximum was reached. A further increase decreased the hot-tearing tendency. The maximum is found at different concentrations for the alloys investigated and depend on the interplay between all alloying elements as well as the cooling rate.

The formation of coarse Fe and Mn containing intermetallic particles during the DC casting of Al-Mn-Ti-Fe-Si alloys is detrimental to the alloys' cracking susceptibility, formability and even final mechanical/corrosion properties. In this talk, SEM/TEM are employed to reveal the microstructural features of the particles including composition, size distributions and morphologies. These experimental measurements are then compared with the predictions from Phase field method and Frequency Distribution Function method (implemented in our in-house software PreciMS). The combination of modeling and experimental approaches is able to shed some lights on the origins of these coarse particles. It is expected that this work is useful for the heat exchanger alloys composition design and casting parameters optimization.

The carbide precipitation nature of an automobile gear material, i.e. Fe-C-Mn-Si-Cr-Mo alloy, during carburization and element addition process are investigated through PE+PA+LR prediction. Results show that, carburization process greatly increases the amount of cementite as well as the hardness index at the surface part. Besides, the addition of Ti and V helps to formation of TiC and (V,Mo)C carbides but suppress the precipitation of cementite, which contributes equivalent hardness at surface and lighter weight to the alloy.

γ-TiAl alloys are regarded as promising candidates in aero and automobile industry due to their superior properties including low density, high specific strength, excellent oxidation resistance and creep strength. However, the application of these alloys is limited by the problems such as micro-segregations, inhomogeneous microstructure, texture and severe cracking tendency. In the article, the effect of thermal cycling in the mushy zone and solidification rates on the microstructure of Ti-48Al-2Cr-2Nb alloy and Ti-45Al-8.5Nb-(W,B,Y) alloy were firstly discussed. Then, the phase transformation behaviors and microstructure evolution of Ti-45Al-8.5Nb-(W,B,Y) alloy was investigated. And based on phase transition sequence, a novel multi-step isothermal treatment (TIT) during solidification process was proposed to control the microstructure of Ti-45Al-8.5Nb-(W,B,Y) alloy and the influence of TIT on lamellar structure, homogeneity, texture and cracking was analyzed.

The current work, describes a numerical methodology to obtain deeper understanding of the kinetics of solidification and the dynamics of Al melt infiltration into porous iron preforms, in order to develop a near net shape process for a new class of highly advanced ductile and fine grained Fe-Al intermetallic. Investigations on macroscale simulations revealed that certain pressure and wetting conditions are beneficial for infiltration into finer structures. Microstructure showed a dependency of the solidification process on diffusion. Estimates for infiltration and solidification times are developed to determine which structures can be infiltrated before solidification stops melt flow.

Macrosegregation refers to the inhomogeneous distribution of alloy elements at the macro scale in solidified castings. It is a common defect in direct chill (DC) casting of aluminium alloys. Since various transport phenomena contribute to the formation of macrosegregation, numerical modelling is a valuable tool to assess their role and study their interplay with process parameters. One such parameter is inlet induced flow and consequently the effect of forced convection on liquid flow and moving solid grains. This complex and coupled flow phenomena cannot be addressed using a 2D simulation of sheet ingots. In this work, we conduct 3D modelling case studies on aluminium sheet ingots to analyze the inlet induced forced convection and its impact on macrosegregation formation. To carry out this investigation, we use a multiphase, multiscale solidification model also accounting for thermal-solutal convection, solidification shrinkage, and equiaxed grain nucleation, growth and transport. We show that forced convection alters significantly grain transport and sedimentation and that macrosegregation formation can be altered by varying the intensity of inlet flow.

The control of the carbon macrosegregation level in steel ingots is important for the structural integrity of the nal component. During solidication, the fragmentation of the columnar dendrites is an important source of equiaxed grains, and has a large in uence on the macrosegregation and the grain structure. The goal of this study is to show that a numerical model that takes into account fragmentation can describe the formation of the structures and the macrosegregation during solidication of a large steel ingot. The present article describes the multiphase numerical model used in the simulations. The simulation results are compared to experimental data from a 9.8 t ingot cast in A5/6 steel by ArcelorMittal Industeel. The model can then be used to explain the formation of the observed structures. For example, we show that the structural transitions, rst from columnar to equiaxed globular and then to equiaxed dendritic at the bottom of the ingot are a consequence of the concurrent transport and growth of the dendrite fragments from the columnar zone. Furthermore, we show that most of the structures are formed very early on during solidication, whereas macrosegregation develops much more gradually

The formation of casting defects, including macrosegregation, shrinkage cavity, porosity, and inclusion, which resulting from the combined action of multi-physics filed, is a complex process. This process including the nucleation, growth, floating (or settling) of both inclusion and equiaxed crystals; growth and fragmentation of dendritic crystals; solidification shrinkage; solute redistribution and migration; etc. Therefore, it is a big challenge to predict these defects together in one model. Based on the dendritic-equiaxed & columnar macrosegregation model, a four-phase solidification model that further consider the gas phase, which supplement the volume reduction of solidification shrinkage, has been established to realize the prediction of shrinkage cavity in metal solidification process. For the prediction of inclusion, in order to distinguish the difference between exogenous and endogenous inclusions, two models had been established respectively: 1) for exogenous inclusions, the coupling of Discrete Phase Model (DPM) and four-phase solidification has been considered, which has the possibility to track the moving of inclusion particle; 2) for endogenous inclusions, inclusion-combined macrosegregation model, which coupling the inclusion growth theory with the multicomponent four-phase solidification model, has been established. The formation process of defects in steel ingots had been investigated by these multi-defects models, which provide reasonable results by compared with experiment results. The interaction behaviors and its mechanisms between shrinkage cavity, inclusion and macrosegregation had been deeply studied.

Within the framework of the ESA GRADECET project, experiments of directional solidication of cylindrical Ti-Al samples were conducted in hypergravity. The experiments were performed in a centrifuge with the apparent gravity (sum of centrifugal and terrestrial gravity) aligned along the cylinder centerline. 3D numerical simulations of aluminum macrosegregation in these samples are presented. A volume-averaging solidication model is used that accounts for centrifugal and Coriolis accelerations in a non-inertial rotating reference system. We compare the melt ow pattern and the macrosegregation formation under terrestrial gravity and under 20~g centrifugation. The results show that the Coriolis acceleration, although very weak, breaks the symmetry of the thermosolutal convection, having an important impact on the nal macrosegregation pattern. The macrosegregation is entirely modied in comparison with a sample solidied under terrestrial gravity conditions. Besides the aluminum segregation intensity increases with the centrifugation level.

In this study, 3D flow fields, sump profiles and macrosegregation patterns are predicted for HDC slabs of an Al–4.5 wt.% Cu alloy. Numerical simulations are used to provide initial estimates of these features and the influence of slab geometry for a constant inlet configuration. In general, the heat conduction across the slab thickness controls the sump depth and general macrosegregation profile in the slab. As a result, the slab width does not significantly influence the qualitative sump shape or macrosegregation features and a 2D model is sufficient for establishing processing trends in HDC.

One of the most powerful tools for studying the development of microstructures is the CALPHAD (Calculated PHAse Diagrams) method because it comprises information derived from thermodynamic principles, presented in a form that makes the data readily accessible. It is much easier and less time consuming to perform these thermodynamic calculations with any of the available software rather than trying to experimentally identify the phase transformations during solidification, but one has to be sure that the temperatures calculated are close to the real system. Even when the thermodynamics tells us what should happen in equilibrium, it does not guarantee that it is what will happen, as the real system may be influenced by kinetics. However, when dealing with complex systems like commercial grades of steel, where many phases may be formed during solidification through various reactions, it is necessary to perform more complex calculations. In this publication two examples will be shown on how the CALPHAD method is used to evaluate the castability of commercial steel grades where complex phases appear during solidification and cooling. The first example will focus on the different phases that can appear in HSLA steels (High Strength Low Alloyed) where the microalloying elements (Nb, V, Ti, Al, etc.) that precipitate as carbo-nitrides at different stages during and just-after solidification are related to the so-called second ductility trough and may cause cracking at the last stages of the casting process. Further, the second case show the possible castability issues for AHSS (Advanced High Strength Steels) that are related to the phase transformation and how the CALPHAD method can help on designing new steels.

**Abstract:** A virtual submesh cellular automata method is established to reduce the computational mesh induced artificial anisotropy with high efficiency and the method is applied to simulate the microstructure evolution during laser melting additive manufacturing. The additive manufacturing process is divided into three parts during simulation: random powder deposition, laser melting and rapid solidification. The laser induced temperature field will be calculated by finite difference method and a uniform solute concentration filed will be used for the non-equilibrium solidification. The Lipton-Glicksman-Kurz and Kurz-Giovanola-Trivedi model will be used respectively for the growth of the molten pool boundary and grains nucleated in the bulk. In order to compare with the experiment, the simulation will be implemented with the pulse laser source and the nickel base superalloy. The single crystal substrate was used and the processing parameter to produce single crystal part was predicted by simulation. The final microstructure of parts fabricated by the laser melting additive manufacturing has compared with the simulation result, which shows reasonable agreement. **Keywords:** modified cellular automata simulation, laser additive manufacturing, artificial anisotropy, single crystal prediction

The targets of this study were to determine the effect of vertical location and the composition of inclusions to the occurrence of oxide–sulfide stringers, and to determine the calcium aluminate phases most prone to form stringers during hot rolling of aluminium killed, calcium treated steel. The phases present in the inclusions have a significant effect on the deformation of inclusions during hot rolling, and consequently, on mechanical properties of the steel. A MATLAB script is utilized to identify and locate detrimental stringers from the hot rolled plates. Inclusion analysis data gathered with a scanning electron microscope and exported from IncaFeature software is analyzed. The following properties are presented for each stringer: the number of inclusions and length of stringer, phase fractions and compositions, and the composition of the unfragmented inclusion before hot rolling. According to the results, the longest stringers have total lengths over 200 µm, with almost 20 inclusions. The overall composition of the longest stringers is between C12A7 and C3A calcium aluminates with minor MgO contents. The diameters of the unfragmented inclusions in the slabs, forming stringers during hot rolling, were estimated to be around 20 µm for the longest stringers. From the dataset, plenty of CaO–CaS stringers were also characterized, obviously a result of excess calcium treatment.

Computational process modelling has become an important engineering tool in the casting industry to predict the solidification sequence in complex castings. Used properly, this tool can help reduce manufacturing costs. One of the challenging issues in developing casting simulations of the low pressure die casting (LPDC) process for automotive wheels is to quantify the heat transfer coefficients (HTC) within the cooling channels in a die. When water is used as the cooling media, the HTCs exhibit a complex, non-linear behaviour due to the boiling phenomena that occur making it possible to extract a significant amount of heat from the die in a short period of time and influence the solidification of a wheel. Primarily, constant heat transfer coefficients have been used to describe this heat transfer in casting models up until now, but an opportunity exists to improve the transient description of heat transfer in channels cooled with water. In this paper, HTC’s in a lab-scale physical analogue model of die cooling will be characterized as a function of initial die temperatures.

A method for detecting sticker breakouts in continuous casters is proposed based on extensive feature engineering and advanced analytical modelling. It is observed that the temperature patterns generated during a sticker breakout have a consistent signature, owing to the underlying physical processes involved. The aim of the present exercise is to capture this latent physical phenomenon in a reliable and robust way. The temperatures measured by thermocouples in the mold are used to extract features of physical significance and operational importance. The features were custom built iteratively to capture the difference between true and false patterns. These features are consequently used to develop a Gradient Boosting model to detect sticker breakout. The model is trained on previously raised sticker alarms that are manually tagged as true or false alarms for a time period of around $3$ years ($1500$ alarm files). The sample space of the non-alarm (tagged $0$) is increased by using data collected during normal operation. The GBM model shows an overall superior performance as compared to the existing in-place logic. While ensuring that no true alarms are missed, the GBM model reduces the false alarms by around $80$%. The reduction in false alarms imply a huge production advantage while the feature engineering involved in the modelling process makes this a unique exercise in the realm of continuous casting research. The data used in the analysis was generated in continuous casters operating in Tata Steel Jamshedpur.

Asymmetric flow of the molten steel in a continuous slab casting mold may be detrimental to the quality of the solidified end product. Depending on the severity of the biased flow, local mold powder entrainment, non-homogeneous solidification around the perimeter of the initial shell or non-optimal inclusion seclusion may occur. In particular, nozzle clogging in the last slabs before an SEN exchange may cause strong and asymmetries in the upper regions of the mold. To avoid costly downgrades of steel quality in these slabs, it is vital to maintain stable and symmetric fluid flow conditions in the mold. The EMBR and the FC Mold are two flexible electromagnetic devices able to produce asymmetric braking/stirring along the width of the mold in slab casting, and in this way allow counteraction against e.g. biased mold flow or local excessive flow speeds. Numerical computations of mold fluid flow and magnetic flux have been carried out to quantify the required fields to symmetrize biased flow scenarios caused by e.g. SEN clogging. In conjunction with steel plant trial feedback, the simulation results have been used to setup control algorithms for the EMBR and FC Mold.

As nonmetallic inclusions (NMI) are unavoidable in steel production and commonly have a negative impact on product quality. Metallurgical research has always aimed at either the adjustment of inclusion morphology or inclusion separation to minimize their adverse effects. In the 1990s the conscious nucleation of particles to either promote or inhibit the subsequent growth of other phases became a new research issue. Today, inclusions are successfully used as grain refiner in the solidification of ferritic and austenitic steels providing that their size, number, morphology and chemical composition are precisely adjusted. Ce2O3 and AlCeO3 are considered active to trigger a fine grained solidification structure for many years now. Depending on the introduced amount of Ce the inclusion type present changes. To gain a better understanding for the conditions that lead to these favorable particles and evaluate the reliability of thermodynamic data, the following poster is presenting theoretical considerations based on results from a ChemApp–based kinetic inclusion prediction model that allows the quantification of the influence of chemistry taking into account microsegregation. These results are complemented by laboratory experiments in a vertical tube furnace, where Ce and Al were introduced simultaneously to a pure iron melt containing 80 ppm O and 40 ppm S. The comparison of simulation and experiment reveals that the simulation is fit to be used as guidance even if the agreement is not perfect. The deviations can mostly be attributed to the dissolution process of cerium. Nevertheless these results are invaluable for designing and performing future experiments with the goal of specific cerium inclusions and a refined primary solidification structure.

Laser Beam Melting (LBM) processes benefit from significant progress in recent years. Currently, manufacturing of ceramic parts for applications at high temperature in aeronautical industries can be planned. However, understanding of defect formation is required in order to optimize manufacturing strategy. In this work, level-set modelling is proposed to simulate tracks development during LBM processes. Thermo-mechanical solution is performed in both powder and dense domains. Fluid flow is computed considering the surface tension and Marangoni forces. In addition mechanical resolution is achieved to investigate stress evolution in the rear part of the track. Applications are developed on alumina material. The influence of laser power, scanning velocity and physical properties are investigated and discussed. Validations of the heat source model are proposed by comparisons of melt pool dimensions and shapes with experimental measurements. A coherent evolution of the track morphology is shown when varying process parameters or material properties.

Since it is not straightforward to directly observe nucleation at the initial stage of solidification in experiments, investigation from computational approach is strongly desired. In this study, influence of grain refiner in heterogeneous nucleation of undercooled Al melt is investigated by molecular dynamics calculations. Particularly, we focus temperature, size of the refiner, and anisotropy in surface orientation of the grain refiner. It is confirmed that the growth rate of FCC Al at the (0001) plane is much larger than that at the other surfaces at all temperatures calculated in this study. Moreover, epitaxial growth of HCP Al appears on the surface of (101̅ 0) and (112̅ 0) planes at large undercooling temperature.

We have performed the large-scale molecular dynamics simulations for nucleation, solidification and microstructure formation [1] and found various new insights such as the heterogeneity in homogeneous nucleation [2] and decrease of the averaged grain boundary mobility during the grain growth [3]. Although these studies were performed using pure metallic system for simplicity, the most of practical structural materials consists of alloys. Therefore, nucleation and solidification processes for various alloy systems are investigated by molecular dynamics simulation in this study. Regarding the homogeneous nucleation from undercooled melt of Ni-Al alloy, the B2 structure is directly nucleated from the undercooled melt of Ni-50mol%Al, whereas no specific compound is not nucleated but mixture of bcc and fcc phases appeared from Ni-25mol%Al. The heterogeneous nucleation in Al melt inoculated with Ti and Ti3Al particles as grain refiners are performed and the relationship between the particle size and the undercooled temperature needed for the heterogeneous nucleation are discussed. The preferential facet of the hcp Ti surface for the heterogeneous nucleation of Al melt is also investigated. Moreover, interfacial properties such as the solid-liquid interfacial energy and the kinetic coefficient are closely examined for Fe-Cr system by focusing on the concentration dependence. Our cutting-edge results will be introduced in the presentation. [1] Y. Shibuta, et al. JOM 67 (2015) 1793. [2] Y. Shibuta, et al, Nature Comm. 8 (2017) 10. [3] S. Okita, et al., Acta Mater. 153 (2018) 108

Melt conditioning, i.e. shearing of liquid metal using a rotor-stator device, refines the grain structure without the need for grain refiners. Melt conditioning as applied to direct-chill (DC) casting, forming the MC-DC process, has recently been demonstrated to be an effective method of producing high quality light alloy billets. The optimisation of melt conditioning parameters through experiments is expensive and difficult due to the large variability in DC casting and melt-conditioning parameters, diameters of billets, and compositions of the treated alloy melt. In this contribution, we present a new numerical model of melt conditioned direct-chill (MC-DC) casting that considers grain motion and use this model to determine the position of the mixer inside the sump that maximises the temperature gradient across the mushy zone. This numerical model is implemented using the OpenFOAM library. The model is based upon a continuum formulation that avoids the need for tracking phase interfaces, thereby making the model computationally affordable and attractive for optimization studies. The model is validated using temperature and sump profile measurements in conventional DC casting from the literature and in-house measurements in MC-DC cast billets. Turbulence is handled using a Large Eddy Simulation (LES) to accurately resolve the effect of turbulence on grain redistribution; this flow model has been validated using particle image velocimetry (PIV) measurements in a water tank. The optimization search is performed over the global design space using an Evolutionary Algorithm (EA). Melt shearing results in an increased temperature gradient across a shortened slurry zone and a shallower sump, consistent with previous experiments in both Mg and Al alloys. The consequent uniform cooling rate in the slurry zone contributes to a finer, more uniform grain structure in the as-cast billet. The optimization study will be extended to other design parameters (rotation speed, rotor-stator geometry, operating temperature, alloy composition etc.)

The high pressure die casting process is extensively used to manufacture light metal parts with high productivity. A major drawback of the process is the relatively high variability in mechanical properties and poor repeatability between casting cycles, limiting the achievement of weight reduction through lighter design. Although it has been established that mechanical properties are adversely affected by casting defects, the origin of the relatively high randomness in the HPDC process is not well understood. Numerical simulation is a powerful and costeffective tool to address this question, as it gives access to quantities that are difficult to obtain experimentally. A numerical simulation approach based on the finite element casting software ProCAST has been developed. The model was applied to the casting of aluminium tensile test samples, which were used to measure the tensile properties of the alloy. Simulation permitted the study of fluid flow, solidification and defect formation during each stage of the HPDC process: pouring, injection and cooling. Air entrapment and porosity distribution in the cast part were predicted. The results were compared with temperature measurements, porosity observations and solid distribution in the sleeve prior to injection. Although the results are still very preliminary, some trends could be established between the level of turbulence of the melt during injection and reduced elongation.

The accumulation of iron in molten aluminium is one of the main concerns for the recycling and casting industries because it leads to the formation of undesirable Fe-rich intermetallic compounds which are detrimental to mechanical properties. Many methods have been developed in the past to reduce the iron accumulated in molten aluminium scrap, but they all suffer from poor efficiency. Hence, a more efficient method is urgently needed to mitigate the negative effect of high iron levels in the melt, thereby avoiding downgrading secondary aluminium to low quality products or the dilution with expensive primary aluminium. This contribution provides a study of the Fe-rich intermetallic compounds developed in aluminium casting alloys with high levels of Fe as a function of melt processing conditions. Results show that the formation of the Fe-compounds is not only dependent on the cooling rate and holding time before solidification, but more on the initial melt treatment as it enhances the nucleation and growth of the Fe-phases. Elemental addition of Mn leads to the formation of large and compact intermetallic particles, but at slow rate. Physical melt treatment by intensive high shearing produces a much faster nucleation and results in a fine dispersion of smaller iron containing intermetallic particles. The latter could be used either to increase the tolerance to iron contamination or to facilitate the iron removal process, providing huge benefits for the recyclability of scrap aluminium alloys as it would allow the transformation of low-grade feedstock into a low cost and small carbon footprint material for high quality castings.

It has been reported that native MgO particles in Mg alloy melts can act as heterogeneous nucleation substrates such that grain refinement of Mg alloys is achieved. A recent study showed the addition of Ca, combined with the native MgO particles, significantly improves grain refinement of Mg and its alloys. However, the mechanism underlying the grain refining phenomenon is not well understood due to the lack of direct experimental evidence. In this work, we investigated the segregation of Ca atoms at the Mg/MgO interface and its effect on grain refinement in Mg-0.5Ca alloys by utilizing advanced analytical electron microscopy. The experimental results focus on the chemical and structural information at the interface between MgO and the Ca solute. Adsorption layers rich in Al, N and Ca have been detected on {1 1 1} facets of MgO particles, with the lattice structure resembling the structure of MgO. It is suggested that the significant grain refinement improvement can be attributed not only to the growth restriction due to the presence of Ca addition but also to the specific chemistry and structure of the adsorption layers.

Laser Additive Manufacturing (LAM) can directly produce near-net-shape metallic components using commercial alloy powders. However, primary solidification during LAM is far from equilibrium due to the ultra-fast laser-powder interaction (<50 ms), mixing of solutes during melting and micro-segregation upon rapid solidification. Our understanding of these phenomena and the resultant microstructural features formed cannot be fully elucidated using traditional a posteriori characterization, necessitating in situ and operando characterization. We have developed a LAM Process Replicator (LAMPR) that allows real and reciprocal space synchrotron imaging of LAM. The X-ray diffraction study presented exploits the quasi-steady-state and layer-wise AM to enable the detection of the primary solidification of tracks under conditions distant from the near equilibrium transformations observed from conventional processing. This is complemented with in situ and operando x-ray imaging of the same process. The results can be used to help design new alloys and guide the optimisation of processing parameters to exploit the full potential of LAM.

This work focuses on bimetallic casting between A356 alloy melt and profiles of AA6xxx wrought aluminum alloys through a gravity casting process, with the aim to improve the component’s mechanical properties. However, combining two aluminum alloys is difficult due to the stable aluminum oxide present on the surface of the aluminum inserts and the advancing liquid melt front. The oxide layer strongly reduces the wettability between aluminum melt and solid metal. It will also prevent diffusion and formation of a metallurgical bond. In order to obtain sound metallic bonding between the two alloys, different surface treatments, including flux coating and chemical treatments of the profiles have been tested. The influences of preheating temperature and melt flow modes on the quality of the bimetallic casting have been addressed. Based on a detailed microstructure characterization of the bonding layer in the casting, by using Optical Microscopy (OM), Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray Spectroscopy (EDS), the solidification structure development at the interface has been discussed. Results showed that when flux coating was applied, magnesium diffused to the insert surface and prevented formation of a metallurgical bond. Without flux coating, a metallurgical bond was achieved due to slight melting of the insert surface.

DISTEMP solidification numerical model of the continuous casting steel has been a useful tool to explain billet surface and internal quality at Sidenor steel company. Work has been carried out along the years to adjust the appropriate heat transfer coefficients in the mould and secondary cooling to get a reliable description of the CC billet solidification. The numerical model has assisted to explain the influence of casting parameters on the formation of different defects in the CC process as the billet surface cracks for microalloyed steel grades; intercolumnar cracking for high temperature interval steel grades or the formation of central pipe/segregation for high carbon bearing steel grades. The solidification model has helped to find the optimum casting parameters to improve as cast semis quality

Hot isostatic pressing(HIP) is an effective method to eliminate the shrinkage in castings. The morphology of shrinkage is complex, and there are structures such as sharp corners and small passages which would lead to a large number of elements and easily divergent calculation results. Therefore, the application of numerical simulation in HIP is limited. To solve the non-convergence problem, the real shrinkage is often simplified as a sphere. However, this simplification ignores the characteristics of the shrinkage and makes the simulation results unreliable. In this paper, the technique of morphology-equivalent ellipsoid is applied to the numerical simulation of shrinkage evolution during HIP. Firstly, the 3D morphology of shrinkage in Ti6Al4V alloy castings is obtained by micro computed tomography. The radius of sphere and the geometric size and orientation of morphology-equivalent ellipsoid are calculated by corresponding equivalent techniques. Secondly, the numerical simulations of HIP for the Ti6Al4V castings before and after the equivalent method are carried out. The volume evolution of three kinds of shrinkages are recorded and compared. The results show that the volume evolution of the morphology-equivalent ellipsoid is closer to that of the real shrinkage, the feasibility of the morphology-equivalent ellipsoid and the limitations of sphere are verified.

During continuous casting of steel, the strand is exposed to severe thermal and mechanical stresses. These may trigger formation and growth of cracks, which deteriorate the slab properties. Particularly, at temperatures in the region of the second ductility trough, deformation can become critical. A low ductility level is often observed in a temperature range of 600°C to 1200°C, where the bending of the strand is commonly performed. One of the concepts for prevention of this problem is known as “Surface Structure Control Cooling” (SSCC). SSCC aims to intensively chill the external section of the strand just below the mould. In the present study, the SSCC heat treatment is compared with the regular cooling conditions for the concast strand. Hot tensile tests, which have been found very useful in assessing a steels susceptibility to cracking, are performed on a thermo-mechanical simulator. Both thermal histories are applied to microalloyed steel with variations of the test temperature and the results are summarized in hot ductility curves. Furthermore, the microstructures are investigated by optical microscopy and the precipitation states are studied by transmission electron microscopy (TEM).

Ming HE 1, 2, Xianliang LI 1, 3, Qingwei WANG1, 2, Qiang WANG 1, * 1 Key Laboratory of Electromagnetic Processing of Materials (Ministry of Education), Northeastern University, Shenyang 110819, China 2 School of Metallurgy, Northeastern University, Shenyang 110819, China 3 School of Materials Science and Engineering, Northeastern University, Shenyang 110819, China * Corresponding author: wangq@mail.neu.edu.cn Abstract: In the electromagnetic induction controlled automated steel teeming (EICAST) technology of ladle, the position and thickness of the blocking layer are critical factors to determine the coil installation position and the coil length. Therefore, it is also an important factor affecting the successful implementation of this new technology. In this paper, the influence of the molten steel refining temperature and the ladle containing time of molten steel on the position and thickness of the blocking layer were studied by numerical simulation method. The molten steel refining temperature and the ladle containing time are determined by the steel grades and the production processes, the position and thickness of the blocking layer can only be adjusted by changing the elements of alloy filled in the nozzle. Therefore, the influence of alloy compositions on the position and thickness of the blocking layer was also analyzed. To verify the accuracy of numerical simulation results, a large experimental platform was used to imitate the actual working condition of the ladle with the electromagnetic steel teeming system. The results show that the position of the blocking layer moves down with the increase of the molten steel refining temperature and the ladle containing time of molten steel. The thickness of the blocking layer decreases with the increase of the molten steel refining temperature, however, it increases with increasing of the ladle containing time of molten steel. In addition, the position and thickness of the blocking layer can be largely adjusted by changing the elements of the alloy filled in the nozzle, which is very important for the industrial implementation of the EICAST technology. Keywords: EICAST, refining temperature, containing time of molten steel, blocking layer, numerical simulation

The ice-templating of porous materials has received a great deal of attention during the past few years. This simple process, where a material suspension is simply frozen and then sublimated, provides materials with unique porous architectures, where the porosity is almost a direct replica of the frozen solvent crystals. The basic principle of ice-templating is the assembly of a second phase, usually particles, triggered by their progressive concentration increase in the inter-crystal space. This concentration mechanism can be used to induce ordering or self-assembly of various types of building blocks between the ice crystals, providing thus materials with an increasingly elaborate architecture with improved functional properties. I will present first a few examples of such materials and phenomena investigated in our group. The interaction of objects with a moving solidification front is however a common feature of many industrial and natural processes such as metal processing, the growth of single-crystals for photovoltaics and microelectronics, the cryo-preservation of cells, the formation of sea ice, or the preparation of frozen food. Solidification fronts interact with objects with different outcomes, from the total rejection to the complete engulfment of objects. Being able to understand and control the solidification dynamics and microstructure is of primary importance in these domains. I will discuss how the recent developments in the lab with cryoconfocal microscopy may help us make progresses in these domains and revisit some of these old but still relevant problems.

For the simulation of phase transformations and thermal problems in general, a set of input data is required that for the majority of alloys is unavailable with sufficient accuracy. ”Remedies” are to use data of the pure major alloying element, or use constant values that were measured at room temperature, neglecting their temperature dependence. The effort to gain temperature dependent data for thermal conductivity, thermal diffusivity and heat capacity is considerable, requires several measurement devices and a high degree of expertise. At present it is unlikely that this effort will be undertaken for numerous alloy systems and compositions. At FSU Jena a measurement method is being developed that allows to measure thermal conductivity, thermal diffusivity and heat capacity as a function of temperature in a single experimental run by evaluating transient and steady state temperature profiles. A state-of-the-art infrared camera is used to measure the temperature distribution of a rod-shaped sample in a temperature gradient with high resolution. Transient states are evaluated employing an inverse method to acquire thermal diffusivity, the steady state is evaluated for thermal conductivity and the temperature dependent heat capacity is calculated from these two entities. The accuracy of the measured data is comparable to those gained by conventional methods. This is demonstrated for Ni and 70:30 brass.

With the advent of high-energy X-ray sources, it is now possible to follow solidification processes in three dimensions and as a function of time. The ability to observe and quantify the solidification process in metals on sub-second time scales and micron spatial scales in three dimensions provide fundamentally new insights into this complex phase transformation. Only through three-dimensional measurements is it possible to quantify the complicated interfacial morphology and topology of solidification microstructures. We illustrate the power of this approach by examining solidification processes over a range of time scales. At the shortest timescales, the morphologies of free-growing hyperbranched dendrites in Al-Zn and Al-Zn-Cr alloys are examined. Unlike classical dendrite morphologies, by scaling the interfacial shape distribution with a time-dependent characteristic length we find that the morphology of these dendrites is approximately self-similar during growth. At longer time scales, the evolution of the topology of a dendritic mush and processes that alter it, such as fragmentation, is examined. Finally, spatial correlations of interfacial curvature are used to explain why the t1/3 temporal power law for the evolution of a characteristic length scale during coarsening is so robustly observed even when the microstructure of the mush is not self-similar, and thus when theory predicts that a temporal power law should not exist.

Solidification microstructure of metal castings determines their as-cast mechanical properties. Accurate prediction of as-cast grain structure and key defects (such as porosity) is critical in the design and manufacturing of metal castings using an integrated computational materials engineering (ICME) approach. In this talk, a three-dimensional (3-D) model based on cellular automaton (CA) and process simulation will be presented for predicting grain growth coupled with hydrogen porosity evolution during solidification of aluminum alloys in high pressure die casting (HPDC). The 3-D CA model integrates the concurrent nucleation and growth of grains as well as those of hydrogen porosities. The diffusions of both solute and hydrogen are considered in the model. A test specimen casting, consisting of different wall thicknesses, was simulated using a finite element based software ProCAST®. The thermal history of the simulated casting was extracted and used in subsequent mesoscale CA modeling to simulate the evolution of microstructure during HPDC. The grain morphology, grain density and grain size were obtained, and the porosity size and distribution were computed by CA modeling. The effects of cooling rates on final grain size and percentage of porosity were discussed. Electron backscatter diffraction (EBSD) analysis was performed on different wall thicknesses samples to validate the simulated results of grain size and distribution. X-ray micro computed tomography (microCT) technique was used to characterize the porosity morphology and distribution qualitatively and quantitatively. The 3-D simulated microstructure results, including grains and porosities, are in excellent agreement with the experimental results, which means the present model can be used in ICME design and development of aluminum castings.

Advanced numerical simulation models for continuous casting of steel were developed in Finland by Casim Consulting and Aalto University in cooperation with the University of Oulu and the steel industry. The aim was to develop models that are scientifically rigorous, but also computationally fast enough to be used in online applications. The models developed are a transient three-dimensional heat transfer model, CastManager, and a solidification and microstructure model, IDS. The computing time of these models are short, and they are integrated together in one online concept. This concept is installed in the automation systems of four slab casters in Finland. Testing and validation work is in progress. The system simulates the important heat transfer, solidification and microstructural phenomena in continuous casting online. The future aim is that this information will be used for online quality control and for optimizing the process conditions to avoid formation of defects. Many quality indices have already been developed. A steady state version of the CastManager tool has also been developed, called Tempsimu.

Numerical simulation technology has been widely used in the field of casting because of its visibility and forward-looking. However, for numerical simulation prediction of complex shape castings, the designer usually considers pouring under ideal conditions. The thermal property parameters used are uniform standardized parameters, and less consideration is given to the physical property parameters and boundary condition differences caused by process fluctuations in the casting process. Therefore, the numerical simulation prediction results are quite different from the actual production casting defects, which restricts the application and development of numerical simulation technology in casting process design and optimization. In this paper, a single-piece simulation and quality control method for complex shape casting is proposed considering the condition of process fluctuations. Firstly, the sensor, industrial data acquisition and casting single-piece full-cycle management technology is used to accurately obtain the key process parameters such as alloy casting temperature and time, initial temperature of casting and auxiliary materials during the forming process, and the volatility characteristics of the process parameters are derived. Secondly, the thermal property parameter reverse technique is used to carry out the pouring temperature measurement experiment of the standard test piece, and The orthogonal test and the anti-heat conduction method are applied to accurately obtain the thermal property parameters, and series of simulation of the whole process for casting process is performed. Thirdly, combined to the defect level analysis method, Fuzzy evaluation method and deep neural network method, the relationship model between key process parameters and differences is established, and then correct the model through differential reverse feedback to achieve accurate prediction of individual casting defects. Finally, based on the results of accurate prediction, multi-objective optimization algorithm is applied to find volatility characteristics of key parameters and sensitivity and influence mechanism of defects, and combining defect repair method of adaptive operation of casting to digitally control casting products quality in whole process. The research and implementation of the single-piece simulation and quality control method will provide scientific theory and technical support for high-quality and high-performance complex shape casting.

One of the crucial ingredients of today’s numerical simulation technologies is their cross-scale and cross-platform capabilities for material processes applications. Handling of simultaneous evolution paths at various scales\times (i.e., multi-scaling) during complex material processes where materials microstructure & microchemistry interact with meso and macro events, is one of awkward challenges of computational material engineering. The material phase change and also its thermal energy evolutions are also drastically increasing the complexity of the numerical simulations. The introduction of multi-resolution and multi-scale numerical schemes in recent years and their ground-breaking potentials for computational material science applications have vividly raised the expectations for more resourceful future virtual tools. A crucial point in implementing these novel numerical techniques for simulation of material processes is their flexibility towards the modelling approach (i.e., discrete, continuous...) and also their compatibility with solver-independent platforms. As these multi-resolution\physical techniques should provide some answers to the best ways of designing future high-performance materials along with optimisation of new & existing material processes and also improvement of their life-time performance, a broad & well-structured research work is required. Hence, the proposed multi-resolution framework herein, has been developed based on analytical & numerical techniques built on sound physical and mathematical foundations developed during the last few decades. The combination of recently developed concepts of dynamic\evolving domains along with cross-scale grid overlapping\interfacing and also sound parallel computing routines have been employed to address the multi-scale challenges of material processes simulations. In the research work herein, analytical and computational aspects of multi-resolution simulation framework for dynamic casting processes (i.e., continuous and semi-continuous casting) are presented and physical\mathematical basis of the analytical-computational solidification and cooling modules are elaborated. Industrial applications of the techniques are also envisaged using parallel-processing and fast computing facilities for full-scale casting applications.

Numerical simulation of casting filling process with complex shape is time-consuming. Compared with the traditional SOLA-VOF method, the lattice Boltzmann method (LBM) calculates the pressure field by particle distribution functions instead of the correction of the velocity and pressure fields, which greatly simplifies the calculation process. In addition, the LBM provides a flexible approach which can be easily parallelized. In this study, the LBM is employed to simulate casting filling process. An implementation of a volume-of-fluid (VOF) method within the lattice Boltzmann framework is proposed to capture the free surface of the casting filling process. A Smagorinsky large eddy simulation (LES) model is adopted to improve the numerical stability of the LBM. An adaptive time stepping technique is implemented to ensure an efficient and stable simulation. The model is validated by the experimental and simulation results of Campbell box filling process. The filling process of complex casting is simulated, and the result is compared with the filling process obtained by the SOLA-VOF method. The prediction accuracy and reliability of free surface profile is analysed.

The manufacturing of special steel forgings normally follows the basic technical process from casting, homogenization annealing, pre-heating, forging and post-heat treatment after forging. Such treatment obviouly is high-cost, energy dissipation and long manufacturing cycle. In terms of the formation mechanism of internal casting defects such as macrosegregation, shrinkage cavity and coarse grains in large steel ingot, this study proposes a novel casting-forging integrated technology (CFIT), in which the solidifying ingot is forged before its complete solidification. It was found that composition and microstructure were both inhomogeneous when deformed by CFIT with a higher liquid fraction above 0.5, and even A circle-like C, Cr, Mo and V macrosegregation band appeared in the forging. In contrast, a uniform microstructure and mechanical properties were obtained at the liquid fraction of 0.3. Further, to obtain the appropriate liquid fraction of 0.2 before forging and the time period of individual step, the plant-wide numerical simulations of CFIT by the arbitrary Lagrangian-Eulerian method and three-dimensional thermo-mechanical analysis were carried out when foring the 19t 16Mn thread tightening ring in industry. It shows that the forging by CFIT has slightly higher strength and hardness, and comparable impact energy and plasticity compared with the traditional processes. More significantly, the current technology presents highly economic, energy-conservation and short-period advantages in producing special steel forgings.

Graphite degeneracy in heavy-section spheroidal graphite cast irons is mostly associated with the formation of chunky graphite which consists of large eutectic cells with interconnected graphite strings. At low level, appearance of chunky graphite is limited to its non-aesthetic effect on machined surfaces, while at higher level it is detrimental for mechanical properties of the components. Chunky graphite is often related to high silicon levels and too high cerium additions during the spheroidization treatment. The appearance of this defect may be limited by controlled additions of antimony that is thought to tight the excess of cerium, but other impurities and low level elements may have to be considered during melt preparation. This contribution proposes a review of recent results and approaches on chunky graphite appearance, primarily but not exclusively in the case of heavy-section cast irons. Based on this literature review and series of experimental data, a predictive index for evaluating the risk of chunky graphite appearance is proposed. Lines for further research work aimed at a better understanding of graphite degeneracy are finally suggested.

An experimental device conducts thermal analysis and volume change measurements in a single ceramic cup with cast iron quality as the variable. The recorded data are processed using specialized software. Experiments compare solidification patterns for white [WI], grey [GI] and ductile [DI] irons, to correlate the most important events between the cooling curves and contraction curves, to evaluate the sensitivity to shrinkage formation. All of the irons have similar values for initial expansion up to the start of eutectic freezing [0.437 – 0.443%]. Graphite formation promotes expansion [WI-0.002%, GI-0.109%, DI-0.596%], resulting a difference in the reached maximum expansion [WI-0.465%, GI-0.552%, DI- 1.032%], placed between the end of eutectic recalescence and the end of solidification. Higher graphite expansion, greater the shrinkage sensitivity: open shrinkage increased while the density of a casting when considering total shrinkage and micro-shrinkage formation decreased. Prolonged graphitization at the beginning of eutectic reaction increased the expansion and, consequently, shrinkage sensitiveness. More graphite formation at the end of this stage also increased expansion, but this phenomena contributed to reduce of shrinkage level, due to the better access of liquid iron to compensate contraction holes. Special metallurgical treatments can favour a strong graphitization process at the end of solidification, with beneficial effects on the castings soundness. Keywords: white, grey and ductile cast irons; solidification; thermal analysis; cooling & contraction curves; graphite expansion; shrinkage

Deleterious effect of sulphur excess in gray iron is shortened by Mn addition. The suitable Mn addition according to sulphur in the metal is preformed according to the different addition rules available in open literature. With Mn addition is intended to avoid FeS formation by MnS precipitation that also prevents the undesirable effect of sulphur dissolved in the melt in the morphology of graphite flakes. To achieve the latter effect, it is thinking that MnS precipitation must occurs before eutectic reaction were flakes of graphite are formed. In order to get more basis for the choice of the rule of Mn addition, this study was focused to provide experimental evidence of MnS precipitation sequence in a hypoeutectic gray iron; Fe-3.3%C-2%Si-0.012P with 6 different Mn/S ratios: 28, 16, 4, 4.6, 2.6 and 0.7. Partially solidified samples of these alloys were obtained using a Bridgman furnace with quenching capabilities. Experimental evidence shown that the temperature of precipitation of MnS is in a reasonable agreement with the temperatures that could be calculated using thermodynamic data extrapolated from 1600 °C.

Tailored heterogeneous distributions of microstructural features enable extraordinary material performance in biological and physiological structures such as trees, the aortic arch, human teeth and dinosaur skulls. In ductile iron, a heterogeneous distribution in size and morphology of graphite nodules and variations of the fractions of ferrite and pearlite are created during solidification, and varies as a function of parameters such as local cooling rate, segregation and flow. In the current work, the size distribution as well as the orientation and relation between graphite nodules is obtained by a three-dimensional reconstruction of a ductile iron microstructure from X-ray tomography. The effect of the nodule morphology and clustering on the localization of plastic strains is studied numerically using finite element analysis of the reconstructed microstructure. Real castings have a variation in geometry, solidification conditions and are subjected to variations in loads. A framework for optimized geometry and solidification conditions in order to design and deliver castings with tailored local material performance is proposed.

One technology that is often employed in continuous casting of high-carbon steel billets to minimize centre- (or macro) segregation is hard secondary cooling. Investigations unanimously show, that hard cooling significantly reduces macro-segregation, but a mechanism for the reduced segregation is rarely given. In this paper the solidification of high carbon tire cord grade C80D cast as a 150x150 mm billet is calculated using the proprietary SMS Group solidification simulation package CHILL using steel properties calculated with the Thermo-Calc Software package and TCFE steels database. The obtained cooling rates in the billet for hard and soft secondary cooling are used to run solidification simulations considering solute redistribution using the diffusion module DICTRA. It is shown that for cooling rates achieved in continuous casting the steel solidifies far away from equilibrium. The solidification profile and solidus temperature lie in between the Scheil solidification model and the para-equilibrium Scheil model with carbon defined as a fast diffusing element. The calculated cooling rates and temperature gradients are used to simulate the solidification microstructure 20 mm from the billet surface using the phase field approach and the MICRESS® software package linked to Thermo-Calc through the TQ interface. This model clearly shows, that the most probable mechanism by which hard cooling reduces segregation is trapping of solutes between the intricately branched dendrite microstructure that results from the steep temperature gradients achieved when applying hard secondary cooling.

IDS (Inter-Dendritic Solidification) is a thermodynamic-kinetic software package that simulates phase changes, compound formation/dissolution, and solute distribution during solidification of steels as well as during their cooling/heating process after solidification. The software package also simulates solid-state phase transformations related to the austenite decomposition process at temperatures below 900/600 °C, and calculates thermophysical material properties from the liquid state down to room temperature. These data are needed in other models, such as heat transfer and thermal stress models, whose reliability heavily depends on the input data. The software package also features a database for thermodynamic, kinetic and microstructure data, as well as for several material properties. Owing to the short calculation times, the IDS tool is suitable for online applications. This paper presents IDS and its modules with the latest developments and validations, along with examples of modeling results.

During solidification of alloys with a temperature gradient in a stati magnetic field, the thermoelectric magnetic(TEM) force may be induced. The TEM force will produce convection in the molten metal and stress in the solid during the solidification. The convection and stress may play an important role in the formation of the micro-and macro-structures of the alloy. The influence of the thermoelectric magnetic force on the unidirectional solidification of Al-Cu alloys under a static magnetic field has been investigated experimentally. It is shown that the TEM force induced the instability of solid/liquid interface, which is the sound verification of the ATG instability model proposed by Asalor, Tiller and Granfield in 1990`s. The experiments showed that the dendrite growth was modified seriously and the dendrite spacing increased along with increase of the magnetic field. The columnar- equiaxed transition during directional solidification of the alloy was promoted by the imposing of a astatic magnetic field, due to the TEM force and convection. The static magnetic field was imposed in the additive manufacturing of superalloy with laser melting. It is shown that the magnetic field accelerated the CET transition and the crystal grains were refined. The result was attributed to the strong TEM force in the process.

Applying magnetic field during solidification of metallic alloys process under high thermal gradient may influence drastically microstructures of solidified metal. In fact, Magnetic Lorentz forces due to the coupling of electromagnetic field with thermo-electric currents are acting on the movement of solid grains and on the movements of fluid around grains, which finally produce re-organized and modified microstructures. We have developped 2D and 3D models to analyze and quantify numerically these effects. 2D and 3D Finite element models are based on non linear, time dependant algorithms, coupling electrokinetic to thermal and laminar fluid dynamic equations. Magnetic field is imposed and may be constant or varying at low frequency. Coupling terms coming from magnetism are Lorentz forces in fluid dynamic equations and induced currents in electrokinetics equation. Different approaches have been followed, working either with simplified geometrical grains (spherical, hexaedral ..), either with realistic grains coming from 2D X radiography or 3D tomography on Al7%Cu alloy sample. Solid grains are immersed numerically in the liquid using Chimera method. Results of simulation will present grain trajectories and movement during time and will be compared to analytical results on simplified geometries and to results coming from AlCu alloy solidification experimental processes.

The presence of small amount of the Fe in the Al-Si alloy causes the formation of the platelet-shaped Al5FeSi intermetallic phase, which is detrimental to the alloy mechanical properties. A series of directional solidification experiments on Al-Si-M alloy were performed under diffusion or controlled convection to study the interaction of fluid flow and the formation of the intermetallic phase and their effect on microstructure and segregation[1]. The blocking effort of the intermetallic phase on fluid flow in the mushy zone leads to the decrease of the permeability of the mushy zone both in flow normal and parallel to the primary dendrites and a decreased species transport. A two-phase model is used to study the effect of the intermetallic phase on the fluid flow and segregation. In the two-phase solidification model, the columnar dendrites are approximated by step-wise growing cylinders with constant primary and secondary arm spacing. The growth kinetics of the columnar phase is governed by the diffusion of the rejected solute ahead of the solid-liquid interface. The conservations for mass, enthalpy, momentum and species are solved respectively. The influence of the intermetallic phase on the fluid flow in the mushy zone is approximated with the anisotropic permeability law with an additional coefficient (where represents the impact factor of the intermetallic phase). Due to the insignificant amount of the Fe in the alloy, the simplified binary alloy Al-7%Si alloy is used to perform current simulation. The unidirectional solidification process can be qualitatively reproduced with the two-phase solidification model. Two inward circulation flow pattern ahead of the mushy zone, which induces the strong positive segregation in the center of the sample, is observed. The fluid velocity in the mushy zone is suppressed due to the decrease of the permeability caused by the formation of the intermetallic phase. At the same time, the segregation severity is reduced for the decreasing of the mass transfer in the mushy zone. The decreasing of the velocity will decrease the feeding ability, which makes it more likely to form some pores and porosities. References [1] S. Steinbach, L. Ratke, G. Zimmermann, L. Sturz, A. Roosz, J. Kovács, Y. Fautrelle, O. Budenkova, J. Lacaze, S. Dost, Investigation of the effect of fluid flow on microstructure evolution in Al-Si-Fe alloys: The MICAST project, in: Decennial International Conference on Solidification Processing, 2017.

The rapid solidification processing is frequently used to fabricate novel microstructures. In this work, the formation of novel microstructures in two Al-Cu alloys, i.e. hypoeutectic Al-26wt%Cu and hypereutectic Al-45wt%Cu alloy, in a steady magnetic field (SMF) was investigated using the conventional quenching technique. It was found that the application of the SMF led to appearance of a bulky θ-Al2Cu phase in Al-26wt%Cu and bulky supersaturated α solid solution Al-45wt%Cu alloy, respectively, which are impossible to appear under normal solidification conditions. The cooling curves of the quenched samples show that the cooling rate with a SMF is significantly higher than that without a SMF, and a larger undercooling before the onset of eutectic solidification was achieved in the SMF. The microstructure formation in the SMF can be explained using the phase diagram with metastable extensions.

Supercooled liquids may offer fascinating phenomena as compared with the normal state. In the case of supercooled paramagnetic liquids, completely different phenomena in high magnetic fields have be observed thanks to the high undercooling leading to higher magnetization and very strong magnetic coupling in the liquids. However, due to the restriction of the magnetic field intensity and maximum undercooling, the instability evolution in uniform magnetic field and magnetic field liquid are still not directly evidenced. In this study, high static magnetic field up to 25 T is applied to the metastable miscibility gap of Co-Cu alloys which owns liquid-phase separation at undercooled state, leading to the formation of α Co liquid phase at lower temperature that very close to the Curie temperature. With the application of strong magnetic field, the undercooled liquid are highly magnetized and Rosensweig instability with different shape and patterns are evidenced. The size distribution, volume fraction and instability pattern are strongly influenced both by the undercooling and field intensity. The mechanism of magnetic field on instability pattern formation and magnetism of liquid metals are discussed.

Electromagnetic effect produced in alloy melt and dendrite growth under the action of a pulsed magnetic field are researched by experimental and simulation. Simulation results show that alternating gradient electromagnetic forces act on the melt in front of the interface of the growing dendrites and the maximum electromagnetic force appears near the front of the solid/liquid interface, which forms a local electromagnetic oscillation and a strong disturbing effect on dendrite growth. Experimental observation shows that the stable growth of dendrites was disturbed, which led to occurrence of columnar to equiaxed transition (CET) and refinement of solidified grains. In addition, the pulsed magnetic field also produces electromagnetic convection in the melt which promotes the CET. As the results, the dendrites are observably refined even to spheroidal grains with the pulsed magnetic field. The CA-FD calculation shows that the electromagnetic convection causes the primary dendrite to rotate in solidification process. Because of the interface front is subjected to the action of multi-directional melt flow, the optimal growth of crystal is inhibited, which leads to the formation of spheroidal grains solidification.