Synchrotron X-ray radiography was used to study tensile and compressive deformations of semi solid Al-Cu and/or Fe-C alloys. In the case of tensile deformation of globular Al-Cu sample at ~60% solid, relatively high strain regions were formed even at mean strain of 0.005. The normal strain rate at the regions was 10 times as high as mean normal strain rate (3.45×10^–3 s^–1). At mean strain of 0.04, tensile deformation was localized in the high strain region, resulting in the formation of internal cracking in the plane normal to the tensile axis. On the other hand, in the case of compressive deformations of globular Al-Cu sample at ~55% solid and polygonal Fe-C sample at ~73% solid, shear bands with decreased solid fraction were formed at the domains tilted by approximately 45 degrees with respect to compressive plane. Rearrangement of solid particles including translation and rotation caused the shear induced dilation at the shear domains. Shear strain was localized at the shear domain with decreased solid fraction. Deformation of the polygonal solid particle of Fe-C sample caused a force to transmit over a longer distance than for the globular Al-Cu sample. Shear fracture finally occurred due to inadequate liquid flow into the expanding spaces between solid particles caused by shear-induced dilation. The solid/solid interaction including impingement between solid particles and rearrangement has significant role in the compressive deformation. These observations demonstrated that the mechanism of cracking formations induced by compressive deformation was totally different from that in the tensile deformation.

This study demonstrates in-situ measurement of solute partition coefficient in multicomponent alloys, using X-ray transmission imaging and X-ray florescence spectroscopy. The developed technique was applied to determine partition coefficients of Cr, Ni and Mo between δ phase and liquid phase in Fe-17.4Cr-12.6Ni-2.0Mo-1.6Mn-0.6Si alloys (mass%). In the observation and measurement, X-ray fluorescence spectra were directly obtained by irradiating the incident X-ray beam (23keV) to the solid phase or the liquid phase near solid/liquid interface. The partition coefficients were simply determined from the X-ray florescence analysis. In addition, successive measurements during unidirectional solidification allowed to measure change in partition coefficient along solidification path. During the solidification, partition coefficients of Cr, Ni and Mo changed from 1.01 to 1.08, from 0.76 to 0.70 and from 0.86 to 0.74, respectively. The present study proved that the developed technique was applicable to determine partition coefficients of 3d and 4d transition elements of which characteristic X-ray energies ranged from 4 to 20keV.

In-situ observation of flow of residual liquid has been deduced from the movement of bubbles generated at the end of solidification. Succinonitrile-water alloy system was used for experiment. At the final stage of solidification, some bubbles formed at the interdendritic region, because of the limit of solubility of air in the liquid phase. These bubbles grew and traveled as the solidification proceeded, keeping the position where the fraction solid was approximately constant. The loci of some bubbles were characterized, changing the solidified structure from columnar to equiaxed. In case of columnar dendrites, the bubbles moved along and normal to the primary trunks. On the other hand, in case of fine equiaxed grains, the loci of bubbles were very smooth. In case of coarse equiaxed grains, the loci of bubbles were very complex. Since the bubbles could not penetrate the equiaxed grains, they went around the equiaxed grains along the boundary. Therefore, the path was circuitous and complex. These observations suggested that the complex shape of solidified structure may increase the resistance of liquid flow at the final stage of solidification.

Methods for evaluating dendrite structure of an Al-5mass%Si alloy were investigated based on the fractal dimension and dimensionless perimeter. A phase-field simulation was carried out and calculated columnar dendrite showed a self-affine fractal characteristics. However, difference was small between self-affine parameters for x and y directions that mean columnar dendrites can be regarded as a self-similar fractal. Unidirectional and isothermal solidification experiments were carried out and it was shown that the fractal dimension and dimensionless perimeter of dendrites were determined only by local solidification time, and irrespective of solidification manner and dendrite morphology. This result demonstrated the potential of the fractal dimension and dimensionless perimeter as parameters for estimating local solidification time of an ingot. The tensile test showed that the tensile strength increased with increase in fractal dimension or dimensionless perimeter of dendrites in test specimens. Permeability of dendrite arrays were estimated by using the dimensionless perimeter as the tortuosity factor of interdendritic liquid channels. Obtained permeability corresponded to reported values.

In this study, the solidification path of a Fe-Ni-Cr-Mo-Cu alloy was investigated to determine the partition coefficient of each alloying element between the solid and liquid phases. The solidification path was confirmed to be single austenite mode whose solidification structure consisted only of austenite phase in the cooling rates from 0.06 to 0.69 K/s. The relation of λ₂=43.6 V^–0.41 between secondary arm spacing (λ₂) and cooling rate (V) was obtained by thermal analyses and microscopic observation.

The content of every alloying element in the primary γ phases was measured by EDS. Random sampling method, as many points were randomly analyzed, was applied to have content distributions of the alloying elements. The randomly measured data were rearranged with an increasing order except for Ni. Reference to Scheil’s equation could give us the partition coefficients (k_R). The k_R values were mostly in accordance to the equilibrium partition coefficients (k_E) obtained after holding a specimen in the solidification temperature for an hour. However, the k_R values tended to be closer to unity than the k_E values implying that segregation was predicted less with the k_R values. This fact leads to the presumption that the k_R values can evaluate the solidification sequence including the other factors such as diffusion of elements during and after solidification. Thus, it can be proposed that the k_R values are useful to predict microscopic segregation of the actual product under a given solidification and cooling condition.

A prediction method for microsegregation in Fe-based alloys was developed based on an approach of machine learning called Deep Learning. A set of model and algorithm of Deep Learning suitable for description of microsegregation was constructed by employing training data obtained by one-dimensional finite difference calculations for interdendritic microsegregation. It is shown that the developed method enables accurate prediction of the microsegregation behavior in Fe-based binary and ternary alloys with the solute atoms of C, Si, Mn, P and S. The present results demonstrate that Deep Learning offers a promising way of constructing an easy-to-use approach for prediction of microsegregation with high accuracy. Importantly, it is expected that the present method can be extended to describe effects of microstructural processes on microsegregation behavior.

An approach of machine learning called Deep Learning is utilized for construction of a prediction method of microsegregation behavior in Fe-based binary alloys with solute atoms of C, Si, Mn and P. Training data for the machine learning are obtained by quantitative phase-field simulations for directional solidification. Therefore, effects of microstructural evolutions on the microsegregation behavior are taken into account in the present method. Importantly, this method can be coupled with a macrosegregation model. The simulation result of the macrosegregation model is quite different from those obtained by a conventional macrosegregation model with the Scheil model and a model with a prediction method constructed from the training data of one-dimensional finite difference calculations for the microsegregation. This fact highlights the importance of accurate description of microsegregation behavior in prediction of macrosegregation.

A numerical model was developed to predict solidification grain structures and macrosegregation based on a three- dimensional cellular automaton finite difference method coupling with flow calculation of natural convection. For validation of the proposed model, simulations of unidirectional solidification cooled on the bottom surface of mold were performed at Al -10wt.%Mg alloy. Columnar grain structures have formed from the bottom to the top in alloy melt. During solidification, Mg-rich plumes rising in the melt were seen due to a subsequent upward flow, which caused by the thermosolutal buoyant force. Once the plume occurred in the melt above the mushy zone, the morphology of columnar grains varied and the grains became coarse. Mg-rich channels forming in the mushy zone were observed. Such region could delay solidification inside liquid-rich channels and results in a freckle defect. Examined the average Mg concentration profiles of the cross section from the bottom to the top, the negative segregation occurred in the middle of solidification and the positive segregation occurred at the end of solidification. From these results, it was confirmed that the proposed model was effective to predict the macrosegregation coupling with the grain structure formation. Moreover, the influence of the anisotropic permeability on macrosegragation formation was investigated in the present simulations. As the results, it was confirmed that the anisotropic permeability should be considered to predict quantitatively macrosegregation.

Direct simulations of macrosegregation were performed using a numerical model to predict solidification grain structures and macrosegregation based on a three-dimensional cellular automaton finite difference method coupling with flow calculation of shrinkage flow and natural convection. In order to investigate the relationship between solidification structures and macrosegregation, the simulations for a special mold used in model experiments of Sato et al. [Tetsu-to-Hagané, 99(2013), 101], which can form macrosegregation in the central region of the small ingot, were carried out. In the simulations, the bridging of columnar grains formed in the center of ingot during solidification, and then the positive segregation was generated in the region below the bridging. On the other hand, the negative segregation was generated in the region above the bridging. The primary factor of this macrosegregation was the shrinkage flow with the formation of bridging, and the degree of positive and negative segregation was enlarged by the presence of natural convection. As the results, it was confirmed that the shrinkage and the bridging of solidification structures played an important role for the formation of macrosegregation.

Effect of thermo-solutal convection on the macrosegregation in continuously cast slabs was investigated by numerical simulation using a finite-volume scheme. The thermo-solutal convection and solidification shrinkage induced flow was considered in the numerical model. The thermos-solutal flow induced the vortex flow during the solidification. The solute-rich liquid was washed out from the mushy zone by the vortex flow and the V-shaped segregation pattern was formed. The solidification shrinkage induced flow from top side to bottom side was dominant at the last stage of the solidification. However, the segregation pattern induced by this flow could not be calculated by well accuracy because of the grid size and convergence error.

Heat transfer analyses of casting processes offer an effective way of understanding solidification processes which is important for prevention of formation of casting defects and for control of segregations. The accuracy of the simulations depends on the accuracy of input parameters such as thermal conductivity and heat transfer coefficient. It is not straightforward to evaluate such parameters, especially thermal conductivities of bulks and heat transfer coefficient with high accuracy. It is very important to develop a reliable method for estimation of these parameters. In this study, particle filter, which is a method of data assimilation, is applied to estimation of thermal conductivity of melt and heat transfer coefficient during alloy solidification in mold casting and its applicability is systematically investigated on the basis of twin experiments. It is shown that a constant heat transfer coefficient can be accurately estimated in this method by utilizing a single cooling curve measured in the mold or in the melt near the mold, while the thermal conductivity can be accurately estimated from a single cooling curve measured in the melt. In addition, this method can be utilized for estimation of time-dependent heat transfer coefficient without any assumption and/or approximation on its time dependence. Importantly, the thermal conductivity and time-dependent heat transfer coefficient can be estimated simultaneously with high accuracy.

A high-speed video camera was installed in a model experimental apparatus for horizontal centrifugal casting process. Using this newly developed apparatus, in-situ observation has been performed with transparent organic substances. The former data for the loci of an equiaxed grain observed at the rotation rate of 250 rpm indicated like a line graph, because the event was recorded with 30 frames per second (fps). On the other hand, the present loci of an equiaxed grain recorded with 500 fps is smooth and very easy to characterize. This installation enabled us to rotate the experimental apparatus up to 1000 rpm, where the acceleration due to centrifugal force is the same order of magnitude as the real production process. The fluctuation of free surface and the movement of the equiaxed grain has been precisely characterized by the high-speed video camera up to the rotation rate of 1000 rpm. It has been found that the fluid flow can be stabilized as the rotation rate increased. Therefore, the formation of equiaxed grains decreased and the length of columnar dendrites increased with increasing rotation rate. These observation has been first achieved in this study because the experimental apparatus for horizontal centrifugal casting process has been equipped with the high-speed video camera.

Centrifugal force is applied to a molten metal during a centrifugal casting process. Therefore, grains of primary solidified phase, inclusions or other additional agents are separated if centrifugal force is strong. The centrifugally separated structure is strongly influenced by a relationship between solidification rate of castings and centrifugal separation speed. However, it is difficult to observe or measure what happens in the process because the process is carried out under high temperatures and the material is usually opaque. Therefore, we tried to observe the centrifugal separation behavior during the centrifugal casting process by using 2-phase flow simulation based on MPS (Moving Particle Semi-implicit) method which is one of particle methods. We simulated centrifugal separation behaviors in the process and investigated an influence of density difference between the phases, particle size, viscosity and rotation speed. As a result, the centrifugal separation behavior was well simulated by the particle method, and also well evaluated by Stokes’s law by considering an influence of an apparent viscosity increase caused by an error included in the MPS method.