Power modules are utilized for electric power control and play a key role in efficient energy conversion. A structural reliability problem of wire bonding, wire-liftoff, in power modules becomes important at high temperature operation. Wire-liftoff is a thermal fatigue phenomenon caused by thermal stress due to mismatch of coefficients of thermal expansion between a wire and a chip material. According to experimental studies, a saturation phenomenon of wire-liftoff lifetime was observed in power modules above the maximum junction temperature of 200℃. In the present paper, we propose a new failure model based on the nonlinear fracture mechanics parameter T*-integral range ΔT*. For practical use of the failure model based on ΔT*-integral, we propose a simple calculation method of ΔT*-integral that can be calculated by using a commercial finite element computer code such as Marc, Ansys, etc.

This paper presents a numerical solution for shape determination to maximize flow velocity in unsteady natural convection fields to maximize flow velocity. The square error integral of the actual flow velocity distributions in the prescribed sub-domains during the specified period of time was used as the objective functional. Shape gradient of the shape determination problem was derived theoretically using the Lagrange multiplier method, adjoint variable method, and the formulae of the material derivative. Reshaping is carried out by the traction method proposed as an approach to solving shape optimization problems. A numerical analyses program for the shape determination was developed based on FreeFem++, and the validity of proposed method was confirmed by results of 2D numerical analyses.

This paper presents a numerical solution to shape optimization for elastic fields considering Fluid-Structure Interaction (FSI). The mean compliance minimization problem in order to achieve stiffness maximization on elastic fields is formulated on volume constraint condition. The shape gradient of the shape optimization problem is derived theoretically using the adjoint variable method, the Lagrange multiplier method and the formulae of the material derivative. Reshaping is carried out by the traction method proposed as an approach to solving shape optimization problems. Numerical analyses program for the shape optimization is developed based on FreeFem++, and the validity of proposed method is confirmed by results of 2D numerical analyses.

This paper describes a numerical solution to the shape optimization problems of viscous flow fields. The lift maximization problem for an isolated body in uniform flow was formulated in the domain of unsteady-state viscous flow fields. The shape gradient of the shape optimization problem was derived theoretically using the adjoint variable method, the Lagrange multiplier method and the formulae of the material derivative. Reshaping was accomplished using a traction method that was proposed as a solution to domain optimization problems. Numerical analysis program for the shape optimization was developed based on FreeFem++, and the validity of proposed method was confirmed by results of 2D numerical analyses.

This paper presents numerical solution to shape optimization for viscous flow field considering Fluid-Structure Interaction (FSI). A minimization problem for total dissipation energy in viscous flow field is formulated on volume constraint condition. The shape gradient of the shape optimization problem is derived theoretically using the adjoint variable method, the Lagrange multiplier method and the formulae of the material derivative. Reshaping is carried out by the traction method proposed as an approach to solving shape optimization problems. Numerical analyses program for the shape optimization is developed based on FreeFem++, and the validity of proposed method is confirmed by results of 2D numerical analyses.

Image-based FEM is now widely used in the research works of variety of materials such as porous materials, but also used in the biomechanics analysis in the orthopedic and dental fields. So far, micro-CT, TEM tomography and FIB-SEM have been used. This study presents a novel image-based FEM using second harmonic generation (SHG) images for human bone, where collagen fiber could be observed in 0.5 mm x 0.5 mm x 0.15 mm region. The resolution of the sliced SHG images in this study was 0.83 μm x 0.83 μm x 20 μm. The purpose of the down-scaling FEM is to put the macroscopic boundary conditions on the small 3D model reconstructed from SHG images, by bridging the gap between the micro-CT image-based macroscopic model and SHG image-based model. The developed down-scaling FEM analysis system provided us with the strain distribution in the collagen fiber region.

The impact of stock price fluctuations on the global economy increases gradually since Lehman shock in 2008. The importance of research on stock price prediction has been increasing year by year. The stock price prediction method using Bayesian network is proposed in this study. The results show that the result of the proposal algorithm is more accurate than the conventional method such as time series analysis method, Neural Network, and so on.

Grammatical Swarm (GS), which is one of the evolutionary computations, is designed to find the function or the program satisfying the design objective. The algorithm of GS has one difficulty. Since the design variables are defined in the real-valued numbers, the real-valued variables have to be rounded to the integer-valued ones. The aim of this study is to present the real-coded Grammatical Swarm (RGS) in order to solve the above problem. The real-coded GS (RGS) is compared with the traditional GS in the symbolic regression problem. The best parameters of both algorithms are determined and then, their results are compared. The results show that RGS can find a better solution than traditional GS.

Quenching is one of the important manufacturing processes for steel to increase the durability, but the thermal deformation after rapid cooling requires a post-processing to fix the dimensions of the product. Hence, quenching simulation, which solves a multi-physics problem considering heat conduction, phase transformation and thermal deformation, is used in industries. To increase the accuracy and to understand the sources of numerical error, a novel sensitivity analysis was carried out with respect to the variability of chemical composition of steel.

Oscillation of microbubbles in bubbly liquids induces a dispersion effect of waves into weakly nonlinear pressure waves and its propagation process is described by a KdV-Burgers (KdVB) equation for a long wave. We numerically predict an evolution of waveform by solving the KdVB equation derived by Kanagawa et al. (2010, 2011) via a finite difference method. As a result, (i) an initial Gaussian waveform is distorted due to a nonlinear effect and a steepening of waveform is observed on about 8000 period; (ii) the dispersion effect is observed at a steep part of waveform and its part becomes sharp as the increasing of initial void fraction; (iii) the initial waveform change into an attenuated soliton on about 130000 period due to a dissipation effect and the attenuation of soliton is noteworthy as the decreasing of initial void fraction; (iv) a balance of the nonlinear, dispersion, and dissipation effects thereby forms an attenuated soliton, and its amplitude strongly depends on the value of initial void fraction. Although the nonlinear, dissipation, and dispersion effects were qualitatively balanced in the KdVB equation, we conclude that these effects independently appear in sound field.

In powder compacting, there is sometimes a demand to maximize the fill ratio. The random particle packing software Meshman_ParticlePacking under development at our company has the ability to achieve high packing ratios of over 60%, but with multiple particle sizes, high packing ratios cannot always be achieved. Now, it has become possible to achieve the high filling rate which could not be achieved conventionally by the method of filling in order from the large particle size. In addition, we searched for a method to efficiently search for the filling rate at or near the maximum with a specified particle size distribution by using Bayesian optimization, which is a type of machine learning, and report the results.

In this paper, a parameter-free optimization method for frame structures is proposed to minimize time-dependent response, which is one of the important mechanical properties. The objective is to minimize the transient vibration displacements at a specified area under the volume constraint. This time-dependent optimum design problem is formulated as a distributed-parameter optimization problem, and the sensitivity function with respect to the design velocity field of frame structure is derived based on the material derivative method, the Lagrange multiplier method and the adjoint method. The derived sensitivity function is applied to the H1 gradient method for frames, which is a gradient method in the function space to determine the optimal shape variation of frame structures. With the proposed method, the optimal frame shape for time-dependent response problems such as a forced-vibration, a free-vibration or an impact response can be obtained while minimizing the objective functional and maintaining the smoothness of frame shape. A numerical design example is shown to verify and validate the effectiveness of the proposed method.

In this study, we examined the estimation accuracy for flow field in shallow water area using Kalman filter FEM that is considering the moving boundary. The linear shallow-water equation is employed as the governing equation, and the Galerkin method and the selective lumping method are respectively applied to discretize the governing equation in space and time. In the process of numerical calculation, the Kalman gain matrix is first computed. Subsequently, the flow field is estimated using Kalman gain matrix. In order to examine the estimation accuracy, we used non dimensional parameter which is referred Courant number that is used the discussion of computational stability, and residual squares sum of true value and estimation value. As the result, we were obtained a result that is the residual squares sum of true value and estimation value are curvilinearly increasing. Moreover, we confirmed the Courant number that is accord the true value and estimation value nearly.

In this study, we present the influence of filtering method in three dimensional topology optimization based on the density method for minimization of strain energy. As the computational model, the cantilever beam is employed, and the results of topology optimization are compared by usual filtering and smoothing technique of sensitivity distribution based on the Poisson’s equation. The isoparametric element is used to subdivide the design domain, and the Galerkin finite element procedure is employed to discretize the governing equation for the elastic deformation. As a real problem, we analyze topology optimization for stent structure model.

A twin-screw extruder is a widely-used device for manufacturing polymer materials. For the optimization of extruder design and operation condition, numerical simulations of flows in twin-screw extruders have been desired. In this study, MPS (moving particle semi-implicit) method is used to simulate highly viscous incompressible flows with free surfaces in a twin-screw extruder. To handle complicated geometric shapes of the screw, recently-developed integral wall model is adopted for the treatment of solid wall boundaries. As a result, fluid flows driven by the rotation of screws were obtained. Therein, a flow through a small gap between screws and a cylinder was reproduced.

Variations or uncertainties of parameters to be considered in numerical simulation of fiber reinforced plastic composites are roughly classified into physical parameters, geometric parameters, and boundary conditions. In this study, damage progress analysis was performed in consideration of damage only to plastic base material in short fiber reinforced plastic. The analysis is performed by a first-order perturbation based stochastic homogenization method (FPSH method), and Meshman_ParticlePacking (Insight, Inc.) was used to create the model.

In recent years, the demand for production methods for directly producing final products from 3D CAD data using materials equivalent to final products in 3D printers has rapidly increased, so development of design technology using numerical analysis for 3D printer shaped articles is required. In this paper, we aimed to reproduce the deformation behavior by elastic-plastic analysis for resin like polypropylene which is a kind of 3D printer material and has physical properties close to that of polypropylene. As a result, by linking the optimal design software and the CAE software, we constructed a system to identify material constants by response surface optimization method, and confirmed its validity.

Cracks and destruction of glass occur, so it can be said that it is important to consider methods for evaluating the features and safety of glass. Although the features and safety of the glass have been experimentally evaluated, quantitative evaluation by numerical analysis is also necessary. Peridynamics is attracting attention as to the breakdown phenomenon of glass. This method is an analytical method that expresses a continuum by a particle model, and reproduction and analysis of destruction phenomena are expected. In this paper, we simulate DCDC method which is experimental method to investigate the mechanical properties of glass by Peridynamics. By comparing the analytical results on crack growth with experiments, we will examine the usefulness of Peridynamics in numerical analysis of glass. Solver uses Peridigm, an open source software using Peridynamics, to reproduce the DCDC method with a 2D and 3D model of Ordinary state-based PD (ODSBPD) and Non-Ordinary state-based PD (NODSBPD), which makes it easy to reproduce cracks. Analysis by changing the loading pressure showed the same tendency as in the experiment that the crack length became larger as the pressure became larger and crack propagation stopped when the energy was in equilibrium state. Therefore, it was found that the DCDC method can be reproduced by Peridynamics.

The phenomena of droplets colliding and scattering on the blades of the steam turbine were calculated using the general-purpose thermal fluid analysis software STAR-CCM+ by Lagrangian multiphase and liquid film model. The simulation results were verified by an experiment in which water was sprayed on the blade rotated at the rated speed and a pressure distribution sensor collects the impact of colliding droplets detached from the blade. The simulated distribution of droplets colliding and scattering on the front of the blade was close to the value measured by the pressure distribution sensor in the experiment. But the small droplets detached from the blade tip were not correctly measured due to the limited sensitivity of the sensor, and they need further verification．

High tensile steel, such as dual-phase steel consisting of ferrite and martensite, is still widely used in several industries. From a research and development point of view, it is important to clarify relationship between morphology or characteristics of each constituent phase and macroscopic mechanical properties. For the pursuit of such research, multiscale simulation is considered to be one of the effective tools. In this study, two types of 3-dimensional microstructure RVE models of dual-phase steel with different computational region size were generated based on observation result by serial sectioning method. Using these models, multiscale FE simulation based on homogenization elasto-plasticity theory was conducted, and effect of RVE model size on simulation results such as microscopic mechanical behavior of each phase or macroscopic strength was investigated.

This paper describes a new adaptive finite element method with deep learning. This method makes it possible to obtain better results from two analyses with a coarse mesh and corresponding refined mesh used in the adaptive finite element method. The proposed method consists of three phases. In the first phase, a lot of data set, each for different analysis setting, are collected, each of which includes three analysis results with a coarse mesh, corresponding fine mesh refined in the adaptive analysis and a very fine mesh enough to get converged results. Then in the second phase, a hierarchical neural network is trained to estimate the results with a fine mesh from the two results with a coarse and refined meshes. And finally, in the third phase, the trained neural network is utilized to estimate much better analysis results from two new analysis results with a given coarse mesh and its refined mesh used in the adaptive analysis. Basic performance of the proposed method is successfully tested through simple 2D stress analysis.

In our preceding study, we have applied the smoothed particle hydrodynamics (SPH) method to the problem of the fatigue crack propagation and proposed a method to solve the propagation of the planar cracks in the 3D body. Here, the method is applied to the non-planar cracks in the 3D body. To solve the propagation of the non-planar crack, we employ the information of the slope and the position of the crack surface in addition to the crack length. To confirm the validity of the proposed method, a fatigue test of CT specimen with an additional horizontal hole is carried out and the result is compared with the computed result. The computed result shows almost the same crack shape as the tested result.

Shape and topology optimization of the structure subjected to pressure load requires maintaining the smoothness of the boundary at which the pressure load is applied during the optimization process, and the direction of the load position changes with the variation of the boundary. Therefore, it is a difficult structure optimization problem. In this paper, we present the simultaneous shape and topology optimization method for the pressure load problem. In this method, optimization of the boundary subjected to the pressure load is set as a distribution type shape optimization problem by the traction method, and the optimization of the other region is set as a topology optimization problem by the SIMP method. With the proposed method, the light-weight structure can be obtained while maintaining the smooth of the boundary shape to deal with the pressure load problem. A design problem is presented to confirm the validity and the practical utility of the proposed method.

In this study, an adaptive mesh refinement (AMR) method is applied to phase-field lattice Boltzmann simulation of the dendrite growth with liquid flow and solid motion. Furthermore, a parallel computation using multiple graphics processing units (GPUs) is implemented for the method. To evaluate computational efficiency of the multi-GPUs AMR method, the growth of dendrite settling in undercooled melt is simulated. As a result, it is confirmed that the phase-field lattice Boltzmann computation is efficiently accelerated by the multi-GPUs AMR method.

This work will mention about an efficient aerodynamical sound prediction technique based on the computational fluid dynamics. The aerodynamical sound can be evaluated by the CFD (Computational Fluid Dynamics) technique to solve the full system of fluid dynamics equations. However, such computations often require huge computation costs caused by difference of characteristic speed between sound wave motion and variations of flow fields, and its unsteady computational approach. It seems to be a hindrance of application of CFD approach for aerodynamical sound predictions to the practical industrial design work. Present work proposes a numerical model for evaluation of sound power distribution use with the turbulence energy and the vorticity distributions which can be computed by only steady state CFD computations to reduce the computation cost. Proposed model is based on the Proudman’s idea. The model is applied to noise level predictions of ten different shape fans, then the variation of noise level between fans are predicted and compared with experimental observations. Present model can provide more accurate results compare with similar previous model.

Three-dimensional numerical simulation of incompressible, magnetohydrodynamic flows under traveling magnetic fields are carried out, which takes into account the coupling with the electromagnetic field in the solid and gas regions. A numerical scheme is constructed by combining the Galerkin finite element method and the edge-element based finite element method, which are applied to the discretizations of Navier-Stokes equations and the electromagnetic field equations, respectively. The solution algorithm for fluid flow is based on an explicit fractional step approach and the simultaneous relaxation of velocity and pressure to satisfy the continuity equation. The electromagnetic field equations are formulated with the use of the magnetic vector potential which is defined on the edge elements. Numerical simulations of the MHD flows in an annular channel are carried out under traveling magnetic fields. It is confirmed that the magnitude and oscillation amplitude of the z-component of the velocity becomes larger with the increase in the Hartmann number. It is also shown that the frequency of the velocity pulsation is doubled than the frequency of the externally applied traveling magnetic field.

From the mechanical point of view, welding processes are modeled as a coupling problem between heat conduction and thermal elastic–plastic problems. Such a welding mechanics problem generally requires a large amount of computational time due to its nonlinearity as well as a large number of time steps with a moving heat source. To overcome this difficulty, we are developing a parallel welding simulator based on the domain decomposition method. The domain decomposition method is one of the large-scale parallel finite element methods, and has been used for various problems such as heat conduction and structural mechanics. In the present paper, the methodology of the domain decomposition method for multi-pass welding problems is presented. In finite element analysis of multi-pass welding problems, a zero-stiffness element approach is usually used. In this approach, finite elements inside the welding pass do not have stiffness initially, and they obtain stiffness after the moving heat source arrives there. These zero-stiffness elements cause floating degrees of freedom, which should be constrained in the algorithm of the domain decomposition method.

Offshore floating structures need to be stable on the sea and structurally secure. Therefore, it is necessary to understand the behavior of the structure by waves. In such analysis, it is necessary to solve the fluid-structure coupled problem. The SPH method, which is a kind of particle method, is useful for solving these problems. However, it is known that the SPH method has problems such as vibration appearing from continuity of steady flow and vibration of solution generated from numerical error. In order to solve these problems, the δ-SPH method adds an artificial viscosity term to the equation of motion and an artificial density diffusion term to the equation of continuity. In this study, numerical analysis is performed using general SPH method and δ-SPH method respectively, and the results are compared.

Tissue morphogenesis in development is mechanically regulated by multi-cellular activities such as proliferation and constriction. Despite identifying the underlying molecular and cellular behaviors, the regulatory mechanism of how various tissue morphologies are robustly formed even under fluctuation is still unclear. In this study, we proposed a novel approach to capture the overall view of morphogenesis based on nonlinear continuum mechanics. We described the landscape of strain energy accumulated in the morphogenetic process, in which mechanically admissible process is represented by a path on the energy landscape. This approach enables us to understand the variety and robustness in tissue morphogenesis from mechanical viewpoints by exploring the behavior of local stable states on the energy landscape.

Mechanical meta-materials are the materials or structures which do not exist in nature but are artificially created to generate abnormal mechanical properties. Here we numerically studied the mechanical properties and found a novel phenomenon; when the material is pulled under linearly increasing stress, the material undergoes collapse transition at a given critical stress, and the stress-strain curve exhibits “S” shape. In order to understand the dynamical behavior of the system, we applied the normal mode analysis to the mechanical metamaterials by using numerical diagonalization. As a result, we found that there exists one floppy (zero frequency) mode as well as conventional acoustic modes at smaller frequencies and optical modes at large frequencies.

Conventionally, to improve printing quality of continuous-type inkjet printer, it has been necessary to repeat many experiments manually. Therefore, to streamline the improvement process, we developed an bayesian optimization system to determine an appropriate control. This system is composed of ink-droplet flight simulation using OpenFOAM, response surface method with Kriging model, and multi-objetive genetic algorithm. In this study, expected improvement and mutual information were implemented as aqusition functions. Optimizations with four different settings were performed. As a result, an optimization with expected improvement was most efficient and one with mutual information had advantage in terms of solution diversity.

In order to improve the electrical performances of polycrystalline bulk superconducting materials, which are attracting much attention for several applications, we need to control the polycrystalline microstructure formed by the solid-state sintering. Although some numerical simulation models based on the phase-field (PF) method have been proposed to reveal the microstructural evolution in the solid-state sintering, very few experimental verifications of the simulation results have been reported. The purpose of this study is to carry out the experimental verification of the PF simulation results of the solid-state sintering in polycrystalline bulk superconducting materials. For the quantitative comparison between the simulation and experimental results, we perform three-dimensional PF simulations of the solidstate sintering including many particles. In this article, we report the PF simulation with parallel computing using Message Passing Interface (MPI) library to reduce the computation time. The results of the simulation demonstrate that the simulation results comparable to the experimental results can be obtained by the PF model, since the PF model can analyze three-dimensional rigid motion and grain boundary formation in the solid-state sintering.

Decrement of the engineering popular in Japan causes a risk of maintaining infrastructures inappropriately. Integrity of nuclear power plants is one of the most urgent issues to keep a stability of a generation of electrical energy. In order to accurately evaluate a stability of a generation of electrical energy, supposing even a seismic load, a fracture behavior with a large plastic area must be grasped. J-integral has application limit. J-integral cannot evaluate a fracture behavior with a large plastic area. Physical quantities and constitutive equation that can evaluate fracture behavior with a large plastic area are studied. A ductile fracture under very low cycle fatigue and monotonic load increasing is mainly investigated with some specimen size and shape in numerical analysis. However, comparison and examination of numerical analysis and experiment are not sufficiently conducted. In order to set a boundary conditions in numerical analysis correctly, to measure a situation of an experiment precisely is important. In this study, a fracture toughness test of a three points bending is performed under monotonic load increasing with SEN(B) specimen of stainless steel SUS304. Crack growth length, crack front shape, deformation of specimen, load and load point displacement is taken measurement. These are for the generation phase analysis, i.e., precise measurement results in the experiment are useful for boundary conditions in a numerical simulation. In the three points bending test, a plastic deformation near a load point and a load direction for the SEN(B) specimen are obtained for FE analysis.

This paper presents a topology optimisation of cloaking device. Unlike the standard topology optimisation which minimises the scattered field in an observation region, the present one minimises the scattering cross section. With this formulation, we can avoid trial-and-error procedures for finding an appropriate observation region. We also propose an efficient evaluation of the scattering cross section and its topological derivative by exploiting the optical theorem, which accerelate the sensitivity analysis. Along with the detailed formulations, we demonstrate a numerical example of the topology optimisation.

By applying the renormalized molecular dynamics (RMD), we have developed the zooming analysis method. This method first calculates a whole domain on the renormalized (coarse-grained) system, then re-calculates a part of the domain on the less renormalized system. The method allows a more detailed analysis on a local domain of interest while reducing the computational cost. Applying the method to a supersonic flow through a Laval nozzle showed that the results of a whole domain analysis and a zooming analysis were agreed within 7% difference or less. Also, by applying the method to a condensing flow through the nozzle, clusters of particles were observed on the zooming analysis, which did not appear on the whole domain analysis in low resolution. The computation time was reduced to less than 30% of a high resolution case.

In this paper, we report a magnetic-fluid coupling analysis method for magnetic granules based on the Magnetic Beads Method (MBM). The MBM enables analysis of the magnetic interaction between the magnetic granules without spatial discretization. By combining the particle dynamics method and the free surface flow analysis method, analysis of the magnetic granules behavior in fluid under an external magnetic field has been achieved. A magnet separator is used in machine tools such as grinding machines to recover fine magnetic granules from a coolant fluid. We performed three dimensional magnetic field-fluid coupling analysis of the magnetic separator. The result has confirmed that the magnetic granules contained in the fluid stick to a drum surface by a magnetic field produced by permanent magnets. The number of magnetic granules sticking to the drum increased as the granule diameter increases. A recovery rate of the magnetic granules was in good agreement with experimental.

To apply Renormalization Molecular Dynamics (RMD) for elastic analysis, we need to arrange molecular along the shape. RMD is accurate if molecular arrangement close to the face centered cubic structure (fcc). There is also a method of using a commercially available “mesher”, but in many cases the molecular arrangement is not as intended. Therefore, we propose a method to arrange molecular in complicated shapes, such as gears and link mechanisms, and generate a mesh without breaking the fcc structure as much as possible.

In actual solidification of alloys, multiple equiaxed dendrites grow with motion. In our previous study, the phase-field lattice Boltzmann method (PF-LBM) that can simulate formation process of equiaxed structure by considering growth, motion, collision, and coalescence of multiple dendrites, was developed, where double-well potential was employed in the quantitative phase-field model for binary alloy solidification. When we simulate a series process from solidification to grain growth, application of double-obstacle potential, rather than the double-well potential, is desired in view point of computation. In this study, we employ the double-obstacle potential for the PF-LBM to perform accurate and efficient simulations. The accuracy of the model is evaluated by performing some simulations.

A key property for the design of new shape memory alloys is their working temperature range that depends on their transformation temperature T₀. In previous works, T₀ was predicted using a simple linear regression with respect to the energy difference between the parent and the martensitic phases, ΔE_p-m. In this paper, we developed an accurate method to predict T₀ based on machine learning assisted by the first-principles calculations. First-principles calculations were performed on 15 shape memory alloys; then, we proposed an artificial neural network method that used not only computed ΔE_p-m but also bulk moduli as input variables to predict T₀. The prediction error of T₀ was improved to 49 K for the proposed artificial neural network compared with 188 K for simple linear regression.

In recent years, the manufacturing industry has increased its use of evaluation by numerical analysis in the upstream process of development in order to reduce manufacturing rework. Furthermore, with the aim of improving product quality and reduce development man-hours, automatic determination of design parameters using optimization algorithms is also being actively studied, and there are some cases where numerical analysis is used when evaluating objective functions. However, due to the difficulty of analysis and computational cost, shape optimization on a part-by-part basis and overall optimization with a reduced dimensional model are the main focus. Therefore, we combined the in-house large-scale structure simulator and the optimization technology, and developed the shape optimization system based on overall structure analysis. Optimization was performed targeting plastic injection molding machines, with the shape of the linkage as the design variable and the energy consumption minimization during mold open / close operation as the objective function. As a result, it was possible to obtain a part shape that can greatly reduce energy consumption.

This paper presents topology optimization of magnetic cores for EV wireless charging device. In this optimization, the magnetic core shape for wireless power transfer (WPT) is represented by the Gaussian basis function. The core shape is optimized so as to maximize the coupling factor. In addition, robustness against misalignment between transmitting and receiving coils is taken into account in the optimization. It is found that the optimized core shape effectively increases the net magnetic flux interlinked with the receiving coil, and outperforms the conventional core shape.

In this study, we propose a parameter-free shape optimization method of a three dimensional rigid-plastic solid for minimizing plastic work, where the optimum free-form is obtained under the volume constraint without shape parameterization. We formulate this shape optimization problem as a distributed-parameter optimization system with both constraint conditions of the rigid plastic governing equation and the volume. The shape gradient function for this problem is theoretically derived by the Lagrange multiplier method, the adjoint variable method and the material derivative method. The shape gradient function is numerically determined by the Cauchy stress tensor obtained from the state equation (or the rigid plastic equation) and the strain rate tensor obtained from the adjoint equation with the stiffness variation term. The shape gradient function determined is applied to the H1 gradient method, a parameter-free gradient method in the function space, to determine the optimum shape variation by applying the fictitious distributed load proportional to the shape gradient function in the normal direction to the design boundary. A numerical example is presented to demonstrate the effectiveness and practical utility of the proposed method and the shape optimization system developed.

Carbon steel is used in wide spectrum of industrial products that sustain our society, and its characteristic is primarily determined upon solidification. In a recent study, phase transformations which is referred to as δ-γ massive like transformation were uncovered by in situ observation of the phase transformation when peritectic transformation was expected to occur in solidification processes. This reaction is a solid-solid transformation between δ phase and γ phase without any liquid phase, and thus the strain energy due to difference in densities between BCC-structured δ phase and FCC-structured γ phase would not be relieved. Thus, δ-γ massive like transformation may leads to solidification cracking. For tailoring solidification processes, resultant microstructures and materials properties, the detailed mechanism behind the phase transformation needs to be understood well. It has already been revealed in previous study that this transformation is dominated by nucleation of γ phase out of δ matrix phase. In this study, using phase field method with intra-nucleus inhomogeneity taken into account, we revealed that the inhomogeneity of γ nucleus reduced strain energy and facilitate nucleation. We also revealed how inhomogeneity of matrix δ phase affect to the γ phase growth.

Mechanical properties of metals used for structural materials are significantly affected by the dynamics recrystallization (DRX) progressing in the hot plastic work. Therefore, it is industrially important to predict the microscopic behavior of materials in the deformation process during heat treatment. In the previous research, we proposed the temperature dependent DRX model based on multi-phase-field (MPF) and dislocation-based crystal plasticity models and showed some numerical results. However, numerical simulation was conducted in an incorrect manner. In the preset report, we revise the numerical process and identify numerical parameters again. In addition, we perform a multiphysics analysis for pure Ni and investigate the temperature dependence of the microstructure formation by comparing the obtained nominal stress-strain curves during DRX with an experimental result.

This paper presents a numerical method for two-phase flow with complex boundary of solid object. Fluid flows are governed by the weakly compressible Navier-Stokes equations under the conditions of isothermal state and low Mach number. Finite difference method is employed for discretization on a staggered Cartesian grid. Two-phase interface is captured by conservative Allen-Cahn equation, which is solved by lattice Boltzmann method. Through the bounce-back scheme and an extension technique for phase field function, exact mass conservation at the solid boundary is achieved. After a validation of accuracy for bubble rising problem, the gas-liquid flow with rotating plate and water drop falling on a cylinder are simulated to demonstrate the practicality and stability.

A Magnetic Resonance Imaging (MRI) system is a diagnostic imaging device using magnetic field. In the MRI system, pulse current is energized to gradient coils in a strong static magnetic field generated by a superconducting magnet. Since electromagnetic force (i.e. Lorentz force) is generated in the gradient coils and the superconducting magnet, the MRI magnet system vibrates. Especially, the vibration of MRI system disturbs a homogeneity and stability of the magnetic field. As the magnetic field disturbances causes error magnetic field and affect image quality, it is important to evaluate them in a design process. We are evaluating them by an oscillating magnetic field analysis technology that combines structural analysis and magnetic field analysis. In this study, using excitation forces taking into account operation conditions, the difference of the error magnetic field by the structures of the superconducting magnet was evaluated.

A lot of numerical studies have so far been conducted to systematically investigate the effects of anisotropic grain boundary properties on grain growth. Nevertheless, conclusive knowledge on the correlations between boundary properties and grain growth characteristics is not yet established, which is mainly due to insufficient statistics coming from the limited computer resources. In this study, by utilizing parallel graphics processing unit computing on a supercomputer, we perform large-scale simulations of anisotropic grain growth with more than three million initial grains. The multi-phase-field grain growth model is employed, enabling accurate and efficient treatment of grain growth. Misorientation-dependent anisotropies in grain boundary properties are introduced by simple models. Through a series of grain growth simulations with the misorientation-dependent properties, the anisotropy effects on the growth behaviors are elucidated in detail.

4H-SiC is receiving attention as a next-generation material for power devices. However, the spontaneous expansion of stacking faults due to carrier activation, which is called bipolar deterioration, remains as a serious problem. This phenomenon is considered to be explained by the magnitude relationship between stacking fault energy and the energy gain by carrier traps to stacking faults. However, since direct measurement of stacking fault energy at finite temperature is difficult, quantitative relation is not yet clear. In this study, we estimated the temperature dependent stacking fault energies of single Shockley-type stacking fault and double Shockley-type stacking fault by calculating free energy of perfect and faulted crystals using ab-initio calculation and harmonic vibration approximation model. For a single Shockley-type stacking fault, it was found that the stacking fault energy decreased with temperature rises, and was lower than the experimental value above room temperature, which is consistent with the result predicted from the energy gain using band theory. For the double Shockley-type stacking fault, the stacking fault energy is decreased to almost 0 mJ/m² at 1300 K, and increased slightly at above 1300K. In order to estimate the stacking fault energy at high temperature more precisely, it is necessary to consider the thermal expansion.

In order to prevent fatal traffic accidents triggered by sudden loss of consciousness of drivers, a non-contact brain wave sensor installed on the head restraint is studied. A textile microstrip patch antenna is fabricated and attached on the head restraint, then effects of its deformation due to the contact force between the head and the restraint are studied. Microwave is radiated through the antenna and reflected signal is measured by a spectrum analyzer and a return loss bridge. Visual evoked potential (VEP) is used to evaluate the effect of the deformation in the study; a checkerboard pattern flickering at 20Hz is imposed on a subject and the neuronal activities are measured by both conventional EEG and proposed sensor. Frequency response with peaks at 20Hz and 40Hz are observed by both methods. A numerical model is also used to perform electromagnetic analysis to study the response of brain tissue activated by the VEP.