Recently, there has been a movement to make bicycle helmets mandatory. One of the authors, Hagiwara et al. produce work helmets made of cardboard. And the helmet has already been well received. As since it follows the iron rule of "Using axial crushing of the pillar structure", a helmet is comparing to the conventional curved structure, it is very light and sufficient energy absorption can be obtained. For bicycles, energy absorption characteristics in all circumferential directions of the head is questioned. Here, we made a new hat based on Namako folding, and confirmed its effectiveness. In addition, we will try to prepare and construct a simulation model for the energy absorption effect to verify whether the helmet is resistant to impact.

In Japan, the fan industry is more than 1200 years old. Skills have been passed down from generation to generation to this day. With the aging of the population, these skills have been lost due to the lack of continuity of the new generations. That is why the Meiji Institute for Advanced Study of Mathematical Sciences (MIMS), in collaboration with several local thousand-year-old companies, have proposed to re-establish the fan manufacturing industry with the use of new technologies in processes where human intervention is not essential and manufacturing processes can be accelerated by cutting production costs. The work focuses on the measurements of the surface roughness of the bamboo that is used as the skeleton of the fans and preliminary results are shown with four methods of recognition of the roughness of their surfaces from the processing and automatic recognition of their images.

The numerical simulations of the mode vibrations of the violin are performed by the finite element method using the software. The geometry of the violin is scanned by micro-CT scanner and the orthotropic properties of spruce and maple, such as Young’s modulus, the modulus of rigidity and Poisson’s ratio, are set in the parameters of the numerical simulation. The main vibration mode are shown in this paper.

Physics-informed neural network (PINNs or PINN) is supervised learning for approximating the initial-boundary value problems of partial differential equations. This research focuses on the magnetostatic field problem derived from Maxwell's equations and evaluate the performance of large-scale computation and parallel computation of PINN. In finite element analysis, which is a conventional numerical analysis method, the magnetostatic field is formulated with the magnetic vector potential ‘A’ as an unknown. However, since the three-dimensional magnetostatic field problem has indefiniteness, it is necessary to solve a singular problem or construct an equation with the indefiniteness removed. This study also applies PINN to partial differential equations based on the A-formulation.

In order to educate people on disaster prevention, it is necessary to create a simulated experience that motivates people to engage in disaster prevention. In this study, we focus on a visualization system utilizing the results of numerical analysis. For simulations of disaster phenomena, various numerical analysis methods are used depending on the target, and parallel computing is employed to handle large-scale data. We apply parallel computation scheme to the marching cube method with a signed distance function and aim to generate a generic and fast 3D model of distributed data associated with large-scale parallel numerical analysis. The performance of the proposed method is evaluated for the surface mesh generation of Stanford bunny model.

Numerical analysis is effective for quantitative strength evaluation of composite materials such as CFRP. Simulations that distinguish between fibers and resins cannot be performed in a realistic amount of time. Therefore, multiscale simulators based on homogenization theory are employed. Simulations of hydrogen tanks are expected to require an enormous amount of computation time, even when using the homogenization method. In this study, parallel computation scheme is applied to reduce the computation time. When using multiple types of different microstructures, the conventional domain decomposition method for macroscopic structures may lead an unbalanced amount of computation. We evaluate to apply a static load balancing technique to parallel computation of an two-scale homogenization simulation.

The deflated conjugate gradient (CG) method is one of the linear solvers for numerical simulations, which can improve the convergence rate since the condition number would be reduced by inputting a known independent basis of the solution space. In this study, we measured the number of iterations and computation time for structural analysis using the deflated CG method with subdomain eigen-mode deflation. As the results, when the appropriate number of known basis are input, the proposed method leads better convergence and computational performance.

We aim to analyze a flapping-wing motion of a free-flight insect and have developed a new framework of s-version of finite element method (SFEM) called B-spline SFEM (BSFEM), which achieves both localized high accuracy and low computational cost. To analyze a large-scale problem such as a flapping free flight, distributed memory parallel computers are often used. In this study, we achieve parallel computing using BSFEM based on the graph structure representing the interactions between computational nodes. A strong scaling test is performed to evaluate its parallel performance.

The Matrix Power Kernel (MPK) plays an important role in the Algebraic Multigrid-Conjugate Gradient (AMG-CG) method, when using the Chebyshev polynomial smoother. In this study, we propose a new kernel for the MPK, which combines a Depth-First Search (DFS)-based MPK implementation with a reordering that has a recursive structure. Our MPK implementation reduces the need for synchronization among OpenMP threads, thereby simplifying the implementation and improving the performance. Results from numerical experiments show that our method consistently reduces the cache misses and the computation time across all OpenMP thread conditions, outperforming the baseline methods. The proposed method also improves the acceleration rate of MPK.

When solving sparse matrix linear equations by direct methods, algorithms which take advantage of their sparsity are required. The multifrontal method is a parallelizable method with few fill-ins in the numerical decomposition of matrices, which is advantageous in terms of memory consumption and efficiency when dealing with large systems. Supernodes are a concept for grouping dense portions of a decomposed matrix, and are now also incorporated in the computation of the multifrontal method. In this presentation, we will consider the use of supernodes in the multifrontal methods, especially for unsymmetric matrices.

The conjugate projected gradient (CPG) method is an iterative solver for a linear problem with linear multipoint constraints. In the present study, a formulation of the projection procedure is rewritten for the dual Lagrange method, which is used in a mortar finite element method. A block matrix with respect to a slave surface of the constraint matrix, C^T , becomes a diagonal matrix due to the biorthogonal shape functions, which is used in the dual Lagrange multiplier method. By utilizing this property, the projection may be conducted without solving a linear problem for a coefficient matrix, C^TC .

As silicon semiconductor devices are more integrated and miniaturized, silicon wafers for substrates used in the devices are required to be even flatter. Chemical-Mechanical Polishing (CMP) is an essential process in the manufacturing of silicon wafers to determine wafer’s flatness. Predictive simulation technology of polishing amount distribution is needed to improve wafer’s flatness efficiently. Especially this distribution depends on mechanical of properties of polishing pad. However, the pad’s compressive deformation in CMP process remains to be delineated. In this study, we have measured the pad’s deformation in situ and identified pad’s properties by viscoelastic characteristics formulation. Then, we have developed a simulation model of dynamic pressure on a wafer in consideration of pad’s viscoelastic characteristics using finite element method (FEM). As a result, the simulation model has reproduced the measured values and confirmed validity. Therefore, it is clarified that one of the keys for the accurate prediction of polishing amount distribution is the pad’s deformation in CMP process.

Thermal CVD is a promising process for epitaxial growth of silicon carbide (SiC), which is widely used in various types of electric devices these days. For improving CVD processes, CFD have a key role because to understand the kinetics, flow and heat is significantly important. In this paper, we examined the applicability of multiphysics features of Simcenter STAR-CCM+ for a simple 2D SiC-CVD reactor model, which including the coupling of reacting flow and electromagnetic field calculation. Moreover, a simple design exploration with Simcenter HEEDS was performed using a reduced order model (ROM) which is built with Simcenter ROM Builder from the results of 2D CFD.

Aggregates of Si vacancies (void defects) formed in the process of Si single crystal growth by the Czichralski (CZ) method are known to cause quality deterioration of Si wafers. Although first-principles calculations are widely used as an atomic-scale simulation method, there are limitations to first-principles calculations of large-scale defects in terms of computational cost. In this study, we analyzed the energetic stability of large-scale void defects (maximum number of Si vacancies: 969) in Si single crystals using a machine-learning interatomic potential. Our calculation results revealed a strong correlation between the energetic stability and the number of dangling bonds in the large-scale void defects.

One of the crucial challenges in GaN epitaxial growth by metalorganic vapor phase epitaxy (MOVPE) method is the control of carbon (C) incorporation, which affects the resistivity of the individual layer of high-power and high-frequency devices based on GaN. Incorporating C into the GaN epilayer depends on various process parameters, such as growth temperature, pressure, carrier gas, flow rates of source gases, and reactor type. To determine the process parameters that control incorporated C concentration and uniformity, we have developed a novel numerical model for C incorporation into the grown GaN epilayer using MOVPE method. The developed model considers C incorporation through the decomposition of Ga source gas. We have implemented this model in a commercial crystal growth simulator, Virtual Reactor Nitride Edition, developed by STR. This study used experimental data to evaluate the dependence of temperature, growth rate, and V/III ratio of C incorporation predicted by the simulations. These results showed a strong correlation between the C concentration and these parameters, and good agreement with the experimental data.

Fluororesin has low dielectric constant and dielectric tangent. For that reason, it has attracted attention as a material for high frequency substrate that reduce transmission loss in high-frequency bands in 5G communications. However, Fluororesin has a high coefficient of thermal expansion (CTE), which raises concerns about the dimensional stability of the substrates during the lamination process. Therefore, in this paper, we propose a simulation method for evaluating the dimensional stability of the substrates considering lamination process by the finite element method (FEM) for copper-clad laminates (CCL). In this method, the laminated molding of the substrate is expressed by the birth and death of elements, and the deformation of the substrate due to it is removed by forced displacement. The validity of this method was confirmed by comparing the calculation results with the test results.

With the recent development of data science, by utilizing surrogate models using simulation results as learning data, simulation results can be reproduced with certain accuracy in a very short time. By using this feature, it will be possible to use surrogate models in the initial stage of design and/or examination of manufacturing conditions. both of which have been difficult to apply conventional simulations due to the time required for calculation. This paper investigates the potential of surrogate model in application to these kind of problems by a case study of electroplating processes..

In recent years, the speed required for product development has increased significantly over the years. With the increasing levels of sophistication in requirements, data-driven development utilizing past data has been attracting attention in the automotive industry. Compared to physical testing, simulation is characterized by the ease of obtaining information from calculations and analyses, but on the other hand, handling huge amounts of data is a challenging issue. In this paper, a novel process to search for similar behavior of parts of interest from dozens of crash simulation results is proposed with order reduction by modal decomposition techniques.

Coastal erosion has long been occurring in Japan. Wave dissipating blocks have been installed on many beaches as a countermeasure against beach erosion, and a variety of shapes exist. In this study, the effect of different shapes on the flow around the blocks was analyzed numerically using the SPH method, and the reaction force on the wall surface after passing through a group of blocks was compared for each block shape to evaluate the wave dissipation effect. As a result, it was confirmed that the hexapod blocks were highly effective in dissipating waves. In addition, a coupled SPH-DEM analysis was performed to account for sandy soil. The validity of the coupled model was confirmed based on the experimental paper, and it was confirmed that the hexapod blocks had the highest wave dissipation effect in both the SPH method analysis and the coupled analysis, but the dissipation effect of the hexapod blocks was more enhanced in the coupled analysis that considered the sandy substrate.

When a thin shell structured is buckled, the incremental change of external work and internal strain energy are not balanced each other. In such a case, the arc-length method is widely used to proceed the calculation. On the other hand, when the implicit integration does not converge, the explicit integration can be used to follow the last converged result of the implicit integration. But if this auto-switching process follows the pre-setting condition, the result of calculation would be highly influenced. In this paper, the effect of the auto-switching process against to its result is reported.

Doyle-Fuller-Newman (DFN) model has been used as P2D (pseudo 2-dimensional) to predict basic electrochemical behavior of a lithium-ion battery (LiB). However, some effects such as structural deformation induced by swelling cannot be observed in P2D, and implementation of P4D (pseudo 4-dimensional) was tried in some of the studies. In this paper, some advanced studies on LiB electrochemistry are presented using the P4D module implemented in a commercial FEM code Abaqus/Standard. Battery casing deformation and its related effects such as porosity change are observed in TECM (Thermal-Electrochemical-Mechanical) simulation results. TECMP (TECM-Pore fluid flow) simulation results show axially non-uniform ion concentration distribution, which does not resolve at all after one hour of a break process.

In recent years, faster switching speeds have made it possible to reduce the size and increase the efficiency of power electronics devices. On the other hand, higher switching speeds have led to an increase in electromagnetic noise (EMI) generated by the devices. To prevent prolonged development time due to this EMI, front-loading development utilizing simulation is effective, and various EMI analysis methods have been proposed. In particular, a method for modeling the entire device with a three-dimensional electromagnetic field simulation using FEM is beneficial in verifying and improving the layout design before prototyping. In this method, modeling of magnetic components, one of the important components in power electronics, becomes a challenge. In this paper, we propose a modeling method for magnetic components to predict EMI by 3D electromagnetic field simulation for high frequency applications. With the proposed method, the measured and simulated impedance characteristics of a choke coil used in a boost chopper in industrial equipment in the frequency range from 150 kHz to 30 MHz agree with each other within an error of less than 15%.

This study presents the numerical analysis of a standing-wave thermoacoustic engine using the general-purpose simulation software COMSOL Multiphysics^®. The thermoacoustic phenomena arise from thermal interactions between acoustic waves and solid boundaries of narrow flow channels with temperature gradients. The thermoacoustic technology has been the focus of numerous research studies during the past few decades because of its simplicity and potential use in renewable energy systems. The Computational fluid dynamics (CFD) simulations play a critical role in the numerical modeling of thermoacoustic phenomena, but the basic problem concerning a CFD simulation of a whole system is the computational cost. In this paper, CFD and thermoviscous acoustics simulations are performed, and the results are compared in terms of simulation accuracy and computational cost.

Since the tire is driven by frictional contact with road surfaces, the contact condition between tire and road is a very important factor for the tire performances. Conventionally, the finite element analysis is used for predicting or evaluating the tire performances, such as the tire cornering force, traction and breaking forces, that road models is constructed by a flat surface. However, considering of tire rolling on real road surfaces, it is difficult to predict or evaluate tire performances by using a conventional flat surface road model because typical real road surfaces are rough conditions and the grip force of the tire is caused by contact between rough road surfaces and the tire tread rubber. In this study, to evaluate the contact conditions between the tire and the road surfaces constructed by based on real road surface measurement data, the dynamic tire rolling simulation is carried out by a general purpose finite analysis software Ansys LS-DYNA.

Uncertainty Quantification (UQ) is expected to be the key to evaluating the realistic reproducibility of models. However, the computational costs associated with considering a large number of variability factors and establishing the accuracy and validity of simulation models can be challenging. This paper introduces the integration of deterministic numerical simulations with measured data, utilizing machine learning and the uncertainty quantification tool SmartUQ. This approach makes it possible to quantitatively evaluate various uncertainty factors and easily introduce probabilistic analysis to understand complex real-world phenomena.

There is growing interest in using discontinuous fiber reinforced thermoplastics (FRP) in the production of mass-produced automobiles. FRP is a material that combines the benefits of lightweight and easy-to-process resin with the high strength of reinforced fibers. However, the stiffness and strength of FRP can vary depending on the direction (orientation) of the fibers, which are distributed anisotropically. This can have a significant impact on the accuracy of structural analysis. While simplifying the analysis model may reduce computational costs, it can also decrease accuracy. To address this, a Verification & Validation process was used to evaluate the accuracy of finite element analysis models using multi-scale analysis, taking into account the mechanical properties of both the resin and fiber in FRP materials.

Flow in human bronchus was experimentally and numerically investigated. Curved pipe was connected bronchial entrance to simulate human pharynx. The flow around the first branch of the bronchus was examined by experiments and numerical simulation. In the experiment, velocity distribution was measured by PIV. In numerical simulation, three-dimensional RANS calculation was performed. Separated flow and flows toward two branches are confirmed by experiments in the first branch region. Cross-sectional distributions of turbulent kinetic energy and pressure was obtained by numerical simulation. The effects of Dean vortices on these flow structure were investigated.

A new finite element formulation based on the edge center-based strain smoothing element using 4-node tetrahedral mesh (EC-SSE-T4) is presented for large deformation analysis of nearly incompressible solids. The present method combines EC-SSE-T4 and the cyclic smoothing technique using the selective reduced integration (SRI) to avoid numerical issues caused by incompressibility. EC-SSE-T4 helps to obtain accurate deviatoric strain/stress distributions, whereas the cyclic smoothing technique helps to suppress pressure checkerboarding; then, the volumetric locking is resolved by the SRI.

A method is proposed to easily and accurately calculate incremental variational formulations (IVFs) that describe the behavior of inelastic materials. The IVF requires first- and second-order numerical differentiation to obtain the internal variables by the Newton-Raphson method. In addition, first- and second-order numerical differentiation is also required to obtain stresses and consistent tangent modulai. Therefore, in this paper, we investigate the implementation of generalized CSDA (Complex-Step Derivative Approximation), which is highly accurate and easy to implement, for these differential operations. To confirm the validity of the proposed method, we applied it to an analytical example of a large elasto-plastic model. As a result, the proposed method was able to capture the behavior of the elasto-plastic model with high accuracy, and the accuracy was equivalent to that of the Hyper Dual Number (HDN) reference solution. Furthermore, the convergence of the nodal force residuals was also good, indicating that the analysis was performed with high accuracy.

Material testing of rubber requires the measurement of uniaxial and multiaxial properties, but multiaxial property measurement requires a dedicated testing machine. Therefore, the measurement of multiaxial properties by compression testing is discussed.The compression test is one of the test methods to measure the mechanical properties of rubber. The compressive load-strain relationship in compression tests may vary depending on the friction coefficient and Poisson's ratio, so care must be taken when applying the relationship to the identification of material constants. In this study, numerical experiments using FEM analysis were conducted using JIS compression tests as an example to investigate the dependence of the measurement results on each property. As a result, it was confirmed that the measurement results up to 25% compressive strain were independent of these properties in compression tests under conditions that most rubber materials satisfy (Poisson's ratio of 0.495 or higher and friction coefficient of 0.75 or higher). It was also confirmed that under those conditions, the compression test could measure differences due to multiaxial properties.

In this paper we report an efficient implementation method of the preconditioners for the full-wave electromagnetic finite element method using general purpose calculation libraries. Preconditioning methods suitable for parallel processing like an extended node patch preconditioner, which we have reported, require suitable implementation such that independent subsets are effectively selected and parallelized. The extended node patch preconditioner is sufficiently efficient when properly parallelized under sophisticated partitioning and processing scheduling. To optimize partitioning subsets into independent groups allotted to CPU cores and parallelization schedule we propose to use general purpose numerical analysis module which plays optimal assignment of each stencil calculation and ordering. We implemented this approach and verified the effectiveness.

This paper deals with a high-frequency electromagnetic - heat conduction coupled analysis of a numerical human body model. The iterative substructuring method has been considered to be an efficient parallel computing method.

This paper deals with a parallel finite element analysis by Hierarchical Domain Decomposition Method (HDDM) of an electromagnetic field problem with a quadruple precision floating point number. A Quasi-Minimal Residual (QMR) method for real symmetric matrices and a QMR_SYM method for complex symmetric matrices are implemented in the quadruple precision for the parallel computing. Then, the HDDM with the quadruple precision arithmetic for the electromagnetic field problem is implemented by the C language. As a result, the convergence of the iterative method is improved, and the minimum residual norm is reduced by approximately five orders of magnitude in a high-frequency electromagnetic field analysis.

This paper introduces a novel optimization method for the integrated design of electrical machines. The proposed method can simultaneously optimize the discrete variables that define the machine configurations and the continuous variables describing the detailed shape of parts. The discrete variables are represented by a tree structure and the continuous variables are determined at the leaf node using Monte Carlo Tree Search (MCTS). Multi-objective optimization is performed in conjunction with MCTS to obtain Pareto solutions composed of different machine structures and shapes. The obtained Pareto solutions are stored in the tree and evaluated to update the search strategy to effectively select the combinations of the discrete variables. This method is applied to the optimization of permanent magnet (PM) motors. It is shown that this method is effective for the multi-objective integrated optimization of the PM motors.

Bone maintains its strength by remodeling, which enables to repair microdamage and adapt its morphology to the surrounding mechanical environment. To effectively prevent a decrease in bone strength due to aging and diseases, it is necessary to understand how bone strength is maintained and decreased. However, the observation of changes in damage distribution and bone morphology within a living body is challenging. The objective of this study is to observe their changes through remodeling by simulating microdamage repair and mechanical adaptation. To achieve this, we constructed a mathematical model in which remodeling is regulated in response to damage and stress, and simulated remodeling in a cancellous bone. As a result, we observed the loss of some trabeculae, while other trabeculae maintained their load bearing function by mechanical adaptation concurrent with damage repair. The bone loss occurred because the rate of damage accumulation exceeded the rate of repair, leading to the spread of damage deep within the trabeculae. These results suggest that the loss of trabeculae through damage accumulation contributes to a decrease in bone strength. Our simulation platform, which enables to observe the process of microdamage repair and mechanical adaptation, will be a basis for understanding the effects of mechanical states on changes in bone strength by remodeling.

The blood flow in the renal microvasculature may be determined by vascular branching geometries that affects distribution of red blood cells. In this study, a two-dimensional particle method simulation of blood flow from the interlobular arteries to the afferent arterioles was performed to clarify the effects of an asymmetric branching geometry on the distribution of red blood cells. In a blood vessel model, an interlobular artery as the parent vessel bifurcated into 10 afferent arterioles as the daughter ones. Red blood cells and blood plasma were considered as blood components. As a result of simulations, red blood cells were attracted toward the daughter vessels. This caused uneven hematocrit distribution over daughter vessels. In the case that all daughter vessels branched in the same direction, the hematocrit decreased toward the downstream; in contrast, hematocrit increased if daughter vessels branched alternately in opposite directions.

The phase-contrast magnetic resonance imaging (PC-MRI) is a diagnostic tool capable of providing valuable insight into physiological and pathophysiological flows, but due to its multimodal acquisition process and large range of parameters, the sources of the intrinsic artifacts are challenging to distinguish. Furthermore, the physical principle of the PC-MRI (magnetization precession and relaxation), as described by the Bloch Equation, does not possess analytical solution in flow fields due to non-zero convective terms. This issue is often bypassed by the use of a Lagrangian approach, capable of solving for individual particles along the flow, but limited due to the high computational cost and reduced resolution. Thus, this study aims to develop a Eulerian approach to solve the Bloch equation in flow fields. The Bloch equation was discretized by the Discontinuous Galerkin Method (DGM) and solved in a step-by-step manner. Numerical examples of 1-dimensional magnetization motion in a constant velocity showed that the L2 relative errors of the numerical solution was below 1.5% when compared to a single particle trajectory, and below 1.9% when compared to a grid throughout the domain with low undershooting and overshooting observed at discontinuities.

Phase-contrast magnetic resonance imaging (PC-MRI) is a powerful tool for flow measurements in vivo and provides a volume-averaged velocity vector in each voxel of the MRI images (partial volume effect: PVE). Several flow data assimilation approaches between MRI and computational fluid dynamics (CFD) simulations have been proposed, while the observation model of the MRI velocity fields is not well established and the PVEs of the MRI images have been not fully considered. Therefore, this study evaluates the impacts of the PVE on MRI flow data assimilation. The flow data assimilation was formulated by minimizing the error between MRI and CFD flow velocity fields with respect to the inlet boundary following the adjoint variable method. Here, we considered two types of error functions with and without PVE and conducted numerical examples of the two-dimensional steady flow fields with three cases of MRI resolutions. Numerical examples successfully demonstrated that the effects of PVE are non-negligible in the case of low-resolution MRI images.

In this study, we revealed a cause of overestimation of the permeability coefficient, and proposed a method to correct it. The theoretical solution of a one-dimensional mass diffusion problem with a permeable membrane was dealt to consider the cause of the overestimation. A relational expression between the permeability coefficient and the concentration gradient indicates that the overestimated concentration gradient increases the error of the permeability coefficient than the underestimated one when they are comparable. The treatment for this overestimation of the permeability coefficient is discussed in the presentation.

The cerebral microvascular network plays an important role in oxygen supply to the cerebral cortex and has a complex and characteristic structures. The physiological significance of these characteristic geometries are open questions. This study aims to develop a computational model of the cerebral microvascular network based on oxygen transport efficiency using topology optimization. We considered the topology optimization problem of the blood flow domain in the cerebral cortex to maximize the oxygen transport efficiencies in the brain tissue. Numerical examples demonstrated that acquired vessel structures became complex with many local branching with increasing the inlet pressure. These complex vessel shapes led to induce uniform oxygen supply in a whole brain cortex

Autonomous Vehicles (AV) and connectivity technologies have been on the rise these days. On April 1st, 2023, LEVEL4 AV has been legalized to drive on public road in Japan. AV has a lot of features to contribute to solving social problems such as social welfare for vulnerable road users, lack of labor force, and alleviation of traffic jams, etc. In this study, the mixed-Autonomy condition: an environment that both AV and Human-Driven Vehicles run around the network, is in focus. Altruism in the route choice process is implemented on the microscopic traffic simulator, and the influence of CAV on the traffic network has been evaluated.

Series of traffic accidents and traffic congestions happen every day in big cities like Tokyo. Therefore, it’s necessary to simulate the traffic condition and research on a method to control vehicles’ behavior well. Besides, autonomous driving is being developed rapidly nowadays and researchers often use deep learning to study trajectory prediction and path planning for autonomous vehicles. In this research, we use the shortest path search algorithm and deep reinforcement learning to control vehicles’ behavior in a traffic simulator SUMO. Regarding the local behavior which contains their speed and acceleration, we utilized deep reinforcement learning to control it. Regarding global behavior, which is path planning, we used a method combining Dijkstra algorithm and deep reinforcement learning. The vehicle agents in the simulator have better behavior after training. They can have acceleration and path selection that shorten their driving time when they encounter different traffic situations.

From a geometric point of view, it is known that there are similarities between dislocations and disclinations. In this study, we consider the stress field difference of dislocation array and disclination dipole is not only in geometry but also in mechanics. We constructed models of dislocation arrays and disclination dipoles, and numerically analyzed their stress fields to verify the stress field difference of dislocation array and disclination dipole. It was found that the dislocation density of the dislocation array becomes larger, the difference between the stress field of the dislocation array and the disclination dipole becomes smaller. And the magnitude of the Frank vector was found to have no effect on the difference between the two stress fields. We also verified whether interaction exist or not between lattice defects. For those purposes, two models with the same dislocation interval but different arrangements were constructed, and the stress fields of each were analyzed numerically. The stress fields of the two models showed different distributions, indicating that there is interaction between dislocations.

TRIP steels have an advantage as structural materials in which strength and ductility can be combined through deformation-induced martensitic transformation. However, these properties on grain-size dependency have not been clarified. In this report, we introduce a grain-size dependency into the energy barrier of phase transformation in the phase-field and dislocation-crystal plasticity model proposed by the authors previously, then investigate computationally the effect of grain size on strength and ductility of TRIP steels from the viewpoint of microstructure formation and dislocation accumulation.

This study integrates machine-learning potentials (MLP) into crack propagation simulations, comparing the obtained atomic structural changes and stress distribution in four crack systems with the ones by embedded-atom-method (EAM) and modified-EAM (MEAM) potentials. The MLP predicts stress concentration at crack tips, with increased load leading to bond breakage and cleavage, implying brittle fracture. In contrast, EAM and MEAM suggest ductile failure, with observed dislocation ejection and gradual crack opening due to structural changes. Overall, MLP exhibits a distinct fracture tendency compared to conventional potentials.

The computational simulation of the plastic deformation behavior of the polycrystalline material accompanying the friction is performed using the second-order homogenization finite element (FE) method. The dynamically updated boundary condition (BC), which can describe the relative slip displacement under counter-friction force, is introduced to reproduce the progress of nodal displacement and force BCs during the rolling process. To directly solve the displacement under the BC accompanying friction force, the components of the stiffness matrix constructed by the macroscopic FE structure are edited using the friction coefficient and the contact angle. The elasto-viscoplastic mechanical behavior of the microscopic polycrystalline structure is described by the conventional crystalline plasticity FE method, and the relative scale-depended micro- to macroscopic FE model is established using the second-order homogenization method. The computational simulations of the rolling process with different friction coefficients were performed using the established FE model. The computational results represent increase and decrease in the rolling force owing to changes in the contact area.

Thermoelectric conversion using the anomalous Nernst effect (ANE) has excellent potential for application in energy harvesting technology. Compared to the typical thermoelectric effect known as the Seebeck effect, the thermoelectric figure of merit in ANE is about ten times smaller; material search with a large ANE response is a critical challenge aiming to achieve its application. In this study, we implement a high-throughput material search for evaluating ANE and perform this material screening up to 1400 magnets. We find that D03 type Heusler compound Fe3Ga shows 6 microvolts per kelvin at room temperature. Motivated by the recent discovery of a large ANE material, we also analyze the origin of the enhancement of ANE from the topological electronic structure. We find that not only the Weyl points but also stationary points in the energy dispersion of the nodal lines play a crucial role in enhancing ANE. These present results will give us a useful guiding principle to design magnets showing a large ANE.

The delamination energies of bimetal interface are calculated by DFT calculation, substituting an atom on the interface with carbon, phosphorus, silicon and vacancy. The delamination energies are calculated with the supercell of stacked/separated 2x2x5 unit lattices of various bcc (001)/fcc(001) combination. The effect of the atom substitution is also investigated with the supercell for the bcc(001) and fcc(001) surface. The tendencies of surface and delamination energies are not same, suggesting the bonding effect at the interface.

Graphene sheets (GS), which is one of low-dimensional nanocarbon materials, and the wrinkling and bending deformation induced by lattice defects can significantly affect the mechanical properties of GS. Therefore, it is essential to understand the deformation mechanism of wrinkles and bending in order to control the mechanical properties of GS. The purpose of this study is to evaluate the validity of introducing the continuum theory equation used for colloidal films to monolayer GS with lattice defects, which can derive strain energy from parameters related to curvature, in order to elucidate the deformation mechanism of GS. Comparison of the strain energy calculated by continuum theory with the potential energy calculated by the molecular dynamic simulation shows that the energy distribution is similar in the elastic range of the model. This result is expected to contribute to the elucidation of the deformation mechanism of GS and further development of new materials.

Indentation cutting are performed on amorphous PP, PE, PS, and PET with a rigid Fe tool by molecular dynamics simulation. Amorphous blocks or “work” are made by randomly growing molecular chains and compressing them with elastic walls. During indentation, the amorphous work bent in V shape, and doesn’t show remarkable reaction force against the Fe tool. Molecular chains of PET and PE are straightened just below the cutting edge, while those of PP and PS keeps rounded shape due to side group of CH3 and C6H6.