Recent advancement in parallel computing enables the precise finite element analysis of steel frames using solid elements. However, the finite element models used in the analysis have not been verified sufficiently. In this study, first, static finite element elastic-plastic buckling analyses of a square steel tube column subjected to a prescribed lateral displacement are performed with different meshes in order to verify the analysis model. It is shown that the accuracy and computation time of the analysis depend not only on the number of mesh divisions but also on the aspect ratio of each finite element. Then, dynamic elastic-plastic buckling analyses are performed for different meshes and different time increments. In the dynamic buckling problem, which is a kind of slow dynamics problem, accurate results can be obtained using a fine mesh but with a rather large time increment.

View full abstract

We have proposed a new method of forecasting pollen dispersal by combining the non-linear least squares method with neural networks. Because making highly accurate estimates of pollen count in cedar forests is critical for improving the accuracy of forecasting pollen dispersal, we propose a method to estimate pollen count by using observed aerial pollen concentrations measured in living areas and using convection-diffusion equations to calculate aerial pollen concentrations that reproduce the observed values. To forecast pollen dispersal, we made use of neural networks’ learning and decision-making functions. We created neural networks that could use their learning function to estimate the pollen count from the weather condition in areas of interest. To estimate the amount of pollen dispersed in an area of interest on a specific date, we used the most suitable neural network that can estimate the pollen count to estimate the pollen count on that specific date, and then calculated the aerial pollen concentration in the area of interest. We confirmed that the calculated concentrations tended to match the observed concentrations. Thus we confirmed the feasibility of a new method of predicting aerial pollen concentration that combines the non-linear least squares method with neural networks.

View full abstract

As a Lagrangian meshfree method, the MPS(Moving Particle Semi-implicit) method has been shown useful in engineering applications widely. In this paper, by using the Taylor series expansion, generalized schemes for any order spatial derivatives with higher order consistency, convergence, and completeness conditions are developed. Applying new schemes for numerical tests, calculating first and second derivatives of linear and non-linear functions, demonstrates higher order convergence regardless of whether particles are distributed regularly or randomly spread involving domain boundaries. Furthermore, application of new spatial derivative schemes enhances computational accuracy and stability for numerical analysis of incompressible flow with the free surface.

View full abstract

A 3-dimensional multiphysics coupling scheme is developed for analyzing fluid-structure interaction problems based on the explicit MPS method and Hamiltonian MPS method. The interaction between the fluid and the structures is symmetrical. For verification the coupling scheme, a buoyancy case, an elastic baffle case and an elastic gate case are calculated. The results are compared with the experiment and other methods’ results.

View full abstract

A welding process simulation considering the weld pool and dropped metal during welding which have a fluid flow with thermal conduction and a free surface is performed by using the hybrid particle and grid method with Explicit MPS. In this study, first, the applicability of Explicit MPS to the welding process is studied. Second, a hybrid particle and grid method is developed. In the hybrid method, particles can be used in the weld pool and the area located near the weld pool, while grid elements are used in the other areas. For high-speed simulation of the weld pool, all of calculation processes are parallelized and accelerated by a GPU (CUDA). The welding process simulation using the hybrid method can reduce the calculation cost reasonably within 1 to 2 % accuracy or so on.

View full abstract

The purpose of this paper is to improve efficiency of multi-objective topology optimization using bar-system representation Genetic Algorithm (GA). We propose a new GA using elite initial individuals produced using a SIMP (Solid Isotropic Material with Penalization) method with a weighted sum method. The SIMP method for a multi-objective topology optimization is one of the most established methods that use the sensitivity analysis. Although the SIMP method is easily implemented and it is computationally effective, it may be difficult to find a proper Pareto-optimal set in a multi-objective optimization. In the present paper, GA is adopted to obtain the Pareto-optimal set. To build more evenly distributed global Pareto-optimal set and reduce GA computational effort, new individuals that resemble topology of the Pareto-optimal set of SIMP are introduced for initial pool of GA. The proposed method is applied to a structural topology optimization example and compared with the results of the traditional method that uses standard random initialization for initial pool of GA.

View full abstract

Alkali Silica Reaction (ASR), chloride damage of reinforced steel and other aging degradations in concrete structures are strongly dependent on both of chemical phenomena and mechanical behavior including fracture. In this study, a coupling simulator has been developed in order to investigate the degradation mechanism of these concrete aging damages. In the coupling simulator, non-stationary diffusion analysis for Alkali ion is solved firstly, and then the expansion force act on the aggregate is evaluated by referring consistency of the alkali ion. In the second step, nonlinear damage analysis is conducted. The nonlinear damage analysis evaluates cracking region with the damage parameter. The following diffusion analysis changes diffusion coefficient in the cracking region by referring the damage parameter. The above procedures are coupled at each time step.

View full abstract

In this paper, numerical integration technique of meshfree approach for plate bending analysis is discussed. Stabilized conforming nodal integration (SCNI) is high accuracy and efficient technique for integrating the stiffness matrix in the meshfree approach. However, high gradient region of stresses/strains cannot be accurately represented in the plate bending analysis when SCNI is adopted. To overcome the problem, sub-domain stabilized conforming integration (SSCI) is introduced. Hybrid technique of SCNI and SSCI is also proposed to enhance the accuracy and the computational efficiency. Some plate bending problems are demonstrated and the quantitative evaluation is shown in the numerical examples.

View full abstract

In this paper, the authors propose the Free-Kick Support System, which solves the inverse problem of the free-kick simulator. The free-kick simulator is the system which physically computes the motion of soccer ball by the information of ball position, velocity, rotating angle, rotating speed and the aerodynamic properties of soccer ball. Using the Neural Network (NN) and the Self-Organizing Map (SOM), the Free-Kick Support System solves the inverse problem of the free-kick simulator and the output is available for the practice of the free-kick and making a strategic plan of real soccer games. Furthermore, the authors propose the evaluation method of the modular network SOM (mnSOM) for quantitatively evaluating the learning effects, and show that the mnSOM is useful for the Free-Kick Support System.

View full abstract

This paper proposes a new discrete analysis method for fracture simulation using structural elements based on the discrete element concept. The suggested structural elements are suitable for fracture simulations because the element stiffness is evaluated by means of relative displacements at the interface between discrete elements (particles). We first formulate the structural elements in consideration of the position and the angle of interface between discrete elements and discuss the integration of element stiffness. The fundamental properties for elastic problem are examined in chapter 3. The eigen-values and eigen-vectors of stiffness matrix are analyzed to investigate the relationship between basic deformation mode and integration order. Also the accuracy of the structural element for beam bending problem is assessed. Finally, after explaining the modeling of fracture in this paper, two representative numerical examples are presented to demonstrate the applicability to fracture simulations for structures involving multiple cohesive cracks.

View full abstract

Particle method such as Smoothed Particle Hydrodynamics (SPH) and Moving Particle Semi-implicit method (MPS) can handle quite complicated physical problems involving dynamic changes of free surface and crack propagation, and it is wildly expanding its applications not only in the fluid dynamics but also in the solid mechanics. Beside of these advantages, particle methods are not so easy to treat boundary conditions, like pressure Neumann condition and slip or no-slip condition in fluid dynamics. This is one of the typical difficulties in mesh-less type method. In addition, particle simulation model may include an incompatible step-shaped boundary line, which is made by using a simple pre-processing of particle model. Although the simple and robust pre-processing is one of the advantage of particle simulation, the step-shaped boundary may generate un-realistic numerical solution across the real boundary line especially in the fluid dynamics . Recently, pressure Neumann condition in SPH is re-focused with fixed ghost boundary method using virtual makers. In this paper, the fixed ghost boundary method is modified to treat the incompatible step-shaped boundary particles by referring the real physical boundary line. The accuracy and efficiencies of proposed method are validated by comparison between a numerical solution and experimental results.

View full abstract

This paper describes the development of a parallel model refinement tool REVOCAP_Refiner and its application to efficient large scale finite element analyses. The model refinement tool is implemented into parallel solvers as common library, and model refinement is performed completely in parallel without any communication among processors. A user does not need to handle the refined large scale model directly, and a process of producing large scale mesh can be shortened dramatically. To demonstrate its practical performances, it is implemented into a parallel solid solver named FrontISTR and a parallel flow solver named FrontFlow/blue, respectively, and some numerical examples are given.

View full abstract

The domain decomposition method (DDM) is a well-known effective method for parallel computing of the finite element analysis. Several studies have considered applications of the DDM to various phenomena such as stress and deformation, fracture, temperature, viscos flow, electromagnetic fields, and others. Besides, the DDM needs to solve a Schur complement equation using a preconditioned iterative method. Although a simplified diagonal scaling preconditioning is widely used for the Schur complement equation, its simplicity has never been discussed. Therefore, this study focuses on the diagonal scaling preconditioning for the DDM. We propose a new explanation of the simplified diagonal scaling preconditioning, and some numerical examples are demonstrated.

View full abstract

This paper presents a structural design method of metallic waveguides loaded with ferrite material using the level set-based topology optimization method. Metallic waveguides are widely used to control the wave propagation, especially in microwave application, such as waveguide filters, T-junctions. Ferrite materials display a frequency-dependent permeability due to magnetic resonance phenomenon that can be altered by changing the magnitude of an externally applied DC magnetic field. Thus, metallic waveguides with ferrite inclusions are expected to offer some advantages, such as tunable operating frequencies. Anisotropic permeability tensor of ferrite is modeled using Landau-Lifshitz model in analysis. The optimization problem is formulated as to maximize the transmission power at prescribed frequencies using S-parameters. To confirm the validity and utility of the presented method, we apply it to the waveguide filter design problem and T-junction design problem, and the results show that the presented method successfully finds optimized configurations that maximize transmission power of waveguides.

View full abstract

This paper proposes a new isotropic damage model for quasi-brittle materials and demonstrates its performance in crack propagation analysis for concrete. The suggested damage model is based on fracture mechanics for concrete, and is therefore capable of simulating the quasi-brittle fracture behavior with the fracture energy. We first formulate the damage model in 1-D problem by borrowing the ideas of traction-separation law based on the fracture energy of concrete. And then the damage model is extended to 2-D or 3-D problems by introducing the modified von-Mises equivalent strain. The fundamental characteristics of the suggested damage model are examined in chapter 3. We evaluate the energy balance in 3-point bend test specimen with a single-edge notch, and verify the mesh objectivity of finite element solution with the damage model. Finally, a benchmark test characterized by mixed-mode fracture, which is called Nooru-Mohamed Test, is performed to demonstrate the capability of the suggested damage model.

View full abstract

Adiabatic quantum computation has been proposed as quantum parallel processing with adiabatic evolution by using a superposition state to solve combinatorial optimization problem, then it has been applied to many problems like satisfiability problem. Among them, Deutsch and Deutsch-Jozsa problems have been tried to be solved by using adiabatic quantum computation. In our previous paper, it has been shown that the adiabatic quantum computation in Deutsch problem is modified by using a cubic step function instead of a linear step parameter. In this paper, it is proposed to solve Bernstein-Vazirani problem more efficiently by the same cubic method to obtain a solution with higher observation probability of 99.6%.

View full abstract

In this study, performances of a projection method for the interaction system of an incompressible fluid and a flexible elastic structure are evaluated. The proposed projection method splits the monolithic equation system into the equilibrium equations and the pressure Poisson equation (PPE) using the intermediate velocity in the nonlinear iterative procedure. The monolithic equation system has the drawbacks of the large degrees of freedoms and the ill-condition. The proposed formulation reduces these drawbacks. The performance of the proposed projection method is demonstrated using a channel with a flexible flap, which is one of typical test problems, and a flapping flexible wing in a closed tank with comparable fluid and structural mass densities. It follows from the results for these numerical examples that the proposed projection method is efficient for the cases, where some conventional monolithic and partitioned methods would suffer the numerical difficulties.

View full abstract

The multi-phase-field method is one of the most powerful numerical simulation methods to study microstructure evolutions in practical polycrystalline materials. However, the computational cost for the multi-phase-field simulation is much higher than the conventional phase-field simulation, since the same number of time-evolution equations with multiple phase-field variables must be solved. In this study, a multiple-GPU computing technique using a programming language CUDA and the MPI library is newly developed to accelerate the multi-phase-field simulation. In order to hide communicational time for CPU-to-CPU and CPU-to-GPU communications, we propose an original overlapping method between computation and communication. Furthermore, we decompose whole computational domain into sub-domains to distribute computational load to multiple GPUs. The multi-phase-field computation for each sub-domain are efficiently performed by using multiple stream executions in CUDA. The three-dimensional grain growth simulation performed using our multiple-GPU computing technique with the overlapping method demonstrates that the communicational time can be hidden completely and good weak and strong scalings are achieved.

View full abstract

Moving Particle Semi-Implicit method (MPS) is applied to the analysis of one and two dimensional scalar wave equations. The author and his coworkers are going to analyze the non-linear characteristics of waves and the physical non-linear phenomena caused by waves. Particle methods, e.g. MPS or Smoothed Particle Hydrodynamics method (SPH), are mesh free numerical methods and well suited to analyze the large non-linear dynamics. But wave analyses by MPS are minimal; therefore we will address linear waves and investigate the wave characteristics of the analysis using MPS. First, we inquired into the stability conditions and the dispersion characteristics of MPS wave analysis. In the process of the analysis, we found if the particles are positioned regularly and r_e=2Δh in one-dimensional and r_e≤√2Δh in two-dimensional, the formulation by the MPS is as same as the central difference scheme of a finite difference method. Here, Δh and r_e is a distance between neighboring particles and a radius of weight function, respectively. Related to the stability conditions, an interesting phenomenon was found that in spite of the Courant number α>1, stable cases existed when r_e was large. Second, we investigated the dispersion properties vs. parameters α, θ and r_e in detail. Here, θ shows a propagating direction. Consequentially, in order to achieve the phase velocity error ε[%]<1, a wavelength has to be discretized about over 20Δh, under conditions of α<1, r_e≤4Δh in one dimensionin case, and α≤1 ⁄ √2, r_e≤3Δh in two dimensionalin occasion. Last, the results of sound wave propagation analyzed by MPS were good agreed to the analytical predictions.

View full abstract

This paper presents a temperature-dependent visco-hyperelastic analysis scheme for pressure-sensitive adhesive with an Eulerian finite element method. All the basic equations are numerically solved in the Eulerian framework because it allows arbitrarily large deformations. Visco-hyperelasticity is formulated using Simo’s finite-strain viscoelastic model, where hyperelasticity is modeled as a strain energy function based on Yamashita-Kawabata model. The left Cauchy-Green deformation tensor is temporally updated from the Eulerian velocity field without material points. Temperature-dependence is described with the time-temperature superposition principle of Williams, Landel, and Ferry. To validate the proposed approach, we simulate uniaxial tension tests under different tensile speed and temperature conditions.

View full abstract

Generally, wall boundary model in particle method is modeled by using fixed particles. However, high resolution is required from the viewpoint of shape reproducibility when boundary is complicated geometry. On the other hand, the boundary model for a wall expressed by triangle patches utilized for CAD has been often used in MPS based commercial codes. The problem is improved by the approach to some extent in that case. However, there is no useful method to deal with the large deformation of wall boundary. In this research, the wall boundary model which does not use a distance function was developed, and it was applied to the SPH method. Furthermore, it was shown that the fluid behavior in the wall boundary with large deformation as a restriction condition could be calculated. The validity of the proposal calculation method was presented.

View full abstract

The present study proposes an analytical sensitivity analysis for a so-called multi-scale topology optimization introduced to minimization of compliance of three dimensional structural problems. The multi-scale topology optimization is a strategy to optimize topology of microstructures applying a decoupling multi-scale analysis based on a homogenization method. The stiffness of the macrostructure is maximized with a prescribed material volume of constituents under linear elastic regime. A gradient-based optimization strategy is applied and an analytical sensitivity approach based on the adjoint method is proposed to reduce the computational costs. It was verified from a series of numerical examples that the proposed method has great possibility for microscopic advanced material designs.

View full abstract