We propose a characteristic Galerkin scheme using B-spline basis functions for advection diffusion problems. One of significant features of the B-spline basis functions is that the B-spline basis functions of p degrees have at most C^p-1 continuous derivatives although classical finite elements based on Lagrange polynomials have only C⁰ continuous derivatives. The feature works better for the characteristic Galerkin scheme. In order to confirm effectivity of the present scheme, we perform some numerical experiments and compare results of the present scheme with the SUPG scheme using the same basis functions.

View full abstract

To farther accelerate the three-dimensional two-scale stress analyses of composite material with iterative conjugate gradient method and voxel finite element method, we propose a code implementation for GPU. In the framework, several specialized algorithms can be employed; i.e., element-by-element technique utilizing explicit element-to-nodes connectivity, FFT pre-conditioner accelerating the convergence of the iteration, and so on. These algorithms are implemented for GPU computations by CUDA to gain more speed compared to PC computations. Numerical experiments have exhibited that the GPU computations are several times faster than the PC computations.

View full abstract

The estimation of risk due to rockfall is often done empirically. Rockfall simulation helps to describe the motion of rockfall on a slope. This paper details a typical simulation method and analysis of the manner of rockfall. Fundamental idea is concretely shown to consider the roughness of the slope surface to reproduce the state that is close to the real slope. A treatment method of the roughness at the moment of the collision between a rockfall and a slope was developed. The method expressed the roughness of the slope by adding an angle in the normal vector of the triangular plane. The validity of proposed simulation method becomes clear by the application to rockfalls on a simple slope.

View full abstract

The present study addresses a numerical method for estimating macroscopic material properties by predicting the time variation of a three-phase porous microstructure, composed of Nickel (Ni) and Gadolinium doped ceria (GDC), due to sintering. The phase-field method (PFM) is applied for simulating the sintering process and the effect of creep deformation of the constituent materials is taken into account by introducing the strain energy to the PF free energy functional. The numerical results show that the temporal changes by sintering of the microstructure decrease the total length of triple phase boundaries and that the strain energy deteriorates the wettability between Ni and GDC more than a little.

View full abstract

Large scale analysis with Finite Element Method (FEM) needs much memory because of construction of a global matrix. This motivates us to employ Element by Element (EbE) scheme, which does not require to construct global matrix. However, in the EbE scheme, general useful preconditioners for global matrix cannot be applied. To enhance effectiveness of the EbE scheme, variants of the EbE schemes, such as Grouped EbE, Clustered EbE, and preconditionings for them have been proposed by many researchers. In the Grouped EbE scheme, parallel implementation of the EbE scheme becomes easier. The clustered EbE scheme, which is regarded as a generalization of standard EbE scheme, can decrease number of iterations. Therefore, in this paper, we propose an efficient Eisenstat SSOR preconditioning used in the Grouped EbE scheme, and demonstrate its effectiveness.

View full abstract

This paper proposes a new topology optimization method using a three-dimensional mesh generator based on the level set method. Basic details of the level set-based topology optimization method are briefly discussed. A topology optimization of the linear elastic problem is formulated using the level set method. Based on the formulation, the design sensitivities are derived using the adjoint variable method. Using the derived sensitivities, topology optimization algorithm is constructed where the Finite Element Method (FEM) is used to solve the equilibrium equations and to update the level set function. A numerical implementation for generating finite element mesh based on the level set method is proposed. Three-dimensional examples are provided to confirm the validity and utility of the proposed topology optimization method.

View full abstract

Dynamic response considering fluid structure interaction (FSI) is crucial in many engineering fields and some of the FSI phenomena are treated as an acoustic fluid and structure interaction (AFSI) problem. This paper describes a new parallel simulation system for the solution of large-scale AFSI problems using partitioned iterative coupling methods. Here, a new parallel coupling technique with partitioned iterative methods is developed, while employing existing parallel solvers of the ADVENTURE system, i.s. parallel structural solver, ADVENTURE Solid and parallel Poisson solver, ADVENTURE Thermal. The developed system runs efficiently in parallel computing environments and shows accurate and robust performance in solving large scale AFSI problems.

View full abstract

In order to exploit the potential applications of gold nanowaires, it is important to reveal their mechanical properties. Although Molecular Dynamics (MD) simulations can treat dynamic conformational modifications in atomic scale, the analysis time scale is strictly limited. In this paper, Parallel Replica (PR) method which can extend the analysis time scale is employed to investigate mechanical properties of a gold nanowire at slower strain rates. We performed tensile-loading simulations at wide strain-rate range (1.0×10⁵∼1.0×10¹²s^-1) using MD and PR method, and revealed that the deformation behavior changes depending on the strain rate. Furthermore, we predict the yield strain at practical strain rate based on the transition state theory.

View full abstract

Material searching technology using optimization technique has been developed in order to speed up material design. This technology employs a response surface method which is used in an optimized calculation instead of material search using experiments and simulation in the past. The proposed technology has two key features: (1) Kriging model which can approximate non-linear functions is used as a response surface method in order to search the target material, (2) the Kriging model can be constructed by a fewer sampling points by selecting appropriate points in displayed design space. The effectiveness of the proposed technology was demonstrated by applying it to the material searching of electrode filter. It was found that the proposed technology reduced the material searching time by 70% compared with that for previous methods.

View full abstract

This paper presents a new approach to optimize the turbine layout of wind farms. We have made use of the 3 dimensional CFD simulation to replace the wake superposition model, which makes the present study substantially different from the existing works. The CFD computation, which is the most time-consuming part and has been thought not affordable in a genetic algorithm (GA) based optimization system, is accelerated by GPUs. We have developed an automatic system for wind farm layout optimization on the Tsubame 2.0 supercomputer which is equipped with massive GPUs, and demonstrated that the implementation of the 3D CFD simulations in a GA optimization computation is practically possible if the CFD can be solved on GPUs at a speed dozens of times faster.

View full abstract

Adiabatic quantum computation has been proposed as a quantum algorithm with adiabatic evolution to solve combinatorial optimization problem, then it has been applied to many problems like satisfiability problem. Among them, Deutsch problem has been tried to be solved by using adiabatic quantum computation. In this paper, we modify the adiabatic quantum computation and propose to solve Deutsch problem more efficiently by a method with higher observation probability.

View full abstract

In this research, a new distributed memory parallel algorithm of the explicit MPS (Moving Particle Simulation) method is presented. The analysis region is divided for a distributed memory parallel computation using ParMETIS. Two communication techniques of an overlapping method and a non-overlapping method are estimated by parallel scalability tests. Since we find that load balance is most important for the distributed memory parallel algorithm of the explicit MPS method, we find that the non-overlapping method is more effective than the overlapping method. As a result, we have been able to do the MPS analysis of 268 million particles in 38 seconds per one time step. Performance during large scale simulation is examined by computing tsunami wave run-up on a virtual gulf area using up to 58 million particles.

View full abstract

An MPS formulation for Reissner-Mindlin shell is developed in the framework of Hamiltonian particle dynamics. The model conserves linear momentum, angular momentum and total energy of the system. Free vibration of a free-fixed end square plate is tested for the convergence of the formulation. Moreover, the computation time is reduced from cubic of the resolution that total three-dimensional process is needed to quadratic by using this model.

View full abstract

A thermodynamics-based constitutive model, which accounts for both crystallographic slip and deformation twinning, is developed for a single crystal of hcp metals within the framework of finite crystal plasticity. While the volume fractions of stress-free twin deformations are introduced as internal variables, the free-energy involves the bulk energy of separate phases and the surface energy at twin interfaces, which are introduced as functions of the internal variables, in addition to the standard hardening-related energy in crystal plasticity framework. After the formulation is described in detail, a series of numerical examples is presented to verify the performance of the proposed model in predicting the deformation twinning, the successive deformation process and the twinning-induced stress responses. The results are studied with reference to the theoretical consequences and the experimental results reported in the literature.

View full abstract

A patch antenna is a typical microstrip antenna composed of a dielectric block sandwiched between two thin metal plates, the metal patch and the ground plane. Although the shape of the metal patch critically determines antenna performance, the design of such shapes is often based on trial and error, and an effective method for obtaining optimized shapes is needed. Furthermore, as operating frequency increases, it becomes more difficult to numerically evaluate antenna performance, due to the skin effect, a phenomenon whereby the penetration of high frequency electric current into a metal surface decreases exponentially. In this paper, we construct a topology optimization method of metallic patch antennas that obtains high performance configurations. In this method, a transition boundary condition is implemented so that the skin effect can be treated as surface impedance. As a result, the proposed method can successfully optimize metal shapes that deliver high antenna performance.

View full abstract

The forming mechanism of the factory-roof fracture in circumferential notched specimen is numerically investigated by using the smeared crack model based on the finite element method. In this numerical investigation we obtained the appearance of factory-roof like cracks in the specimen with the steep notch, and we numerically found out the critical condition of notch-slope for forming the factory-roof fracture patterns. Furthermore, the relationship between angle of external-torsion and numerical iteration were investigated. As a result, we obtained some characteristic tendencies for each case of the fracture shaped like an X or a factory-roof.

View full abstract

Distribution of the pre-existing cracks has an influence on fracture patterns in rock mass. Although this complex distribution can be considered on the basis of the Discrete Fracture Network (DFN) model, there are no numerical techniques with which we can simulate a rock mass with both high accuracy and high robustness for crack propagation based on the model. In this study, we propose a novel discrete crack propagation analysis method for the DFN model. In this method, the pre-existing cracks are assumed at edges of triangular element, and both of the Generalized finite element approximation for continuous areas and the Moving Least Squares approximation for discontinuous ones are employed. Some numerical examples by using the proposed method are shown in this paper.

View full abstract

An immersed boundary method for simulating compressible viscous flows is presented. The boundary conditions on the immersed boundaries are imposed by a ghost point treatment. The immersed boundaries are represented as a sharp interface. An adaptive selection technique of interpolating polynomials is used to evaluate the values at the ghost points. The present approach effectively avoids numerical instabilities caused by matrix inversion and leads to a robust means of interpolation in the vicinity of the boundaries. The immersed boundary method is implemented in a finite-difference solver for the direct numerical simulation of the compressible Navier-Stokes equations on non-staggered Cartesian grids. The accuracy and fidelity of the solver are examined by the three-dimensional numerical simulation of the thermal convection in a rotating spherical shell. The numerical results are compared with a well-resolved simulation on the spherical coordinate grids.

View full abstract