This paper proposes a new method of genetic algorithms (GAs) for discrete optimization problems. For discrete optimization problems, the performance of Distributed GAs (DGAs) are not so good. We propose a new method of increasing the performance of DGAs for discrete optimization problems. The features of the proposed method, Global Crossover based DGA (GCDGA), are multiple crossover operations applied to the elite individuals and DGA without migration. We apply GCDGA to job-shop scheduling problems (JSPs). The experiments on JSPs showed that GCDGA has a better performance than the conventional GAs, and GCDGA provides an efficient distributed scheme in GAs for discrete optimization problems.
In the previous papers(13, 14, 15), I introduced new stress rate, namely, linear stress rate in order to formulate nonlinear beam element. This stress rate is non-symmetric, but its material stiffness matrix holds symmetry. Then, it has only the geometric stiffness of rigid rotation as the geometric stiffness. Because the study of the strain and the constitutive equation for this stress rate is necessary, I will show that the stress rate is consistent with linear strain, which is linear about the right stretch tensor, and that the constitutive equation consists of the linear strain and the transpose of the first Piola-Kirchhoff stress. Then, the simple shear problem will be shown to consider this strain.
Proposed method can reduce hardness for conventional receding horizon control to treat complicated state equation. Gradient method had a defect to take long time to converge into optimal solution. This can be turned into advantage by the proposed method. This paper describes how to treat gradient inside Receding Horizon Control. Gradient method can be operative in real time control using the proposed method.
The purpose of this study is to develop a method to reduce residual vibration which the base excitation due to a rigid motion causes when a flexible structure reaches a target position under the rigid motion. The vibration of structures is composed of vibrational components of infinite modes. The first mode of vibration is a target in reducing the residual vibration in the present study. Conditions on acceleration of a base excitation due to a rigid motion not to cause residual vibration were derived for a mechanical system of single-degree-of-freedom. The validity of the derived conditions was verified performing calculations of time history responses when a mechanical system moves at the acceleration satisfying their conditions.
In recent years, the Lattice Boltzmann Method (LBM) has been developed as an alternative numerical approach in Computational Fluid Dynamics (CFD). In particular, this method is promising for simulations of multiphase and multi component fluid flow involving complex interfacial dynamics. Unlike the conventional CFD methods based on NS equations, the LBM is based on the mesoscopic particle’s kinetic equation. This method has some advantages such as the simplicity of the algorithm (high efficiency on parallel processing), flexible reproduction of interfaces between phases. The conventional LBM, however, requires regular structured grids. But, for complex flow field, unstructured grid should be used. In this paper, we describe a computational scheme based on two-dimensional unstructured grids from the point of view of Finite Volume Methods (FVM). As examples of tests of this scheme, numerical simulations for two-dimensional Lid-driven cavity flow, and Motions of drops in a shear flow are presented.
A software infrastructure was developed with distributed object technologies that enable us to connect heterogeneous computing environments. The software infrastructure, which was implemented with CORBA, Java, C++ and XML, establishes seamless connection between structural analysis components. The components include model generator, matrix calculation solver and visualization tool which are often running on the various computing hardware such as parallel computer, graphics workstation and so on. Software wrappers are integrated with the components in order to provide multi-protcol networking and asynchronous job control. A multi-faced problem solver for structural design and an eigenvalue analysis system for rotating shaft were developed onto the infrastructure to evaluate its applicability.
Numerical analysis was performed to estimate fracture toughness values of SM490A steel in the wide temperature range including the ductile-brittle transition region. Microscopic fracture surface observation revealed that cleavage fracture was observed on the fracture surface at the lower shelf and transition regions, while dimple fracture was observed at the upper shelf region. In the numerical analysis, void volume fraction was taken as a dimple fracture criterion, and a set of critical plastic strain, stress triaxiality and maximum principal stress was taken as a cleavage fracture criterion. The numerical analysis showed that the dimple fracture took place in the upper shelf region. The dimple fracture toughness values agreed well with the experimental results. In the transition and lower shelf regions where the cleavage fracture was observed, the numerical analysis results also well agreed with the experimental results that the fracture toughness values were gradually increased with the temperature.
A special finite difference method for nonlinear dynamic response analysis of semi-infinite foundation soil using mapping which transforms semi-infinite domain into finite domain is presented here. For the region of engineering interest, mapping is isometric, and for far field, shrink mapping which transforms infinite interval into finite interval is adopted. At first, the responses of semi-infinite foundation soil with linear constituting model are computed, and compared with theoretical results and those of existing method. Good agreements are obtained among the results of the proposed method, Lamb’s theory and FEM with extensive mesh model. Then the responses of infinite foundation soil are computed by the present method, using small and large mesh model. The results of small and large mesh models agree well with each other, demonstrating the effectiveness of the proposed method.
Cellular automata simulation is one of the widely used calculation methodologies to form optimum shape of structural bodies. It can construct complex whole structure by applying locally and iteratively a simple calculation rule so-called, “state transition rule” or “local rule”. Therefore, it has been perceived as a promised approach of forming optimum shape that could not be predicted by resident design knowledge. In many researches, however, state transition rules are just used for controlling birth and death of structural elements and the validity of formed shapes as structural bodies was not considered in those state transition rules. Extra structural simulations, typically FEM analyses, are necessary for calculating stress, strain and deflection in the formed bodies in order to determine that the structures can satisfy design requirements. Those analyses are global process and not suitable to locality principle of cellular automata and limits possibility of shape forming of cellular automata. In this context we propose a new state transition rule that can express structural behaviors for two-dimensional linear elastic body. Also we show numerical results with two typical examples and clarify its accuracy and effectiveness.
It is hoped that engineering information can be browsed with the various situations and systems from the improvement of digital engineering technology, the trend of Semantic Web and information technology. In this paper, the integration framework of the new approach based on the standardized contents is developed for this requirement. Its framework is W2E (Web Engineering Environment) based on the Web Master Model (WMM, analysis information and 3-dimensional model were modeled with XML (eXtensible Makeup Language)) for aiding engineering on Web environment. WMM was realized by the development of W2E to integrate analysis information and 3-dimensional model by the standardized information expression, to connect between the applications and to reuse model across general XML parser.
The purpose of this research is to develop a procedure of performing stress analysis on tanks composed of a few layers made of anisotropic materials considering plastic deformation, and high-pressure hydrogen tanks installed in fuel-cell vehicles. In the present report, a procedure of performing triaxial-stress analysis on tanks composed of two layers made of orthotropic elastoplastic materials was developed. The equations to calculate stresses, strains and displacements were first derived using constitutive equations, equations of force balance and boundary conditions. Then computational program was developed using the derived formulation. The validity of the derived formulation and the developed program was proved by comparing the analytical results calculated by the devloped program with the numerical ones of finite-elemet analysis.
A technique for identifying the structural parameters from the vibration parameters such as mass, damping and stiffness is developed and tested using the numerically simulated noise free and noisy data, as well as experimented data. The eigen parameters and the transfer functions are obtained by the RD, FFT and the ERA method used. The mass, damping and stiffness matrix of the structural system are assumed to be linear functions of structural parameters to be determined and non-iterative identification method for noise free data is derived. Then the procedure is combined with Kalman filtering procedure to deal with noisy data. Structural parameters are identified well for data with small noise. For the data contaminated by large observation noise and unknown force, use of transfer functions obtained from force vibration are effective.
A continuous jet changes to droplets where jet breakup occurs. In this study, x-y two-dimensional numerical analysis of jet breakup is performed using MPS method (Moving Particle Semi-implicit Method) which is a particle method for incompressible flows. The continuous fluid surrounding the jet is neglected. Dependencies of the jet breakup length on the Weber number and the Froude number agree with experiment. It is indicated that MPS method can reproduce these dependencies of the jet breakup length despite the difference of the space dimensions. The size distribution of droplets is in agreement with the Nukiyama-Tanasawa distribution which has been widely used as an experimental correlation. Effects of the Weber number and the Froude number on the size distribution are also obtained from the calculation.
As the requirement of large-scale structure analysis becomes stronger, the expectation of parallel computing becomes higher. How to program for parallel processing highly depends on the computer’s architecture, especially memory architecture and parallel interface. This paper discusses parallelisation strategies of general eigenvalue analysis not only based on MPI for distributed memory but also based on OpenMP for shared memory. The performance of MPI, OpenMP, and MPI+OpenMP hybrid parallelised codes are compared on SGI Origin 2000 and Compaq SC series. As a result, we can give that all these approaches work equally well, while the parallelisation with OpenMP has a merit of simplicity; the efficiency of hybrid parallelisation was affected by type of parallel computer and scale of problem.
Several mesh generation methods are proposed in this paper for the numerical analysis of vehicle electric circuit using the geometric properties of 3-dimensional layers. The aim of mesh generation is to obtain 3-D finite elements with better geometry. In this paper the authors propose four mesh generation methods to generate hexahedral prism and tetrahedral meshes. Basic tools for mesh generation are 2-dimensional and 3-dimensional Delaunay triangulations, which are geometric subdivisions using points. Through simple numerical experiments four methods are compared and their efficiency is discussed.
Hyperelastic material is regarded as incompressible material. Solving of simultaneous equations which consist of incompressible restraint condition and the equation of motion is necessary to incompressible material analysis. Mixed FEM is generally used in hyperelastic material analysis. In this method the simultaneous equations are directly solved. Meanwhile in incompressible fluid analysis decoupled FEM which has developed from MAC method in FDM and has superior features is used besides mixed FEM. In this method the simultaneous equations are indirectly solved by introducing decoupled equations which consist of predictor step, corrector step and Poisson equation. In this study decoupled FEM for incompressible hyperelastic material is constructed. The hyperelastic material of Mooney-Rivlin model and GSMAC-FEM, decoupled FEM suitable for calculation on a large scale, are especially used. Lagrangian method and Newmark-β method is introduced to Eulerian basic equation to apply decoupled FEM to solid analysis. The discretized equations are verified in plane strain problem.
This paper presents a numerical analysis method for boundary shape optimization problems involving moving the nodes of specified natural vibration mode shapes of linear elastic continua to assigned positions and describes the applicability of the method to chassis-like frame structures. Based on the theoretical investigation done in the previous work, problems of moving the nodes of natural vibration mode shapes are formulated as minimization problems of the square integrals of the specified natural vibration mode shapes on the assigned sub-boundaries to which it is desired to move the nodes. Reshaping is accomplished by the traction method that was proposed as a solution to boundary shape optimization problems of domains in which boundary value problems of partial differential equations are defined. With this method, problems of moving the nodes of the natural vibration mode shapes with first-order bending of chassis-like frame structures are solved. An investigation into the effect of the support conditions in modal analyses reveals the superiority of using a vertically free support condition.
In this paper, element moving algorithm for finite element analysis of shallow flow is proposed, and its algorithm and verification results are shown. This algorithm moves nodes on shoreline according to the movement of brink. And the nodes on shoreline are exchanged one by one when the transformation of the element grows. The features of the present algorithm are as follows. 1) The generation of the distorted element is limited to the vicinity of the shoreline. 2) Because restructuring, division and integration of elements are not caused in the process of the calculation, the load of the element movement processing to the analysis of the flow is small. An uniform slope and a coupled slope were taken up for a verification example. And the results were shown that this method was able to hold bottom shape of tidal flat compared with boundary moving method by Lagrangian method.
In this paper, an effective numerical method to solve a transient dynamic and acoustic problem for railway structures due to the impact forces is presented. After the dynamic response of the structure is obtained by FEM and the velocity response of the surface of the structure is transformed to the frequency domain, the sound pressure in the air outside the structure is obtained by BEM using the velocity boundary condition at each frequency. The transient sound pressure at the arbitrary points outside the structure is obtained by the superposition of the sound pressure in the frequency domain. In the superposition, the inadequate higher frequency decided from the element size of boundary mesh is omitted to get a reasonable solution and save computation time. The submodel in the whole structure is employed to get a transient acoustic solution effectively for an actual railway structure. Based on the present method, a transient dynamic and acoustic analysis program, ASA/ACOUSTICS, has been developed. Numerical examples are demonstrated.
Robust Incomplete Cholesky decomposition developed by Ajiz is based on the idea of stabilization for diagonal entries, and no breakdown can occur during the incomplete decomposition. In this paper, we consider convergence of new preconditioning which combine relaxation with dropping procedure for diagonal entries for the purpose of speed-up of the preconditioned CG iteration. Through numerical experiments for some realistic structural problems, the new approach with quasi robustness insures convergence rates of PCG method.
The research of the square roots of NOT gate and the XOR gate encourages me to explore the other square root gates such as Hadamard transformation and Toffoli gate. There is some regularity in the square root gate of the quantum logic gate. The operation of the square root gate is found to consist of the operations of identity operation and the usual quantum logic gate.
The Ginzburg-Landau and Maxwell equations which describe superconductivity in materials are solved numerically by using Algebraic Multi Grid. Numerical efficiency between the Algebraic Multi Grid and the Conjugate Gradient method are compared. In certain parameter region, it is found that the Algebraic Multi Grid is superior to the Conjugate Gradient method.
In this paper, a preconditioning technique that is based on the finite differential approximation of the Hessian for non-linear conjugate gradient method is proposed and it was adapted to electronic structure calculation within finite element framework, i.e. the real space formula, which has the favorable nature for parallel computing. In the conventional electron calculations, namely the band structure methods, the conjugate gradient method has been also used with preconditioning. However, such preconditioning cannot be applied to the real space method. The efficiency of our new approach was explored in the computation of hydrogen electron structure. It has turned out that the proposed method was five times faster than no-preconditioning conjugate gradient method at maximum.
It is important to plan an optimal trajectory in order to save the dissipated energy of a manipulator controlled with Point-to-Point motion. In this paper, the optimal angle functions of the joints minimizing the Joule’s and mechanical heat losses due to Coulomb friction are taken into consideration. It is difficult to obtain the optimal angle functions analytically, because Coulomb friction is discontinuous to angular velocity. Therefore the angle functions have been expanded by the Fourier series, and the coefficients have been determined by a Genetic Algorithm (GA). However, applying the conventional GA arises some serious problems when Coulomb friction is strong. This paper proposes a novel GA that divides the search generations into some stages, and increases the number of Fourier coefficients according to the progress of the stages. The optimal angle functions minimizing the dissipated energy can be obtained by using the novel GA, even if Coulomb friction is strong.
In this paper, we discuss and present a newly-developed computer-assisted documentation module in a distributed problem solving environment (PSE) for partial differential equation (PDE) based problems. We have proposed a PSE for PDE-based problems in Trans. JSCES No.20010018, (2001), in order to extend the conventional PSE concept. The distributed PSE system inputs a problem information including PDEs, the initial and boundary conditions and a discretization and computation scheme, and outputs a program flow and also a C-language source code for the problem. The PSE contains all the information of the problem itself, PDEs, discretization scheme, mesh information, equation manipulation results, designed program structure, variable and constant definitions and program itself. Therefore it is possible to construct a documentation module for the designed and generated program in the PSE, and the generated document describes the important information for users to understand the problem itself and the program. This new idea explores the capability of PSEs and can be realized within the framework of the PSE.
This paper describes a fast and parallel stack-based linear genetic programming system for local contact search algorithms of finite element impact analyses. Here the present GP system is used to find the function approximating local coordinates of the contact point in the local contact search process of the master-slave type contact algorithm. The proposed system is superior in computing speed and in accuracy of finding approximating function than the pointer-based GP system which was proposed by the authors in the previous paper. The fundamental formulation of this system is first described in detail, and then its basic performance is demonstrated through some sample analyses on a PC cluster.
A formulation of two-dimensional geometrical non-linear finite element with drilling degrees of freedom is presented and some analyses using the proposed element are conducted. The formulation is based on a rate form of the virtual work principle and obtained by a simple extension of standard FEM. A convergence of analysis solution and robustness for element distortion are investigated and the results are compared with those of standard displacement based first and second order elements. The proposed element provides a good solution convergence as well as second order standard elements although number of D.O.F. of the element is less. Additionally, it is shown that high-precision solution is given when the mesh is strongly distorted.
The purpose of this study is to develop a procedure for carrying out analysis on torsional vibration of a mechanical system including piston-crank systems such as a reciprocating compressor. In the formulation, we first derived variable moment of inertia of a piston-crank system about the crankshaft through revolution of the crank. We adopted the mechanical system consisting of a motor, a fly wheel and multiple piston-crank systems as an analysis model. We also derived equations of motion using the derived variable moment of inertia of a piston-crank system and considering the boundary conditions that both ends of the crankshaft are free. The program for eigenvalue analysis and the one for time history responses were developed according to the formulation. The dynamic characteristics of a model system consisting of a motor, a fly wheel and two piston-crank systems were investigated by performing eigenvalue analysis. The time history responses on torsional vibration of the crankshafts of the model system were also calculated.
The balancing domain decomposition method (BDD) is known to be an efficient linear solver that can be applied to large-scale parallel finite element analyses. In this study, a method to incorporate a set of linear multipoint constraints (MPC) into the BDD is proposed. In this method, all the DOFs that are related to MPCs are converted into the interface DOFs. Then, the interface problem with a set of MPCs is solved by the conjugate projected gradient method. The effect of the MPCs for the interface DOFs is imposed on the coarse grid problem in the BDD by using the penalty method. Results of illustrative examples show good convergence properties of the present scheme.
Higher-order elements are applied to GSMAC-FEM, i.e., a decoupled FEM that was developed from HSMAC method in FDM. And it is evaluated in terms of the accuracy, the memory capacity, and the convergence of simultaneous relaxation iteration in two-dimensional lid-driven cavity flow at Reynolds number 5000. We propose a new diagonal component method for the discretized Poisson equation, not assuming that the shape of element is quadrilateral and the interpolation function for pressure is constant in element. And we also propose a new simultaneous relaxation method of velocity and pressure under consideration of the satellite elements of a pressure node. The numerical method realized by these ideas does not lose the advantage of GSMAC-FEM. As the result combinations of the interpolation functions for velocity and pressure suitable to GSMAC-FEM are found.
The Multi Agent Simulation Technique (MAST) was applied to establish a plausible model for evaluation of the project management. Both features of the simultaneous distributed processing and the so-called bottom-up modeling, not by giving a series of whole governing equations, are crucially important points that MAST can provide to shed accurate light on an analysis of the project management. The artificial society consists of three types agents such as Staff who actually deals with Processes, Leader who manages a section having several Staffs and Project Manager who oversees a project holistically. There are existing Processes, which are distributed on the 2D artificial society considering the assumed characteristics for the project whether the project is related to relatively closed area or wide ranging, or the project has a deep dependent chain through several processes or not. The simulation was done from the viewpoint of the project organization and the Process characteristics. The obtained result well captured and reproduced features observed in the real project organization such as Functional, Matrix and Task Force type respectively.
In this paper, we apply the equation discovery system to inverse analysis, and present the method of deriving the formula which can calculate unknown parameter directly. This method is the modification of RF5, and numerical model identification system using genetic algorithm and information criterion. It is assumed that the model is expressed with multidimensional polynomial equation. The number of terms and dimensions are coded into genetic algorithm. The coefficients for each model are obtained by least square method. The fitness for each chromosome is evaluated with an information criterion c-AIC. Numerical simulations are performed to demonstrate the validity and usefulness of this method.