In large scale truss structures, application of Genetic Algorithms (GA) for solution of optimal layout problems, become probressively difficult as the genome length increases. This makes it necessary for implementation of a hierarchical representation technique where the genome does not directly encode the problem solution (in the entire search space), but rather encodes a representative subset of the actual search space. The method presented in this paper encodes a Cellular Automata (CA) whose "local rules" are being searched by the GA. As the GA encodes "only" a fixed set of local rules, to evolve the CA, the genome length is independent of the scale of the problem or the size of the search space (scale of the truss structure).

View full abstract

A method has been developed that gives advanced Tersoff interatomic potential parameters describing nonequilibrium atomic structures. A genetic algorithm is used to optimize many potential parameters that are fitted to the first principle cohesive energies of various systems, including bulk systems with atomic defects and amorphous, surface, or cluster systems, which are under stress. This optimization method converges towards a set of potential parameters that well describe not only crystabls but also amorphous systems. In this work, the Tersoff potential for carbon is discussed.

View full abstract

For the numerical simulation of mechanical behaviors of structures with local heterogeneity, a global/local modelling technique using the finite element mesh superposition technique and shell-solid connection is proposed. Not only the global deformation but also three-dimensional stresses nearby the local heterogeneity can be analyzed precisely with less efforts than conventional methods. The finite element mesh superposition technique makes it possible to overlay local fine solid mesh which considers the local heterogeneity on global rough solid mesh which neglects the local heterogeneity. Aplate structure that contains dissimilar material in a local portion is analyzed by the proposed modelling technique. The results agreed well with those calculated using the conventional three-dimensional fine mesh model.

View full abstract

The efficient and robust distributed processing of finite element method can be realized on the network of various computers by using substructure method.Java, one of object oriented language, is applied to the programing of FEM, and then the program is executed on the various computers. The variation causes the late calculation because of the difference of calculation ability among them. The adjustment of substructure is used to solve the problem of the latency. Some troubles caused on one of the computers affect the whole analysis in the distributed processing. The robusticity of the distributed processing is realized to avoid the troubles by the mutual saving system among computers.

View full abstract

Optimization method for various phenomena on nonlinear problems is presented by using multilayer neural network in this paper. The each gradients of objective function for the output signals of the network is assumed in order to define the effects of learning on back propagation, and then it comes to derive optimum design variables on the problems including inelastics by self-teaching. Furthermore, it is also presented that the idea of probability can realize efficient optimization by adding fluctuation to the network. The results of the presentation are applied to the reduction problem of residual stress of welding by use of thermo-elasto-plastic analysis and show its applicability to the problems.

View full abstract

An automatic FEM procedure based on a geometry-oriented h version is presented for a general 3D solid structure combined with many parts of various forms. A simple and efficient method to create the geometry of a solid structure with many parts combined is described. Physical conditions for the structural analysis are given directly on the geometry. Then simply by giving the total number of elements, all input data including the element mesh and constraint equations to combine elements between parts required for a finite element structural analysis program are generated, and the initial analysis is made automatically. By giving only the allowable ratio of the finite element error, the adaptive mesh refinement is made and the detailed analysis is also carried out automatically. A practical PC program based on the present method has been developed. Numerical examples are demonstrated to show the validity and effectiveness of the proposed method.

View full abstract

If initial temperature is assumed to be constant, a domain integral is not required for solution of unsteady thermal stress problems without heat generation using the boundary element method (BEM). However, under heat generation or nonuniform initial temperature distribution, the domain integral is necessary. In this report it is shown that the problem of unsteady heat conduction with heat generation and nonuniform initial temperature can approximately be solved without use of the domain integral by a triple-reciprocity boundary element method. In this method, the distribution of heat generation and initial temtepature are interpolated by using integral equation.

View full abstract

New integrals of fundamental solution of three-dimensional Laplace equation are derived by using Gauss' divergence theorem. These are useful for boundary element method. One of the integrals is single layer potential of a constant triangular element and others are single and double layer potential of a linear triangular element. There are two advantages of these integrals. First, coordinate transformation and subdivision of a triangular element are not necessary to evaluate these integrals. Second, it is possible to evaluate formulas of single and double layer potential effectively, because the formula of double layer potential is related to the formula of single layer potential. The validity of these integrals is confirmed by comparing with numerical integration of fundamental solution over a triangular element by using Lachat algorithm. The effective gradient formulas are derived by differentiation of these integrals analytically. Present integrals can be applied for not only collocation BEM but also Galerkin BEM and fast multipole BEM.

View full abstract

The present study investigated the relationship between the interatomic potential and the thermal properties, including the melting point and the thermal expansion, by using molecular dynamics simulations. Tantalum, which had a high melting point, was used as the test material. The potential functions tested in the present study were made based on the concept of FS potential, while the function was modified from the original one in order to control the thermal properties. The melting point was determined by using NPT ensemble simulations with the initial conditions that the solid phase and the liquid phase coexisted in the calculating region. The thermal expansion characteristic could be estimated qualitatively from the change of potential energy with an isotropic expansion/compression under the still atoms condition. The melting point was influenced by amplitude of vibration of atoms, and it was decreased with decreasing of the modified elastic moduli. An empirical potential function for Tantalum was derived, which satisfied some properties near 0 K, the lattice constant at 2500 K and the melting point.

View full abstract

Due to the strain-induced martensitic transformation during plastic deformation, a transformation-induced plasticity (TRIP) steel possesses such favorable mechanical properties as high strength, ductility and toughness. Since its favorable mechanical properties are realized under quite restricted circumstances, the prediction and control of the deformation processes is indispensable to determine the expected mechanical properties of TRIP steel. However, it is very difficult to obtain the TRIP steel which only have the martensitic structure before deformation and the stress-strain relationship of each phases under various temperatures cannot be measured by the experiment. So the identification of the parameters in the constitutive equations of TRIP steel is necessary. Here, a method employing a finite-element simulation and the non-linear least square method with constraint conditions was proposed to identify the constitutive parameters of TRIP steel from the true stress-plastic strain and the volume fraction of martensite-plastic strain relations obtained by the uniaxial tensile test and computational simulations. Then, the application of this procedure employing virtual data of an ideal TRIP steel which possesses more favorable mechanical properties is discussed for investigating improvement of the mechanical properties of TRIP steel.

View full abstract

Combined normal stresses (σ₁ and σ₂) were applied to tubular specimens normal and parallel to the material axes. In all cases, acoustic emissions (AE) were measured in details for distinguishing damage accumulation. The S-S relationships in both directions were not strongly affected by a biaxial stress ratio, α(=σ₁/σ₂) in comparison with the S-S relationships under tension/shear biaxial loading since the load was primarily sustained by fibers under tension/tension biaxial loading in the material axes. The S-S curves in both direction are represented by two straight lines intersecting with each other at their knee points while the secondary knee does not appear as soon as the first knee has been observed except at α=1/1. Consequently, the initial yield stress condition is represented by two lines parallel to either σ₁ or σ₂ axes. They are drawn based on the knee stresses at α=1/0 and 0/1. The maximum stress theory fits the experimental failure envelope better than the Tsai-Wu failure criterion.

View full abstract

The finite difference method based on inteqration along the bicharacteristics is the suitable method for the stress wave propagation. The advantage of this method is attributed to the fact that the characteristic surface coincides with wave front. In this paper, by introducing a laminated composite model and the finite difference method based on integration along the bicharacteristics, propagation of the stress waves in a nonhomogeneous plate is approximately analyzed. The material properties in the plate change continuously in the thickness direction. Numerical calculations are carried out for nonhomogeneous plates composed of alumina and aluminum alloy. From numerical results, it is found that impact damage resistance of the nonhomogeneous plate is improved by controlling the distribution of the components.

View full abstract

The electromagnetic interference (EMI) may cause disruptions in robots, personal computers, cellular phones, medical appliance and other electronic instruments. It is therefore important to develop the materials where shield the electromagnetic wave in the environment from the EMI generated by electronic devices. This study was performed to obtain an instruction for the development of smart materials having shielding effectiveness for electromagnetic wave. In the plastic composites used, fine copper wires were oriented with various angles and separations. The fundamental characteristics of shielding effectiveness for electromagnetic wave investigated for the basic plastic composite and carbon fiber reinforced composite. The shielding effectiveness was measured using a copper shielding box located in an electromagnetic shielding room. It was found out that the shielding effectiveness for electromagnetic wave depends on the oriented angle of copper wire. The shielding effectiveness decreased with increasing the wire separation. For carbon fiber reinforced composites, the shielding effectiveness was improved by the contact of carbon fiber each other.

View full abstract

To improve the fatigue reliability for spheroidal graphite iron, FCD370, by WC plasma-sprayed coating, the influence of the surface modification on fatigue fracture mechanism was experimentally studied. The fatigue properties of FCD 370 are found to be obviously improved by WC plasma-sprayed coating, so that they can no longer be estimated by using a concept of "S-N Globe" which has successfully been applied to many surface-treated materials. Through fractography using SEM, it was found that the fatigue crack initiated at the debonding graphite at the interface between the substrate and sprayed coating. Furthermore the meso-crack growth was considerably influenced by the graphite due to its irregular profile. A "P-S-N Globe" was aplied by combining the experimental analysis with the linear notch mechanics and reliability engineering.

View full abstract

The influence of WC plasma-sprayed coating on fatigue behavior was examined to improve the fatigue reliability of spheroidal graphite iron, FCD 500 and FCD 700. The fatigue strength of FCD 500 and FCD 700 is obviously improved by WC plasma-sprayed coating. Through detailed observation by SEM, it was found that the T-type crack was formed on the coating surface. The T-type crack initiation is attributable to the debonding between the coating surface and the subsurface. It could be concluded that the improvement in the fatigue properties of FCD 500 and FCD 700 would be caused by the prevention of propagation of mesoscopic crack due to sorbite which had been created by the WC plasma-spraying heat.

View full abstract

In order to elucidate the mechanism of superlong fatigue failure in the regime of N>10⁷ cycles, the fracture surfaces of specimens of heat treated hard steel, SCM 435 and 0.46% medium carbon steel were investigated by optical microscope, SEM and AFM. It has been revealed that specimens having longer life have a particular morphology beside the inclusion at fracture origin. The particular morphology looks optically dark by the observation of optical microscope and it has been named the optically dark area, ODA. The ODA looks a rough area in the observation by SEM and AFM. The relative size of ODA to the size of inclusion at fracture origin increases with increase in fatigue life. Thus, ODA has a crucial importance for the mechanism of superlong fatigue failure. It has been assumed that ODA is made by the fatigue due to cyclic stress coupled with hydrogen which is trapped by the inclusion at fracture origin. To verify the hypothesis, specimens annealed at 300°C in a vacuum (VA specimens) and quenched in a vacuum (VQ specimens) are prepared to desorp the hydrogen trapped by inclusions. The specimens VA and VQ, had a much smaller ODA than conventionally heat treated specimens. Thus, it has been concluded that hydrogen trapped by inclusion is the crucial factor which causes superlong fatigue failure of high strength steels.

View full abstract

In our previous paper, as the results of the high cycle rotating fatigue tests of HSP-treated SUS 316 L steel, the fatigue crack initiation was observed on inside of the specimen by fractographical investigation using the scanning electron microscope (SEM). In this paper, in order to investigate the mechanism of the internal crack initiation and propagation behaviors, elastic-plastic finite element method (FEM) analysis was carried out. For the FEM analysis, the residual stress and the yield stress distributions models of the HSP-treated specimen were established based on the residual stress by X-ray diffraction measurement. As the results of analysis, the depth of maximum stress of the specimen by FEM agreed well with the depth of the crack origin by SEM. Also internal crack propagation behaviors could be satisfactorily explained by the FEM analysis.

View full abstract

In a previous study, to minimize or eliminate the errors and noises associated with a full-field experimental measurement and subsequent fringe analysis such as moire interferomery, the authors have derived a variational principle minimizing the experimental measurement errors. Concepts of restoration forces and energies were also presented. Furthermore, the authors have developed an intelligent hybrid method based on this variational principle. In this study, first, a fringe-analysis method is developed together with the two-dimensional FFT. Then, experimentally recorded moire fringe patterns are analyzed by the fringe-analysis method. The conventional and present intelligent hybrid analyses are caried out using the analyzed fringe information as input data. The present method verifies the automatic detection of experimental noises and errors, and the simultaneous automatic elimination of those errors. This method also made it possible to obtain fairly smooth visualization of higher-order information such as the stress and strain distributions.

View full abstract

Crack growth behavior in serviced first-stage nozzles of an 80 MW class gas turbine was studied. Crack growths in the outer wall and trailing edge were mainly observed in the nozzles inspected in this research. Cracks propagated along the dendrite or grain boundary. Precipitation of chromic carbides was also observed, and it seemed to affect crack growth. The small punch method which is a destructive test using a small 10 mm square and 0.5 mm thick specimen, was suitable for evaluating the degradation of the gas turbine nozzle. Crack growth behavior was also analyzed from inspection data. Crack growth in nozzles operated in DSS mode was more severe than that in WSS mode. These cracks were identified as the thermal fatigue cracks that propagate by temperature variations at start-up and shut-down of the turbines. The crack propagation in the outer wall has a tendency to arrest when the crack length reaches about half width of the wall. This crack arrest will be caused by thermal stress releasing with crack growths. A statistical trend analysis based on 3σ gave a conservative assessment result and was found to be a reasonable assessment method for the gas turbine nozzle cracking.

View full abstract

A crack growth analysis method for gas turbine first-stage nozzle was discussed in order to improve the residual life assessment method of the nozzle. A cobalt base superalloy was used for the trap-ezoidal strain waveform including the compressive strain hold. Crack growth rate was accelerated with increase of the compressive hold period. A crack growth analysis was carried out considering the stress profile in the nozzle which has a gradient due to the hot combustion gas and inner cooling. Crack growth rate calculated in this analysis was modified with considering the compressive strain hold effect. The predicted crack growth was coincided with the inspection data within factor of 1.5. Furthermore, a simplified method to predict the crack length after service operation was also proposed and it was verified by using the inspection data.

View full abstract

In this paper, totating bending fatigue tests are carried out in order to clarify the difference of initiation and growth behaviors of cracks in shot-peened carbon steel specimens and unpeened ones. The material used is an annealed S 45 C carbon steel. The difference between the two is clarified through the successive observations of the specimen surface by the replication technique and the measurement of the residual stress during the fatigue test. The effect of shot-peening on the growth behavior of a crack is discussed based on the fact that it is divided into three parts:the weakening effect in the damaged layer near the surface, the strengthening effect in the work-hardened layer with residual stress and the zero effect in the non-work-hardened zone.

View full abstract

The micro hardness testing machine was applied to detect micro cracks on the model vessel under cyclic thermal transient with creep and fatigue. Micro hardness by the micro indentation test was measured along thickness of the model vessel. The micro hardness value decreased in the region of intergranular cracks from surface of the vesel and increased near the crack tip in comparison with no crack region. The decrease of micro hardness value in the crack region corresponded to creep-fatigue damage D≩1 under the cyclic thermal transient, where the creep-fatigue damage was calculated by the linear damage fraction rule. The micro indentation method was applicable to the detection of micro cracks.

View full abstract

Morphological aspects of thermally induced brittle cracks are revisited. The formation of a growth pattern of cracks, which are subjected to the thermal loading due to the penetration of surface cooling, had been extensively discussed for the system of parallel interacting straight cracks. In the present paper, attentions are focused on the transition from straight to wavy crack propagation of edge cracks due to the penetration of surface cooling in a brittle solid. Curved crack extension is predicted based on the local symmetry criterion, where step-by-step finite element calculations are carried out by the automatic mesh generation using the paving method. Based on the variety of numerical results, the influencing factors which control the wavy crack paths are identified. Qualitative discussions are also made by comparing the numerical results with some experimental observation.

View full abstract

It is practically important to know the value of stress intensity factor of hole or notch with a crack. In this paper, the problems of the tension of a semi-infinite plate having a 60° V-shaped notch with a crack or an infinite plate having a double 60° V-shaped hole with two cracks are analyzed by body force method. Furthermore, the error due to the use of equivalent crack length, which is the sum of notch depth and crack length, is discussed. Finally, a simple method for calculating F₁ of a semi-infinite plate having a 60° V-shaped notch with a crack of an arbitrary size is presented.

View full abstract

A method of elastic theory with inherent strain is proposed for estimating residual stress and deformation in a pipe due to welding. Experiment of deflection was done for deciding conditions of numerical analysis on base of thermal elasto-plastic theory. Inherent strains were calculated under above useful conditions. By combining elastic theory and stress source of inherent strain, new equations are derived for calculating deflection and residual stress generated by circumfedential welding of pipe. The results of numerical analysis show a good agreement with the results predicted by the proposed equations.

View full abstract

The main aim of this paper is to give important information for understanding the mechanism of the axial crushing of thin-walled structures under dynamic loading, from the safety point of view in car crushing. For this, the axial crushing experiments of circular tube specimens were conducted by using the dynamic-loading device with the load sensing block to evaluate the properties of dynamic progressive buckling of two types of materials. The thin-walled circular tube specimens of steel and aluminum alloy were chosen, stressing mainly on the behavior of steel specimens. The effect of impact velocity on the magnitude of axial stress corresponding to the first buckling of the dynamic progressive buckling, and the effect of the differences between those material properties on the behavior of the dynamic progressive buckling were clarified. The absorbed energy per unit volume of the steel specimen was also examined. It was found that the ratio of the thickness to the diameter of the specimen affects greatly the absorbed energy per unit volume and the deformation mode of dynamic progressive buckling.

View full abstract

A new ultrasonic evaluation method of axial load is proposed for short and highly stressed bolts used for automobiles. The velocity-ratio method need not require time-of-flight measurement under stress free state, however, a considerable error is caused by the variation of bolt length which is unknown in the practical measurement. The proposed method estimates the bolt length variation and reduces the error of the axial load due to this variation. Based on the measured relations of the axial load with the time-of-flight of longitudinal and transverse waves, the variation of the unknown bolt length is estimated. By this correction, the variation of the axial load can be reduced to a half of those measured.

View full abstract

Deformation and fracture characteristics of an interface in a BCC metal (α-Fe) under hydrogen diffusion are investigated by the molecular dynamics (MD) analysis. Influences of strain, existence of interface and hydrogen concentration on diffusion of hydrogen from a free surface are studied. Two types of Fe-H potentials gave different results. By comparing these results with a physical H diffusion mechanism, a new model of H diffusion combining two Fe-H potentials was developed. In this model, deformation and fracture of an interface are relatively promoted than the provious models using a single Fe-H potential. It is shown major factors of interface fracture are applied strain and hydrogen concentration.

View full abstract

Magnetostriction curves of invar alloy are measured under various levels of uniaxial stress. Because of its low thermal expansion, the invar alloy is used in parts which suffer from large temperature change. To avoid the failure by thermal fatigue, thermal stress should be nondes-tructively evaluated. Since the invar alloy is ferromagnetic, the magnetization is accompanied by the magnetostriction. For ferromagnetic steel, we have already suggested to use the maximum magnetostriction in nondestructive stress measurement. If the magnetostriction of the invar slloy ahows the maximum value during the magnetization, it is expected to be used for the stress measurement. In this study, we measured stress dependence of the magnetostriction of invar alloy. We revealed that the knee point appears in the magnetostriction curve, while the maximum magnetostriction will not arise. It was found that the knee point magnetostriction and the minimum magnetostriction, which appears in the normal direction to the magnetization, can be used as parameters in the nondestructive stress measurement.

View full abstract

A parallel distributed optimization method for the minimization of the total resource of a system with discrete elements is proposed, and a theoretical and experimental investigations are carried out in this paper. The distributed optimization algorithm consists of two processes, namely the resource reduction process and the resource addition process. In the former process, each element discards its critical resource margin with respect to global and local constraints, while in the latter process, a small amount of resources are added to all the elements. The proposed method is successively applied for optimizing truss structures, and the method is found to be very robust and suitable for parallel processing.

View full abstract