This paper deals with the ultimate strength of unstiffened and longitudinally stiffened plates subjected to biaxial in-plane forces. Firstly, the elastic buckling strength and corresponding buckling modes of stiffened plates are investigated in order to deduced the required minimum relative stiffness of a stiffener, and to decide the initial deflection modes of analytical models for analyzing the ultimate strength of stiffened plates parametrically. Then, the ultimate strength of unstiffened and stiffened plates are investigated through the elasto-plastic and finite displacement analyses on the basis of a finite element method. Finally, a simple and approximate interaction curve for the ultimate strength of unstiffened and stiffened plates is proposed.

View full abstract

Semi-unified constitutive equations are proposed for use in the analysis of rate and time dependent deformation of structural steel SS 41, at high temperatures. The material is assumed to behave as an elastic viscoplastic material. The effect of creep is introduced as contraction of yield surface, at each step of loading. Material parameters for the model can be obtained from standard creep and tension compression tests. Experiments have been performed on SS 41 steel and details of tension compression test have been reported. A numerical scheme has been developed to determine the strain increment for given stress increment using given material constants. Simulated results are compared with the experimental data and are found to be in good agreement.

View full abstract

Analysis of a medium which consists of irregularly shaped regions is a major problem in the DEM studies. A new method for solving this problem is proposed. We developed the Extended Distinct Element Method (EDEM) and applied it to problems of fracture behavior in media composed of irregularly shaped regions. An aggregate of circular (2-D) or spherical (3-D) elements connected by pore-springs constitutes an irregularly shaped region, and is called the “Combined Discrete Element (CDE)”. The concept of the CDE has greatly broadened application of the EDEM as the behavior of an anisotropic medium composed of variously shaped regions can be easily studied; moreover, cracks within the CDE as well as complete fracture of the CDE can be simulated automatically.

View full abstract

The object of this paper is to propose a stochastic method for evaluating the magnitude of future earthquakes taking account of the nonstationariness in earthquake occurrence. For this purpose, the strain energy accumulation in a focal region was estimated by means of earthquake data of the past 100 years in Japan. Furthermore, distributions of maximum ground acceleration were derived by means of the attenuation law. As a result, we found that the distibutions of maximum ground acceleration fit type III extreme value distributions and that the expected values of those distributions depend on the strain energy accumulation significantly. Finally, it is pointed out that the nonstationariness in earthquake occurrence should be taken into consideration in order to evaluate the earthquake load in design.

View full abstract

Two simple stochastic models of spatially varying ground motion are established and calibrated from strong motion array data at four sites in Shizuoka Prefecture, Japan. These adopt a stationary random field model of each event's ground displacement in time and space. These models are applied to estimate maximum relative displacement between two points on the ground surface. Good agreement is found with observed relative displacement, both root mean square values and maximum values.

View full abstract

On a high speed railway like Shin-kansen, it is so important to suppress the roughness of rail serface from the viewpoint of running safety, growth of irregularities, noise and vibration. For this purpose, the rail bending-up in cold work has been developed. In this process, following problems were noticed. (1) The lift-up of rail in bending-up. (2) The difference of effects in bending-up between by head catch and by bottom catch. (3) The safety from failure of rail head in head catch. (4) The measure to protect rail from punching due to rail catch. Through both the analyses with use of Bernuilli-Euler beam theory and FEM and the comparison of them with experimental data, sufficient results are obtained.

View full abstract

The ultimate strength of steel plates with longitudinal stiffeners of open or closed cross-section subjected to bi-axial in-plane forces is experimentally investigated in this study. Six stiffened plate specimens corresponding to 1/3 times a stiffened plate panel of the steel deck plate in an actual bridge were tested by using an experimental apparatus newly developed for this study. The test results verify an interaction curve for the ultimate strength which have been developed by the authors for predicting the ultimate strength of stiffened plates under biaxial in-plane forces.

View full abstract

This paper deals with the ultimate strength of composite members with box cross section subjected to pure torsion. Through the experiments by four specimens of which one is thin-walled steel, two are concrete, and one is composite members, the following conclusions are obtained; (1) The ultimate torsional moment of composite member is about 1.2 times the cumulative torsional moment of the fully plastic torsional moment of the outer steel plate and the collapse torsional moment of the isolated concrete section. (2) However, when the ultimate shear stress of confined concrete is taken as 0.5 times the compressive strength of concrete, the ultimate torsional moment can be approximately evaluated by the cumulative torsional moment.

View full abstract

To build knowledge-based systems more effectively, it is required to improve the performance of the acquisition of knowledge, especially in engineering domain, the acquisition of quantitative knowledge. One of the most promising candidates is the application of machine learning techniques to (semi-) automatic acquisition of quantitative knowledge.
The purpose of this paper is to propose an inductive machine learning method for the formulation of experimental equations from noisy empirical data. It combines statistical techniques and heuristics to reduce the search space of possible equations effectively. As an example of its application, the estimation of sloshing impact pressure at flat roof surface of tanks is taken up, which one of the authors reported previously.

View full abstract

An assembled structure, for instance precast concrete slab or precast rock-shed roof which is made of several precast concrete members laterally tightened by prestressing steel tendons can not be regarded as a perfectly continuous structure. The reason is that such a structure acts as if the precast members were connected through the spring corresponding to prestressing steel tendons after the maximum normal stress becomes zero at the tip of contact surface between precast members. This paper presents the experimental and analytical results on the static behavior of the above-mentioned structure, for which a beam made of two precast concrete members longitudinally tighted by the prestressing is substituted. To obtain the analytical results the rigid body spring model which is suitable for non-linear problem is applied.

View full abstract

Appropriate length and interval of bolts in the long bolt tension-type connections are investigated by parametric finite element analyses. Both length and interval of bolts do not only significantly affect the connection behaviours, but are very important factors on the design of the connections. The recommendations of the length of the connection and the interval of bolts are provided by the values of nondimensional geometric parameters.

View full abstract

Goal programming has mainly been used for the system control of structures and may be a useful optimization technique for various multiobjctive problems in optimum structural design. It is the purpose of this paper to inspect the convergency and accuracy of the nonlinear goal programming technique for large scale optimization problems of which the objective functions are the displacements of structures as well as the structural weight. Furthermore, one-sided goal programming technique is applied to the minimum weight design of structures subjected to restrictions on the displacements, and its utility is investigated in comparison with the ordinary single-objective optimization technique by numerical calculations.

View full abstract

A practical minimum weight design algolithm is presented for such large scale problems as the design of spatial framed structures under multiple loading cases. The stress, local buckling of plates, slenderness ratio, displacement and minimum size constraints specified by the design codes are considered in the optimum design. For the development of a practical programming system with high efficiency, new mathematical and numerical methods, such as modified SLP, surplus variables and the other several technique, are employed. Efficiency and reliability of the design system are demonstrated by numerical calculations for large scale problems in the design of actual bridges.

View full abstract

This paper presents a method for the evaluation of seismic hazard, defined as the relationship between the intensity of earthquakes at a given location and their annual probability of exceedance, which was developed for the probabilistic safety assessment of nuclear power plants.
The proposed method has the following characteristic features:
1) A combined usage of historical earthquake records and active fault data is made with a consideration of data availability and seismological conditions in Japan.
2) For evaluating the probability of occurrence of earthquakes in a given period, a time-dependent model is introduced to take into account the effect of strain energy accumulation in the active fault.
A sensitivity study with this method indicated that the evaluated seismic hazard is strongly dependent on the selection of attenuation model and the uncertainty parameter in the attenuation model.

View full abstract

For assessing the seismic reliability of large scale lifeline networks we developed an algorithm which had only polynomial complexity to enumerate paths in the network. Using this algorithm we have developed two program packages. One is composed of a vectorized program to increase computing efficiency and the other is a program coded by a new approximate calculation scheme, named as the point matching method. The vectorized program can solve seismic reliability of a network with up to 3500 elements using the super computer, FACOM VP-400E. If the number of elements of network exceed 3500, only the point matching method can provide a feasible solution from the practical stand point of computing time. This program can also ran on a 32 bit desktop computer with the main memory of 16MB and the operation time of 5.3 MIPS. This program package offers you an interactive capability, to select supply and demand nodes as well as show them on screen. To check the applicability of the program package, we considered two gas supply networks composed of 3529 and 19522 elements.

View full abstract

A modal analysis procedure based on a FE-BE method in the time domain is applied to vertical vibration of a dam-foundation system. The procedure has an advantage to analyze not only underdamped modes but also overdamped modes. For the case of six different impedance ratios between the dam and the foundation, impulsive responses and thier Fourier spectra are calculated to evaluate modal characteristics, such as vibration mode, vibration frequency, and radiation damping. As a result, it is found that the radiation damping for the vertical motion is half of that for the horizontal vibration of the system.

View full abstract

In this paper, the storage requirements and performance consequences of different implementations of the conjugate gradient methods for the solution of sparse systems arising from finite element structural analysis are reviewed.
Although the ICCG method has good convergence rates, it is not so effective in computing times because of the expence of computing forward and backward substitution at each iteration. Since algorithms for using the incomplete Choleski procedure are unsuitable for vectorization, it is concluded that the SCG method will be of advantage over the ICCG method on supercomputer systems.
The convergence characteristics of the ICCG method depends on the accelerating factor introduced into the resulting diagonal elements. This paper presents a possible means to decide the parameter which achieves the optimal convergence.

View full abstract

A hybrid method of importance sampling and Kalman filtering is proposed for evaluation of probability of failure of structural systems. In this method, data are effectively sampled from an importance sampling density function, and with them as observation data, probabilities of limit state functions are adaptively updated by using Kalman Filter.
Efficiency and stable convergency of estimated probabilities to true ones are examined through numerical examples.

View full abstract

This paper proposes a practical evaluation method of seismic stresses developed in the cross section of shield tunnels in homogeneous soft grounds. The seismic stresses are analyzed by means of subjecting a ring-with-elastic-support model of the tunnel lining to ground displacement and shearing stresses which herein represent seismic loads. These seismic stresses are described through mathematical formulae using a series of Bessel functions. The accuracy of the method is checked through FEM analyses.
For designers' convenience in making a rough estimate of seismic stresses, approximate formulae are presented by simplifying the exact formulae.
The fundamental characteristics of seismic stresses developed in tunnels are also investigated by parametric studies.

View full abstract

This study consists of the following two topics: one is a basic consideration on the extended kalman filter in identification problems, and the other is a proposal of a new local iteration procedure called Extended Kalman filter-weighted local iteration method (EK-WLI method).
In the former study, roles of each factor involved in the kalman filter algorithm was clarified using a two component state vector equation.
In the latter one based on the results of the former study, a local iteration procedure was developed which was found to be effective in reducing the enor due to the linearization of the observation equation on the identified state vector.

View full abstract