Fundamental characteristics of stress wave propagating in the fabric reinforced laminates may be more complex than that of non-woven reinforced plastics or the homogeneous materials. Especially, the heterogeneous construction and accompanied internal fracture introduce severe complexities into the behaviour. The shock propagation due to impact tensil displacement in the lamina was measured to understand the essential characteristics of composites. Also, the shock propagation characteristics were analyzed by FEM (Finite Element Method) by considering the stiffness drop caused by the occurence of micro-fracture as a problem of material non-linearity. It was made clear the stress wave propagating in the warp fibre bundle arranged along the load direction increased sharply by the occurence of micro-fracture in the weft fibre bundle. The stress waves transmitting in both warp and weft fibre bundle diminished suddenly by the inter-separation between them. The stiffness drop that is generally called“knee”appeared clearly in the numerical results.
Reflecting its three dimensional fabrication and accompanied material anisotropy, the plane weave reinforced laminates was modelled as a combined structure composed of heterogeneous and homogeneous model units. Also, the debonding between warp and weft and the interlayer separation under the three point bending were analyzed by the FEM (Finite Element Method) supposing the critical adhesive strength between them. Numerical results that can well approximate the occurence of interlayer separation accompanying microcracks were obtained. And it was made clear that the primary factor bringing in the sudden decrease of bending load differed depending upon the stacking number of lamination. From the above results as well as the fact that the constituent model units can be selected as the needs of the case demand, the quasi-microscopic analysis is considered to be an effective CAD (Computer Aided Design) method for the design of multi-plied or large sized composites.
Recently, much attention has been paid to the high-performance fiber reinforced plastic flywheels for their ability of storing kinetic energy. In the present paper, in order to overcome the cracking failure due to the tensile stress normal to fibers in the circumferentially filament-wound (FW) discs, the hibrid discs wound by glass-fiber in the inner side and carbon-fibers in the outer side were studied. The main results are summarized as follows. (1) The analytical formulae for predicting the residual stress which stimulates the crack initiation along fibers were verified experimentally by using various kinds of specimens. The curing stress can be estimated reasonably by assuming the temperature drop during cure as-ΔT=Tc-20°C-Tr where Tc is the maximum curing temperature and Tr is the room temperature. (2) The cracking and buckling failures after the curing process can be explained in a good agreement with the analytical predictions with ΔT above mentioned. (3) The cold fitting of a metal disc inside of the filament-wound disc is another effective technique to decrease the radial tensile stress. The critical rotationary velocity of these hybrid discs was analysed to obtain the optimal dimension and constitution. (4) The use of soft epoxy resin of rubber type was found to be effective to reduce the residual stress and suppress the crack initiation.
There are three kinds of interfaces in FRP: fiber/matrix, matrix/matrix and FRP/FRP interfaces. Since these interfaces are aligned in a series in the flatwise direction of FRP, the individual property of the interface or the relation among the three interfaces can be evaluated by applying tensile load in this direction. In this paper, the relation between the fiber/matrix and the FRP/FRP interfaces and its dependence on matrix were studied by this method in the temperature range from -40°C to 120°C. Two types of matrices selected were unsaturated polyester resins which show obvious difference in their adhesiveness and thermal properties. The reinforcement was roving glass cloth. The fiber/ matrix ratio and the condition of heat treatment (FRP/adhesive) were changed. The results obtained are as follows. (1) The temperature dependence of the brittle or ductile behavior of the interface was not affected by the location of the weakest interface. But it was changed by the thermal properties of the matrix which depended on the condition of heat treatment. (2) The temperature dependence of the location of the weakest interface was affected by the fiber/matrix ratio. The location of the weakest interface changed also with the type of matrix and the condition of heat treatment. (3) The temperature dependence of the bond strength was affected by the fiber/matrix ratio. The bond strength varied with the type of matrix, the condition of heat treatment and the location of the weakest interface. (4) In the case of the fiber/matrix ratio=100/0, the bond strength changed with the temperature dependence of the bond strength. On the other hand, in the case of the fiber/matrix ratio=0/100, the bond strength changed with the temperature dependence of the location of the weakest interface.
Investigations were performed on the fatigue properties of satin woven glass cloth FRP under the pseudo random loading condition which was programmed in two different ways. Both the constant stress amplitude and the random loading tests were carried out by a hydraulic fatigue testing machine. The FFT analysis was conducted in order to identify the statistical nature of random load to be applied on the specimen. Then, discussions were made on the validity of evaluating the fatigue life under the Gaussian narrow band random loads with the aid of equivalent stress amplitude based upon Miner's rule. Also, the effect of number of cycles assigned for each stress step on the actual fatigue life in the program loading test was examined. As a result, it was shown that the fatigue life under various program loads could be estimated to some extent by the equivalent stress amplitude.
By using Fokker's Bond Tester, resonance characteristics (Δf) of the coupled system between. plastic materials and a transducer were experimentally and theoretically studied. As the materials, fiber reinforced plastics (FRP), plastics/metal lapped sheets and FRP with artificial defects were used. Δf of such materials were compared with those of sound polymethylmethacrilate sheets. Δf of FRP with artificial defects were also compared with those of FRP delaminated by fatigue tests. Possibility of detection of such delamination was examined experimentally. A characteristic curve of Δf of polymethylmethacrilate sheets was found to resemble the modified curves of FRP, plastics/metal lapped sheets and etc.. The curves of Δf of FRP with artificial defects were nearly equal to that of FRP delaminated by fatigue tests. The size of the smallest defects in FRP which can be detected by the instrument was proportional to the bottom area of transducers. The size was influenced also by the thickness of the upper and the lower zones of specimens.
In reliability analyses of machines and structures, it is necessary to know quantitatively the fatigue life distributions of the fabricated parts or the structural members, though these distributions are usually determined through statistical analysis of fatigue data. However, complete samples are hardly available for the parameter estimation, since Type I censoring is often introduced into fatigue tests at low stress levels around the fatigue limit. Another difficulty is the fact that the probability of failure may be saturated to a finite level in such low stress levels. In this study, a mathematical expression was made on the fatigue life distribution near the fatigue limit, and the parameter estimation was performed on this distribution by modifying the correlation coefficient method established for complete samples in the previous paper. The usefulness of this procedure was verified by a good agreement of the estimated distributions with those of experimental fatigue data on several kinds of metallic materials.
Various kinds of welded joints are being used in many structures such as ships, bridges and constructions. It is important in reliability analysis of such structures to clarify the statistical fatigue property of the welded joints. In this study, fatigue tests were carried out on the butt-welded joints of SM50A steel and a theoretical interpretation on the fatigue life distribution was attempted, assuming that a butt-welded joint is composed of a number of sliced specimens with different fatigue strengths. Main results obtained are summarized as follows; (1) The median crack initiation life of the butt-welded joint specimens coincided with that of the sliced specimens, when the crack initiation was defined by a 0.2mm crack in the sliced specimens or the equivalent state of stress intensity factor in the joint specimens. (2) The distribution of crack initiation lives of the butt-welded joints can be theoretically derived by combining the concept of extreme distribution and the distribution model on the number of fatigue cracks. The theoretical distribution of crack initiation lives of the joints is in good agreement with the general trend of the experimental results. (3) If the distribution of crack initiation lives and the crack growth law are given experimentally, one can obtain analytically the distribution of final fatigue lives. The fatigue life distribution of the sliced specimens can be explained by the theory established in this study.
The reduction and the scattering of fatigue strength of the butt welded joint with defects consisting mostly of blow holes were clarified through fatigue tests on 60 specimens for each of the three kinds of steels: JIS SM50Y, SM58Q, and H. T. 80. The specimens were prepared by machining from a full size butt welded plate so as to include weld defects intentionally in the groove. The quality of butt weld, or the defect severity, was evaluated on the basis of the ratio of the total area of weld defects on the fractured surface to the cross sectional area. The results showed that the fatigue strength decreased remarkably when the defects were included and the severity was small, and the rate of decrease was accelerated for the joints of higher strength steel. The fatigue strength of butt joints containing surface weld defects was 20% lower than that of the joints with inner defects. The coefficient of variation of fatigue strength varied from 10 to 13% among three kinds of butt welded joints with the defect severity of 1% or less.
As welded structures become larger, more complicated and multifarious and, furthermore, their operating conditions become more severe than before, the quality assurance of welds becomes more important. But the quality of welds cannot be assured in a straightforward manner, because (1) There are many factors which cannot be quantified easily, (2) There are a wide variety of factors which interact with each other in very complicated manners. In the present study, an attempt was made to solve these problems by applying the graph theory, because the graph theory permits a logical approach. A basic procedure for selecting the most important factor controlling at each manufacturing stage was presented to obtain the required quality of welds. In this procedure, a special care was taken about the fact that the quality assurance is closely related with standardization. The main feature of this method is that it does not necessarily call for the quantification of factors. It is expected that this method will greatly reduce the labor of quality control at the manufacturing stages.
Many studies have been made on optimum structural design based on the reliability concept. However, they are limited to a simple type of structures. This may be attributable to the difficulty of reliability analysis and the lack of efficient algorithm to obtain optimum solutions for optimum design problems. Murotsu et al. developed an effective method of reliability analysis for redundant truss structures, using a Matrix method. As for an efficient algorithm, Murotsu et al. also proposed a basic idea of an optimum design procedure. This paper extends the proposed design procedure to the optimum allocation of reliability for large-scale redundant truss structures, combined with the method of reliability analysis. The problem treated in this paper is to allocate reliability to each member of a structure so that the structural weight is minimized under a constraint on the reliability of the structure. Optimum allocation of reliabilities to the members is attained by iterating the following two steps until optimality conditions are satisfied. The first step is to adjust the allocated reliabilities of the members such that the structural weight may be minimized under a fixed structural reliability. The strategy for the adjustment is determined by considering the sensitivity of the structural weight to the variation in the assigned reliabilities of the members. The second step is to determine the design variables corresponding to the specified reliabilities of the members. The proposed optimum design procedure is applicable to the cases of normal and non-normal distributions. The design examples were provided for statically indeterminate plane and space truss structures, and the validity of the proposed method was demonstrated.
As glass filaments are susceptible to flaw at surface, the distribution of the strength of glass fibers is sensitive to the experimenter's skill in handing. The mean value of strength increases rapidly with the development of the skill first, and then becomes to attain a constant value. The standard deviation also increases with increasing skill first, but after that it deceased and reaches finally a constant value. As a result, the coefficient of variation decreases monotonically till the high degree of skill is attained. The weakest link model of a mixed flaw type was proposed here to analyze these changes in the strength data. This model consisted of a link of two kinds of flaws; the original flaws and those introduced during specimen preparation. The strength distribution of the link with either type of flaws was assumed to be Weibull, and the difference in the Weibull distribution between the original flaws only and the combined flaws was taken to give a measure of the skill. The theoretical results showed a good agreement with the experimental results.