To secure integrity of industrial product, various uncertainties surrounding the design technology must be treated properly. To do this, the uncertainties underlying during the following phases: evaluation of loads, modeling technique of systems of interest, evaluation of material strength, maintenance and inspection and so on, have to be clarified and effective measures to grasp the degree of the uncertainty and to reduce the inherent uncertainty have to be established. The present paper shows the classification of the uncertainties and the fundamental relation with the design technology, and mentions a reliability design method based on probabilistic and statistical techniques, which has been getting popular in many engineering fields, and finally raises some future problems of design methods from the viewpoint of the uncertainties.
If the storaged substances in a pressure vessel like H2, LNG and so on explode, the disaster range is expanded because the explosion power is increased and the fragments of the shells are scattered far and wide. As a result, there are some cases where the actual disaster range may be larger than the calculated one by the conventional method. No report on the disaster of chemical plants has apparently been published. Hence, a computer simulation program which is able to calculate the disaster range of explosion in pressure vessels has been developed, and the accident of gas-explosion which occurred in 1986 has been analyzed by this program. It is shown that the calculated disaster range almost coincides with the actual one and it is recognized that this program is very useful for safety evaluation for chemical plants with many pressure vessels.
The computer program of disaster simulation was developed and the calculational results were shown in the previous paper. But, the function of extinguishment was not included in the developed program. Therefore, a computer simulation program which takes extinguishment into consideration has been developed and their effects are analyzed by using sprinklers in this paper. The total amount of supplying water and the amount for each tank have to be determined to calculate the effect of fire-extinguishment. Therefore, the distributions of supplying water for each tank are assumed as follows; (1) Water is distributed only to the nearest tank from the first broken tank. (2) Water is distributed proportionally to each broken time of tanks that can be calculated by heat of fire. (3) Water is distributed in order to delay the time when the second tank is broken. As an example, the damage progression for a model chemical plant based on the water distribution by the above methods is simulated. As a result, it is clear that the water supply of type (3) is suitable to extinguish. It is revealed that the proposed program in this paper is very useful for safety evaluation and protection against calamities for plants.
This paper attempts to improve a fuzzy expert system which was developed previously to help maintenance engineers to evaluate the damage state of reinforced concrete bridge deck. One of important items in the maintenance program is that the repairing methods are automatically selected. The present system can provide several appropriate repairing methods by taking account of the damage cause, damage degree, and damage propagation speed. Since the system is built on a 32 bit engineering workstation and is written with Common Lisp, anyone can use this system without difficulty. Furthermore, adding the ability of executing backward reasoning in the procedure of fuzzy inference, it is possible to reduce the computation time to a quarter of that required for the previous system.
It has been shown from the results of round robin tests conducted under VAMAS project that the fatigue property may vary from one researcher to another due to the difference of evaluation models used. It is important to clarify the effect of the S-N evaluation model used on the reliability analysis. In this paper, the fatigue data are fitted to five regression models and the goodness of fit is investigated. The fatigue data are picked from the fatigue database which was provided by JSMS. No significant difference could be found concerning the goodness of fit against the models. However, the results differed slightly depending on the factor used in the goodness of fit. Moreover, the effect of the model on the evaluated fatigue life is investigated for the fatigue under random loads. The fatigue life evaluation was formulated for three regression models and the fatigue life under various rms values was calculated for several materials. The effect of the fatigue model on the evaluation life was clarified.
Fatigue life of a plate with collinear two cracks is theoretically investigated with the random variation of crack propagation resistances taken into consideration. Firstly, the crack propagation process is modeled as a bivariate stochastic process governed by a randomized Paris law. Secondly, a joint probability density function of the two crack lengths is derived from the Fokker-Planck equation by using a difference scheme, the initial condition being an approximate solution. Finally, failure probability of the plate is numerically evaluated, and its sensitivity to the initial crack state is examined.
Rolling fatigue tests were carried out on hot isostatically pressed silicon nitride bearing balls. At three load levels of the maximum Hertzian contact pressure pmax=5900, 6740 and 7110MPa, 13 balls each were tested, and the life distribution was studied. It was found that the rolling fatigue life followed approximately a 2-parameter Weibull distribution with the shape parameter of α≅1 at each load level. The relationship pmaxκL50=const (κ≅10) was found between pmax and the median life L50. Microscopic observation of flaking was also performed, and two types of flaking were found. One is flaking with the depth of the order of 100μm, and the other is a shallower one. The depth of the former type was found to coincide approximately with the depth at which the maximum shear stress occurred. At lower load levels, the former type was dominant. Frequency of appearance of the latter type increased with an increase in load level. Comparing the shape parameter of α≅1 and the exponent of κ≅10 mentioned above with those for the cyclic bending fatigue of plain specimens of silicon nitride, it was suggested that the mechanism of rolling fatigue of silicon nitride was different from that of cyclic bending fatigue of the same material.
Mechanical properties at high strain rates were investigated by considering the specific strengthening mechanisms involving the effects of grain size, solid solution and second phase particles. Aluminum and its alloys were tested in tension or compression at strain-rates between 10-5 and 2000s-1 at room temperature. The conclusions were as follows: (1) Aluminum tested at high strain rates could not exhibit strengthening due to grain size refinement according to the well known Hall-Petch equation. (2) Solutes such as magnesium were able to increase the yield stress of aluminum at high strain rates. The yield stresses at high strain rates for Al-Mg alloys increased with magnesium addition, but the variation in strengthening rate with an additional content of magnesium was almost similar to that at low strain rates. (3) The treatment of strengthening by the second phase particles was clearly valid to improve the yield stress of Al-Mg-Si alloys at high strain rates. Basically, a decrease in the mean spacing between the second phase particles led the yield stress at high strain rates to be high. It was suggested that an optimum control in distribution of second phase particles was the most valid way for an improvement of the yield stress at high strain rates comparing with other strengthening methods.
Progressive plastic buckling appears in the axial symmetric behavior of circular tubes subjected to static loads or low velocity impacts, when the inertia effect is negligible. For high velocity impacts in which inertia cannot be neglected, the studies by Tanaka and Kurokawa revealed the appearance of the same modes as those in static tests. However, a series of impact tests on tubes by Florence et al. recorded the phenomenon of dynamic plastic buckling in which the deformation consists of many waves superposed on a uniform plastic deformation. This phenomenon was found for moderately thick circular tubes subjected to high velocity axial impacts. Ren et al. observed that dynamic plastic buckling developed on thin circular tubes. Thus, the plastic buckling behavior for high velocity impact tests on circular tubes consist of both progressive plastic buckling and dynamic plastic buckling, but the detailed behavior is not clear yet. The purpose of this study is to clarify the characteristics for the transition process from progressive plastic buckling to dynamic plastic buckling. In particular, the following features are examined: (1) The comparison and consideration of various experimental methods used in the previous impact test results. (2) The influence of end condition on the static buckling behavior of circular tubes. (3) The consideration of some impact test results.
In this paper crushing tests of circular and square tubes under impact loading are conducted to investigate the energy absorption ability by using a newly constructed impact machine for crushing tests. The difference between the static and impact crushing behaviours of aluminium specimens is found not significant, while the results of steel specimens are strongly affected by the dynamic effect. The energy absorption ability of steel square tubes under impact loading is much higher than that under static loading because the crushing modes in static and impact tests show a big difference each other, which would come from the strain rate effect of steel, inertia force, imperfection of specimens and so on. It may be difficult to discuss the dynamic/static ratio in crushing tests only by the strain rate effect of material used. The absorbed energy by steel specimens is absolutely higher than aluminium specimens, while the absorbed energy by unit mass shows the advantage of aluminium because the mass density of aluminium is one third of steel. Circular tubes have a better energy absorption ability than square tubes under the condition that the size and area of cross section are approximately equal.
It is well known that two metallic plates are bonded momentarily at a high velocity impact. It is guessed that the plastic work occurring on the contacting surface is converted into heat so that the bonding proceeds due to the heat. Although this phenomenon is observed on an explosive welding, it is considered that the bonding mechanism of the dynamic compaction is also the same. The exposive welding and the dynamic compaction have been already used industrially. However, the bonding mechanism is not clarified sufficiently. The aim of this paper is to investigate the formation of bonding area at the impact welding using a longitudinal impact test. In the longitudinal impact test, a projectile accelerated by a gas gun collides with a target so that the welding occurs at the interface between two metals. The material used for the projectile is a commercial pure aluminum and the target materials are mild steel, copper and stainless steel. The bonding area is observed using a scanning acoustic tomograph and the bonding strength is examined by a quasi-static tension test. Additionally, the profile of the crater on the target after impact is also measured. It appears that the formation of the bonding area is dependent on the deformation processes of the projectile and the target, as indicated by the following results. The radial deformation of the impact face of the projectile is very restricted by the friction force at the interface or the deformation of the target so that the bulging accompanied with the folding of the projectile side occurs and the mushrooming profile appears after impact. Although a few areas considered to be bonding ones are found from an ultrasonic image after the bonding, the bonding force of the central part is much stronger than that of the other parts. Subsequently, it is presumed that the bonding mechanism of the copper target differs from those of the other targets.
To study the plastic zone size of dynamic fracture surface, instrumented Charpy impact tests of high carbon bearing steels were conducted. The plastic zone size beneath the fracture surface was evaluated by the X-ray diffraction parameters. The results gave the following three experimental conclusions. The first, a good correlation exists between the plastic zone size and the dynamic stress intensity factor as ωy=α(Kd/σY)2, and the values of α depend on materials. The second, α can be calculated from the equation α=0.15[(B0+2.12)/(BF+2.12)]2, where B0 is the average half-value breadth of the base material and BF is the maximum half-value breadth beneath the fracture surface. The third, both ωy and α can be simultaneously measured on the fracture surface, and Kd can be evaluated by using the above two equations. This technique was applied to an actual fracture surface and confirmed to be very effective for failure analysis. It is advantageous over other test methods and it will find wider application in the industry.
To obtain the dynamic Mode II interlaminar fracture toughness of CFRP under impact loading, an impact bend test using split Hopkinson pressure bar (SHPB) method was applied to the end notched flexure (ENF) specimen, and the estimation of GII was investigated. The dynamic behavior of the ENF specimen during impact bending was analysed by the finite element method, and the history of the energy release rate GII of the specimen was calculated using the crack-closure method. The energy release rate GIICC calculated by the crack-closure method was compared with the energy release rate GIIST estimated by the formula for the static ENF test and the load history of the output bar of SHPB. The results obtained are as follows. For the step input loading, the ratio GIIST/GIICC oscillates strongly around unity and the oscillation continues after considerably long time, while for the ramp input loading with a finite rise-time, the oscillation is damped more quickly and the ratio GIIST/GIICC approaches to unity more gradually as the rise-time increases. Therefore, the dynamic Mode II fracture toughness can be estimated by the formula for the static ENF test if the ramp input loading wave with an appropriate rise time is applied.