In order to examine the behaviour of fatigue cracks in carbon steel subjected to rotary bending and cyclic torsion, both the technique utilizing slip-initiation phenomenon in the material and the copper electroplating method of stress analysis are applied. Based on the above results, the fatigue-notch factor of cracked specimens is compared with the stress-concentration factor at crack tip. It is found that more accurate results can be obtained by the slip-initiation technique than by the electroplating method, if the measuring region at the crack tip is identical. The change in metal properties caused by the formation of fatigue cracks has no effect on the crack propagating stress in pure iron, but it affects notably the crack propagating stress in eutectoid steel.
An attempt was made to utilize a computer for data registration, reservation and comparison of metal fatigue test results. More than 1000 sets of data, each of which contains the program fatigue test results of one test specimen recorded on three punched cards or 80×3 columns, were punched out and then transfered to a magnetic tape as well as to a punched tape. They were processed through a newly developed FORTRAN program. The process described here should be valuable for establishing a standard way of data presentation.
Fluctuating tensile load test was carried out on the stranded wire rope which was applied to the hanger rope for the long span suspension bridges. The S-N curve and the fatigue limit diagram were plotted. The tendency of increase of broken wire and the dynamic extension of the wire rope during the test were also measured to find out the progress of fatigue of the wire rope. The results obtained are summarized as follows (1) The S-N curve by which the fatigue strength of the wire rope can be appreciated is obtained, and the fatigue limit is about 21kgf/mm2. (2) Mean stress σm does not practically affect the fatigue strength. (3) The wire breaking due to fatigue occurs more frequently on the low cycle side than the high one. (4) The wire breaking due to fatigue occurs concentratively in the core rope than in the outer one. (5) The wire breaking due to fatigue occurs shortly before the wire rope breaks. (6) It is difficult to find out the extent of fatigue damage and the indication of breaking by means of the change of dynamic extension.
In order to clarify the fatigue process of comparatively hard materials under a combined stress state, electron and optical microscopic observations were made successively on torsional fatigue damage on the surface of quenched and tempered 0.54%C steel by using the two stage plastic replica method which makes it possible to observe the fatigue process in reversed sequence, with special attention being given to the starting points of fatigue cracking. The process was then compared to that under rotating bending fatigue. The following conclusions were reached: (1) The process of crack initiation is such that the region to become a crack in the future is damaged as a whole without increasing the size at the specimen surface and is gradually turned into a crack. This process of crack initiation is essentially different from that of crack propagation, and the difference between these two processes was also observed in annealed steel. The process of crack initiation under torsional fatigue or rotating bending fatigue of the material used is, however, slightly different from that of annealed steel; the process in the material used is such that minute portions disrupted by the repetition of slip appear and scatter discretely over the region to become a crack in the future, and then they are joined together to become a crack. (2) The torsional fatigue crack initiated in the direction of maximum shearing stress propagates slightly in the same direction and then follows the direction perpendicular to that of maximum tensile stress. In this instance, the feature of the deformation of the material in the vicinity of the crack tip is similar to that of rotating bending fatigue at the beginning of the tensile mode of crack propagation.
The effect of two metallurgical factors, i.e., inclusions and the rolling-induced ferrite-pearlite laminated structure, of SB56M carbon steel on the fatigue process of specimens with or without a slit was examined under cyclic torsional stressing. Specimens with different directions of anisotropy relative to the specimen axis were prepared. A half of them were annealed, and the rest were spheroidized in order to eliminate or decrease the effect of the laminated structure. Micro- and macroscopic observations on processes of fatigue damage were made on the surface of the specimens subjected to either low- or high-cycle fatigue stress. Both the inclusions and the laminated structure were responsible on the micro- and macroscopic scales for anisotropic features of fatigue damage and failure, degree of which was affected by the applied stress levels, although the two factors played different roles in the fatigue process. In the case of smooth specimens, resistance of the pearlite bands in the laminated structure to propagation and interlinking of slip-band type microcracks initiated in the ferrite bands controlled the rate of damage and the paths of crack propagation at high stress levels. At low stress amplitudes the pearlite resistance to propagation of microcracks originated at inclusions were the most influential factor in the failure process of annealed specimens, while in spheroidized specimens, in which the laminated structure was diffused to a great degree by this heat treatment, the macroscopic anisotropy in failure appearances almost disappeared. In the case of slitted specimens, anisotropy in the crack propagation rates was observed for annealed specimens: the cause of this anisotropy may be interpreted in terms of the pearlite resistance to crack propagation. This anisotropy was almost diminished by spheroidizing treatment, although an anomaly was observed for a particular (45°) combination of the direction of the lamination and the specimen axis. The inclusions hardly affected the crack propagation process for both annealed and spheroidized specimens, except for the final stage of rapid crack propagation.
The fatigue process of steel is characterized by the existence of fatigue limit and coaxing effect. Many researches have been made from the viewpoint of strain ageing brought about by interstitial solute atoms of carbon and nitrogen. The effect of strain ageing on fatigue limit, however, is not fully understood. This experiment is undertaken from the viewpoint that fatigue is a competitive process of two mechanisms of damage and strengthening of the material. Two techniques are applied for this purpose. The change in the solute content of C and N atoms is measured by internal friction. Artificial ageing and reversion are applied in order to change the density of Cottrell atmosphere at dislocations. The experimental results are as follows: The solute atomic content in Ferrite decreases as the fatigue process advances. The material is hardened by strain ageing to the extent of 1∼1/2 times of strain hardening. The fatigue limit is explained by the competitive relation between active slip systems and strain ageing.
Existence of the fatigue limit in steels has been explained in connection with the change in grain size or strain ageing. In this work, the effect of strain ageing on the fatigue strength of mild steels with different sensitivities to strain ageing was investigated. In order to know the mechanism of the effect of strain ageing on fatigue, the measurement of plastic strain amplitude, εpa, and observation of slip lines on the specimen surface were performed during the high cycle fatigue test under constant stress amplitude. The important results obtained were as follows: (1) The steel with the higher sensitivity to strain ageing had the higher fatigue strength. (2) In the steel with the high sensitivity to strain ageing, εpa increased very rapidly in the early stage of fatigue, then decreased immediately. On the other hand, in the steel with the low sensitivity, εpa increased acceleratively after some incubation time, which depends on the cyclic stress level. (3) Such a behavior of εpa is well correlated with the change in the density of slip lines during the test.
Rimmed, silicon-killed and aluminum-killed steels (0.1%C) were fatigued in rotating bending. The solute content was modified by annealing, low-temperature quenching and slow cooling. The strain ageing potentials which were measured in terms of the change in static yield strength were in the range of 4.3∼17.5%, when specimens were prestrained. by 2.5∼6% and then held at 100°C for 40min. The fatigue limits were observed to be in the range of 14∼24kg/mm2. Although both the fatigue limit and the strain ageing potential increased by quenching, generally, no close relationship between them was apparent. A much better correlation was found between the fatigue limit and static yield strength. However, a knee in σ-N curve, at which stage-I cracks became non-propagative, shifted towards a small number of cycles with the increase in strain ageing potential. The fatigue limit of quenched specimens, in which stage-I cracks had been introduced by cyclic stressing above the limit, Was enhanced by statical ageing at 100°C for 40min. before fatigue. Contrary to this, with the slowly cooled steel, no strengthening effect due to ageing was observed. For the fatigue limit of quenched specimens, the same strengthening effect as statical ageing is to be expected by strain ageing during stressing for a large number of cycles. The increase in the limit by quenching, statical ageing and compressive residual stress was estimated to be about 9, 1∼2 and 2.5∼5kg/mm2, respectively. It is concluded that strain ageing during fatigue can make a contribution to the arrest of propagating stage-I cracks, but the effective contribution to the increase in fatigue limit due to quenching is made by substitute hardening and compressive residual stress rather than strain ageing during fatigue.
The manifold repeated rotating bending fatigue tests are carried out at two stress levels on three low carbon steels, which are rimmed, Si-killed and Al-killed steels having the different ageing abilities. These tests constitute a characteristic feature that the cumulative frequency of programmed loading cycles is mainly occupied by the understress lower than the fatigue limit. The fatigue life under programmed loading is plotted against the equivalent stress, which is estimated on the basis of the modified Miner's law, both on log-log and semi-log diagrams. All experimental points of rimmed steel fall on the right side of the straight line extrapolated from the linear part of S-N curves on both diagrams. Those of Si-killed and Al-killed specimens seem to lie mostly around the above extrapolated straight line, but some experimental points in the longer life range tend to deviate from the straight line toward the shorter life direction on log-log diagram. The cumulative cycle ratios based on Miner's law are mostly larger than one in rimmed steel, while in Si-killed and Al-killed steels the greater part of these values are less than one. These facts indicate that in the former the understress contributes to the increase of fatigue strength for the higher stress above the fatigue limit, while in the latter it gives fatigue damage to the specimen. The results of surface structure observation and hardness measurement in the fatigued specimen suggest that the strain ageing gives a remarkable effect on fatigue strength. The fatigue test results mentioned above can be explained along this line.
The main objectives of the present study are to investigate the effect of overstressing on fatigue limit and then to make clear the effect of strain aging in such case. For these purposes, by using two kinds of low carbon steels, Al-killed and rimmed steel having different strain-aging abilities, fatigue tests were carried out under rotating bending and torsion. The slip line and crack initiation curves, as well as the S-N curves, and the damage curves were drawn for each case. The variations of hardness and temperature on the surface of the specimens during stress cycling were measured. The most important result obtained from these experiments is that the damage curve for the rimmed steel with higher strain-aging ability lies relatively in the side of shorter life as compared with that of the Al-killed steel with lower strain-aging ability. Therefore, it seems that the strain aging during stress cycling is not effective to the damage on fatigue limit caused by overstressing.
The authors carried out some fatigue tests under a set of program load which include peak load. The conclusions are as follows: (1) The effect of loading sequence on the fatigue life becomes greater, when the lowest stress level σ1 in program load becomes larger in this research. (2) The value of Σ(n/N) sometimes becomes considerably smaller than unity when the lowest stress level σ1 is small. (3) It appears that the understress that is lower than the endurance stress can reduce the fatigue life of the specimen in a program load. (4) The increasing rate of horizontal deflection of the fatigue specimen is greater than that of vertical deflection, under the program load which include large stress.
The effect of periodic overstressing on fatigue crack propagation was studied with sharp notched specimens of a low carbon steel, and the result was discussed in the light of the linear cumulative damage concept. A very small number of cycles of overstress applied intermittently during a very large number of cycling of understress had an extremely detrimental effect; namely, they yielded a remarkable acceleration of crack propagation (about one hundred times) and a very small value of the cumulative cycle ratio (less than one hundredth). Fractographic examinations revealed that the mechanism of crack propagation under the periodic overstressing was quite different from that under steady stress conditions, consistently with the high rate of crack propagation. Questions were raised on the conventional fatigue design based on the results with smooth specimens under steady loads, and also on the way of presenting the fatigue test data under program loads based on the modified Miner type laws.
A fatigue test has been performed using a rotating bending testing machine on the smoothed specimens of two kinds of low carbon steels, CH10A and CH10R, which show the endurance limits with well-defined“knee”. Then, the mechanism of endurance limits has been investigated by means of electron microfractography and of fracture mechanics. The main results obtained are as follows: (1) The crystallographic facets of Stage 1 cracking are found in all specimens both broken and unbroken under stresses near the endurance limits. (2) The endurance limit is not the critical stress for cracking, but for transition from Stage 1 to Stage 2 cracking. (3) The criterion of transition from Stage 1 to Stage 2 cracking can not be expressed only by the stress intensity factor. A model for the mechanism of transition has been devised and the effect of active slip planes which cross the Stage 1 crack plane has been discussed.
In this study, the initiation and propagation behaviors of cracks emerged in 18% Ni maraging steel sheets were investigated under cyclic tensile loading. A shallow notch was provided to the sheet specimen in order to facilitate the direct observation by an optical microscope. The main results obtained are summarized as follows. (1) Fatigue cracks initiate mainly from the inclusions at the notch root near the middle of the thickness, and the fracture first proceeds along the prior austenitic grain boundaries or the packet boundaries. Then the main crack propagates, forming ridge patterns with dimples until the catastrophic shear fracture occurs. (2) An exponential relation holds between the stress intensity factor range and the crack propagation rate even in the earlier stage of the propagation. The value of the exponent is about 2.63 and is different from that in the progressed stage obtained in the previous research. (3) Secondary cracks are observed everywhere, which intersect the fracture surface. These cracks have no characteristic orientation in the region of the ridge pattern, but they tend to appear in parallel to the striations at the place where the latters are formed.
Corrosion fatigue behaviors have been studied on a low carbon steel exposed to continuous or alternate immersion in 1% NaCl under low stresses in the long life range up to 109 cycles corresponding to life time up to 104 hours. S-N curves of the long life corrosion fatigue are composed of two curves having different gradients, those belonging to the short and long lives being corresponded to the corrosion fatigue stage and to the corrosive crack-extension stage, respectively. In the former stage mechanical repetitive stresses extend fatigue cracks directly, whilst in the latter corrosive dissolution at crack tips brings crack extension. Fractured surface morphology reveals also the change corresponding to the S-N curves, the inter-and transgranular mixed modes in the corrosion fatigue stage being replaced by the corrosive dissolution modes in the corrosive crack-extension stage. The life cycles under the alternate immersion are longer than those under the continuous one when compared at the same net immersing duration. This is due to the delayed growth of corrosion pits and the increased wedge action of corrosion products in the corrosion fatigue stage and to the delayed extension of crack tips by corrosive dissolution in the corrosive crack-extension stage.
A new technique, called “unloading elastic compliance method”, to detect the fatigue crack length as well as the crack closure behavior inside of the specimen by using strain gages has been developed. The crack propagation rate and the crack closure behavior at a low stress intensity level were investigated on various kinds of materials using the above mentioned method, and it was shown that the crack propagation rate above the region of 10-5mm/cycle could be represented by the effective stress intensity range, ΔKeff. However, the crack closure behavior at the threshold level was found to be strongly dependent on the material properties.
It was expected by the authors that the existence of the ‘knee’ on the S-N curve would be related to the instability of the structure of metals and the sudden bend of the P-N curve would correspond to this knee. With this expectation in mind, the effects of the age hardening ability of materials on the S-N diagram and on the P-N diagram were studied by fatigue testing. Two kinds of rimmed steel were prepared for testing; specimens A, in which the age hardening ability was attained through heat treatment, and specimens B, in which the ageing process in specimens A was completed by leaving them in room temperature. The S-N diagram and the P-N diagram were made from rotating fatigue testing (the test speed of 1500 cycles per minute). As a result, for the test on specimens B, the knee did not appear on the S-N diagram and the P-N diagram showed a logarithmic normal distribution. On the other hand for the specimens A, the knee appeared in the S-N diagram near N=106 and also the bending point in the P-N diagram near N=5.8×105. It was concluded that these effects were due to the ageing ability of the material.
The main purpose of the present study was to investigate the fatigue life distribution at stress levels around the endurance limit σw. Three multiple-type rotating bending testing machines were manufactured, each of which can test eighteen specimens simultaneously. Approximately twenty specimens were tested at each respective stress level around the endurance limit σw with the maximum number of stress cycles of about 5×108. The specimens used were heat treated rimmed steel CH10R bars. The results were analyzed using the Weibull distribution of mixed type. The major conclusions are as follows; (1) In the stress range above σw, there are fixed relationships between the parameters of Weibull distribution and the stress, that is, the shape parameter m is constant, and the scale and location parameters, a and γ, have linear relationships on log a-σ and log γ-log σ diagrams. (2) The Weibull distribution of mixed type F(N)=p1F1(N)+p2F2(N), p1+p2=1, is well fitted to the fatigue life distributions at the stress levels around σw. The parameters m1, a1 and γ1 of F1(N) are given by extending m-σ, log a-σ and log γ-log σ relations in the stress range well above σw to the stress range around σw. And F2(N) appears only at the stress levels around σw, whose parameter m2 is constant and parameters a2 and γ2 have different linear relationships on log a2-σ and log γ2-logσ diagrams.
Firstly, it is discussed in this paper how to estimate statistically the shape and scale parameters in the fatigue life distribution assumed to be a two-parameter Weibull type. As the result, the former is found to be well estimated by TOSE, i, e., the two-ordered-statistic estimator first developed by I.C. Whittaker, et al., and the latter by MLE, that is, the maximum likelihood estimator. Secondly, the method is discussed for determining the design safe life to meet the desired reliability goal. Further, the actual fatigue life distribution and the dependency of the shape parameter upon the testing conditions are considered by applying the aforementioned reliability analysis technique to the empirical fatigue life data on S25C, S35C, S45C and S50C carbon steels under constant amplitude stressing, programmed loading or superposed stressing. The importance is also pointed out that the design safe life in the structure has to be determined on the basis of TTFF, that is, the time to the first failure of a component.
From a standpoint of axle design, a detailed statistical information about the fatigue limit of press-fitted axles is required. Little data are, however, available for the design of press-fitted axles. The objective of the present paper is to obtain a basic information on the scatter of fatigue limits of the press-fitted axles for the statistical evaluation of design safety factors, and then to confirm the allowable stresses given in JIS (Japan Industrial Standard) for the axle design. In the present paper, the rotating bending fatigue test was made on the normalized or quenched and tempered medium carbon steel axles. The stair-case method was applied to investigate the statistical properties of fatigue limit of the press-fitted axles. The results show that the standard deviations which are the measure of scatter in the fatigue limit are 0.22kg/mm2 for the normalized and 0.63kg/mm2 for the quenched and tempered steel axles.