Three series of specimens were specially prepared so that they may contain various degrees of MnS non-metallic inclusions, other parameters being kept constant especially the amount of manganese in solution as much as possible. Fatigue tests were carried out with a rotating bending fatigue machine. The results obtained are as follows: (1) The definite relationship could not be recognized between the inclusion ratings and the statistical nature of fatigue life of steel, which is in accordance with the previous work by one of the present authors, but contrary to the result by Epremian and Mehl. (2) The ferrite grain size is a predominant metallurgical factor in the statistical nature of fatigue fructure of steel. (3) For the same stress amplitude, the scatter of fatigue life increases and the probability of fructure decreases with the decrease of grain size. (4) For the equivalent stress amplitude, the scatter of fatigue life appears not to be affected by inclusion ratings and ferrite grain size. (5) The longitudinal fatigue strength of carbon steel is not so much affected by the content of MnS as inclusions. The effect by ferrite grain size or other statistical parameters may also be large.
In the course of the development of the modern light construction, the spectrum fatigue tests have become more and more important. The authors made several spectrum fatigue tests by changing stress in normal distribution under rotary bending. The stress spectrum was obtained by changing the dead-loads. Two kinds of annealed mild steels were tested and the following results were obtained. (1) The relation between the total number of stress repetition Ng and the meam amplitude of alternating stresses σa was plotted in normal S-N curve. Then the endurance limit σaw obtained from the σa-Ng curve was expressed as follows: σaw=σw-σr where, σw is the endurance limit of the original S-N curve and 2σr is the difference between the maximum and the minimum amplitude of stress on the normal distribution of stress. (2) Total number of cycles to failure Ng is not affected by the order of stressing in load spectrum or by the load spectrum pattern, and calculated by the following formula: Ng=N1/α1+α2N1/N2+α3N1/N3+…+αkN1/Nk where, α1, α2, …, αk are the cycle ratio of stresses σ1, σ2…, σk in one load spectrum, and N1, N2, …, Nk are the number of cycles to failure for the stresses σ1, σ2, …, σk in the original S-N curve. (3) The results obtained by Corten-Dolan's method for calculating the total number of cycles to failure Ng in the spectrum fatigue. tests were found in safety side in almost all experiments.
In this report, the construction and the capacity of a newly designed torsional fatigue testing machine for wire is stated. This testing machine has the following characteristics: (1) This machine can automatically record the development of torsional deformation. (2) Mean stress can be given to the wire specimen. (3) A slender wire can be used as a specimen so that data obtained in this experiment may be applied to the coiled spring.
The present investigation has been made in order to determine the influence of induction-hardening on the fatigue strength of shrink-fitted specimens. Cyclic direct stress fatigue tests were made on shrink-fitted specimens 20mm in diameter of NiCrMo steel. The results obtained were summarized as follows: 1. The reduction ratio 2.3 was obtained by comparing the endurance limit for a shrink-fitted specimen with that of a plain specimen. 2. The endurance limit of a shrink-fitted specimen was increased by about 2.1 times by induction-hardening. 3. The fatigue strength of the induction-hardened specimen with a shink-fitted member may be estimated approximately from the modified Goodman diagram, in which the residual compressive stress on the surface layer of specimen is assumed to be mean stress.
The present investigation was carried out for the size effects of test piece and various testing conditions on the torsional fracture behaviour, using a newly developed dynamic torsion machine. Dynamic torque-twist curves were obtained photographically at various temperatures. The maximum rate of straining was 60per second and the range of temperatures from +200°C to -196°C. Yield and fracture strength and plastic strain to fracture were not affected by the variation of gauge length of the test piece, but yield and fracture strength increased with decreasing of the diameter. The size effects on torsional strength were also appeared for hollowed specimens as well as for solid ones. In the static torsion test, farcture type at comparatively low temperature had a ductile manner and the cleavage fracture appeared only at -196°C under the tested conditions. On the other hand, in the dynamic test, the specimens used for the investigation took a ductile fracture, excepting a specimen having a larger grain than the A.S.T.M. No 4. The torque-twist curves in static test or room temperature dynamic test dropped at maximum torque after drop in load. To the contrary, in the low temperature dynamic test, after reaching the maximum stress, torque suddenly decreased with increasing twist angle and then kept a constant value until the beginning of fracture. Ruptured specimen with cleavage type due to the dynamic torsion test had been broken at maximum torque. These strange phenomena in the fracture manners and the torque-twist curves were discussed and assumed to be due to the temperature rise in the tested part connected with the rapid large plastic deformation.
Setting treatment is ordinarily given to helical springs in order to improve the fatigue strength and to suppress the fatigue deformation. The effectiveness of setting treatment on helical springs has been proved by several experiments, and recently the authors examined this effect on curved leaf springs, which verified the same availability. The beneficial effect of setting seems to arise primarily from two factors, favorable residual stress and work hardening. As to the measurement of residual stress in helical springs, no calculation formula according to the dissecting method is found as yet due to the complexity of their shape. In practice, the helical springs shaped by the coiling machine are likely to have triaxial residual stresses as well as torsional one, being in an extremely complicated state. In this study, as a first step, the problem was limited to measure only the torsional residual stress introduced by setting, and along this line a new calculation formula was derived. Helical springs of the material SUP 2 were heat treated in two ways, that is, tempered at the temperature of 450°and 500°C separately, and then subjected to setting treatment. Fatigue tests were made in partly pulsating compression with both springs, which showed an effectiveness of setting distinctly. Residual stress measurements were carried ont using a new formula, at the stress levels above and below the fatigue limits, and the changes in residual stresses were examined. The results were in quite good agreement with those obtained previously.
We have studied on a relation between minimum creep rate and time duration to rupture of unplasticized polyvinyl chloride. Although it has been shown by many reports on metals that they are reciprocally proportional to each other, we have recognized through our experiment that the relation is as follows; tr=M/(εmin)θ, where tr is time duration to rupture, εmin is minimum creep rate, M and θ are coefficients which change their values depending on testing temperatures, two ranges of εmin and conditions of heat treatment of specimens. Values of coefficient θ for normal specimens become larger than 1 in the lower range of minimum creep rate, though it is almost equal to 1 in the higher range in which time duration to rupture is approximately proportional to reciplocals of minimum creep rate. For the specimens kept at 60°C for 5 days, no range is recognized where the value of θ is almost equal to 1. A master rupture curve of normal specimens is drawn in the range of temperature -10 to 40°C, derived from the formula proposed by the authors in the preceding paper and from the above mentioned parameters.