It is a recent tendency that high temperature power plants with more power and higher efficiency, are more in demand, especially for the steam turbine installations. But such increase in power and efficiency is largely dependent on metallurgical advances. Designers and engineers need to know of limitations in regard to the senility of the metals they are considering of using. However, they are obliged to deal with the short time creep rupture data, for the long time creep rupture test requires much time and money. Such short time data needs extensive care in extrapolation for a longer period of time. Using many data collected by Nihon Gakujitsu Shinko-Kai, the long time creep rupture strength for 18-8 Ti steel and 18-8Cb steel was estimated statistically by various methods, i.e., the extrapolation method by one straight line, Larson-Miller parameter method, Dorn's parameter method, Manson-Haferd parameter method and the extrapolation method by broken line. Although the parameter methods are more convenient because the time and temperature can be expressed with one parameter, the reliability is rather low. The most reliable one in these methods was the extrapolation method by broken line. It is considered that metallurgical changes result in breaks on the straight line. One break appeared at 600°C for 18-8Ti steel and two breaks for 18-8Cb steel. The break point of 18-8Ti steel and the second break point of 18-8Cb steel correspond to precipitating of σ phase. A more reliable extrapolation became possible from these break points, and the slope of each straight line obtained experimentally. Calculation was conducted with the use of this extrapolation method of 105hr rupture strength at various temperatures for 18-8Ti steel and 18-8Cb steel.
To obtain the high temperature strength data of Austenitic stainless steels, creep rupture tests at 650°C for as long as 10000 hours are in progress with five steels-two 18Cr-12Ni-2.5Mo type steels, one 18Cr-12Ni-0.25Ti type steel, and two 18Cr-12Ni-1Nb type steels, heat treated at 1100°C for 1 hour and water-quenched. Three specimens were tested under each stress level to investigate the scattering of data. The results obtained from the experiments which were conducted for 3000 hours are reported in this paper. The change of the tensile and the impact properties of these steels during the aging and creep at 650°C were also investigated. The impact values of these steels fell to one half of its initial values during several thousand hours of aging at 650°C and the 18Cr-12Ni-2.5Mo steel showed a larger drop of impact value and more increase in tensile strength during the aging than the two other types of steel. To study the strengthening mechanism in austenitic stainless steels at elevated temperature, the precipitation and dislocation behaviour during the aging and creep of a 18Cr-12Ni-2.5Mo and a 18Cr-12Ni-1Nb steel were observed by transmission electron microscopy. In the specimens of both steels, solution-treated at 1 100°C, the dislocation density was rather high because of the thermal strain produced in the quenching in water. In the 18Cr-12Ni-2.5Mo steel, M23C6 formed on dislocations during the aging, and after aging for 1000 hours, the dislocation loops surrounding the coarsened precipitates of M23C6 were observed, which suggested that the precipitates threw-out dislocations during its growth (“precipitation growth dislocations”). In the specimens aged under the creep stress, much more dislocatinos tangled around the precipitates than in those aged without stress, and this showed that moving dislocations were arrested by the precipitates during the creep deformation. In the 18Cr-12Ni-1Nb steel, many dislocations were produced around the large undissolved NbC particles in the solution treatment, because of the difference in the thermal expansion coefficient between NbC and the matrix. The NbC formed on dislocation during the aging, and the NbC precipitates also threw-out precipitation growth dislocations”, on which very fine NbC precipitates were formed. In the specimens deformed under creep stress, numerous dislocations tangled around the precipitates of NbC which formed on dislocations produced during the solution treatment and in the earlier stage of creep deformation. From these observations, it was comfirmed that in these steels, precipitation of M23C6 or NbC contributed to the strengthening against creep at 650°C in typical precipitation hardening mechanism. In both steels, very thin platelets of M23C6 or NbC formed on stacking faults were observed, which appeared more frequently in the specimens aged under creep stress than in those aged without stress. Segregation or precipitation of the alloying elements on stacking faults also seemed to have some role in the strengthening of austenitic stainless steels at elevated temperature. Fewer dislocations were observed in the specimens ruptured after the creep deformation, and this would suggest the possibility of escaping of dislocations from precipitates during the ternary creep, but further studies are required about the sudden increase of creep strain and the initiation of cracks in the beginning of tertiary creep.
The effect of heat treatment on the creep rupture strength of 21/4Cr-1 Mo steel is well known. In order to realize the extrapolation of long-term creep-rupture strength, the microstructure in variously heat-treated specimens before and after testing and their creep-rupture strengths were investigated. The heat treatments are of 2 series, that is‘as quenched’and‘as quench-tempered’with the following cooling rate, water quenching, air cooling, 600°C/hr, 100°C/hr and 15°C/hr. The tempering was made at 700°C (1292°F) for 5hr. Testing temperature was limited to 566°C (1050°F). The kinds, shapes, composition and phase change of the precipitated carbides were investigated by the electron diffraction method of the extracted replica, and by chemical analysis, of the electrolytically separated residue. The relation between these results and the rupture time was considered. (a) The precipitates obtained during heat treating were Fe3C, Mo2C, Cr7C3 and M23C6, among which acicular carbide of Mo2C were formed in every specimen, particularly in ferrite region. M23C6 was found in the bainite structure chiefly. Sometimes Fe3C and Cr7C3 coexisted. (b) Among the‘as-cooled’test pieces, the air-cooled of bainite structure showed the highest creep-rupture strength, decreasing with the decrease of cooling rate. In the test-pieces rapidly cooled such as water-quenching, air-cooling, and 100°C/hr, which showed large decreasing slopes in the stress-time rupture curve, Fe3C and Mo2C precipitated abruptly and grew up. (c) In the tempered pieces, the stress-time rupture curve showed almost the same slope, and the strength decreased with the decreasing of cooling rate. The estimated values for 104hr were 9.4kg/mm2 for the air-cooled and 7.6kg/mm2 for the slow-cooled at 15°C/hr. (d) Both the stress-time rupture curves for some of the‘as-cooled’and the‘cooled-and tempered’crossed at the long duration of almost 104hr. The rupture strengths over this cross point were of much interest from the extrapolation point of view. (e) The elongation of the ruptured test pieces, both the water-quenched and the air cooled, decreased rapidly with the increase in testing duration. This is presumed to be correlated with the precipitation at the grain boundary and the phase change in the steel. This is the example of strong creep resistance showing short rupture time by grain boundary precipitation. Though some of main factors have been made clear from this experiment, possible other factors should be investigated for complete understanding of high temperature properties.
With the progress of high effective machinery, the fatigue for the plastic region becomes important. Herein, it is considered to be significant to detect the correlation between the constant strain cycle fatigue and the constant stress cycle fatigue, and this may make it possible to presume the mechanism of the fatigue for the plastic region. In the case of the constant plastic strain cycle fatigue, the formula for fracture has been formed and has been confirmed through many experiments. In the case of the constant plastic stress cycle fatigue, however, the formula for the fracture has not yet been estimated. This paper deals with the experiment of the constant strain cycle and constant stress cycle fatigues for the plastic region by the method of applying the tension and the compression on a Ni-Cr-Mo steel at a room-temperature, 350°C and 550°C. The conclusions obtained from the foregoing are as follows. 1) In the process of developing the Yokobori's theory, it was found that the static true strain at fracture corresponded to the range of strain at N=1/2 cycle as proposed by D.E. Martin. 2) The formula for fracture due to the constant stress cycle fatigue has been induced as follows. Assuming the static tensile stress-strain curve to be expressed by the formula σ=a+blogeεa, b; constant and using the formula N1/2Δεp=c then σa=(bloge1/√2+σf)-b/2logeN σf; the true static tensile stress at the fracture corrected by the method of P.W. Bridgman σa; the stress amplitude N; the cycle till the fracture 3) The calculation on this formula as above was confirmed to coincide with the experimental data at each temperature. The static tensile true stress at fracture; σf corresponded to the stress amplitude at N=1/2 cycle. 4) Further, the relation between the total energy and the total strain absorbed in the specimen until the fatigue fracture is discussed in the text of this paper.
In the design of boilers and other high temperature and high pressure installations, most of the tubes under internal pressure are still designed on the basis of the creep or creep rupture data of simple tension bar specimens. In practice, these tubes are loaded under multiaxial stresses. It is, therefore, a very interesting and important problem to elucidate the correlations between the creep or creep rupture strengths under simple tension and the multiaxial stress state. In the present study, the creep rupture tests on thin-walled tubular specimens of a low carbon steel and two sorts of low alloy steels, 1.25 Cr-0.5 Mo and 2.25 Cr-1 Mo steels, under internal pressure at constant temperature were carried out. The tensile creep rupture tests on bar specimens cut from the same tubular materials were also made, and their correlation stated above has been studied. The results obtained are summarized as follows. (1) Generally speaking, the creep rupture data of tubular specimens of Cr-Mo steels agreed most closely with the simple tension data on bar specimens of the materials when the equivalent stress on the wall of the tube was calculated by the average diameter formula, though according to the ASME formula, this agreement was obtained with somewhat less accuracy. (2) The visible cracks in the ruptured area were few in number and were very faint and small in size when the rupture occurred in the short duration of test. On the other hand, the visible cracks were larger and distributed widely over a bulging area when it took longer time to rupture. (3) The values of hoop strains and the micro-structures of ruptured tubular specimens were similar to those of simple tension tests. (4) It seems that the rupture is governed by the maximum shear stress criterion, since the ruptured surfaces in all the tests were the planes having the angle of about 45°with the tangential planes of the tubes. (5) The axial elongation in the ruptured tubular specimens were seldom observed.
In recent years so much attention has been called to the problem of creep or creep rupture in pressure vessels and tubes that operate at high temperature, and numerous theoretical and experimental works on this problem have hitherto been reported. There are, however, several open questions that must be discussed from the standpoint of multiaxial creep behaviors. The authors have taken interest in this subject and have been carrying out the creep tests of thick-walled cylinders under internal pressure at high temperature. The tests have been made with 0.19% carbon steel at the temperature of 450°C. The tubular specimens had a 400mm cylindrical gauge length, with the outside diameter about 50.45mm and the ratio 2 of outside to inside diameter (OD/ID). The specimens of uniaxial creep tests were cut from the same tubes. The internal pressure was supplied by the hand-operated water-pump and was maintained within ±1% by the pressure controller. The circumferential elongation on the outside diameter of the tubular specimen was measured by dial gauges and mirror-telescope instruments. The results of the tests revealed that the circumferential strain on the outside diameter versus the time relations for the tubular specimens were similar as the strain versus the time relations for the uniaxial tension. The axial creep for the tubular specimens was negligible. In addition to this, this report deals with the results of the measurement of residual stress distributions in the thick-walled cylinders after the creep tests. The measurement was carried out by the well-known Sachs' method. The X-ray method was also used for the residual stress determinations on the outer surface of the cylinder, and a fairly good agreement was obtained between the stress values measured by the two methods. The results obtained were as follows: No residual stress in the cylinder before the creep tests can be measured except near the outer and inner surfaces. The residual stress distributions after the creep tests under 1000kg/cm2 for 1 hour are similar as those under 1200kg/cm2 for 104 hours; that is, the circumferential and axial residual stresses change monotonously from compression on the inner surface to the tension on the outer surface, and the radial residual stress, which is smaller than the others, is compressive in the wall. The stress distributions in creep conditions determined by these measured residual stresses correspond to the steadystate stress distributions. It was also found that the circumferential and axial residual stresses measured by means of X-rays remained constant on the surface of the cylinders after the creep tests under 1000kg/cm2 for 1, 10 and 100 hours respectively. From these results it is found that, as far as the creep of the cylinders at 450°C under 1000kg/cm2 is concerned, the stress distributions change from the initial elastic state to a steady state in a short time. It is the authors' belief that the time required for the stress distributions almost to reach the steady state will be relatively short in the case of creep at high temperature and high stress. However, further discussions on a transient stage will be necessary from the point of view of multiaxial creep theory.
The present work has dealt with the stress relaxation characteristics of 12Cr-1 Mo-1 W-1/4 V steel which is used for high temperature equipments. The effect of initial stress on the stress relaxation properties of this steel was studied at 550°C by means of“Step-Down flow Rate”test1) in which the conventional lever type creep testing machines were used. Stress relaxation tests were also processed at several temperatures with constant stress initially applied, by the use of the specially designed stress relaxation testing machines which automatically kept the total extension of the test specimens constant. The magnitude of stress by which the residual stress of the test specimens was reduced with time manually (in“Step-Down”test) or automatically (in automatic type“Relaxation”testing machine), was 1.27kg/mm2 respectively. The test temperature was maintained constant to ±0.5deg. C at each of the tests and 0.06kg/mm2. continued to the period of about 1500 hours. Microstructure change of this steel after long term test procedure was observed with optical and electron-microscopic method. The results obtained are as follows: (1) The group of the residual stress-time curve obtained from“Step-Down”test changing the initially applied stress at given temperature, showed the tendency that each curve of the group approached one another with time. (2) The curves in log-log plots of stress versus strain rate which were derived from the residual stress-time curves of this steel consisted of two or three straight line portions. The stress dependence of the strain rate was larger in“the first straight portion”which indicated the earlier stage of the tests, than in the“second straight portion”which showed the later stage of the tests. (3) These stress versus strain rate curves plotted in log-log scale with different initial stress were nearly in parallel with one another at both the stages. (4) The stress versus minimum creep rate curve obtained from the creep test of this steel was also shown by two straight lines which had nearly the same slope as in the corresponding straight line portions of the stress-relaxation tests at the same temperature. (5) The group of residual stress-time curves obtained from the stress relaxation tests changing the test temperatures with the same initial stress, showed that the stress relaxation process of this steel was thermally activated, and that the activation energy for stress relaxation calculated from the curves in log-log plots of stress versus strain rate, was about 74000cal/mole in“the first straight portion”(stress range of more than 20kg/mm2) which was much the same as that for the self-diffusion of Alpha-Iron; 73000cal/mole9), and was about 58000cal/mole in“the second straight portion” (stress range of less than 20kg/mm2). (6) Some electron-microscopic structure change was observed in the specimen after these tests such as carbide precipitations along the needle-like portions of the tempered martensite phase and inside the Delta-Ferrite phase, and it was found that these structure change would affect the stress relaxation properties of this steel.
Most of the studies on the stress relaxation of metals reported so far are relevant to the relaxation of uniaxial tensile stress. The more general problem of complex stress relaxation has hitherto attracted but little attension and only a few works on this have been reported by E. A. Davis, J. Marin and A. E. Johnson et al. The experimental works in these investigations were carried out with only two materials, alminum and magnesium alloys. However, no information of this sort is available on steel yet. In the present study, the relaxation tests for combined tension and torsion stresses were carried out on the thin-walled tube of a low alloy steel (21/4Cr-1Mo) at 500°C. In the experiment, the stress ratio of torsional stress τ to axial tensile stress σ was kept at constant value ranging from zero (simple tension test) to infinity (pure torsion test) during the tests. For the relaxation condition, the total axial strain was maintained constant for the cases of simple tension and combined stress system and the total torsional strain for the test of pure shear. The analysis was also made by using the von Mises effective stress and effective strain rate for multiaxial stress state in order to elucidate the correlation between the relaxations under the complex stress system and that under simple tensile stress. The results of the experiment and the analysis are summarized as follows. The von Mises effective stress relaxation of complex stress system under the condition of constant total axial strain is equivalent to that of the tension test under the same initial stress. Under the condition of constant total shear strain, the effective stress of combined stress state relaxes ranidly [2(1+ν)/3]1/n times more than the tension test, where ν is the Poisson's ratio and n is the exponent of time dependence of the creep strain expressed in a power function. It is impossible in general to maintain the total strains in many directions at the same time under the condition of constant stress ratio, hence it is to be noticed that stress relaxation depends on its relaxation condition of strain.
In recent years, the problems of the dynamic creep have become of increased interest with the use of alloys at high temperature. With a view to predicting the dynamic creep from the static creep data, the analysis based upon the strain hardening theory, was reported in the previous papers by the authors. In the papers, it was found that the analysis could be applied to the materials of relatively stable structure, that is, carbon steels, some ferritic and austenitic steels and commercially pure titanium. In the case of supper alloys of precipitation hardening type, however, considerable discrepancy was observed between the theory and the experiments, and this discrepancy seemed to result from the acceleration of precipitation hardening or the strain retardation due to alternating stress. In the present paper, the dynamic creep tests to elucidate the effects of alternating stress frequency on the strain retardation have been carried out within the frequency range from 20 to 4000cpm. In the experiments, for the higher frequency range from 800 to 4000cpm the alternating stress has been applied to a specimen by the centrifugal force of rotating eccentric mass as reported previously, and for the lower range of 20 and 80cpm the alternating stress has been applied by an eccentric disc and lever mechanism. In both cases, the application of sinusoidally alternating stress has been accurately attained. The tests have been carried out with a low carbon steel at the temperature of 450°C. It is found from the test results that, taking the stress amplitude and the mean stress as the same, the creep curves for the various alternating frequencies coincided approximately with the creep curve under the corresponding equivalent static stress analytically obtained.
Several investigations have been published on the bending and torsion creep strength of alloy steels. The present study was carried out to provide data for torsion and bending creep strength of 304 type stainless steel. Specimens were solution-treated by water quenching after heating at 1050°C for 40min. The torsional creep was measured at several temperatures from 200°C to 600°C of the solid cylindrical specimens of 5mm in diameter, and the bending creep at temperatures from 600°C to 650°C of the plate specimens of 5mm or 7mm in thickness and 10mm in width. The machines used for both the torsion and bending creep testing had been made in our laboratory. The results obtained are summarized as follows: (1) The torsion creep limit for a mean creep rate of 1×10-4%/hr and 5×10-4%/hr during 200 to 300 hours were 6.5 and 10.8kg/mm2 for temperature of 500°C, and 4.5 and 8.9kg/mm2 for temperature of 600°C. (2) on the torsional creep tests under a torsional stress of 11∼16kg/mm2 at 200°C∼400°C, a large initial transient creep was found, but at 200°C and 400°C the creeping did not proceed after about 200 hours, and at 300°C the creeping was continued even after 730 hours. The creep rate at 300 hours in the temperature of 300°C was 1.2×10-4%/hr for 11kg/mm2, 1.1×10-4%/hr for 13kg/mm2, and 1.6×10-5%/hr for 16kg/mm2. (3) The relationship between the mean bending creep rate during 200 to 300 hours υ (near steady creep) and nominal bending stress σ was represented by following equations. υ=1.8×10-8·σ3.7(%/hr) for 600°C υ=4.6×10-9·σ4.9(%/hr) for 650°C From this it was found that the bending creep limit for a creep rate of 1×10-4%/hr and 5×10-4 %/hr was 10.1kg/mm2 and 15.9kg/mm2 at the creep temperature of 600°C, and 7.5kg/mm2 and 10.4kg/mm2 at 650°C.
This investigation was carried out in relation to the effects of heat-treatment and nickel content on the tensile properties, creep and fatigue strengths, at the temperature ranging from room-temperature to 400°C on AC5A (Y alloy), AC8A (Low-Ex.) and AC8B (Alcoa D 132) cast aluminium alloys for piston. The chemical compositions and mechanical properties at room-temperature including the heat-treatments are given in Table I and II, respectively. The specimens were cast into a permanent mould which is in accordance with Japanese Industrial Standard (JIS) H0321. The alloys were compared on the basis of their tensile properties at elevated-temperature after 1000 hour soaking. The creep tests were carried out with the lever-type creep-rupture tester, and the creep strengths were calibrated at 0.1% per 1000 hours. The fatigue tests were conducted on Ono's high-temperature fatigue tester (3000rpm) and fatigue strengths were determined at 107 cycles. The fatigue strength showed lower value at the temperature ranging from room-temperature to 300°C than yield strength. In the case of AC8A and AC8B alloys, however, the creep strength showed the lowest value of the temperature above 200°C. As for the effect of heat-treatment conditions on their strengths, higher values were given by T6 treatment at the temperature up to 200°C, and T5 treatment showed slightly higher strengths than by other treatments at further elevated-temperature. The authors would recommend T5 treatment for the heat-treatment of those alloys for piston. The strengths of nickel-free alloys decreased a little compared with nickel-containing alloys at elevated-temperature. It would appear from the above results that elimination of nickel from the alloys is undesirable. Alcoa F132, however, which has a reduced nickel content of less than 0.50%, is currently in use in U.S.A., and in consideration of the difference in strengths of nickel-free and nickel-containing alloys, it may be acceptable for nickel content of these alloys to approach the lower limit of the current specification.
The studies in metallic and non-metallic materials have made considerable progress in recent years. Still they have some problems concerning their properties which have not been made clear. In particular, it has become increasingly important to clarify the behaviour of the materials when they deform at high strain rates. The high strain rate compression testing apparatus which is manufactured for trial, and the experimental results which are obtained by using the apparatus to clarify the effect of the temperature and the strain rate on the strength of aluminium, are described in this paper. The principle of the experimental method is as follows. The mechanical parts consist of a projectile apparatus, a striking bar, an input bar, an output bar and a throw-off bar. The specimen is mounted between the input bar and the output bar. The impact load is provided with the striking bar which is projected from the projectile apparatus. A pulse of compression caused by impact, travels into the input bar, compresses the specimen, and is transmitted to the output bar and finally to the throw-off bar. The throw-off bar absorbes the surplus energy. The striking bar, the input bar and the output bar are all steel bars of 17mm diameter and 500mm length. The stress and strain to be produced in the specimen can be obtained from the analysis of the stress-time patterns in the input and the output bars. The patterns are drawn on the screen of the two-elements synchroscope by the strain gauges cemented at the midpoints of the input bar and the output bar. The specimens are machined to 17mm diameter and 17mm length from the 99.997% purity polycrystalline drawn aluminium bar. The velocities of the striking bar are 2.5m/sec, 3.3m/sec and 4.2m/sec for this experiment. The testing temperatures are room temperature, 220°C, 290°C and 360°C. The temperature of the specimen is measured by two Chromel-Alumel thermocouples fastened to the specimen's surface. In each impact test, a specimen is impacted repeatedly with the same velocity of the striking bar. The following results are obtained within the range of strain rate of 50∼200/sec. (1) The compressive flow stress of aluminium becomes higher with the increase of the strain rate, and becomes lower with the elevation of the temperature. (2) The static flow stress is more sensitive to temperature compared with the dynamic flow stress. (3) The ratio of the compressive flow stress to cause a certain amount of strain at elevated temperature to the flow stress at room temperature is adopted as a parameter to represent the temperature sensitivity on the compressive strength. The ratio for the static load is far smaller than the ratio for the dynamic load. (4) Comparing the dynamic flow stress with the static one at the same temperature and at the fixed strains, it is concluded that the ratio, σdynamic/σstatic, has the higher value with the rise of temperature. But the value has the tendency to approach 1 rapidly with the increase of strain.
It is very important to clarify the creep strength of the members which have non-uniform stress field, for example, notched members. In connection with this, the creep deformation of cold-worked aluminium specimen of 40mm width subjected to eccentric tension at 200°C is treated in this paper. In this experiment, the creep tests under eccentric loading were carried out under a series of condition of eccentricity, e=2.5, 5.0, 7.5, 10.0, 12.5 and 15.0mm. The load is applied to the specimen in the longitudinal direction, and the loading axis is kept apart from the center axis of the specimen. A technique of photograting method previously proposed by the authors is applied here in order to know the plastic strains produced in the specimen during the creep. By using the values of strain obtained at 60, 180, 300 and 420min. of elapsed time, the corresponding stress distributions are determined with a simple assumption that the stress is constant during each period divided by the times when the strains are measured. The calculations are carried out in both cases, one case is under the assumption of uniaxial stress, and the other is of biaxial stresses. The additional calculation results with the consideration of time-hardening and strain-hardening theory are also obtained. The results obtained are as follows: (1) In the transverse direction to the specimen axis, the strain distribution is linear regardless of the eccentricity and the elapsed time. (2) In the transverse direction, the point, where the strain is zero, is not fixed with the elapse of time. But the movement of the point is considered to be negligible in the case of large eccentricity. (3) The relation between the above-mentioned zero strain point and eccentricity and the amount of eccentricity with transition of time can be explained quantitatively using the experimental fact described in (1). (4) Except the case of e=2.5mm, the stress distribution in the specimen is almost unchanged during the test. (5) Both the stress distributions calculated with the assumption of uniaxial stress field and of biaxial stress field agree with slight variations with one another. It is presumably possible to treat the creep deformation of the specimen subjected to eccentric tension as under uniaxial stress. (6) The tensile load calculated from the stress distributions obtained from the measurement of strain is slightly higher than the externally applied load. The creep deformation of the notched specimen has the contrary tendency. This means that the essential cause of notch strengthening lies not in the stress gradient in the transverse direction, but in other factors, such as the steep stress gradient in the specimen axis or the biaxial stress field.
The sheath and core material which constitutes fuel element through bonding generally functions under different conditions of temperature and atmosphere. When the rigidity of the sheath material is poorer than that of the core material, the sheath is subject to deformation to match the core under service. Hence, it seems very important to clarify such relative deformation considering the thermal fatigue behavior from the view point of the design, safety and usage of the fuel element. In our research “the universal thermal fatigue testing machine” was used which was originally designed by the authors in order to investigate the relative deformation and other various thermal fatigue phenomena in general. The experimental procedures were carried out so as to put Uranium and Magnox A-12 in separated positions. The difference in thermal strain which occurred from the difference in thermal cycling temperature ranges was accurately transmitted to Magnox A-12, and its induced stress behavior, microstructure and failure were observed. The results were as follows. (1) The restricted deformation of Magnox A-12 caused by Uranium. When Magnox A-12 is restricted by Uranium it occasions compressive stress in Magnox at higher temperature range, and the stress in its case is very small in spite of the sudden increase in elongation which is brought about by heating Uranium above its transformation temperature. This fact shows that the deformation of Magnox advances by creep. The increase of the cycling period from 0.5 to 4hr. per cycle promotes tensile creep and stress relaxation at lower temperature range. The maximum stress induced in Magnox under the restricted relative deformation caused by thermal cycling of 190°↔575°C for Uranium and 150°↔450°C for Magnox is higher even after 320 cycles than that of the original state. These results show that Magnox is easily subject to the deformation of Uranium and that the Magnox load on Uranium is very small. (2) Themal fatigue tests on Magnox A-12. The difference in stress and strain values which occur between tension and compression side during the thermal fatigue test, i.e., the non-symmetrical character of stress-strain relationship, increases with the lowering of the thermal cycling temperature, and it is distinguished especially for the thermal fatigue tests of temperature range with the lower temperature limit than 225°C. At higher temperature limit, for instance 525°C, the decrease of the fatigue life is very remarkable and its crack occurs at grain boundary. (3) Observation of microstructure in Magnox A-12 Visible basal slips and twins have occured already after one cycle, and the position of deformation has moved toward grain boundary from grain interior with the increase of both cyclic number and cyclic temperature range. In the case of higher thermal cycling temperature range, such as 275↔475°C, the grain growth and the grain boundary flow especially becomes remarkable, resulting in remarkable wrinkles on the surface. The occurrence of wrinkles breeds had influence upon the heat transfer from the core. During the whole experiments, cavitation which was discovered in ordinary creep tests could not be found. This fact seems to depend upon the difference of loading direction between the creep tests and these experiments. Some peculiarities of the above mentioned behaviors and some problems in Magnox A-12 used for sheath material were discussed also from the crystallographic anisotropic point of view.