ICF10B

Particle size distribution The authors [23] have proposed a method where a size distribution of particles in materials can be estimated from information on a cutting plane. The method is applied to eutectic Si, Fe compound and porosity in aluminum cast alloy. The size distribution of particles in materials can be assumed as follows. MV0(R1)= ¯NV0 expn−³ R1 λ ´ νo (8) Here, MV0(R1) is the number of particles per unit volume whose radius, R, is larger than R1, ¯NV0 the total number of particles and λ and ν are the characteristic constants of the particle-size distribution. According to the below-mentioned fatigue test, the fracture origins of the present materials were the surface defects. Therefore the characteristics of a particle cut by the specimen surface are examined from the size distribution of particles in the material. The particle cut by the specimen surface is projected perpendicular to the first principal stress and the projected area is corrected on the basis of a fracture mechanics consideration. Then the corrected area is identified with areaP as shown in Fig.4. The number of cut particles on the surface whose √areaP is larger than √areaP 1 ,MS0(√areaP 1 ), can be expressed as MS0(√areaP1)=λ ¯NV0 Z 1 0 t √1 −t2 nΓ³1+ 1 ν , ³ √areaP 1 λ√θ− ´ ν´ +Γ³1+ 1 ν , ³ √areaP 1 λ√θ+ ´ ν´odt (9) θ+ = π 2 +2p1−t2 , θ− =arcsin(t) −tp1−t2 (10) Next, let us consider the probability distribution function of the maximum √areaP value on area A, √areaPmax. The average number of particles whose √areaP is larger than√areaP 1 , is A·MS0(√areaP 1 ). The probability that the √areaPmax is smaller than√areaP 1 , means the probability that there is no particle whose √areaP is larger than √areaP 1 . Using the Poisson’s distribution, the probability is exp[−A· MS0(√areaP 1 )]. Therefore, the probability that the √areaPmax is smaller than √areaPmax1 , FSmax(√areaPmax1 ), is expressed as follows. FSmax(√areaPmax1 )=exp[ −MS(√areaPmax1 )] (11) Prediction of high-cycle fatigue life Although the fatigue life consists of the fatigue crack nucleation life and growth life, it is supposed in this paper that the former life can be neglected and the main crack propagates from the maximum surface defect in the present materials. Moreover, although the scatter of the fatigue life consists of the scatter of the maximum defect size and the scatter of material properties in the crack propagation process, it is supposed in this paper that the former scatter can be considered as the main factor. That is; the fatigue life, Nf, is expressed from Eqn.6 as follows. Nf = E C5 σS³ σS σa´ n lnh d qπ 8√areaPmaxi (12) Here, d is a diameter of a specimen. The scatter of the fatigue life depends on the scatter of √areaPmax in Eqn.11 and it can be evaluated with Eqn.10. Determination of C5 With Eqn.7 for estimating C5 in Eqn.6, the fatigue crack growth rate can be estimated within about ±50% error except for low strength steels. However, if the fatigue life data of the specimen with a small blind hole is measured under a certain stress amplitude, C5 can be determined more accurately with Eqn.11. Moreover, if d`/dN is measured under the stress amplitude, the validity of Eqn.6, where n = 6 is supposed, can be examined. FATIGUE EXPERIMENT A commertial JIS-AC4B alloy was used. The chemical compositions of the alloy is shown in Table 1. Figure 5 and Table 2 show the tensile properties and micro-Vickers hardness of α-Al after T6 treatment. Fatigue tests were carried out on the plain specimen shown in Fig.6. For an examination of Section 2.4, a specimen with a small blind hole (φ0.3×0.15mm) was used for an observation of fatigue crack growth. The specimen was machined, polished withfine emery paper, alumina(3µm) and diamond paste(1µm), and polished chemically. A Ono-type rotating bending fatigue machine, whose capacity was 15Nm and frequency was about 50Hz, was used for fatigue tests. The stress used in the analysis of experimental results is the nominal stress

RkJQdWJsaXNoZXIy MjM0NDE=