The Predictive Modeling of Metal Fatigue LifeUnder Random LoadingThe Predictive Modeling of Metal Fatigue LifeUnder Random LoadingThe Predictive Modeling of Metal Fatigue LifeUnder Random LoadingThe Predictive Modeling of Metal Fatigue Life Under Ra

• Main Authors: Yeng-Ti Huang, 黃彥迪
• Other Authors: Horng-Yith Liou
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/54849617200939505903

碩士 === 清雲科技大學 === 機械工程研究所 === 93 === The present thesis is centered on the study of predictive modeling related to metallic fatigue life under random loading. The contents of the thesis consist of three major parts. The first part is associated with the establishment of a series of mathematical models by the incorporation of random vibration theory as well as Morrow’s nonlinear damage rule into the traditional S-N relationship. With a view to determining the maximum stress needed in applying the mathematical formula, either Gumbel’s asymptotic theory of statistical extremes or Lambert’s empirical assumption is included. The accuracy of computational algorithms is surveyed by the experimental data of a batch of specimens made of 7075-T651 aluminum alloy. The results computed by these derived formula justify their applicability within a certain extent of accuracy. The second part relates the reliability assessment to the fatigue life, where both the parametric and nonparametric models are under study. The merit exists in the fact that this approach allows an understanding of the fatigue scatter feature rather than the “unique” life expectations through the formula derived in the first part. In the third part, the grey theory is utilized for the prediction of fatigue crack lengths under the specified numbers of cycles. Two different routes, including “on-line” as well as “off-line” predictions, are carried out. Experimental crack growth data of Type 4340 steel alloy are used to survey the feasibility of both aforesaid methods. The on-line prediction indeed achieves the excellent effect of length estimation ahead of obtaining the experimental outcome on the next particular number of cycles. On the other hand, the off-line prediction provides a conservative estimation with the aid of few experimental data.