عنوان مقاله [English]
Typically, fatigue failure of a component happens at a notch where the stress level rises because of the stress concentration effect. Notch is usually defined as a geometric discontinuity. Notch may be introduced either by design or by the manufacturing process. A hole in a component is an example of a design notch. Material and fabrication defects, such as weld defects, inclusions, casting defects, or machining marks, are notches which are introduced due to the manufacturing process. A variety of methods are available to predict fatigue failure of notched specimens. The purpose of the present investigation is to rank such methods according to their predictive capability. Tensile tests were conducted on CK45 steel specimens and mechanical properties and stress-strain curve were obtained. Rotating bending fatigue tests were performed at room temperature on smooth and notched cylindrical specimens, and S-N curves were obtained. To better investigate the notch-size effect on fatigue life of the cylindrical specimens, two different notch geometries were used. Based on the obtained experimental S-N curve for smooth specimens, fatigue strength reduction factor or fatigue life for notched specimens were predicted by Neuber, Peterson, maximum stress, critical distance (Point method, Line method, Area method) and weakest-link (Area method, Volume method) theories and comparison of experimental results were considered. The predictions by maximum stress and stress gradient methods were conservative, which is some consolation to engineering designers; nevertheless, the errors were high. Neuber, Peterson and critical distance methods are based on stress gradient in notch root radius and predictions by these methods were not accurate. Also, Peterson, Neuber and Stress gradient methods are not consistent with finite element results. The predicted fatigue strength reduction factors by weakest-link methods were the closest to the experimental results. Finally, the weakest-link theory is recommended in terms of predictive capability, availability of the required materials data, and the compatibility with FEA stresses to predict fatigue strength factor among the studied methods in this research.