posted on 2014-12-15, 10:37authored byP. M. Hughes
This work focusses on the problem of predicting the fatigue strength of engineering components where the geometry and stress distribution differ from conventional notched fatigue test specimens. The traditional approach based on the use of a net section nominal stress is therefore difficult to apply. The use of estimated local surface stresses and strains can lead to misleading and often highly erroneous predictions of the fatigue strength at high (107) cycles, particularly for components manufactured in cast iron, since this type of material exhibits a relatively high notch insensitivity in fatigue tests.;The work is of a theoretical nature, but makes use of the availability of a significant quantity of recent experimental data on cast iron test specimens of various grades, geometries and loading conditions. An approach based on the calculation of the condition for crack arrest has been investigated as a means of predicting fatigue strength. The finite element method has been used to help calculate the range of stress intensity factor (K) values for cracks in notched fatigue specimens of various geometries. The K value at a number of different crack sizes is compared with the apparent short crack threshold value. If the K value for a growing crack becomes less than the threshold value (Kth) then the crack is assumed to arrest.;Comparisons between the predicted strengths and the estimated values from the test data show that a crack growth assessment procedure based on a continuous surface crack geometry at the notch root gives a reasonable approximation to the data and can account for the effects of geometry, stress distribution and mean stress. The inclusion of a short crack model makes the procedure particularly relevant to assessments of cast iron components, however, limited comparisons with notched data for other materials were also favourable.