Ductile cast iron (DCI), due to its good static mechanical properties achievable thanks to the improvements in casting technology, is becoming a viable alternative to steel in many structural components, such as crankshafts, wind turbine parts, pipes, pumps. Moreover, compared to steel, DCI is easier to cast in complex shapes, it is less dense and can meet the mechanical requirements without expensive heat treatments. Many of the applications where DCI is a possible replacement for steel involve cyclic loads. Unfortunately, the fatigue properties of cast iron are generally inferior to those of steel due to the peculiar microstructure of DCI, characterized by the presence of graphite and various types of casting defects such as shrinkage porosity and inclusions and other impurities that act as stress concentrators. In addition, the wide variability in the morphology of these defects translates into a wide scatter in fatigue strength data that negatively affects the mechanical reliability of cast iron parts. Many practical engineering components typically present notches and grooves that locally intensify stresses and that need to be accounted for in the prediction of fatigue life. In other words, when designing mechanical components in DCI, both the effects of the peculiar microstructure and of the geometry should be considered. Unfortunately, till today reliable design criteria for notched DCI members are not available. A promising approach for the prediction of the notch and defect sensitivity of the fatigue limit is the theory of critical distances.
The aim of this work is to investigate the applicability of the theory of critical distances to ferritic and pearlitic ductile cast irons. The critical length parameter is estimated from fully reversed fatigue tests carried out on plain and notched specimens designed to maximize the stress gradient at the notch. The influence of the defects (shrinkage pores) and of the graphite particles on the fatigue limit of the plain specimens is estimated with the Murakami √area model. The critical distance parameter estimated with such procedure is then compared with that obtained using the threshold stress intensity factor range measured from crack growth tests.