Short Fatigue Crack Propagation from Surface Pores of an Additive Manufactured Ti-6Al-4VTuesday (26.05.2020) 17:20 - 17:40
Today, it is common to achieve bulk density above 99% by using powder-based metal additive manufacturing (AM) techniques. Yet, process-induced defects remain as one of the most important source of failure under cyclic loading. This work reports fatigue testing of titanium alloy Ti-6Al-4V, focusing on the behavior of surface porosity, which could be labelled as a “semi-natural” defect. It is typically formed during post-processing stage, when process-induced, embedded pores become exposed to the free surface by the necessary material removal work to achieve desired geometry or required surface quality. Surface pores have been shown in literature to be more detrimental than embedded defects, hence the important of this study.
A typical route is followed to obtained surface pores: First, cylindrical rods of Ti-6Al-4V with 14mm diameter and 120mm length manufactured by using Electron Beam Melting (EBM) process. Afterwards, flat dog-bone samples were created by subsequent machining and polishing work to remove machining marks. Surface pores on the gauge section have been counted and quantified by using optical imaging. Silicon rubber compound replicas are utilized during the load-controlled, uniaxial fatigue testing to monitor crack initiation and short crack behavior. Based on the initial analysis, following conclusions are made:
(1) Planar type, crack-like Lack of Fusion (LoF) defects are more detrimental than pores under fatigue loading. In one test, a LoF defect found outside the gauge length led to failure in spite of many pores observed within the gauge section.
(2) Non-propagating cracks initiating from pores were observed, which further corroborates the usage of Kitagawa-Takahashi approach for critical analysis of AM Ti-6Al-4V in presence of defects.
(3) Crack growth follows a tortuous path initially, suggesting microstructural influence during the short crack regime. This effect needs to be considered in predictive modelling either by mechanistic models, e.g. crystal plasticity, or empirical factors as commonly used in engineering designs.