High cycle fatigue behavior and life prediction for additively manufactured 17-4 PH stainless steel: Effect of sub-surface porosity and surface roughness

The high potential of additive manufacturing (AM) techniques offers novel opportunities and unexplored design freedom. However, typical internal defects and poor surface quality inherent to AM process not only cause a lower fatigue resistance, but also more scatter in fatigue data; thus, hindering adoption of AM to fatigue critical applications. This study investigates the effect of surface quality and sub-surface porosity on high cycle fatigue behavior of 17-4 precipitation hardening (PH) stainless steel (SS) fabricated using laser beam powder bed fusion (LB-PBF) process. Parts were fabricated in three conditions: net-shape (NS) specimens, oversized specimens, and cylindrical rods. The oversized specimens and cylindrical rods were, respectively, further shallow machined (SM) and deep machined (DM) to the dimensions and geometry of net-shape specimens. The population of defects was investigated via optical microscopy of polished sections, X-ray micro-CT scan analysis, and fractography of fracture surfaces after fatigue tests. The fatigue crack growth (FCG) properties were generated at three stress ratios of R = -1, 0.1, 0.7 to determine the Kitagawa-Takahashi diagram and propagation curve. The polished sections showed the presence of large sub-surface, close-to-surface pores in the NS specimens, while SM and DM conditions had smaller and more uniformly distributed porosity. Critical defects detected on the fracture surfaces were small pores in machined specimens, and relatively large surface irregularities in NS specimens. Machining process, both in SM and DM conditions, enhanced the fatigue performance of the material as compared to that of NS condition. However, in terms of level of machining allowance, no further enhancement in fatigue performance was observed for DM specimens as compared to that of SM ones. Fatigue assessment for both net-shape and machined conditions was obtained performing FCG simulations based on the typical surface features and volumetric defects. Simulation results yielded correct estimates for both net-shape and machined specimens.