Evaluating the impacts of surface roughness and microstructure on the size effect in two additively manufactured stainless steels

Laser powder bed fusion (L-PBF) enables direct manufacturing of components with complex geometries and thin walls, but many authors report size-dependent mechanical properties that may complicate design. Size effects are commonly attributed to surface roughness, microstructure, and/or internal defects, but the relative importance of each is still not fully understood. To systematically study these effects, L-PBF specimens made of two microstructurally-distinct stainless steels (17-4PH and 316L) were manufactured and mechanically tested, in varied heat-treatment conditions and across a range of thicknesses and build angles. It was found that the size-dependent mechanical properties are efficiently predicted by the ratio of surface roughness to specimen thickness, R a / t , in a way that is relatively microstructure-agnostic. This metric is particularly useful for predicting ultimate strength and elongation, while microstructure moderates its predictive power for yield strength, especially in the more processing-sensitive 17-4PH. When considered in isolation, thickness or surface roughness had weaker correlations with mechanical properties and, importantly, tended to correlate much more strongly with one steel's properties than the other's. Comparing this comprehensive dataset with summary data from other researchers highlights the utility of R a / t , and provides semi-quantitative estimates of the relative impacts of porosity, microstructure, and R a / t on size-dependent properties. • A comprehensive study evaluated the contributors to size-dependent mechanical properties in laser powder bed fusion-based stainless steels. • A simple ratio of macroscale build characteristics efficiently predicted properties, despite drastically different microstructures. • Relative effects of macroscopic and microscopic specimen features were semi-quantitatively evaluated using new and published datasets.