Additively manufactured components produced using the laser powder bed fusion (PBF-LB) process have a characteristic residual stress distribution that depends on the geometry of the manufactured structure. However, it is currently not possible to take residual stresses into account in the structural simulation and optimization of PBF-LB components, as there is no generally valid model for predicting the geometry-dependent residual stresses. Within the project CAEaddFert a virtual test bench is build that aims to predict the strength of additively manufactured components, considering specific material properties and residual stresses.

For this reason, the residual stresses are initially measured using X-ray diffraction analyses on primitive geometries, both on the surface and in depth. To be able to map residual stresses in complex geometries, e.g. in topology optimization, a topological similarity algorithm is developed that recognizes the investigated geometries in the complex component.

Since a complex component can be based on many different primitive geometries and many measurements are required for this, the residual stress state is initially determined in this work using a shot-peened PBF-LB part with a Y geometry manufactured from 316L. The surface curvature has a major influence on the residual stresses on the surface, whereby the residual stresses in the depth showed similar relative curves. Compressive residual stresses are present at the surface, which increase with increasing surface curvature up to a maximum of 750 MPa. From a depth of 50 to 150 µm, there is a transition from compressive to tensile residual stresses. This was used to determine a standard function of the residual stress depth curve on surfaces with little curvature, which is then modified as a function of the surface curvature. Once the residual stress depth curves have been determined by global curvature detection on the surface of the complex component, the residual stresses are assigned to the 3D finite elements inside the component as a function of the surface distance. These element-by-element residual stresses can be considered in the structural simulation in Abaqus by means of an initial stress distribution. To map the directional dependency, the orientation tensor, which points in the direction of the distance vector, is stored for each element.

This simplified consideration of residual stresses is iteratively integrated into the process of density-based topology optimization in TOSCA with the aim of obtaining a component design with reduced weight that complies with permissible stresses.

The work presented on the virtual test bench for components for additive manufacturing aims to integrate additively manufactured components into the conventional CAE-supported virtual product development process and thus increase their lightweight construction potential.