Residual stress and the resulting deformations, as well as process disturbances, continue to be one of the major problems in Laser Powder Bed Fusion (LPBF) processes of metals. Simulations are meanwhile well established to predict their formation and the induced distortions. They can be used to adapt critical part geometries, the support structure or the manufacturing process even before the parts are manufactured.

In particular methods based on the inherent strain approach enable predictions of even macroscopic parts in tolerable computation times, i.e. in only several hours even for parts with industrial relevant size and complexity. Therefore, the method is widely discussed in the literature, e.g. [1,2,3]. Nevertheless, the predictions are not always reliable. This contribution identifies a critical assumption of classical inherent strain based LPBF simulations and finally proposes an extension of inherent strain methods to adress this issue: The temperature history is assumed to be approximately equal. Depending on factors such as the part geometry or the process environment, it is demonstrated that this is not always the case. Different temperature histories significantly influence the formation of residual stress and distortion. Experimental results confirm this hypothesis.

A multi scale (thermal and thermo-mechanical) simulation framework was developed and utilized to investigate the emergence of non-uniform temperature distributions and histories (Figure 1), as well as their impact on the evolving inherent strains. The framework is based on the Finite Elements Method and includes several advanced simulation techniques, e.g. adaptive meshing. The investigation results in an extension of inherent strain methods, by linking it to a reduced order thermal simulation model and choose the applied inherent strain loads with respect to the calculated temperature. This enables predictions even for parts with non-uniform temperature histories and extends its scope of application.