PSI - Issue 77

Roman Hofmann et al. / Procedia Structural Integrity 77 (2026) 237–247

245

Roman Hofmann et al. / Structural Integrity Procedia 00 (2026) 000–000

9

4000

Reference Index Reorder

Time Reorder Voronoi

3500

3000

2500

2000

Amplitude (N)

1500

1000

10 3

10 4

10 5

10 6

10 7

Number of Cycles (-)

Fig. 8. Fatigue S-N plot for specialis fatigue specimens. Specimens manufactured with linear scan (reference), index reorder, time reorder and voronoi scan strategy. Round markers: failure, right-facing marker: run-out

layer tapers o ff ), whereas the Voronoi strategy distributes exposure more evenly, suppressing localized thermal build-up and steep temperature gradients within a layer. • Cantilever deflection (representative for residual stress): By reordering the scan of the upper layer above the internal support of the cantilever the deflection can be reduced. Thus the simple index reordering reduced the deflection by 13 %. Time Reorder was inconclusive on this simple geometry, consistent with limited e ff ective reordering, thus adjustments or the combination of both reorder methods are needed for optimization. Beside re ordering segmentation, as applied with the Pilger scan strategy, achieved the largest reduction at approximately 25 %. Voronoi increased deflection when a contour was applied around each vornoi cell; without contouring, deflection was comparable to the reference. With regard to the issue of residual stresses, further optimization is required in relation to the spacing, order and size of the protrusions, since the thermographic images indicate a homogeneous distortion of the component. • Fatigue behavior: The novel strategies substantially reduced scatter relative to the linear scan reference. Index Reorder and Time Reorder nearly overlapped (reflecting similar reordering sequences), while Voronoi trended slightly better. These results indicate improved reproducibility for fatigue-critical applications. The scanning strategies presented are developed for quick optimization and are designed to facilitate seamless inte gration into existing systems, obviating the necessity for defined machine configurations, intricate estimations through simulation, or voluminous data sets for artificial intelligence. Consequently, the full potential of the in-house slicer was not fully realized in this study. Nevertheless, the following hypothesis concerning practicability and transferability can be formulated: • Index and Time Reorder are simple and readily portable to other slicers, machines and already prepared files. • Pilger (back-step–inspired) can be superimposed on any linear hatch by segmenting vectors and reversing seg ment direction, enabling controlled heat input without changing global orientation. • Voronoi o ff ers geometry-aware partitioning and parameter zoning; however, implementation details (e.g., cell contouring) strongly influence outcomes. Implementation is more di ffi cult here and greater intervention in a slicer is necessary. • Strategies are composable: For example Voronoi fields can host time- or index-based reordering, enabling finer control of local thermal histories. Overall, the results show that scan strategy changes a ff ects thermal history, residual stress, and fatigue scatter, sup porting the ambition to move PBF-LB / M closer to ”First-Time-Right” manufacturing, without expensive equipment.