Date of Award


Publication Type

Doctoral Thesis

Degree Name



Mechanical, Automotive, and Materials Engineering


Additive Manufacturing, Collision-Free Partitioning, Direct Energy Deposition, Supportless Fabrication, Thin-Wall Hemisphere


Jill Urbanic




Although multi-axis bead deposition-based additive manufacturing processes have been investigated in many aspects in the literature, a general process planning approach to address collision detection and prevention still needs to be developed to fabricate complex thin-wall geometries in a supportless fashion. In this research, an algorithm is presented that partitions the surfaces of the part and finds the appropriate tool orientation for each partition to avoid collisions. This algorithm is applied to segment the surface of a thin-wall hemisphere dome and fabricate it without the need of support structures. Two main fabrication strategies are developed: wedge-shaped partitioning, and a rotary toolpath. A five-axis toolpath and a 2+1+1-axis toolpath is introduced to fabricate the partitioned build scenarios. A rotary (1+3-axis) toolpath is also developed. It is concluded that planar slicing brings limitations to reduce the number of partitions that can be modified by a constant step over toolpath. On one hand, the partitioning strategy provides an opportunity to fabricate geometries in a supportless fashion by direct energy deposition additive manufacturing, on the other hand, it introduces physical properties challenges such as surface roughness and hardness variations. Process planning, data collection, and experimental/numerical procedures are implemented to investigate the surface roughness variations (Ra measurement) of fabricated domes. Hence, two solutions are developed using Matlab programming. A mount solution uses the magnified pictures of the exposed surface edges of mount samples as input data. The other solution uses a 3D point cloud of the surface. The innovation of the 3D point cloud solution is the distance factor that is applied in the calculations. The results of this solution are compared to the mount solution. Since the input data of the mount solution is more accurate, the results are more reliable than the 3D point cloud method. The Ra variation diagrams show lower Ra values for the 5-axis sample and the highest values for the rotary sample. Large surface irregularities are noticed at the transition points between partitions, which escalates the roughness values drastically in the region. The sudden alteration of the tool orientation between partitions causes these surface irregularities. Additionally, process planning, data collection, and experimental/numerical analyses are developed to explore hardness variations of the fabricated domes along the slicing direction. The hardness diagram of the 2+1+1-axis sample shows a recognizable pattern for partitions 2-4. The hardness is around 200 (HV) within the partitions but drops to 150 (HV) at the transition points between partitions. Partitions 5-8 show a less recognizable pattern. Although the rotary sample is fabricated in 3 intermittent fabrication sections, it does not show any significant pattern related to the sectioning. The statistical analysis of the hardness shows the highest standard deviation for the 5-axis sample and the least for the rotary one. Finite element analysis of the hardness and residual stress are performed by the ESI Sysweld software for 144 beads of the 2+1+1-axis sample. To reduce the calculation time (a factor of 15 times), a variable mesh size of the beads and substrate are introduced. This means that the element size of the beads grows for the regions farther from the measurement region. The resultant hardness diagram predicts the peak and valley of the experimental diagram for the partitions 1-4, but it misses some patterns for partitions 5-8. Fast Fourier transformation analyses of the surface roughness and experimental/numerical hardness data show a repetitive pattern by the wavelength of the partition length. The preparation time and accuracy of the finite element analysis results reveal that an experimental fabrication and measurement test is preferred at this time, or a new method of numerical analysis is required. This research clearly illustrates the challenges associated with building a complex component and understanding its characteristics. On one hand, splitting the part geometry by different partitioning shapes facilitates the fabrication of the geometries in a supportless fashion. However, this fabrication strategy introduces inconsistency in the mechanical properties. Hardness variations generated by a partitioning strategy needs to be dealt with (possibly by a post-heat treatment). Surface quality at the transient points needs to be investigated more. This foundational research highlights the process planning challenges associated with metal bead based deposition processes, and highlights relevant challenges for similar process families.