Understanding the formation, effect, and mitigation of voids in laser powder bed fusion additive manufacturing
Open Access
- Author:
- Cummings, Christine
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 31, 2024
- Committee Members:
- Robert Kunz, Professor in Charge/Director of Graduate Studies
Edward Reutzel, Chair & Dissertation Advisor
Jay Keist, Outside Field Member
Allison Beese, Outside Unit Member
Guhaprasanna Manogharan, Major Field Member
David Jeffrey Corbin, Special Member - Keywords:
- additive manufacturing
laser powder bed fusion
voids
Ti-6Al-4V
x-ray computed tomography
defect mitigation - Abstract:
- Laser powder bed fusion (L-PBF) additive manufacturing (AM) enables the production of complex and customized components on demand. The L-PBF process is, however, susceptible to stochastic process anomalies that may form persisting voids. Voids and other flaws may compromise the mechanical properties of a part and initiate premature failure. The layer-by-layer construction of AM parts provides the opportunity to detect the formation of a void and subsequently repair the void region. Considering both stochastic and surrogate (programmatically induced) voids, this dissertation provides novel insights into how voids form and persist in L-PBF, the effect of voids on mechanical and fatigue properties, and how different methods of repair may eliminate flaws in-situ. These key insights help to close the loop for in-situ defect mitigation in L-PBF. Layer-wise images or emission sensors can be used to identify process anomalies that indicate the formation of a void or other build defect. One indicator of build quality investigated in this dissertation is swelling, an overgrowth of printed material that may be characterized from post-recoat layer-wise images. As swelling is relatively easy to identify, swelling may be a computationally inexpensive indicator of build quality or flaw locations. Specific instances of swelling are linked to voids and geometric deviations. In addition, the build plate location, processing parameters, and cooling rate contribute to the prevalence of swelling. To understand how process anomalies contribute to void formation, surrogate voids were programmatically induced by locally increasing or decreasing the laser power when processing in pre-determined volumes. The results indicate that L-PBF is resilient to low power process anomalies (consisting of an 80% reduction in laser power) that span fewer than three layers. During nominal processing, the melt pool penetrates ~120 µm into the processing surface, effectively healing lack-of-fusion voids within two layers of the processing surface. L-PBF is, however, sensitive to high power process anomalies (consisting of a 94% increase in laser power). The location and morphology of persisting voids suggests that the high power hatches cause an unstable keyhole-mode melt pool that penetrates 8+ layers. These findings provide insight into how various process anomalies influence the size, morphology, and persistence of voids. L-PBF components are susceptible to premature fatigue failure triggered by internal and surface connected voids, with larger voids and voids nearer to the surface more likely to initiate failure. By incorporating near-surface surrogate voids into fatigue specimens, we investigated the effect of void size, shape, and surface proximity on the number of cycles to failure. The results highlight the dominance of near-surface voids (within 100 µm of the surface) at initiating fatigue failure, even when significantly larger voids were present in the part (200-500 µm voids, ~250 µm from the surface). These results provide insight into the relative criticality of voids detected in Ti-6Al-4V components. To understand whether voids detected mid-process can be eliminated in-situ, we programmatically induced and attempted to mitigate surrogate lack-of-fusion voids. Our results indicate that lack-of-fusion voids spanning up to four layers below the processing surface can be eliminated, such that persisting voids are not detectable in high-resolution XCT, and larger voids can be significantly reduced in volume by increasing the laser power (to 129% nominal) when processing in the void region on the layer following the top of the void. Re-melting, or re-processing void regions was also effective at healing voids. These results indicate that lack-of fusion voids can be eliminated in-situ and that up to four layers of process monitoring data may be used to identify probable void locations prior to void elimination. To understand the impact of re-melting on the mechanical properties of Ti-6Al-4V, we re-melted a portion of the gauge section in tensile specimens. Re-melting with a laser power higher than nominal (125-150%) and otherwise nominal parameters generated significant amounts of porosity. Re-melting with nominal parameters or with parameters corresponding to 75% normalized enthalpy did not generate significant levels of porosity, however, the re-melt affected the mechanical properties, generally decreasing the ultimate tensile strength and strain at break. This change may be a concern for void mitigation. If sufficient regions of a part are re-melted to repair detected voids, the part performance may be changed. Limiting mitigation to regions identified as probable void locations that surpass a high confidence threshold may reduce the effect of re-melt on properties. For some applications, however, altering properties via re-melting with varied processing parameters could be beneficial, tuning the properties of a portion of a part to suit the part’s application. Understanding the nuances of void formation and mitigation could enable the consistent creation of high quality, high-density L-PBF components.