Page 147 - MSAM-4-2

Materials Science in Additive Manufacturing Bi-modal powder spreading behavior of ceramics

A B C

Figure 5. Comparison of simulated and experimental average particle sizes (D (50)) across different axis positions for (A) 5 µm powder, (B) 20 µm powder,
and (C) bimodal powder blends. The bottom diagram illustrates the measurement grid used for sampling, showing the specific X-axis and Y-axis positions
within the build platform

size was statistically significant. All the powders showed a simulation treats particles as rigid, dry, perfectly spherical
statistically significant (p<0.001) difference in the powder bodies, which do not fully represent the irregular shape
size along the spreading direction. This confirms the of real ceramic powders, resulting in higher discrepancies
preferential deposition of smaller powder at the beginning between simulation and experiment. Finally, the
of the powder bed in the spreading direction. However, preferential behavior of powder deposition in the direction
there was no statistically significant (p>0.65) difference of spreading, which can affect the packing density of the
in powder sizes perpendicular to the powder spreading powder bed and final part density, was observed across all
direction for 5 µm, 20 µm, and bimodal powders. conditions.

For bimodal powder, the powder sizes in the The spatial variation in particle size observed along the
experiments closely match the simulation. While both the powder spreading direction can be attributed to several
simulation and experiments showed a preferential powder particle-scale mechanisms inherent to granular flow
70
deposition in the spreading direction, the average particle dynamics. One major contributor is granular convection,
size varied by 9.58% between the simulation (8.23 µm) also referred to as the Brazil nut effect, where smaller
and the experiment (7.51 µm). The variation was smaller, particles tend to percolate downward, while larger particles
6.41%, in the direction perpendicular to the spreading migrate to the surface and are displaced toward the end of
between the experiment (8.34 µm) and simulation the spreading direction. This sorting can be exacerbated by
(8.91 µm) average particle size. On the other hand, for the formation and collapse of the powder heap in front of
both unimodal powders, the average powder size in the the counter-rotating roller, where momentum differences
experiments deviated significantly from the simulation: between fine and coarse particles lead to differential settling.
16.32% for 5 µm powder and 12.27% for 20 µm powder. Wang et al. described the inhomogeneity resulting from
36
The deviation between the simulated and experimentally the powder bed using a roller from the powder burst
measured particle sizes for unimodal powders arises phenomenon, which arises from the conflicting motion
primarily from the simplifying assumptions employed due to the rotation of the roller. In additon, van der Waals
in DEM simulations. The simulation represents the forces and moisture-induced cohesion play a significant
PSD using a limited number of discrete particle sizes to role, particularly for finer alumina powders. 71,72 The loss
approximate the unimodal distribution. Furthermore, the on the ignition test confirmed that finer powders exhibit

Volume 4 Issue 2 (2025) 8 doi: 10.36922/MSAM02510016