Assessment of Ceramics 3D Printing
The purpose of the following assignment was to test the effect of 3D printing with ceramic resin on part shrinkage, overhang slumping, and lattice designs.
​
1. Introduction
The ceramic resin had been rarely used at our university facilities, and thus, our project focused on testing the printing capabilities and limitations of using ceramic resin on the Formlabs printer. Our experiment was divided into 3 parts to test: 1) shrinkage, 2) effects of overhang slumping, and 3) effects of lattice. Formlabs reported a 15% shrinkage primarily in the z-direction, and so, the first experiment will be verifying this statistic. The second experiment will examine slumping or warping of the overhang features with and without supports. The third experiment will test the printability of lattice designs since it is known that ceramic resin prints self-supporting parts and organic geometries better.
​
2. Iterations
Three different geometries were modeled using OnShape to observe the printability of sharp corners and the effects of various geometries on shrinkage: a cube, a cylinder, and a rounded cube (Figure 1). Two sets were printed, one set with supports and one set without. We followed Kara’s advice that structures with 2-3 mm wall thickness print better. Additional designs were made to experiment the effects of shrinkage on lattices as well. Three different lattices (hexagonal honeycomb, neovius, and Schwarz) were created using nTopology (Figure 2). One set of the lattices was a cube, and another was a cylinder. The type and period of lattice were chosen so that the lattice would be self-supporting.
Figure 1. Geometries for shrinkage testing.
Figure 2. Lattice geometries
To test the effects of overhang, we designed a part to have overhangs of two different angles (45° and 60°). Three copies would be printed: one copy would have supports, one with manually-added supports, and one with no supports (Figure 3). Our first iteration did not have a base that would allow the part to stand in the kiln once the supports were removed. Normally, the supports printed would be removed while being intact. However, it was not possible to do so because the supports were fragile. A second design was made in which a base was added at the bottom of our part. For the manually-added supports, instead of manually adding them on PreForm, we decided to design our own supports in CAD and print them, which would make removing the supports a lot easier.
Figure 3. Geometries for overhang testing.
3. Results
3.1. Shrinkage
The team was able to qualitatively observe that the ceramic parts shrunk in size after they were fired. In order to quantify the shrinkage effect, measurements were made prior to and after the kiln process, and the percentage change was calculated. Table 1 summarizes the average shrinkage percentages, and the results are slightly higher (18.5% in the Z direction) than was Formlabs have reported.
Table 1. Summary of average shrinkage of each part dimension.
3.2. Lattice
All except one lattice structure printed successfully without support structures and were able to be fired successfully as well. One of the lattice geometries had a part chipped off the base in between the printing and kilning process when the parts were considered to be green and very fragile. In terms of print quality, the lattice parts printed and fired very well and came out as expected. On top of observing the lattice print quality, the team also measured the dimensions of the lattice parts before and after the kiln process to observe shrinkage. Different lattice designs had no effect on the shrinkage. Table 2 summarizes the results of the lattice shrinkage.
Table 2. Summary of average shrinkage of each part dimension for the lattice designs.
3.3. Overhang and Slumping
Three overhangs were printed and kilned (Figure 4). All original parts were designed to have two overhang angles of 45° and 60°. The aim was to see if these overhangs would experience more drastic slumping without support structures. For ones without supports, the more extreme overhang of 45° resulted in a larger slumping, 29% change without support, compared to a 7% change in the 60° overhang as shown in Table 3. For ones with manual supports, the more extreme overhang of 45° resulted in a larger slumping, 33% change without support, compared to a 17% change in the 60° overhang. The percentages are larger than those of the previous part which may be due to the nubs hanging off the overhangs that could add some weight to it. Furthermore, the base warped and one edge lifted 3°. As expected, the auto-supported overhang part resulted in minimal percentage change or slumping as shown in Table 3. Visually, you can tell, in Figure 4b, that the part stands taller and is less warped than the other overhang parts.
Figure 4a. Overhang parts without supports (left), with custom supports (center), and with auto supports that did not print prior to the kiln process (right).
Figure 4b. Overhang parts after kilned without supports (left), with auto supports (center), and with custom supports (right),
Table 3. Percent change of the angles for the overhangs.
4. Discussion
4.1. Shrinkage
From this experiment’s results, the shrinking amount was found to differ somewhat from both the Formlabs website, and their user guide document. Thus, an interesting experiment for the future is to look at how different part sizes and how different contact surface area would affect the shrinkage percentage - looking into see if maybe a larger part shrinks more than a smaller part or if there is a consistent shrinkage ratio.
​
4.2. Lattice
The ceramic resin printed only five out of six of our designed lattices, with one failing unexpectedly after a few layers; however the five lattices printed completely and successfully. The reason why the lattice did not print is still undetermined by the team. In future iterations of this particular aspect of the experiment, the specific volume lattice that failed could be further investigated - the experiment can look at different densities of the lattice to see where the failure came from. It would also be interesting to see if a more porous lattice geometry would result in a poorer print quality compared to a dense geometry.
​
4.3. Overhang and Slumping
Unfortunately, the customized support on Preform did not successfully print, so this experiment was unable to compare a custom designed support system and a Preform custom support. In the future, it would be interesting to look into more complex overhang geometries and comparing different support geometries; such as tree support versus pillar supports. Because of the tendency for support removal to damage the actual part, the team should focus on implementing support structures that can be easily separated from the part but will come in contact with the part in a way that will physically support the structure. More work could also be done to investigate the creation of setters, which could be used to support many different geometries in the kiln in useful and creative ways.
​
5. Conclusion
In conclusion, with the project aim to verify Formlabs’ statistic that ceramic parts shrink around 15% in the Z-direction in the kiln, the experiment revealed results of on average 18.5% shrinkage in the Z-direction, which is close to 15% considering there could many different variables between the experiments Formlabs ran compared to this one. Lattice structures and geometries of both TPMS and volume lattice nature were designed to be self-supporting and more organic, printed and fired successfully as expected based on literature of the ceramic material. Slumping was also confirmed to happen without support material, existing more so in extreme overhangs. Furthermore, parts with more surface area in contact with the kiln platform resulted in warping. Future iterations and experiments can be designed to obtain more information on how to best optimize 3D printed parts with ceramic resin. From this experiment, we can conclude that ceramic parts print best as organic shapes, such as the lattice, instead of rigid structured geometries with large overhangs or sharp angles that are prone to cracking and slumping.