Date Published: April 25, 2020
Publisher: Springer International Publishing
Author(s): Prashanth Ravi.
Fill density is a critical parameter affecting the functional performance of 3D printed porous constructs in the biomedical and pharmaceutical domain. Numerous studies have reported the impact of fill density on the mechanical properties, diffusion characteristics and content release rates of constructs. However, due to the way in which slicing toolpath calculations are performed, there is substantial deviation between the measured and slicing fill density for relatively small sized constructs printed at low fill densities (high porosities). The purpose of the current study was to investigate this discrepancy using a combination of mathematical modeling and experimental validation.
The open source slicer Slic3r was used to 3D print 20 mm × 20 mm × 5 mm constructs at three identified slicing fill density values, 9.58%, 20.36% and 32.33% (exact values entered into software), in triplicates. A mathematical model was proposed to accurately predict fill density, and the measured fill density was compared to both the predicted as well as the slicing fill density. The model was further validated at two additional slicing fill densities of 15% and 40%. The total material within the construct was analyzed from the perspective of material extruded within the beads as well as the bead to bead interconnects using the predictive model.
The slicing fill density deviated substantially from measured fill density at low fill densities with absolute errors larger than 26% in certain instances. The proposed model was able to predict fill density to within 5% of the measured fill density in all cases. The average absolute error between predicted vs. measured fill density was 3.5%, whereas that between slicing vs. measured fill density was 13%. The material extruded in the beads varied from 86.5% to 95.9%, whereas that extruded in the interconnects varied from 13.5% to 4.1%.
The proposed model and approach was able to predict fill density to a reasonable degree of accuracy. Findings from the study could prove useful in applications where controlling construct fill density in relatively small sized constructs is important for achieving targeted levels of functional criteria such as mechanical strength, weight loss and content release rate.
Three-dimensional (3D) printing, also known as additive manufacturing or rapid prototyping, enables the fabrication of complex geometries without the need for any part-specific tooling or dies [1, 2]. 3D printing refers to a family of layered manufacturing processes including the process known as Material Extrusion (ME). In ME, a thermoplastic filament is melt-extruded through a nozzle in pre-defined paths to build the object in a layer-by-layer fashion . The software used to generate these pre-defined paths, commonly referred to as toolpaths, is called a slicer. A Standard Tessellation Language (STL) file to be fabricated is typically imported into a slicer, and the toolpaths for printing the model are generated by the slicer based on settings specified by the user. Slic3r , a ME slicer developed in 2011, is used extensively by ME printing users because it allows customization of many characteristics in the printed part [5–10].
The extruder calibration was verified based on the error between predicted vs. measured weight at 100% Slic3r fill density. At 100% fill density Slic3r overlaps adjacent beads by 0.043 mm as seen in the G-code and accurately calculated using Eq. 3. The error between predicted and measured weight at 100% Slic3r fill density was 0.2% on average which demonstrates excellent extruder calibration (Table 1). The measured weight was on average 1.1% larger than the fully dense weight on the other hand. The XYZ dimensions of the 15 samples were all quite close to the CAD design without any significant deviation (Table 2). The weights within each Slic3r fill density percent group were also highly consistent.
Table 1Measured weight, predicted weight and absolute percent error for the 100% fill density samplesSampleSlic3r Fill Density %Fully Dense Weight (g)Measured Weight (g)Predicted Weight (g)Abs. % Error (Predicted vs. Measured)Abs. % Error (Dense vs. Measured)11002.5202.5602.5480.51.121002.5202.5512.5480.11.131002.5202.5502.5480.11.1Table 2Measured weight and dimensional measurements for the samples printed at 5 fill density percent values in triplicatesSampleSlic3r Fill Density %X (mm)Y (mm)Z (mm)Measured Weight (g)19.5820.0819.994.980.32229.5820.0820.004.900.32339.5820.0819.914.860.320120.3620.0820.134.940.603220.3620.1020.134.920.599320.3620.0520.144.970.602132.3320.0220.034.900.881232.3320.0420.044.930.875332.3319.9820.004.900.872115.0019.9819.944.920.404215.0020.0519.974.930.407315.0020.0419.904.920.411140.0019.9319.994.921.057240.0019.9119.964.931.063340.0019.9420.014.921.056
The aim of the current study was to understand the difference between fill density percent set in Slic3r vs. the fill density percent measured in the actual part. The printing process was highly repeatable as seen from the consistency in weights between samples printed at the same fill density percent values in Tables 1 and 2. During extruder calibration, the overlap between adjacent beads generated within Slic3r is necessary to fill all the void spaces between the oblong shaped cross-section of beads and create a true 100% dense construct. With an overlap of 0.043 mm the center-center distance between beads became 0.4–0.043 = 0.357 mm. Slic3r extrudes plastic during this 0.357 mm travel from current bead to the next bead as well and that is why the measured weight is larger than 2.520 g, which is the estimated weight for a fully dense 20 mm × 20 mm × 5 mm model printed at 100% fill density. Because the predicted weight calculation includes the bead-to-bead interconnecting extrusions in the calculation, it can be seen that it is much closer to the measured weight (0.2% average error) than the fully dense weight (1.1% average error). The low error between predicted and measured weight helped verify extruder calibration as well as establish confidence in the predictive model for weight.
At low fill densities the measured fill density was found to deviate substantially from the slicing fill density with average absolute errors greater than 26% in certain instances for a 20 mm × 20 mm × 5 mm cuboidal construct. The predictive fill density percent model in the study estimated fill density to well within 5% of the measured fill density percent values at all five low fill density settings, since the model is based on the actual slicer (Slic3r) toolpath and takes into account the material extruded in the bead-to-bead interconnects. The approach presented in this study could be used to predict construct fill density with reasonable accuracy based on slicing parameters in Slic3r. The methodology could prove useful in biomedical and pharmaceutical applications requiring accurate control of fill density for relatively small sized constructs to meet targeted functional criteria such as compressive strength and content release rate.