Effect of infill density and pattern on the specific load capacity of FDM 3D-printed PLA multi-layer sandwich

2.1 Procedure of research

The specimens were manufactured by 3D printing employing various process settings (four levels of infill ratio and filling pattern). At first, the fabricated samples were subjected to tensile testing to determine the effect of pattern and filling density on tensile strength. Secondly, Young’s modulus values have been examined in the elastic range using the previous fill and pattern variations at maximum load without breaking the specimen. Further, additional measurements were needed, such as scaling the weight of the inspected specimens using a digital lab balance. From the obtained results, material properties were determined for finite element analysis. Therefore, the outcomes of these tests were used as the input parameters for the virtual bending tests, which were modeled utilizing Ansys Workbench software (finite element method). The specimens of modeling were in the shape of a sandwich. Thus, in addition to investigating the same 3D printing parameters concerning filling ratio and pattern, the influence of layer thickness of the sandwich inner and outer layers was assessed. The aim of this is to achieve an optimal specific load capacity by optimizing material consumption. This would help to improve the product properties and simultaneously to reduce the production time and cost. More details about the 3D printing settings used as well as the testing and modeling parameters are mentioned in the following sections.

2.2 3D printing of specimens and process settings

According to the fused deposition modeling technology, the printed specimens were made by a 3D printer (CRAFTBOT3). The process was executed at a temperature of 60 °C and 195 °C for the workbench and printing head, respectively. The software employed for slicing the STL file (3D printing model) is Slic3r v.1.3.0. The printing material is a PLA filament (branded Filaticum) with yellow color having a diameter of 1.75 mm. The manufacturing of specimens was accomplished at a print speed of 60 mm/s using a nozzle diameter of 0.4 mm.

The tensile test specimens were fabricated following the ISO 527-2:2012 standard [24], with the dimensions demonstrated in Figure 1a. Two process settings which are filling patterns and infill percentage were employed at four levels. The four patterns used are Rectilinear, Hilbert Curve, Concentric, Honeycomb (see Figure 1b). The specimens of each pattern were manufactured in four different filling ratios, which are 40%, 60%, 80%, and 100%. The Honeycomb pattern was unattainable with 100% filling because this setting is not applicable in the slicing software. Therefore, the number of conditions utilized to produce all specimens was 15, as presented in Figure 1c, with three identical samples for each condition to yield reliable results (45 models in total).

Figure 1:

(a) ISO 527 standard specimen for tensile tests, (b) patterns used when fabricating the specimens, (c) actual appearance for 3D printed tensile samples under various pattern and infill density, (d) extensometer used to determine the modulus of elasticity.

2.3 Tensile test and elastic modulus determination

The specimens were examined with a universal tensile test machine Zwick/Roell Z100, according to the standard and test parameters of ISO 527 [25]. Mechanical experiments were carried out in Knowledge Transfer Centre at Szent István Campus, MATE University, Hungary.

The modulus of elasticity was determined experimentally within the elastic range using an extensometer (see Figure 1d). The extensometer is an instrument attached to the tensile test specimen for measuring the elongation, which implicitly helps determine Young’s modulus accurately. Since the sample is not fractured during this test, the measurements can be repeated frequently employing only one specimen of every examined condition. The measuring procedure requires meticulous adjustment, as an incorrectly adjusted tensile force can cause fiber breakage. This issue can lead to measurement inaccuracies that are not visible, hearable, or noticeable at all to the eye or ear. The chosen limit of measurement is up to 80% of the yield strength (the maximum elastic force). Considering this value carefully, the measurement of the modulus of elasticity was repeated 10 times for each condition (15 conditions). These 150 measurements were conducted to determine reliable values for Young’s modulus.

Accurate and realistic material properties are required to achieve precise and reliable data for modeling. Hence, the mass of every specimen was measured accurately using a digital balance branded Sartorius. It was needed for the final optimum calculations of every piece.

2.4 Modeling of bending tests

2.4.1 Specimen structure and varying parameters

A virtual sandwich (multi-layered) specimen ISO-178 Standard compliant was created by Ansys Workbench software. A part of the structure cross-section of this sample is illustrated in Figure 2a. The total thickness of the specimen is constant as 4 mm according to the standard. The outer (top and bottom) layers have the same filling (100%) and pattern (Rectilinear) for all specimens. The height of the outer layers is changing and designated based on the thickness of the inner layer (core). In other words, the thickness of the outer layers (dark gray parts) depending on the internal layer height (v). When v increases, the thickness of the external (top and bottom) layers decreases, and vice versa. Starting at 0.2 mm, the core thickness has been increased 15 times up to 3 mm with an increment of 0.2 mm. This is to examine the influence of varying the inner layer thickness on the product properties at the current model. Consequently, when the inner layer is at the minimum thickness value set (0.2 mm), the outer layers’ thickness is 3.8 mm (1.9 mm for top and 1.9 mm for bottom). Whereas, at maximum inner layer thickness (3 mm), the outer layers become 1 mm (0.5 mm on each), and so on for the remaining thickness values in between.

Figure 2:

(a) Illustration of the sandwich specimen’s structure, (b) test method of ISO 178 bending standard, (c) the mesh of finite element model.

Moreover, the pattern and filling density of the core were also changed, imitating the parameters examined in the experimental (tensile) tests. Therefore, the applied variations were Concentric, Hilbert Curve, Honeycomb, and Rectilinear in terms of patterns, and 40%, 60%, 80%, and 100% regarding the infill ratios. As a result, the previous calculations (such as modulus of elasticity and mass) can be applied during the modeling tests. Throughout the modeling, the test procedure was implemented according to ISO 178 bending standard [26] for determining the specific load capacity, as displayed in Figure 2b.

3.1 Tensile strength and elastic modulus evaluation

The results of tensile tests for all examined patterns and filling ratios are shown in Figure 3. It can be seen that the values of force versus elongation are pretty affected by different 3D printing settings. Generally, increasing the infill percentage has increased the strength of the test pieces irrespective of the filling pattern. Fernandez-Vicente et al. confirmed a corresponding trend [27]. In terms of the influence of various patterns, the Concentric pattern at 40 and 60% infill percentage proved to be much stronger than the other samples’ styles. However, Honeycomb performed excellently at 80% filling, while Rectilinear specimens revealed the best at 100% density. On the other hand, the Hilbert curve pattern disclosed the weakest performance in all cases. Its samples at 40, 60, and 80% filling ratios were almost capable of withstanding half of the tensile force that the Concentric specimens could afford. This might be attributed to the fibers’ deposition direction of these patterns since most of the fibers in the Concentric were longitudinally oriented toward the direction where the specimens are being pulled. More deposited filaments parallel to the tensile test force would result in stronger tensile strength [28]. On the contrary, the Hilbert curve pattern contains discontinuous and short beads [29]. Even in other filling patterns, the placement of fibers was not longitudinal (side by side) with the tension direction. Thus, due to the deviation of fibers direction, torque may develop among them within a particular cross-section, and a higher tension arises at the connection of the fibers. This would be followed by a larger elongation with a smaller force and then a fracture.

Figure 3:

Tensile test curves (force vs. elongation) for all examined patterns and filling ratios.

Regarding the elongation, Concentric and Rectilinear patterns independently have almost offered identical values for all examined densities. It can be seen that the Hilbert curve filling has constantly demonstrated inferior elongation as compared to other patterns. Further, the Honeycomb specimens were elongated nearly as much as the Rectilinear and Concentric pattern at the infill ratio of 60 and 80%. However, it showed a brittle behavior (much lower elongation) at the 40% infill percentage. This indicates that both infill pattern and density are considerably essential for the elongation properties of the material. Nevertheless, the results showed that the infill ratio parameter has more impact on the elongation capacity.

The engineering tensile strength values were calculated and presented versus the filling density in Figure 4. Each point is the average of three basic measurements. The tensile strength is increasing as the filling ratio increases. A similar tendency was reported in a recently published study [30]. The Concentric pattern generally reported the highest tensile strength values (Max. 55.5 MPa at 100% filling density), which were an overall average of 34.31%, 9.97%, and 12.32% higher than the Hilbert curve, Honeycomb, and Rectilinear, respectively. The performance has significantly improved for the Honeycomb and Rectilinear specimens at 80% and 100%, consecutively. The overall attitude is in good agreement with the literature [29, 31].

Figure 4:

Engineering tensile strength results against the filling density.

The manufacturer specified the material’s tensile strength as 52 MPa for an injection molded solid body. Since the cross-sectional area of the specimen (at the gauge section) is 40 mm² (10 mm × 4 mm). Thus, the force required to rupture per square millimeter is 52 N. Based on that, the maximum load that the material can withstand was defined when testing the elastic modulus. It is noteworthy that the perimeter cross-section area is not identical with the inner filling cross-section area, not even for 100% infill, due to the anisotropy of 3D printing. The modulus of elasticity was examined within the elastic range (see Figure 5a), where the stress does not cause permanent deformation in the material.

Figure 5:

(a) Elastic zone during the modulus of elasticity test, (b) screenshot taken from the testing machine software for Young’s modulus (E _t column) measurements of concentric pattern specimen at 40% density.

Using an extensometer, the modulus of elasticity was measured for all pattern and filling density specimens examined. For achieving credible results, 10 measurements were conducted (repeated) on each condition. As an example, Young’s modulus measurement results of the Concentric pattern specimen at 40% density are exhibited in Figure 5b. This figure is a screenshot of the data obtained and was taken from the machine software used. Row no. 1 shows a measurement error due to improper installation or mounting of the extensometer. The other 10 measurements show the validity of the results.

3.2 Virtual bending tests investigation

The values of modulus of elasticity measured were fed into the boundary conditions of Ansys Workbench. The modeled bending test specimen (ISO 178) has the same conditions (patterns and fillings) applied in the ISO 527 tensile samples. The force that the structure can be loaded was not known in advance. Therefore, a load test (named F force first) was used to determine the maximum stress. The value of the initial force obtained from the load test is 50 N. Since the system is linear, it is possible to determine the allowable stress and then the maximum force that the specimen can withstand. Hence, for determining the maximum force:

where F _first is the first force (50 N, obtained from the load test aforesaid), σ _first is the first tensile strength in (MPa), F _max represents the maximum force in (N), and σ _max is the maximum tensile strength in (MPa). From this:

The stress values obtained and 3D printing settings (patterns and fillings) were used in Ansys (see Figure 6). In addition to the pattern and filling density parameters, the layer thickness variation of the sandwich specimen’s inner layer (core) was also examined according to the manner explained in Section 2.4.1. The maximum bending force was calculated for each layer order based on the finite element results and the measured tensile strengths.

Figure 6:

The stress obtained from the ISO-178 piece.

As the examined settings involve four levels of patterns, four levels of infill densities, and 15 levels of layer thicknesses, the total number of the resultant values would be 225 (Bearing in mind that Honeycomb at 100% density is impossible to print as aforementioned, thus the feasible measurements are 240 − 15 = 225). The acquired values were included in an excel table. Depending on the boundary conditions and the filling, the values of the maximum stress named “Max.force” are presented after the substitution into the formula with an initial load of 50 N. The external (100% Rectilinear, constant) and internal (variable) layer density and pattern were implemented during the calculations.

As an instance of the results, a screenshot for the data obtained (measured by Ansys) after testing a specimen of one pattern at one filling ratio (40% Concentric) and under 15 different levels of layer thicknesses is displayed in Figure 7 (upper part). It can be seen that as the layer thickness of the sandwich’s inner layer (column B) increases, the amount of the maximum force generated decreases continuously. This applies to the maximum force of the specimen’s outer and inner layers (columns C and D, respectively). A maximum force of 46.16 N could be measured at the outer layer when the core height is 0.2 mm. At the same time, the inner filling layer can withstand a maximum force of 313.7 N. This significant variance, in this particular case, was mainly due to the huge difference in total thickness and infill density between the outer layers and the core. As the total thickness was 3.8 mm for the cumulative outer layers, whereas only 0.2 mm for the core. In addition, the infill density of the outer layers was 100%, while 40% for the core. These obvious differences led to the maximum force measured for the outer layers being remarkably higher than the maximum force that the core can bear. On the other hand, it was observed that at 3 mm inner layer thickness, there was almost no difference between the maximum force measured in the outer and inner layers. The ratio of maximum load capacity to weight (force/mass in N/g) was also presented (column G). Increasing the core thickness caused decreasing in the mass due to the density of the inner layer (40% for the shown data). Further, it was expected that less material (decreasing the filling density) leads to the reduction of the load capacity. However, it was found that the “force/mass” increased with the decrease of weight, as it reached a maximum value (5.47) at 2.6 mm core thickness. This indicates that decreasing specimen weight up to a certain extent resulting in an increase in specific load capacity. Therefore, the point where the highest force/mass value was found refers to the optimum inner layer thickness. Column H represents the percentage of the core thickness (between 0.2 mm “5%” and 3 mm “75%”) out of the specimen’s total height (4 mm “100%”). Since the highest force/mass value was found at 2.6 mm core thickness, this means that the optimal point was achieved when the inner layer thickness was 65% of the total sample height. The curve of developing the specific load capacity (force/mass ratio) versus the inner layer thickness was drawn in Figure 7 (lower part).

Figure 7:

Evaluation of data obtained for determining the optimum specific load capacity (the specimen is 40% concentric).

The procedure carried out in Figure 7 was repeated on all settings examined (filling density and pattern). The curves of developing the specific load capacity versus the core thickness are presented in Figure 8a–d. Obviously, it can be seen that the best specific load capacity was always achieved at the infill density of 40%, irrespective of the filling pattern. After the regression, a third-degree polynomial trend line could be observed on all 40% curves since R² is close or equal to 1. This means that the polynomial model fitted the data with a high coefficient of determination (R²).

Figure 8:

The curves of developing the specific load capacity in front of the inner layer thickness under various filling densities at (a) concentric pattern, (b) Hilbert curve pattern, (c) honeycomb pattern, and (d) rectilinear pattern.

In all exhibited patterns, the filling density of 100% has no optimum point (except for Honeycomb since it is already not existed as aforementioned) because the relationship was almost flat (without increment). The main reason for this is that the outer sandwich’s layers also have the same filling ratio (100%). Therefore, the material properties did not improve despite the increasing thickness. That was more evident in the Rectilinear sample where the line was relatively straight due to the pattern of both outer and inner layer, as well as the filling ratio, were the same.

The highest load-bearing capacity values were reported at the inner layer thickness of 2.6 mm (65% of total thickness), 1.8 mm (45% of total thickness), 1.6 mm (40% of total thickness), and 2 mm (50% of total thickness) for Concentric, Hilbert Curve, Honeycomb, and Rectilinear patterns, respectively, all under 40% filling ratio. Further, it was also noticed that after the highest point, there was a considerable decline in specific load capacity, especially at the filling density of 40 and 60% for most patterns. This intimates the significance of searching the optimal point, which seems not associated with a specific thickness for all conditions.

In summary, it is clear from the diagrams, shown in Figure 8, that there is an optimum for each filling density and pattern. This study attempted to provide the best values (optimal), which were represented by the maximum specific load capacity paired with the inner layer thickness (employing the procedure described in Figure 7). The core thickness in which the optimal specific load capacity values were found for each infill density and pattern are tabulated in Table 1.