1 Introduction

Additive manufacturing (AM) better known as 3D printing became a widely used solution to create parts rapidly with certain geometries, which can only be made with difficulty using traditional methods [1]. Fundamentally, AM is a set of technologies that create parts by local addition of material, as opposed to subtractive processes, where the existing material is truncated to obtain the desired geometry. As a result, the geometry, that can be formed, is not limited by the machining tools; for example, in the case of drilling or milling, the part is created by the layer-upon-layer building process [2]. Over the years, several technologies have been invented which can be considered AM technology [3]. The connection between them is that the material used can solidify or fuse locally and thus create the desired structure layer by layer. These techniques can be classified into subgroups of AM, based on what kind of materials they use or how they form the layers, but these can vary based on different people’s perspectives. One thing is for sure, fused deposition modeling (FDM) is the most popular technology among all of them [4, 5]. It belongs under the material extrusion subcategory and uses thermoplastic raw material. Due to its simplicity, it is a relatively inexpensive process, and the production and procurement of the raw material are also cheap, which is why it is already excellent for home use or company applications as well. In the FDM process, one of the big advantages is that hollow bodies can also be produced, which meet the same requirements as the solid version but can save a significant amount of material. For such hollowed parts, infill can be set which can help ensure the rigidity of the structure [6,7,8]. The commercial slicing software that creates the necessary g-code for printing has some built-in default infill pattern and the percentages of the infill can also be set.

Numerous research articles covered the fracture mechanism of FDM-printed parts [9, 10]. Also, the effect of infill patterns and percentages on the mechanical properties of the printed parts has been investigated in many previous studies [11, 12]. Tanveer et al. [13], in their review, concluded that, in general, in the case of hollow parts, the air gaps, inside the structure, act as crack propagators, and therefore, the mechanical properties cannot be equal to solid samples. Fernandez-Vicente et al. [14] investigated the tensile properties of specimens printed with three different infill patterns and ratios. They found that as the filling increases, the yield strength and tensile strength both increase, however, nonlinearly. The change from 50% infill to 100% infill means a bigger improvement in mechanical resistance, like 20–50%. After creating the same study, Farazin and Mohammadimehr [15] extended this conclusion by mentioning that at a high infill percentage, the parts became brittle and had lower strain fractures. Tanveer et al. [16] performed a similar comparison and found the same conclusion; however, they tested specimens with a combination of different infill ratios. The resulting mixed cross sections were able to provide better mechanical performance and keep the infill density low. Abbas et al. [17] carried out some compression tests to compare the infill ratios. They found that the transition of the compression strength has a linear relationship. Birosz et al. [18] extended this by testing the specimens with applied loads parallel and perpendicular to the layering direction. They concluded that regardless of the infill pattern the load direction does not influence the compression strength. Apart from checking the mechanical properties of the general infill patterns, several research works focused on creation of topology optimization based [19], bio-inspired [20], or self-supporting infill structures [21] as well. Even though FDM is a rather simple technology, the properties of the manufactured parts are influenced by many other factors, such as layer thickness, raster angle, nozzle diameter, post-processing, annealing, and even the color of the raw material. Therefore, a general comparison could be challenging, and it must be aimed that all variables not examined from the point of view of the experiment should be fixed.

Although the mechanical properties of 3D-printed parts with different infills have already been examined based on many aspects, one important factor is typically overlooked. The infill was set in the Slicer software, and in it usually, the construction of it is made by one continuous extruded line. Consequently, the infill percentage refers to the density of these lines. In this paper, a scaled infill thickness technique is presented and compared based on tensile tests. With a proper scaling technique, the print time and complexity of the part can be reduced, while maintaining the same mechanical resistance.

2 Materials and methods

For the experiments, Prusament PLA Gray (Prusa Research, Prague, Czech Republic) was used. Table 1 contains the general properties of the PLA filament. This type of thermoplastic filament is one of the most frequently used raw materials for FDM printing due to its favoring printing behavior (better build plate adhesion and smaller thermal expansion).

Table 1 PLA material properties [22]
Full size table

To print the samples, Prusa i3 mk3s 3D printer was used. The following parameters have been set for production: 0.4 mm nozzle diameter, 0.15 mm layer thickness, 215 °C print temperature, 60 °C build plate temperature, 100% infill ratio (the creation of different infills explained in the subsection below), and four contour lines. To create the necessary g-code, Prusaslicer 2.5 has been used. Each specimen was laid on its largest surface on the build tray, this way, during the measurement, the load on the layers acts parallel, and thus, the effect of anisotropy can be eliminated.

The tensile test was performed on a Zwick Roell Z100 Tensile Test Machine, based on the ISO 527–1:2012 standard (with tensile speed: 10 mm/min). The results were evaluated using testXpert software. The standard tensile test samples were based on ISO 20753:2019 standard, Type-A dog bone shape with an extended cross section. The extended cross section was necessary, so the effect of the infill pattern can be better investigated (Fig. 1).

Fig. 1
figure 1

Size of the test specimen

Full size image

2.1 Methodology to create uniform infill

As it was mentioned above, dog bone samples have been prepared for the test. However, the structure of the printed pieces is more complex. The 3D printer draws the contour of each layer first. These contours are made out of one, two, or more connected lines in offset. These contours provide the shell of the geometry, and basically, it determines the surface of the part. For the study, the number of contour lines was set to two. It is a default setup in most of the slicer software since it ensures that the infill will not have any effect on the outside surface of the part. Because, with only one contour line it could be possible that due to the overlapping the infill toolpath would deform the already deposited contour line, resulting in the dimensional inaccuracy of the part. Once the contours have been created, the printer fills it in according to the set infill pattern and ratio. Usually, the bottom and first layers are also part of the shell, therefore, to produce them the slicer automatically sets the infill to 100%. When the infill ratio is set lower than 100%, the program creates the infill from one continuous line based on the selected pattern. By the change in the ratio, only the size of the air gaps is modified, but the structure is still created by one continuous line. For example, as can be seen in Fig. 2, the part was sliced by setting 2 contour lines and a Grid infill pattern with a 50% fill ratio.

Fig. 2
figure 2

Two contour lines and infill made with one line without a stop

Full size image

Since the load is taken up by only one line, regardless of the ratio and pattern of the internal infill, its effectiveness is therefore questionable. However, this 50% infill can be achieved by building up the inner structure using more adjacent lines. In that way, the wall thickness of the pattern would be bigger and the sizing of the pattern (in the upper-mentioned case the Grid lap distance) would be bigger, proportionally. The two most frequently used patterns are the Grid and Honeycomb, and hence, in this study, the effect of the scaling on these was investigated. When using a Grid, the infill is created by nets forming an angle of 90 degrees to each other, while Honeycomb is a nature-based hexagonal formation, which, in theory, has excellent load-bearing capability. The patterns can be seen in Fig. 3. Scaling can be made by adjusting the thickness (V) and length (L) of a unit strut that builds up the pattern. Since the printer is equipped with a nozzle with a diameter of 0.4 mm, each wall (strut) can be 0.4 mm or its multiplied value thick. Therefore, for the experiment, the investigated wall thicknesses were 1.2 mm, 1.6 mm, 2.0 mm, and 2.4 mm. Both the Grid and Honeycomb patterns are two-dimensional, which means the toolpath to create the infill is the same on each layer, and the lines are deposited on top of each other. Therefore, the infill ratio can be determined by calculating the area of a pattern on a layer and comparing it with a solid layer. For comparison, a 75% infill ratio was chosen, and since it is relatively difficult to calculate the density value from this, a simplified calculation method has been used. The infill structure is built up from the multitude of struts marked with red color, and hence, the size of the strut is proportional to the size of the whole pattern. For calculation, the background area can be used (area of blue and red together).

Fig. 3
figure 3

Infill structures a Grid pattern, b Honeycomb pattern, c Grid unit cell, and d Honeycomb unit cell

Full size image

The following equations can be used to calculate the necessary lengths for the four selected thicknesses and three infill density values (1, 2):

$${\mathbf{G}\mathbf{r}\mathbf{i}\mathbf{d}:\mathrm{A}}_{\mathrm{whole area}} \cdot \mathrm{Infill Ratio}=\mathrm{L} \cdot \mathrm{V}-\frac{1}{2}{\mathrm{v}}^{2}$$
(1)
$$\mathbf{H}\mathbf{o}\mathbf{n}\mathbf{e}\mathbf{y}\mathbf{c}\mathbf{o}\mathbf{m}\mathbf{b}:{\mathrm{A}}_{\mathrm{whole area}} \cdot \mathrm{Infill Ratio}=\mathrm{L} \cdot \mathrm{V}-\frac{1}{2}{\mathrm{v}}^{2}\mathrm{tan}30^\circ$$
(2)

In the above equations, \({\mathrm{A}}_{\mathrm{whole area}}\) refers to the area of one whole unit cell (blue + red), L is the length of the unit strut (red), V is the thickness of the unit strut, and \(\mathrm{Infill Ratio}\) is an arbitrary number (0–100). By solving the equations, the following length values have been obtained (Table 2).

Table 2 Possible thickness and length values of the unit struts
Full size table

2.2 Investigated area

To eliminate some adverse effects, these patterns have been manually added to the geometry of the tensile test specimens. These dog bone specimens have a 4 × 30 mm cross-section region of 50 mm length, and it is expected that the fracture and the corresponding elongation would happen in this area. Therefore, the scaled pattern was only applied to this region, and the rest of the sample was preserved as the original. To achieve this, a 50 × 28.4 mm region in the middle of the specimen was replaced by the created pattern. The value of 8.4 mm came from the concept that an aforementioned shell must be created, and the default setup is that the shell is created by two contour lines, and hence, on the two sides, 0.8 mm was left untouched. The other printing parameters are the ones mentioned above, including the 100% infill, which ensures a solid structure at the two ends of the specimens so that the proper grip can be achieved during the tensile test.

The figure above shows that the density of the formed patterns is sufficient for the tests, and the asymmetry inside the experimented area does not have a significant effect on the results (Fig. 4).

Fig. 4
figure 4

Modified test specimen with the added experimented area

Full size image

3 Results and discussion

Figure 5 contains the results of the tensile tests, Table 3 contains the maximum loading forces collected, and Fig. 6 shows the broken test pieces.

Fig. 5
figure 5

Results of the tensile tests a Grid, and b Honeycomb

Full size image
Table 3 Maximum measured force
Full size table
Fig. 6
figure 6

Broken test specimens

Full size image

3.1 Grid pattern’s results

By checking the results of the Grid pattern, it can be seen that Young’s modulus is seemingly the same for each tested strut thickness. This indicates that the robustness of the pattern does not affect the flexural modulus of the parts. It could be a surprising result since the initial assumption was that the pattern constructed by thinner “lines” would more likely bend under load. It is important to mention here that this kind of infill structure had no parallel trusses to the applied load direction, other than the contour lines. The lack of change in flexural modulus could be explained by the fact that if the strut thickness is smaller, the pattern (L) is also smaller. This could counterbalance the bending effect of the infill. Furthermore, it can be stated that the maximum resistible load is more or less the same for each pattern construction. The most resistant was the one with a 2.4 mm strut thickness (V), the least resistant was the one with a 1.6 mm strut thickness (V), the difference between them was less than 14%.

3.2 Honeycomb pattern’s results

The results of the Honeycomb pattern variants show a highly different load resistance capability. Seemingly, this structure is more sensitive to scaling, and it is difficult to state a correlation between the strut thickness (V) and length (L) and the mechanical resistance. The least resistant was the structure with the second highest V value, and its flexural modulus was also outstandingly small compared to the others.

3.3 Peculiarities of the manufacturing process

By examining the broken pieces, some very important conclusions can be drawn. Firstly, in general, when evaluating the results of the tensile test, the broken surfaces have been investigated. Fundamentally, there are two main breaking scenarios, as can be seen in Fig. 7. The first one is when the first crack starts due to the pattern (discontinuity of the material), and the breaking happening along the pattern in a horizontal direction breaking the intermediate struts. The other one is when the failure happens longitudinally in a strut.

Fig. 7
figure 7

Break types a pattern induced and b along the struts

Full size image

As it was mentioned in the Introduction section, in the earlier research works, during FDM 3D printing interlayer adhesion plays a crucial role. Since the layers are formed from the extruded lines, their adhesive connectivity affects the strength of the parts. As per the g-code generation algorithm, the first boundary lines are drawn by the print head, and then, it fills the regions between these boundary lines according to the infill setting in each layer. In this case, the infill ratio in the slicer software was 100%, and thus, the filling was created by deposited lines next to each other following a zigzag path. To create one layer, two contour lines (perimeter) were used. The different structures of the printed pattern variants extracted from the slicer software can be seen in Fig. 8.

Fig. 8
figure 8

Generated tool path for the layer formation a Grid V12, b Grid V16, c Grid V20, and d Grid V24, e Honeycomb V12, f Honeycomb V16, g Honeycomb V20, h Honeycomb V24

Full size image

It can be seen that in the case of the pattern variants which have lower strut thickness than 2.4 mm, there is no infill created to the adjacent shapes (cubic or Honeycomb). Therefore, they are connected by their external contours, which were proven that have the least adhesion. On the other hand, in the case of the variants with a 2.4 mm V value, a connective infill is created between the unit shapes. By comparing the measurement results and the broken pieces, it can be determined that the first braking type (pattern induced) occurred in the samples where this type of solid infill prevails between the cells. In all other cases, the fracture occurs along the struts, due to the poor adhesion of the contour lines. And hence, these two cases must be separated, since the maximum mechanical resistance in the V = 2.4 mm pieces is due to the load capacity of the raw material, and in the V < 2.4 mm, it can be traced back to the adhesion properties. Another remarkable peculiarity is that both the V16 and V20 pattern types are made out of 4 contour lines and there is no infill between the adjacent unit cells. The slicer’s toolpath generation algorithm detects that the gap between the cells is not enough to create a zigzag solid infill. Therefore, for the V20 specimens its deposits, the material with less adjacent overlapping and the extrusion rate is higher. This would result in less adhesion which explains why the V20 specimens gave less load resistance.

Figure 9 contains the print time for each pattern variant. It can be stated that as the pattern size increases (V and L), the print time can be reduced drastically. This could lead to more effective printing while taking into account that sometimes if the feature of the part is too small, the selected upscaled pattern might not fill perfectly the regions between the contour lines. Referring back, this was the main reason why the cross section of the standard specimen was increased, and a filling ratio of less than 75% was not investigated.

Fig. 9
figure 9

Print times

Full size image

4 Conclusion

In this paper, a simplified infill scaling method has been presented and has been tested on tensile test specimens. Based on the results, the following conclusions can be drawn:

  • It is possible to create the same infill ratio by defining different unit pattern sizes. If the thickness of the pattern’s struts is bigger, the size of the unit pattern must also be increased, to balance the space occupation.

  • Compared to the slicer’s automatic infill generation, in our solution, the effect of contour lines is more outstanding. If a part is created by this proposed scaled infill, it must be ensured that the number of contour lines must be selected in a way that a solid infill can be formed between the unit patterns, to achieve an appropriate mechanical resistance.

  • If the size of the unit pattern is bigger, it will lead to a reduced print time; however, it must always be borne in mind that for parts with smaller details, the increased size pattern might not be suitable

Although, with the presented pattern size optimization, a print time decrease was achieved without any change in the mechanical resistance, it must be borne in mind that the maximal size of the unit cell is affected by the geometry of the part. If the part has some narrow feature, where the distance between two opposite walls is small, then the unit cell cannot be equal or bigger than this distance. Therefore, an adaptive infill generation would be a possible future goal to extend the usability of the proposed method.