Compressive strength of cylindrical specimens generated by topology optimization: influence of finite element mesh refinement

ABSTRACT

Topology optimization (TO) is a mathematical method that optimizes a part’s layout, maintaining the project’s spatial constraints, considering loads, boundary conditions, and manufacturing to maximize its performance. Through mesh analysis, non-essential areas in the structure are identified, allowing efficient material removal without compromising its integrity. However, it is not usual to use TO without changing the part layout, focusing only on removing its internal mass, as is done in this paper. Therefore, two sets of cylindrical ABS samples were generated: one with an internal structure obtained by TO, varying the density of finite element mesh through the SolidWorks Software, and another with an internal structure chosen in the slicing stage before 3D printing. Then, the samples were subjected to the same compression tests. According to the results, all models support loads above the 10 kN specified. Unlike theoretical expectations, the best performance was from the coarse mesh TO, which was statistically equivalent to the tri-hexagonal and grid infill patterns. However, TOs made with medium and fine mesh pushed the limit of the piece shape, making the wall very thin and reducing its resistance. Using the infill pattern proved to be simpler to execute and less time-consuming to manufacture.

Keywords:
3D Printing; Additive Manufacturing; MEX; Infill Pattern; Simulation

1. INTRODUCTION

Topology optimization (TO) is a design feature that aims to find the best possible distribution of material in the part by reducing the amount of raw material used in the manufacturing process so that it satisfies the conditions previously established in the design. In the case of structural analysis, the goal is to make the part lighter, but with the desired strength. To do this, the program simulates mechanical performance, considering the specifications and design constraints, including the clamping points and the forces acting on the geometry of the part. For the application of TO, the model of the object under study is transformed into a geometric mesh that defines its shape.

1.1. Mesh density

In the model, each space is constituted by a finite element interconnected to others, whose common points are called nodes. Geometry is composed of a finite number of smaller segments that can have varied, triangular or tetrahedral shapes, for example, with known properties, in order to simplify processing.

The Finite Element Method (FEM) solves the problem of the physical continuity of objects by means of interpolations to generate meshes. For finite element computing, the mesh is first created from the surface data, discretizing the contour of the solid (vertices, edges and faces), generating a geometric mesh with information from the object of study [¹]. Only then is a mesh generated in the volume that, according to KURZ et al. [²], this introduces geometric errors such as gaps and overlap [³].

A study published by LIU and GLASS [⁴] evaluated the influence of finite element mesh density on the accuracy of static, modal, and impact analysis for fundamental structural components such as plates and beams. In these simulations, some differences were found between the meshes with different refinements. The simulation was performed on a plate virtually submitted to bending. The results of the bending in the plate, with the less dense mesh, showed an error of approximately 0.22% in relation to the more refined one, while the Von Mises stress values had more significant differences, reaching 1%. The theory of finite element analysis assumes that the displacements of the elements of the lattice are compatible with its size and that the forces are in equilibrium, while the stress is calculated from the displacement. That is why there is a difference in precision in the analyses, since the stresses exerted are constant in each element. For the simple geometry studied by LIU and GLASS [⁴], these differences are not significant, from an engineering point of view, but the processing time of the simulations was 40 times longer for the denser mesh. However, for more complex geometries, a higher mesh density is required so that the material deformation is more accurately represented. Therefore, the quality of the mesh used in the TO is decisive for the accuracy of the results, as the presence of inadequate meshes could trigger errors and distortions in the analyses. For this reason, choosing a mesh that is suitable in size and shape, with the right degree of refinement, is a critical factor in the optimization process.

1.2. Topology optimization simulation

The terminology TO was coined in 1988 by BENDSØE and KIKUCHI [⁵], based on homogenization theory. The principle is that it is possible to predict the macroscope behavior of a non-homogeneous material, such as a composite, if the property of the simplest repetitive unit, the unit cell, is known, considering it homogeneous. For this, FEM is used, as it makes the properties of the cells equivalent. Many approaches related to TO using FEM have emerged, among them: Density Methods, Level Set Methods, Phasefield Methods and Topological Derivatives, which were compared in a review written by SIGMUND and MAUTE [⁶]. Of these, the approach applied in this article is Density Methods.

Simulators using the topology optimization tool can be found in many Computer Aided Engineering (CAE) programs, such as: ABAQUS, SolidWorks, Ansys, Fusion, among others [⁷]. TYFLOPOULOS and STEINERT [⁸] found that 31% of the programs are open, of the remaining 69%, 44% have the student version. In this comparative study, it was found that 83% of commercial software uses Density Methods or Solid Isotropic Material with Penalty (SIMP), as well as SolidWorks, student version, used in this work.

In the SIMP method, Young’s modulus (E) is related to the density of the material by an interpolation of the law of powers, Eq. 1. When the relative density of the material (ρ_e) is equal to 1, the material is required (solid black element) e; when ρ_e = 0, the material can be removed (empty white element). This relationship proved to be physically acceptable for materials with a Poisson ratio equal to 1/3, if the penalty factor (p) is equal to 3. With this, there is a reduction in the contribution of elements with intermediate densities (gray element) [⁹].

$$E\left({\rho }_e\right)={\rho }_e^p{E}_0$$ (1)

According to the SIMP method, the volume is proportional to the density, but the overall stiffness (K_SIMP) has a lower value than the proportional one, Eq. 2. This is because in computation it is necessary to impose a minimum relative density, ρ_min, in order to avoid the singularity problem in the FEM, with 0 < ρ_min ≤ ρ_e ≤ 1, where N is the number of elements in the design domain and K_e is the rigidity of the element. For example, for an element with relative density assigned ρ_e = 0.5, penalty factor (p) = 3, and ρ_min = 0.001, the global stiffness matrix is calculated by multiplying K_e by 0.12587 [¹⁰].

$$K_{SMP(\rho )}={\displaystyle {\sum }_{\rho e=1}^N\left[{\rho }_{min}+\left(1-{\rho }_{min}\right){\rho }_e^p\right]}{K}_e$$ (2)

In the SolidWorks program used, during each iteration it was possible to search for a design for one of the following optimization goals: best stiffness to weight ratio (standard); minimize maximum displacement; or minimize mass with displacement constraint [¹⁰]. This is possible because, from the known displacement range in an optimal structure, it is possible to optimize the stiffness distribution in order to maximize the elastic strain energy. Similarly, for a known stress distribution of an optimal structure, it is possible to minimize the energy [¹¹].

In the standard case, where the aim is to maximize global stiffness, the goal is to minimize the global compliance of the structure (C), i.e., to minimize the sum of the deformations of all elements when subjected to a Force (F^T). Equations 3 and 4 consider the premises of the SIMP method (Equation 1) [¹²].

$$F^T=K(\rho ){\overline{u}}_e={\rho }_e^p{K}_e{\overline{u}}_e$$ (3)

$$C(\rho )=F^T{\overline{u}}_e^T={\rho }_e^p{K}_e{\overline{u}}_e{\overline{u}}_e^T$$ (4)

In this process, the optimization algorithm performs a sensitivity analysis to evaluate the effect of varying material densities through Eq. 5. Where ρ contains the relative densities of the elements ρ_e, u_e is the global displacement vector of the element node and u_e^T means the same relative the external force. In this analysis, for each element the algorithm combines the stiffness (Ke) of the material with a relative mass density factor “ranging from 0.0001 (for a hollow element with no load-bearing capacity) to 1.0 (for a solid element with load-bearing capacity)” [¹⁰]. Elements with low material density (< 0.3) are not of significant structural importance and may be eliminated from SolidWorks. While elements of high relative mass density (> 0.7) contribute to the overall stiffness [¹⁰].

$$\frac{dC}{d{\rho }_e}=-p{\left({\rho }_e\right)}^{p-1}{K}_e{\overline{u}}_e{\overline{u}}_e^T$$ (5)

According to the SolidWorks manual [¹⁰], “if the calculated sensitivity of each element is done independently and does not consider the connectivity between the elements, it can lead to discontinuity and disconnection of the volumes from the main geometry. This is known as the chessboard effect.” To prevent this from happening, it is necessary to apply a filter.

Optimization iterations continue until the variations of the objective function are converged and the iterations meet the convergence criteria. The target mass (M) is given by Eq. 6, where v_e is the volume of the element.

$${\displaystyle {\sum }_{e=1}^N{v}_e}{\rho }_e\le M$$ (6)

The application of TO to generate lighter structures is more common, both by changing the layout of the piece [⁸, ¹³, ¹⁴], as well as equipment in the field of mechanics [⁷, ¹⁵]. However, the search for internally porous structures, which this article deals with, is a more recent development, as presented by WU et al. (2021) [¹⁶], who showed the application of TO in multi-scale structures found in nature in bones or in bamboos, considered a porous infill design.

As the geometry generated in topology optimization usually has a complex layout, additive manufacturing has been the natural choice for the manufacture of parts, as pointed out by ZHU et al. [¹⁴], who reviewed TO applied to additive manufacturing (AM), considering the constraints of the manufacturing process. The combination of AM and TO allows the manufacture of products and alternatives that were previously impossible to achieve and test, since traditional processes have more geometric limitations.

Thus, the purpose of the article is to compare structures with internal porosity generated by TO, from different mesh densities, with those obtained with the infill patterns by 3D printing

1.3. 3D printing infill patterns

Print parameters are all the boundary conditions surrounding the printing process, defined before the printing process starts. These conditions can be related to the material to be printed, the 3D model and even the printer and must be introduced into the slicing software when preparing the model for printing [¹⁷]. They influence the quality of the surface, the mechanical properties of the final part, and even anisotropy [¹⁸].

The infill parameter defines how much material will be deposited inside the object, that is, between the walls of the piece. This filling can have several variations, both in quantity, patterns, even in materials, in the case of printers with 2 printheads [¹⁹]. When a project is printed with 0% infill, it is often said to have been printed in vase mode. This type of printing is very common on decoration items that will not suffer much effort, since they do not have great mechanical resistance. On the other hand, an object with 100% infill theoretically has the highest mechanical resistance that one can obtain with printing. However, completely filling a template brings several problems to the project, the main one being the printing time, which increases significantly in cases where the printing is massive. For these reasons, it is customary to vary the percentage of infill density of the models, balancing it with patterns that make it as resistant as possible.

Different infill patterns on the same printed part result in different mechanical properties. A study conducted by DOMÍNGUEZ-RODRÍGUEZ et al. [²⁰] resulted in a variation of more than 10 MPa, comparing honeycomb and rectangular filled samples during a compression test. CABREIRA and SANTANA [²¹] found variations in tensile strength from 2.4 to 1.1 MPa and impact strength from 3.8 to 1.5 kJ/m², between the rectilinear, grid, honeycomb and triangle infill patterns in PLA, with the rectilinear pattern being considered the best in terms of mechanical property.

TO is a complex process that involves analyzing the geometry of the part, the material used and the loading conditions to which the part will be subjected. On the other hand, parts with variable infill patterns are simple to produce. Comparing parts whose interior was made by infill patterns and topology optimized is justified by the need to evaluate the efficiency of TO as a tool for reducing internal mass. Therefore, this article tests TO to identify if it achieves a better optimization of the internal filling of the part, compared to the traditional way in which, after modeling, the choice of filling occurs in the slicing step, before 3D printing. This will be done by comparing printing time and the compressive strength.

2. MATERIALS AND METHODS

The printing material chosen was acrylonitrile-butadiene-styrene terpolymer (ABS), which is one of the options presents in the library of materials made available by SolidWorks. The material data can be found in Table 1. The ABS filament used for the printing was Natural Premium Ivory from the 3DFila brand.

Table 1
SolidWorks Version 2022 Library and 3DFila [²²] ABS Property Values.

PROPERTIES	SOLIDWORKS	3DFILA	UNITS
Specific mass	1020	1060	kg/m3
Elastic modulus	2000	2300	MPa
Filament		1323	MPa
Printed test specimen			MPa
Poisson’s ratio	0.394		–
Shear Modulus	318.9		MPa
Tensile stress at yield		18.9	MPa
Tensile strength	30		MPa
Tensile stress at break (printed test specimen)		33.8	MPa

To test the TO, in order to identify if it would achieve a better optimization of the internal filling of the part, compared to the traditional way, in which the choice of infill occurs in the slicing stage, before 3D printing, the procedure began in the generation of the specimens.

The process of obtaining the specimens begins with modeling via CAD software (SolidWorks), in which the standard cylindrical specimen for compression tests was designed, with a diameter of 2.5 cm and a height of 5 cm, following the proportion of 1:2, as defined by the ASTM standard (2015) [²³]. After the model was finalized, the process of making the specimens was branched into two fronts: the first, related to the preparation of topology optimized samples, while the second, with varied infill patterns, as shown in Figure 1.

Figure 1.
Sequence of sample manufacturing steps.

For the first branch, SolidWorks resources were used to generate specimens with meshes of 3 degrees of refinement: the first with the lowest possible number of nodes (1,926 nodes), the second with an intermediate value (10,410 nodes) and the last at the maximum level of refinement (58,564 nodes) made available by the software. With these three variations, the TO process began, in which each of the meshes was submitted to an analysis process to determine an optimized internal structure. For this, simulations were made by the method of mass reduction by 50% for each of the specimens, considering that: all faces would be fixed/unchanged, the applied load would be 10 kN and the symmetry would be in the XZ plane.

To perform the optimizations, a computer with an AMD Ryzen 5 5600X 3.7GHz, 6-Cores 12-Threads processor was used, obtaining simulation times of 45, 150 and 5,400 seconds for the coarse (S1), medium (S2) and refined (S3) mesh bodies, respectively.

After performing the topology study in SolidWorks, the most relevant structural nodes for the integrity of the mechanical properties of the bodies were listed, as well as the nodes that could be removed without causing significant losses in the strength of the structure. In the diagrams prepared by SolidWorks, the areas that could have the material removed are displayed, labeled by color, on a scale that varies between what should be kept and what could be removed. This diagram was drawn up according to the Von Mises stress, calculated for each of the study points. For slicing, the files with the standard triangle language format (.stl) of the optimized final parts were exported to the Ultimaker Cura Software, in which the 3D models were sliced, as well as prepared for 3D printing.

The second branch starts from the 3D model to the slicing step, generating three samples from the original model, based on the change in the infill through the Ultimaker Cura Software [²⁴]. This software was responsible for dividing the 3D model into layers, following the specified printing characteristics. The fillings chosen to be applied to the bodies were: triangular (S4, consisting of triangles interconnected by the faces), trihexagonal (S5, small triangles and hexagons) and grid (S6, the lines go in two directions making an angle of 90 degrees). These infill patterns were applied to the samples, so that the final masses of the pieces were reduced in a proportion similar to the mass reduction applied with the TO process, which for this study was 50%. Table 2 presents the printing parameters used in the manufacture of the samples.

Table 2
3D printing parameters used in slicing.

PARAMETERS	VALUES
Layer height	0.2 mm
Print speed	50 mm/s
Nozzle temperature	240°C
Bed temperature	70°C
Nozzle size	0.4 mm
Filament diameter	1.75 mm
Outlines count	3

The sample used as a reference was manufactured with a massive grid infill pattern, without mass reduction, as well as the topology optimized samples. At the end of the slicing, a Gcode file was generated, which translates all the specifications and characteristics of the samples directly to the 3D printer. Table 2 presents the printing parameters used in the manufacture of the samples.

For each of the seven samples, five specimens were printed using the Ender3 printer, totaling thirty-five 3D prints. In these prints, compressive strength tests were carried out using a 100 kN load cell, with displacement at a speed of 10 mm/min in a universal testing machine of the EMIC DL 10.000 brand. Fisher’s F test was applied to perform the statistical comparison of the means, with α = 0.05.

3. RESULTS AND DISCUSSION

First, in order to verify the efficiency of the printing process, the diameters of the specimens were measured, and the coefficients of variation (CV) were between 0.2% and 0.5%. The CV of the maximum strength of the samples was between 2 and 3%. The processing times for topology optimization and printing of the specimens are described in Table 3.

Table 3
Processing and printing time of the specimens.

SAMPLES	PROCESSING TIME (MINUTES)	PRINTING TIME (MINUTES)
S0	N/A	164
S1	10	170
S2	42	184
S3	83	185
S4	N/A	109
S5	N/A	110
S6	N/A	110

There is an increase of more than 50% in the printing time of the S1, S2 and S3 samples, generated by TO, compared to the printing times of the bodies with defined infill patterns. For an application where the internal structure is not a design concern, triangular, grid, and tri-hexagonal fills are more economically viable options due to the reduced printing time.

The performance of the samples in terms of compressive strength can be seen in Figure 2. When comparing the force versus displacement curves, it is possible to perceive 3 distinct groups of mechanical behavior, according to modulus of elasticity and yield stress, whose values are shown in Table 4: the reference (S0); the group of samples with intermediate behavior (S1, S4, S5 and S6), with the S4 (triangular) sample being more fragile than the others and the lowest performing (S2 and S3).

Figure 2.
Force versus displacement curves of all studied samples.

Table 4
Comparison between the values obtained and the standard sample S0.

SAMPLE	MODULUS (MPa)	YIELD STRENGHT (MPa)	MAXIMUM FORCE (kN)	FORCE RELATIVE TO S0 (%)	DISPLACEMENT IN MAXIMUM FORCE (mm)	DISPLACEMENT IN RELATION TO S0 (%)
S0	1354.1	45.0	27.870	100	23	100
S1	832.6	25.7	13.490	48.4	7.3	31.7
S2	725.3	21.0	10.990	39.4	3.0	13.0
S3	690.9	19.2	10.550	37.9	3.5	15.2
S4	824.2	25.0	12.600	45.2	2.4	10.4
S5	814.7	25.4	13.760	49.4	11.2	48.7
S6	831.7	25.6	13.350	47.9	8.4	36.5

Table 4 shows the values of the maximum force and displacement of each sample in relation to the S0 sample, since it was printed with a solid interior and without mass reduction. The reference sample (S0) showed a maximum strength of 28 MN, i.e., a compressive strength of 55 MPa, which is very reasonable, since SolidWorks indicates a tensile strength of 30 MPa for the ABS used in the simulation. As noted by BANJANIN et al. [²⁵], in compression test, strength is dictated by the property of the material, as the filament is squeezed, whereas in a tensile test, strength depends heavily on the adhesion between the layers. However, according to GUESSASMA et al. [²⁶], during compression there is a detachment during loading due to lateral expansion, which depends on the orientation of the filaments in each layer in the plastic deformation stage. Only in the case of the S0 sample, the plastic deformation led to a strength gain, since the maximum strength was greater than the yield stress.

When comparing the data obtained for each of the samples with the reference (S0), a significant reduction in the mechanical compressive strength of all samples is perceived. These losses are expected and occur proportionally due to the 50% reduction in the mass of the samples, as well as the creation of internal structures with cavities, both in optimizations and infill patterns.

As shown in Table 3, the best compressive strength results were obtained by samples with varying infill patterns in relation to optimized ones. In terms of absolute values, the trihexagonal pattern (S5) had the lowest percentage reduction of the maximum force, resisting the equivalent of 49.4% of the maximum force supported by S0. However, this sample is statistically equivalent to the optimized S1 and grid-infilled samples, S6.

Sample S1 sustained a significantly higher load than S2 and S3 in percentage values relative to S0 of 9% and 10.5%. Samples S2 and S3 did not show significant differences between them. The S5 and S6 samples prepared with different infill patterns did not vary significantly from each other, but compared to the S4 sample relative to S0 they were higher by 4.2% and 2.7%, respectively.

It was expected that the optimized sample with the coarsest mesh (S1) would present the worst performance, since the removal of a cell would represent the biggest quantity. According to FREND [²⁷], the most resistant infill pattern in the XY horizontal direction is the trihexagonal, the gyroid is in second place, the grid in third and the triangle in fourth. According to the author, the triangle pattern has the disadvantage of having to interrupt the flow at intersections. Giroid, which involves wavy lines in alternating directions, was not chosen for testing because this pattern takes longer to print and has a more complex geometry. Thus, the best performance of the tri-hexagonal and the worst performance of the triangle pattern was already expected, but it was not expected that the tri-hexagonal would be equivalent to the grid. All of them presented a yield force greater than the load applied in the simulation, of 10 kN.

Among the samples of different mesh densities, the S1 sample stood out, since it achieved a mechanical resistance similar to those obtained for bodies with filling patterns, being significantly superior compared to the triangular filling (S4). This was probably due to the fact that the Solidwork software had a different strategy in this case, since it increased the wall, instead of creating a central column as it did in the S2 and S3 samples, demonstrating a dependence on the solution in relation to the type of mesh. According to BENDSØE and SIGMUND [¹¹], “there is a lack of closedness of the admissible set of designs” caused by a greater number of holes when the mesh is refined, which would generate an internal structural layout similar to the microstructures that theory predicts would be a centralized column.

Figure 3 presents the diagrams with the Von Mises stress distributions, calculated for each element of the mesh along the internal structure of the parts. The highest values found are those of the green region, whose mass remained in the optimized part. In descending order, the values were S1 (152.6 MPa), S3 (140.9 MPa) and S2 (116.0 MPa), resulting in values above the experimental value found in this study (Table 3) and in the results presented by BANJANIN et al. [²⁵], which were 17.781 MPa in the compression tests performed on specimens printed in ABS. The removed parts of the part are in dark blue, whose minimum stresses are 9.885 × 10⁻² Pa for S1, 4.020 × 110⁻² Pa for S2 and 5.119 × 10⁻² Pa for S3., well below the yield stresses found.

Figure 3.
Von Mises stresses of topology studies.

Figure 4 presents the images obtained for the specimens of the samples studied, in which it is possible to observe the lateral views, cross-sections and details of the cracks of the specimens in question.

Figure 4.
Images of the specimens after compression tests.

It can be seen that the specimen of the S0 sample, despite being theoretically massive, has layers on the inside, resulting from the manufacturing process by 3D printing. It is noted that the specimens tend to break in the area called the sewing line, in which the printer performs the layer change vertically. This line can be considered a concentration of tension in the body, which when subjected to effort often tends to break.

Analyzing the behavior of the fissures of the sample bodies, it can be seen that only at S0 does the specimen complete rupture. Despite this, it is noted that the cracks of all bodies also start in the sewing line, reinforcing the hypothesis that this is the defect from which there is crack propagation of the piece. In addition, except for S5, all samples showed the formation of a bulge in the vicinity of the rupture section, however, these bulges were present at different heights for each of the cases. For the case of S5, it is observed that the deformation of the specimen is presented by the formation of several undulations along the entire length of the specimen, demonstrating a better distribution of the deformation along the structure. Among the topology optimized samples, it is noted that S3 is the one with the most distinct behavior, because although it also has a bulging region, it presented a less uniform structure than those formed in the other cases. In addition, there was also a rupture in the transverse direction of the specimen, just above the bulging.

Due to the internal geometries generated by the optimization process, the cross-sections of the cavities vary when the vertical axis of the specimens is traveled. In Figure 5, it is possible to see the different cross-sections of the optimized bodies, which justifies the rupture at different specimen heights for each sample.

Figure 5.
Top view of the cross-sections of topology optimized samples.

One of the printing parameters used was the fixed amount of three outer layers of material on all faces of the specimens. Due to the variation in the geometry of the cross-sections mentioned above, this parameter could not be implemented in samples S2 and S3, since the optimization generated walls with a thickness smaller than the three defined layers, which are 2 and 1 layers, respectively. In this situation, no matter how much this boundary condition has been defined, the slicer inserts only the number of layers consistent with the spaces in which there are no cavities. For these bodies, a reduction in the maximum force supported by the bodies was observed, indicating an influence of the thickness of the walls on the resistance of the bodies that had not been predicted.

4. CONCLUSIONS

Highlighting the influence of the degree of refinement of meshes on the compressive strength of printed bodies, it was borne in mind that the mechanical strength of a piece would be given by a relationship directly proportional to the degree of refinement of the polygonal mesh of the model. When performing the mechanical tests of the specimens, results different from the theoretical expectations were found, since the S1 sample, with a lower degree of refinement, was the one that most resisted the compression load, when compared to the standard sample. It is estimated that one of the main factors that influenced this difference between theory and practice was the wall thickness of the specimens. After all, as mentioned in the development of the methodology, a constant of three outer layers was used on all sides of the models so that they could be printed. However, because the internal structures of the geometries generated in the S2 and S3 samples are very close to the external area, the edge layers ended up being reduced at certain points of the parts during the slicing process. It is believed that this area was responsible for giving greater resistance to the model and, for this reason, when it ceased to exist, there was a drop in the absolute value of this mechanical property.

Regarding the internal filling patterns, it was expected that the tri-hexagonal type would give the pieces the greatest mechanical resistance among those chosen, however, it was equivalent to the grid. Both are considered highly efficient ways to optimize the internal geometry of a body, without significant losses in the mechanical properties of the part. However, the triangular infill pattern proved to be the least resistant of the three.

Finally, when comparing an internal structure optimization using topology optimization with one that uses changing infills, it is noted that, although all of them support loads above the design loads, predefined as 10 kN, and have similar resistance ranges, there is an advantage in the infill patterns, compared to structures with optimized cavities. The infill proved to be simpler to be executed, in addition to consuming less time to manufacture the part in question.

5. BIBLIOGRAPHY

• ^[1] FURHT, B., Encyclopedia of Multimedia New York, Springer-Verlag, 2006. doi: http://doi.org/10.1007/0-387-30038-4.
  » https://doi.org/10.1007/0-387-30038-4
• ^[2] KURZ, S., SCHÖPS, S., UNGER, G., ET AL., “Solving maxwell’s eigenvalue problem via isogeometric boundary elements and a contour integral method”, Mathematical Methods in the Applied Sciences, v. 44, n. 13, pp. 10790–10803, Mar. 2021. doi: http://doi.org/10.1002/mma.7447.
  » https://doi.org/10.1002/mma.7447
• ^[3] CUILLIÈRE, J., FRANCOIS, V., “Integration of CAD, FEA and topology optimization through a unified topological model”, Computer-Aided Design and Applications, v. 11, n. 5, pp. 493–508, 2014. doi: http://doi.org/10.1080/16864360.2014.902677.
  » https://doi.org/10.1080/16864360.2014.902677
• ^[4] LIU, Y., GLASS, G., “Effects of mesh density on finite element analysis”, SAE Technical Papers, v. 2, n. 4, Aug. 2013. doi: http://doi.org/10.4271/2013-01-1375.
  » https://doi.org/10.4271/2013-01-1375
• ^[5] BENDSØE, M.P., KIKUCHI, N., “Generating optimal topologies in structural design using a homogenization method”, Computer Methods in Applied Mechanics and Engineering, v. 7, n. 2, pp. 197–224, Mar. 1988. doi: http://doi.org/10.1016/0045-7825(88)90086-2.
  » https://doi.org/10.1016/0045-7825(88)90086-2
• ^[6] SIGMUND, O., MAUTE, K., “Topology optimization approaches: a comparative review”, Structural and Multidisciplinary Optimization, v. 48, n. 6, pp. 1031–55, Aug. 2013. doi: http://doi.org/10.1007/s00158-013-0978-6.
  » https://doi.org/10.1007/s00158-013-0978-6
• ^[7] SLAVOV, S., KONSULOVA-BAKALOVA, M., “Optimizing weight of housing elements of two-stage reducer by using the topology management optimization capabilities integrated in solidworks: a case study”, Machines, v. 7, n. 9, pp. 1–15, Jan. 2019. doi: http://doi.org/10.3390/machines7010009.
  » https://doi.org/10.3390/machines7010009
• ^[8] TYFLOPOULOS, E., STEINERT, M.A., “Comparative study of the application of different commercial software for topology optimization”, Applied Sciences (Basel, Switzerland), v. 12, n. 611, pp. 1–23, 2022. doi: http://doi.org/10.3390/app12020611.
  » https://doi.org/10.3390/app12020611
• ^[9] BENDSØE, M.P., SIGMUND, O., “Material interpolation schemes in topology optimization”, Archive of Applied Mechanics, v. 69, n. 9-10, pp. 635–654, May. 1999. doi: http://doi.org/10.1007/s004190050248.
  » https://doi.org/10.1007/s004190050248
• ^[10] SOLIDWORKS, https://www.solidworks.com/pt-br/domain/design-engineering, accessed it September, 2022.
  » https://www.solidworks.com/pt-br/domain/design-engineering
• ^[11] BENDSØE, M.P., SIGMUND, O., Topology optimization theory, methods, and applications, 2nd ed., Berlin, Springer-Verlag, 2004. doi: http://doi.org/10.1007/978-3-662-05086-6.
  » https://doi.org/10.1007/978-3-662-05086-6
• ^[12] KAZAKIS, G., KANELLOPOULOS, I., SOTIROPOULOS, S., ET AL., “Topology optimization aided structural design: Interpretation, computational aspects and 3D printing”, Heliyon, v. 3, n. 10, pp. e00431, 2017. doi: http://doi.org/10.1016/j.heliyon.2017.e00431. PubMed PMID: 29226262.
  » https://doi.org/10.1016/j.heliyon.2017.e00431
• ^[13] MENG, L., ZHANG, W., QUAN, D., ET AL., “From topology optimization design to additive manufac-turing: today’s success and tomorrow’s roadmap”, Archives of Computational Methods in Engineering, v. 27, n. 3, pp. 805–830, Mar. 2020. doi: http://doi.org/10.1007/s11831-019-09331-1.
  » https://doi.org/10.1007/s11831-019-09331-1
• ^[14] ZHU, J., ZHOU, H., WANG, C., ET AL., “A review of topology optimization for additive manufacturing: status and challenges”, Chinese Journal of Aeronautics, v. 34, n. 1, pp. 91–110, Aug. 2021. doi: http://doi.org/10.1016/j.cja.2020.09.020.
  » https://doi.org/10.1016/j.cja.2020.09.020
• ^[15] PISARCIUC, C., DAN, I., CIOARA, R., “The influence of mesh density on the results obtained by finite element analysis of complex bodies”, Materials (Basel), v. 16, n. 7, pp. 2555, Mar. 2023. doi: http://doi.org/10.3390/ma16072555. PubMed PMID: 37048854.
  » https://doi.org/10.3390/ma16072555
• ^[16] WU, J., SIGMUND, O., GROEN, J.P., “Topology optimization of multi-scale structures: a review”, Structural and Multidisciplinary Optimization, v. 63, n. 3, pp. 1455–1480, Mar. 2021. doi: http://doi.org/10.1007/s00158-021-02881-8.
  » https://doi.org/10.1007/s00158-021-02881-8
• ^[17] SONG, Y., LI, Y., SONG, W., ET AL., “Measurements of the mechanical response of unidirectional 3D-printed PLA”, Materials & Design, v. 123, pp. 154–164, 2017. doi: http://doi.org/10.1016/j.matdes.2017.03.051.
  » https://doi.org/10.1016/j.matdes.2017.03.051
• ^[18] DIZON, J.R.C., DIZON, J.R.C., “”Mechanical characterization of 3D-printed polymers”, Additive Manufacturing, v. 20, pp. 44–67, Dec. 2018. doi: http://doi.org/10.1016/j.addma.2017.12.002.
  » https://doi.org/10.1016/j.addma.2017.12.002
• ^[19] ZIVIC, F., MITROVICA, S., “The influence of the 3D printing infill and printing direction on friction and wear of polylactic acid (pla)under rotational sliding”, Journal of Friction and Wear, v. 42, n. 2, pp. 106–111, Jan. 2021. doi: http://doi.org/10.3103/S1068366621020124.
  » https://doi.org/10.3103/S1068366621020124
• ^[20] DOMÍNGUEZ-RODRÍGUEZ, G., KU-HERRERA, J., PEREZ HERNANDEZ, A., “An assessment of the effect of printing orientation, density, and filler pattern on the compressive performance of 3D printed ABS structures by fuse deposition”, International Journal of Advanced Manufacturing Technology, v. 95, n. 5-8, pp. 1685–1695, Nov. 2018. doi: http://doi.org/10.1007/s00170-017-1314-x.
  » https://doi.org/10.1007/s00170-017-1314-x
• ^[21] CABREIRA, V., SANTANA, R.M.C., “Effect of infill pattern in Fused Filament Fabrication (FFF) 3D Printing on materials performance”, Matéria (Rio de Janeiro), v. 25, n. 3, pp. e12826, 2020. doi: http://doi.org/10.1590/s1517-707620200003.1126.
  » https://doi.org/10.1590/s1517-707620200003.1126
• ^[22] 3DFILA, PLA: tudo o que você precisa saber sobre esse Filamento 3D Fila, https://blog.3dfila.com.br/pla-tudo-sobre-o-filamento-pla/, accessed it September, 2022.
  » https://blog.3dfila.com.br/pla-tudo-sobre-o-filamento-pla/
• ^[23] ASTM INTERNATIONAL, ASTM-D-695: Standard Test Method for Compressive Properties of Rigid Plastics, Pennsylvania, USA, ASTM, 2015.
• ^[24] ULTIMAKER, https://ultimaker.com/software/ultimaker-essentials, accessed it September, 2022.
  » https://ultimaker.com/software/ultimaker-essentials
• ^[25] BANJANIN, B., VLADIĆ, G., PÁL, M., ET AL., “Consistency analysis of mechanical properties of elements produced by FDM additive manufacturing technology”, Matéria (Rio de Janeiro), v. 23, n. 4, pp. e12250, Jan. 2018. doi: http://doi.org/10.1590/s1517-707620180004.0584.
  » https://doi.org/10.1590/s1517-707620180004.0584
• ^[26] GUESSASMA, S., BELHABIB, S., NOURI, H., ET AL., “Anisotropic damage inferred to 3D printed polymers using. fused deposition modelling and subject to severe compression”, European Polymer Journal, v. 85, pp. 324–340, Oct. 2016. doi: http://doi.org/10.1016/j.eurpolymj.2016.10.030.
  » https://doi.org/10.1016/j.eurpolymj.2016.10.030
• ^[27] FREND, T., 14 Cura Infill Patterns: Strongest to Weakest, https://3dprintinggeek.com/cura-infill-patterns/, accessed it September, 2022.
  » https://3dprintinggeek.com/cura-infill-patterns/

Publication Dates

• Publication in this collection
  27 Jan 2025
• Date of issue
  2024

History

• Received
  18 Sept 2024
• Accepted
  21 Nov 2024