Additively manufactured polymer punches for deep drawing: performance assessment and infill design Open Access

The replacement of metal components with polymers in manufacturing has become an increasingly popular trend across various industries due to advancements in polymer technology (Ligon et al., 2017). This shift is driven by several factors, including the need for weight reduction; polymers are significantly lighter than metals, which makes them ideal for applications where reducing weight is crucial, such as in the automotive (Patil et al., 2017) and aerospace industries (Dyer and Kumru, 2023). Lighter materials contribute to improved fuel efficiency and performance (Rajak et al., 2019). From a cost-efficiency perspective, polymers are generally less expensive to produce and process compared to metals. They often require less energy to manufacture, and their lower density can lead to savings in material costs (Parandoush and Lin, 2017). Moreover, polymers are inherently resistant to corrosion and chemical attack, unlike metals, which require coatings or treatments to prevent rust and degradation. This makes them suitable for use in harsh environments (Ohtsuka, 2012). Another advantage is on design flexibility; polymers offer greater design flexibility than metals, allowing for the creation of complex shapes and intricate designs through techniques like injection moulding. This flexibility enables manufacturers to consolidate multiple metal parts into a single polymer component, reducing assembly time and cost (Gonzalez-Gutierrez et al., 2018). Common polymers used as metal replacements are Polyamide (Nylon), Polycarbonate, Polyetheretherketone (PEEK), Polypropylene, Acrylonitrile Butadiene Styrene (ABS) and Thermoplastic Elastomers (TPE) (Tan et al., 2020).

In manufacturing of goods, the use of polymer-based materials is applied to create tooling (such as moulds, dies and fixtures) quickly and efficiently, often used in prototyping and short production runs (Masood and Song, 2004). Polymer tooling is generally less expensive than traditional metal tooling; this makes it ideal for prototypes and low-volume production, where the high cost of metal tooling cannot be justified (King and Tansey, 2002). Moreover, the use of polymer materials allows for faster production times, reducing the lead time from design to the finished tool (Equbal et al., 2015); polymers are easier to shape and modify than metals, allowing for complex geometries and intricate designs to be produced more easily (Boparai et al., 2016).

To promote the use of polymer tools for rapid tooling application in this research the authors tested it in the production of aluminum cups by deep drawing process. In literature the tools were already tested demonstrating their advantages in the production of steel and aluminum cups (Giorleo and Ceretti, 2022); authors test performance of different polymer reinforced and not finding good solution using Nylon reinforced with carbon fiber in the production of DC04 cups (Bergweiler et al., 2020); Athale et al. showed that GF-PC polymer AM tools can be used to stamp HSS 590 steel sheets relevant for automotive applications at volumes appropriate for prototyping (Athale et al., 2023). Other authors developed and verified a method for predicting the failure mode of a cylindrical cup drawing die fabricated from polyurethane in the production of AISI 304 and Al 1100 cups; the authors verified that plastic deformation occurs primarily due to the wrinkles in the sheet metal when no blankholder is used and the drawing die fails when the maximum principal stress reaches the flexural strength of the die material (Park and Colton, 2005). Schuh et al. presented a cupping test that confirms Polylactic acid (PLA) tools as material sufficiently stable for sheet metals and provide similarly good results as metallic tools in terms of formability (Schuh et al., 2020). In addition, Frohn-Sörensen et al. tested PLA versus nylon demonstrating not only best cups accuracy with PLA but also that a considerable degradation of the Nylon tooling was observed from the small batch of 30 pieces, leading to an additional deviation of cup accuracy (Frohn-Sörensen et al., 2022).

In advance to the good results already obtained a benefit to produce these tools with additive manufacturing is the possibility to realize an infill strategy to do not modify the external geometry of tools saving material used and production time (Tanveer et al., 2022). The concept of lightening in tool design refers to reducing the weight of tools while maintaining or enhancing their performance, durability and usability (Suteja, 2021). Nowadays, there are different infill strategy that could be applied as a function of pattern and density modifying polymer mechanical properties (Mishra et al., 2021; Gunasekaran et al., 2021); however, a not deeper investigation was done about their application in deep drawing processes.

Recent research has also explored numerical and multiscale modeling approaches to investigate the mechanical response of material extrusion components as a function of infill architecture and density. Finite element homogenization techniques have been proposed to predict the effective elastic properties of printed lattice structures under different raster angles, build orientations and infill ratios (Gonabadi et al., 2022). Similarly, thermomechanical simulations have been used to estimate process-induced distortions and dimensional deviations associated with infill design variations (Al Rashid and Koç, 2023). More advanced multi-scale homogenization frameworks incorporating strain-gradient theories have also been developed to capture the influence of architected microstructures on the macroscopic response of additively manufactured composites (Sarar et al., 2024). These approaches demonstrate the potential of numerical modeling to explore the expanded design space enabled by additive manufacturing. Nevertheless, the majority of these studies focus on the prediction of elastic properties, dimensional accuracy or static structural response of printed components. Limited attention has been devoted to the cyclic mechanical performance and progressive degradation mechanisms of additively manufactured polymer tools operating under forming conditions, where contact stresses, axial compression and abrasive wear interact over repeated forming cycles.

Despite the encouraging results reported in the literature, most existing studies primarily demonstrate the feasibility of using additively manufactured polymer tools for sheet metal forming, often focusing on short production runs or single design configurations. However, a systematic understanding of how internal infill design parameters influence punch deformation mechanisms, wear evolution and dimensional stability under repeated forming cycles is still lacking. In particular, limited attention has been devoted to correlating infill-driven mechanical response with progressive geometric deviations of both the tool and the formed parts. Since material extrusion enables tailored internal architectures without modifying external tool geometry, infill pattern and density may significantly affect stiffness, load distribution, elastic recovery and resistance to abrasive wear during deep drawing.

To address this gap, the present study goes beyond feasibility assessment and investigates the role of infill pattern and density as governing parameters of punch deformation and wear mechanisms. A two-stage experimental campaign was designed: a controlled screening phase to identify the most promising infill configuration based on dimensional accuracy, and an extended batch production phase (up to 99 cycles) to analyse progressive tool degradation and stabilization behaviour. By combining dimensional measurements of cups and punches with statistical analysis, the study provides new insight into the interplay between axial compression, elastic deflection and abrasive wear in additively manufactured polymer punches. These findings contribute to a more mechanistic understanding of infill-driven performance in polymer additive tooling for deep drawing applications.

The experimental campaign was structured in two sequential phases to both identify the most suitable punch configuration and evaluate its stability under repeated use. In the first step, named screening experiments, the influence of two different infill strategies (triangular and gyroid) and two density levels (35% and 50%) was investigated. For each combination, 10 cups were produced, and the tenth cup was selected for dimensional analysis. This choice was motivated by the intention to assess punch performance after a small prototyping batch rather than after a single draw. The evaluation focused on cup radius, drawing depth, fillet radius and the overall condition of the punches. The configuration that provided the best compromise between weight reduction and cup accuracy was then selected for the second phase. In this second step, named batch production experiments, six punches were manufactured with the optimal infill and density identified in the screening campaign. Each punch was used to produce a different number of cups – specifically one, five, 10, 25, 50 and 99 – to capture the effect of progressive tool deformation and wear on cup dimensional accuracy.

The experimental tests consist in deep drawing of a an Al1050 blank sheet with initial diameter of 70 and 1 mm thick. The blank was fixed between a blankholder and a die to avoid typical deep drawing defect as wrinkling; blanks were formed by a cylindrical punch having diameter equal to 39 and 6 mm of fillet radius; the target was to produce a cup 20 mm depth. All the main dimension of cup (a) and tools (b-d) designed for this research are presented in Figure 1 while Table 1 resumes main data of process.

Material selected for blankholder and the forming die was 45NiCrMo16 steel; the punches were produced with Material Extrusion process in Onyx that was a commercial name of PA12 filled with short carbon fiber (SCF Nylon); to test the effect of lightening two different infill geometry (Triangular, Gyroid) and two different density level (35% and 50%) were designed. The selection of the triangular and gyroid infill geometries was motivated by their fundamentally different structural characteristics and their widespread adoption in material extrusion processes. The triangular pattern is based on a truss-like architecture, promoting directional stiffness and efficient load transfer along linear paths. In contrast, the gyroid pattern is a triply periodic minimal surface, characterized by continuous curvature and smoother stress redistribution. Comparing these two configurations allowed the investigation of how distinct internal load paths influence punch deformation and wear mechanisms without altering the external tool geometry. The chosen infill densities (35% and 50%) were selected to represent a realistic design window for load-bearing polymer components. Densities below approximately 30% generally lead to insufficient stiffness for forming applications, while values above 60% reduce the lightweight and economic advantages of additive manufacturing without reaching the mechanical response of fully dense parts. The selected range therefore reflects a compromise between structural integrity and material/time efficiency, consistent with industrial practice. Table 2 summarizes the properties of the materials used while the Table 3 collect the data related to the punch design and manufacturing. The punches were produced by material extrusion (also referred to as Fused Filament Fabrication, FFF), machine Mark 2 (Markforged, MA, US).

The representation of a layer with different pattern (a,b for gyroid and c,d for triangular) and density (a,c for 35% and b,d for 50%) is visible in Figure 2. The internal patterns allow to reduce the quantity of row material with a saving in terms of cost ranging between 38% and 54%. Furthermore, fill pattern reduces the time necessary to produce the punch, especially with triangular shape. During punch printing, the extrusion temperature was set to 275 °C, the maximum printing speed was 280 mm/s, the layer height was 0.1 mm and the nozzle diameter was 0.4 mm. Two wall layers were set for the punch shell, each printed with a line width of 0.4 mm, resulting in an overall shell thickness of approximately 0.8 mm. This configuration was selected to ensure sufficient surface stiffness while maintaining the lightweight design enabled by the internal infill. Each drawing test was repeated 10 times to investigate the possible effect of the wear or deformation of the punch on the geometrical features of the cups.

A preliminary visual inspection was carried out to identify any visible defects or significant differences among the produced cups. The quantitative evaluation of the drawing process was then performed through a geometrical inspection of both cups and punches using a CMM Machine Cyclone Series 2 (Renishaw, Wotton-under-Edge, UK). For comparison, measurements were taken from the 10th cup and from the punches after forming, focusing on the cup/punch radius, cup drawing depth and punch height.

Specifically, the cup/punch radius was obtained by acquiring the x–y coordinates of 300 points along a circumference located 10 mm from the cup bottom or punch top (corresponding to 50% of the drawing depth). The drawing depth of the 10th cup and the punch height were determined from a cross-sectional profile (about 100 points), isolating 25 points positioned away from the fillet to minimize edge effects. Figures 3(a) and (b), illustrates the procedure adopted to perform these dimensional measurements. The radius data were used to detect potential wear or radial deformations of the punches, which could lead to enlarged diameters or higher dispersion, whereas drawing depth measurements provided information on the extent of punch compression during forming, reflected in changes in punch height and consequently in cup depth.

An additional parameter of interest was the cup fillet radius, which served as an indirect indicator of punch deformation in the fillet region. To estimate this value, the z-deviation of the experimentally measured fillet profile was compared with the corresponding ideal CAD geometry. However, it was observed that variations in drawing depth could artificially increase this deviation. To address this, a reference CAD curve was generated using the experimentally measured drawing depth as the starting point. The deviation in the z-direction, calculated with respect to this adapted CAD reference, was defined as the ΔCAD. Figure 4(a) shows a representative comparison between the experimental curve, the CAD model and the reference CAD. To localize the deformation with greater detail, the fillet was further divided into four equal zones, as illustrated in Figure 4(b). Specifically, zone 1 corresponds to the region closest to the cup bottom, while zone 4 extends to the transition with the vertical wall of the cup.

The accuracy data were analyzed using Minitab software (Minitab, PA, US). For cup radius and drawing depth, a two-way (screening step) and one-way (batch production step) ANOVA was performed to verify the statistical significance of the process parameters, followed by Tukey’s post hoc test to evaluate differences among the tested levels. The results were displayed through interval plots, which allowed visualization of the mean values and their variability across conditions. For the fillet radius, which was expressed as the ΔCAD, an individual value plot was used. This approach was adopted because the parameter is defined as a single value for each test condition, making it unsuitable for statistical grouping. ΔCAD was further analyzed as a function of the four predefined fillet zones, to identify the regions most affected by tool deformation.

In this section, the results of the experiments are listed. The preliminary analysis executed by a visual inspection on cups produced and on punches after drawing process is illustrated in Figure 5.

The different punches used in this initial experimental campaign, together with the 10th cup produced for each configuration, are shown in Figure 5.

Observing the 10th cup produced by each combination (Figure 5) reveals that the cup obtained by Test G35, using a punch with gyroid infill at 35% density, is noticeably different from the others. This cup has a higher flange, a lower depth and a larger fillet radius. These defects suggest that the punch underwent compression during the drawing process, leading to the reduced depth and the corresponding increase in flange height. In addition, the increase in the fillet radius indicates that this compression was accompanied by radial expansion. The other cups do not show any significant differences. In examining the punches, the main defect found was in Test G35, where a blob is present on the cylinder surface. The results of the statistical analyses carried out on the measurements of the cup radius, drawing depth and fillet radius are presented in Figure 6. As previously explained, the measurements were taken from the tenth cup produced for each configuration. The main p-values from the ANOVA, used to assess the effect of the infill strategy and density on the measured parameters, are reported in Table 4.

Merging the results presented in Figure 6 and Table 4, it is possible to observe clear trends in the influence of infill strategy and density on the measured parameters. With regard to the cup radius, the ANOVA results highlight the statistical significance of all the considered factors, with slightly higher p-values associated with the infill strategy, indicating a lower influence compared to density. The interaction between the two factors exhibits the lowest p-value, confirming a combined effect. As shown in Figure 6(a), the Tukey test reveals that the only configuration providing a significantly lower cup radius is G35, while all other configurations yield comparable values, with differences not exceeding 0.05 mm. A similar analysis can be conducted for the drawing depth. In this case, all parameters exhibit p-values below 0.001, indicating a strong statistical significance. As illustrated in Figure 6(b), the gyroid configuration at 35% density results in lower drawing depth compared to the other configurations. The Tukey test further indicates that T35 and G50 behave similarly, whereas the highest drawing depth is achieved with T50. The analysis of the cup fillet radius, expressed in terms of relative ΔCAD, follows a trend consistent with that observed for drawing depth. The ANOVA confirms the statistical significance of all factors, while Figure 6(c) shows that the G35 configuration again exhibits the highest ΔCAD values, corresponding to lower dimensional accuracy. Conversely, the triangular infill provides lower ΔCAD values, as confirmed by the Tukey test. A more detailed evaluation across the four fillet zones [Figure 6(d)] indicates that zones 1 and 2, located near the bottom of the cup, present smaller deviations compared to zones 3 and 4 in the upper region of the fillet, where deformation effects are more pronounced. Based on these results, it can be concluded that the triangular infill provides overall better performance compared to the gyroid configuration, as it allows higher dimensional accuracy by maximizing cup radius and drawing depth while minimizing ΔCAD. Increasing the infill density generally improves performance; however, the Tukey analysis indicates that, except for drawing depth, the T35 and T50 configurations are statistically comparable.

Considering that drawing depth can be compensated by adjusting the punch stroke, and given that the difference between the mean values is limited to 0.35 mm, the T35 configuration was selected for the subsequent batch production phase. In addition, T35 offers reduced material consumption and shorter printing time compared to T50, making it more suitable for cost-efficient small-batch production. This selection also enables the evaluation of whether a lower-density configuration can maintain adequate dimensional stability over extended production cycles.

To further evaluate the performance of the selected punch configuration under repeated use, a batch production campaign was conducted. The objective of this phase was to investigate whether dimensional accuracy and tool integrity could be maintained over an extended number of forming cycles.

The cups produced together with their respective punches are shown in Figure 7(a). The experimental campaign proceeded without any critical issues, and all punches successfully completed the planned production batch. Visual inspection revealed no macroscopic defects either on the punches or on the cups produced. From a qualitative perspective, however, a comparison between the first and the 99th cup highlights a variation in both drawing depth and fillet radius [Figure 7(b)]. Another noteworthy difference concerns the punch morphology: the punch used for only one cup (P01) compared to the punch that formed 99 cups (P99). As illustrated in Figure 7(c), the step-like effect measured along the punch fillet – typical of FFF processes – appears less pronounced in P99. This observation suggests that the tool underwent flattening and, when subjected to the abrasion typical of the process, experienced wear that progressively reduced the characteristic step morphology of FFF-produced surfaces.

With regard to the statistical analysis presented in Figure 8, several considerations can be drawn about the influence of the increasing number of cups produced on the parameters under investigation. It should first be emphasized that all p-values obtained from the one-way ANOVA were below 0.001, clearly demonstrating the high statistical significance of the number of cups produced.

With regard to the statistical analysis presented in Figure 8, several considerations can be drawn about the influence of the increasing number of cups produced on the parameters under investigation. All p-values obtained from the one-way ANOVA are below 0.001, confirming the high statistical significance of production volume. The evolution of the cup radius shows a gradual decreasing trend as production progresses [Figure 8(a)]. However, the Tukey test indicates that most production levels belong to overlapping statistical groups, suggesting that the values remain comparable up to approximately the 50th cup. A statistically significant difference is observed only for the 99th cup. It is important to note that the magnitude of this variation is limited, with a mean difference of approximately 70 µm between the first and the last cup, which remains within acceptable dimensional tolerances. A similar interpretation applies to the drawing depth. Although the ANOVA confirms statistical significance, the relatively high dispersion of the data reduces the practical relevance of the trend suggested by the interval plot in Figure 8(b). Most production levels belong to multiple overlapping groups, with only the first and 99th cups showing a clear distinction. This indicates that the first cup exhibits a slightly greater drawing depth than the 99th one, while intermediate levels remain largely comparable. The overall variation is limited, with a mean difference of approximately 0.24 mm, which can be considered negligible in the context of the process. The parameter ΔCAD, representing the deviation of the cup fillet radius, shows the highest sensitivity to production volume. As illustrated in Figure 8(c), ΔCAD increases progressively with the number of cups produced. In this case, the relatively low variability within each production level allows the Tukey test to distinguish multiple groups, with overlaps occurring mainly between consecutive levels. A more detailed analysis of ΔCAD as a function of the fillet zones [Figure 8(d)] reveals that zones 3 and 4, corresponding to the upper part of the fillet, are the most affected by the loss of dimensional accuracy during batch production.

Overall, the batch production results demonstrate that, although dimensional deviations increase with the number of forming cycles, their magnitude remains limited and does not compromise the functional validity of the cups up to 99 cycles. The observed behaviour suggests the presence of an initial stabilization phase followed by a steady-state regime, in which the punch maintains consistent performance. Among the investigated parameters, the cup radius and ΔCAD are the most sensitive indicators of tool deformation and wear, particularly in the upper fillet zones, whereas the drawing depth exhibits only minor and non-critical variations. These findings provide valuable insight into the progressive deformation and abrasion mechanisms affecting polymer-based punches under repeated forming conditions.

The discussion of the experimental results highlights the combined role of deformation and wear mechanisms in determining the performance of polymer-based punches during repeated deep drawing operations.

As shown in Figures 7 and 8, the cups produced progressively exhibited reductions in radius and accuracy of the fillet profile, while drawing depth remained largely stable and within process tolerances. These findings suggest that the punch undergoes a combination of axial compression and surface abrasion, which are the primary drivers of dimensional variation over multiple cycles. The presence of an initial stabilization phase, followed by a steady-state behavior after approximately 25 strokes, further supports the hypothesis of an early elastic–plastic adjustment of the punch before wear mechanisms become dominant.

Figure 9 provides confirmation of these trends through punch measurements. The interval plots indicate an overall reduction in punch radius with increasing production volume, consistent with the observed decrease in cup radius [Figure 9(a)]. Although a slight increase in the mean punch radius is visible at P05, the Tukey post hoc analysis shows that P01 and P05 belong to the same statistical group, indicating that this variation is not statistically significant and should be attributed to normal experimental dispersion rather than to a physical expansion mechanism. The statistically relevant reduction becomes evident only at higher production levels, particularly at P99. At the same time, punch height decreases significantly within the first 25 cycles and then stabilizes, demonstrating the effect of initial flattening under load [Figure 9(b)]. This behavior corresponds to the early deviations observed in cup geometry, after which the forming process proceeds more consistently.

The analysis of the punch fillet profile offers further insight into the mechanisms involved. The reduction of the fillet radius with production volume mirrors the increase in relative ΔCAD observed in the cups, indicating a progressive rounding caused by abrasive wear [Figure 9(c)]. However, the comparison between the fillet radius of the 99th cup and that of the corresponding punch [Figure 9(d)] reveals a discrepancy: the cup exhibits a larger variation than the punch itself. This suggests that part of the deformation is elastic and occurs during the forming process, with the punch temporarily deflecting under load and partially recovering after unloading.

Overall, the discussion of Figures 7–9 demonstrates that the dimensional deviations of the cups are directly linked to the progressive deformation and wear of the punches. While both axial compression and abrasion contribute to the observed trends, the evidence of elastic deformation highlights the complex mechanical response of additively manufactured polymer tools, where porosity and reduced stiffness compared to metals result in time-dependent behavior under cyclic loading. From an application standpoint, although deviations accumulate over 99 cycles, their magnitude remains within acceptable limits, confirming the suitability of these tools for prototyping and small-batch production.

In summary, the combined analysis of cups and punches demonstrates that dimensional deviations originate from the interplay of axial compression, abrasive wear and elastic deflection. These mechanisms act with different intensities throughout the production cycle, leading to an initial stabilization followed by a steady-state regime. Despite these progressive changes, the dimensional accuracy of the cups remained within acceptable tolerances, confirming that polymer punches manufactured by material extrusion can provide reliable performance for prototyping and small-batch deep drawing applications.

From an industrial perspective, the present results suggest that infill-optimized polymer punches can be effectively used for prototyping and small-batch production under moderate forming loads. For harder alloys, thicker sheets or more complex geometries, higher forming forces and more pronounced stress concentrations are expected, which may accelerate axial compression and abrasive wear. In such cases, increased infill density, alternative reinforcements or locally reinforced architectures may be required to maintain dimensional stability. Therefore, while the identified trends are expected to remain qualitatively valid, quantitative performance limits will depend on the specific forming severity.

This study investigated the performance of lightweight polymer punches manufactured by material extrusion for aluminum deep drawing, with particular focus on the influence of infill architecture on deformation and wear behavior. The results demonstrated that the internal structure of the punch plays a critical role in determining its mechanical response. In particular, the triangular infill configuration ensured higher dimensional stability compared to the gyroid pattern, highlighting the importance of load path organization within the tool. Increasing infill density led to improved stiffness and reduced geometric deviations; however, the triangular configuration at 35% density provided performance comparable to 50% density while offering advantages in terms of reduced material consumption and shorter printing time.

The batch production experiments, conducted up to 99 forming cycles, showed that dimensional deviations remain limited throughout the process. An initial stabilization phase was observed, followed by a steady-state regime in which the punch maintained consistent performance. The dominant degradation mechanisms were identified as axial compression, elastic deflection and abrasive wear, with cup radius and relative ΔCAD emerging as the most sensitive indicators of tool degradation.

Despite these promising results, the study is limited to cylindrical cups in Al1050 and to two infill configurations tested under controlled conditions. Moreover, forming forces were not directly measured, and deformation mechanisms were therefore inferred from the evolution of geometric parameters. Future work will focus on extending the analysis to different materials, geometries and reinforcement strategies, as well as integrating load monitoring and numerical–experimental approaches to better correlate internal stress states with progressive tool degradation.