Additive Manufacturing of Carbon Fiber Cores for Sandwich Structures: Optimization of Infill Patterns and Fiber Orientation for Improved Impact Resistance

Abstract

Carbon fiber-reinforced composites (CFRCs) are widely used in aerospace, automotive, and defense applications due to their high strength-to-weight ratio and excellent mechanical performance. In this study, cores and sandwich panels were fabricated via fused filament fabrication (FFF) using co-polyester filaments reinforced with 20 wt.% short carbon fibers. The mechanical response of the structures was evaluated under low-velocity impact (LVI) conditions using instrumented drop weight testing at energy levels ranging from 2 to 20 J. A three-factor, three-level full factorial experimental design was employed, considering build orientation (flat vs. upright), infill pattern (trihexagonal vs. triangular), and impact energy as factors. The maximum contact force was selected as the primary response variable. The results revealed that upright-printed specimens exhibited significantly improved impact resistance compared to flat-printed ones, with increases in peak force of up to 28% for cores and over 68% for sandwich structures. Among the tested infill geometries, the triangular pattern outperformed the trihexagonal one across all configurations and energy levels. The combination of upright orientation and triangular infill proved to be the most effective, providing enhanced energy absorption and reduced rear-side damage, especially under higher impact energies. These findings offer valuable insights into the design of lightweight, impact-resistant structures produced by additive manufacturing, with direct implications for structural components in demanding engineering environments.

1. Introduction

The development of additive manufacturing (AM), commonly known as 3D printing, is a game changer, revolutionizing the aerospace [1], defense [2], and other engineering industries such as robotics, biomedical, electronics, textiles, and sports [3] by enabling the fabrication of high-performance, lightweight components. The overall performance, especially the surface finish and mechanical properties of the 3D-printed part, are profoundly influenced by the layer and fiber orientation [4], infill density [5,6] of the core, and other process parameters, such as the cooling rate, extrusion temperature, print speed, and nozzle temperature [7].

The continuous pursuit of lightweight and high-performance structures remains a critical challenge in numerous industrial sectors, and sandwich structures, composed of rigid outer face sheets and a lightweight core, offer an exceptional strength-to-weight ratio and superior impact energy absorption capabilities, making them highly desirable for these demanding applications [8,9]. The advent of AM, particularly fused filament fabrication (FFF), has revolutionized the design and fabrication of such structures, enabling the creation of complex core geometries with unprecedented design freedom and customization that are difficult to achieve with conventional manufacturing methods [9,10,11,12,13,14].

This study advances the state of the art by systematically investigating the low-velocity impact response of novel 3D-printed sandwich structures incorporating triangular and trihexagonal cellular core patterns, fabricated using a co-polyester filament reinforced with 20 wt.% short carbon fibers. While previous research has explored various cellular core topologies, such as honeycomb [14,15], auxetic [8,9,12], gyroid [10,13], or truss structures [9,16], our work introduces and investigates, using statistical tools such as Design of Experiments (DoE), specific infill patterns, like triangular and trihexagonal, to expand the space for optimized energy absorption.

Polyamide 12 reinforced with carbon fibers (Nylon 12CF) was used to create a re-entrant hexagonal honeycomb sandwich assembly [17]. The Nylon 12 (unreinforced) and Nylon 12CF (reinforced) samples were fabricated through AM and were tested for their quasi-static behavior. The results demonstrated that the negative Poisson’s ratio structure reinforced with CF exhibited improved failure stress and energy absorption efficiency compared to the unreinforced structure. The results from the finite element simulations were found to be consistent with the experimental outcomes.

The beetle’s forewing trabecular structure inspired the implementation of the study [18]. The core material was responsible for the rate dependence observed in the force–displacement graphs.

It is important to cite another study inspired by the same structure, which focuses on the effect of various design parameters on the impact performance of fully 3D-printed sandwich panels [19]. In that study, an asymmetric panel, with a thicker face sheet at the back and a variating cell wall thickness core, showed maximum impact energy dissipation.

In the study by Guerra Silva et al. [20], a honeycomb structure was fabricated completely using an FFF printer, and the carbon fiber-reinforced sandwich panels exhibited higher stiffness than those reinforced with glass fiber. As predicted by the theoretical models, a higher fiber content led to higher values of flexural modulus.

Arslan et al. [21] characterized the flexural properties of sandwich panels fabricated using filament extrusion additive manufacturing technology with composites of polyphenylene sulfide (PPS) and polyetherimide (PEI). A digital image correlation setup was used to evaluate crack propagation and full-field strain during the flexural test, revealing crack initiation due to strain concentrations in the core region.

The response of materials to the dynamic application of load can be investigated by referring to different tests, including Charpy [22], Izod [23], drop weight [24,25], high-speed ballistic test [26], and explosion/shock wave tests [27].

A central and distinctive element of our research lies in its comprehensive investigation into the effect of printing orientation (upright versus flat) on the impact performance of these 3D-printed cellular sandwich structures. Anisotropy is a well-known characteristic of 3D-printed parts produced via FFF, especially when using composite materials, where fiber alignment along the extrusion direction can significantly influence mechanical properties [28,29,30]. Despite this, the relationship between the printing orientation during material characterization and its direct effect on accurate modeling of the impact response of 3D-printed cellular structures is not yet fully understood, and many studies tend to use a single printing orientation for their main structures [30]. It is crucial that recent works, such as that by Fisher et al. [30], have highlighted that material models derived from default printing orientations (e.g., ±45°) can under-predict maximum impact forces and over-predict impact duration, potentially leading to an underestimation of impact severity and greater damage in real-world applications.

Our research directly addresses this gap by exploring the impact behavior across different printing orientations for the sandwich core itself, moving beyond material characterization to the structural level, as increasingly required in real-world 3D-printed composite sandwich applications such as energy-absorbing liners for bicycle and motorcycle helmets [31,32], lattice-reinforced crash boxes in automotive structures [33], and lightweight protective shells in civil and military applications [34].

In summary, this work provides novel insights into the mechanical behavior of 3D-printed sandwich structures with specifically designed cellular cores under low-velocity impact, with a unique focus on the critical role of printing orientation.

2. Materials and Methods

2.1. Materials

ColorFabb XT-CF20 is a high-performance Eastman Amphora AM1800 co-polyester-based 3D printing filament filled with short carbon fibers [35]. It is basically composed of co-polyester matrix and 20 wt.% carbon fibers. The polymer matrix acts as a binder for the reinforcing carbon fibers, offers high toughness, thermal resistance, and flexibility, and is perfect for components which demand high stiffness.

The skins of the sandwich structures were fabricated using HexPly^® M79 carbon fiber prepreg (Hexcel Corporation, Stamford, CT, USA), specifically designed for out-of-autoclave processing. The material consists of a 210 gsm 3K carbon fiber fabric in a 2 × 2 twill weave, pre-impregnated with a low temperature curing epoxy resin matrix. Curing was performed in an ECP Industrial OVEN301-NL (Easy Composites, Stoke-on-Trent, UK) following the manufacturer’s recommended low-temperature cure cycle: a heating rate of 5 °C/min up to 90 °C, followed by an isothermal dwell of 300 min at 90 °C.

2.2. Methods

The specimens for the impact tests were fabricated into two separate categories: cores alone and sandwich panels, using FFF with the Ultimaker S5 3D printer (Utrecht, The Netherlands), with dimensions of 70 × 70 × 10 mm³ for cores and 70 × 70 × 12 mm³ for sandwiches.

The 3D printing parameters considered for the present work are summarized in

Table 1. Printing parameters for the manufacturing of the specimens.

Parameters	Values
Bed temperature	70 °C
Layer height	0.15 mm
Nozzle diameter	0.4 mm
Print speed	40–70 mm/s
Temperature of nozzle	250 °C
Infill density	20%
Infill pattern	Triangular/Trihexagonal

.

The 3D printing orientations of the core were varied as ‘upright’ and ‘flat’. To study the effect of infill pattern on the impact resistance, ‘trihexagonal’ and ‘triangular’ designs were analyzed. The trihexagonal infill pattern combines triangles and hexagons, generating a 2D mesh that provides balanced material efficiency and strength (Figure 1). On the other hand, the triangular infill pattern consists of a web of equilateral triangles, known for their ability to distribute loads evenly and their integral rigidity. A 20% moderate infill density confirmed a balance between strength and weight reduction, allowing a comparative study of the impact responses under different loading situations. The fabricated structures and their corresponding labels are listed in

Table 2. Details of the fabricated specimens for the drop weight test.

Specimen Type	Material	Orientation	Infill Pattern	Specimen ID
Core	XT-CF20	Upright	Trihexagonal	U-TH
Core	XT-CF20	Flat	Trihexagonal	F-TH
Core	XT-CF20	Upright	Triangular	U-T
Core	XT-CF20	Flat	Triangular	F-T
Sandwich	XT-CF20	Upright	Trihexagonal	SU-TH
Sandwich	XT-CF20	Flat	Trihexagonal	SF-TH
Sandwich	XT-CF20	Upright	Triangular	SU-T
Sandwich	XT-CF20	Flat	Triangular	SF-T

. The printed parts and manufactured sandwiches are shown in Figure 2.

The dynamic impact characterizations were carried out using the Instron/Ceast 9340 instrument tower (Figure 3). The tests were conducted at room temperature, utilizing a mass of 3.055 kg and three different energy levels (2, 5, and 10 J for the cores only, and 2, 10, and 20 J for the sandwiches) by varying the release height of the striker. The impactor head used for the study was hemispherical with a diameter of 12.7 mm. Preliminary testing indicated that an impact energy of 10 J was sufficient to induce perforation in the cores alone. This value was therefore established as the upper limit, with two additional energy levels selected to ensure substantial separation between test conditions while maintaining an impact velocity above 1 m/s. For the sandwich configurations, the same minimum and maximum energies defined for the cores were retained; however, the upper limit was increased to 20 J to account for their greater structural resistance. All tests were performed in accordance with the ASTM 5420 standard.

Design of Experiments (DoE)

To investigate the impact behavior, expressed in terms of peak force, for both the cores and the sandwiches, two replicated general factorial designs were considered, respectively. In detail, for the cores, the factors investigated (independent variables) in the experimental design were:

• Orientation (factor A)—Categorical factor varied at two levels (a = 2) corresponding to {upright, flat};
• Infill Pattern (factor B)—Categorical factor varied at two levels (b = 2) corresponding to {trihexagonal, triangular};
• Energy (factor C)—Numerical factor varied at three levels (c = 3) corresponding to {2, 5, 10} J.

While, for the sandwiches, the factors investigated (independent variables) in the experimental design were:

• Orientation (factor A)—Categorical factor varied at two levels (a = 2) corresponding to {upright, flat};
• Infill Pattern (factor B)—Categorical factor varied at two levels (b = 2) corresponding to {trihexagonal, triangular};
• Energy (factor C)—Numerical factor varied at three levels (c = 3) corresponding to {2, 10, 20} J.

The two experimental plans are summarized in

Table 3. Experimental plan for cores characterization: factors and levels.

Factor	Symbol	Type	Unit	Levels	Low Level (−1)	Center Level (0)	High Level (+1)
Orientation	A	Categorical	[-]	a = 2	Upright	-	Flat
Infill Pattern	B	Categorical	[-]	b = 2	Trihexagonal	-	Triangular
Energy	C	Numerical	[J]	c = 3	2	5	10

and

Table 4. Experimental plan for sandwiches characterization: factors and levels.

Factor	Symbol	Type	Unit	Levels	Low Level (−1)	Center Level (0)	High Level (+1)
Orientation	A	Categorical	[-]	a = 2	Upright	-	Flat
Infill Pattern	B	Categorical	[-]	b = 2	Trihexagonal	-	Triangular
Energy	C	Numerical	[J]	c = 3	2	10	20

for the cores and sandwiches, respectively. Furthermore, the statistical significance for the investigated factors and their possible interaction were examined through an analysis of variance (ANOVA) as soon as the values for the investigated response were collected for each experimental scenario.

For both experimental designs, the number of replications was set to n = 3, resulting in a total of N = a × b × c × n = 36 experimental runs.

To determine the most effective combination of the factors investigated on the response variable, i.e., the peak force, an optimization process was conducted using the desirability function method. This technique involved converting each response y_i into a corresponding desirability function d_i, which ranged between 0 and 1. A desirability value of d_i = 1 indicated that the response met its desired target, whereas d_i = 0 meant that the response fell outside the acceptable limits. The goal was to adjust the design parameters in a way that maximized the overall desirability. For this analysis, specific constraints were defined to optimize the values of peak force, while the factors were varied within the range of the experimental design, covering all examined levels. A summary of the constraints used in the optimization process is provided in

Table 5. Set constraints for the optimization process based on the desirability functions.

Parameter	Goal	Lower Limit	Upper Limit	Lower Weight	Upper Weight	Importance
Orientation	In range	Upright	Flat	1	1	3
Infill Pattern	In range	Trihexagonal	Triangular	1	1	3
Energy	In range	2	20	1	1	3
Peak Force	Maximize	Lowest value collected	Highest value collected	1	1	3

.

3. Results and Discussion

3.1. Impact Performance of Core Specimens

3.1.1. Experimental LVI Test Results for Core Specimens

The force–time and force–displacement curves for the 3D-printed cores as a function of infill pattern and impact energy are shown in Figure 4, Figure 5, Figure 6 and Figure 7.

The curves show the presence of extensive damage due to impact with energies greater than 2 J, as evidenced by the long contact time and a marked drop in the load after reaching the maximum load. The damage degree (ratio between absorbed energy and impact energy) was in fact found to be close to unity already at 5 J, while complete perforation was reached at 10 J, as evidenced by the open force–displacement curves. The two printing configurations, flat and upright, showed different behaviors with the same type of infill: the upright configuration was the best in terms of impact resistance (lower damage degree and better maximum contact force) than the flat one. As far as the type of infill is concerned, the architecture based on “triangles” showed a higher maximum contact force than that based on “trihexagons.”

Peak force (Fmax) is the extreme force the test specimen can withstand before a maximum deformation or failure. The shorter the time to peak force, the more brittle the failure. As evident from Figure 8, all four tested specimens (F-T, F-TH, U-T, and U-TH) exhibited a progressive response as the impact energy increased from 2 J to 10 J, i.e., the peak force increased with increasing impact energy. This behavior indicated an increase in impact resistance and reflected the material’s strength.

The 3D-printed core specimens with an upright orientation performed well compared to the flat ones, showing only slight deviations under higher impact energy (10 J). Various studies have found that the upright orientation is superior and highly effective for short CF-reinforced additive manufacturing. As an example, Becerra et al. [36], studied the effect of build orientation on 3D printing using short carbon fiber-filled nylon. Upright orientations showed good agreement and accuracy with theoretical 3D printing models.

The triangular infill pattern was found to be superior to the trihexagonal pattern. A similar positive outcome for the triangular infill pattern was reported by peer researchers who conducted extensive studies on the crashworthiness of 3D-printed parts [37]. Another recent study on the dynamic and mechanical performance of 3D-printed continuous carbon fiber-based onyx composites also proves the prominent role of the triangular infill pattern under tensile and drop weight impact loadings [38].

The aforementioned two studies strongly support the capability of the triangular infill pattern considered in the present work.

But the response of the ‘flat’ oriented cores is noteworthy. Although they did not exhibit a sudden spike in peak loads (as observed in the upright specimens), they maintained their performance progressively up to higher impact energy levels (10 J). This trend was evident in both the ‘flat-triangular (F-T)’ and ‘flat-trihexagonal (F-TH)’ specimens.

3.1.2. DoE-Based Analysis and Optimization for Core Specimens

These findings for the tested cores were also confirmed by the effects diagrams reported in Figure 9, Figure 10 and Figure 11 for the interactions AB, AC, and BC, respectively. In detail, it is worth noting that, while at 2 J all of the tested configurations exhibited comparable behaviors in terms of the investigated response (i.e., peak force), as the impact energy (factor C) increased, the Upright configuration demonstrated higher impact resistance when combined with the triangular infill pattern compared to the trihexagonal pattern. Conversely, when the cores were fabricated in the flat configuration, they all behaved similarly regardless of the type of infill pattern used. Furthermore, from the interaction BC effects diagram (see Figure 11), greater variability in the investigated response can be observed for the {flat; trihexagonal} configuration, particularly at impact energies of 5 or 10 J.

These results were also analyzed using ANOVA, and the findings are summarized in

Table 6. ANOVA table for the impact tests of the cores (investigated response: peak force).

Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	1.612E6	11	1.466E5	34.42	<0.0001	Significant
A—3D Printing Orientation	3.033E5	1	3.033E5	71.22	<0.0001	Significant
B—Infill Pattern	3.094E5	1	3.049E5	71.59	<0.0001	Significant
C—Impact Energy	4.271E5	2	2.135E5	50.14	<0.0001	Significant
AB	1.490E5	1	1.490E5	34.99	<0.0001	Significant
AC	1.185E5	2	59247.11	13.91	<0.0001	Significant
BC	1.619E5	2	80933.89	19.00	<0.0001	Significant
ABC	1.476E5	2	73820.42	17.33	<0.0001	Significant
Pure Error	1.022E5	24	4259.01
Cor Total	1.715E6	35
Mean	65.26	R2	0.9404
St. Dev	1255.78	Adj-R2	0.9131

.

Indeed, according to the ANOVA results, all of the investigated factors, i.e., A, B, and C, were influential (p-value < 0.0001) for the investigated response. Furthermore, all the interactions AB, AC, BC, and ABC were statistically significant (p-value < 0.0001). In addition, no anomalies were observed from the model adequacy check on the residuals. In the end, the high R² value of 0.9404 confirmed that most of the variability observed in the investigated response, peak force, was due to the variations in the aforementioned significant factors and their interactions.

Figure 12, Figure 13, Figure 14 and Figure 15 illustrate the development of damage on both surfaces (front and rear) of the 3D-printed cores as the impact energy increases.

As mentioned above, the improved impact resistance of the upright configuration was evident, particularly with a triangular-type infill. In general, the trihexagonal infill could reduce the extent of damage even in the flat configuration by absorbing the impact energy and limiting the delamination of the top layer on the surface opposite the point of impact. This was particularly evident in the flat-trihexagonal core. Damage on the non-impacted side was not present up to 5 J for upright configurations. In the case of the flat printing orientation, only a bulge could be noticed at the non-impacted side up to 5 J with the trihexagonal infill.

The sandwich structures, fabricated in two different configurations, flat and upright, were impacted at three different energy levels (2, 10, and 20 J), highlighting the differences in behavior attributable to the core lattice structures.

3.2. Impact Performance of Sandwich Panels

3.2.1. Experimental LVI Test Results for Sandwich Panels

The force–time and force–displacement curves for the two configurations as a function of infill pattern and impact energy are shown in Figure 16, Figure 17, Figure 18 and Figure 19.

Figure 20 summarizes the peak force of the different configurations, while Figure 21 illustrates the trends in maximum contact force, comparing different configurations at the lowest and highest tested energy levels (2 and 20 J).

3.2.2. DoE-Based Analysis and Optimization for Sandwich Panels

The results from the sandwich structures confirmed the same behavior observed in the case of the 3D-printed cores without sandwiching. These results were also consistent, once again, with the ANOVA analysis carried out for the sandwiches. The results obtained are summarized in

Table 7. ANOVA table for the impact tests of the sandwiches (investigated response: Peak Force).

Source	Sum of Squares	Df	Mean Square	F-Value	p-Value
Model	1.732E7	11	1.575E6	175.27	<0.0001	Significant
A—3D Printing Orientation	4.845E6	1	4.845E6	539.18	<0.0001	Significant
B—Infill Pattern	1.216E6	1	1.216E6	135.39	<0.0001	Significant
C—Impact Energy	9.058E6	2	4.529E6	504.04	<0.0001	Significant
AB	3.507E5	1	3.507E5	39.04	<0.0001	Significant
AC	1.643E6	2	8.217E5	91.45	<0.0001	Significant
BC	1.231E5	2	61570.40	6.85	0.0044	Significant
ABC	86740.77	2	43370.38	4.83	0.0173	Significant
Pure Error	2.156E5	24	8985.09
Cor Total	1.754E7	35
Mean	107.84	R2	0.9828
St. Dev	2335.84	Adj-R2	0.9768

.

Even in this case, the factors A, B, and C were statistically significant (p-value < 0.0001) for the investigated response. Furthermore, all the interactions AB, AC, BC, and ABC were also influential (p-value < 0.0001). Once again, no anomalies were detected from the model adequacy check on the residuals. Eventually, even in this case, most of the variability observed in the investigated response, peak force, could be attributed to the changes in the significant factors and interactions, as R² = 0.9828 was very high.

The effects diagrams for the investigated response, i.e., peak force, from the impact tests performed on the sandwiches are reported in Figure 22, Figure 23 and Figure 24. This confirmed the same trend described for the cores.

In the end, Figure 25 and Figure 26 present the results obtained from the optimization process, which was performed using desirability functions with the purpose of identifying the best performing combination of the investigated factors. It turned out to be {upright; triangular} for both cores and sandwiches. The only exception was for the cores tested at 2 J. Indeed, in this case, the best performing solution was the {upright; trihegaxonal} one. However, it must be highlighted that the desirability function value was quite similar to the {upright; triangular} configuration. This result was consistent with the previously discussed findings, i.e., that at 2 J, all of the tested core configurations exhibited comparable behaviors in terms of peak force (see Figure 9a).

For all configurations, an impact energy of 2 J was insufficient to cause appreciable damage, as evidenced by the symmetrical force–time curves without load drops and by the observations of the impact surfaces (Figure 16, Figure 17, Figure 18 and Figure 19). The upright printing orientation exhibited superior linear stiffness and peak force compared to the flat orientation (Figure 16 and

Table 8. Average values and standard deviations of linear stiffness for impacted cores and sandwich structures.

Specimen ID	Linear Stiffness (N/mm)
Core
F-T_2J	1018.4 ± 9.8
F-T_5J	1094.2 ± 48.9
F-T_10J	1103.4 ± 3.7
F-TH_2J	881.4 ± 76.7
F-TH_5J	820.9 ± 90.9
F-TH_10J	586.9 ± 43.8
U-T_2J	1588.2 ± 8.4
U-T_5J	1613.5 ± 31.8
U-T_10J	1626.2 ± 54.1
U-TH_2J	1253.8 ± 72.2
U-TH_5J	1357.4 ± 88.7
U-TH_10J	1304.5 ± 83.2
Sandwich
F-T_2J	1148.5 ± 54.7
F-T_10J	1221.4 ± 66.0
F-T_20J	1015.2 ± 60.6
F-TH_2J	1169.3 ± 28.8
F-TH_10J	1025.3 ± 20.7
F-TH_20J	832.5 ± 63.3
U-T_2J	1828.1 ± 54.2
U-T_10J	1371.8 ± 65.4
U-T_20J	1386.0 ± 58.1
U-TH_2J	1750.4 ± 17.5
U-TH_10J	1360.5 ± 41.0
U-TH_20J	1318.9 ± 35.4

). As the impact energy increased, the difference in stiffness became less pronounced due to the onset of progressive damage on the impacted surface (Figure 27, Figure 28, Figure 29 and Figure 30). This held true for both the cores and the sandwich structures. With the exception of the F-TH core, for which a significant decrease in linear stiffness (defined as the average slope of the impact curve until the first load drop takes place [39]) was observed with increasing impact energy, the other cores did not exhibit a marked variation. This was indicative of superior resistance to impact damage, despite failure of the top face sheet at impact energies of at least 5 J (Figure 12, Figure 13, Figure 14 and Figure 15). This trend was also evident in Figure 5, where several peaks were visible already at 5 J, denoting progressive core damage with alternating load drops and recoveries. It is further worth noting that both flat configurations showed the presence of debonding between the core and the bottom face sheet on the surface opposite to the impact, which was more marked in F-TH.

Conversely, the upright configurations displayed better impact resistance, also in terms of linear stiffness, with the triangular infill offering the best performance, without progressive degradation as a function of impact energy. The trihexagonal infill, on the other hand, exhibited a force–displacement curve (Figure 7) that already at 5 J presented a double peak, an indication of progressive core cracking and reduced impact resistance. These behaviors were also observed in the sandwich structures, highlighting, however, that a large fraction of the impact energy during low-velocity impact was absorbed by fracture of the top face sheet, as no damage was observed on the back face sheet up to an impact energy of 20 J, especially in the flat configuration. This behavior explains the marked reduction, already at 10 J, in linear stiffness for the sandwich structures (Table 8). This was also evident from the different shapes of the force–displacement curves at 20 J (Figure 21b), where only the flat configurations, for both infill types, exhibited two peaks, suggesting contact of the impactor with the top and bottom face sheets, respectively—a feature not recorded for sandwich structures with the upright configuration [40]. Another noteworthy observation is that the second peak had a lower magnitude than the first. The deformation characteristics of the 3D-printed cores under compressive loading were likely to be the main cause of this behavior, which occurred in the form of core cracking/breaking rather than cell collapse, typically observed in cork cores compared to synthetic foams [41]. In this case, at higher strains, the core structure exhibits reduced energy absorption due to the breaking of individual cells, with densification not taking place, thus failing to resist further penetration of the impactor [42]. The fracture of the face sheet is governed by the extent of localized bending that can be accommodated due to indentation-induced deformation [43]. It is therefore reasonable to assume that the different energies required to break the core allowed tailoring of the damage extent in the core material, which was found to be lower in the upright configuration (Figure 29 and Figure 30). This damage was more extensive in the flat configuration and led to increased energy absorption and a higher degree of damage. This energy absorption in both upright configurations (triangular and trihexagonal) was limited to the upper skin, affecting the core at 20 J and causing delamination of the rear skin opposite the impact.

The damage in flat configurations was greater and affected the core, even at an impact energy of 10 J. This contrasted with the behavior observed in ‘upright’ configurations, highlighting a key factor in the mechanical response related to the 3D printing orientation. The impact energy was absorbed through brittle failure of both the fibers and the matrix. This damage was most evident and was localized around the contact area with the hemispherical drop weight. It is important to note that the best configuration was the upright orientation combined with a triangular infill.

The comparative analysis of data from low-velocity impact tests remains inherently challenging due to the wide range of variables influencing the measured response. As reported in several studies [44,45,46,47], these variables include the size, shape, and mass of the impactor; the boundary and support conditions of the test specimens; the geometric configuration of the sandwich panels—particularly the core and face sheet thicknesses—and, most importantly, the material systems employed in their manufacture [42,48]. Consequently, direct comparison between experimental campaigns is often limited unless these parameters are strictly controlled or properly normalized. Despite these challenges, a commonly employed parameter for comparing the energy absorption capacity of different structures is the specific absorbed energy (SAE), defined as the ratio of the total energy absorbed to the mass of the sandwich panel [49]. In this study, the most pronounced weight differences were observed between cores printed in upright and flat orientations, independent of the infill pattern. The upright configuration consistently exhibited a higher mass than its flat counterpart. Specific energy absorption (SEA) values were evaluated for both cores—excluding the 10 J energy level due to penetration—and for the corresponding sandwich structures, with the 20 J energy level similarly omitted.

At the lower impact energy (2 J), cores displayed no statistically significant differences, with SEA values ranging from 50 to 55 J/kg. At 10 J, however, the upright cores exhibited lower SEA values than the flat cores across both infill types, as summarized below:

• U-TH = 146 ± 2 J/kg, F-TH = 174 ± 1 J/kg
• U-T = 132 ± 1 J/kg, F-T = 171 ± 1 J/kg

Since the laminated face sheets were identical across sandwich specimens, these trends were preserved at the sandwich level, with SEA values substantially higher at 10 J:

• U-TH = 262 ± 1 J/kg, F-TH = 317 ± 3 J/kg
• U-T = 229 ± 5 J/kg, F-T = 313 ± 2 J/kg

These findings indicated that the core printing orientation exerts a stronger influence on energy absorption than the infill pattern. Nevertheless, the trihexagonal infill configuration achieved higher SEA values compared to the triangular pattern.

Despite the challenges in direct comparisons, these values aligned with literature reports. Indres et al. [50] fabricated three PLA-based sandwich panels using fused deposition modeling, one with a regular hexagonal core and two with re-entrant cores at 0° and 90°, and subjected them to low-velocity impacts of 10 and 15 J. The SEA values at 10 J ranged from 27 to 37 J/kg depending on the core type.

Similarly, Karami et al. [51] developed a cost-effective method for integrated PLA sandwich panels with glass fiber-reinforced face sheets. Under 18 J impact, the SEA values ranged from 228 to 601 J/kg, with the highest values observed in the integrated structure without adhesive between the core and face sheet.

4. Conclusions

This study systematically investigated the influence of printing orientation and infill pattern on the impact resistance of CFRP cores and sandwich panels fabricated via FFF 3D printing. The quantitative results clearly demonstrated that both orientation and internal geometry significantly affect performance under low-velocity impact. For the core samples, specimens printed in an upright orientation consistently performed better than the flat ones. At 10 J impact energy, the upright-triangular (U-T) configuration exhibited a peak force of 1711.3 N ± 29.1, compared to 1335.9 N ± 47.0 for flat-triangular (F-T) and 1136.7 N ± 52.1 for flat-trihexagonal (F-TH). This corresponded to relative improvements of approximately 28% over F-T and 50% over F-TH. The triangular infill also demonstrated superior performance across all orientations; for instance, at 5 J, the U-T cores reached 1666.7 N ± 56.2, whereas the U-TH configuration reached only 1258.5 N ± 133.6. These differences are attributed to the more efficient load distribution and rigidity inherent to triangular architectures.

The same trends were confirmed in the sandwich structures. At 20 J, the upright-triangular sandwich (SU-T) achieved a peak force of 3895.0 N ± 98.4, significantly outperforming the flat-triangular (SF-T, 2316.7 N ± 33.9) and flat-trihexagonal (SF-TH, 2166.8 N ± 70.4) specimens. Even at lower energies, the advantage of upright-triangular designs was evident: at 2 J, SU-T reached 1913.1 N ± 90.3, compared to 1621.8 N ± 26.4 for SF-T and 1494.8 N ± 27.8 for SF-TH. On average, the upright-triangular sandwich structure exhibited up to 68% higher peak force than the worst-performing flat-trihexagonal counterpart at the highest energy level tested.

Statistical analysis through ANOVA confirmed the high significance (p < 0.0001) of all main factors (orientation, infill pattern, and impact energy) as well as their interactions, with R² values of 0.94 for core samples and 0.98 for sandwich panels, indicating excellent model accuracy. The optimization via desirability functions further identified the {upright; triangular} combination as the most effective configuration in terms of impact resistance across most energy levels.

These findings offer strong, data-driven guidance for designing lightweight, AM structures capable of resisting LVI. They highlight the critical importance of combining optimal internal architectures with strategic build orientations to exploit the full potential of CF-reinforced FFF composites.

Future research should explore fatigue performance, the potential of hybrid infill architectures, and the long-term environmental durability of the printed components.