Analysis of Microstructural and Wear Mechanisms for 3D-Printed PET CF15 Using Box–Behnken Design

Abstract

We examined the impact of 3D-printing parameters, such as the deposition pattern, deposition speed, and layer height, on the tribological performance measured through the coefficient of friction and cumulative linear wear. Optimizing these factors can significantly influence material wear and friction, which is critical for ensuring durability and functionality in practical applications like a cylindrical gear assembly for a vertical-articulated robot. The purpose of the study was to investigate these relationships by employing the Box–Behnken design (BBD) method to systematically analyze the effects of these parameters, while also using scanning electron microscopy (SEM) for detailed microstructural characterization. The findings aim to provide insights that can guide the development of more efficient and wear-resistant 3D-printed materials. The strong impact of layer height on CLW was noted, showing that lower layer heights can either improve or worsen wear depending on the combination of speed and pattern, with layer height playing a dominant role in determining wear performance. Lower speeds and specific patterns, particularly lines and concentric patterns, tend to result in higher COF values. The validation test results, with a COF of 0.2215 and CLW of 29.2075, closely align with the predicted values of 0.2064 and 27.3, showing small percentage errors of 7.3% for COF and 6.5% for CLW.

1. Introduction

1.1. Literature Review

Additive manufacturing (AM), often referred to as 3D printing, has revolutionized manufacturing by enabling the production of complex, custom parts with significant reductions in material waste and production time. Among the various materials used in AM, Polyethylene Terephthalate (PET) has gained attention due to its excellent chemical resistance, mechanical strength, and ease of recycling [1,2]. However, for applications demanding higher performance, such as automotive, aerospace, and industrial sectors, the mechanical properties of PET often require enhancement. One effective approach for achieving this is through the incorporation of carbon fibers (CF), which improve the composite’s strength, stiffness, and wear resistance [3,4,5,6].

The successful utilization of this composite in AM processes, particularly FFF (Fused Filament Fabrication), requires careful adaptation of manufacturing parameters. For instance, annealing has been shown to enhance the interlayer tensile strength of carbon fiber-reinforced composites, such as PETG (Polyethylene Terephthalate Glycol-modified) and PLA (Polylactic Acid), by influencing the microstructure and reducing residual stresses [7]. Similarly, the choice of layer height can significantly impact the dimensional accuracy of 3D-printed parts, as demonstrated in studies on various thermoplastics, including PLA, PETG, and Acrylonitrile Butadiene Styrene (ABS) [8,9]. Moreover, the incorporation of carbon fibers has been reported to enhance the surface properties of PETG composites, further improving tribological performance [10]. Infill patterns and raster angles also play an important role in mechanical and surface characteristics, affecting the printed material’s structural integrity and accuracy [9,11]. The mechanical properties and dynamic behavior of PETG composites reinforced with carbon fibers have been studied in [12,13], highlighting the need to optimize printing temperature and speed for enhanced performance.

Microstructural characterization plays a very important role in understanding how these printing parameters impact the distribution and orientation of carbon fibers within the PET matrix. For example, studies demonstrated that the morphology and orientation of reinforcements, such as graphite skeletons, significantly influence wear and friction performance, emphasizing the need for precise control over the printing process [14]. Comparisons of traditional and additive manufacturing techniques for materials like 17–4 PH stainless steel highlight the enhanced wear performance achieved through tailored microstructures in additively manufactured components [15]. Research on carbon fiber-reinforced Polyether Ether Ketone (PEEK) composites showed that the alignment and dispersion of fibers are key factors in optimizing tribological performance [16]. Additionally, investigations into the deflection angles of graphite layers in 3D-printed composites further underscore the importance of microstructural orientation in determining wear behavior [17]. A well-optimized microstructure can enhance the load-bearing capacity of the composite, reduce wear rates, and improve overall durability. Additionally, understanding the wear mechanisms in PET-CF15 composites under different operating conditions is essential for predicting their performance in real-world applications.

PET has become a popular choice in AM due to its ease of processing and good balance of mechanical properties. However, as highlighted in [12], the use of PET in AM is still relatively novel, and further research is needed to fully understand the material’s behavior during and after printing. The mechanical properties of PETG were tested, showing that lower layer thickness, with adjusted feed rate and infill density, improved both tensile and flexural strength. Layer thickness had the greatest impact, contributing the most to mechanical performance.

Carbon fiber-reinforced polymers (CFRPs) are widely recognized for their superior mechanical properties, including high tensile strength and stiffness. A study performed by Shofner et al. [18] demonstrated that the addition of carbon fibers to thermoplastic matrices significantly improves the composite’s mechanical performance, particularly in load-bearing applications.

The influence of varying amounts and lengths of carbon fiber to ABS on the mechanical properties of FDM-fabricated parts, focusing on improvements in tensile and flexural performance was investigated by Ning et al. [19]. The results showed that incorporating carbon fiber improved tensile strength and Young’s modulus but reduced toughness, yield strength, and ductility. The best tensile strength was observed with 5% carbon fiber, while 7.5% provided the highest Young’s modulus. Longer carbon fibers (150 μm) further enhanced tensile strength, but reduced toughness compared to shorter fibers. Flexural properties improved significantly with 5% carbon fiber. However, higher fiber content, especially at 10%, increased porosity and negatively affected overall performance.

Despite the growing interest in 3D printing, there remain a limited number of studies focused on the tribological behavior of 3D-printed parts, particularly in references [20,21,22,23,24,25,26,27]. Research exploring how specific printing variables, such as infill percentage and layer thickness, affect tribological and frictional performance is scarce.

One paper [27] investigated the tribological behavior of 3D-printed PLA parts, focusing on the effects of printing parameters (infill percentage, layer thickness) and post-processing (annealing) by measuring the coefficient of friction, cumulative wear, and surface roughness. The results showed that annealing increased the coefficient of friction, especially for samples with 75% infill and a layer thicknesses of 0.15 mm (+147%) and 0.2 mm (+198.95%). Annealed samples had smoother surfaces, as indicated by lower roughness values. Statistical analysis revealed that layer thickness had the greatest impact on friction, while infill percentage influenced the wear in as-built samples, and layer thickness dominated in annealed samples. Similarly, the research in [26] aimed to evaluate how process parameters affect the tribological and frictional behavior of 3D-printed ABS and PLA parts, using design of experiments (DOE) techniques. The optimization results indicate that the best parameters for minimizing friction and wear are an infill percentage of 50% with a layer thickness of 0.1 mm for ABS, and an infill percentage of 50% with a layer thickness of 0.15 mm for PLA. These findings, based on universal tribometer tests, help identify the optimal printing conditions for reducing material consumption in additive manufacturing. Also, the impact of printing parameters on the tribological and friction behavior of 3D-printed ABS and PLA samples was studied in [28].

A related study [25] primarily investigated the friction characteristics of 3D-printed samples, revealing that the coefficient of friction was consistently higher in the transverse direction than in the longitudinal direction, regardless of applied load or sliding speed. Additionally, a comparison between PLA and ABS showed that PLA samples consistently exhibited lower friction coefficients than ABS, irrespective of printing direction, load, or sliding speed.

Another study [23] explored the effects of scaffolding angle and raster gap on friction behavior, focusing on the coefficient of friction and wear rate. The research introduced graphite flakes into ABS to improve material properties. The results indicated that the scaffolding angle significantly influenced behavior only with a positive gap; with a negative gap, no notable effect was observed. However, incorporating graphite increased the friction coefficient while diminishing wear resistance.

The objective of work performed by Maguluri et al. [21] was to identify optimal 3D printing parameters to minimize the wear rate of PLA, focusing on extrusion temperature, fill density, and nozzle speed. It was concluded that infill percentage had the most significant effect on wear rate, followed by extrusion temperature and nozzle speed. The optimal parameters for minimal wear were an infill percentage of 100%, an extrusion temperature of 220 °C, and a nozzle speed of 40 mm/s.

Perepelkina et al. [29] demonstrated the significant impact of 3D-printing settings on the strength, stiffness, and surface quality, all of which influence the tribological properties of printed parts. Their findings showed that white filament had the highest friction tendency, while black filament printed at a 45° orientation resulted in the greatest wear depth.

It can be observed that in FFF, ABS and PLA are commonly used due to their ease of manufacturing and many studies are focused on their tribological behavior. However, PETG (polyethylene terephthalate glycol) is also known for its excellent tribological performance [6,30,31]. PETG is highly impact and chemical resistant, making it suitable for industrial applications. It provides excellent layer adhesion during printing, resulting in parts with high structural integrity and wear resistance. Additionally, PETG’s recyclability and low moisture absorption offer sustainability and practicality advantages over PLA and ABS, despite the latter two having high strength, as shown in [32,33].

The research in [30] examines the tribological performance of PETG parts coated with specialized 3D-printing materials, produced via FFF technology. It finds that while thicker coatings improve surface stability and reduce roughness, they do not significantly alter the friction coefficient. The study highlights that coating thickness affects surface quality moderately, and applied load influences wear groove depth and width, offering insights for optimizing tribological properties in additive manufacturing. Similarly, the tribological behaviour of fused deposition modelling (FDM) samples made from PETG was examined by Batista et al. [30] and the friction coefficient of PETG was found to be comparable to that of nylon in low-load tests.

Although PETG has not traditionally been regarded as a significant tribopolymer for friction or wear, the study [34] reveals that the FDM-AM process induces oxidation in PETG, which enhances its wear resistance. Additionally, annealing further improves this property. Overall, additively manufactured PETG exhibits favorable wear behavior, making it a viable candidate for bulk tribopolymer applications, particularly when combined with friction-reducing additives.

The BBD is a statistical approach used to evaluate the effects of multiple factors on one or more response variables. It is particularly useful for studying the impact of various process parameters and optimizing conditions in experimental research. The BBD is employed to understand how different input factors affect outcomes and to find the optimal settings that lead to the best performance or quality. It is commonly used in experiments where the goal is to optimize a process, such as in manufacturing or materials science. It involves a set of experiments with combinations of factor levels, including midpoints and extreme values. The BBD does not require the testing of all possible combinations, making it more efficient. The main advantage of this design is that fewer experiments are needed compared to full factorial designs, making it cost-effective and time-efficient. The method is suitable for experiments with three or more factors, and it generates a response surface model that can predict outcomes under different conditions, aiding in process optimization and decision-making. Many scientific works [35,36,37,38,39] related to 3D-printing parameter optimization used this method in the performed investigations.

1.2. Bibliometric Analisis

The present work aims to fill gaps in the literature by analyzing the tribological properties of PET-CF15 composites, focusing on the effects of printing parameters, with an emphasis on analyzing the resulting microstructure and wear mechanisms. By systematically varying key printing parameters (layer height, deposition speed and pattern), we aim to optimize the composite’s performance, providing insights that could lead to broader adoption of PET CF15 in industrial applications such as automotive parts, aerospace components, robotics, etc.

After searching the relevant paper, a cluster analysis was performed using VOSviewer version 1.6.20, as seen in Figure 1 presenting the co-occurrence keyword diagram. In the co-occurrence analysis, we included author keywords, while setting the minimum number of keyword occurrences to five, resulting in a map with 23 thresholds.

This VOSviewer co-occurrence map shows the relationships between key terms related to 3D printing, tribology, and mechanical properties. It can be observed that the term “3D printing” appears as a major node, indicating that it is a core topic of current research. The map connects 3D printing with adjacent terms such as “tribology”, “additive manufacturing”, and “fused deposition modeling”, suggesting significant interest in general mechanical and manufacturing aspects. The network shows terms like “PLA”, “ABS”, and “PEEK” linked to 3D printing and tribology. However, while there is research into the tribological behavior of these polymers, the specific exploration of parameter interactions on wear performance, as conducted in the present study, appears less emphasized. The connection between “wear”, “friction”, “surface roughness”, and “hardness” points to studies involving the mechanical properties of printed parts. Nevertheless, the interactions between deposition parameters and how they influence combined tribological outcomes, such as coefficient of friction and cumulative linear wear, are not strongly highlighted, suggesting an opportunity for deeper exploration. While “mechanical properties” and “wear resistance” appear in the graph, there is a visible gap in studies focusing on the use of detailed analytical tools like scanning electron microscopy (SEM) for correlating microstructure with wear behavior. On the other hand, the term “fused deposition modeling” connects to concepts like “flexural strength” and “polymer composites”, but there is no clear link emphasizing comprehensive studies that assess the combined impact of deposition pattern, speed, and layer height on wear properties. This underlines the novelty of our approach, which methodically evaluates these parameters using a Box–Behnken design (BBD).

In the keyword overlay visualization network, shown in Figure 2, keywords closer to red indicate more recent associations with the corresponding research year.

This second VOSviewer co-occurrence map includes a temporal dimension, highlighting how research has evolved over the past few years, as shown by the color gradient from 2020 to 2023. The color gradient reveals that research on topics such as “ABS”, “PLA”, and “wear” has been well-established since around 2020, shown in blue and green. However, more recent focus areas (yellow) are moving toward “mechanical properties”, “additive manufacturing”, and “composites”, indicating an expanding interest in how 3D-printed components can be engineered for specific functional traits. The relatively newer appearance of terms like “mechanical properties” and “tribology” suggests a growing emphasis on understanding the performance of 3D-printed materials under various conditions. Our study contributes to this recent trend by focusing on tribological performance, specifically examining COF and CLW.

A very important aspect is that, when incorporating the keyword “PET” (polyethylene terephthalate) into the research scope, the analysis yielded only two relevant papers in the context of 3D printing, tribology, and wear studies (namely [40,41]). This significant limitation underscores a clear gap in current research. While the VOSviewer co-occurrence maps display active research around materials such as “PLA”, “ABS”, and “PEEK”, the minimal presence of studies involving PET highlights its underexplored nature in the field of additive manufacturing and tribological assessments. Given PET’s promising properties, such as durability, recyclability, and potential for high-performance applications, the lack of literature focusing on its tribological behavior in 3D-printed parts represents an opportunity for novel investigation.

2. Experimental Details

2.1. Design of Experiments

The experimental design performed with Minitab 19 software, was based on the BBD, a type of RSM that allows the exploration of the interaction between multiple independent variables using a reduced number of experimental runs [39,42]. This design was chosen due to its efficiency in investigating complex relationships between factors while requiring fewer experimental trials than a full factorial design. The Box–Behnken design (BBD) was selected for this study due to its specific advantages in optimizing tribological properties, particularly for three independent variables. BBD is well-suited for exploring the relationships between multiple process parameters and their quadratic effects without requiring a full factorial design, which can be resource intensive. This method efficiently reduces the number of experimental runs compared to central composite design (CCD) or full factorial approaches while still providing reliable interaction and response surface information.

The three independent variables and their levels are as follows:

• Surface pattern deposition (A): linear, rectilinear, concentric;
• Deposition speed (B): 30, 40, 50 mm/s;
• Layer height (C): 0.10, 0.15, 0.20 mm.

The process parameters and levels considered for experimentation are shown in

Table 1. Significant factors and their corresponding levels.

Process Parameter	Level 1	Level 2	Level 3
Surface pattern deposition	Linear (coded −1)	Rectilinear (coded 0)	Concentric (coded 1)
Deposition speed [m/s]	30	40	50
Layer height [mm]	0.10	0.15	0.20

.

In the BBD, each factor is varied at three levels, and a total of 15 experimental trials are required. This includes combinations of high and low levels for pairs of factors while holding the third factor constant, as well as replicates at the center points of the design to estimate experimental variability and assess model accuracy.

Using the Box–Behnken design of experiments, the full factorial set of 27 experiments was reduced to 15 experiments. The experimental design is presented in Table 1.

The BBD for three factors generates a total of 15 experiments, including the following:

• Twelve unique combinations of the factors (runs 1–12),
• Three center points (middle level for all factors), which are repeated to assess experimental variability (runs 13–15).

The BBD approach was also chosen for its ability to generate predictive models that describe the relationships between the independent variables and the resulting tribological characteristics (coefficient of friction and wear). By applying regression analysis to the experimental data, second-order polynomial equations can be developed to predict the response variables (coefficient of friction and wear across the entire design space).

These models enable the estimation of optimal printing parameters within the specified ranges, facilitating a deeper understanding of how surface pattern, deposition speed, and layer height interact to affect the tribological performance of PET CF15.

In this study, a second-order polynomial regression equation was used to predict the behavior of PET CF15 under various additive manufacturing conditions. The general form of the regression equation used in the BBD is as follows [35]:

$$Y={\mathsf{\beta }}_0+{\displaystyle \sum _{i=1}^k{\mathsf{\beta }}_i{X}_i+}{\displaystyle \sum _{i=1}^k{\mathsf{\beta }}_{ii}{X}_i^2+}{\displaystyle \sum _{i<j}^k{\mathsf{\beta }}_{ij}{X}_i{X}_j}$$

(1)

where

• Y is the predicted response (COF, W),
• β₀ is the intercept (constant),
• β_i are the linear coefficients for each independent variable X_i (i.e., surface pattern deposition, deposition speed, layer height),
• β_ii are the quadratic coefficients that account for non-linear effects,
• β_ij are the interaction coefficients that describe how two factors combined affect the response variable,
• X_iX_j represent the levels of the independent variables.
• The regression equation captures both the linear effects of the individual factors and the interactions between them. For instance,
• Linear terms (β_iX_i) show the direct impact of each variable on the response,
• Quadratic terms (β_iiX_i²) describe non-linear relationships, highlighting how the response changes when increasing or decreasing a variable beyond the center point,
• Interaction terms (β_ij X_iX_j) illustrate how combinations of factors influence the response.

For the experimental study, the material polyethylene terephthalate with 15% carbon fiber reinforcement was considered (PET-CF, BASF, Ludwigshafen, Germany), simulating a tribotest for robotic arms using gear transmission and automated systems for precise motion control while minimizing weight. The specimens, shaped as discs with a 15 mm radius (Figure 3), were 3D printed using a Raise E2 3D printer (Raise 3D, Shanghai, China) and a 0.4 mm ruby nozzle to mitigate the risk of premature wear from the carbon fibers in the filament.

The printing parameters, kept constant, were as follows:

• A layer thickness of 0.2 mm,
• Two outer shells,
• Four bottom and top layers,
• A filling percentage of 50%,
• Deposition temperature of 270 °C.

This configuration was chosen because the tribological properties being investigated were more influenced by surface quality than by the internal structure of the printed samples.

By applying the BBD (with 15 runs and 3 center points) involving 3D printing of PET-CF: the deposition speed, layer height and deposition pattern were the varying parameters (

Table 2. The Box–Behnken design of experiments.

Std Order	Run Order	Pt Type	Blocks	Pattern	Speed	Layer Height
14	1	0	1	0	40	0.15
4	2	2	1	1	50	0.15
12	3	2	1	0	50	0.20
15	4	0	1	0	40	0.15
9	5	2	1	0	30	0.10
5	6	2	1	−1	40	0.10
2	7	2	1	1	30	0.15
7	8	2	1	−1	40	0.20
3	9	2	1	−1	50	0.15
1	10	2	1	−1	30	0.15
8	11	2	1	1	40	0.20
6	12	2	1	1	40	0.10
13	13	0	1	0	40	0.15
11	14	2	1	0	30	0.20
10	15	2	1	0	50	0.10

), the method helped to efficiently determine the optimal combination of these variables for a desired performance outcome.

The three factors were set as follows (Table 2):

• Deposition patterns for the top and the bottom layers (−1—lines, 0—rectilinear, 1—concentric),
• Deposition speed (30 mm/s, 40 mm/s, 50 mm/s),
• Layer height (0.1 mm, 0.15 mm, 0.2 mm).

2.2. Tribological Test Setup

The wear tests were developed according to the torque transmission of a cylindrical gear assembly for a vertical-articulated robot (joint shaft), whose geometric characteristics are detailed in

Table 3. Geometrical specification of the gear application.

Geometric Parameters	Values
Number of teeth	41
Module	1.25 mm
Pressure angle	20°
Type of gearing	External
Tip diameter	53.750 mm
Pitch diameter	51.250 mm
Pressure angle	20° mm
Width	13 mm
Shaft mounting diameter	10 mm

and Figure 4. Friction coefficients were determined using a CSM Instruments THT pin-on-disc tribometer (Freiburg in Breisgau, Germany), as illustrated in Figure 3 and data acquisition was made by using the InstrumX Software, version 6.0 (CSM Instruments, Peseux, Switzerland).

The friction pair consisted of a 30 mm diameter disc made from PET-CF material and a 4 mm cube of AISI 4130 alloy steel, simulating a gear assembly composed of these materials. Tribological tests were conducted under the following conditions: a normal load of 10 N, producing a contact pressure equivalent to the Hertzian pressure in gears (0.613 MPa, Equation (2)) when considering a nominal torque generated by a 250 W electric motor operating at 575 RPM; a friction distance of 250 m and a linear speed of 0.60 m/s. All tests were performed at room temperature (20 °C) in ambient air with 48% relative humidity.

$${\sigma }_H=\sqrt{{\displaystyle \frac{F_n}{l_k}}·{\displaystyle \frac{1}{\rho }}·{\displaystyle \frac{1}{\pi \left(\frac{1-{\vartheta }_1^2}{E_1}+\frac{1-{\vartheta }_2^2}{E_2}\right)}}}$$

(2)

where

• σ_H = Hertzian contact stress or maximum contact pressure (in Pascals, Pa);
• F_n = Normal force applied to the surfaces (in Newtons, N);
• l_k = Contact width or length (in meters, m);
• ρ = Equivalent radius of curvature at the contact point (in meters, m); this depends on the curvature of both contacting bodies;
• ν₁\ν₂ = Poisson’s ratios of the two contacting materials (PET-CF and AISI 4130 alloy steel), dimensionless constants representing the material’s ability to expand or contract laterally when compressed;
• E₁/E₂ = Young’s moduli of the two contacting materials (in MPa).

The coefficient of friction (μ) was calculated as the ratio of the tangential friction force to the normal force, while cumulative linear wear was determined by measuring the difference between maximum and minimum penetration (measuring the depth of wear at its highest and lowest points, excluding outliers). For each set of printing parameters, three friction pairs, classified as 4th, were tested. Continuous measurements were recorded at a sampling rate of 9.5 Hz during the tests to monitor both the coefficient of friction and cumulative linear wear.

2.3. Microstructural Characterization Method

The microstructural analysis using scanning electron microscopy (SEM) of 3D-printed surfaces made from carbon fiber-reinforced composite materials provides detailed insights into the influence of wear on these advanced structures. SEM facilitates the observation of microscopic topographical and morphological characteristics that result from wear and degradation. In this study, the evaluation focused on the effects of wear on the surfaces of components made from carbon fiber-reinforced composites, a material recognized for its strength and lightweight properties. SEM enables the examination of microcracks, fiber detachment, and other structural defects caused by wear. These investigations helped in the determination of wear impact on the structural and functional properties of the components.

For the microstructural analysis of 3D-printed surfaces made from carbon fiber-reinforced composite material, a scanning electron microscope HITACHI S-3400 (Hitachi High-Technologies Corporation, Tokyo, Japan) was configured with specific settings. Imaging was performed using a backscattered electron composite signal, with an accelerating voltage of 15,000 volts and no additional deceleration voltage applied. The emission current was set to 16,000 nanoamperes, ensuring adequate image brightness. The pressure in the SEM chamber was maintained at 30 (units unspecified), essential for optimal operation. The condenser 1 setting was at 80,000, optimizing the electron beam focus. The scan speed was chosen as “Slow3” to achieve the best possible resolution, and the calibration scan speed was set to 25.

3. Results and Discussions

3.1. Microstructural Characterization Results

The analysis using SEM of the carbon fiber-reinforced composite materials highlighted significant details regarding the microstructure and wear behavior of these materials. The results obtained revealed the presence of microcracks and delamination at the interface between the matrix and fibers, demonstrating critical points where mechanical stress and environmental factors exert their influence. The SEM images of the surfaces of the carbon fiber-reinforced composite samples are presented in their initial state in

Table 4. Initial SEM images of carbon fiber-reinforced composite specimens (mv—micro-voids, dl—delamination, ovl—overlapping zone).

SEM Images			Defined Parameters
			RunOrder 1 PtType 0 Blocks 1 Pattern 1 Speed 40 Layer height 0.15
			RunOrder 2 PtType 2 Blocks 1 Pattern 1 Speed 50 Layer height 0.15
			RunOrder 3 PtType 2 Blocks 1 Pattern 0 Speed 50 Layer height 0.20
			RunOrder 5 PtType 2 Blocks 1 Pattern 0 Speed 30 Layer height 0.10
			RunOrder 6 PtType 2 Blocks 1 Pattern -1 Speed 40 Layer height 0.10
			RunOrder 7 PtType 2 Blocks 1 Pattern 1 Speed 30 Layer height 0.15
			RunOrder 11 PtType 2 Blocks 1 Pattern 1 Speed 40 Layer height 0.20

, and in

Table 5. SEM images of the wear track on carbon fiber-reinforced composite specimens (ox—oxidative zones, pd—plastic deformation, db—debris, dl—delamination).

Sem Images			Defined Parameters
			RunOrder 1 PtType 0 Blocks 1 Pattern 1 Speed 40 Layer height 0.15
			RunOrder 2 PtType 2 Blocks 1 Pattern 1 Speed 50 Layer height 0.15
			RunOrder 3 PtType 2 Blocks 1 Pattern 0 Speed 50 Layer height 0.20
			RunOrder 5 PtType 2 Blocks 1 Pattern 0 Speed 30 Layer height 0.10

, after simulating the wear process.

The selected images indicate the presence of micro-voids and delamination, oxidations, plastic deformation and debris, which are common issues encountered in carbon fiber-reinforced composites that can significantly affect their mechanical performance and overall durability. The presence of micro-voids reduces mechanical properties such as strength and stiffness, impairs fatigue resistance, leading to premature failure, and alters the thermal and electrical conductivity of the composite material [43,44]. Delamination refers to the separation of layers within a composite material, which can occur between fiber layers or at the fiber–matrix interface. Delamination significantly weakens the composite’s overall structural integrity, potentially leading to catastrophic failure under loading conditions since delaminated areas cannot effectively transfer loads.

The fibers show a good level of alignment, which generally improves the mechanical properties of the material. The surface presents some roughness, indicating wear that occurred during the experiments. The wear marks exhibit surface flattening without deep penetration, regardless of the printing settings used, suggesting that these materials possess good wear resistance. Upon examining the matrix–fiber interface, there are no noticeable signs of separation, indicating a stable bond overall. All the samples were analyzed microscopically, the appearance of wear marks showing no significant differences.

Table 6. SEM structure of optimized sample.

SEM Images			Optimized Parameters
			RunOrder V Pattern 1 Speed 43 Layer height 0.10

presents the SEM images of the final sample, optimized in terms of the settings used for 3D printing, conducted after analyzing the experimental data obtained.

The sample made by optimizing the printing parameters shows signs of wear without surface deformation, or the outer layer showing signs of flattening.

3.2. Experimental Tribological Results

Table 7. Mean of COF and cumulative linear wear experimental results according to BBD.

Run Order	Pattern	Deposition Speed	Layer Height	COF	CLW
1	0	40	0.15	0.299	78.7097
2	1	50	0.15	0.373	32.9287
3	0	50	0.20	0.321	96.0912
4	0	40	0.15	0.118	42.1567
5	0	30	0.10	0.206	63.3729
6	−1	40	0.10	0.177	30.4561
7	1	30	0.15	0.405	28.6970
8	−1	40	0.20	0.305	70.3184
9	−1	50	0.15	0.320	28.7556
10	−1	30	0.15	0.301	57.7046
11	1	40	0.20	0.196	41.5728
12	1	40	0.10	0.181	27.3728
13	0	40	0.15	0.182	25.1284
14	0	30	0.20	0.284	40.6397
15	0	50	0.10	0.191	37.7734

presents a detailed summary of the data gathered during the experimental study, highlighting the measured results and variables under various controlled conditions. This comprehensive presentation facilitates a thorough analysis and interpretation of the experimental findings.

The COF results indicate a clear variation with changes in deposition speed and layer height, with the deposition pattern showing a less pronounced effect. As the deposition speed increased (from 30 mm/s to 50 mm/s), the COF generally tended to increase. This can be attributed to the higher velocity at which the material is deposited, which leads to increased surface roughness and less adhesion between the contacting surfaces. This increase in roughness could result in more frictional resistance during sliding contact. Specifically, runs 2 and 9, which had a deposition speed of 50 mm/s, showed higher COF values. The highest COF was obtained for run 7 (COF = 0.405), where a combination of 30 mm/s speed and 0.15 mm layer height was used.

The lowest COF value (0.118) was recorded for run 4, where both a lower deposition speed (40 mm/s) and a smaller layer height (0.15 mm) were used, which likely contributed to a smoother interface and less resistance during contact.

Layer height also played a role in the COF variation. A higher layer height (0.20 mm) generally resulted in an increase in COF, which can be explained by the larger surface irregularities and less dense microstructure in the material. This could be due to the increased roughness and the subsequent interaction between asperities on the surfaces, leading to higher friction. In terms of CLW, the data reveal a similar trend where CLW increases with layer height and deposition speed but is more sensitive to the deposition speed. For example, run 2, which had the highest deposition speed (50 mm/s) and a moderate layer height (0.15 mm), exhibited a CLW of 32.93 mm, significantly lower than other runs with slower deposition speeds. This suggests that although higher speeds might contribute to higher COF, they also result in less material wear, potentially due to quicker material deposition and lower overall heat generation during the process.

Conversely, when deposition speed was reduced to 30 mm/s, higher wear was observed, especially for run 5 (CLW = 63.37 mm). Additionally, for runs with a higher layer height (0.20 mm), such as run 3 and run 8, wear was relatively higher, possibly due to the increased roughness and larger voids created during the deposition, leading to greater material loss.

The deposition pattern did not show a strong influence on CLW, as runs with different patterns (0, 1, −1) exhibited relatively consistent wear values across varying deposition speeds and layer heights.

The friction–time curve for experiment No. 3 (Figure 5) is representative for the 3D-printed PET-CF material and demonstrates an overall stable trend, starting with an initial coefficient of friction of 0.167 and averaging 0.261 throughout the test, with minimal fluctuations (standard deviation of 0.011). While the friction coefficient remains steady for most of the duration, a slight gradual increase is observed, indicating potential surface wear or debris accumulation over time. Occasional spikes in the curve may reflect temporary contact instabilities or irregularities in the material’s surface. Overall, the curve suggests consistent frictional performance with minor wear-related adjustments as the test progresses.

3.3. Box–Behnken Statistical Results

After collecting the DoE required data, the results were investigated by applying the option Analyze Response Surface Design from Minitab 19 [45]. The first equation from

Table 8. BBD predictive regression models (P—pattern, S—speed, L—layer height).

Response	Box–Behnken Model	R2
COF	COF = 1.327 + 0.142 P − 0.0781 S + 4.89 L + 0.0572 P2 + 0.000929 S2 − 16.8 L2 − 0.00127 P*S − 0.565 P*L + 0.0260 S*L	94.7%
CLW	CLW = 365 − 33.3 P − 6.77 S − 2615 L − 12.6 P2 + 0.009 S2 + 3943 L2 + 0.830 P*S − 58 P*L + 40.5 S*L	91.8%

has a coefficient of determination of 94.7% and highlights that layer height (L) is the most influential factor, both in linear and quadratic terms. While increasing layer height initially increases COF, the strong quadratic term suggests that further increases lead to a reduction in COF. Deposition speed (S) generally lowers COF, and pattern (P) has a moderate, positive effect. Interaction effects, especially between pattern and layer height, play a significant role, indicating that the factors do not act independently. The second equation from Table 8 indicates a coefficient of determination of 91.8% and indicates also that the most influential factor is layer height (L), both in linear and quadratic terms. Figure 6 illustrates the standardized effects of various factors on the coefficient of friction (COF), with a significance level of α = 0.05. The quadratic effect of deposition speed (BB) has the largest influence on COF, followed by layer height (C), neither of which exceed the statistical significance threshold of 2.571. The quadratic effects of pattern (AA) and layer height (CC) also have notable impacts but are less influential than BB and C. Other factors, including interactions between pattern, speed, and layer height (AC, BC, AB), as well as the linear effects of pattern (A) and speed (B), fall below the significance threshold, indicating that they do not have a statistically significant effect on COF under the experimental conditions. Overall, deposition speed, particularly its quadratic component, and layer height are the most critical factors affecting friction behavior.

Initially, increasing the layer height greatly reduces wear, but beyond a certain point, wear starts to increase again. Pattern (P) also plays an important role in reducing wear, while speed (S) has a smaller, primarily negative effect on wear. The interaction between pattern and layer height significantly reduces wear, while the interaction between speed and layer height increases it. The overall trend is that controlling layer height is crucial for minimizing CLW, but interactions between parameters must be carefully managed to optimize performance, also noted in Figure 7.

The three surface plots from Figure 8 visualize the relationships between different parameters affecting “CLW”, which is the response variable. The first plot shows how CLW changes with different combinations of pattern and speed, holding the layer height constant at a central point of 0.15. The surface has a curve, indicating a nonlinear relationship between CLW, pattern, and speed. For example, concentric patterns at slower speeds have a positive effect on wear compared to rectilinear or line patterns. The second plot shows how CLW changes with different combinations of pattern and layer height, while holding speed constant at a central point of 40 mm. The curvature of the surface shows that variations in deposition pattern combined with different layer heights result in changes to CLW. For example, a line pattern at a higher layer height increases CLW values compared to a concentric pattern at a lower layer height. The third plot shows how the combination of deposition speed and layer height affects CLW, with deposition pattern held constant at 0 (rectilinear). In the second and third plots, layer height has a strong impact on CLW. Lower layer heights may improve or worsen the CLW depending on the other parameters (pattern and speed).

Figure 9 presents the surface plots of COF, which stands for the coefficient of friction (COF). The layout of the plots is similar to the previous one, showing the relationship between the deposition pattern, speed, and layer height with respect to COF. The hold values are as follows:

• Pattern: 0 (Rectilinear);
• Speed: 40 mm/s;
• Layer height: 0.15 mm.

This first plot shows how the COF varies based on deposition pattern and speed, with layer height held constant at 0.15 mm. The surface shows some non-linear interaction between the two variables, with lower speeds and patterns close to −1 (lines) and 1 (concentric) appearing to result in higher COF values. The second plot examines the impact of deposition pattern and layer height on COF, with speed held constant at 40 mm/s. There is a pronounced curvature, indicating that certain combinations of pattern and layer height yield lower or higher COF values. For instance, at lower layer heights and certain patterns, COF appears to decrease. The third plot shows how the COF is influenced by the interaction between speed and layer height, with deposition pattern held constant at 0 (rectilinear). The surface shows a clear change in COF as both parameters vary, with higher speeds and larger layer heights tending to increase the COF.

3.4. Response Optimization and Validation of the BBD

The process parameters, including deposition pattern, deposition speed and the layer height were optimized for both coefficients of friction and wear using a composite desirability function. For each response, the optimization criteria were similar, aiming to minimize the coefficient of friction and wear of the PET-CF 3D-printed samples (as detailed in

Table 9. Optimization goals.

Response	Goal	Lower	Target	Upper	Weight	Importance
Wear	Minimum		25.1284	96.0912	1	1
COF	Minimum		0.1180	0.4050	1	1

). The plot illustrates the multi-response optimization for minimizing both cumulative linear wear (CLW) and the coefficient of friction (COF) using a composite desirability function. The optimal settings are displayed at the top in Figure 10, with a pattern of 1, speed of 43 mm/s, and layer height of 0.1. The composite desirability value, “D = 0.8189”, represents the overall optimization level. The individual desirability values for CLW and COF are also shown, where CLW is minimized at 27.3205 with a desirability of 0.96911, and COF is minimized at 0.2064 with a desirability of 0.69204. The red vertical lines highlight the optimal parameter levels for achieving these results.

The optimized solution for minimizing both the coefficient of friction (COF) and cumulative linear wear (CLW) using a composite desirability function. The optimal settings include a speed of 42.9293, a layer height of 0.1, and fit values of 27.3205 for CLW and 0.206385 for COF, resulting in a composite desirability score of 0.818940.

Therefore, the analysis titled “Multiple Response Prediction” presents the predicted values for the cumulative linear wear (CLW) and coefficient of friction (COF) based on specific settings for pattern, speed, and layer height. The pattern is set to 1, the speed to 43 mm/s, and the layer height to 0.1.

• The predicted CLW is 27.3, with a standard error of the fit (which measures the accuracy of the predicted response) of 17.4, a 95% confidence interval (CI) of (−17.4, 72.0), and a 95% prediction interval (PI) of (−40.1, 94.8).
• The predicted COF is 0.2064, with a standard error of the fit of 0.0640, a 95% CI of (0.0419, 0.3709), and a 95% PI of (−0.0417, 0.4545).

The validation test results, with a coefficient of friction (COF) of 0.2215 and cumulative linear wear (CLW) of 29.2075, closely match the predicted values. The predicted COF was 0.2064, and the validation value shows a small percentage error of approximately 7.3%. For CLW, the predicted value was 27.3, with the validation result showing a percentage error of about 6.5%. Both the COF and CLW values from the validation fall within the 95% confidence intervals of their respective predictions, indicating good agreement and validating the accuracy of the model within these parameters.

4. Conclusions

This paper investigates the influence of 3D-printing parameters like deposition pattern, deposition speed and layer height on the tribological performance quantified through coefficient of friction and cumulative linear wear by applying the BBD method and microstructural characterization using SEM.

• The analysis of the experimental data, conducted through Minitab 19’s Response Surface Design feature, revealed that layer height has the strongest influence on the coefficient of friction (COF), with a high coefficient of determination (94.7%). Initially, a rise in layer height causes COF to increase, but the pronounced quadratic effect shows that further increases lead to a decrease in COF. Deposition speed generally reduces COF, while deposition pattern has a moderate positive impact on its value.
• The SEM image analysis of worn samples revealed the presence of multiple wear mechanisms. Abrasive wear is suggested by the visible grooves and scratches, indicating that harder particles or asperities moved across the surface, removing material. Additionally, some areas showed signs of plastic deformation, where the material was permanently deformed under stress.
• The validation samples’ results, printed with the optimal parameters of BBD analysis, with a coefficient of friction (COF) of 0.2215 and cumulative linear wear (CLW) of 29.2075, closely align with the predicted values. The predicted COF was 0.2064, showing a small percentage error of approximately 7.3%, while the predicted CLW was 27.3, with a percentage error of about 6.5%. Both COF and CLW validation values fall within the 95% confidence intervals of the predictions, confirming the model’s accuracy and reliability within these parameters.
• The micro characterization of the validation samples generally revealed smoother worn surfaces compared to the initial samples, indicating improvements in wear resistance under the same tested conditions.