Process–Microstructure–Property Relationships in FDM-Printed PLA Components: An Experimental and Predictive Study

Process–Microstructure–Property Relationships in FDM-Printed PLA Components: An Experimental and Predictive Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Process–Microstructure–Property Relationships in FDM-Printed PLA Components: An Experimental and Predictive Study Ruchira Chakraborty, Vishal J Hawale, Rominkumar A Vaghasiya, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9557043/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Fused Deposition Modeling (FDM) is widely used in additive manufacturing, yet printed components often exhibit poor mechanical performance due to process-induced microstructural defects. This study investigates the influence of key printing parameters, particularly print-bed temperature, on the process–structure–property relationship in polylactic acid (PLA). Mechanical properties were evaluated by tensile testing, and microstructural features were analyzed by optical microscopy and FE-SEM. Thermal behavior was examined using differential scanning calorimetry (DSC), revealing that increasing bed temperature reduces cold crystallization and promotes partial in-situ crystallization. An optimal condition at 70 °C produced the highest tensile strength, attributed to enhanced interlayer diffusion near the glass transition temperature and reduced void formation. FE-SEM observations confirmed improved filament fusion at elevated temperatures. The experimental dataset was also used to train a machine learning model to predict mechanical performance from process parameters. This integrated approach provides valuable insights into optimizing FDM printing for improved structural integrity and reliability. Fused Deposition Modeling (FDM) Microstructural Defects Fracture Morphology Mechanical Properties Process Parameters Machine Learning Figures Figure 1 Figure 3 Figure 4 Figure 5 Figure 6 Figure 8 1. Introduction Fused Deposition Modeling (FDM) has emerged as one of the most widely used additive manufacturing techniques due to its simplicity, cost-effectiveness, and ability to fabricate complex geometries through layer-by-layer deposition of thermoplastic filaments [ 1 , 2 ]. As a result, it is widely used for prototyping across academia, industry, and consumer applications. However, despite its widespread adoption, FDM has not yet matured into a reliable manufacturing method for load-bearing applications, primarily because printed components exhibit inferior and inconsistent mechanical properties.[ 3 ] The mechanical performance of FDM-printed parts is governed by a complex interplay of process parameters that influence inter- and intra-layer bonding. Since parts are formed through successive deposition of molten filaments, the adhesion between adjacent filaments in the x–y plane and across layers in the z-direction plays a critical role in determining structural integrity [ 4 , 5 ]. It is well established that printed components exhibit anisotropic behavior, with significantly weaker mechanical properties along the build (z) direction compared to the planar (x–y) direction [ 6 ]. Factors such as filament diameter, extrusion conditions, and deposition patterns further affect fusion and solidification kinetics, thereby influencing mechanical strength. For instance, Sudin et al. demonstrated that filament size, governed by nozzle diameter, directly affects bonding quality and mechanical performance [ 7 ]. Among various parameters, print-bed temperature plays a crucial role in controlling interlayer adhesion by influencing the thermal history of deposited filaments. Elevated bed temperatures can delay solidification, allowing increased molecular diffusion and improved bonding between layers. Conversely, non-uniform cooling and thermal gradients can induce residual stresses, warping, and dimensional instability, ultimately compromising mechanical performance [ 8 – 10 ]. These competing effects make it challenging to identify optimal processing conditions. Consequently, FDM parameter optimization is often carried out using time-consuming trial-and-error approaches, which are inefficient and material-intensive. Furthermore, optimal parameters vary with polymer type, necessitating repeated experimentation for different materials. To address this complexity, researchers have increasingly employed artificial intelligence and machine learning (AI/ML) techniques to model and predict the mechanical behavior of FDM-printed components [ 11 ]. Prior studies have demonstrated the effectiveness of ML algorithms in correlating process parameters such as infill density, layer thickness, and nozzle diameter with mechanical properties [ 12 – 15 ]. While these approaches show promising predictive capability, they often rely on empirical correlations and lack a fundamental understanding of the underlying microstructural mechanisms that govern material behavior. This limits the interpretability and generalizability of such models. In this study, we aim to bridge this gap by integrating experimental characterization with data-driven modeling to establish a comprehensive process–structure–property relationship for FDM-printed polylactic acid (PLA). Samples were fabricated under varying process parameters and characterized using microscopy and thermal analysis to capture microstructural and molecular-level changes. Mechanical performance was evaluated through tensile testing, and the resulting dataset was used to train and compare multiple machine learning models for predicting mechanical behavior. By linking processing conditions to microstructural features and mechanical outcomes, this work provides a more physically grounded framework for optimizing FDM processes and improving the reliability of polymer-based additive manufacturing. 2. Materials and Methods 2.1. Materials eSUN’s PLA filament spool (diameter 1.75mm, melting temperature 210–220°C), DSC pans (aluminum). 2.2. Methods 2.2.1. Sample preparation 3D-printed rectangular specimens were computer-aided designed in SolidWorks (v.2022), with a length of 8 cm, a width of 5 cm, and a thickness ranging from 1 mm to 4mm. Thereafter, the CAD design was converted to .stl format for slicing. The .stl file was sliced using slicing software (Simplify 3D v.4.1.1) to generate G-code for 3D printing of samples. Each sample was printed in triplicate to reduce the likelihood of random variations. The printing parameters, such as fan speed, were kept at 100%; the extruder temperature was 220°C, the printing speed was 45mm/sec, and the nozzle diameter was 300µm. Three sets of experimental analyses were performed by varying the printing parameters: mechanical strength, morphological, and DSC analyses. The first varying printing parameter was print bed temperature; samples were printed at bed temperatures ranging from 60°C to 90°C to determine their impact on mechanical properties.[ 16 , 17 ] The sample printed has employed grid infill patterns. Further, other specifications of the 3D printing process are mentioned in Table 1 (Supplementary). The second parameter was the infill pattern; samples were prepared with Grid (raster angles of -45° and 45°), Full Honeycomb (raster angles of 0° and 120°), Fast Honeycomb (raster angles of 0° and 90°), and Triangular patterns (-60°, 0°, and 60°). Other printing process specifications are mentioned in Table 2 (Supplementary).[ 18 ] The final varying parameter was infill percentage; the samples were printed with grid and triangular infill patterns, each with three distinct values: 60%, 70%, and 80%. The other printing parameters are mentioned in Table 3 (Supplementary).[ 19 ] All prepared samples were evaluated for their mechanical properties using a universal testing machine. The samples were also imaged by scanning electron microscopy before and after mechanical testing, and by optical microscopy for morphological characterization. 2.2.2. Tensile Testing All the above samples, with the same length and width of 25 x 80 mm and varying thicknesses of 1 mm, 2 mm, 3 mm, and 4mm, were subjected to tensile testing. Testing was conducted using an INSTRON 5967 universal testing machine (UTM) with a maximum load capacity of 30 KN and a closed chamber maintained at 30°C. To ensure accuracy and consistency, the samples were positioned in the UTM grip to align perfectly with the direction of the applied load. A standardized strain rate of 2 mm/min was applied until fracture. Throughout the testing process, force-elongation data were continuously recorded to capture the mechanical response of the samples under tensile stress. 2.2.3. Morphological analysis The FDM-printed samples were imaged using an optical microscope and a scanning electron microscope. The samples were examined using a Field Emission Scanning Electron Microscope (FE-SEM) (Nova nanoSEM-450) at 15–20 kV and at different magnifications, after being placed on carbon tape and sputter-coated with platinum to improve visualization of the non-conducting PLA samples. Furthermore, post-tensile testing, the fractured regions of the samples were also imaged using FE-SEM to assess alterations during testing. The captured images were processed using ImageJ software for morphological analysis. 2.2.4. Differential Scanning Calorimetry 6 µg of the sample was taken from the base layer of the 3D-printed samples, and a PLA spool was used as a control sample and analyzed with DSC (make and model: NETZSCH DSC 200F3). Thereafter, the temperature was raised from 30°C to 250°C in steps of 10°C. Heat transfer was recorded as the temperature changed. 2.2.5. ML for Mechanical property prediction of FDM printed parts The data used in this study were collected through tensile testing of various FDM-printed samples, each fabricated under different printing parameters. Key mechanical properties, including yield stress, Young’s Modulus (YM), ultimate tensile strength (UTS), and others, were measured during the tests. The dataset underwent rigorous quality control measures to ensure data accuracy and consistency, and to avoid missing values. To predict the mechanical properties of FDM-printed samples from collected input data, a suite of machine learning models was developed. These models were designed to handle multiple inputs and outputs, enabling a comprehensive analysis of the relationships between printing parameters and mechanical properties. 3. Results and Discussion 3.1. Physical characterization The actual thickness of PLA-printed samples at different bed temperatures was measured using a Vernier caliper. During physical characterization, the thickness of each sample was measured at three distinct locations. The thickness of the samples printed at different bed temperatures was compared with the specified CAD model thickness, as shown in Fig. 2 A. An increase in bed temperature led to changes in the thickness of printed samples relative to the specified thickness. Enhanced adhesion between layers was observed at higher temperatures due to improved fusion or increased chain penetration [ 20 ]. Thus, there is a likelihood that thickness reduction increases at higher temperatures, depending upon heat conduction across layers. However, no discernible trend was observed with increasing bed temperature, primarily due to PLA’s limited heat conductivity (0.0643 W/(m·K)). [ 21 ] Moreover, changes in thickness were influenced by the layer thickness used during printing, as the G-code segmented the given thickness into multiple layers based on nozzle diameter and layer thickness. To avoid any discrepancies, the individual layer thickness was kept constant. When comparing average thickness across temperatures (Fig. 2 A), no strong correlation was observed. At the same time, comparing thicknesses at 60°C and 90°C revealed a significant difference. For a calculated 4 mm sample, the thickness decreased from 4.28 ± 0.06 mm at 60°C to 3.93 ± 0.18 mm at 90°C, representing an approximate 8% decrease. Similarly, for 3mm, 2mm, and 1mm samples, thickness decreases by 1.97%, 2.71%, and − 7.07%, respectively. (Fig. 2 A) The porosity analysis of the samples shows the relationship between the experimental average porosity and the CAD model porosity (Fig. 2 B). The 60% infill percentage in the sample implies that the material is 40% porous. However, there was a significant deviation between the experimental average porosity and the porosity given in the CAD model. Further, it was observed that porosity was minimally affected by variations in bed temperature. (Fig. 2 B) The observed porosity range was 41.86% to 49.14%. Among the samples printed with the patterns, the full honeycomb infill consistently had a thickness lower than the specified value. In contrast, the grid infill was thicker than the specified value. (Fig. 2 C) Porosity values for these samples range from 42.5% to 49.12%. (Fig. 2 D) These elevated porosity values (> 40%) compared to CAD may correspond to micro-voids or microstructural defects that develop in samples during the printing process and may affect their mechanical strength. [ 22 ] 3.2. Mechanical Strength Analysis We generally expect a trend in mechanical strength when comparing parts printed at different print bed temperatures; this trend is often not observed. The results we obtained indicate a general decrease in Young’s Modulus and UTS with increasing bed temperature, except at 70°C, where UTS reached 8.14 ± 0.68, the highest value. (Fig. 3 A, B) A similar phenomenon was also reported by Liparoti et al., who demonstrated that print-bed temperatures around 70°C promote polymer chain diffusion and interlayer entanglement, thereby improving mechanical performance. In contrast, excessively high temperatures may negatively affect morphology and bonding.[ 23 ] The improved UTS is linked to increased adhesion between adjacent printed layers, driven by the segmental mobility of macromolecules, and is highest in a material when heated near its Tg, which can lead to improved adhesion between polymeric surfaces [ 20 ]. As the segmental mobility of polymer chains increases, segments can penetrate the interface, influencing adhesion, which depends on the extent of interdiffusion and chain interpenetration. Beyond 70°C, the adhesion forces between polymer molecules decrease, possibly due to altered interactions with the printing surface. Specifically, when PLA is printed on glass, adhesion forces reach a plateau at 80°C, suggesting stabilization of intermolecular bonds [ 24 ]. This may be why the material printed at a 70–80°C bed temperature shows the highest UTS, Young’s modulus, tensile strain at break, and other mechanical properties. (Fig. 3 A-C) ANOVA of the UTS data yielded a p-value of 0.018 (p < 0.05), affirming that bed temperature significantly impacts UTS. A similar analysis was conducted for Young’s modulus (YM) and percentage elongation at break, (Fig. 3 A, C) depicts the influence of bed temperature on YM. Conversely, percentage elongation at break demonstrated a negative correlation with bed temperature, as illustrated in Fig. 3 C. Statistical analysis, indicating a p-value of 0.028, supported this observation. The change in infill pattern was implemented to investigate its impact on mechanical properties. The average Young’s modulus (YM) was highest for triangular infill at 0.95 ± 0.05 GPa and lowest for grid infill, also at 0.95 ± 0.05 GPa. (Fig. 3 D). This disparity can be attributed to raster orientation: triangular infill follows 0°, 60°, and − 60° orientations, while grid infill follows 45° and − 45° orientations. Furthermore, ultimate tensile strength (UTS) reached its peak with full honeycomb infill at 12.87 ± 1.0 MPa and was lowest for grid infill at 8.04 ± 1.82 MPa, as demonstrated in Fig. 3 (D-F). All infill patterns have different raster angles, which significantly affect the ultimate tensile strength of the samples.[ 25 ] Percentage elongation at break was highest with grid infill at 7.73 ± 1.55% and lowest with triangular infill at 4.11 ± 0.35%. (Fig. 3 F) It’s notable that during loading, the 45° and − 45° raster orientations in the grid infill demonstrated greater potential to withstand the applied load, resulting in enhanced tensile properties. (Fig. 3 F) E. E. Cho et al . claim that changes in layer height significantly impact tensile properties.[ 26 ] For instance, in a full honeycomb structure, three raster angles for a single layer resulted in three layers being pressed into a 0.3mm space. This compression led to the formation of a compact layer, thereby increasing tensile strength [ 26 ]. A similar observation was made by A. Rodríguez et al., who found that increasing the infill percentage led to proportional increases in ultimate tensile strength (UTS), elongation percentage, modulus, and yield stress.[ 27 ] Furthermore, the study revealed that infill had a greater impact on these parameters than primary layer thickness and other printing parameters.[ 27 ] 3.3. Microstructural analysis One reason for the highest UTS at 70°C, compared to other print-bed temperatures, may lie in microstructural defects arising from uneven heat conduction through the printed layers. To further analyze, the morphological properties of the 3D-printed material were examined (Fig. 4 ). After tensile testing, the fractured areas were imaged with FE-SEM, as shown in Fig. 4 (A). Micron-sized voids were observed on the fractured portions of the samples, primarily between the gaps in adjacent filaments (Fig. 4 (A) and (B)). The size and shape of the voids varied, with the size distribution shown in Fig. 4 D. Further, owing to the 45° and − 45° raster angles chosen, air gaps are observed at the junction of filaments, as shown in 4(C). The air gaps may occur due to printing speed, non-uniform filament melting, a high solidification rate, or improper filament arrangement [ 28 ] The large air gap may also lead to stress concentration sites, which can initiate crack propagation and early fracture of samples during mechanical testing. However, during cutting, some voids may have collapsed, leading to incomplete observation of all voids in the samples. From the SEM images, it is observed that the number and size distribution of micro-voids are not uniform. If the bed temperature is high, then the solidification rate of the printed filament would be slow, leading to uniform cooling. Due to the high bed temperature, heat transfer from the bed to the printed samples will be high, potentially reaching the maximum layer. Furthermore, slow solidification of the extruded polymer melt improves adhesion between layers. All the samples prepared at four different temperatures have 7 layers. Figure 4 shows SEM images of the distribution of micro-voids in the three layers (bottom to top) of the four samples. All samples exhibit a rough fracture pattern, indicating that, during loading, this material resists breaking and thus has good mechanical strength. The average number of voids measured in the samples prepared at a bed temperature of 60°C is 12; it is 10, 9, and 8 for the samples prepared at bed temperatures of 70, 80, and 90°C, respectively. The observed voids ranged in diameter from 1.7 µm to 11.88 µm, with an average diameter of 4.13 µm. The cause of these voids may be the low extrusion temperature and the non-uniform cooling of the filament after extrusion [ 29 – 31 ]. The formation of voids across the various printed layers is probably due to the fact that, during filament solidification, the outer layer solidifies first, and then the solidification propagates inward. However, this outer solidified area acts as a constraint or fixed boundary.[ 32 ] The microstructural analysis showed that increasing the bed temperature reduced the number of voids in printed filaments, indicating improved filament fusion. (Fig. 5 ) However, excessive thermal exposure may promote crystallization and restrict chain mobility. Therefore, an optimal balance between reduced void density and sufficient chain mobility will result in enhanced mechanical strength of the printed components. 3.4. Differential Scanning Calorimetry Analysis Differential scanning calorimetry (DSC) was conducted to evaluate the thermal behavior of PLA samples printed at different bed temperatures. Approximately 6 mg of material was extracted from 4 mm printed specimens, ensuring consistent sampling. The DSC thermograms (heat flow vs temperature) exhibit three characteristic transitions of semi-crystalline polymers: the glass transition temperature (Tg), cold crystallization (Tcc), and melting temperature (Tm). Table 1 Heat Flow characteristics of the different samples printed at varying print-bed temperatures Sample Tg (°C) Tm (°C) ΔHcc ΔHm Xc (%) PLA Spool 65 176.36 0.64 0.92 0.29 60°C 65 176.36 0.59 0.81 0.23 70°C 65 176.40 0.04 1.07 1.10 80°C 65 176.35 0.03 1.00 1.04 90°C 65 176.37 0.03 0.86 0.89 The PLA spool material shows a Tg of ~ 65°C and Tm of ~ 176°C, consistent with reported values. (Table 1 ) The sample printed at 60°C exhibits thermal behavior similar to that of the PLA Spool or the not-printed sample, indicating minimal structural modification during printing. However, samples printed at higher bed temperatures (70–90°C) exhibit reduced cold-crystallization peaks, suggesting partial crystallization during printing. This reduction in ΔHcc indicates that elevated temperatures promote in-situ molecular rearrangement, thereby decreasing the extent of post-print crystallization. The degree of crystallinity (Xc) increases from 0.23% at 60°C to a maximum of 1.10% at 70°C, followed by a slight decline at higher temperatures. Although the absolute values are low due to prior baseline correction, the relative trends remain. The 70°C sample demonstrates an optimal balance between amorphous and crystalline phases, as reflected in both thermal and mechanical behavior. Notably, the highest tensile strength is observed for the 70°C sample. This can be attributed to the proximity of the bed temperature to the Tg of PLA, where polymer chains possess sufficient mobility to enable interlayer diffusion and molecular entanglement.[ 33 ] This enhances interfacial bonding between deposited filaments, improving mechanical performance. At higher temperatures (80–90°C), increased thermal exposure may lead to premature crystallization or chain relaxation, limiting interlayer diffusion and reducing effective bonding, despite comparable melting behavior.[ 34 ] These results highlight the critical role of processing temperature near Tg in governing the structure–property relationship in FDM-printed PLA. (Fig. 3 B) However, identifying optimal process parameters through conventional experimentation is time-consuming and resource-intensive. Therefore, integrating artificial intelligence (AI) offers a promising approach to predict and optimize printing conditions by quantitatively mapping process–structure–property relationships. 3.5. Machine learning modeling The performance of various machine learning models was evaluated based on their predictive accuracy for the mechanical properties of 3D-printed samples. The input parameters considered were infill pattern, infill percentage, number of layers, bed temperature, actual thickness, and raster angle. The output parameters included Young’s modulus (YM), Ultimate Tensile Strength (UTS), maximum load, and yield stress. The dataset was split into a training set comprising 70% of the data and a testing set containing the remaining 30%. The random state was set to 42 to ensure consistent data splitting. The correlation heat map visually depicts the strength and direction of relationships between input parameters (such as infill pattern, infill percentage, and bed temperature) and output parameters (modulus, UTS, yield stress, and maximum load) in the 3D printing process, as shown in Fig. 6 . Strong correlations are indicated by darker shades, providing insights into which input factors significantly influence the mechanical properties of printed parts. A strong correlation was found between input parameters such as infill pattern, infill percentage, raster angle, and number of layers, and modulus, UTS, yield stress, and maximum load. So, these input and output parameters were used to develop prediction models. The selected machine learning models and their hyperparameters are listed in Table 2 . The performance of each model was assessed using Mean Squared Error (MSE) and R-squared values as evaluation metrics Table 2 .[ 35 , 36 ] The Polynomial Regression model with degree 2 performed best among the tested models, with the lowest MSE of 0.003 and the highest R-squared of 0.888. This indicates that the Polynomial Regression model effectively captured the dataset's variance and accurately predicted mechanical properties based on input parameters. For small datasets, polynomial regression efficiently captures the nonlinear trends.[ 37 ] Although other models showed reasonable performance, none surpassed the predictive accuracy of Polynomial Regression. Table 2 Evaluation matrix for different machine learning models Sl. No. ML Model Mean Squared Error R-squared 1 Ridge Regression Model 0.011 0.726 2 Linear Regression Model 0.01 0.758 3 Polynomial Regression Model 0.003 0.888 4 Decision Tree Regression Model 0.006 0.813 5 Random Forest Regression Model 0.005 0.863 6 Support Vector Regression Model 0.009 0.753 Figure 8 compares the machine learning models' predicted and measured mechanical properties of the 3D-printed material. The Decision Tree Regression and Random Forest Regression models also demonstrated competitive results, suggesting their potential usefulness in predictive models for 3D printing processes. Promising results for the Decision Tree Regression and the Random Forest Regression models were also observed by Jayasudha et al and Ziadia et al.[ 38 , 39 ] Due to the limited dataset size, simpler models such as polynomial regression were preferred to avoid overfitting, although more complex ensemble methods, such as XGBoost, may achieve higher accuracy with larger datasets, as was reported by Alsakarneh et al. [ 40 ] 4. Conclusion and Future Work This study provides a comprehensive understanding of the mechanical properties of FDM-printed materials, focusing on microdefects, their formation mechanisms, and their influence on structural performance. Microscopic observations revealed variations in extruded filament dimensions across printed regions due to non-uniform cooling during deposition. These variations affect dimensional accuracy and promote air-gap formation at raster intersections, thereby influencing mechanical behavior. Statistical analysis confirmed that print-bed temperature significantly impacts mechanical properties. Among the infill patterns, the full honeycomb structure exhibited the highest ultimate tensile strength, the grid pattern showed maximum elongation, and the triangular pattern demonstrated the highest Young’s modulus. Fractographic analysis using FE-SEM revealed micro-voids within filaments, with void density decreasing at higher bed temperatures, indicating improved interlayer fusion. These voids form due to differential cooling, where the outer layer solidifies first and constrains the contracting inner region, leading to internal defects that act as stress concentrators. DSC analysis indicated weak glass transition and crystallization behavior, suggesting predominantly amorphous structures in printed PLA. The experimental dataset was used to develop machine learning models for predicting mechanical properties, with polynomial regression (degree 2) performing best (MSE = 0.003, R² = 0.888). Additionally, increasing the number of layers reduced the influence of bed temperature on the upper layers, altering cooling dynamics. Overall, the results highlight the strong interdependence between process parameters, microstructure, and mechanical performance, supporting the use of AI for predictive optimization in additive manufacturing. Declarations Author Contribution R.C , V.H, RV performed the experiments, RC and PK wrote the main manuscript, prepared the figures and table, HSN reviewed and edited the manuscript, PK conceptualized and supervised the work and edited the manuscript Acknowledgement VH, RV, and RC thank NIT Rourkela for financial support through MTech and PhD scholarships. P.K. acknowledges the Science and Engineering Research Board (SERB), Department of Science and Technology, Government of India, for providing financial support through a start-up research grant (SRG/2021/000859). Data Availability All data supporting the findings of this study are available within the paper and its Supplementary Information. References Dave HK, Patel ST (2021) Introduction to Fused Deposition Modeling Based 3D Printing Process. 1–21. https://doi.org/10.1007/978-3-030-68024-4_1 Penumakala PK, Santo J, Thomas A (2020) A critical review on the fused deposition modeling of thermoplastic polymer composites. Compos B Eng 201:108336. https://doi.org/10.1016/j.compositesb.2020.108336 O’Connor L (2025) Comparative analysis of the mechanical properties of FDM and SLA 3D printed components. J Micromanufacturing. https://doi.org/10.1177/25165984251364689 Patel KS, Shah DB, Joshi SJ et al (2024) Effect of process parameters on the mechanical performance of FDM printed carbon fiber reinforced PETG. J Mater Res Technol 30:8006–8018. https://doi.org/10.1016/j.jmrt.2024.05.184 Bellehumeur C, Li L, Sun Q, Gu P (2004) Modeling of Bond Formation Between Polymer Filaments in the Fused Deposition Modeling Process. J Manuf Process 6:170–178. https://doi.org/10.1016/S1526-6125(04)70071-7 Vanaei S, Rastak M, El Magri A et al (2023) Orientation-Dependent Mechanical Behavior of 3D Printed Polylactic Acid Parts: An Experimental–Numerical Study. Machines 2023, Vol 11, 11:. https://doi.org/10.3390/machines11121086 Sudin MN, Daud NM, Ramli FR, Yusuff MA, Science (2023) Eng Technol 3:33–43. https://doi.org/10.54327/set2023/v3.i1.71 Chen G, Wang D, Hua W et al (2023) Simulating and Predicting the Part Warping in Fused Deposition Modeling by Thermal–Structural Coupling Analysis. 3D Print Addit Manuf 10:70. https://doi.org/10.1089/3dp.2021.0119 Amelot A, Colman M, Loret JE (2023) A thermodynamic framework for the additive manufacturing of crystallizing polymers. Part I: A theory that accounts for phase change, shrinkage, warpage and residual stress. Int J Eng Sci 183:103789. https://doi.org/10.1016/j.spinee.2018.01.019 El Moumen A, Tarfaoui M, Lafdi K (2019) Modelling of the temperature and residual stress fields during 3D printing of polymer composites. Int J Adv Manuf Technol 2019 104(5):1661–1676. https://doi.org/10.1007/s00170-019-03965-y Singh B, Kumar R, Singh Chohan J (2022) Multi-objective optimization of 3D Printing process using genetic algorithm for fabrication of copper reinforced ABS parts. Mater Today Proc 48:981–988. https://doi.org/10.1016/j.matpr.2021.06.264 Özkül M, Kuncan F, Ulkir O (2025) Predicting Mechanical Properties of FDM-Produced Parts Using Machine Learning Approaches. J Appl Polym Sci 142:e56899. https://doi.org/10.1002/app.56899 Wei H, Tang L, Qin H et al (2024) Optimizing FDM 3D printing parameters for improved tensile strength using the Takagi–Sugeno fuzzy neural network. Mater Today Commun 38:108268. https://doi.org/10.1016/j.mtcomm.2024.108268 Ulkir O, Akgun G, Karadag A (2025) Mechanical Behavior Prediction of 3D-Printed PLA/Wood Composites Using Artificial Neural Network and Fuzzy Logic. Polym Adv Technol 36:e70103. https://doi.org/10.1002/pat.70103 Laurenti M, Bavasso I, Palazzi E et al (2025) Predicting the mechanical behavior in FDM printing of biopolymers through boosting artificial neural networks. Mater Des 257:114475. https://doi.org/10.1016/j.matdes.2025.114475 Chalgham A, Ehrmann A, Wickenkamp I (2021) Mechanical Properties of FDM Printed PLA Parts before and after Thermal Treatment. Polym 2021 13. https://doi.org/10.3390/POLYM13081239 İncesu R, Akderya T (2024) The Influence of Printing Speed and Temperature on the Mechanical, Absorptive, and Morphological Properties of PLA-Based Hybrid Materials Produced with an FDM-Type 3D Printer. Polymers 2024, Vol 16, 16:. https://doi.org/10.3390/POLYM16192771 Chacón JM, Caminero MA, García-Plaza E, Núñez PJ (2017) Additive manufacturing of PLA structures using fused deposition modelling: Effect of process parameters on mechanical properties and their optimal selection. Mater Des 124:143–157. https://doi.org/10.1016/J.MATDES.2017.03.065 Gunasekaran KN, Aravinth V, Kumaran CBM et al (2021) Investigation of mechanical properties of PLA printed materials under varying infill density. Mater Today Proc 45:1849–1856. https://doi.org/10.1016/J.MATPR.2020.09.041 Spoerk M, Gonzalez-Gutierrez J, Sapkota J et al (2018) Effect of the printing bed temperature on the adhesion of parts produced by fused filament fabrication. Plast Rubber Compos 47:17–24. https://doi.org/10.1080/14658011.2017.1399531 Barkhad MS, Abu-Jdayil B, Mourad AHI, Iqbal MZ (2020) Thermal Insulation and Mechanical Properties of Polylactic Acid (PLA) at Different Processing Conditions. Polymers 2020, Vol 12, 12:. https://doi.org/10.3390/POLYM12092091 Wang X, Zhao L, Fuh JYH, Lee HP (2019) Effect of Porosity on Mechanical Properties of 3D Printed Polymers: Experiments and Micromechanical Modeling Based on X-ray Computed Tomography Analysis. Polym 2019 11:11. https://doi.org/10.3390/polym11071154 Liparoti S, Sofia D, Romano A et al (2021) Fused Filament Deposition of PLA: The Role of Interlayer Adhesion in the Mechanical Performances. Polym 2021 13:13:1–18. https://doi.org/10.3390/polym13030399 Jabbari E, Peppas NA (1994) Polymer-polymer interdiffusion and adhesion. J Macromolecular Sci Part C 34:205–241. https://doi.org/10.1080/15321799408009635 Algarni M (2021) The Influence of Raster Angle and Moisture Content on the Mechanical Properties of PLA Parts Produced by Fused Deposition Modeling. Polym 2021 13:13:1–12. https://doi.org/10.3390/polym13020237 Ganeshkumar S, Kumar SD, Magarajan U et al (2022) Investigation of Tensile Properties of Different Infill Pattern Structures of 3D-Printed PLA Polymers: Analysis and Validation Using Finite Element Analysis in ANSYS. Materials 2022, Vol 15, 15:. https://doi.org/10.3390/ma15155142 Rodríguez-Panes A, Claver J, Camacho AM (2018) The Influence of Manufacturing Parameters on the Mechanical Behaviour of PLA and ABS Pieces Manufactured by FDM: A Comparative Analysis. Mater 2018 11:11. https://doi.org/10.3390/ma11081333 Wickramasinghe S, Do T, Tran P (2020) FDM-Based 3D Printing of Polymer and Associated Composite: A Review on Mechanical Properties. Defects Treatments Polym 2020 12:12:1–42. https://doi.org/10.3390/polym12071529 Gajjar T, Yang R, Ye L, Zhang YX (2024) Effects of key process parameters on tensile properties and interlayer bonding behavior of 3D printed PLA using fused filament fabrication. Progress Additive Manuf 2024 10(2):1261–1280. https://doi.org/10.1007/s40964-024-00704-y Mencarelli M, Sisella M, Puggelli L et al (2025) Sensitivity Analysis of 3D Printing Parameters on Mechanical Properties of Fused Deposition Modeling-Printed Polylactic Acid Parts. Appl Mech 2025 6:6. https://doi.org/10.3390/applmech6010017 Bermudo Gamboa C, Martín Béjar S, Trujillo Vilches FJ, Sevilla Hurtado L (2023) Geometrical analysis in material extrusion process with polylactic acid (PLA)+carbon fiber. Rapid Prototyp J 29:21–39. https://doi.org/10.1108/RPJ-09-2022-0294 Antony Samy A, Golbang A, Harkin-Jones E et al (2021) Prediction of part distortion in Fused Deposition Modelling (FDM) of semi-crystalline polymers via COMSOL: Effect of printing conditions. CIRP J Manuf Sci Technol 33:443–453. https://doi.org/10.1016/j.cirpj.2021.04.012 Omer MAE, Shaban IA, Mourad AH, Hegab H (2025) Advances in interlayer bonding in fused deposition modelling: a comprehensive review. Virtual Phys Prototyp 20:2522951. https://doi.org/10.1080/17452759.2025.2522951 Yin J, Lu C, Fu J et al (2018) Interfacial bonding during multi-material fused deposition modeling (FDM) process due to inter-molecular diffusion. Mater Des 150:104–112. https://doi.org/10.1016/j.matdes.2018.04.029 Rashidi HH, Albahra S, Robertson S et al (2023) Common statistical concepts in the supervised Machine Learning arena. Front Oncol 13:1130229. https://doi.org/10.3389/FONC.2023.1130229/FULL Chicco D, Warrens MJ, Jurman G (2021) The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Comput Sci 7:1–24. https://doi.org/10.7717/PEERJ-CS.623/SUPP-1 Nikzad MH, Heidari-Rarani M, Rasti R, Sareh P (2025) Machine learning-driven prediction of tensile strength in 3D-printed PLA parts. Expert Syst Appl 264:125836. https://doi.org/10.1016/J.ESWA.2024.125836 Jayasudha M, Elangovan M, Mahdal M, Priyadarshini J (2022) Accurate Estimation of Tensile Strength of 3D Printed Parts Using Machine Learning Algorithms. Processes 2022, Vol 10, 10:. https://doi.org/10.3390/PR10061158 Ziadia A, Habibi M, Kelouwani S (2023) Machine Learning Study of the Effect of Process Parameters on Tensile Strength of FFF PLA and PLA-CF. Eng 2023, Vol 4, Pages 2741–2763 4:2741–2763. https://doi.org/10.3390/ENG4040156 Alsakarneh A, Obaidat S, Mumani AA, Tamimi MF (2025) Predicting the dynamic tensile response of FDM materials using machine learning. Discover Appl Sci 2025 8(1):51. https://doi.org/10.1007/S42452-025-08049-Z Additional Declarations No competing interests reported. Supplementary Files Supplementary.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9557043","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":637657462,"identity":"faf33587-ec98-4cdd-ba36-fb169c5be07b","order_by":0,"name":"Ruchira Chakraborty","email":"","orcid":"","institution":"National Institute of Technology Rourkela","correspondingAuthor":false,"prefix":"","firstName":"Ruchira","middleName":"","lastName":"Chakraborty","suffix":""},{"id":637657463,"identity":"f4070304-c5e5-4293-8fbd-4df8bd17cd27","order_by":1,"name":"Vishal J Hawale","email":"","orcid":"","institution":"National Institute of Technology Rourkela","correspondingAuthor":false,"prefix":"","firstName":"Vishal","middleName":"J","lastName":"Hawale","suffix":""},{"id":637657464,"identity":"59aa9e8c-e0c8-40e4-b443-9b2cd0fadeb7","order_by":2,"name":"Rominkumar A Vaghasiya","email":"","orcid":"","institution":"National Institute of Technology Rourkela","correspondingAuthor":false,"prefix":"","firstName":"Rominkumar","middleName":"A","lastName":"Vaghasiya","suffix":""},{"id":637657465,"identity":"2e72a135-b692-4931-adbe-1eccb88c405c","order_by":3,"name":"Himansu Sekhar Nanda","email":"","orcid":"","institution":"Indian Institute of Information Technology Design and Manufacturing Jabalpur","correspondingAuthor":false,"prefix":"","firstName":"Himansu","middleName":"Sekhar","lastName":"Nanda","suffix":""},{"id":637657469,"identity":"c598cd58-64af-4a78-84cb-2541e97ce6ad","order_by":4,"name":"Prasoon Kumar","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAElEQVRIiWNgGAWjYBACCRTeBxCfh7kBxOYhSgvjDAYJCQYeRhK0MPOABKBacALJGdmJnyt33JOXb28+9th2h0Wdec/BBoYfNQwy5ji0SEvkbpY8e6bYcMOZY+nGuWckJGTONjYw9hxj4LHEYZmcRO4Gyca2BMYNEjlm0rltEhIS/ECH8TYw8BgcwKll80+gFvv5M4BaLKFaGP/i0QJ02DaQLYkNN4BaGEFaeBsbmPHZItnzdptl45mEZKBf0iR72yQkZ/AcbDgsc0wCpxaJ47mbbzbuSLCdDwwxiZ9tdfwSPMkHH76psbHHpQUMMCLiAHqyIKxlFIyCUTAKRgEyAAAPHFR/RAdbRAAAAABJRU5ErkJggg==","orcid":"","institution":"National Institute of Technology Rourkela","correspondingAuthor":true,"prefix":"","firstName":"Prasoon","middleName":"","lastName":"Kumar","suffix":""}],"badges":[],"createdAt":"2026-04-28 17:23:29","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9557043/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9557043/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109339375,"identity":"e8c5b4a8-ef3f-446a-86b9-babe6219be01","added_by":"auto","created_at":"2026-05-15 18:17:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":224969,"visible":true,"origin":"","legend":"\u003cp\u003e(A) Schematic explaining the process and components of FDM-based 3D-Printing (B) Infill Patterns definition\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/eb96f2152ff8fa25a6367440.png"},{"id":109339379,"identity":"ae9b4030-855b-40ac-81f9-5e4bc967c764","added_by":"auto","created_at":"2026-05-15 18:17:01","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":57042,"visible":true,"origin":"","legend":"\u003cp\u003eMechanical Strength characteristics Young’s Modulus, UTS, Percentage elongation at break of the 3D-printed parts (A-C) at different print-bed temperatures (D-F), in case of different infill patterns (G-I), in case of different infill%\u003c/p\u003e","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/85da96e377cb88a98ff36b38.png"},{"id":109405315,"identity":"1bf5c394-255e-49fa-bea7-e85f79b139bb","added_by":"auto","created_at":"2026-05-17 13:16:43","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":693497,"visible":true,"origin":"","legend":"\u003cp\u003eMicrostructural features of 3D-printed PLA after tensile testing printed at different print-bed temperatures (A) 60°C, (B) 70°C, (C) 80°C, (D) 90°C\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/d8564753d982b982b474a59c.png"},{"id":109339381,"identity":"f337fb38-751a-44fb-9be1-458dacf1739d","added_by":"auto","created_at":"2026-05-15 18:17:01","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":33284,"visible":true,"origin":"","legend":"\u003cp\u003eDistribution of the micron-sized voids in the printed samples at different print bed temperatures\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/e51bb946987bd6cd6ed24555.png"},{"id":109405420,"identity":"6f5f9cbb-f3d2-4bfe-a198-208353adf0b5","added_by":"auto","created_at":"2026-05-17 13:17:56","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":62816,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e(A) \u003c/strong\u003eDSC thermograms of 3D printed parts at different print-bed temperatures\u003cstrong\u003e (B) \u003c/strong\u003eCrystallinity % of the samples for varying process parameters\u003c/p\u003e","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/c1490dbb08a4d5790c70030d.png"},{"id":109339384,"identity":"05cc10c6-910d-4fbf-9d32-2999b53213fb","added_by":"auto","created_at":"2026-05-15 18:17:01","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":186201,"visible":true,"origin":"","legend":"\u003cp\u003eMechanical property prediction by different ML models (A) Ridge Regression model, (B) Linear Regression model, (C) Decision Tree Regression model, (D) Polynomial Regression model, (E) Random Forest Regression model, (F) Support Vector Regression model.\u003c/p\u003e","description":"","filename":"floatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/c15308a0f98cedb961c47d91.png"},{"id":109405878,"identity":"aa57dbb3-6b45-40d2-a250-5ff60f455fc2","added_by":"auto","created_at":"2026-05-17 13:20:43","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1429067,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/f92962d6-04dc-4dfe-a4c6-0fba7853eba7.pdf"},{"id":109339377,"identity":"c3edd5f7-be22-498f-8edf-93b3f817cf7d","added_by":"auto","created_at":"2026-05-15 18:17:01","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":17566,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-9557043/v1/b20e484c6396bbe58f28fba7.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Process–Microstructure–Property Relationships in FDM-Printed PLA Components: An Experimental and Predictive Study","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eFused Deposition Modeling (FDM) has emerged as one of the most widely used additive manufacturing techniques due to its simplicity, cost-effectiveness, and ability to fabricate complex geometries through layer-by-layer deposition of thermoplastic filaments [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. As a result, it is widely used for prototyping across academia, industry, and consumer applications. However, despite its widespread adoption, FDM has not yet matured into a reliable manufacturing method for load-bearing applications, primarily because printed components exhibit inferior and inconsistent mechanical properties.[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] The mechanical performance of FDM-printed parts is governed by a complex interplay of process parameters that influence inter- and intra-layer bonding. Since parts are formed through successive deposition of molten filaments, the adhesion between adjacent filaments in the x\u0026ndash;y plane and across layers in the z-direction plays a critical role in determining structural integrity [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. It is well established that printed components exhibit anisotropic behavior, with significantly weaker mechanical properties along the build (z) direction compared to the planar (x\u0026ndash;y) direction [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Factors such as filament diameter, extrusion conditions, and deposition patterns further affect fusion and solidification kinetics, thereby influencing mechanical strength. For instance, Sudin et al. demonstrated that filament size, governed by nozzle diameter, directly affects bonding quality and mechanical performance [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAmong various parameters, print-bed temperature plays a crucial role in controlling interlayer adhesion by influencing the thermal history of deposited filaments. Elevated bed temperatures can delay solidification, allowing increased molecular diffusion and improved bonding between layers. Conversely, non-uniform cooling and thermal gradients can induce residual stresses, warping, and dimensional instability, ultimately compromising mechanical performance [\u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. These competing effects make it challenging to identify optimal processing conditions. Consequently, FDM parameter optimization is often carried out using time-consuming trial-and-error approaches, which are inefficient and material-intensive. Furthermore, optimal parameters vary with polymer type, necessitating repeated experimentation for different materials. To address this complexity, researchers have increasingly employed artificial intelligence and machine learning (AI/ML) techniques to model and predict the mechanical behavior of FDM-printed components [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Prior studies have demonstrated the effectiveness of ML algorithms in correlating process parameters such as infill density, layer thickness, and nozzle diameter with mechanical properties [\u003cspan additionalcitationids=\"CR13 CR14\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. While these approaches show promising predictive capability, they often rely on empirical correlations and lack a fundamental understanding of the underlying microstructural mechanisms that govern material behavior. This limits the interpretability and generalizability of such models.\u003c/p\u003e \u003cp\u003eIn this study, we aim to bridge this gap by integrating experimental characterization with data-driven modeling to establish a comprehensive process\u0026ndash;structure\u0026ndash;property relationship for FDM-printed polylactic acid (PLA). Samples were fabricated under varying process parameters and characterized using microscopy and thermal analysis to capture microstructural and molecular-level changes. Mechanical performance was evaluated through tensile testing, and the resulting dataset was used to train and compare multiple machine learning models for predicting mechanical behavior. By linking processing conditions to microstructural features and mechanical outcomes, this work provides a more physically grounded framework for optimizing FDM processes and improving the reliability of polymer-based additive manufacturing.\u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Materials\u003c/h2\u003e \u003cp\u003eeSUN\u0026rsquo;s PLA filament spool (diameter 1.75mm, melting temperature 210\u0026ndash;220\u0026deg;C), DSC pans (aluminum).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Methods\u003c/h2\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.2.1. Sample preparation\u003c/h2\u003e \u003cp\u003e3D-printed rectangular specimens were computer-aided designed in SolidWorks (v.2022), with a length of 8 cm, a width of 5 cm, and a thickness ranging from 1 mm to 4mm. Thereafter, the CAD design was converted to .stl format for slicing. The .stl file was sliced using slicing software (Simplify 3D v.4.1.1) to generate G-code for 3D printing of samples. Each sample was printed in triplicate to reduce the likelihood of random variations. The printing parameters, such as fan speed, were kept at 100%; the extruder temperature was 220\u0026deg;C, the printing speed was 45mm/sec, and the nozzle diameter was 300\u0026micro;m. Three sets of experimental analyses were performed by varying the printing parameters: mechanical strength, morphological, and DSC analyses. The first varying printing parameter was print bed temperature; samples were printed at bed temperatures ranging from 60\u0026deg;C to 90\u0026deg;C to determine their impact on mechanical properties.[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] The sample printed has employed grid infill patterns. Further, other specifications of the 3D printing process are mentioned in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e (Supplementary). The second parameter was the infill pattern; samples were prepared with Grid (raster angles of -45\u0026deg; and 45\u0026deg;), Full Honeycomb (raster angles of 0\u0026deg; and 120\u0026deg;), Fast Honeycomb (raster angles of 0\u0026deg; and 90\u0026deg;), and Triangular patterns (-60\u0026deg;, 0\u0026deg;, and 60\u0026deg;). Other printing process specifications are mentioned in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e (Supplementary).[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] The final varying parameter was infill percentage; the samples were printed with grid and triangular infill patterns, each with three distinct values: 60%, 70%, and 80%. The other printing parameters are mentioned in Table\u0026nbsp;3 (Supplementary).[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eAll prepared samples were evaluated for their mechanical properties using a universal testing machine. The samples were also imaged by scanning electron microscopy before and after mechanical testing, and by optical microscopy for morphological characterization.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.2.2. Tensile Testing\u003c/h2\u003e \u003cp\u003eAll the above samples, with the same length and width of 25 x 80 mm and varying thicknesses of 1 mm, 2 mm, 3 mm, and 4mm, were subjected to tensile testing. Testing was conducted using an INSTRON 5967 universal testing machine (UTM) with a maximum load capacity of 30 KN and a closed chamber maintained at 30\u0026deg;C. To ensure accuracy and consistency, the samples were positioned in the UTM grip to align perfectly with the direction of the applied load. A standardized strain rate of 2 mm/min was applied until fracture. Throughout the testing process, force-elongation data were continuously recorded to capture the mechanical response of the samples under tensile stress.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.2.3. Morphological analysis\u003c/h2\u003e \u003cp\u003eThe FDM-printed samples were imaged using an optical microscope and a scanning electron microscope. The samples were examined using a Field Emission Scanning Electron Microscope (FE-SEM) (Nova nanoSEM-450) at 15\u0026ndash;20 kV and at different magnifications, after being placed on carbon tape and sputter-coated with platinum to improve visualization of the non-conducting PLA samples. Furthermore, post-tensile testing, the fractured regions of the samples were also imaged using FE-SEM to assess alterations during testing. The captured images were processed using ImageJ software for morphological analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e \u003ch2\u003e2.2.4. Differential Scanning Calorimetry\u003c/h2\u003e \u003cp\u003e6 \u0026micro;g of the sample was taken from the base layer of the 3D-printed samples, and a PLA spool was used as a control sample and analyzed with DSC (make and model: NETZSCH DSC 200F3). Thereafter, the temperature was raised from 30\u0026deg;C to 250\u0026deg;C in steps of 10\u0026deg;C. Heat transfer was recorded as the temperature changed.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003e2.2.5. ML for Mechanical property prediction of FDM printed parts\u003c/h2\u003e \u003cp\u003eThe data used in this study were collected through tensile testing of various FDM-printed samples, each fabricated under different printing parameters. Key mechanical properties, including yield stress, Young\u0026rsquo;s Modulus (YM), ultimate tensile strength (UTS), and others, were measured during the tests. The dataset underwent rigorous quality control measures to ensure data accuracy and consistency, and to avoid missing values. To predict the mechanical properties of FDM-printed samples from collected input data, a suite of machine learning models was developed. These models were designed to handle multiple inputs and outputs, enabling a comprehensive analysis of the relationships between printing parameters and mechanical properties.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"3. Results and Discussion","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.1. Physical characterization\u003c/h2\u003e \u003cp\u003eThe actual thickness of PLA-printed samples at different bed temperatures was measured using a Vernier caliper. During physical characterization, the thickness of each sample was measured at three distinct locations. The thickness of the samples printed at different bed temperatures was compared with the specified CAD model thickness, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA. An increase in bed temperature led to changes in the thickness of printed samples relative to the specified thickness. Enhanced adhesion between layers was observed at higher temperatures due to improved fusion or increased chain penetration [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Thus, there is a likelihood that thickness reduction increases at higher temperatures, depending upon heat conduction across layers. However, no discernible trend was observed with increasing bed temperature, primarily due to PLA\u0026rsquo;s limited heat conductivity (0.0643 W/(m\u0026middot;K)). [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] Moreover, changes in thickness were influenced by the layer thickness used during printing, as the G-code segmented the given thickness into multiple layers based on nozzle diameter and layer thickness. To avoid any discrepancies, the individual layer thickness was kept constant. When comparing average thickness across temperatures (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA), no strong correlation was observed. At the same time, comparing thicknesses at 60\u0026deg;C and 90\u0026deg;C revealed a significant difference. For a calculated 4 mm sample, the thickness decreased from 4.28\u0026thinsp;\u0026plusmn;\u0026thinsp;0.06 mm at 60\u0026deg;C to 3.93\u0026thinsp;\u0026plusmn;\u0026thinsp;0.18 mm at 90\u0026deg;C, representing an approximate 8% decrease. Similarly, for 3mm, 2mm, and 1mm samples, thickness decreases by 1.97%, 2.71%, and \u0026minus;\u0026thinsp;7.07%, respectively. (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA)\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe porosity analysis of the samples shows the relationship between the experimental average porosity and the CAD model porosity (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). The 60% infill percentage in the sample implies that the material is 40% porous. However, there was a significant deviation between the experimental average porosity and the porosity given in the CAD model. Further, it was observed that porosity was minimally affected by variations in bed temperature. (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) The observed porosity range was 41.86% to 49.14%. Among the samples printed with the patterns, the full honeycomb infill consistently had a thickness lower than the specified value. In contrast, the grid infill was thicker than the specified value. (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC) Porosity values for these samples range from 42.5% to 49.12%. (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD) These elevated porosity values (\u0026gt;\u0026thinsp;40%) compared to CAD may correspond to micro-voids or microstructural defects that develop in samples during the printing process and may affect their mechanical strength. [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.2. Mechanical Strength Analysis\u003c/h2\u003e \u003cp\u003eWe generally expect a trend in mechanical strength when comparing parts printed at different print bed temperatures; this trend is often not observed. The results we obtained indicate a general decrease in Young\u0026rsquo;s Modulus and UTS with increasing bed temperature, except at 70\u0026deg;C, where UTS reached 8.14\u0026thinsp;\u0026plusmn;\u0026thinsp;0.68, the highest value. (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, B) A similar phenomenon was also reported by Liparoti et al., who demonstrated that print-bed temperatures around 70\u0026deg;C promote polymer chain diffusion and interlayer entanglement, thereby improving mechanical performance. In contrast, excessively high temperatures may negatively affect morphology and bonding.[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eThe improved UTS is linked to increased adhesion between adjacent printed layers, driven by the segmental mobility of macromolecules, and is highest in a material when heated near its Tg, which can lead to improved adhesion between polymeric surfaces [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. As the segmental mobility of polymer chains increases, segments can penetrate the interface, influencing adhesion, which depends on the extent of interdiffusion and chain interpenetration. Beyond 70\u0026deg;C, the adhesion forces between polymer molecules decrease, possibly due to altered interactions with the printing surface. Specifically, when PLA is printed on glass, adhesion forces reach a plateau at 80\u0026deg;C, suggesting stabilization of intermolecular bonds [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. This may be why the material printed at a 70\u0026ndash;80\u0026deg;C bed temperature shows the highest UTS, Young\u0026rsquo;s modulus, tensile strain at break, and other mechanical properties. (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA-C)\u003c/p\u003e \u003cp\u003eANOVA of the UTS data yielded a p-value of 0.018 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05), affirming that bed temperature significantly impacts UTS. A similar analysis was conducted for Young\u0026rsquo;s modulus (YM) and percentage elongation at break, (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, C) depicts the influence of bed temperature on YM. Conversely, percentage elongation at break demonstrated a negative correlation with bed temperature, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC. Statistical analysis, indicating a p-value of 0.028, supported this observation.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe change in infill pattern was implemented to investigate its impact on mechanical properties. The average Young\u0026rsquo;s modulus (YM) was highest for triangular infill at 0.95\u0026thinsp;\u0026plusmn;\u0026thinsp;0.05 GPa and lowest for grid infill, also at 0.95\u0026thinsp;\u0026plusmn;\u0026thinsp;0.05 GPa. (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD). This disparity can be attributed to raster orientation: triangular infill follows 0\u0026deg;, 60\u0026deg;, and \u0026minus;\u0026thinsp;60\u0026deg; orientations, while grid infill follows 45\u0026deg; and \u0026minus;\u0026thinsp;45\u0026deg; orientations. Furthermore, ultimate tensile strength (UTS) reached its peak with full honeycomb infill at 12.87\u0026thinsp;\u0026plusmn;\u0026thinsp;1.0 MPa and was lowest for grid infill at 8.04\u0026thinsp;\u0026plusmn;\u0026thinsp;1.82 MPa, as demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e (D-F).\u003c/p\u003e \u003cp\u003eAll infill patterns have different raster angles, which significantly affect the ultimate tensile strength of the samples.[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] Percentage elongation at break was highest with grid infill at 7.73\u0026thinsp;\u0026plusmn;\u0026thinsp;1.55% and lowest with triangular infill at 4.11\u0026thinsp;\u0026plusmn;\u0026thinsp;0.35%. (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF) It\u0026rsquo;s notable that during loading, the 45\u0026deg; and \u0026minus;\u0026thinsp;45\u0026deg; raster orientations in the grid infill demonstrated greater potential to withstand the applied load, resulting in enhanced tensile properties. (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF) E. E. Cho \u003cem\u003eet al\u003c/em\u003e. claim that changes in layer height significantly impact tensile properties.[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e] For instance, in a full honeycomb structure, three raster angles for a single layer resulted in three layers being pressed into a 0.3mm space. This compression led to the formation of a compact layer, thereby increasing tensile strength [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. A similar observation was made by A. Rodr\u0026iacute;guez et al., who found that increasing the infill percentage led to proportional increases in ultimate tensile strength (UTS), elongation percentage, modulus, and yield stress.[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e] Furthermore, the study revealed that infill had a greater impact on these parameters than primary layer thickness and other printing parameters.[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.3. Microstructural analysis\u003c/h2\u003e \u003cp\u003eOne reason for the highest UTS at 70\u0026deg;C, compared to other print-bed temperatures, may lie in microstructural defects arising from uneven heat conduction through the printed layers. To further analyze, the morphological properties of the 3D-printed material were examined (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). After tensile testing, the fractured areas were imaged with FE-SEM, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e(A). Micron-sized voids were observed on the fractured portions of the samples, primarily between the gaps in adjacent filaments (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e(A) and (B)). The size and shape of the voids varied, with the size distribution shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD. Further, owing to the 45\u0026deg; and \u0026minus;\u0026thinsp;45\u0026deg; raster angles chosen, air gaps are observed at the junction of filaments, as shown in 4(C). The air gaps may occur due to printing speed, non-uniform filament melting, a high solidification rate, or improper filament arrangement [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] The large air gap may also lead to stress concentration sites, which can initiate crack propagation and early fracture of samples during mechanical testing. However, during cutting, some voids may have collapsed, leading to incomplete observation of all voids in the samples.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFrom the SEM images, it is observed that the number and size distribution of micro-voids are not uniform. If the bed temperature is high, then the solidification rate of the printed filament would be slow, leading to uniform cooling. Due to the high bed temperature, heat transfer from the bed to the printed samples will be high, potentially reaching the maximum layer. Furthermore, slow solidification of the extruded polymer melt improves adhesion between layers. All the samples prepared at four different temperatures have 7 layers. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows SEM images of the distribution of micro-voids in the three layers (bottom to top) of the four samples. All samples exhibit a rough fracture pattern, indicating that, during loading, this material resists breaking and thus has good mechanical strength. The average number of voids measured in the samples prepared at a bed temperature of 60\u0026deg;C is 12; it is 10, 9, and 8 for the samples prepared at bed temperatures of 70, 80, and 90\u0026deg;C, respectively. The observed voids ranged in diameter from 1.7 \u0026micro;m to 11.88 \u0026micro;m, with an average diameter of 4.13 \u0026micro;m. The cause of these voids may be the low extrusion temperature and the non-uniform cooling of the filament after extrusion [\u003cspan additionalcitationids=\"CR30\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. The formation of voids across the various printed layers is probably due to the fact that, during filament solidification, the outer layer solidifies first, and then the solidification propagates inward. However, this outer solidified area acts as a constraint or fixed boundary.[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe microstructural analysis showed that increasing the bed temperature reduced the number of voids in printed filaments, indicating improved filament fusion. (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e) However, excessive thermal exposure may promote crystallization and restrict chain mobility. Therefore, an optimal balance between reduced void density and sufficient chain mobility will result in enhanced mechanical strength of the printed components.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e3.4. Differential Scanning Calorimetry Analysis\u003c/h2\u003e \u003cp\u003eDifferential scanning calorimetry (DSC) was conducted to evaluate the thermal behavior of PLA samples printed at different bed temperatures. Approximately 6 mg of material was extracted from 4 mm printed specimens, ensuring consistent sampling. The DSC thermograms (heat flow vs temperature) exhibit three characteristic transitions of semi-crystalline polymers: the glass transition temperature (Tg), cold crystallization (Tcc), and melting temperature (Tm).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eHeat Flow characteristics of the different samples printed at varying print-bed temperatures\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSample\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTg (\u0026deg;C)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTm (\u0026deg;C)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eΔHcc\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eΔHm\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eXc (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePLA Spool\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e176.36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.92\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e60\u0026deg;C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e176.36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.23\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e70\u0026deg;C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e176.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e80\u0026deg;C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e176.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.04\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e90\u0026deg;C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e176.37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe PLA spool material shows a Tg of ~\u0026thinsp;65\u0026deg;C and Tm of ~\u0026thinsp;176\u0026deg;C, consistent with reported values. (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) The sample printed at 60\u0026deg;C exhibits thermal behavior similar to that of the PLA Spool or the not-printed sample, indicating minimal structural modification during printing. However, samples printed at higher bed temperatures (70\u0026ndash;90\u0026deg;C) exhibit reduced cold-crystallization peaks, suggesting partial crystallization during printing. This reduction in ΔHcc indicates that elevated temperatures promote in-situ molecular rearrangement, thereby decreasing the extent of post-print crystallization. The degree of crystallinity (Xc) increases from 0.23% at 60\u0026deg;C to a maximum of 1.10% at 70\u0026deg;C, followed by a slight decline at higher temperatures. Although the absolute values are low due to prior baseline correction, the relative trends remain. The 70\u0026deg;C sample demonstrates an optimal balance between amorphous and crystalline phases, as reflected in both thermal and mechanical behavior.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eNotably, the highest tensile strength is observed for the 70\u0026deg;C sample. This can be attributed to the proximity of the bed temperature to the Tg of PLA, where polymer chains possess sufficient mobility to enable interlayer diffusion and molecular entanglement.[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e] This enhances interfacial bonding between deposited filaments, improving mechanical performance. At higher temperatures (80\u0026ndash;90\u0026deg;C), increased thermal exposure may lead to premature crystallization or chain relaxation, limiting interlayer diffusion and reducing effective bonding, despite comparable melting behavior.[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e] These results highlight the critical role of processing temperature near Tg in governing the structure\u0026ndash;property relationship in FDM-printed PLA. (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB) However, identifying optimal process parameters through conventional experimentation is time-consuming and resource-intensive. Therefore, integrating artificial intelligence (AI) offers a promising approach to predict and optimize printing conditions by quantitatively mapping process\u0026ndash;structure\u0026ndash;property relationships.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e3.5. Machine learning modeling\u003c/h2\u003e \u003cp\u003eThe performance of various machine learning models was evaluated based on their predictive accuracy for the mechanical properties of 3D-printed samples. The input parameters considered were infill pattern, infill percentage, number of layers, bed temperature, actual thickness, and raster angle. The output parameters included Young\u0026rsquo;s modulus (YM), Ultimate Tensile Strength (UTS), maximum load, and yield stress. The dataset was split into a training set comprising 70% of the data and a testing set containing the remaining 30%. The random state was set to 42 to ensure consistent data splitting. The correlation heat map visually depicts the strength and direction of relationships between input parameters (such as infill pattern, infill percentage, and bed temperature) and output parameters (modulus, UTS, yield stress, and maximum load) in the 3D printing process, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e. Strong correlations are indicated by darker shades, providing insights into which input factors significantly influence the mechanical properties of printed parts. A strong correlation was found between input parameters such as infill pattern, infill percentage, raster angle, and number of layers, and modulus, UTS, yield stress, and maximum load. So, these input and output parameters were used to develop prediction models.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe selected machine learning models and their hyperparameters are listed in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. The performance of each model was assessed using Mean Squared Error (MSE) and R-squared values as evaluation metrics Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e] The Polynomial Regression model with degree 2 performed best among the tested models, with the lowest MSE of 0.003 and the highest R-squared of 0.888. This indicates that the Polynomial Regression model effectively captured the dataset's variance and accurately predicted mechanical properties based on input parameters. For small datasets, polynomial regression efficiently captures the nonlinear trends.[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e] Although other models showed reasonable performance, none surpassed the predictive accuracy of Polynomial Regression.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEvaluation matrix for different machine learning models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSl. No.\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eML Model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMean Squared Error\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eR-squared\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRidge Regression Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.011\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.726\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLinear Regression Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.758\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePolynomial Regression Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.003\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.888\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDecision Tree Regression Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.813\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRandom Forest Regression Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.005\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.863\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSupport Vector Regression Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.009\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.753\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e compares the machine learning models' predicted and measured mechanical properties of the 3D-printed material. The Decision Tree Regression and Random Forest Regression models also demonstrated competitive results, suggesting their potential usefulness in predictive models for 3D printing processes. Promising results for the Decision Tree Regression and the Random Forest Regression models were also observed by Jayasudha et al and Ziadia et al.[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e] Due to the limited dataset size, simpler models such as polynomial regression were preferred to avoid overfitting, although more complex ensemble methods, such as XGBoost, may achieve higher accuracy with larger datasets, as was reported by Alsakarneh et al. [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Conclusion and Future Work","content":"\u003cp\u003eThis study provides a comprehensive understanding of the mechanical properties of FDM-printed materials, focusing on microdefects, their formation mechanisms, and their influence on structural performance. Microscopic observations revealed variations in extruded filament dimensions across printed regions due to non-uniform cooling during deposition. These variations affect dimensional accuracy and promote air-gap formation at raster intersections, thereby influencing mechanical behavior. Statistical analysis confirmed that print-bed temperature significantly impacts mechanical properties. Among the infill patterns, the full honeycomb structure exhibited the highest ultimate tensile strength, the grid pattern showed maximum elongation, and the triangular pattern demonstrated the highest Young\u0026rsquo;s modulus. Fractographic analysis using FE-SEM revealed micro-voids within filaments, with void density decreasing at higher bed temperatures, indicating improved interlayer fusion. These voids form due to differential cooling, where the outer layer solidifies first and constrains the contracting inner region, leading to internal defects that act as stress concentrators. DSC analysis indicated weak glass transition and crystallization behavior, suggesting predominantly amorphous structures in printed PLA. The experimental dataset was used to develop machine learning models for predicting mechanical properties, with polynomial regression (degree 2) performing best (MSE\u0026thinsp;=\u0026thinsp;0.003, R\u0026sup2; = 0.888). Additionally, increasing the number of layers reduced the influence of bed temperature on the upper layers, altering cooling dynamics. Overall, the results highlight the strong interdependence between process parameters, microstructure, and mechanical performance, supporting the use of AI for predictive optimization in additive manufacturing.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eR.C , V.H, RV performed the experiments, RC and PK wrote the main manuscript, prepared the figures and table, HSN reviewed and edited the manuscript, PK conceptualized and supervised the work and edited the manuscript\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eVH, RV, and RC thank NIT Rourkela for financial support through MTech and PhD scholarships. P.K. acknowledges the Science and Engineering Research Board (SERB), Department of Science and Technology, Government of India, for providing financial support through a start-up research grant (SRG/2021/000859).\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAll data supporting the findings of this study are available within the paper and its Supplementary Information.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eDave HK, Patel ST (2021) Introduction to Fused Deposition Modeling Based 3D Printing Process. 1\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-030-68024-4_1\u003c/span\u003e\u003cspan address=\"10.1007/978-3-030-68024-4_1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePenumakala PK, Santo J, Thomas A (2020) A critical review on the fused deposition modeling of thermoplastic polymer composites. Compos B Eng 201:108336. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compositesb.2020.108336\u003c/span\u003e\u003cspan address=\"10.1016/j.compositesb.2020.108336\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eO\u0026rsquo;Connor L (2025) Comparative analysis of the mechanical properties of FDM and SLA 3D printed components. J Micromanufacturing. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/25165984251364689\u003c/span\u003e\u003cspan address=\"10.1177/25165984251364689\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatel KS, Shah DB, Joshi SJ et al (2024) Effect of process parameters on the mechanical performance of FDM printed carbon fiber reinforced PETG. J Mater Res Technol 30:8006\u0026ndash;8018. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jmrt.2024.05.184\u003c/span\u003e\u003cspan address=\"10.1016/j.jmrt.2024.05.184\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBellehumeur C, Li L, Sun Q, Gu P (2004) Modeling of Bond Formation Between Polymer Filaments in the Fused Deposition Modeling Process. J Manuf Process 6:170\u0026ndash;178. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S1526-6125(04)70071-7\u003c/span\u003e\u003cspan address=\"10.1016/S1526-6125(04)70071-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVanaei S, Rastak M, El Magri A et al (2023) Orientation-Dependent Mechanical Behavior of 3D Printed Polylactic Acid Parts: An Experimental\u0026ndash;Numerical Study. Machines 2023, Vol 11, 11:. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/machines11121086\u003c/span\u003e\u003cspan address=\"10.3390/machines11121086\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSudin MN, Daud NM, Ramli FR, Yusuff MA, Science (2023) Eng Technol 3:33\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.54327/set2023/v3.i1.71\u003c/span\u003e\u003cspan address=\"10.54327/set2023/v3.i1.71\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen G, Wang D, Hua W et al (2023) Simulating and Predicting the Part Warping in Fused Deposition Modeling by Thermal\u0026ndash;Structural Coupling Analysis. 3D Print Addit Manuf 10:70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1089/3dp.2021.0119\u003c/span\u003e\u003cspan address=\"10.1089/3dp.2021.0119\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmelot A, Colman M, Loret JE (2023) A thermodynamic framework for the additive manufacturing of crystallizing polymers. Part I: A theory that accounts for phase change, shrinkage, warpage and residual stress. Int J Eng Sci 183:103789. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.spinee.2018.01.019\u003c/span\u003e\u003cspan address=\"10.1016/j.spinee.2018.01.019\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEl Moumen A, Tarfaoui M, Lafdi K (2019) Modelling of the temperature and residual stress fields during 3D printing of polymer composites. Int J Adv Manuf Technol 2019 104(5):1661\u0026ndash;1676. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00170-019-03965-y\u003c/span\u003e\u003cspan address=\"10.1007/s00170-019-03965-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSingh B, Kumar R, Singh Chohan J (2022) Multi-objective optimization of 3D Printing process using genetic algorithm for fabrication of copper reinforced ABS parts. Mater Today Proc 48:981\u0026ndash;988. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.matpr.2021.06.264\u003c/span\u003e\u003cspan address=\"10.1016/j.matpr.2021.06.264\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e\u0026Ouml;zk\u0026uuml;l M, Kuncan F, Ulkir O (2025) Predicting Mechanical Properties of FDM-Produced Parts Using Machine Learning Approaches. J Appl Polym Sci 142:e56899. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/app.56899\u003c/span\u003e\u003cspan address=\"10.1002/app.56899\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei H, Tang L, Qin H et al (2024) Optimizing FDM 3D printing parameters for improved tensile strength using the Takagi\u0026ndash;Sugeno fuzzy neural network. Mater Today Commun 38:108268. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.mtcomm.2024.108268\u003c/span\u003e\u003cspan address=\"10.1016/j.mtcomm.2024.108268\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUlkir O, Akgun G, Karadag A (2025) Mechanical Behavior Prediction of 3D-Printed PLA/Wood Composites Using Artificial Neural Network and Fuzzy Logic. Polym Adv Technol 36:e70103. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/pat.70103\u003c/span\u003e\u003cspan address=\"10.1002/pat.70103\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLaurenti M, Bavasso I, Palazzi E et al (2025) Predicting the mechanical behavior in FDM printing of biopolymers through boosting artificial neural networks. Mater Des 257:114475. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.matdes.2025.114475\u003c/span\u003e\u003cspan address=\"10.1016/j.matdes.2025.114475\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChalgham A, Ehrmann A, Wickenkamp I (2021) Mechanical Properties of FDM Printed PLA Parts before and after Thermal Treatment. Polym 2021 13. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/POLYM13081239\u003c/span\u003e\u003cspan address=\"10.3390/POLYM13081239\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eİncesu R, Akderya T (2024) The Influence of Printing Speed and Temperature on the Mechanical, Absorptive, and Morphological Properties of PLA-Based Hybrid Materials Produced with an FDM-Type 3D Printer. Polymers 2024, Vol 16, 16:. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/POLYM16192771\u003c/span\u003e\u003cspan address=\"10.3390/POLYM16192771\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChac\u0026oacute;n JM, Caminero MA, Garc\u0026iacute;a-Plaza E, N\u0026uacute;\u0026ntilde;ez PJ (2017) Additive manufacturing of PLA structures using fused deposition modelling: Effect of process parameters on mechanical properties and their optimal selection. Mater Des 124:143\u0026ndash;157. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/J.MATDES.2017.03.065\u003c/span\u003e\u003cspan address=\"10.1016/J.MATDES.2017.03.065\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGunasekaran KN, Aravinth V, Kumaran CBM et al (2021) Investigation of mechanical properties of PLA printed materials under varying infill density. Mater Today Proc 45:1849\u0026ndash;1856. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/J.MATPR.2020.09.041\u003c/span\u003e\u003cspan address=\"10.1016/J.MATPR.2020.09.041\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSpoerk M, Gonzalez-Gutierrez J, Sapkota J et al (2018) Effect of the printing bed temperature on the adhesion of parts produced by fused filament fabrication. Plast Rubber Compos 47:17\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/14658011.2017.1399531\u003c/span\u003e\u003cspan address=\"10.1080/14658011.2017.1399531\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarkhad MS, Abu-Jdayil B, Mourad AHI, Iqbal MZ (2020) Thermal Insulation and Mechanical Properties of Polylactic Acid (PLA) at Different Processing Conditions. Polymers 2020, Vol 12, 12:. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/POLYM12092091\u003c/span\u003e\u003cspan address=\"10.3390/POLYM12092091\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, Zhao L, Fuh JYH, Lee HP (2019) Effect of Porosity on Mechanical Properties of 3D Printed Polymers: Experiments and Micromechanical Modeling Based on X-ray Computed Tomography Analysis. Polym 2019 11:11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/polym11071154\u003c/span\u003e\u003cspan address=\"10.3390/polym11071154\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiparoti S, Sofia D, Romano A et al (2021) Fused Filament Deposition of PLA: The Role of Interlayer Adhesion in the Mechanical Performances. Polym 2021 13:13:1\u0026ndash;18. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/polym13030399\u003c/span\u003e\u003cspan address=\"10.3390/polym13030399\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJabbari E, Peppas NA (1994) Polymer-polymer interdiffusion and adhesion. J Macromolecular Sci Part C 34:205\u0026ndash;241. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/15321799408009635\u003c/span\u003e\u003cspan address=\"10.1080/15321799408009635\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlgarni M (2021) The Influence of Raster Angle and Moisture Content on the Mechanical Properties of PLA Parts Produced by Fused Deposition Modeling. Polym 2021 13:13:1\u0026ndash;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/polym13020237\u003c/span\u003e\u003cspan address=\"10.3390/polym13020237\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGaneshkumar S, Kumar SD, Magarajan U et al (2022) Investigation of Tensile Properties of Different Infill Pattern Structures of 3D-Printed PLA Polymers: Analysis and Validation Using Finite Element Analysis in ANSYS. Materials 2022, Vol 15, 15:. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ma15155142\u003c/span\u003e\u003cspan address=\"10.3390/ma15155142\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRodr\u0026iacute;guez-Panes A, Claver J, Camacho AM (2018) The Influence of Manufacturing Parameters on the Mechanical Behaviour of PLA and ABS Pieces Manufactured by FDM: A Comparative Analysis. Mater 2018 11:11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ma11081333\u003c/span\u003e\u003cspan address=\"10.3390/ma11081333\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWickramasinghe S, Do T, Tran P (2020) FDM-Based 3D Printing of Polymer and Associated Composite: A Review on Mechanical Properties. Defects Treatments Polym 2020 12:12:1\u0026ndash;42. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/polym12071529\u003c/span\u003e\u003cspan address=\"10.3390/polym12071529\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGajjar T, Yang R, Ye L, Zhang YX (2024) Effects of key process parameters on tensile properties and interlayer bonding behavior of 3D printed PLA using fused filament fabrication. Progress Additive Manuf 2024 10(2):1261\u0026ndash;1280. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40964-024-00704-y\u003c/span\u003e\u003cspan address=\"10.1007/s40964-024-00704-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMencarelli M, Sisella M, Puggelli L et al (2025) Sensitivity Analysis of 3D Printing Parameters on Mechanical Properties of Fused Deposition Modeling-Printed Polylactic Acid Parts. Appl Mech 2025 6:6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/applmech6010017\u003c/span\u003e\u003cspan address=\"10.3390/applmech6010017\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBermudo Gamboa C, Mart\u0026iacute;n B\u0026eacute;jar S, Trujillo Vilches FJ, Sevilla Hurtado L (2023) Geometrical analysis in material extrusion process with polylactic acid (PLA)+carbon fiber. Rapid Prototyp J 29:21\u0026ndash;39. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1108/RPJ-09-2022-0294\u003c/span\u003e\u003cspan address=\"10.1108/RPJ-09-2022-0294\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAntony Samy A, Golbang A, Harkin-Jones E et al (2021) Prediction of part distortion in Fused Deposition Modelling (FDM) of semi-crystalline polymers via COMSOL: Effect of printing conditions. CIRP J Manuf Sci Technol 33:443\u0026ndash;453. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cirpj.2021.04.012\u003c/span\u003e\u003cspan address=\"10.1016/j.cirpj.2021.04.012\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOmer MAE, Shaban IA, Mourad AH, Hegab H (2025) Advances in interlayer bonding in fused deposition modelling: a comprehensive review. Virtual Phys Prototyp 20:2522951. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/17452759.2025.2522951\u003c/span\u003e\u003cspan address=\"10.1080/17452759.2025.2522951\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYin J, Lu C, Fu J et al (2018) Interfacial bonding during multi-material fused deposition modeling (FDM) process due to inter-molecular diffusion. Mater Des 150:104\u0026ndash;112. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.matdes.2018.04.029\u003c/span\u003e\u003cspan address=\"10.1016/j.matdes.2018.04.029\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRashidi HH, Albahra S, Robertson S et al (2023) Common statistical concepts in the supervised Machine Learning arena. Front Oncol 13:1130229. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/FONC.2023.1130229/FULL\u003c/span\u003e\u003cspan address=\"10.3389/FONC.2023.1130229/FULL\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChicco D, Warrens MJ, Jurman G (2021) The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Comput Sci 7:1\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7717/PEERJ-CS.623/SUPP-1\u003c/span\u003e\u003cspan address=\"10.7717/PEERJ-CS.623/SUPP-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNikzad MH, Heidari-Rarani M, Rasti R, Sareh P (2025) Machine learning-driven prediction of tensile strength in 3D-printed PLA parts. Expert Syst Appl 264:125836. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/J.ESWA.2024.125836\u003c/span\u003e\u003cspan address=\"10.1016/J.ESWA.2024.125836\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJayasudha M, Elangovan M, Mahdal M, Priyadarshini J (2022) Accurate Estimation of Tensile Strength of 3D Printed Parts Using Machine Learning Algorithms. Processes 2022, Vol 10, 10:. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/PR10061158\u003c/span\u003e\u003cspan address=\"10.3390/PR10061158\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZiadia A, Habibi M, Kelouwani S (2023) Machine Learning Study of the Effect of Process Parameters on Tensile Strength of FFF PLA and PLA-CF. Eng 2023, Vol 4, Pages 2741\u0026ndash;2763 4:2741\u0026ndash;2763. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ENG4040156\u003c/span\u003e\u003cspan address=\"10.3390/ENG4040156\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlsakarneh A, Obaidat S, Mumani AA, Tamimi MF (2025) Predicting the dynamic tensile response of FDM materials using machine learning. Discover Appl Sci 2025 8(1):51. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/S42452-025-08049-Z\u003c/span\u003e\u003cspan address=\"10.1007/S42452-025-08049-Z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Fused Deposition Modeling (FDM), Microstructural Defects, Fracture Morphology, Mechanical Properties, Process Parameters, Machine Learning","lastPublishedDoi":"10.21203/rs.3.rs-9557043/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9557043/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFused Deposition Modeling (FDM) is widely used in additive manufacturing, yet printed components often exhibit poor mechanical performance due to process-induced microstructural defects. This study investigates the influence of key printing parameters, particularly print-bed temperature, on the process–structure–property relationship in polylactic acid (PLA). Mechanical properties were evaluated by tensile testing, and microstructural features were analyzed by optical microscopy and FE-SEM. Thermal behavior was examined using differential scanning calorimetry (DSC), revealing that increasing bed temperature reduces cold crystallization and promotes partial in-situ crystallization. An optimal condition at 70 °C produced the highest tensile strength, attributed to enhanced interlayer diffusion near the glass transition temperature and reduced void formation. FE-SEM observations confirmed improved filament fusion at elevated temperatures. The experimental dataset was also used to train a machine learning model to predict mechanical performance from process parameters. This integrated approach provides valuable insights into optimizing FDM printing for improved structural integrity and reliability.\u003c/p\u003e","manuscriptTitle":"Process–Microstructure–Property Relationships in FDM-Printed PLA Components: An Experimental and Predictive Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-15 18:16:56","doi":"10.21203/rs.3.rs-9557043/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9dd716d6-040b-41cd-af52-1b00507c5524","owner":[],"postedDate":"May 15th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-10T07:35:19+00:00","index":14,"fulltext":""},{"type":"reviewerAgreed","content":"280444465365072340103144368530500003998","date":"2026-05-08T08:19:22+00:00","index":13,"fulltext":""},{"type":"reviewerAgreed","content":"192206822187570401436495804477945798824","date":"2026-05-06T16:27:15+00:00","index":11,"fulltext":""},{"type":"reviewersInvited","content":"5","date":"2026-05-06T15:42:00+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-29T11:43:54+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-29T11:43:00+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-05-15T18:16:56+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-15 18:16:56","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9557043","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9557043","identity":"rs-9557043","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}