Integrated FDM optimization with multivariate capability analysis for dimensional and compressive mechanical properties Authors: Moath Alatefi, Abdulrahman M Al-Ahmari, Mustafa Saleh, Abd Elatty E Abd Elgawad, Abdullah Yahia AlFaify Published: 2025 ABSTRACT AbstractFused deposition modeling (FDM) is an additive manufacturing (AM) technology capable of producing functional parts with complex geometries. However, optimizing both dimensional and mechanical quality characteristics is challenging due to the influence of multiple process parameters. This study aims to determine how key FDM parameters affect the multivariate dimensional and mechanical quality characteristics and to establish an integrated framework for optimizing these responses simultaneously. An experimental design based on response surface methodology (RSM) was implemented to optimize four process parameters: layer thickness, extruder temperature, plate temperature, and printing speed. Cylindrical PLA samples adhering to compression standards were fabricated, and both dimensional characteristics (length and diameter) and mechanical characteristics (compressive strength and modulus) were evaluated. Multivariate process capability indices (MPCIs) were then estimated to assess overall process capability. The results revealed that layer thickness and extruder temperature are the most influential parameters affecting MPCIs. The optimal dimensional MPCI was achieved with a low layer thickness, high extruder temperature, low plate temperature, and low printing speed. Also, the proposed model explained 88% of the total variability in mechanical MPCI. This research introduces an integrated RSM–multivariate process capability analysis approach for the simultaneous optimization of multiple correlated quality characteristics in FDM, which improves both dimensional precision and mechanical performance in AM processes.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-30098-y.Keywords: Additive manufacturing, Process parameter optimization, Response surface methodology (RSM), Multivariate quality characteristics, Process capability index (MPCI), PLA material, Statistical quality controlSubject terms: Engineering, Materials science, Mathematics and computing Introduction and literature review The quality characteristics of parts produced using FDM should be optimized for enhancing customer satisfaction and reducing process variation. The optimization process is conducted by investigating the FDM parameters that can affect the dimensional and mechanical properties of its outputs, such as layer thickness, extruder temperature, and printing speed1,2. Due to the multiple quality characteristics in the FDM process, multivariate process capability analysis is used to evaluate outputs using single index. Process capability analysis measures the ability of a process to produce products within desired specifications3. The first univariate process capability index was proposed by Kane3. It has been used to evaluate manufacturing process4, manufacturing systems5, assembly lines6 and additive manufacturing processes7,8. Moreover, many multivariate process capability indices (MPCIs) have been proposed and applied to evaluate process outputs with multiple quality characteristics9–11. The true performance of manufacturing process can be expressed using MPCIs12–14. In the context of evaluating multivariate additive manufacturing process, Moath et al15 have proposed a framework to evaluate FDM process with multivariate quality characteristics using MPCIs. Also Wadea et al16 examine the performance of metal additive manufacturing process through evaluating electron beam melting process using MPCIs. The FDM process has been investigated in the literature using different process outputs, such as dimensional17 and mechanical properties18. Researchers have explored various methods to optimize FDM fabrication conditions. For instance, a combination of a definitive screening design (DSD) and an artificial neural network (ANN) has been proposed19,20. This approach identifies optimal fabrication conditions for better dimensional accuracy and mechanical properties. Mahmood et al.21 examine the impact of process parameter variations on the geometrical characteristics of the produced parts using Taguchi’s design of experiment statistical technique. Taguchi’s design of the experiment has also been used by Alafaghani and Qattawi3, Moza et al.4, and Kumar et al.22. The primary impacts of four processing factors in the FDM process—the infill percentage, infill pattern, layer thickness, and extrusion temperature—are examined in Alafaghani and Qattawi23 work using Taguchi’s DOE. The findings indicated that a lower extrusion temperature, thinner layers, a lower infill percentage, and a hexagonal infill pattern were necessary for improved dimensional accuracy. Moreover, FDM and other material extrusion processes have several performance-related limitations which hinder part quality. The surface texture of printed components often varies significantly with build orientation and layer height, leading to non-uniform roughness, weak bonding, and pore formation across inclined and vertical surfaces24. Moreover, interlayer bonding strength is highly dependent on nozzle temperature and flow rate, where suboptimal bonding conditions can reduce flexural performance and increase surface irregularities25. Additionally, optimization studies on flexural strength and elastic modulus using advanced multi-objective algorithms confirmed that process parameters—particularly deposition angle, infill density, and temperature—govern the trade-off between mechanical stiffness and flexibility in PLA parts26. These challenges highlight that bonding quality, porosity, and surface finish remain critical factors influencing mechanical and dimensional behavior in material extrusion. Moreover, RSM has been widely used in the literature to optimize AM processes. It was first introduced by Box and collaborators27,28, and it has been applied in many fields, such as chemical process optimization29, the food industry30, and 3D printing31. RSM is a collection of computational and mathematical techniques for building experimental models. It works effectively for describing and assessing problems when several factors increase response and where the goal is to enhance it. It provides less experimental outcomes but more knowledge, nonetheless. The most investigated process parameter is layer thickness due to its significant effect on FDM process characteristics, as shown in Table 1. In addition to the layer thickness, the effect of extruder temperature has been widely studied32 Printing speed and plate temperature have a proven effect on dimensional deviations, of polylactic acid (PLA) printed parts, as stated by Abas et al.33. Many other process parameters have been investigated in the literature, such as infill density and pattern34,35 and table position36 Table 1FDM process parameters in literature.ParametersReferencesParametersReferencesLayer thickness (mm)21–23,32–34,36–40Infill pattern21,32,34–36,41Extruder temperature (°C)21,32,33,35,37,41Infill density (%)21,32–35,38,41Build Plate temperature (°C)21,33Building direction32,36Printing speed (mm/s)21,32,33Orientation angle22,33,34,36,37,39,40Chamber temperature (°C)21,42Table position36,43Number of shells21,33,35,38 FDM process parameters in literature. The characteristics of 3D printed parts in the literature include dimensional accuracy, geometric and mechanical properties, as shown in Table 2. Dimensional accuracy was the most investigated in the literature in the field of FDM process optimization21,22,32–39,41. Chohan et al.39 conducted research on the use of genetic algorithms to estimate the optimal value of dimensional accuracy for polymeric bio-composites and to optimize the FDM process parameters. Furthermore, Abas et al.33 and Qattawi et al.41 stated that 0.2 mm layer thickness is optimal for PLA 3D printed parts dimensional accuracy, whereas Vyavahare et al.11 stated that 0.1 is optimal for ABS material. Table 2FDM quality characteristics and materials.Response and materialReferencesResponse and materialReferencesDimensional accuracy21,22,32–39,41,51,52Time and Energy32,36,37Mechanical properties23,32,36,37,41,42,44–46PLA33,35,41,46Geometric characteristics21,47ABS34,37,46,48 FDM quality characteristics and materials. Furthermore, the effect of printing parameters of FDM process on mechanical properties have been investigated in many studies. In this regard, Ala’aldin et al.23 investigated the influence of key FDM process parameters on mechanical properties of PLA parts. Their results demonstrated that higher extrusion temperature, optimized layer thickness, and higher infill percentage enhance part strength. In a related study, Camposeco et al.36 conducted an experimental investigation on optimizing multiple FDM responses, including mechanical strength which demonstrated that the integrated desirability approach enabled simultaneous optimization of process parameters achieving a balanced improvement. Recent studies have extensively examined the influence of FDM process parameters on the mechanical performance of printed parts. Farashi et al.40 reported that reducing layer thickness and aligning the print orientation parallel to the build plate significantly improve tensile strength, while higher extrusion temperature stabilizes mechanical results. Panda et al.42 highlighted the role of extrusion velocity and layer thickness in deformation behavior, indicating their strong influence on part strength. Torres et al.44 found that layer thickness and infill density are key factors affecting torsional strength. Similarly, Sharma et al.45 demonstrated that layer thickness and infill percentage have the most substantial effects on tensile and compressive strength of ABS specimens. Kholil et al.46 showed that the maximum compressive strength was achieved at a 0.15 mm layer thickness for PLA, while thicker layers reduced strength. Popović et al.47 identified 190 °C extrusion temperature and 40 mm/min speed as optimal for maximizing tensile performance in PLA parts. Nathaphan et al.48 confirmed that low layer height, slow printing speed, and minimum shell count yield the highest compressive yield stress in ABS components. Stojković et al.49 emphasized that layer height dominates tensile strength improvement across materials, outperforming the influence of annealing time or temperature. Qi et al.50 demonstrated that incorporating CaCO₃ and TCP fillers into PLA enhances compressive modulus and strength in lattice structures,. Collectively, these studies underline that layer thickness, infill density, temperature, and orientation are the most critical determinants of mechanical performance in FDM-printed parts. Despite extensive studies on the influence of FDM parameters49,53, most previous works focused on optimizing single quality characteristics, either dimensional or mechanical. However, 3D-printed parts by the FDM process have multivariate characteristics. Also, the interdependence among these characteristics has rarely been addressed. The literature in this research revealed that, there is currently a lack of research on process capability analysis, which uses more than one quality characteristics to optimize the quality of AM process products. Therefore, this research introduces an integrated methodology that combines RSM and multivariate process capability analysis to simultaneously optimize multiple correlated quality characteristics and evaluate the overall process capability of the FDM process. With the existence of more than one quality characteristic, optimizing MPCIs ensures the process capability to produce parts within specification limits14,54. Furthermore, using multivariate tests revealed more realistic results due to the existing correlation between the product’s quality characteristics55. This study aims to determine the optimal process conditions that optimize MPCI of manufacturing processes. The research presents the FDM process, product design, and used materials in Section "Materials and methods". The third Section explains the proposed methodology. Sections "Process parameters" and "Manufacturing material" analyze and discuss the results of this research. Finally, Section "Research methodology" concludes this research and provides directions for future research. Materials and methods Printing process and device specifications An example of a part with specific features is shown in Figure 1. A 3D CAD model is first designed and exported as an STL file, which defines the tessellated surface of the planned object. Also, the object is sliced into horizontal layers resulting in a machine-readable file containing all the needed instructions for printing process. The printing parameters were changed according to the design experiment in section 2.2. Resulting in 30 combinations for FDM process optimization. The current study used RSM, which enables the detection of curvature in the response. Additive manufacturing workflow. An open-access FDM printer (Original Prusa i3 MK3S+, Prusa Research, Czech Republic) was used to fabricate all test specimens. The printer is equipped with a 0.4 mm brass nozzle, a build volume of 250 × 210 × 210 mm, and a heated bed capable of reaching up to 120 °C. The material used was PLA filament with a diameter of 1.75 mm. The printer operates using G-code generated through PrusaSlicer software, allowing precise control of process parameters such as layer thickness, extrusion temperature, plate temperature, and printing speed as shown in Figure 2. Utilized 3D printing machine. Process parameters Process parameters are adjusted during the slicing procedure, after CAD modeling. Table 3 presents the considered FDM parameters—layer thickness (LT), extrusion temperature (ET), plate temperature (PT), and printing speed (PS)—and their corresponding levels. The levels of these variables, as detailed in Table 3, were selected based on the supplier’s recommendations for the used material (PLA)56, the FDM printer’s capability, and findings from the literature. For instance, the LT range (0.15–0.25 mm) was selected as reported in57–59. The ET range (195–225 °C) was chosen according to the supplier’s recommendations56 and previous studies59,60. The PT range (55–65 °C) was determined based on preliminary experiments, supplier guidance56, and58. Finally, the PS range (30–60 mm/s) was selected according to the printer’s capability and58. Other FDM parameters were kept constant, including a nozzle diameter of 0.4 mm, raster width of 0.45 mm, raster angle of ±45°, and an infill density of 100%. Table 3The process parameters and their respective levels.Parameter−101Layer thickness (LT), mm0.150.20.25Extruder temperature (ET),°C195210225Build plate temperature (PT), °C556065Printing speed (PS), mm/s304560 The process parameters and their respective levels. Manufacturing material Polylactic acid (PLA) filament was used in this study. It is produced through lactic acid condensation and water evaporation or by polymerizing it through a ring-opening mechanism. PLA is produced from sustainable sources, however, it has not been widely used due to structural and operational issues61. It is the ideal substance for 3D printing applications due to its thermal and mechanical characteristics when it has been annealed. Ethyl acetate can be used in vapor chambers to smooth PLA surfaces because of its low boiling point. This study used 1.75 mm-diameter PLA filament (Prusament PLA Galaxy Silver filament), acquired from Prusa Polymers in the Czech Republic. Research methodology The steps of the research methodology are shown in Table 4. Based on the previous discussion in section 2.2, FDM process parameters have been selected for optimization. Dimensional accuracy and mechanical properties of the FDM process are optimized in this research. Therefore, two dimensions (length and diameter) and two mechanical properties (compressive strength and modulus) have been selected to be investigated as quality characteristics. According to the literature discussion, layer thickness (LT), extruder temperature (ET), Plate temperature (PT), and printing speed (PS) are the most common process parameters used to optimize the FDM process. Table 4Research methodology steps.StepDescription1FDM process parameters selection2Identify Multivariate quality characteristics to be optimized.3Review the process parameters and ranges in the literature.4Designing the experiments using response surface methodology5Printing process6Measuring identified Multivariate quality characteristics7Estimating MPCIs for experiment8Analyzing MPCI data and identifying best process operation conditions Research methodology steps. The proposed methodology used RSM to produce 3D parts under different operation conditions. Then, the multivariate dimensions and mechanical properties of each combination are measured. Both multivariate dimensional and mechanical quality characteristic data were used to estimate MPCIs for each combination. MPCIs for dimensional and mechanical properties are used separately as responses data to get RSM results to optimize the process parameters. The factorial experiment is used to investigate the main effect and interactions of process parameters. Detailed discussions of the design of experiments and MPCI estimation are explained in the next sections. Research design Figure 3 shows the research design in which the process parameters have been identified and investigated using RSM design. Furthermore, the dimensional and mechanical quality characteristics are measured to be assessed later using MPCIs. The multivariate quality characteristics data were collected for each combination to be expressed as a single value (MPCI). Research design. This research aims also to model and interpret the variation of the 3D printing process by statistical models that describe the relationships between a manufactured product’s quality and process parameters. Determining the most significant factors can lead to process optimization by selecting the optimal values of process parameters that maximize the MPCIs. Multivariate process capability index estimation The majority of 3D-printed parts have multiple correlated quality characteristics, emphasizing the need to analyze them together to produce a single quality index. A multivariate capability index estimation algorithm is presented in this study. The algorithm uses the following equation first to standardize quality characteristics data: p represents the number of quality characteristics, is the standardized data for the ith quality characteristic, denotes the original data for the ith quality characteristic, represents the estimated mean of the ith quality characteristic, and represents the estimated standard deviation of the ith quality characteristic. also, quality characteristic mean and standard deviation are used to standardize the specification limits of corresponding quality characteristic. this ensures that all of the quality characteristics data have a standard normal distribution, with zeros mean vector and correlation matrix forming a multivariate normal data. standard multivariate normal distribution is used to generate very large sample data using matlab software. next, a comparison is made between each generated standard multivariate normal vector and the specification limits vector. any generated vector that deviates from the parameters is noted as nonconforming. thus, the number of nonconforming vectors is divided by the sample size to calculate the percentage of nonconforming (PNC): Finally, MPCI is estimated using PNC using the following equation62: Where is the inverse cumulative density function of standard normal distribution (CDF). Note, the above equation is changed for an upper-sided process as follows: Response surface methodology (RSM) In order to print strong parts with accurate dimensions, the parametric analysis, and FDM parameter optimization were conducted using RSM. MPCI is used as a response to optimize the considered process, which is computed in equation (3). Through the use of experimental design, the impact of the FDM parameters on dimensional accuracy and mechanical properties was assessed. This study used RSM’s face-centered central composite (FCC) design, a three-level, second-order experimental design, to examine how the four chosen FDM parameters affect the output response (MPCI). The RSM design with the four FDM printing parameters (i.e., LT, ET, PT, and PS) listed in Table 3 resulted in 30 experiments, with five repetitions for each experiment. The 30 RSM experiments were composed of 16 cube points, 8 axial points, and 6 center points. In the RSM, a polynomial model is typically used to represent the actual relationship between process parameters and responses as follows63: However, the higher order of model equations should be considered in the presence of curvature. Where y is the response, are process parameters, are quadratic and interaction terms respectively. Results and discussion Dimensional accuracy optimization MPCI results and preliminary analysis The design of RSM revealed a total of 30 combinations, as shown in Table 5, and five samples in each combination were printed. The height and diameter of each printed sample have been measured. Therefore, the data of each combination was used to simulate large samples that were used to estimate the MPCI of the combination. The results of MPCIs are presented in Table 5, which range from 0.07 to 2. It is worth noting that the process is considered capable of producing parts with the required specifications if the MPCI is greater than one. In this regard, the worst MPCI is resulted when the FDM process set to a higher layer thickness (0.25 mm), low extruder temperature (195 ), low plate temperature (55 ), and low printing speed (30 mm/s). At the same time the best(higher) MPCI was obtained when the process parameters set to a low layer thickness (0.15 mm), high extruder temperature (225 ), low plate temperature (55 ), and low printing speed (30 mm/s). Table 5Estimated MPCIs.PSETPT556065LT0.150.20.250.150.20.250.150.20.25301950.920.071.160.812101.482252.000.561.150.52451951.112101.731.601.240.651.321.571.261.371.401.392251.33601951.150.661.810.872101.492251.501.451.060.81 Estimated MPCIs. Table 6 shows the averages and standard deviations of MPCIs for each factor level. The results show that the medium level of layer thickness factor is the most capable setting for the FDM process (MPCI=1.393). Also, it is more stable when interacting with factors, which is indicated by the low variability of the results at this level (STD = 0.164). Generally, low and medium levels of layer thickness have the most MPCI, whereas a high level of layer thickness factor is not capable (MPCI=0.711). Furthermore, the levels of extruder temperature revealed that low extruder temperatures are not capable of producing parts within the specification limits, whereas medium and high extruder temperatures are marginally capable. It is worth noting that all plate temperature levels are either marginally or reasonably capable of producing parts within specification limits. The results of the plate temperature could be an indicator of the lack of significant effect of different plate temperatures in MPCI estimation. However, different levels of printing speed parameter revealed different MPCI estimations. Showing that the process is not capable at the low level of printing speed, reasonably capable at medium and marginally capable at high level. Table 6Level’s means and standard deviations.LTETPTPrinting speedLevel0.150.20.25195210225556065304560Mean1.41.40.70.951.381.151.121.331.060.961.331.20STD0.40.20.40.4650.30.50.60.30.40.60.30.4 Level’s means and standard deviations. RSM results Analyzing the effect of FDM process parameters on MPCIs revealed that there are significant relations between most of process parameter and estimated MPCI, as shown in the RSM results in Table S 1 in the supplementary file. The results showed that the layer thickness, extruder temperature and printing speed have a significant effect on MPCI (p-value < 0.05). Moreover, the interaction effects of the quadratic and two factors on process parameters have been investigated to predict the optimal effect of process parameters on MPCI. The only significant quadratic effect among all parameters is the effect associated with layer thickness. Also, the interaction of layer thickness with printing speed resulted in a significant effect on MPCI. Also, the plate temperature has a significant effect on MPCI only when it interacts with the extruder temperature. We can rely on the results and model of this experiment because the lack of fit in the model is not significant (p = 0.165). MPCI prediction model The regression equation, represented by formula (7), contains the coefficients corresponding to each investigated input variable used in the FDM process. These coefficients express the effect strength and direction of each input variable on MPCI. An increase in MPCI would be predicted by a positive coefficient, whilst a decrease in MPCI would be predicted by a negative coefficient. The summary of the model fitting is presented in Table S2 in the supplementary file. The standard deviation of the values between actual and fitted values is represented by s= 0.20. Also, R-squared value shows that about 82% of the variation in the data is explained by the model. Predicted R2, which is used to predict the actual values, is about 59%. Model adequacy RSM depends on many assumptions that must be met in order to draw valid conclusions from these tests. Normality assumption is the most important in statistical analysis. In this research, the normal probability plot of residuals is illustrated in Figure 4, in which the residuals of the RSM test are plotted against normal probability. Since the plotted points form a straight line, the researcher could not ensure that the data departed from a normal distribution. Furthermore, RSM assumes that the residuals have constant variances over the response range. However, in this research, the variance is not exactly equal over the response range, but it has no increasing or decreasing pattern. Therefore, a constant variance could be assumed, and there were no clear outlier points. As illustrated by Figure 4, in the residual versus order plot, the residuals are fairly distributed around the zero value. There is no increasing or decreasing pattern which ensures the randomness of errors. Model adequacy. Main factor and interaction effects The main effects of the process parameters have been investigated in order to identify the best process setting conditions. Therefore, Figure 5 shows the plots of the main effects of four process parameters on MPCI, which revealed that different layer thickness levels have a curvature pattern when estimating MPCI. Most MPCI results when the layer thickness is medium or low, whereas it decreases with increasing layer thickness. In contrast, extruder temperature, plate temperature, and printing speed seem to have a linear effect on MPCI. Both extruder temperature and printing speed have a positive relation with MPCI; that is, increasing extruder temperature or printing speed will increase MPCI. Also, plate temperature showed a linear negative relation with estimated MPCI. Main effect plots. Also the interaction effects of different process parameters should be studied to optimize FDM process in terms of MPCI estimation. The RSM model showed that there are only two significant interaction effects, as shown in Figure 6. The left side of the figure shows the interaction effects of layer thickness with other factors on MPCI. Layer thickness levels are indicated in the bottom left corner on the X-axis, whereas other factors are indicated by the graph color keys on the right of the graph. The results revealed that higher MPCI could be found when setting lower layer thickness with medium extruder temperature, low plate temperature, and high printing speed. The only significant interaction of layer thickness is with printing speed. Moreover, extruder temperature has a significant interaction with plate temperature, and the results recommend setting a high extruder temperature with a low plate temperature for better MPCI. Interaction plots. Process analysis Figure 7 shows contours and surface plots for all pairs of process parameters, which are used to visualize and model the changes in response variables with two input parameters. The figure includes a plot key to distinguish the desired levels of parameters that maximize MPCI. It shows there is a possibility for obtaining very high MPCI (> 1.75) with a high level of extruder temperature and a relatively low level of layer thickness. However, the RSM model revealed that the interaction of extruder temperature and layer thickness is not significant. Moreover, contour plots showed that speed pairs with all other parameters could reveal high optimum MPCI. It is worth noting that the low levels of layer thickness always lead the process to produce products with dimensions far away from the nominal, regardless of the other parameter settings. (a) Contours plots. (b) Surface plots. Furthermore, the surface plots show that pairs of layer thickness with all other process parameters always contain curvature. This indication ensures the importance of including layer thickness in optimizing the FDM process. Surface plots showed the robustness of relatively low levels of layer thickness. Regarding the highest MPCI, it could be predicted from surface plots that it is potentially high at a low level of plate temperature and high printing speed. Process optimization Figure 8 shows the process response patterns of all considered process parameters. Figuring out the response pattern allows the researcher to identify potential values of process parameters that would result in high and low-sensitive MPCIs. Table 7 presents the optimized parameter settings and corresponding statistical outputs (Fit, SE Fit, Confidence and Prediction Intervals) derived from the RSM optimization model, in which the average MPCI is 1.919, whereas the range of confidence interval ranges from 1.6 to 2.1. Optimized FDM process parameters. Table 7FDM process parameters for optimal MPCI.Optimal settingsLT (mm)ET (°C)PT (°C)Speed (mm/s)0.1832255560ResponseFitSE fit95% CI95% PIMPCI1.9190.115(1.680, 2.158)(1.426, 2.411) FDM process parameters for optimal MPCI. The optimization process was conducted by varying the values of process parameters outside the experimental ranges and estimating MPCI. Consequently, the optimization algorithm predicted so many different combinations of process parameters with their corresponding MPCI as shown in Table S3 in the supplementary file. Therefore, the highest setting recommends a higher value for the layer thickness, with high extruder temperature, low plate temperature, and high speed, resulting in a very high MPCI. This is not usual since all indicators discussed previously showed that low layer thickness is preferred for high MPCI. All other process parameter combinations suggest using low levels of layer thickness and different levels of other parameters. Model validation To further validate the robustness of the optimization approach, additional FDM samples were printed using different combinations of the process parameters. These new specimens were subjected to the same quality characteristic measurements, and the results were compared to the predictions made by the optimized RSM model. Table S4 in the supplementary file shows the parameters of validated samples. The estimated MPCI values for the validation samples are very close to predicted MPCIs values as shown in Figure 9. Consequently, the proposed model can be used to precisely predict MPCIs based on different FDM parameters. Comparison of predicted and actual MPCIs. Mechanical properties optimization Mechanical experiments Compression testing of PLA materials involves subjecting samples to compressive forces to evaluate their mechanical properties, such as compressive strength and modulus of elasticity. It’s used to assess the suitability of PLA for various applications, such as 3D printing. This research is optimizing the compressive strength and modulus of elasticity of 3D printed parts that are produced using the FDM process and PLA material. Compressive strength refers to the maximum compressive stress a specimen can sustain without failure. Moreover, the modulus of elasticity or Young’s modulus, is a measure of a material’s stiffness or resistance to deformation under an applied load. It quantifies the relationship between stress and strain in the elastic region of the material’s stress-strain curve. The compression test has been conducted using the Zwick\Roell compression testing machine shown in Figure 10. Zwick/Roell compression testing machine. Compression testing ensures the quality and consistency of PLA materials used in various applications by establishing standards for compressive strength and modulus of elasticity. However, the compressive strength and modulus of elasticity are affected by the 3D printing process parameters. Therefore, experiments need to be proposed that investigate the effect of varying the process parameters on the compressive strength and modulus of elasticity. Compression testing starts by preparing PLA samples using the FDM process for the desired shape and size. Based on the length-to-width aspect ratio specified in the ASTM D695-15 standard for compressive testing of rigid polymeric materials, ensuring that the length is twice the sample width50. The designed part used in this research is shown in Figure 1, which is a cylindrical part of diameter 12.7 mm and length 25.4 mm. The sample is placed between two flat plates of a compression testing machine. Ensure that the platens are aligned properly and the sample is centered, as shown in Figure 11 a. Moreover, a small initial load is applied to the sample to ensure proper contact between the platens and the sample, as in Figure 11 b. Finally, a compressive force is applied to the sample at a constant rate until deforming 30% of the sample as in Figure 11 c. Record the applied force and the corresponding deformation or strain. Compression test process. Furthermore, the compressive strength is calculated using the recorded data and following formula: Estimating MPCIs requires establishing specification limits for the compressive strength and modulus of elasticity characteristics. Since the greater the better compressive strength and modulus of elasticity, only lower specification limits will be designed. The lower specification limits are calculated using the natural tolerances of the process. In this regard, three sigma specification limits are calculated using the mean and standard deviation of the collected data of all combinations. For example, the lower specification limit for a given quality characteristic is calculated as follows: Where LSL is the lower specification limit of a given quality characteristic; μ is the mean of quality characteristic data; σ is the standard deviation of quality characteristic data; and is the sample size of each combination. MPCIs results The compression test was applied to the 30 combinations of the proposed RSM model. Five replications were tested in each combination. The results of the compression test are the strength and modulus, which have been used to estimate the MPCIs of mechanical properties (Quality characteristics) for each combination, as shown in Table 8. Table 8MPCIs of mechanical properties.No.Process parametersMPCINo.Process parametersMPCILTETPTPSLTETPTPS10.221055450.468160.221060452.00020.221060301.032170.2522565600.53630.1521060452.000180.1522555601.31840.222560452.000190.1519555300.16350.221060452.000200.1522555302.00060.219560450.736210.1519565602.00070.221060452.000220.2519565600.00080.2521060451.308230.1519565300.68290.221060602.000240.221060452.000100.221065451.557250.2519555600.000110.251956530−0.392260.221060452.000120.2522555600.546270.221060450.701130.251955530−1.209280.2522565302.000140.1522565302.000290.2522555301.343150.1522565602.000300.1519555602.000 MPCIs of mechanical properties. RSM results The ANOVA results of MPCI for mechanical properties indicate that the model is highly significant, explaining 88% of the total variability in the response, as shown in Table 9 (model’s p < 0.001 and a high F-value of 19.27). The linear terms account for 54% of the variability. Layer thickness factor is the most influential, explaining 26% of the total variability followed by extruder Temperature (25%). While the printing speed (PS) and plate temperature (PT) factors have a minor effect and are not significant. Table 9ANOVA results of MPCI for mechanical properties.SourceDFSeq SSContributionAdj SSAdj MSF-valueP-valueModel825.218288.01%25.21823.1522719.270Linear415.346953.56%15.34693.8367323.450LT17.483326.12%7.48337.4832945.740ET17.141824.92%7.14187.1418143.650PT10.64112.24%0.64110.641093.920.061PS10.08070.28%0.08070.080720.490.49Square14.499215.70%4.49924.4991827.50PT*PT14.499215.70%4.49924.4991827.502-way interaction35.372118.75%5.37211.7906910.940LT*ET11.64425.74%1.64421.6442310.050.005LT*PS11.3854.83%1.3851.385038.460.008ET*PS12.34288.18%2.34282.3428114.320.001Error213.436111.99%3.43610.16362Lack-of-Fit172.59289.05%2.59280.152520.720.718Pure error40.84332.94%0.84330.21083 ANOVA results of MPCI for mechanical properties. The nonlinear square term for PT contributes 16% to the total variability and is highly significant, and the 2-way interaction terms contribute 19% indicating the importance of nonlinear effects in the model. The error term accounts for 12% of the total variability, while the lack-of-fit is not significant (p-value = 0.72), indicating that the model fits the data well. The model summary indicates that the regression model has a strong fit, with an R-squared value of 88%, as shown in Table 10. However, the predicted R-squared is 78%, which is notably lower than the R-squared, suggesting that while the model fits the current data well, its ability to predict new, unseen data is limited. Table 10Model summary for MPCI of mechanical properties.SR-sqR-sq(adj)PRESSR-sq(pred)0.488.01%83.44%6.377.99% Model summary for MPCI of mechanical properties. The regression equation demonstrates that LT has the most substantial negative impact on MPCI. This is indicated by its high negative coefficient, suggesting that increases in LT significantly decrease MPCI. PT also positively affects MPCI, while Speed and ET have a minor positive impact on MPCI. The interaction between LT and ET is positive, showing that the effect of LT on MPCI is strengthened when ET is higher. Overall, the equation highlights the complex interplay of these variables in influencing MPCI. Response optimization The main effects plot for MPCI, in Figure 12, shows that Layer Thickness (LT) exhibits a linear relationship, where MPCI decreases with increasing LT. Extruder Temperature (ET) displays a positive linear trend, indicating that higher temperatures improve MPCI, likely due to better material flow and bonding. Conversely, Plate Temperature (PT) shows a nonlinear relationship, suggesting that higher and lower temperatures may slightly reduce MPCI. Speed of printing has a positive linear impact. These insights imply that balancing LT and PT while maximizing ET and Speed, could lead to optimal process performance. Main effect plot of MPCI for mechanical properties. The interaction plot for the MPCI, in Figure 13, illustrates that while MPCI decreases as LT increases, higher extruder temperatures can mitigate this negative impact, suggesting an optimal combination of high ET with low LT. The LT and Speed interaction suggests that higher speeds are more beneficial at lower LT, while their positive impact diminishes with increased LT. The ET and Speed interaction indicates that high extruder temperatures combined with lower speeds yield the best MPCI. Interaction plot of MPCI for mechanical properties. The optimization plot for the MPCI in Figure 14 indicates that the maximum MPCI can be achieved with a desirability of one. Specifically, the optimal LT is around 0.15 mm, showing that a low layer thickness is ideal while increasing LT leads to a decrease in MPCI. The ET should be set to its lowest value of 195 °C, as lower extruder temperatures positively influence MPCI. For PT, the plot suggests a modest setting of 60 °C, which helps prevent potential quality issues such as warping that might occur at higher temperatures. Lastly, the optimal Speed is at its maximum value of 60 mm/s, indicating that higher speeds contribute to better process performance by enhancing efficiency. These optimized settings maximize MPCI, ensuring high-quality and consistent production. Optimization plot for the MPCI. Insights and implications The significant impact of layer thickness (LT) is critical in determining the mechanical robustness of the printed components. The observed nonlinear relationship indicates that there is an optimal LT value that maximizes MPCI. This finding aligns with existing literature, which suggests that thicker layers may introduce defects or inconsistencies in bonding, leading to weaker mechanical performance. The consistent increase in MPCI with higher ET levels suggests that elevated temperatures facilitate better material flow and bonding, which are crucial for enhancing compressive strength and modulus of elasticity. This result highlights the need for precision in maintaining optimal temperature settings. However, while ET had a clear positive effect, the marginal impact of PT suggests that its role might be more complex. Although higher PT levels slightly decreased MPCI, possibly due to issues like warping or uneven cooling. Also, its significance in interaction with other parameters like ET indicates that PT cannot be overlooked. This nuanced impact necessitates further exploration to understand its role in multi-faceted printing scenarios. The interactions between PT and Speed suggest that while higher speeds generally improve MPCI, this benefit is maximized only when PT is carefully controlled. This interaction highlights the importance of a holistic approach in parameter optimization, considering not just the individual effects but also how these parameters influence each other. The study’s optimization findings provide practical guidance for the FDM printing of PLA parts. The identified optimal settings not only maximize mechanical performance but also enhance production efficiency. The application of these optimized parameters could lead to more consistent product quality and improving overall process sustainability. Overall, this study demonstrates the utility of using MPCI as a comprehensive metric for evaluating and optimizing the mechanical properties of 3D-printed components. Comparing research’s results with previous research The results of this study are consistent with previous research emphasizing the strong influence of layer thickness and extrusion temperature on the quality of FDM-printed parts. Similar to the findings of Ala’aldin et al.23, Nathaphan et al.48, and Farashi et al.40, lower layer thickness significantly improved dimensional accuracy and mechanical strength, while higher extrusion temperatures enhanced material bonding and reduced interlayer voids. However, our results provide a more comprehensive insight by quantifying these effects through the MPCI, showing that optimal dimensional capability is achieved at a low layer thickness, high extrusion temperature, low plate temperature, and low printing speed. Consistent with Sharma et al.45 and Torres et al.44, the influence of printing speed and interaction effects among parameters were also confirmed to be significant. Unlike earlier works that optimized individual responses, the proposed RSM-multivariate process capability analysis integration enables simultaneous optimization of dimensional and mechanical quality characteristics, demonstrating an enhanced capability to achieve overall product precision and performance stability. Conclusion This research demonstrated a novel and integrated approach to optimizing the FDM process by combining response surface methodology (RSM) with multivariate process capability analysis. Unlike traditional studies that optimize each quality characteristic separately, the proposed methodology simultaneously considers multiple correlated quality characteristics, providing a more realistic assessment of overall process capability. The findings revealed that layer thickness and extruder temperature are the most influential parameters affecting both dimensional and mechanical performance. These models were further validated experimentally, confirming the reliability of the proposed approach. However, the study is limited to a specific polymer material (PLA) and a defined range of process parameters. Extending the analysis to other materials, geometries, and additive manufacturing technologies would enhance the model’s predictive robustness. Additionally, incorporating other multivariate and intelligent optimization techniques—such as machine learning or adaptive control—could further improve process prediction and automation. From an industrial perspective, the proposed methodology offers a practical framework for manufacturers to evaluate and optimize FDM processes more systematically. It can be implemented for process monitoring, parameter tuning, and quality assurance, leading to improved dimensional precision and mechanical strength. This work thus contributes to advancing data-driven quality optimization and capability assessment in additive manufacturing environments. Supplementary Information Supplementary Information. Author contributions Conceptualization, M.A. and A.M.A.; methodology, M.A. and A.A.; software, M.A. M.S. and A.E; validation, A.M.A., A.A., and M.S.; formal analysis, M.A.; investigation, M.A. and M.S.; resources, M.S. and A.E; data curation, M.A.; writing—original draft preparation, M.A.; writing—review and editing, A.A.; visualization, M.A.; supervision, A.M.A. and A.A.; project administration, A.M.A. and A.A.; funding acquisition, A.M.A. and A.A. All authors have read and agreed to the published version of the manuscript. All authors have read and agreed to the published version of the manuscript. Funding “This study received funding from the Raytheon Chair for Systems Engineering. The authors are grateful to the Raytheon Chair for Systems Engineering for funding.” Data availability The data presented in this study are available in this article. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Information. Data Availability Statement The data presented in this study are available in this article.