Optimization of Mechanical Properties and Surface Characteristics of PLA + 3D Printing Materials

,


Introduction
Recently, additive manufacturing (AM) attracts many researchers and manufacturers to investigate and utilize this promising technology [1]. Many merits are provided by AM that enlarge demand for this process such as improving productivity, producing intricate parts, and minimizing warehouses and waste materials [2,3]. Also, a wide range of materials and techniques can be employed in the AM. Accordingly, this technology has found a good market in many sectors such as automotive, defense, medical, and aerospace due to its fexibility compared with conventional technologies [4][5][6][7]. In general, some traditional manufacturing processes involve material removal (as in metal cutting processes) to produce the desired geometry of the machined part. Terefore, annually industrial sectors in various felds lose a huge amount of materials in the form of chips. In contrast, additive manufacturing adds materials through a layering process to produce the required part with dimensions according to the specifcation [8]. Fused deposition modeling (FDM) is one of the most applied additive manufacturing (AM) processes, in which melted flament is extruded through a nozzle and deposited on the platform to produce layered products immediately from the part CAD model [9]. FDM is applied in diferent areas, particularly the biomedical sector, to process prostheses, implants, drugs, etc. [10]. FDM process utilizes various kinds of thermoplastic flaments with round crosssections such as polylactic acid (PLA), improved polylactic acid (PLA+), polyethylene terephthalate glycol (PETG), and acrylonitrile butadiene styrene (ABC). PLA biodegradable material is processed from grown plants, including corn, cassava, and potato, using bacterial fermentation [11]. PLA+ is a modifed version of PLA with good impact resistance and adherence between printed layers, making it suitable for 3D printing functional products. It demonstrated its potential as biomaterials in various medical applications, including cardiovascular implants, regenerative medicine or tissue engineering, orthopedic treatments, cancer therapy, dental specialties, drug carriers, skin and tendon mending, and medical equipment and tools [12,13].
Te bonding between two successive layers on the print platform occurs in FDM at four diferent stages as follows: contact between surfaces, growth of neck, difusion interface, and randomizing [14].
Optimization plays a vital role in manufacturing as well as almost other sectors [15]. A set of works have been published previously in the feld of 3D printing parametric and optimization studies. Te most studied response of the fnal 3Dprinted product was mechanical properties. Some parameters were studied by the researchers in the literature besides printing temperature and speed like raster angle, infll density (%), infll pattern, and layer height. Te raster angle refers to the angle confned between the deposited raster path and the axes of the base platform. Te infll density stands for the volume of deposited material inside the part being 3D printed. Te geometry and structure of the deposited material inside the part represent the infll pattern. Te layer height is the thickness of the individual successive deposited layer. For instance, the infll density (%), printing temperature, raster angle, and layer height have been examined by Leite [16] to show their impact on the obtained ultimate tensile strength (σu), Young's modulus (E), elongation%, and toughness of PLA tensile specimens. Twenty-four experimental runs were carried out and the results were analyzed statistically by using analysis of variance (ANOVA). Te mechanical properties of ABS 3D-printed parts were improved via optimizing the layer thickness and infll pattern by using the Taguchi approach [17]. Physical and mechanical properties of four resins were investigated by Christian and Ezekielle [18]. Te ultimate tensile and bending strengths were evaluated.
Te PLA flaments are preferred over ABS ones [19] in terms of strength and stifness and 3D printability, but both are most popular 3D-printed materials and considered by previous studies [20][21][22].
Te mechanical properties of some 3D-printed materials were investigated by Tanikella et al. [23]. Te external and internal textures of the 3D-printed specimens were tested to fnd out the best layer height taken into account, the specimen mass, and extrusion variables. Te mechanical properties of PLA material were verifed with the fndings of the fnite element simulation [24]. Also, Lubombo and Huneault [25] investigated the PLA mechanical properties at low infll density. Te performance of PLA material was improved in terms of mechanical and electrical properties by reinforcement with carbon nanoparticles (CNPs) [26]. Te infuence of diferent percentages of infll density and printing orientation of PLA+ specimens were also studied [27]. In another study, the impact of printing angles and volume fraction of wood-PLA flament composite were assessed based on the obtained mechanical properties [28]. 25% of wood flaments were more efective on the mechanical properties of the produced composite.
A fused deposition modeling was applied to 3D print of PLA at printing speed, temperature, number of layers, and thickness of 200 mm/s, 200°C, 30 layers, and 0.2 mm, respectively [29]. Te specimens were printed with grid, trihexagon, triangle, and quarter cubic patterns at 60% infll density and have been examined with compression and low rate impact tests. Grid pattern gained 72 MPa as maximum compression strength, while triangle structure made impact resistance and Young's modulus to reach 7.5 J and 0.68 GPa. Te infll structures have been optimized by using the topology optimization method [30][31][32]. Te fndings recommended the use of gradient as optimum infll instead of fxed distance ones where it maintained better mechanical properties and reduced printing time.
Also, the infll pattern and other 3D printing parameters were optimized by another work [33]. Te printed samples showed signifcant diference in terms of tensile and bending strengths based on the applied pattern, speed, orientation, and feed rate. Te PLA was strengthened with ammonium perchlorate by printing with FMD technology using complex and combustible patterns [34]. Te study demonstrated the performance of energetic and structural processed composite. A cubic specimen of PLA was printed at 20-60% infll density to examine its response to diferent densities in terms of printing time where 60% produced higher value of 227 min [35]. Te higher strength of 3D-printed work part was achieved by the correct tuning of printing orientation and raster angle for diferent 3D-printed materials [36].
According to the cited works, little attention was paid to the optimization of the mechanical and surface qualities of the 3Dprinted PLA+ material by using a multiobjective optimization scheme. Terefore, the current study attempts to optimize the mechanical properties and surface characteristics of 3D PLA+ material by using the integrated Taguchi-grey relational analysis (GRA) method as a multiobjective optimization.

Materials and Procedures
PLA+ flaments were printed by using 3D printer in the form of tensile test specimen according to the ASTM-D628 standard. Te PLA+ flament with diameter of 1.75 mm was provided by the Hello 3D Chinese company. Te total number of printed specimens was 9 according to the applied Taguchi method with L9 orthogonal array.
Taguchi's design of experiment is a powerful tool capable of reducing the cycle time of design and manufacturing stages. At the same time, it enables us to identify the signifcant controllable parameters that infuence the process through the analysis of variance (ANOVA). Furthermore, it is a systematic and easy-to-apply method that provides a minimum and sufcient number of experiments for the investigated case study in which the time is saved and the cost of experiments is reduced [37,38].
According to the Taguchi method, two diferent tracks were suggested for problem analysis [37] as follows: (1) Analysis of variance (ANOVA).
ANOVA evaluates the signifcance of one or more parameters by comparing their variables' output response means at various parameter levels. On the other hand, the S/N ratio determines the response variability concerning the target value subjecting to diferent noise conditions.

2
International Journal of Chemical Engineering For the maximization problem (higher is better), equation (1) is applied to calculate the maximum response. While equation (2) is used for the minimization problem (smaller is better). In the current study, the target is maximizing the ultimate tensile strength (σu), Young's modulus (E), and tensile strain of the 3D-printed PLA+ material. In contrast, it is targeted to minimize the surface roughness of the same material as follows: where η, y i , and n denote the S/N ratio, experiment (i), and experiment's total number, respectively. In the current study, the Taguchi experimental design scheme with L9 orthogonal array was applied using Minitab-17 statistical software. Terefore, nine specimens were printed by using 3D printer (model Ender3 Pro), as shown in Figure 1. Te thermoplastic flaments that are utilized in FDM must have a circular cross-section, as stated above. Te printer has a digital display to select and choose the appropriate printing parameters. Te extruder has two wheels, namely, a tensioner and a feeder to guide the flament towards the thermal chamber for heating to a predetermined temperature over the melting point and then extruding via the nozzle onto a glass platform. Te feeder is typically attached to the stepper motor, which powers this process. Te tensioner grasps the flament. Te thermal chamber is supplied with a heater having a thermocouple to carry out the heating and maintain the desired temperature. Te nozzle throat is encased in a heat sink to avoid flament melting before entering the thermal chamber. It dissipates heat from the nozzle head to avoid potential extruder clogging. Te cooling fan accelerates the dissipation of heat from the nozzle opening. Additional stepper motors are fxed to tune the extrusion track of melted flament on the driven platform, which moves downward in the Z-direction during the layer-building process [39].
Te standard tensile specimens were printed under four printing parameters with three levels each as illustrated in Table 1. Te factors A, B, C, and D stand for four printing parameters, namely, printing temperature (°C), printing speed (mm/min), infll density (%), and infll pattern. Te levels of these parameters were chosen with equal interval. When applying the Taguchi approach with L9 orthogonal array using the parameters and their levels in Table 1, Table 2 is constructed.
Te surface roughness in terms of Ra was measured by using stylus roughness tester prior the tensile test. Te mechanical properties of PLA+ were determined by carrying out the tensile test using universal tensile test machine. Te ultimate tensile strength (σu), Young's modulus (E), and tensile strain were calculated directly from the plotted stressstrain curves of the nine specimens.

Results and Discussion
Te results of mechanical properties and surface roughness of PLA+ 3D-printed specimens are collected and presented in the table and curve form as depicted in Table 3 and Figure 2. Tese tables and curves will be subjected to statistical, parametric, and optimization analysis in this section. Te aims are to determine the degree of signifcance for the developed statistical model and its corresponding factors and their contributions in the production of the output response, to conduct parametric analysis for the infuence of printing parameters on the mechanical and surface properties of the 3D-printed specimens, and fnally, to fnd out the optimum 3D printing settings which generate higher mechanical properties and fner surface roughness.

Statistical and Parametric Analysis of 3D Printing of PLA+ Materials.
To perform statistical and parametric analysis, it is so important to construct and present the ANOVA results and the main efect plots of the mean. Tus, Tables 4-7 illustrate the ANOVA results for the ultimate tensile strength (σu), Young's modulus of elasticity (E), tensile strains (ε) (%), and Ra, respectively.
Te DF, Adj SS, Adj MS, F-value, and P value terminologies of the ANOVA Tables 4-7 represent the degree of freedom, adjusted square error, adjusted mean square, F, and P values, respectively. Te degree of freedom refers to the amount of the informative data of the model. Te following formulas are used to calculate the degree of freedom for the following: regression model, each factor, error, and ad total as follows: International Journal of Chemical Engineering where n � no. of experiments � 9 and m � no. of factors � 4. Te adjusted square error (Adj SS) determines the variation of the model's diferent factors. Te adjusted mean square (Adj MS) measures all the factor variations of the model. Te F-value is a statistical measure that identifes whether model factors correlate with the output response. A larger F-value means the signifcance of the model or factor. Te probability of P value determines the strength of the data supporting the null hypothesis. More substantial evidence is presented against the null hypothesis via lower probabilities. If the P value is less than the signifcance level (0.05), then the model or term is signifcant, and vice versa.
Te following equations are used by Minitab software to calculate Adj SS and Adj MS as follows: where n � size of sample for group j x j , which is mean of group (j) x � the overall mean of the four group as follows: where x ij is the i th experimental run within group (j) and x j is the mean of group (j) as follows: total adj ss � adj ss of regression + adj ss of error, adj MS � adj ss DF .
With respect to Table 4 of ultimate tensile strength (σu), it is obvious that the model is signifcant with P value quite less than 0.05 at 95% confdence level. Tus, the model can be navigated and enabled for the purpose of studying of its controllable factors. Another point is the factors recorded diferent P values depending on their degree of signifcance. Terefore, only the infll density (%) was the signifcant factor that impacted the ultimate tensile strength (σu) with high percentage of contribution. Two more things have to be mentioned here, which are R-sq and contribution percentage of the factors. Excellent R-sq of 96.6% was achieved and the contribution percentage of the printing temperature, speed, infll density (%), and pattern were as follows: 0.0033%, 0.167%, 96%, and 0.23%, respectively. In other words, the large contribution came from the infll density.
Similarly, the ANOVA Table 5 confrms the signifcance of the Young modulus model as well as the efect of infll density (%) based on the achieved P values. Other factors were also not registered a degree of signifcance. 99.09% was obtained as R-sq, while printing temperature, speed, infll density, and pattern contributed with the following: 0.508%, 0.1052%, 97.665%, and 0.7654%. Pertaining to the ANOVA of the tensile strength of Table 6, the model is signifcant but the printing factors are not. Anyhow, the infll density (%) is still more infuenced than other factors where it recorded a contribution percentage of 49.549% followed by printing speed (1.5289%), infll pattern (1.1963%), and fnally printing temperature (0.9578%). Te achieved R-sq was 89.86%, and it is considered as an acceptable percentage.
Finally, ANOVA results of surface roughness (Ra) were found in Table 7. Te signifcance of the model is confrmed through the low P value (0.029) and at the same time, the table indicates the efect of the printing speed on the surface roughness. Conversely, other factors did not record any signifcance due to relatively large P values, but the infll pattern was more infuence factor compared with temperature and infll density (%). Good R-sq value was achieved for surface roughness with 89.82%. Also, the order of contributors was as follows: printing speed, infll density pattern, infll density (%), and temperature with contribution percentages as follows: 75.677%, 10.67%, 3.146%, and 0.33%, respectively. Tese contributions are compatible with the corresponding F-values. It is noticed that the factors rank is added to Tables 4-7, and this rank was in the same order for the printing parameters and for all responses (i.e., ultimate tensile strength (σu), Young's modulus (E), tensile strain (%), and surface roughness (Ra)). For all outputs, the infll density (%) keeps the class one because it was more infuential and contributing factor to the output responses. Te statistical fndings support and attribute to the trend of  Figure 2 illustrates P1, P2, and Pm beside the plateau zone. P1 and P2 points refer to the frst and second point of yielding as provided with tester software, while Pm represents the maximum achieved strength, which stands for the ultimate tensile strength. Te PLA+ is semicrystalline material that has crystalline and amorphous regions. Te plateau zone in Figure 2 is attributed to unstable crystalline area during tension that work as bufer feld that restrict stress rising and therefore create plateau area. Figures 3-6 give an illustration about how the four responses are afected by the printing factors, where A � printing temperature (°C), B � printing speed (mm/s), C � infll density (%), and D � infll pattern. For instance, in Figure 3, the impact of infll density (%) is visible compared with other factors where large strengths were recorded at 90% density. From the other side, the average strength for each parameter level is near to the mean value of nine runs due to their low contributions.
Te stifness of the PLA+ specimens was also highly affected by the infll density (%) in contrast with infll pattern, printing temperature, and speed, as depicted in Figure 4. Te trend of the mean efect plot for Young's modulus (E) is similar to that of the ultimate tensile strength (σu).
Regarding the main efect plot for the tensile strain (ε) (%), it can be noticed that the infll density infuence is reverse to the corresponding behaviors of the ultimate tensile strength (σu) and Young's modulus of elasticity (E), as shown in Figure 5.
Tis means that increasing the infll density gives noticeable reduction in tensile strain (ε) (%). Also, infll pattern reveals visible impact particularly triangle pattern. Other two factors (i.e., temperature and speed) did not show signifcant contribution to the tensile strain (ε) (%).
Finally, the surface roughness (Ra) is subjected to more coarsening due to the increase in the printing speed, infll density (%), and infll pattern with diferent levels unlike printing temperature, which produce slight increase in the roughness at low and medium levels and return back to the lower value at a high-level temperature, as shown in Figure 6.
To sum up, the 3D-printing parameters afected the four responses with diferent levels and various contribution percentages. However, the most efective parameter that impacted the frst three responses was the infll density (%) from the point of view of ultimate tensile strength (σu), and increasing the infll density (%), which means increasing of solid fraction on the account of empty fraction inside the cross-section area of the reduced section for the tensile specimen. In other words, the applied stress must be larger to reach the ultimate tensile strength (σu) and ended with necking and fracture. From the other side, increasing the infll density promotes the chance of increasing stifness by empowering of bonding between layers and molecules. Also, setting the more infuencing parameter on the low level gives more probability to increase the change in the length over the original gauge length in which the tensile strain increases. At the end, the surface roughness was infuenced greatly with the printing speed than other parameters where fne surface texture was produced at this speed with less valleys and waviness.

Optimization of the 3D-Printing Parameters.
In the previous section, a statistical and parametric analysis of the main fndings was performed to highlight the reliability of the developed model in terms of its signifcance and their corresponding factors, contribution percentages (%), R-sq, and factor rank order. Tis section undertakes the optimization of 3D-printing parameters to identify the optimum mechanical and surface properties of the PLA+ material.
Te Taguchi method was applied in this section as single objective optimization to optimize the mechanical properties and surface roughness of the PLA+ materials separately. Te larger is better was selected as the target for the mechanical properties, while smaller is better was chosen for the surface roughness.
Each of the low printing temperature (A1), high printing speed (B3), high infll density (C3), and triangle pattern (D1) delivered a highest signal to noise ratio, as illustrated by Figure 7. In other words, the optimum setting that may maintain maximum ultimate tensile strength (σu) is 205°C, 80 mm/s, 90%, and the triangle pattern.
Investigation of Figure 8 reveals that the optimum printing parameters that ensure maximum Young's modulus of elasticity (E) are similar to those that yielded the highest ultimate tensile strength (σu): A1, B3, C3, and D3 (i.e., 205°C, 80 mm/s, 90%, and triangle pattern).
Medium printing temperature and printing speed (A2 and B2) with low infll density (%) and triangle pattern (C1 and D1) may keep the tensile strain at the maximum value. Terefore, they represent the optimum parameters for maximum tensile strain where they produced the highest signal to noise ratio as depicted in Figure 9.
At the end, placing the printing temperature, printing speed, infll density (%), and pattern on the low levels (A1, B1, C1, and D1), as shown in Figure 10, sustains the fne surface roughness because these levels achieved large signalto-noise ratio that enable the producing of fne surface.
To sum up and give more illustrative view about the optimized setting of the 3D printing parameters, Table 8 is constructed. Tis table presents the optimized printing parameters for each response independently. In other words, each response has its own independent optimized settings because the Taguchi method is a single optimization method.

Multiobjective Optimization of 3D-Printing Parameters of PLA+ Materials
In the preceding section, the 3D-printing parameters were optimized for each response separately. Tat means the optimum parameters for mechanical and surface properties are diferent for response to response. Te Taguchi method sufers from demerits that it solves only single objective optimization problem. Te current study has four dependent responses that rely on four independent input parameters. Terefore, these input parameters have to be optimized to fnd out the combined optimum mechanical and surface characteristics. In other words, the problem will be multiobjective optimization.
To perform such task, the Taguchi approach was integrated with grey relational analysis (GRA) to provide an 8 International Journal of Chemical Engineering optimum solution for the mechanical and surface characteristics of the PLA+ materials. Tis approach transforms multiresponse to single response optimization [40,41]. It involves diferent steps that must be carried out sequentially to reach the optimized settings. Among them are the following: (i) Normalization of all responses (ii) Finding the grey relational coefcients (GRCs) (iii) Determining the grades of grey relational analysis Larger is better is chosen for the mechanical properties, while smaller is better is selected for surface roughness. Te normalization is preprocessing step aims to limit the responses values between 0 and 1 and convert original response to comparable one. Te mechanical responses (ultimate tensile strength, Young's modulus of elasticity, and tensile strain) is normalized by using equation (7)   International Journal of Chemical Engineering equation (8) is used to normalize the surface roughness response because it is selected as smaller is better.
where x i (k) � grey relational normalized value of i th run and k th response, min y i (k) � lowest value of y i (k) for k th response, and max y i (k) � highest value of the y i (k) for the k th response. Te two formulas are the same shared dominators with diferent numerators because the frst one stand for larger is better while second formula is used for smaller is better. Applying the equations (7) and (8) Table 9. When the normalized k th response of i th experiment equal or near to one points out that this run is the best regardless of the type response and vice versa for the normalized value equal or approach to zero.
Let us make a sample of calculations for the frst value of each response as follows: and so on for other values.
After normalization of all response values, the coefcients of grey relational analysis are determined with the aid of the following equation: where c(k) represents the grey relational coefcient (GRC), ζ is the distinguishability or identifcation factor and its value is between 0 and 1 and it is set at 0.5, ∆ oi is the absolute diference between the two sequences, and target x 0 (k) and comparison target x i (k), ∆ min, and ∆ max is the minimum and maximum of ∆ oi , respectively. Finally, the average sum of c(k) (GRC) is calculated by using equation (11) to fnd the grey relational grade (GRG) as follows: Table 10 presents the calculated grey relational coefcients for the four responses, the average GRC, and corresponding rankings.
Te table mentioned above illustrates that experiment no. 5 recorded the highest grey relational grade with the frst rank. In other words, it is the optimum experimental run that achieved the optimum mechanical properties and surface characteristics with optimized 3D-printing parameters of PLA+ materials. Recalling the parameters and corresponding responses from Table 2 shows the optimum 3D printing parameters as follows: 215°C, 50 mm/s, 90%, and triangle pattern maintained optimum ultimate tensile strength, Young's modulus, tensile strain (24 MPa, 3.14 GPa, and 13.72%), and optimum surface roughness of 3.21 µm. In coded form, the optimum 3D printing parameters that optimized mechanical and surface properties are as follows: A2B2C3D1, where A2 and B2 are the medium levels of printing temperature and speed, while C3 stand for high level of infll density, and fnally D1 refers to the triangle pattern. Te integrated Taguchi-GRA method proved it is powerful as a multiobjective optimization method.

Conclusions
Tis study dealt with the optimization of 3D-printing parameters to maintain better mechanical properties and surface characteristics of PLA+ materials. Based on the  International Journal of Chemical Engineering 13 conducted statistical and parametric analysis besides optimization, the following conclusions were drawn: (1) Te infll density was the most signifcant parameters in terms of obtained ultimate tensile strength (σu), Young's modulus (E), and tensile strain where it contributed with 96%, 99.09%, and 49.5%, respectively (2) Pertaining to the surface roughness, it was infuenced with printing speed which contributed with 75.67% (3) Te integrated Taguchi-GRA method was able to fnd the optimum settings for multiple responses together (4) Te optimum settings that ensure optimum mechanical properties and surface characteristic of 24 MPa, 3.14 GPa, 13.72%, and 3.21 µm are medium printing temperature (A2 � 215°C), medium printing speed (B2 � 50 mm/s), high infll density (C3 � 90%), and triangle pattern (D1 � triangle). (5) Te triangle pattern was the best structure for all responses compared with cubic and concentric patterns.

Data Availability
Te data that support the fndings of the study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.