Accuracy investigation of 3D printed PLA with various process parameters and different colors

The accuracy of three-dimensional (3D) printing is how the dimensions of a measured product are close to its original model’s nominal values. Thus, dimensional accuracy is essential for determining the machine’s reliability to produce each object according to the expected results. In this study, the inﬂuence of 3D printing process parameters on the dimensional accuracy of specimens manufactured using Polylactic acid (PLA) material is investigated. Based on fused deposition modeling (FDM) technology, cylindrical and dog-bone tensile test samples are fabricated at various process parameters, including build orientation, raster direction angle, and layer thickness. PLA ﬁlaments with three different colors (white, grey, and black) are utilized to produce the required test pieces. The dimensional accuracy for cylindrical (diameter and length) and dog-bone (width and thickness) samples have been evaluated. The nominal values are considered the reference to determine the accuracy percentage for each specimen. The weight of all test pieces is also examined, and its precision is assessed. The optimum process parameter settings have been deﬁned to minimize the error percentage in the dimensions of the printed parts. According to the results, a high overall dimensional accuracy of 98.81% was achieved, which indi- cates the ability of commercial FDM 3D printers as an inexpensive and decent quality alternative for producing utilitarian parts. The ﬁlament’s color displayed a notable impact on the test pieces’ weight, where the difference between the heaviest (white) and lightest (black) specimens is almost a percentage of 7.24%. A remarkable inﬂuence was noticed for the layer thickness parameter on the accuracy, meanwhile the raster direction angle parameter appeared no effect when the number of layers and the contour size are the same. The data obtained from this study might help identify the optimum conﬁgurations that guide the production of components using thermoplastic ﬁlaments through FDM 3D printing.


Introduction
3D printing is a modern technology that provides various new possibilities for complex-shaped objects to be made. It helps designers, based on a CAD model, to build real objects [1]. This technology is already making a significant impact in numerous fields such as industry, agriculture, and medicine [2][3][4][5]. Fused deposition modeling (FDM) is one of the most common additive manufacturing techniques, comprising extrusion of thermoplastic filaments for deposition as layer by layer [6]. Among other thermo-plastic (filaments) materials utilized in the FDM technique, ABS, PLA, PETG, and PC are the most popular ones [7][8][9][10]. The mechanical properties of FDM printed components are significantly influenced by the design and processing conditions of the process, often exhibiting anisotropic mechanical properties due to the choice of specific orientation of the infill layer [11]. The effect of the FDM process parameters on the mechanical properties of fabricated parts was extensively investigated in the literature [12][13][14][15][16]. However, only limited research yet have studied the dimensional accuracy of FDM 3D printing and the impact of the process settings on it.
Hyndhavi et al. used grey relational grade analysis to determine the optimum parameters of the employed settings for minimizing the error percentage in length, width, and thickness of the exam- ined parts. Based on the obtained results, the layer thickness of 200 mm, raster direction of 0°angle, and build orientation of 0 and 90°are the best parameters to refine the overall dimensional accuracy. As they concluded that the specimens fabricated using Acrylonitrile Butadiene Styrene (ABS) material exhibited higher accuracy than PLA material [17]. In contrast, Alsoufi et al. showed that the height dimension of ABS test pieces displayed the worst shape error reached almost a maximum percentage of 34.53% comparing to the base model. They examined the dimensional accuracy of FDM 3D printed PLA, PLA+, ABS, and ABS + thermoplastics in rectangular form specimens. This high shape error percentage in the ABS samples was attributed to undesirable warping deformation that occurs as a result of heat shrinkage. Though, the lowest warping deformation in the height dimension was represented by a shape error of 1.47% and 1.33% for PLA and PLA + materials, respectively. In terms of length and width dimensions, PLA + revealed the best accuracy with a minimum shape error of 0.05% [18]. Zhang et al. compared between the accuracies of dental models manufactured utilizing digital light processing (DLP) and stereolithography (SLA) 3D printing techniques at different thicknesses.
The layer thicknesses were ranged from 20 lm to 100 lm when producing all printed dental pieces (22 models). They stated that higher printing accuracy for DLP technology was detected at a layer thickness of 100 lm. Nevertheless, when using SLA technology the printing accuracy increases as the layer thickness decreased [19]. Nuñez et al. studied the dimensional accuracy obtained in FDM 3D printing with ABS + as the model material. They have used two layer thicknesses (0.178 mm and 0.254 mm) and two densities (low and solid) in the experimental tests. They found out that the best dimensional accuracy behavior was observed with the maximum layer thickness of (0.254 mm) and the solid density (100% infill) with a maximum deviation of 36 mm [20].
It can be recognized from the previous works mentioned above that most researchers paid attention to study the accuracy of rectangular shape 3D printed specimens. This work aims to determine the influence of various process settings on the dimensional accuracy of 3D printed objects in cylindrical and rectangular forms. Furthermore, to study the effect of these process settings as well as filament colors on the weight precision of the printed parts, which was never studied elsewhere. To this end, FDM 3D printed specimens in the form of cylindrical and dog-bone tensile test pieces were produced to investigate the dimensional accuracy. The cylindrical samples were made with three different colors (white, grey, and black) and under various build orientation (Horizontal, 45°a ngle, and Vertical . The dimensional accuracy of 3D printing is how the measured product (in terms of dimensions) close to the nominal value, where the nominal value reflects the CAD model's dimensions. Therefore, the achieved dimensional results (for cylindrical ''diameter and length" and dog-bone ''thickness and width" test pieces) were compared with the nominal values of each condition in order to obtain the accuracy percentage. Weighing for all manufactured specimens was carried out to discover whether various print settings and filament colors would affect the weight and its precision. This data is vital in additive manufacturing for determining the reliability of the machine. In other words, to count on the 3D printer to fabricate each product according to the required dimensions.

Preparation of materials and 3D printing
Fused deposition modeling (FDM) technique was employed to fabricate the test specimens with the commercial 3D printer WAN-HAO Duplicator 6. This printer has a nozzle diameter of 0.4 mm and exists in the laboratory of additive manufacturing at Szent Istvan University, Hungary. Polylactic Acid (PLA) filaments were purchased from the 3D printer material manufacturer eSUN. These filaments with a diameter of 1.75 mm and three different colors (white, gray, and black) were utilized for manufacturing samples. A three-dimensional designing software AutoCAD 2020, was used to generate the 3D CAD model of the sample and convert it to the ''.STL" file format. In order to slice the STL file (3D model), Ultimaker Cura 4.3 software was employed. The file exported from the slicing software is the G-code since it is compatible with the 3D printer. The printing was performed at a temperature of 195°C, and the platform was maintained at 60°C.

Process parameters for producing specimens
Models printed were divided into two groups, including cylinders and dog-bone tensile test specimens. Parameters and settings used for printing the samples of these two groups are listed in Table 1. The cylindrical pieces were printed with three colors and at various build orientation (Horizontal, 45°angle, and Vertical) while raster direction angle and layer thickness were fixed at 45/135°and 200 mm, respectively (see Fig. 1 (a)). However, the doge-bone tensile specimens were produced at three print orientations (Flat, On-edge, and Upright), three raster direction angle (0/45°, 45/135°, and 45/90°), and three layer thickness (100 mm, 200 mm, and 300 mm), as illustrated in Fig. 1 (b). All printed parts were manufactured at 100% infill density. A physical appearance for the specimens after manufacturing is exhibited in Fig. 2 (a) and (b). It can be clearly seen that support materials were used due to the tilt in 45°angle cylindrical specimens. The cylindrical test specimens were 3D-printed with a diameter of 8 mm and a height of 15.2 mm. Meanwhile, the dog-bone tensile test pieces' geometry was produced according to the ISO 527-2: 2012 standard type 1B specimen [21] with dimensions of 150 mm by 20 mm by 4 mm.

Accuracy measurements procedure
The measurements were accomplished on the samples of both printed groups (cylindrical and dog-bone) in order to obtain accuracy and precision. The reason for choosing cylinders and dog-bone is to inspect 3D printed parts' accuracy in circular and rectangular shapes. Concerning the cylindrical test pieces, the dimensions (di- Table 1 Parameters used for printing test ameter and length) were measured for all samples. Fig. 3 (a) displays that length was measured at the marked points (three spots), whereas diameter was investigated in the X and Y direction from both ends of the cylinder. Regarding the dog-bone tensile test specimen, the gauge section is the most crucial part. The width (b) and thickness (h) (gauge cross-section) were measured at both ends of the section as well as the middle (total; three positions), as illustrated in Fig. 3 (b). Hence to determine the accuracy, the obtained dimensions results had to be compared with the original CAD design. For a precise evaluation, the same measuring mechanism has been applied on six identical samples (the same manufacturing condition) of the cylinders and three duplicate specimens of the dog-bone pieces. The precision (repeatability) was determined through the standard deviation calculations exhibited by error bars. All samples were weighed by a Sartorius scale (Sartorius Corporation, Gottingen, Germany), and its weight precision was checked as well.

Raw results evaluation
The dimensions for the cylindrical and dog-bone test specimens were measured using a digital Vernier caliper. The raw results obtained for dimensional measurement of cylindrical and dogbone test samples are tabulated in Tables 2 and 3, respectively. As shown in Fig. 3 (a), the length was measured for the cylindrical specimens at three different points. In contrast, diameter measurements were done on the X and Y axis at both cylinder edges. The results were compared with the cylinder dimensions' nominal values: diameter of 8 mm and a length of 15.2 mm. In respect of the dog-bone tensile test pieces, the dimensions of width and thickness for each specimen were investigated at three points on the gauge area, as displayed in Fig. 3 (b). The gauge section dimensions' nominal values, including a width (b) of 10 mm and a thickness (h) of 4 mm, were used for the comparison to evaluate the accuracy. With regard to the weight accuracy calculations, there was no nominal value to compare with, as the slicing program does not display fractions at the amount provided for the sample shown in the original STL file. It only gives integer numbers such as 1 g for current cylindrical test samples and 8 g for the dog-bone ones, which are not accurate values. Thus, the average value for each print condition was taken, and it was considered the base value for comparison purposes with the results of the same group. The results presented in Tables 2 and 3 were averaged, and its accuracy percentage calculated. These accuracy percentage results with their standard deviation (SD) for cylindrical and dog-bone test samples are listed in Tables 4 and 5, respectively. Fig. 4 shows a comparison among the dimensional (diameter and length) accuracy of cylindrical test specimens manufactured in different print orientation (Horizontal, 45°angle, and Vertical) and color (white, grey, and black). From Table 4 and Fig. 4, it can be clearly seen that the worst dimensional accuracy with elevated standard deviation error bar values was observed in the 45°angle specimens. This was expected due to layers positioning, as layers of this specimen were built tilted. Therefore, the possibility of distortion increases due to the influence of gravity throughout curing. Generally, most of the cylindrical samples showed a high accuracy percentage ranging between 98.36% and 99.72% in both diameter and length measurements compared to the nominal value. After overall calculations, a precision of 99.14% and 99.35% against the   basic dimensions were reported for diameter and length consecutively. Hanon et al. reviewed the dimensional accuracy of FDM 3Dprinted cylindrical tribology test pieces made of bronze/PLA composite. An accuracy average of 99.65% for length and 98.19% for diameter was disclosed, which is in good agreement with the present study's results [22]. However, the weight showed a considerable variance among specimens produced in different conditions, as exhibited in Fig. 5. The diversity of colors has displayed a signif-icant effect on the weight of the test pieces. The lightest weight was noticed in the black colored samples, where its weight average is 755 mg, no matter the print orientation. Whereas other color specimens have reported heavier weights, averaged 814 mg and 770.8 mg for the white and grey, respectively. Hence, the distinction between the heaviest (white) and lightest (black) specimens is almost a percentage of 7.24%. Furthermore, comparing the samples' weight for the same color, the 45°angle pieces offered the highest values, while Horizontal and Vertical ones ranged more or less to the same extent.

Accuracy of dog-bone test specimens
A comparison among the dimensional (width and thickness) accuracy of dog-bone test specimens manufactured in different print orientation, raster direction, and layer thickness is exhibited in Fig. 6. High accuracy percentage was remarked for the width dimension in front of the basic (nominal) value. This accuracy reached higher than 99% for most tested samples, except the Onedge print orientation and the 100 mm layer thickness, where slight reductions were detected (97.85% and 98.66%, respectively). That indicates, there is no significant variation was recognized among the width measurements. This can be explained due to the fact that most of these specimens were printed in Flat print orientation, excluding the On-edge and Upright ones. As illustrated in Fig. 1  (b), the 3D printing process of any part includes building shell/wall (outer contour) and inner filling. The shell is the most robust part due to its larger thickness in the horizontal direction (wall thickness). Since the shell completely encloses the width measuring points of the Flat oriented samples. Thus, the width dimension shape error will be less even if the inner fill or layer thickness settings are different. This is due to the reliability of the outer contour, which is in contact with the measuring equipment. A similar interpretation could be attributed to the high accuracy of the Upright samples' width, as the shell is also the contact of the measuring points. Whereas for the case of On-edge printed specimens, the inner fill is the part which is in between the width measuring points. That means the number of layers is the important factor affecting the precision of the width here. Therefore, there was a reduction observed in the width accuracy of the On-edge pieces due to an elevated number of layers (width (10 mm) / layer thickness (0.2 mm) = 50 layers). In terms of dimensional thickness accuracy, a significant influence was obviously noticed in the layer Table 3 The raw results obtained from dimensional measurements of dog-bone test thickness specimens. Theoretically, the greater the number of layers, the poor the dimensional accuracy [19]. Since the number of layers increases as the thickness of the layer decreases. Therefore, it is conceivable that the 100 mm pieces present inferior thickness accuracy (around 94.69%) with a large error bar compared to the 200 mm (98.56%). As distorted growth becomes a dominant factor resulting from more layers, resulting in low dimensional precision [23]. Surprisingly, the 300 mm samples displayed the worst accuracy (91.75%) despite it supposed to be the best among the layer thickness ones, as its number of layers fewer. However, the reason behind it because the top layer is missing. For better understanding, the number of layers needed for forming the thickness of 300 mm specimen is calculated as following; specimen's thickness (4 mm) / thickness of layer (0.3 mm) = number of layers (13.33). The software will neglect this fraction of 0.33 because it did not reach 0.3 mm, as it equals only 0.1 mm. Consequently, the printer does not print a portion of this layer since the layer thickness was already set on a certain value. In connection with the effect of build orientation, all specimens (Flat, On-edge, and Upright) exhibited a high thickness accuracy percentage, ranging between 98.56% and 99.64%. In spite the shell of the On-edge and Upright specimens is the contact of thickness measuring points; nevertheless, the highest accuracy was observed in the Upright. This might be ascribed to the shell (contour) size of the whole piece. As the small size of the contour decreases shrinkage and improves precision [22]. Accordingly, the Upright's thickness accuracy enhanced since its contour size for each layer is much smaller than the contour of On-edge samples. Unlike other print parameters, raster direction angle specimens showed almost the same range of thickness accuracy percentage (98.54% À 98.69%). This implies that the raster direction angle has no impact on the thickness accuracy when the number of layers and the contour size are the same. As an average dimensional accuracy for dog-bone specimens regardless of the printing settings, a percentage of 99.16% for width and 97.57% for thickness were obtained compared to the nominal values. This high dimensional accuracy rate would agree with the findings of Alsoufi et al., where the dimensional accuracy of PLA thermoplastic material in a rectangular form was evaluated. They   indicated that shape error variation on all specimens' faces had not exceeded 3.00% due to the less occurrence of warping deformation than other materials examined in their research [18]. Eventually, a processing accuracy of 98.81% was gained in the current work through the overall average dimensional precision for both cylindrical (diameter and length) and dog-bone (width and thickness) samples.
A comparison among the average weight of dog-bone test specimens manufactured in different print orientation, raster direction angle, and layer thickness is presented in Fig. 7. A slight reduction was observed for the weight of the Upright test piece (9.34 g) as compared to the other print orientations (Flat (9.53 g) and 45°a ngle (9.61 g)), which are almost similar (see Fig. 7 (a)). Further, no significant difference was noticed among raster direction angle specimens ( Fig. 7 (b)), as all of them averaged within the same range (between 9.52 g and 9.61 g). On the other hand, Fig. 7 (c) shows a notable reduction of $ 17.9% in the weight of the 300 mm pieces against the weight of other layer thicknesses (9.34 g for 100 mm and 9.53 g for 200 mm), which were approximately alike. This reduction could be simply attributed to the top layer missing, as has been aforementioned.

Conclusions
This study analyzed the dimensional and weight accuracy of 3D-printed PLA material in three different colors. The experiment comprised the dimensional accuracy evaluation for cylindrical (diameter and length) and dog-bone tensile test (width and thickness of the gauge section) specimens. The weight precision was also examined for both the mentioned samples. The following conclusions have been drawn based on the observations: Commercial desktop FDM 3D printers could be used as an inexpensive and decent quality alternative for producing utilitarian parts with a high processing accuracy of 98.81% when investigated pieces compared to the nominal dimensions. The worst dimensional accuracy for cylindrical specimens (diameter and length) with elevated standard deviation error bars was observed in the 45°angle sample due to tilted layers positioning when models are constructed, as the influence of gravity increases distortion.
Generally, most cylindrical samples reported high accuracy percentage since overall calculations ranged as 99.14% and 99.35% for diameter and length consecutively against the basic dimensions. The filament color displayed a considerable effect on the test pieces' weight, where the distinction between the heaviest (white) and lightest (black) specimens is almost a percentage of 7.24%. For the dog-bone samples, high accuracy and no significant variation were detected among Flat and Upright's width measurements because the shell completely encloses measuring points. In contrast, On-edge exhibited a slight reduction due to inner filling is the contact of width measuring points where the number of layers becomes the dominant factor. A remarkable influence was noticed for the layer thickness parameter. The 100 mm pieces presented inferior thickness accuracy (94.69%) with a large error bar compared to the 200 mm (98.56%). As distorted growth becomes a prevailing factor as a result of more layers, resulting in low dimensional precision. An exception was detected in the 300 lm sample. It revealed the worst accuracy (91.75%) because the top layer is missing due to not reaching the correct value for the specified layer thickness (300 lm). Hence, the fraction value is neglected.
All build orientation specimens exhibited a high thickness accuracy percentage, ranging between 98.56% and 99.64%. The highest value was offered by the Upright since its contour size for each layer is much smaller than on-edge samples. As the small size of the contour decreases shrinkage and improves precision. The raster direction angle parameter appeared no impact on the thickness accuracy when the number of layers and the contour size are the same. A notable reduction of $ 17.9% in the weight of the dog-bone pieces with a thickness of 300 lm versus 100 lm (9.34 g) and 200 lm (9.53 g) could be attributed to the missing upper layer.
Finally, the best overall dimensional accuracies were obtained with the printing settings of ''black color/Horizontal build orientation/[45/135°] raster angle/200 mm layer thickness, (99.63%)" and ''black color/Vertical build orientation/[45/135°] raster angle/200 mm layer thickness, (99.72%)" for diameter and length of cylindrical specimens, respectively, as well as ''Flat build orientation/[45/135°] raster angle/300 mm layer To sum up, 3D printed parts' dimensional accuracy could be improved within the desired range by appropriately controlling the machine process parameters.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.