Structural Integrity Assessment of Stainless Steel Fabricated by GMAW-Assisted Wire Arc Additive Manufacturing

Abstract

Metal additive manufacturing techniques have seen technological advancements in recent years, fueled by their ability to provide industrial use parts with excellent mechanical properties. Wire Arc Additive Manufacturing is a technology that is being widely used in critical industries, and much research is conducted in this field due to the multiple factors involved in the overall process. Within WAAM, gas metal arc welding stands out for its low cost, high production volume, high quality and capability for automation. In this study, a CNC router was retrofitted with a gas metal arc welding setup to facilitate precise metal printing. The flexibility in this process allows for rapid repairs on site without the need to replace the entire part. The literature predominantly focuses on the macro-mechanical properties of GMAW parts, and very few studies try to study the interaction and influence of different process parameters on the mechanical properties. Thus, this study focused on the GMAW WAAM of stainless-steel parts by studying the influence of the wire feed rate, arc voltage and strain rate on the UTS, yield strength, toughness and percentage elongation. ANOVA and interaction plots were analyzed to study the interaction between the input parameters on each output parameter. Results showed that printing stainless steel through the gas metal arc welding process with an arc voltage of 18.7 V and a wire feed rate of 6 m/min resulted in poor mechanical properties. The input parameter that influenced the mechanical properties the highest was the wire feed rate, followed by the arc voltage and strain rate. Printing with an arc voltage of 18.7 V and a wire feed rate of 5 m/min, tested at a crosshead speed of 1 mm/min, gave the best mechanical properties.

1. Introduction

Additive manufacturing (AM), initially known as rapid prototyping in the 1980s, has transitioned significantly over the last four decades into a technology that is used for multiple industrial applications. The advancements in this field have revolutionized critical industries such as the aerospace, healthcare and automotive industries. Moreover, the innovations and developments have led to the optimization of production processes and supply chains across numerous sectors worldwide [1]. Contrary to the traditional subtractive manufacturing, additive manufacturing works on the concept of adding material layer by layer to create a 3D object from 3D data. The workflow of the AM process is depicted below in Figure 1.

According to the ASTM Standards, additive manufacturing is divided into seven categories: VAT polymerization, sheet lamination, material jetting, directed energy deposition, powder bed fusion, material extrusion and binder jetting [2]. VAT polymerization is an additive manufacturing process which employs an ultraviolet light to selectively cure resin layer by layer. The resins used in this process have low viscosity to ensure a smooth flow under the build plate while ensuring the fine-layer thickness of the final part [3]. The sheet lamination (SHL) process uses thin sheets of material that are bonded layer by layer through different energy techniques. This method is generally low in accuracy, but this is compensated by the fact that it can manufacture objects at higher speeds and lower costs [4]. The material jetting technique prints objects by jetting tiny drops of the material from a printhead onto a substrate. The printhead ejects two materials: build material and support material. This technology is commonly used for printing parts in industrial applications because of its low cost and scalability of production [5]. Directed Energy Deposition is the technique in which metals and alloys can be printed using an electric arc or laser to melt the materials. This method is widely adopted for large-scale printing and is known for its cost-effectiveness [6]. Powder Bed Fusion is an AM method that uses a heat source to irradiate, fuse or melt the powder particles. The materials used in this method are metals and polymers. Common related technologies include electron beam melting (EBM), Selective Laser Sintering (SLS), Selective Heat Sintering (SHS) and Direct Metal Laser Sintering (DMLS) [7]. The material extrusion process uses an extruder in the form of a plunger, gear or screw to extrude a feedstock through the nozzle onto the print bed. This method is known for its high material deposition rates and low costs [8]. Binder jetting is another cost-effective method that uses a layer-by-layer approach to deposit the material and binder onto the bed. This process is advantageous due to its ability to print complex parts with a wide range of materials [9].

In recent years, the concept of metal additive manufacturing has been prominent in multiple industries owing to the advantages it has over the conventional methods. Sarynski et al. highlighted the metal additive manufacturing techniques used in current industries and scientific sectors [10]. The methods were categorized based on the energy source used, material being plasticized and base material constituent form. Techniques such as Powder Bed Fusion (PBF), sheet lamination and Directed Energy Deposition are commonly used to produce large parts with advanced materials. They reported that combustion engine components, electric motor components, drivetrain components and exhaust components can be 3D-printed with lower costs owing to the complexity of the geometries in these parts. Aluminum alloys, copper alloys, steels and titanium alloys are the prominent materials used within these industries. Though metal additive manufacturing has certain advantages, Sarynski et al. noted that closed gas vapors within the process caused inherent porosity within the material structure. Moreover, fracture points can be induced when printed with improper melting temperatures. These imperfections can be eliminated through postprocessing techniques such as heat treatment and machining.

Among the metal additive manufacturing techniques, Directed Energy Deposition (DED) has gained traction in multiple industries. DED can be further classified into arc-based, laser-based and electron beam-based DED. A prominent subset of arc-based DED is Wire Arc Additive Manufacturing (WAAM). In this process, the metal wire is melted through an electric arc in the presence of a shielding gas [6]. B. Wu et al. reported that critical industries such as the aerospace, nuclear energy, marine and architecture industries have been using WAAM to manufacture and repair wing ribs, propellors and bridges [11]. Within WAAM, an effective process under much research and study is gas metal arc welding (GMAW). This is the process in which the wire is continuously fed through the nozzle using drive rolls and melted through the arc. GMAW is widely used in industries like the automotive and marine industries. When paired with robotic arms or CNC machines, the applications of these systems are limitless. Cheng et al. reported that the direct welding parameters in GMAW are the wire feed rate, welding speed, voltage and contact tube to work distance [12]. Additionally, the welding current was found to be the defining parameter for the arc power acting on the wire and substrate. Figure 2 shows the GMAW system with the main parameters.

For an effective GMAW process, the metal used must have exceptional weldability for long-term durability. In real-life conditions, these metal-printed parts must survive harsh environmental conditions, such as high temperatures, high salinity, humidity and corrosion. In recent times, different grades of stainless steels, such as lean duplex, duplex and superduplex stainless steel, have been developed [13]. Stainless steels have been extensively used in multiple industries due to their resistance to environmental corrosion, chemical stability and high mechanical properties [13,14,15,16]. Sampath et al. investigated the 3D printing of stainless steel and its weldability for corrosive environments [17]. Their findings pointed out that parts 3D-printed from stainless steel have good weldability and can retain more than 80% of their yield strength and UTS when compared to wrought parts. Abdo et al. conducted a potentiodynamic polarization analysis of 316L SS and 304L SS samples in 3.5% NaCl solution to explore their corrosion resistance [18]. The authors saw that the icorr value of 316L SS was low, pointing towards the high-corrosion-resistive nature of the material. They also observed that 316L SS had broader passive layer formation on the material, which acted as a barrier against corrosion [18].

Jin et al. reviewed the literature to explore the macroscopic characteristics and microstructures of stainless-steel parts printed using the WAAM method [19]. Mechanical properties like the tensile strength and hardness and their dependence with regard to the parameters were explored as well. They saw that the GMAW’s linear energy density for Stainless Steel G-25 94 material was in the range of 0.4–0.87 KJ/mm. Moreover, the authors found that the 316 stainless-steel parts printed with WAAM had a yield strength of 418 MPa and a tensile strength of 550 MPa. The parameters varied in this case were the heat input and the cooling rate. When the microstructures were analyzed, they observed that the shielding gas played an important role in the composition of the material in the final microstructures. Moreover, the WAAM-printed samples were anisotropic in nature, arising due to the directionality in which the layers were deposited. Residual stresses and distortion are certain consequences found in this method [19]. Appropriate heat treatment postprinting is required to relieve internal stresses. Jin et al. reported that residual stresses are known to have a negative effect on the mechanical properties of the part and can cause cracking, warpage or plastic deformation [19]. Dekis et al. saw that residual stresses, distortion, porosity, cracking and delamination are a few of the challenges in WAAM-related processes [20,21]. Parameters such as the heat input control are pivotal in determining the grain size and internal stresses. A lower heat input (HI) is preferred to reduce stress and produce a more favorable microstructure. However, the very strength of the WAAM process is its high deposition rate, requiring a higher HI rate. Thus, the author noted that it is necessary to optimize the HI and cooling rates to reduce the residual stresses built within the structure while maintaining the desired mechanical properties. They also underscored the importance of postprocessing treatments like heat treatment to reduce the internal stress and surface-finishing techniques to increase the dimensional accuracy. Parameters like the wire feed speed (WFS) and travel speed (TS) are seen as critical parameters that affect the porosity and subsequently the tensile strength and hardness of the material. When 420 Martensitic Steel was subjected to austenitizing treatment, it increased the microhardness from 550 ± 12 HV to 670 ± 4 HV, while the UTS increased from 1151 ± 9 MPa to 1903 ± 12 MPa [20].

Kokare et al. explored the environmental impacts, costs and mechanical properties of WAAM-manufactured high-strength, low-alloy steel (ERS70S) parts [21]. From the printed surface, four tensile samples were cut out, and uniaxial tensile tests were performed in accordance with the ASTM E8/E8M Standard. The strain rate used was 1 mm/min. Additionally, a Vickers microhardness test was performed in line with the ISO 6507-1:2023 Standard with a load of 0.5 kgf for 10 s [22]. The tensile tests showed that the horizontal and longitudinal orientations of the specimen gave averages of 1002 MPa and 700 MPa, respectively. Moreover, the yield strengths were found to be 694 MPa and 574 MPa for the horizontal and vertical orientations, respectively. The anisotropic nature of parts printed using this method was evident through the tensile test results. WAAM-printed vertical samples showed a 30% decrease in the UTS and a 20% decrease in the yield stress compared to the horizontal ones. The hardness test showed that the average hardnesses in the horizontal and vertical directions were 313 HV and 302 HV, respectively. Additionally, the LCA of WAAM-fabricated parts showed that manufacturing large parts was not only more cost-effective but also more environmentally friendly than conventional manufacturing methods. Ermakova et al. studied the mechanical properties of WAAM-manufactured low-carbon steel parts [23]. The authors used ERS70S-6 and ER100S-1 mild-steel welding wires for manufacturing the walls through Cold Metal Transfer (CMT)-based WAAM. The base plates used were EN 10025 Steel. The shielding gas used was Argon + 20% CO₂ with a gas flow rate of 15 L/min. The wire diameter was 1.2 mm, and the wire feed speed was 7.5 m/min. The dwell time between each layer print was 120 s. To speed the cooling process, an exhaust fan was kept near the experimental setup. The printed wall dimensions were 355 mm × 24 mm × 140 mm. From the printed wall, Electric Discharge Machining was used to extract specimens for tensile testing and microhardness measurements. Tensile tests were performed in accordance with the ASTM E8M Standard on both horizontal and vertical specimens at a crosshead speed of 1 mm/min. Specimens printed with ER70S-6 showed UTSs of 522 MPa and 518 MPa in the horizontal and vertical directions, respectively. They also had yield stresses of 390 MPa and 365 MPa in the horizontal and vertical directions, respectively. The Young’s modulus in the horizontal direction was 209 GPa, and it was 221 GPa in the vertical direction. In contrast, the ER100S-1 specimens had higher tensile properties than those of the ER70S-6 samples. They showed UTSs of 818 MPa and 815 MPa in the horizontal and vertical directions, respectively. They also had yield stresses of 538 MPa and 536 MPa in the horizontal and vertical directions, respectively. The Young’s modulus in the horizontal direction was 181 GPa, and it was 155 GPa in the vertical direction. The microhardness tests were performed along the height of the wall with 500 g and 2000 g loads. For the ER100S-1 sample, the hardness values were relatively the same across the height, while the samples printed with the ER70S-6 had higher hardness values toward the top of the wall. For an applied load of 500 g, the average hardness values were 158 ± 14 HV and 257 ± 22 HV for the ER70S-6 and ER100S-1 samples, respectively. When the applied load was 2000 g, the hardness values were 151 ± 12 HV and 245 ± 18 HV, respectively. Al-Nabulsi et al. explored WAAM 3D-printed steel parts and studied the mechanical properties of the specimens [24]. The authors used Union K 40 solid-steel wire for the GMAW printing. The dimensions of the printed wall were 345 mm × 75 mm × 20.5 mm. The outer tough surface of the printed part was postprocessed through, milling and grinding to achieve a high-surface-quality finish. The thickness of the plate was reduced down to 17 mm. The dog bone specimens for the tensile tests were cut down from the wall using the aqua jetting method. The dog bone specimens were cut out in accordance with the BS EN 10002 Standard [25]. The mean Young’s modulus for the specimens was found to be 208 GPa. The mean yield strength and UTS of the specimens were 419 MPa and 473 MPa, respectively. Hadjipanetelis et al. investigated the anisotropic stress–strain behavior of WAAM-printed stainless-steel walls [26]. Using the GMAW process, the walls were printed with two stainless-steel wire diameters: 1 mm and 1.2 mm. The 1 mm diameter wire was used to print the walls with a nominal thickness of 3.5 mm. The welding speed was 15–30 mm/s with a feed rate of 4–8 m/min. The 1.2 mm diameter wire was used to print 8 mm nominal thickness walls with a welding speed of 13 mm/s at a feed rate of 5.7 m/min. The currents for both specimens were set at a range of 100–140 A, and the voltage was between 18 and 21 V. The deposition rate was 0.5–2 kg/h with a dwell time of 30 s between each layer. The shielding gas used was 98% Argon and 2% CO₂ with a gas flow rate of 10–20 L/min. From each wall, the authors extracted tensile samples at three distinct orientations with respect to the deposition direction: 0°, 45° and 90°. Additionally, one set of samples was postprocessed for a good surface finish, and the other set was kept as built. The samples were placed under a UTS machine for tensile testing in accordance with the EN ISO 6892-1 Standard [27]. They observed that the as-built samples demonstrated lower mechanical properties than those of the machined ones. The test results showed high levels of anisotropy in the WAAM-manufactured stainless-steel samples. The high degree of anisotropy can be attributed to the strong “crystallographic structure” that developed from the solidification of the stainless steel under the influence of a substantial thermal gradient [26,28,29]. The average Young’s modulus, yield stress and UTS of the 0° sample were 143.3 GPa, 356 MPa and 575 MPa, respectively. The 90° sample had a Young’s modulus of 139.6 GPa, a yield stress of 338 MPa and a UTS of 554 MPa. The average values of the Young’s modulus, yield stress and UTS for the 45° sample were 219.5 GPa, 407 MPa and 626 MPa, respectively. Rani et al. converted a wood-engraving machine into a metal 3D printer to study the mechanical properties of stainless-steel and mild-steel parts manufactured through GMAW [30]. The authors studied the effect of the process parameters on the bead quality and the effect of the shielding gas on the deposition. The travel speeds investigated were 5, 7 and 8 mm/s, while the shielding gases used were Argon and carbon dioxide. They investigated the effect of the torch power by varying the arc voltage and current. The voltages investigated were 15 V, 18 V, 19 V and 25 V. The currents used were 130 A, 160 A, 170 A and 240 A. They observed discontinuous deposition at an arc voltage of 15 V and a current of 130 A printed with a travel speed of 5 mm/s. Additionally, they observed that using Argon as a shielding gas for the GMAW process improved the bead quality, and oxidation was not observed on the surface. In contrast, carbon dioxide facilitated the formation of oxide layers on the surface, which led to non-uniform layers. Though carbon dioxide is readily available and is cost-effective, the use of chemically inert gases for GMAW has been increasing due to their effectiveness against the diffusion of ambient gases within the structures. Hardness test results showed that the highest harness value was observed at the first few layers near the substrate and the layers furthest away from the substrate. The authors attributed this behavior to the fact that the initial layers cool down faster because of the lower ambient temperatures of the substrate materials. As the layers keep building on, the cooling rate is lowered, and coarse-grain formation is facilitated. Thus, the hardness keeps reducing. The last layer showed a higher hardness due to the formation of bainite through dense dislocations. The authors predicted the tensile strengths of GMAW mild steel and stainless steel through the theoretical equation that converts the hardness into the UTS. The average longitudinal and transverse UTSs for mild steel were predicted to be 440 MPa and 432 MPa, respectively. The average longitudinal and transverse UTSs for stainless steel were predicted to be 600 MPa and 597 MPa, respectively.

The existing literature predominantly looks into the microstructures of specimens printed with the GMAW process, and very limited studies explore the tensile mechanical properties of such structures. Additionally, there is a gap in studies that look into the effect of the process parameters on the tensile properties of GMAW stainless-steel parts. Thus, this study aimed to conduct a factorial ANOVA on the effect of process parameters such as the arc voltage, wire feed rate and strain rate on the tensile mechanical properties of stainless-steel parts manufactured through GMAW on a low-cost, retrofitted CNC machine. Additionally, quantitative porosity analyses of the samples were performed using MatLab (Online—Basic Version) to study the failure of the specimens.

2. Experimental Design and Methodology

An extensive literature review was performed initially of the field of GMAW and the use of stainless-steel material for creating functional structures. The mechanical properties reported by the authors were recorded in the review, and gaps within this field were identified. Then, using the GMAW manufacturing process, vertical walls were printed onto a substrate, and the tensile specimens were subsequently machined in accordance with the ASTM E8M Standard. The machine used in this study to print the wall was a CNC router that was retrofitted with a gas welding nozzle. The retrofitted CNC machine aims at providing a cost-effective alternative to manufacturing 3D-printed metal parts with comparable mechanical properties to those produced with advanced metal manufacturing methods. The CNC machine is manufactured by AccTek (Model: AKM 6090 Router), and the setup is such that the bed level remains constant while the carriage and gantry move in the x, y and z directions. The welding is facilitated through a Miller Matic Passport Plus and M-10 Welding Gun that is retrofitted to the carriage in the CNC machine. The Miller Matic Passport Plus setup allows the user to adjust the voltage and the wire feed rate based on the requirements. The setup is also connected to Argon gas to be used as shielding gas during the GMAW process. The G-Code to move the retrofitted print head was saved onto a text file and then uploaded into the Mach3CNC software (Version 3.043.066) to run it on command.

From the literature, it was evident that there are limited studies that focus on the dependance of the output mechanical properties on varying input parameters, such as the arc voltage, the wire feed rate and the testing parameter, the crosshead speed. In this study, walls were printed through the GMAW process using stainless-steel filaments of 1 mm diameter onto a mild-steel substrate. Additionally, the arc voltages under investigation were 18 V and 18.7 V, while the wire feed rates were 5 m/min and 6 m/min. The testing parameters were 1 mm/min and 2 mm/min.

Table 1. Printing parameters and material information.

Parameters	Set Values
Printing Parameters
Printing Speed	600 mm/min
Layer Height	1.5 mm
Dwell Time	60 s
Shielding Gas Composition	100% Argon
Gas Flow Rate	20 L/min
Arc Mode	Short Circuit
Print Material Information
Material	Stainless-Steel ER316LSi
Standard	AWS A5.9
Diameter	1 mm

shows the print material information and the printing settings used in the process. Figure 3 shows the CNC router setup with the retrofitted welding gun.

Table 2. Process parameters.

Parameters	Set Values
Current	197 A, 189 A, 162 A, 156 A
Arc Mode	Short Circuit
CTWD	15 mm
Torch Angle	90 degrees to plate
Interpass Temperature	<140 °C

shows the process parameters used in the study while

Table 3. Chemical composition of welding wire (%) (typical).

Carbon (C)	Silicone (Si)	Manganese/Magnesium (Mn)	Chromium (Cr)	Nickel (Ni)	Molybdenum (Mo)
0.02	0.80	1.6	18.5	11.5	2.2

shows the typical chemical composition of welding wire (%).

Table 4. Mechanical properties.

Yield Strength (MPa)	Tensile Strength (MPa)	Impact Strength (ISO-V/+20 °C)
Min. 400	550–700	Min. 63 J

shows the mechanical properties of the wrought material.

After the walls were 3D-printed, as shown in Figure 4, tensile specimens were cut out using water jet cutting in accordance with the ASTM E8/E8M Standard [31]. Subsized specimens were used with an overall length and thickness of 100 mm and 3.2 mm, respectively. The gauge length was 25 ± 0.1 mm, while the length of the reduced section was 32 mm. Samples were in the vertical and horizontal directions and were cut out from the walls to conduct tensile tests for both the longitudinal and vertical directions. Studies in the literature have performed similar tests which look into both the vertical and horizontal properties, as it helps in detecting the anisotropy. Additionally, the layer adhesion, bond strength between the layers and layer strength were investigated when both directions were considered. Figure 5 shows the directions in which the samples were cut out from the printed walls.

The output parameters investigated were the ultimate tensile strength, yield strength, toughness and strain at fracture. The dependence of these output parameters on the critical input and test parameters was tested, and the results were compared to the literature trends. The approach of this study is depicted in Figure 6.

2.1. Design of Experiments

For this study, three factors were investigated with two levels. Minitab Statistical Software 22 was used for providing the 2³ full factorial experimental table and for the further postprocessing of the data.

Table 5. Parameters investigated.

Parameter Investigated	Values
Arc Voltage	18 V and 18.7 V
Wire Feed Rate	5 m/min and 6 m/min
Crosshead Speed	1 mm/min and 2 mm/min

shows the parameters that were investigated and the different values.

Table 6. Full factorial design of experiment.

Sample No.	Run Order	Arc Voltage (V)	Wire Feed Rate (m/min)	Crosshead Speed (mm/min)	Current (A)	HI (kJ/mm)
1	1	18.7	6	2	197	0.37
2	2	18.7	6	1	197	0.37
3	3	18	5	1	156	0.28
4	4	18.7	5	2	162	0.30
5	5	18	6	2	189	0.34
6	6	18	5	2	156	0.28
7	7	18	6	1	189	0.34
8	8	18.7	5	1	162	0.30

shows the full factorial design implemented in this study. The experiments were conducted in accordance with the run order obtained from Minitab. Each experiment was run independently 3 times, and the mean values were used for analysis. This allowed for randomization within the experiments to minimize the effects of incontrollable external conditions that can affect the results. This eliminates such differences and potential biases that may appear. An analysis of variance (ANOVA) was then used in the subsequent section through Minitab to identify the contributions of the input parameters to the mechanical properties. Additionally, surface plots and interaction plots were created using Minitab to analyze the results of this study.

2.2. Postprocessing and Testing

To prepare the samples from the wall, water jet cutting was used, as it eliminates the heat-affected zones in the structure. Figure 7 shows the water jet cutting process and the samples placed according to the sample number in both the horizontal and vertical directions.

After the ASTM samples were prepared, surface finishing was performed using a surface grinder. This removed the irregularities present in the structure and also reduced the effect of crack propagation from such projections. Figure 8 shows the tensile samples after the surface finishing.

Tensile tests were conducted on the OTS Technik Universal Tensile Tester machine at crosshead speeds of 1 mm/min and 2 mm/min, as per the full factorial experimentation. Based on the gauge length of 25 mm, the corresponding strain rates for the crosshead speeds of 1 mm/min and 2 mm/min were 6.7 × 10⁻⁴ s⁻¹ and 1.3 × 10⁻³ s⁻¹, respectively. The UTM has a maximum load capacity of 500 kgf with a force resolution of ±0.5%. The UTM measures strain through the crosshead displacement. Tests were performed on both the horizontal and vertical samples with the same parameters to investigate the properties of the structures in both directions. The tensile tests were all conducted within the same day under the same environmental conditions (24 °C and 75% humidity) to prevent any biases in the results obtained. In line with Section 7.5 of the ASTM E8/E8M Standard, the specimens’ gripping was restricted to the grip section to avoid any errors in the test results. In this study, the effect of the strain rate was studied to investigate the strain rate sensitivity of GMAW-manufactured parts [31]. Figure 9 shows the UTM used for conducting the tensile tests.

3. Results and Discussion

This study was a full factorial study where every possible combination of the input parameter levels was investigated. The interaction plots obtained from Minitab provide a visual representation of the dependance and interactions between the studied parameters.

3.1. Horizontal Samples

From the experimental results, it was observed that Sample 3 had the highest UTS of 543 ± 4 MPa with a percentage elongation of 46 ± 1%. In accordance with the ASTM E8/E8M Standard, the yield strength was measured using the 0.2% offset method, whereby an offset line is drawn at 0.2% strain parallel to the elastic region of the curve. The intersection of the line on the stress–strain curve was taken as the yield strength. The highest yield strength of 418 ± 7 MPa was observed in Sample 8, which also demonstrated a similar UTS (537 ± 4 MPa) to that of Sample 3. These samples had very low porosity, and the combinations of the parameters involved in these samples demonstrated higher-quality finished walls. It is interesting to note that Sample 1 and Sample 2 showed poor mechanical properties when compared to those of the other six samples studied. Samples 1 and 2 were printed with an arc voltage of 18.7 V and a wire feed rate of 6 m/min. The higher wire feed rate essentially was seen to impact the quality of the printing, as pores were observed within the structures of the samples. The particular combination of a higher voltage (18.7 V) and higher feed rate (6 m/min) resulted in unfavorable build quality and pore formation. The presence of the large pores facilitated crack propagation and premature fracture. The UTSs of the horizontal samples were between the range of 394 ± 4 MPa and 543 ± MPa, which was in line with the trends seen in the literature [19,23,24,26]. The wrought UTS of this particular material is specified as 550–700 MPa, and the UTSs of the printed samples were observed to be lower than that of the wrought UTS. This is due to the presence of porosity, lack of fusion and residual stresses. The toughness and %EL for Samples 1 and 2 were very low due to the high levels of porosity present in the parts. The presence of porosity facilitates brittle fracture behavior. This is evident in Figure 10, where Samples 1 and 2 demonstrate brittle fracture while the other samples show ductile fracture. Figure 11 also shows the porosity seen in Sample 1-H and Sample 2-H, while the other samples show little to no porosity in the structure. The pictures were captured using an optical microscope. There is noise and fluctuations present in the tensile stress-strain curves of several samples. These fluctuations, though minor, are observed between 15 and 20% strain. As Sharma et al. points out, WAAM-manufactured stainless-steel components have common defects, such as porosity and lack of fusion [32]. These defects lead to inconsistent areas within the structure, which may be stronger than the others. These pores act as stress concentrators, resulting in regions of lower load-carrying capabilities. In this study, samples with larger and higher number of pores fractured rapidly, as the pores acted as fracture initiation points. However, certain samples with relatively smaller numbers of pores showed minor fluctuations in the stress–strain curve, as these pores were not large enough to initiate rapid fracture but were adequate to slightly drop the load-carrying capacity. Further analysis of the pores is given in Section 3.3.

Table 7. Output data obtained from tensile testing for horizontal samples.

Horizontal Specimen No.	Arc Voltage (V)	Wire Feed Rate (m/min)	Crosshead Speed (mm/min)	UTS (MPa)	Yield Strength (Offset = 0.2%) (MPa)	Toughness (MPa)	%EL
1	18.7	6	2	416 ± 5	406 ± 6	27 ± 2	13 ± 1
2	18.7	6	1	394 ± 4	390 ± 4	24 ± 2	11 ± 1
3	18	5	1	543 ± 4	407 ± 5	170 ± 4	46 ± 1
4	18.7	5	2	516 ± 4	392 ± 4	215 ± 3	51 ± 2
5	18	6	2	476 ± 8	370 ± 6	138 ± 3	41 ± 1
6	18	5	2	524 ± 6	412 ± 4	146 ± 4	42 ± 2
7	18	6	1	492 ± 3	400 ± 7	100 ± 4	31 ± 1
8	18.7	5	1	537 ± 4	418 ± 7	248 ± 4	57 ± 1

shows the tensile test results of the horizontal samples. These results were uploaded to Minitab to conduct the ANOVA factorial regression of the output parameters with respect to the input parameters and the surface plots.

3.1.1. Analysis of Variance and Interaction Plots for Horizontal Samples

Analysis of variance (ANOVA) is a statistical method that is commonly used in studies to determine the significant factors affecting the response parameters [33,34]. A confidence level of 95% was used for the ANOVA in this case, and corresponding to this, the p-value needed to be lower than 0.05 for the factor to have a significant influence on the output parameter.

Ultimate Tensile Stress (UTS)

An ANOVA of the UTS with respect to the arc voltage, wire feed rate and crosshead speed was performed through Minitab and is shown in

Table 8. ANOVA: UTS versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	3728.2	17.21%	3728.2	3728.2	4.66	0.097
Wire Feed Rate (m/min)	1	14,594.9	67.37%	14,594.9	14,594.9	18.26	0.013
Crosshead Speed (mm/min)	1	143.7	0.66%	143.7	143.7	0.18	0.693
Error	4	3197.4	14.76%	3197.4	799.3
Total	7	21,664.0	100.00%

. It is evident from the p-value that the wire feed rate had a significant impact on the UTS (p-value < 0.05), while the crosshead speed did not have an influence on the UTS. Additionally, the contribution of the wire feed rate to the UTS was seen to be the highest (67.37%), followed by the arc voltage (17.21%) and crosshead speed (0.66%). The error contribution was 14.76%. Furthermore, the main effects plot in Figure 12 shows the average UTS at each level of the three factors. It is seen that the UTS decreases when the arc voltage increases from 18 V to 18.7 V. The same is observed for the wire feed rate, but the drop in the UTS here is much stronger. The average UTS value drops from 530 MPa to 445 MPa when the wire feed rate increases from 5 m/min to 6 m/min. The strain rate has a low influence on the UTS values. The interaction plot for the factors in Figure 12 shows that there is an interaction between the arc voltage and wire feed rate for the UTS. When the wire feed rate is 6 m/min, the effect of increasing the arc voltage is seen to be more profound, as it negatively affects the UTS. However, when the wire feed rate is at 5 m/min, the arc voltage has a minimal effect on the mean UTS. Additionally, there is interaction between the arc voltage x crosshead speed and wire feed rate x crosshead speed. In both plots, the mean UTS decreases with the increasing arc voltage and wire feed rate, irrespective of the crosshead speed. The slope of the crosshead speed, 1 mm/min, is steeper than 2 mm/min.

Yield Strength

For the yield strength, it is seen that the p-values for all factors are greater than 0.05, showing that the factors did not influence the output parameters with 95% confidence, as seen in

Table 9. ANOVA: yield strength versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	34.03	2.12%	34.03	34.03	0.15	0.719
Wire Feed Rate (m/min)	1	504.03	31.36%	504.03	504.03	2.21	0.211
Crosshead Speed (mm/min)	1	157.53	9.80%	157.53	157.53	0.69	0.453
Error	4	911.87	56.73%	911.87	227.97
Total	7	1607.47	100.00%

. However, based on the output parameter values, the wire feed rate’s contribution to the yield strength is 31.36%, followed by the crosshead speed (9.8%) and arc voltage (2.12%). It is interesting to note that the error contribution here is over 50%, which suggests that the model is not able to explain most of the system’s behavior. This error is usually attributed to random errors and can be reduced by including more significant factors, such as the travel speed and gas flow rate, in the model. The main effects plot in Figure 13 shows the significant impact of the wire feed rate on the YS, as seen with the UTS. Interestingly, the crosshead speed here had more significance than the arc voltage, which was not seen in the case with the UTS. As the wire feed rate increased from 5 m/min to 6 m/min, the mean yield strength dipped from 407.5 MPa to 391.5 MPa. Moreover, a higher crosshead speed resulted in a lower mean YS. A higher arc voltage, in contrast, showed an increase in the mean YS.

When the interaction plots are analyzed in Figure 13, it is seen that there is an ordinal interaction between the arc voltage and wire feed rate on the mean YS. For a wire feed rate of 6 m/min, the mean YS increased when the arc voltage increased from 18 V to 18.7 V. However, the opposite trend was observed for the 5 m/min wire feed rate. Increasing the voltage at this wire feed rate showed a dip in the mean YS. The arc voltage × crosshead speed and wire feed rate × crosshead speed interaction plots also show an interaction between each other on the mean YS, as the lines are not parallel. However, the interactions here are fairly minimal. For both crosshead speeds, when the arc voltage was increased, the mean YS increased slightly. However, when the wire feed rate was decreased for both crosshead speeds, the mean YS decreased.

Toughness

The toughness of the material is defined as the total energy the material can absorb before undergoing fracture. This parameter is obtained by calculating the area under the stress–strain curve. The ANOVA of the toughness vs. the input parameters in this study showed that only the wire feed rate had a significant impact on the toughness values (p < 0.05), as seen in

Table 10. ANOVA: toughness versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	196.8	0.43%	196.8	196.8	0.05	0.833
Wire Feed Rate (m/min)	1	30,122.9	65.73%	30,122.9	30,122.9	7.78	0.049
Crosshead Speed (mm/min)	1	30.8	0.07%	30.8	30.8	0.01	0.933
Error	4	15,479.8	33.78%	15,479.8	3870.0
Total	7	45,830.3	100.00%

. The crosshead speed and the arc voltage had almost no impact on the toughness in this case. This is also reflected in the contribution percentages, where the wire feed rate had a contribution of 65.73%, while the other two input parameters had less than 1% contributions. The error’s contribution was 33.78%. The main effects plot in Figure 14 shows that the wire feed rate of 5 m/min gave the highest mean toughness values, while the arc voltage and crosshead speed had much less say in the toughness. With the increasing wire feed rate, the mean toughness dropped from 195 MPa to 75 MPa. Though insignificant, the mean toughness slightly dropped when the arc voltage and crosshead speeds were increased to 18.7 V and 2 mm/min, respectively. An analysis of the interaction plots in Figure 14 shows that there was an ordinal interaction effect between the arc voltage and wire feed rate on the toughness values. For a wire feed rate of 5 m/min, an increasing arc voltage showed a higher toughness value, while with a wire feed rate of 6 m/min, a higher arc voltage resulted in a lower mean toughness. The arc voltage × crosshead speed interaction plot shows a strong interaction, evident through the intersecting lines. At an arc voltage of 18 V, the sample tested with a crosshead speed of 2 mm/min had a higher toughness while at an arc voltage of 18.7 V, and the samples tested with a crosshead speed of 1 mm/min had a higher toughness. It is interesting to note here that the increase in the toughness values is very minimal. Furthermore, the wire feed rate × crosshead speed plot shows that there was a strong interaction. With the increasing wire feed rate, the mean toughness decreased, independent of the crosshead speed. However, because of the intersection, the change in the wire feed rate from 5 m/min to 6 m/min reversed the effect of the crosshead speed. At a wire feed rate of 5 m/min, the toughness values of the 1 mm/min crosshead speed samples were higher. The opposite trend was observed at a higher wire feed rate.

Percentage Elongation

The percentage elongation, or the strain at fracture, represents the amount of elongation the material underwent before fracturing. The ANOVA trends here are very similar to those for the toughness, wherein the wire feed rate had a significant impact on the %EL, while the arc voltage and crosshead speed had no impact (p > 0.05), based on the 95% confidence used in the ANOVA. The ANOVA table for the percentage elongation versus arc voltage, wire feed rate and crosshead speed is presented in

Table 11. ANOVA: percentage elongation versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	105.04	5.23%	105.04	105.04	0.64	0.470
Wire Feed Rate (m/min)	1	1244.61	61.91%	1244.61	1244.61	7.54	0.052
Crosshead Speed (mm/min)	1	0.72	0.04%	0.72	0.72	0.00	0.951
Error	4	659.86	32.83%	659.86	164.96
Total	7	2010.22	100.00%

. The wire feed rate contribution to the %EL was 61.91%, while the arc voltage had a contribution of 5.23%. The error contribution was 32.83%. Samples printed with an arc voltage of 18 V showed a higher mean %EL when compared to those printed with 18.7 V. Furthermore, a lower wire feed rate ensured a higher average %EL in the samples, as seen through the mean effects plot in Figure 15. Additionally, the arc voltage × wire feed rate interaction plot shows that there was an interaction between the two with respect to the percent elongation. For a wire feed rate of 5 m/min, the %EL increases with the increasing arc voltage. However, for a wire feed rate of 6 m/min, the %EL decreases with the increasing arc voltage. Examining the arc voltage × crosshead speed and wire feed rate × crosshead speed plots shows that there is a strong interaction between them. At a lower arc voltage, the crosshead speed of 2 mm/min has a higher %EL, while the values are reversed at a higher arc voltage. Additionally, at a lower wire feed rate, the lower crosshead speed sample has the higher %EL, while the higher crosshead speed sample shows a higher %EL at higher wire feed rates.

3.2. Vertical Samples

Samples cut out in the vertical direction from the printed wall showed lower mechanical properties than those of the horizontal counterparts since the grain orientation was in the transverse direction. Similar to the horizontal samples, Sample 1 and Sample 2 showed poor mechanical properties due to high porosity within the structure. The walls printed with a wire feed rate of 6 m/min at an arc voltage of 18.7 V showed this anomalous behavior due to low interlayer bonding and high porosity. As with the horizontal samples, this particular combination of the printing parameters facilitated pore formation in the samples. The pores acted as stress concentrators, causing the samples to have premature fracture. The highest UTS (495 ± 4 MPa) was observed in Sample 8, printed with an arc voltage of 18.7 V and a wire feed rate of 5 m/min, and tested at a crosshead speed of 1 mm/min. The highest toughness (256 ± 4 MPa), however, was observed in Sample 6, where the samples were printed at 18 V and a wire feed rate of 5 m/min. The highest %EL was seen in Samples 6 and 3, with 56 ± 1% and 52 ± 1%, respectively. Both these samples were printed at 18 V and a wire feed rate of 5 m/min, with the only difference being the crosshead speed. The ductile nature seen in the rest of the samples is due to the lower presence of pores in the structures and the better surface finish of the samples. A poor surface finish leads to early crack initiation and crack propagation, as seen in Samples 1 and 2. It is evident from the stress–strain curve in Figure 16 that Samples 1 and 2 fractured very early when compared to the other six samples. Moreover, as seen with the horizontal samples, Samples 1-V and 2-V also showed brittle fracture while the others showed ductile fracture. The yield strength was calculated using the 0.2% offset method, in compliance with the ASTM E8/E8 M Standard.

Table 12. Output data obtained from tensile testing for vertical samples.

Vertical Specimen No.	Arc Voltage (V)	Wire Feed Rate (m/min)	Crosshead Speed (mm/min)	UTS (MPa)	Yield Strength (Offset = 0.2%) (MPa)	Toughness (MPa)	%EL
1	18.7	6	2	383 ± 4	340 ± 2	24 ± 1	10 ± 1
2	18.7	6	1	340 ± 3	330 ± 4	21 ± 1	10 ± 1
3	18	5	1	472 ± 5	362 ± 4	222 ± 4	52 ± 1
4	18.7	5	2	444 ± 4	342 ± 3	149 ± 2	44 ± 2
5	18	6	2	435 ± 3	325 ± 4	61 ± 2	22 ± 1
6	18	5	2	464 ± 3	365 ± 5	256 ± 4	56 ± 1
7	18	6	1	425 ± 4	350 ± 4	47 ± 1	19 ± 2
8	18.7	5	1	495 ± 4	320 ± 4	119 ± 2	33 ± 2

shows the tensile test results of the vertical samples. When compared with the wrought material, the GMAW-manufactured parts showed lower UTSs and yield strengths. The same trend was also seen in the horizontal samples. This is attributed to the imperfections and defects that occur in the part during the gas metal arc welding process, which, in turn, affect the overall mechanical properties of the sample. Figure 17 shows the high amounts of porosity seen in Samples 1-V and 2-V, while the others showed little to no porosity, allowing them to elongate further before fracture. The device used to capture the image was an optical microscope at low magnification. Minitab was used to analyze the output results based on the ANOVA and interaction plots. Further quantitative analysis of the pores is performed in Section 3.3.

3.2.1. Analysis of Variance and Interaction Plots for Vertical Sample

For the ANOVA of the mechanical properties of the vertical samples, a confidence level of 95% was used, and the p-value was examined to understand the significance of the factor on the output parameter.

Ultimate Tensile Stress (UTS)

An analysis of variance of the UTS versus the input parameters in

Table 13. ANOVA: UTS versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	2207.1	12.54%	2207.1	2207.1	1.86	0.245
Wire Feed Rate (m/min)	1	10,637.6	60.43%	10,637.6	10,637.6	8.95	0.040
Crosshead Speed (mm/min)	1	6.3	0.04%	6.3	6.3	0.01	0.945
Error	4	4752.9	27.00%	4752.9	1188.2
Total	7	17,603.9	100.00%

showed that the wire feed rate played the most prominent role in deciding the UTS with 60.43%. This was followed by the arc voltage (12.54%) and crosshead speed (0.04%). The error contribution was 27%. Moreover, the p-value shows that the significance of the wire feed rate on the UTS was very high (p < 0.05), while the arc voltage and crosshead speed did not affect the UTS with 95% confidence. The main effects plot in Figure 18 portrays that the mean UTS increased with the decreasing arc voltage, wire feed rate and crosshead speed. However, the effect of the strain rate was considered very minimal. The influence of the wire feed rate was seen to be the highest, as the mean UTS dropped from 470 MPa to 397 MPa when the wire feed rate increased from 5 m/min to 6 m/min.

An analysis of the interaction plots for the UTS shows that there is an interaction between the arc voltage and wire feed rate on the UTS. At a wire feed rate of 5 m/min, the mean UTS remained relatively constant, irrespective of the arc voltage. However, at a higher wire feed rate, the UTS dipped when the arc voltage increased from 18 V to 18.7 V. On the arc voltage × crosshead speed plot, there is a small overlap of both the lines, suggesting an interaction between them. At 18 V, the mean UTS is the same for both crosshead speeds, while at a higher voltage, the UTS dips under both crosshead speeds. In the wire feed rate × crosshead speed plot, there is a clear intersection of the lines, pointing to a high interaction between the factors. At a wire feed rate of 5 m/min, the mean UTS is higher with a crosshead speed of 1 mm/min. However, the mean UTS is lower for the same crosshead speed but at a higher wire feed rate.

Yield Strength

With respect to the yield strength, the arc voltage had the highest contribution (32.03%), followed by the wire feed rate (12.68%), as seen through the ANOVA table in

Table 14. ANOVA: yield strength versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	614.25	32.03%	614.25	614.25	2.34	0.200
Wire Feed Rate (m/min)	1	243.10	12.68%	243.10	243.10	0.93	0.390
Crosshead Speed (mm/min)	1	12.25	0.64%	12.25	12.25	0.05	0.839
Error	4	1047.96	54.65%	1047.96	261.99
Total	7	1917.56	100.00%

. The error contribution was 54.65%. The strain rate essentially had no influence on the yield strengths of the vertical samples. Analysis of the p-values shows that none of the factors has a p-value of less than 0.05, suggesting that the significance of these factors on the yield strength is very low. The mean effects plot in Figure 19 suggests that the mean arc voltage dropped from around 350 MPa to 332 MPa when the arc voltage increased from 18 V to 18.7 V. Moreover, a higher feed rate resulted in a dip in the mean UTS, but at a lower rate when compared to the effect of the arc voltage. The strain rate’s influence was seen to be very minimal in this case, with the mean yield strength slightly increasing with the increasing crosshead speed. The presence of high error as part of the contribution suggests the influence of random errors in the data that cannot be explicitly explained by the model.

The arc voltage × wire feed rate interaction plot shows a clear intersection of the lines, pointing to a clear interaction between them. At the 18 V arc voltage, the samples printed with a wire feed rate of 5 m/min had higher mean yield strengths than those of the ones printed with a wire feed rate of 6 m/min. The samples printed with an arc voltage of 18.7 V and a 6 m/min wire feed rate showed marginally better yield strengths than those of the ones printed with a 5 m/min wire feed rate. The yield strengths of the 6 m/min wire feed rate samples remained relatively the same, irrespective of the arc voltage, while there was a drastic dip in the yield strengths when the arc voltage was increased for the 5 m/min wire feed rate samples. On the arc voltage × crosshead speed plot, there is a high interaction, as evident through the intersection of the plots. At an arc voltage of 18 V, the samples tested at the 1 mm/min crosshead speed showed better yield strengths. However, at a higher arc voltage, the samples tested with the 2 mm/min crosshead speed showed better results. The same trend is also seen in the wire feed rate × crosshead speed plot. In this case, the samples tested with a crosshead speed of 1 mm/min showed similar yield strengths, irrespective of the wire feed rate. However, with an increase in the wire feed rate, the samples tested with the 2 mm/min crosshead speed showed a dip in their mean yield strengths.

Toughness

The toughness of the vertical samples was highly influenced by the wire feed rate and the arc voltage, as both the p-values were lower than 0.05, as seen in

Table 15. ANOVA: toughness versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	9361.2	16.37%	9361.2	9361.2	12.48	0.024
Wire Feed Rate (m/min)	1	43,979.8	76.93%	43,979.8	43,979.8	58.65	0.002
Crosshead Speed (mm/min)	1	829.9	1.45%	829.9	829.9	1.11	0.352
Error	4	2999.4	5.25%	2999.4	749.8
Total	7	57,170.3	100.00%

. The crosshead speed did not have any influence on the toughness, since the p-value was greater than 0.05. Additionally, the highest contribution towards the toughness was seen through the influence of the wire feed rate (76.93%), followed by the arc voltage at 16.37%. The crosshead speed’s contribution towards the toughness was very minimal (1.45%). The main effects plot in Figure 20 shows that when the arc voltage increased from 18 V to 18.7 V, the mean toughness dropped from 150 MPa to 75 MPa. Additionally, the wire feed rate was seen to have a much steeper downward slope for the toughness with the increasing wire feed rate. At a wire feed rate of 6 m/min, the mean toughness was just 37.5 MPa, while a toughness of 187.5 MPa was observed when the wire feed rate was 5 m/min. The influence of the crosshead speed was very minimal on the toughness values. The interaction plots for the arc voltage × crosshead speed show that both lines are parallel to each other, indicating no interactions between the factors towards the toughness. The arc voltage × wire feed rate plot shows that there is an interaction between them, as the lines are not parallel. For both wire feed rates, the mean toughness decreased with the increasing arc voltage. With respect to the wire feed rat × crosshead speed plot, there is an interaction present, as the lines seem to intersect each other at a wire feed rate of 6 m/min. At a wire feed rate of 5 m/min, the samples tested at a crosshead speed of 2 mm/min had higher toughness values, while at a wire feed rate of 6 m/min, both the samples showed the same toughness values, irrespective of the crosshead speed.

Percent Elongation

The %EL was highly influenced with 95% confidence by the arc voltage and wire feed rate, as their p-values were 0.007 and 0, which are lower than 0.05, as seen in

Table 16. ANOVA: %EL versus arc voltage, wire feed rate and crosshead speed.

Source	DF	Seq SS	Contribution	Adj SS	Adj MS	F-Value	p-Value
Arc Voltage (V)	1	324.64	13.87%	324.64	324.64	25.12	0.007
Wire Feed Rate (m/min)	1	1926.53	82.28%	1926.53	1926.53	149.07	0.000
Crosshead Speed (mm/min)	1	38.47	1.64%	38.47	38.47	2.98	0.160
Error	4	51.70	2.21%	51.70	12.92
Total	7	2341.33	100.00%

. The crosshead speed, however, had no major influence on the %EL (p > 0.05). The contribution of the wire feed rate was seen to be 82.28%, while the arc voltage contributed 13.87% towards the %EL. The crosshead speed only had a 1.64% contribution towards the %EL. The main effects plot in Figure 21 shows that increasing the arc voltage from 18 V to 18.7 V decreased the mean %EL from 37.5% to 25%. The influence of the wire feed rate is again seen to be higher, as the slope is much steeper in the main effects plot. The mean %EL decreased from around 45% to 15% when the wire feed rate increased from 5 m/min to 6 m/min. In contrast, the higher crosshead speed resulted in a marginally higher %EL. The arc voltage × crosshead speed interaction plot shows that the lines are fairly parallel to each other, indicating little to no interaction between these factors and the %EL. Additionally, the arc voltage × wire feed rate plot shows that there was an interaction between both the factors. When the arc voltage increased, the %EL decreased for both wire feed rates. The wire feed rate × crosshead speed interaction plot shows the same results as those of the toughness plot. It is interesting to note that the mean effects plots and interaction plots for the toughness and %EL are showing fairly the same trends and slopes.

3.3. Porosity in Horizontal and Vertical Samples

The GMAW samples showed varying amounts of porosity within the structure. The presence of pores is common in parts manufactured through this process. Gordon and Harlow and Sharma et al. pointed out that incomplete melting of the stainless-steel material during the WAAM is a factor for this defect within the structure [32,35]. They also highlighted that process parameters such as the deposition route and heat input should be carefully optimized to reduce the possibility of pores. Additionally, pores can be caused due to certain defects in the raw material itself. Contamination of the filler metal and substrate with external factors is a common contributor in GMAW parts [32]. In this study, MatLab (Online—Basic Version) was used to quantify the pores based on an area fraction analysis. Firstly, original images of the thickness cross-sectional areas of all samples were uploaded onto MatLab, and the darkest global reference across all images was found. This served as the threshold for the software to correctly determine the pores from the non-porous elements in the sample. The images taken were of similar lengths and locations across all samples to ensure precise analysis. The images were converted into grayscale, and the pores were detected and overlayed based on the darkest global reference determined by the software. Finally, the percentage of the pore areas was determined based on the number of pixels the pores were covering. Figure 22 shows the images obtained from MatLab for Samples 1-H and 2-H. The red overlay shows the pore morphology and distribution as determined by MatLab from the original uploaded image.

Table 17. Porosity % in the samples (area fraction).

Vertical Specimen No.	Porosity %	Horizontal Specimen No.	Porosity %
1	30%	1	18.57%
2	42%	2	21.12%
3	1.81%	3	0.13%
4	3.61%	4	2.84%
5	6.23%	5	5.12%
6	2.94%	6	1.17%
7	5.4%	7	4.85%
8	0.85%	8	0.28%

shows the porosity percentage of each sample.

From Table 17, it is evident that the porosities in Specimens 1 and 2 of both the vertical and horizontal orientations are high (>18%) when compared to those of the other samples. The presence of high-porous regions within the structure caused rapid fractures, as the stress concentrations at these points tended to be very large. This explains their poor tensile mechanical properties. Moreover, the porosities seen in the vertical samples are much higher than those of the horizontal samples. Since the walls are being built in the vertical direction, the cooling rate after each subsequent layer is uneven. The uneven cooling in certain regions can entrap gas, causing pore formation. Samples 3–8 of both orientations showed relatively low pore formation, showcasing that the selected parameters for these builds discourage pores from appearing. This low-cost retrofitted CNC GMAW setup lacks an enclosed chamber where the temperatures can be strictly controlled. Having such integrations in the setup will promote consistent builds and consistent temperatures across the printing process. This, in turn, will reduce the factors affecting pore formation.

4. Conclusions and Future Work

This paper aimed to address the literature gap in exploring the influence of the arc voltage, wire feed rate and strain rate on the mechanical properties of GMAW WAAM stainless-streel structures for industrial applications. A full factorial L8 study was implemented, and walls were 3D-printed through a retrofitted CNC router setup. This study looked into the feasibility of setting up low-cost GMAW WAAM machines to print parts with comparable mechanical properties. An ANOVA was used to identify the contributions and significance of the input parameters for each output parameter. The following conclusions were drawn from this study:

• Samples printed with an arc voltage of 18.7 V and a wire feed rate of 6 m/min showed the lowest mechanical properties, with certain output parameters being lower than the norm. This anomaly was seen because of the presence of high levels of porosity within the structure when printed with this particular combination of input parameters. The presence of porosity will facilitate faster fracture and also affect the quality of the final part.
• The influence of the wire feed rate was seen to be the highest for all output parameters across both the horizontal and vertical samples, except for the yield strength of the vertical samples. For the yield strength of the vertical samples, the arc voltage had the dominant significance, followed by the wire feed rate.
• The strain rate had a very low influence on the mechanical properties of both the vertical and horizontal samples.
• Samples taken out from the horizontal direction had higher tensile mechanical properties because of the favorable grain orientation. The vertical samples showed marginally lower mechanical properties.
• The highest UTS (543 ± 4 MPa) was observed in horizontal samples printed with an arc voltage of 18 V and a wire feed rate of 5 m/min tested at a strain rate of 1 mm/min. The highest YS (418 ± 7 MPa), toughness (248 ± 4 MPa) and %EL (57 ± 1%) were observed in horizontal samples printed with an arc voltage of 18.7 V and a wire feed rate of 5 m/min tested at a strain rate of 1 mm/min. Using this combination gave a UTS of 537 ± 4 MPa, which is very close to the highest UTS observed (543 ± 4 MPa).
• Overall, samples from the horizontal direction, printed with an arc voltage of 18 V and a wire feed rate of 5 m/min, showed the best results in this study.

As part of future work, more input parameters, such as the gas flow rate, gas material, dwell time, layer height and travel speed, can be investigated in depth to analyze the effect of their interdependence on the tensile mechanical properties. Additionally, Digital Image Correlation can be conducted to gain a wider understanding of the behavior of such samples under tensile loading conditions.