Analysis of the Wall Geometry with Di ﬀ erent Strategies for High Deposition Wire Arc Additive Manufacturing of Mild Steel

: Additive manufacturing has gained relevance in recent decades as an alternative to the manufacture of metal parts. Among the additive technologies, those that are classiﬁed as Directed Energy Deposition (DED) are characterized by their high deposition rate, noticeably, Wire Arc Additive Manufacturing (WAAM). However, having the inability to produce parts with acceptable ﬁnal surface quality and high geometric precision is to be considered an important disadvantage in this process. In this paper, di ﬀ erent torch trajectory strategies (oscillatory motion and overlap) in the fabrication of low carbon steel walls will be compared using Gas Metal Arc Welding (GMAW)-based WAAM technology. The comparison is done with a study of the mechanical and microstructural characteristics of the produced walls and ﬁnally, addressing the productivity obtained utilizing each strategy. The oscillation strategy shows better results, regarding the utilization rate of deposited material and the ﬂatness of the upper surface, this being advantageous for subsequent machining steps.


Introduction
Additive Manufacturing (AM) is a new production concept that has generated increasing interest over the last ten years. A transformation has been promoted in the way that structurally complex parts can be manufactured with unrivaled design freedom [1]. Each AM technology consists of a particular combination of a heat source, feedstock and motion system, making it suitable for numerous applications. According to the standard terminology for AM by ASTM (ASTM F2792), AM technologies for metal components have been classified into four categories, one being the Directed Energy Deposition (DED) [2]. DED by definition is "an additive manufacturing process in which focused thermal energy is used to fuse materials by melting as they are being deposited" [3]. Amongst the DED technologies are wire-feed processes and powder-feed processes [4]. The ability to use wire as a feedstock, compared to powder, offers a reduction in price per kilogram and a high material usage efficiency. Therefore, the manufacture of large parts is ideal for the technologies that use wire as a feedstock, being the most effective additive processes for this purpose [5]. Furthermore, in these technologies no use of powder recycling systems is necessary, so a more environmentally friendly process is achieved. All these characteristics enable easier manipulation of material, thus reducing the safety and health concerns [6,7]. flexible in the generation of parts with more complex geometries that could be undertaken with a more continuous (oscillatory) strategy than multiple discrete trajectories (overlapping).

Material and Set-Up
An Addilan (Durang, Spain) WAAM 3 axis AM machine equipped with a tilt table (+2-axis) and an inert-atmosphere close chamber was used to manufacture two walls; one with each strategy, one overlapping and the other oscillating. This specific closed build chamber is imperative to work with titanium or aluminum and other reactive materials. Although the technology chosen to carry out the work presented in this paper was the GMAW, due to the versatility of this AM machine, other processes, such as GTAW or PAW, were allowed. The GMAW process was based on the melting of the material with an electric arc created between the wire and the part. The energy generated by this electric arc was used to melt the material that was deposited on the top of the substrate and generated wall growth layer by layer. The build envelope for manufacturing parts was 1300 mm × 900 mm × 500 mm with a maximum weight of 300 kg. The additive manufacturing system was equipped with: a MIG/MAG welding power source, Titan XQ 400 AC puls, and a wire feeder, M drive 4 Rob5 XR RE, as the GMAW system (EWM, Mündersbach, Germany). Figure 1 shows a montage of the sensors used; a compact pyrometer Optris (Optris GmbH, Berlin, Germany) and a laser scanner, QuellTech Q4 (QuellTech, München, Germany), which were placed on the torch holder. At every two layers, the laser scanner recorded a cloud of points that reproduced the transversal geometry of the walls used to monitor the geometry of the deposited layers. The pyrometer was directed to the melted pool and it was used to determine cooling time. The cooling time is a non-productive period necessary between layers in order to promote the correct growth of the bead, without causing the collapse of the wall due to excess heat that melts the wall material. A personal computer was used to load the welding parameters and monitor them during the process. Finally, the data was sent to a database to be accessible remotely.
Metals 2020, 10, x; doi: FOR PEER REVIEW 3 of 20 the more commonly implemented overlapping strategy. Thus, it helps to to make the trajectories more flexible in the generation of parts with more complex geometries that could be undertaken with a more continuous (oscillatory) strategy than multiple discrete trajectories (overlapping).

Material and Set-Up
An Addilan (Durang, Spain) WAAM 3 axis AM machine equipped with a tilt table (+2-axis) and an inert-atmosphere close chamber was used to manufacture two walls; one with each strategy, one overlapping and the other oscillating. This specific closed build chamber is imperative to work with titanium or aluminum and other reactive materials. Although the technology chosen to carry out the work presented in this paper was the GMAW, due to the versatility of this AM machine, other processes, such as GTAW or PAW, were allowed. The GMAW process was based on the melting of the material with an electric arc created between the wire and the part. The energy generated by this electric arc was used to melt the material that was deposited on the top of the substrate and generated wall growth layer by layer. The build envelope for manufacturing parts was 1300 mm × 900 mm × 500 mm with a maximum weight of 300 kg. The additive manufacturing system was equipped with: a MIG/MAG welding power source, Titan XQ 400 AC puls, and a wire feeder, M drive 4 Rob5 XR RE, as the GMAW system (EWM, Mündersbach, Germany). Figure 1 shows a montage of the sensors used; a compact pyrometer Optris (Optris GmbH, Berlin, Germany) and a laser scanner, QuellTech Q4 (QuellTech, München, Germany), which were placed on the torch holder. At every two layers, the laser scanner recorded a cloud of points that reproduced the transversal geometry of the walls used to monitor the geometry of the deposited layers. The pyrometer was directed to the melted pool and it was used to determine cooling time. The cooling time is a non-productive period necessary between layers in order to promote the correct growth of the bead, without causing the collapse of the wall due to excess heat that melts the wall material. A personal computer was used to load the welding parameters and monitor them during the process. Finally, the data was sent to a database to be accessible remotely. The wire material used for GMAW manufacturing of the walls in both strategies was ER70S-6 mild steel (DIN 8559SG2). The composition of ER70S-6 has been provided by the supplier in weight %: Mn-1.64, Si-0.94, C-0.06, Cr-0.02, Cu-0.02, Ni-0.02, S-0.016, P-0.013, Mo-0.005, Ti-0.004, Zr-0.002, Fe-bal. The diameter of the wire dwire, to be used as raw material, was 1.2 mm. The substrate plate was S235JR steel (DIN ST37-2) in sheet form with a thickness of 8 mm. The wire material used for GMAW manufacturing of the walls in both strategies was ER70S-6 mild steel (DIN 8559SG2). The composition of ER70S-6 has been provided by the supplier in weight %: Mn-1.64, Si-0.94, C-0.06, Cr-0.02, Cu-0.02, Ni-0.02, S-0.016, P-0.013, Mo-0.005, Ti-0.004, Zr-0.002, Fe-bal. The diameter of the wire d wire , to be used as raw material, was 1.2 mm. The substrate plate was S235JR steel (DIN ST37-2) in sheet form with a thickness of 8 mm.

Deposition Strategy and Metal Transfer Mode
This section describes the two strategies used for conducting tests. One wall has been fabricated in each strategy. The walls' dimensions were l: 220 mm long straight walls with the same w: width (20 mm) and z: height (70 mm). Figure 2 corresponds to the programmed path on the machine axis in both strategies. The first strategy used sought to cover 20 mm in width by overlapping three beads with a rate of 65% (6.9 mm) of the expected width of one single weld pass. The second strategy consisted of an oscillatory triangular movement of the torch. The peak to peak amplitude of the oscillation was 20 mm, which was compensated with a tool offset equal to half of the width of the expected weld pass, and a frequency of 1.5 Hz. In both walls, prior to the return trajectory, a dwell time was waited until the pyrometer read a temperature lower than 400 • C, to make the deposition over a solidified material. Therefore, in the oscillated strategy, a unique dwell time between layers was expected, and, in the overlapped strategy, one dwell time at each bead with a total of three per layer. In both strategies, the return trajectories ran in the opposite direction (starting from the other side of the weld pass) to compensate for the geometry of the welding pass. This decompensation was created due to the starts and stops of the process. On the other hand, thanks to the laser scanner, at every two layers a geometry of the walls has been represented to manually determine the Z position for the next layer trajectory on both strategies. Lasers installed on the machine gave a total amount of 244 points per profile and one profile each 2 mm. The transfer mode selected was the MAG-PULS.

Deposition Strategy and Metal Transfer Mode
This section describes the two strategies used for conducting tests. One wall has been fabricated in each strategy. The walls' dimensions were l: 220 mm long straight walls with the same w: width (20 mm) and z: height (70 mm). Figure 2 corresponds to the programmed path on the machine axis in both strategies. The first strategy used sought to cover 20 mm in width by overlapping three beads with a rate of 65% (6.9 mm) of the expected width of one single weld pass. The second strategy consisted of an oscillatory triangular movement of the torch. The peak to peak amplitude of the oscillation was 20 mm, which was compensated with a tool offset equal to half of the width of the expected weld pass, and a frequency of 1.5 Hz. In both walls, prior to the return trajectory, a dwell time was waited until the pyrometer read a temperature lower than 400 °C, to make the deposition over a solidified material. Therefore, in the oscillated strategy, a unique dwell time between layers was expected, and, in the overlapped strategy, one dwell time at each bead with a total of three per layer. In both strategies, the return trajectories ran in the opposite direction (starting from the other side of the weld pass) to compensate for the geometry of the welding pass. This decompensation was created due to the starts and stops of the process. On the other hand, thanks to the laser scanner, at every two layers a geometry of the walls has been represented to manually determine the Z position for the next layer trajectory on both strategies. Lasers installed on the machine gave a total amount of 244 points per profile and one profile each 2 mm. The transfer mode selected was the MAG-PULS. Within the variants of GMAW, the synergic pulsed GMAW (MIG/MAG) was selected. This working mode is one of the most used these days due to its operating method, which consists of unit current pulses digitally controlled from the power source. This provides identical droplet formation of molten material with a predetermined volume correlated to the electrode wire [16]. The synergic pulsed GMAW's main advantage is to be able to ensure stable burn off of the wire. Figure 3 shows the signal gathered from the power source during drop formation with an OROS data acquisition system (OROS GmbH, Koblenz, Germany) at 12.8 kHz frequency. The current rose with a controlled slope until it reached the peak current in which the droplet was formed, this current fell, making a Within the variants of GMAW, the synergic pulsed GMAW (MIG/MAG) was selected. This working mode is one of the most used these days due to its operating method, which consists of unit current pulses digitally controlled from the power source. This provides identical droplet formation of molten material with a predetermined volume correlated to the electrode wire [16]. The synergic pulsed GMAW's main advantage is to be able to ensure stable burn off of the wire. Figure 3 shows the signal gathered from the power source during drop formation with an OROS data acquisition system (OROS GmbH, Koblenz, Germany) at 12.8 kHz frequency. The current rose with a controlled slope Metals 2020, 10, 892 5 of 18 until it reached the peak current in which the droplet was formed, this current fell, making a stop in an intermediate current for a short period of time until it reached the base current, in which the drop released. The period of the wave was 7.23 msec, which made a droplet deposition frequency of 138.2 Hz. The current parameters were established automatically through the power source database based on the wire speed, shielding gas, diameter and material of the electrode. The selected parameters were: steel, 80/20 Argon CO 2 shielding gas, wire diameter of 1.2 mm and wire feed speed of 8 m/min.   Figure 3 shows the signal acquired from the power source for mild steel deposition with a wire feed speed of 8 m/min, common in both strategies. As it was a pulsed signal, each of the peak-to-base steps produced a drop. Therefore, the energy necessary to produce each drop can be considered constant in both strategies. From now on it will be considered that the number of drops defines the energy deposited in the parts. Furthermore, the following Figure 4 shows the average current and voltage per wall pass. It was observed that they are comparable in both strategies, therefore its product that defines the energy was equal in both cases.

Preliminary Study for Parameters Selection
A preliminary study for parameter selection was carried out, designing a test battery in order to find the optimal process parameters to manufacture the walls. In this test battery, unitary beads were manufactured and scanned with a laser scanner. To design the preliminary experiment, within the range of possible wire feed speeds, 5 levels were chosen and a full factorial was completed, taking three levels of travel speed, as can be seen in Table 1.  Figure 3 shows the signal acquired from the power source for mild steel deposition with a wire feed speed of 8 m/min, common in both strategies. As it was a pulsed signal, each of the peak-to-base steps produced a drop. Therefore, the energy necessary to produce each drop can be considered constant in both strategies. From now on it will be considered that the number of drops defines the energy deposited in the parts. Furthermore, the following Figure 4 shows the average current and voltage per wall pass. It was observed that they are comparable in both strategies, therefore its product that defines the energy was equal in both cases.   Figure 3 shows the signal acquired from the power source for mild steel deposition with a wire feed speed of 8 m/min, common in both strategies. As it was a pulsed signal, each of the peak-to-base steps produced a drop. Therefore, the energy necessary to produce each drop can be considered constant in both strategies. From now on it will be considered that the number of drops defines the energy deposited in the parts. Furthermore, the following Figure 4 shows the average current and voltage per wall pass. It was observed that they are comparable in both strategies, therefore its product that defines the energy was equal in both cases.

Preliminary Study for Parameters Selection
A preliminary study for parameter selection was carried out, designing a test battery in order to find the optimal process parameters to manufacture the walls. In this test battery, unitary beads were manufactured and scanned with a laser scanner. To design the preliminary experiment, within the range of possible wire feed speeds, 5 levels were chosen and a full factorial was completed, taking three levels of travel speed, as can be seen in Table 1.

Preliminary Study for Parameters Selection
A preliminary study for parameter selection was carried out, designing a test battery in order to find the optimal process parameters to manufacture the walls. In this test battery, unitary beads were manufactured and scanned with a laser scanner. To design the preliminary experiment, within the range of possible wire feed speeds, 5 levels were chosen and a full factorial was completed, taking three levels of travel speed, as can be seen in Table 1. After the manufacture of the beads, these have been analyzed, taking into account the homogeneity, continuity, geometrical shape, penetration and dilution. To determine the penetration and dilution, macrographs of the transversal face of the beads were analyzed. Regarding the geometrical shape, to measure the width and height of the beads, the laser scanner results were utilized obtaining images, such as Figure 5a.  After the manufacture of the beads, these have been analyzed, taking into account the homogeneity, continuity, geometrical shape, penetration and dilution. To determine the penetration and dilution, macrographs of the transversal face of the beads were analyzed. Regarding the geometrical shape, to measure the width and height of the beads, the laser scanner results were utilized obtaining images, such as Figure 5a. Nine of the profiles generated by the laser in each scan have been acquired to obtain a cloud of points that reproduces the traversal geometry of each bead (Figure 5a). The width and height of the beads were calculated from each of these profiles and the average of these values (wavg and havg) can be found in Table 2. Furthermore, in Table 2, the product of the inverse of the standard deviations of height and width can be seen. This inverse value is a geometrical feature (GF). The GF indicates the level of homogeneity and continuity of the beads and it was achieved following the Equation (1): Nine of the profiles generated by the laser in each scan have been acquired to obtain a cloud of points that reproduces the traversal geometry of each bead (Figure 5a). The width and height of the beads were calculated from each of these profiles and the average of these values (wavg and havg) can be found in Table 2. Furthermore, in Table 2, the product of the inverse of the standard deviations of height and width can be seen. This inverse value is a geometrical feature (GF). The GF indicates the level of homogeneity and continuity of the beads and it was achieved following the Equation (1): where, In addition, a wetting angle value can be found in Table 2 calculated as Equation (2): Once all the beads were measured employing the laser, the optimal welding parameters were determined, taking into account the maximum GF value and a desirable wetting angle [13]. In previous literature, it has been documented that a desirable wetting angle for additive manufacturing processes is 30 • . Therefore, it was decided that the parameters used in bead 8 are the optimal ones.
After the characterization of the optimal welding parameters, the selected ones can be seen in Table 3. The resulting speeds to cover the wall length of the walls were comparable, as can be seen in the subsequent kinetic analysis. In this analysis, if the length of the path to cover the 200 mm in the X direction of the target wall was compared in the overlapped strategy, the torch ran 627.6 mm (3 mm × 200 mm of each welding pass and idle movement) and 200 mm in the case of oscillation. If we take into account the command traverse speeds (65 cm/min for overlapping and 20 cm/min for oscillating strategy), the addition time was comparable in both cases (1 min). The same wire feed lead to an equal deposition rate.

Mechanical and Microstructural Characterization Method
The walls manufactured by GMAW-based WAAM using both strategies were analyzed. In addition, an analysis of the microstructure of the material was performed and mechanical tests were carried out: the tensile test and the Charpy test.
For the mechanical characterization of both walls, six tensile specimens were extracted in the vertical direction, six tensile specimens in the horizontal direction and four vertical specimens for the Charpy Impact test. Furthermore, another three samples in each wall were utilized to analyse the microstructure and the hardness. Figure 6 shows the arrangement of the specimens on the additive wall. We had the cylindrical specimens for tensile testing and rectangular pieces for the Charpy test. The arrangement of the specimens sought to analyze the influence of the directionality of the material on its mechanical behavior.

Mechanical and Microstructural Characterization Method
The walls manufactured by GMAW-based WAAM using both strategies were analyzed. In addition, an analysis of the microstructure of the material was performed and mechanical tests were carried out: the tensile test and the Charpy test.
For the mechanical characterization of both walls, six tensile specimens were extracted in the vertical direction, six tensile specimens in the horizontal direction and four vertical specimens for the Charpy Impact test. Furthermore, another three samples in each wall were utilized to analyse the microstructure and the hardness. Figure 6 shows the arrangement of the specimens on the additive wall. We had the cylindrical specimens for tensile testing and rectangular pieces for the Charpy test. The arrangement of the specimens sought to analyze the influence of the directionality of the material on its mechanical behavior. Wire Arc Additive Manufacturing showed anisotropy on the macrographic structure (disposition on the wall inner surface) due to the adding of material layer-by-layer. As can be seen in Figure 7, the arrangement of the layers was different in the case of oscillatory and overlapping strategies. The presence of inter-layer boundaries in the case of oscillation were observed since each layer was formed by a single bead, and in the case of the three overlapping beads that form each layer, inter-bead boundaries were noticed. The existence of an interface between the overlapping weld passes meant that there was less continuity inside the wall. To analyze the grain structure, Salmi et al. [17] present remarkable results using ultrasonic burnishing that increases hardness and quality on the surface compared with the typical post-processing techniques. Although these results are promising, in this paper, metallographic samples were extracted by conventional milling from the inner surface of the additive manufactured walls. Then, the specimens were mechanically polished and finally etched with a solution of 2% Nital-Nitric and Ethanol acidto reveal the grain structure, as suggested by MacKenzie [18]. In addition, hardness testing (Vickers hardness) was performed on the same samples at room temperature using a Struers Duramin A-300 machine (Struers, Ballerup, Denmark). Wire Arc Additive Manufacturing showed anisotropy on the macrographic structure (disposition on the wall inner surface) due to the adding of material layer-by-layer. As can be seen in Figure 7, the arrangement of the layers was different in the case of oscillatory and overlapping strategies. The presence of inter-layer boundaries in the case of oscillation were observed since each layer was formed by a single bead, and in the case of the three overlapping beads that form each layer, inter-bead boundaries were noticed. The existence of an interface between the overlapping weld passes meant that there was less continuity inside the wall. To analyze the grain structure, Salmi et al. [17] present remarkable results using ultrasonic burnishing that increases hardness and quality on the surface compared with the typical post-processing techniques. Although these results are promising, in this paper, metallographic samples were extracted by conventional milling from the inner surface of the additive manufactured walls. Then, the specimens were mechanically polished and finally etched with a solution of 2% Nital-Nitric and Ethanol acidto reveal the grain structure, as suggested by MacKenzie [18]. In addition, hardness testing (Vickers hardness) was performed on the same samples at room temperature using a Struers Duramin A-300 machine (Struers, Ballerup, Denmark).
Therefore, the selection of specimens on different orientations on the wall was done. The dimensions of the tensile specimens followed the standard ISO 6892-1 [19], with a narrow diameter of 4 mm and a length of 22 mm, with threads at the ends of M-6. The machine where the uniaxial tensile tests were performed at room temperature was the Instron 5585H, with a maximum load of 100 kN and equipped with a contact extensometer, Instron EX2620-602 (Instron, Norwood, MA, USA). Charpy tests were performed at room temperature on an impact test pendulum Ibertest PIB300 machine (S.A.E. Ibertest, Madrid, Spain). Following standard ISO 148-1, Charpy impact test specimens with a 55 mm length and 10 mm thickness were used. The specimens were selected in the vertical direction from both walls.
A measurement of the geometry of both walls was addressed by means of the volumetric reconstruction of the profiles acquired in each layer. This measurement was made by means of the laser scanner installed and attached to the torch shown in Figure 1. Therefore, the selection of specimens on different orientations on the wall was done. The dimensions of the tensile specimens followed the standard ISO 6892-1 [19], with a narrow diameter of 4 mm and a length of 22 mm, with threads at the ends of M-6. The machine where the uniaxial tensile tests were performed at room temperature was the Instron 5585H, with a maximum load of 100 kN and equipped with a contact extensometer, Instron EX2620-602 (Instron, Norwood, MA, USA). Charpy tests were performed at room temperature on an impact test pendulum Ibertest PIB300 machine (S.A.E. Ibertest, Madrid, Spain). Following standard ISO 148-1, Charpy impact test specimens with a 55 mm length and 10 mm thickness were used. The specimens were selected in the vertical direction from both walls.
A measurement of the geometry of both walls was addressed by means of the volumetric reconstruction of the profiles acquired in each layer. This measurement was made by means of the laser scanner installed and attached to the torch shown in Figure 1.

Process Characterization
To determine the heat input effect on the manufactured walls, it is important, first of all, to understand the process kinetics. In the case of the overlapped wall, the travel speed is constant throughout all the trajectory, resulting in a uniform deposition. On the other hand, a cycle of the torch trajectory in the oscillated wall and its travel speed distribution is shown in Figure 8. In this case, the speed is lower on the sides than in the central part, becoming null in the corners due to the change of direction. Thus, the material deposition is non-uniform along the entire path.

Process Characterization
To determine the heat input effect on the manufactured walls, it is important, first of all, to understand the process kinetics. In the case of the overlapped wall, the travel speed is constant throughout all the trajectory, resulting in a uniform deposition. On the other hand, a cycle of the torch trajectory in the oscillated wall and its travel speed distribution is shown in Figure 8. In this case, the speed is lower on the sides than in the central part, becoming null in the corners due to the change of direction. Thus, the material deposition is non-uniform along the entire path. To obtain the number of drops per advanced millimeter, it would be sufficient to integrate the product of the instantaneous speed and the advance travelled in the x and y direction. In the case of overlap, it should be taken into account that it is necessary to overlap three weld passes to travel the same distance and that the instantaneous speed is constant and equal to the torch speed (Ts) (Equation (3)). In the case of oscillation, starting from the instantaneous speed as a function of the position of the torch registered in the control (V (x, y)), it is multiplied by the differential advance. In both cases, the droplet deposition frequency is given by the values of the power source with the selected transfer mode (f = 138.2 drops/sec). In Equation (4), the integration has been calculated by means of its resolution based on the sum of the instantaneous speeds and positions registered in the monitoring system.  To obtain the number of drops per advanced millimeter, it would be sufficient to integrate the product of the instantaneous speed and the advance travelled in the x and y direction. In the case of overlap, it should be taken into account that it is necessary to overlap three weld passes to travel the same distance and that the instantaneous speed is constant and equal to the torch speed (Ts) (Equation (3)). In the case of oscillation, starting from the instantaneous speed as a function of the position of the torch registered in the control (V (x, y)), it is multiplied by the differential advance. In both cases, the droplet deposition frequency is given by the values of the power source with the selected transfer mode (f = 138.2 drops/sec). In Equation (4), the integration has been calculated by means of its resolution based on the sum of the instantaneous speeds and positions registered in the monitoring system.  Table 4 below shows the comparison of droplets deposited per advanced millimeter in both torch movement strategies. In the case of oscillatory movement, a greater number of drops per advanced mm is observed with a more homogeneous distribution, according to the trajectory, as shown in Figure 8. In addition to the drop transfer analysis, Figure 9 shows the instantaneous deposition energy calculated as the product of the current and the voltage divided by the speed of the torch. This deposition energy is constant in the case of the overlapping strategy and it is variable due to the kinetics of the process on the oscillation strategy, as shown in Figure 8. In order to ensure that the motion system has completed what was programmed in the manufacturing of the oscillated wall, Figure 10 presents the generated bead in the first layer. The width of the bead is slightly less than 20 mm of the target wall, because in layer 0 the substrate is cold and there is greater growth in the Z direction and a slight decrease in width.
The marks appreciated in Figure 10 are generated due to the oscillatory path and, by measuring the distance between two consecutive marks, the actual wavelength of the process can be determined. In order to ensure that the motion system has completed what was programmed in the manufacturing of the oscillated wall, Figure 10 presents the generated bead in the first layer. The width of the bead is slightly less than 20 mm of the target wall, because in layer 0 the substrate is cold and there is greater growth in the Z direction and a slight decrease in width.
The marks appreciated in Figure 10 are generated due to the oscillatory path and, by measuring the distance between two consecutive marks, the actual wavelength of the process can be determined. Five marks of the first bead are bordered by a red box ( Figure 10) and the luminance values of the bordered part of the photo are analyzed, determining that the actual wavelength is 2.30 mm. Therefore, as the theoretical programmed wavelength was 2.22 mm, it can be concluded that the actual trajectory is in accordance with the programmed one. The complete zigzag shape is not observed on the top surface of the bead, since the theoretical programmed wavelength value is small, without producing marks on the forward side direction because they are removed by the next movement. Five marks of the first bead are bordered by a red box ( Figure 10) and the luminance values of the bordered part of the photo are analyzed, determining that the actual wavelength is 2.30 mm. Therefore, as the theoretical programmed wavelength was 2.22 mm, it can be concluded that the actual trajectory is in accordance with the programmed one. The complete zigzag shape is not observed on the top surface of the bead, since the theoretical programmed wavelength value is small, without producing marks on the forward side direction because they are removed by the next movement. To highlight this point, Figure 11 presents the evolution of Z, the measured height of the wall along both strategies. It can be observed in the first layers that growth is greater in the oscillated wall than in the overlapped wall because at the beginning the substrate is at atmospheric temperature and the energy input and cooling times are not yet stable. After these first layers, the overlapped wall presents greater growth than in the oscillated wall, and the number of layers necessary to obtain the desired height is therefore reduced, 40 layers and 48 layers, respectively. The incidence of a more concentrated heat in the oscillatory strategy produces a greater dilution in this case, therefore a greater number of layers is necessary for the same growth in the Z direction. To highlight this point, Figure 11 presents the evolution of Z, the measured height of the wall along both strategies. It can be observed in the first layers that growth is greater in the oscillated wall than in the overlapped wall because at the beginning the substrate is at atmospheric temperature and the energy input and cooling times are not yet stable. After these first layers, the overlapped wall presents greater growth than in the oscillated wall, and the number of layers necessary to obtain the desired height is therefore reduced, 40 layers and 48 layers, respectively. The incidence of a more concentrated heat in the oscillatory strategy produces a greater dilution in this case, therefore a greater number of layers is necessary for the same growth in the Z direction. The manufacturing time is calculated as a summation of the adding time and the idle time, which considers the time to cool down the material between layers and non-productive movements. Figure  12 shows the surface temperatures of the welding pass during the addition and the idle time on both oscillating and overlapping strategies. The summation of the three intralayer times on the overlapping strategy supposes a longer run time than the interlayer time after manufacturing with an oscillatory strategy. The manufacturing time is longer in the case of the overlapping strategy (2h 2min) compared to values of the oscillating strategy (1h 25min). The main differences appear when idle time is compared, since three start-stop cycles must be done on the overlapping strategy. The idle time is 1h15min for overlapping and 32min for the oscillating strategy. On the other hand, overlapping adding time is more productive, reachi  The manufacturing time is calculated as a summation of the adding time and the idle time, which considers the time to cool down the material between layers and non-productive movements. Figure 12 shows the surface temperatures of the welding pass during the addition and the idle time on both oscillating and overlapping strategies. The summation of the three intralayer times on the overlapping strategy supposes a longer run time than the interlayer time after manufacturing with an oscillatory strategy. The manufacturing time is longer in the case of the overlapping strategy (2 h 2 min) compared to values of the oscillating strategy (1 h 25 min). The main differences appear when idle time is compared, since three start-stop cycles must be done on the overlapping strategy. The idle time is 1 h 15 min for overlapping and 32 min for the oscillating strategy. On the other hand, overlapping adding time is more productive, reaching a value of 46 min as traverse speed is faster than the oscillating strategy. Adding time using an oscillating strategy is 52 min. The manufacturing time is calculated as a summation of the adding time and the idle time, which considers the time to cool down the material between layers and non-productive movements. Figure  12 shows the surface temperatures of the welding pass during the addition and the idle time on both oscillating and overlapping strategies. The summation of the three intralayer times on the overlapping strategy supposes a longer run time than the interlayer time after manufacturing with an oscillatory strategy. The manufacturing time is longer in the case of the overlapping strategy (2h 2min) compared to values of the oscillating strategy (1h 25min). The main differences appear when idle time is compared, since three start-stop cycles must be done on the overlapping strategy. The idle time is 1h15min for overlapping and 32min for the oscillating strategy. On the other hand, overlapping adding time is more productive, reachi

Mechanical Characterization
This section shows the mechanical results obtained from the wall test produced by overlapping and by oscillatory strategies. It is well known that solidification defects that may appear in WAAM samples, for example, porosity, lack of fusion, localized brittle zones, and grain coarsening in the heat-affected zones, can promote scatter in mechanical properties [20]. In Figure 13, the stress-strain curves of each test sample are represented. In the obtained values, some scatter is appreciated, especially in elongation. It has been shown that, depending on the thermal history of each area, location-dependent mechanical properties are obtained [21]. As can be seen in Figure 6, six horizontal and six vertical tensile test specimens are cut from the walls. The results show slightly higher ultimate tensile strength (UTS) and yield strength (YS) values in the case of overlapping strategy, especially in the case of tests in the vertical direction. Table 5 summarizes the results of the mechanical testing of the oscillating and overlapping walls. The standard deviation in the results of UTS and YS in the horizontal direction are greater than those observed in the vertical direction, as can already be seen in Figure 13. Charpy tests have been performed in order to analyze the influence of the bead direction on the fracture toughness. Although, in the literature [22,23] results on Charpy testing show larger scatter than the tensile test results in low carbon steels, which is created due to the different microstructures that can be seen throughout the part, for example the acicular ferrite lead to superior impact toughness results than polygonal ferrite [24], in the present study, the impact test results are similar in the two compared walls. On the other hand, Vickers hardness testing has been also performed at these walls. As it was pointed out by Salmi et al. [17] the samples extraction procedure can affect the obtained hardness values. Average values observed in the overlapping strategy are similar to the ones observed in the oscillating strategy. Results are similar to those addressed by [15] and slightly inferior to those reported by the wire supplier for this material as welded.

Mechanical Characterization
This section shows the mechanical results obtained from the wall test produced by overlapping and by oscillatory strategies. It is well known that solidification defects that may appear in WAAM samples, for example, porosity, lack of fusion, localized brittle zones, and grain coarsening in the heat-affected zones, can promote scatter in mechanical properties [20]. In Figure 13, the stress-strain curves of each test sample are represented. In the obtained values, some scatter is appreciated, especially in elongation. It has been shown that, depending on the thermal history of each area, location-dependent mechanical properties are obtained [21]. As can be seen in Figure 6, six horizontal and six vertical tensile test specimens are cut from the walls. The results show slightly higher ultimate tensile strength (UTS) and yield strength (YS) values in the case of overlapping strategy, especially in the case of tests in the vertical direction.  Table 5 summarizes the results of the mechanical testing of the oscillating and overlapping walls. The standard deviation in the results of UTS and YS in the horizontal direction are greater than those observed in the vertical direction, as can already be seen in Figure 13. Charpy tests have been performed in order to analyze the influence of the bead direction on the fracture toughness. Although, in the literature [22,23] results on Charpy testing show larger scatter than the tensile test results in low carbon steels, which is created due to the different microstructures that can be seen throughout the part, for example the acicular ferrite lead to superior impact toughness results than

Analysis of the Wall Geometry
In this section, the final geometry of the manufactured wall is analyzed utilizing the different strategies of deposition: oscillatory strategy and overlapping of linear welds. The flatness as it is defined by [ASME Y14.5 standard] "specifies a tolerance zone defined by two parallel planes within which the surface must lie", having been calculated as the difference between the major distance to mean value and the minor distance to the height on the upper surface of the wall. Figure 14 shows the point cloud that is defined by each of the profile readings made by the laser. The two envelope lines that define the maximum profile and, above all, the minimum usable profile, are also shown. Lastly, the mean profile line between these two envelopes is shown. As can be seen, the flatness observed in the case of the use of an oscillatory strategy is greater than that of overlapping. The profile presented by the oscillation case is parabolic. Therefore, the material flows towards the central zone from the edges of the bead, even though, as shown in Table 4, a greater number of drops are deposited in that zone. This indicates that the material flows backward and upwards into the melt pool in the limit zone, going to the hot zone and avoiding the greater surface tension in the limit zone. The same effect was reported by Ma et al. [25], obtaining a flatter layer with a weaving strategy. This behavior observed in GMAW is different from that modeled for the case of PAW by the authors, Bai et al. [26], in which it was discovered that metal liquid flow was upward inside and outward on the surface.
Regarding the shape observed in the upper layer, it is observed that the wall obtained by means of oscillatory strategy is quasi-symmetrical, due to the nature of the movement of the torch and the movement of the metal in a fluid state. In contrast, in the case of the overlapping strategy, it can be seen how the growth of the lateral welds is less than that of the central weld due to the anisotropic configuration of the wall once each of the welds is manufactured. The production of a wall with a homogeneous top surface is critical for the subsequent face milling process and obtaining the finished wall. In previous works, Veiga et al. [27] have shown how face milling is a more critical machining process in this type of wall than tangential milling. The dimensions of the final wall will be determined by the minimum measured profile, shown in Figure 15, of the parabolic nature of the measured profiles; an effective wall of the rectangular cross-sectional area could be obtained. This cross-sectional area is of variable character depending on the effective height and width in which the target square is inscribed. In Figure 15, the effective crosssectional area in both strategies is shown in red for the target width of 20 mm. The maximum effective height at which a wall of this height would be obtained would be greater in the case of an oscillatory strategy, this height being 72.2 mm compared to the 71.3 mm height observed in the case of walls obtained by overlapping. Although this difference is not very large, it should be noted that due to The dimensions of the final wall will be determined by the minimum measured profile, shown in Figure 15, of the parabolic nature of the measured profiles; an effective wall of the rectangular cross-sectional area could be obtained. This cross-sectional area is of variable character depending on the effective height and width in which the target square is inscribed. In Figure 15, the effective cross-sectional area in both strategies is shown in red for the target width of 20 mm. The maximum effective height at which a wall of this height would be obtained would be greater in the case of an oscillatory strategy, this height being 72.2 mm compared to the 71.3 mm height observed in the case of walls obtained by overlapping. Although this difference is not very large, it should be noted that due to the greater homogeneity of the profile in the case of an oscillatory strategy, a higher maximum effective height is obtained in a wall with a maximum value of measured height lower than that observed in the case of an overlapping strategy. In the following Figure 16, the cross-sectional area value is shown considering the rectangular wall section. This area is calculated as the effective height value for the different wall widths that can be obtained with both strategies. The oscillatory strategy allows better use of the material provided due to the greater homogeneity of the measured profile, even finding Z-level values, corresponding to the height, in the case of the overlapping strategy. In general, this produces a greater effective cross-sectional area in the case of an oscillatory strategy, in addition to a smaller volume of shavings removed to obtain the final wall. This fact entails a lower cost in subsequent machining processes. In the following Figure 16, the cross-sectional area value is shown considering the rectangular wall section. This area is calculated as the effective height value for the different wall widths that can be obtained with both strategies. The oscillatory strategy allows better use of the material provided due to the greater homogeneity of the measured profile, even finding Z-level values, corresponding to the height, in the case of the overlapping strategy. In general, this produces a greater effective cross-sectional area in the case of an oscillatory strategy, in addition to a smaller volume of shavings removed to obtain the final wall. This fact entails a lower cost in subsequent machining processes.

Conclusions
Low-carbon steel (ER70S-6) walls were manufactured with the GMAW-WAAM technique. The oscillatory and overlapping strategies were used and compared in the fabrication of 20 mm × 70 mm × 220 mm walls. Different process aspects and parameters were compared to illustrate a clear idea of the advantages and disadvantages of each one. The main aspects studied were the material deposition, the growth per layer, the heat input, the fabrication time, the microstructure, the mechanical properties and the wall geometry. Some conclusions drawn from this comparison are:

•
In this paper, two manufacturing strategies have been evaluated in terms of geometry and mechanical properties (tensile strength, hardness and Charpy). Similar mechanical values and manufacturing times have been observed. The manufacture of parts with different and complementary strategies increases the adaptation of technology to complex geometries, allowing the application in parts that have crossings, variable wall thickness or curvature.

•
The non-productive idle time establishes the differences observed on productivity in each case. The cooling time (part of the idle time where the wall is at rest until it reaches the desired temperature) is lower in the oscillated wall. The oscillated wall consists of a continuous trajectory on each layer; on the other hand, the overlapping strategy requires three discrete trajectories to cover the same width. After each of these beads, a stop must be made. Therefore, this waste of time makes the overlapping strategy less attractive from a productivity point of view.

•
The tensile tests show similar data in both strategies, but certain differences derived from their operating principle can be concluded. Although the Yield Strength and UTS are slightly higher in the case of overlapping trajectory, the elongation before breakage is less. Besides this undesired behavior observed during its elongation, it is observed that the extraction direction of the specimen affects the measured elongation. In the case of the overlapping wall, the specimens in the vertical direction show lower elongation values than those shown in the horizontal direction. The oscillatory strategy presents more homogeneous values in both directions due to the more homogeneous distribution of the beads.

•
Mechanical properties were tested, and similar hardness measured in both strategies were obtained. Large variation associated with anisotropy of the material deposition and heating cycles is observed on the measured hardness. Regarding the impact study, Charpy testing shows similar results on both the overlapping and oscillating strategies.

•
The flatness in the upper layer is improved thanks to the use of the oscillatory strategy. In the wall made by overlapping, a greater growth of the middle wall is observed, which worsens the

Conclusions
Low-carbon steel (ER70S-6) walls were manufactured with the GMAW-WAAM technique. The oscillatory and overlapping strategies were used and compared in the fabrication of 20 mm × 70 mm × 220 mm walls. Different process aspects and parameters were compared to illustrate a clear idea of the advantages and disadvantages of each one. The main aspects studied were the material deposition, the growth per layer, the heat input, the fabrication time, the microstructure, the mechanical properties and the wall geometry. Some conclusions drawn from this comparison are:

•
In this paper, two manufacturing strategies have been evaluated in terms of geometry and mechanical properties (tensile strength, hardness and Charpy). Similar mechanical values and manufacturing times have been observed. The manufacture of parts with different and complementary strategies increases the adaptation of technology to complex geometries, allowing the application in parts that have crossings, variable wall thickness or curvature.

•
The non-productive idle time establishes the differences observed on productivity in each case. The cooling time (part of the idle time where the wall is at rest until it reaches the desired temperature) is lower in the oscillated wall. The oscillated wall consists of a continuous trajectory on each layer; on the other hand, the overlapping strategy requires three discrete trajectories to cover the same width. After each of these beads, a stop must be made. Therefore, this waste of time makes the overlapping strategy less attractive from a productivity point of view.

•
The tensile tests show similar data in both strategies, but certain differences derived from their operating principle can be concluded. Although the Yield Strength and UTS are slightly higher in the case of overlapping trajectory, the elongation before breakage is less. Besides this undesired behavior observed during its elongation, it is observed that the extraction direction of the specimen affects the measured elongation. In the case of the overlapping wall, the specimens in the vertical direction show lower elongation values than those shown in the horizontal direction. The oscillatory strategy presents more homogeneous values in both directions due to the more homogeneous distribution of the beads.

•
Mechanical properties were tested, and similar hardness measured in both strategies were obtained. Large variation associated with anisotropy of the material deposition and heating cycles is observed on the measured hardness. Regarding the impact study, Charpy testing shows similar results on both the overlapping and oscillating strategies.

•
The flatness in the upper layer is improved thanks to the use of the oscillatory strategy. In the wall made by overlapping, a greater growth of the middle wall is observed, which worsens the flatness values. This fact affects the surface finishing process, from additive surface to finished surface.

•
Finally, the analysis of wall geometry is important because it determines the achieved effective wall geometry and it has an influence on the final part machining. The oscillating wall presents improved flatness, and it is quasi-symmetrical, obtaining a greater homogeneity. Using the oscillatory strategy allows better use of material, achieving a lower cost in subsequent machining. Further studies must be done to evaluate the influence of overlapping distance and frequency in the case oscillating strategies.

Conflicts of Interest:
The authors declare no conflict of interest.