1. Introduction
Wire and Arc Additive Manufacturing (WAAM) is growing in interest in different sectors such as aerospatial, maritime, and automotive, among others, and is being applied to many materials like aluminium alloys [
1,
2,
3], steels [
4], and titanium alloys [
5]. Moreover, apart from materials, technologies used are going forward in order to obtain the best advantages of WAAM: high deposition rate, short time to market, and low Buy to Fly (BTF) ratio, making it suitable for the manufacturing of large parts. WAAM is characterised by the deposition of overlapped layers obtaining a near-net-shape part and it emerges as a practical solution for the manufacturing of high-dimension parts due to its high productivity potential and associated low manufacturing times and costs.
In order to optimise the BTF ratio, a tight selection of parameters needs to be used to minimise the surface waving and to avoid internal defects [
1]. For that, a value of effective area gives a full vision of the percentage of the wall free of waving [
6] and external defects.
Most associated defects in aluminium parts manufactured by WAAM are the porosity and solidification cracks [
2,
7,
8]. Porosity is directly linked to the impairment of mechanical properties such as strength and ductility. This porosity normally appears due to the process parameters, heat input, alloy composition, interpass temperatures, and resulting microstructures [
7].
CMT (Cold Metal Transfer) technology has been widely applied for the WAAM purposes of aluminium alloys, due to the lower heat input compared to other gas metal arc welding (GMAW) processes. The metal transference during CMT is assisted by the novel mechanical oscillation movement of the wire at high frequency [
9]. Its coaxiality makes the deposition by WAAM easier regarding automation, path planning, and system complexity in comparison with other arc welding technologies such as gas tungsten arc welding (GTAW) and plasma arc welding (PAW) that need lateral wire feeding. A recent study [
8] applied CMT to 5356 aluminium alloy and concluded that deposition strategy, shielding gas, and gas flow rate directly affect porosity.
In order to increase WAAM process efficiency, which is critical for the manufacturing of large components, different alternatives have been studied so far, including the increase in the welding current employed in conventional GMAW processes [
10] and the deployment of multiple wire processes, such as tandem GMAW [
11,
12,
13] and double-wire GTAW [
14].
The tandem GMAW arc welding process is characterised by the combination of two independent welding sources that are fully synchronised [
15,
16]. Two filler metals are fed into a single torch hose pack via separate feeders and routed through electrically isolated contact tips [
17]. A single shielding gas nozzle is employed and the two arcs generate a single weld pool. The possibility of isolating each power source and synchronising them allows for implementing CMT, pulsed, and other innovative welding curves independently, leveraging a range of process combinations. In this sense, the ancient double-wire welding GMAW technologies were limited by the fact that the electrodes were not isolated and the same potential should be selected for only standard GMAW welding curves. Being a tandem GMAW arc process, the state of the art CMT-Twin process provides the typical assisted droplet detachment of the CMT process which leads to low spatter and heat input, while either welding speed or deposition rate can be doubled [
18]. Recently, these authors have investigated this innovative welding process for the WAAM of aluminium alloys [
15,
19,
20,
21,
22].
Both tandem GMAW and CMT-Twin have been successfully employed to create in situ personalised alloys by combining different wires [
18,
21,
23]. It also offers the possibility of increasing the deposition rates. This goal was studied in stainless steel by Martina et al. [
24], with CMT doubling the deposition rate for thin parts; however, for the manufacturing of thicker parts without defects, an additional cooling aid was required.
In the case of the CMT unitary torch for 5356 alloy, Köhler et al. described a methodology to obtain sound walls that met the standards related to tensile properties for the wrought material [
25]. In this case, the deposition rates achieved are between 1 and 1.2 kg/h. In order to manufacture big parts by WAAM, CMT-Twin allows for higher deposition rates, but the presence of two electric arcs increases the heat input. In this regard, heat input, heat distribution, and heat accumulation directly affect the geometry of the deposited part [
25]. Due to this, a balance between deposition rate and surface waving needs to be found. According to Tawfik et al. [
1,
26], the obtained volume fraction of pores in 5356 aluminium alloy is directly proportional to heat input. Cracking is also due to the reheating caused by the successive deposition of layers, and it affects the mechanical properties.
Obtained dimensions, microstructures, and mechanical properties highly depend on the cooling conditions during the manufacturing of a part by WAAM [
27,
28]. The successive deposition of layers of the process itself affects the previously deposited layer, creating complex microstructures, but also being a source of possible defects such as pores and segregations [
29]. With the aim of obtaining a homogenous microstructure and mechanical properties in the whole part, active cooling has been studied in recent works [
25].
The use of forced cooling such as a cooling plate has been widely applied to avoid heat accumulation [
28,
30]. Moreover, the control of interpass temperature can be used as a strategy to prevent defects such as pores and cracks. Derekar et al. [
2] concluded that an interpass temperature of 100 °C including a preheating of the substrate at the same temperature before the deposition of the first layer for 5356 and an oscillating strategy reduced the amount and size of pores.
In the fusion line, other defects such as segregations (Al
8Mg
5) at the grain boundaries are also visible in WAAM manufactured parts of 5356 due to a long heat exposure at those zones [
25]. These segregations and other heterogeneities in grain size and hardness affect the obtained mechanical properties in a vertical orientation, being poorer than in the horizontal orientation [
25].
In order to monitor the temperature of the wall for a near-net-shape part by WAAM, other works used thermography as a monitoring technique to maintain a constant interpass temperature, avoiding the use of a constant interpass dwell time [
31]. They observed that a constant interpass dwell time affects the final geometry, causing irregularities in the part weight and height.
The objective of this study was to find the highest deposition rate to manufacture a sound wall using CMT TWIN with 5356 aluminium alloy. Moreover, the effect of cooling conditions was studied by changing the interpass dwell time with the aid of thermography to monitor local temperature in the last deposited layer and the use of active cooling through a cooling plate in order to find an optimised deposition strategy with the lowest manufacturing time. Mechanical properties and microstructure are assessed to verify the best conditions.
2. Materials and Methods
WAAM parts were manufactured in an arc welding robotic cell. A Fronius TransPlus synergic (TPS) 4000 CMT R and 5000 CMT R power sources, two fully digital inverter CMT welding power sources, and a Robacta Twin Compact Pro 30° PA OVT torch from Fronius International (Fronius International, Wels, Austria) were used. A scheme of the torch and the created electric arc is shown in
Figure 1.
The welding torch was attached to a 6-axis ABB robot IRB 4600-45/2.05 model with a IRC5 controller (ABB Ltd., Zurich, Switzerland. The gas shielding of the torch was carried out with Argon (99.999% purity) and the gas flow was set at 25 L/min. A 1.2-mm diameter ER5356 wire from ESAB was employed as filler metal. The substrates for the manufacturing of the walls were of AA6082-T6 alloy. The chemical compositions of the wire and the substrate are shown in
Table 1.
In order to check the temperature during the deposition and heat accumulation, K-type thermocouples were welded to the substrate. Their location is shown in
Figure 2 with 2 stars named T1 related to thermocouple 1 and T2 related to thermocouple 2. The distance between the walls and the thermocouples is 50 mm.
The setup for the manufacturing of the walls is shown in
Figure 3. It includes a cooling plate with internal water conformal cooling connected to a chilling machine. The substrate was clamped to this cooling plate to achieve forced cooling conditions.
For the local thermal analysis, a Flir a655sc microbolometer infrared camera (Teledyne FLIR LLC, Wilsonville, OR, USA) was employed. The spectral range of the camera is 7–14 µm and its temperature calibration is up to 2000 °C. Standard 25.4 mm lenses were employed and the whole system was encapsulated. The frame rate employed was 16 frames/s. The camera was mounted on a tripod and placed at one side of the component at an approximate distance of 40 cm from the deposition zone. The temperature was always captured at the last deposited layer. The measurement of the temperature was obtained after 30 s once the torch passed.
For the first assessment of welding parameters to optimise the deposition rate, walls of 10 layers were built. In a second assessment, taller walls were manufactured (70 × 130 mm2) to check the influence of heat accumulation and mechanical properties in horizontal and vertical orientations. Pulse mode was used for the leading wire, whereas CMT mode was applied to the trailing wire.
Two different deposition strategies were used: circling and hatching. Circling strategy means an overlapping of round circles with 2 mm of amplitude and 3 Hz frequency. Ignition and extinction of the electric arc was alternated between layers in order to avoid material accumulation. A scheme of both strategies is shown in
Figure 4.
The orientation of the torch is perpendicular to the substrate. The offset in Z direction was continuously adjusted in a range of 2–2.5 mm in order to maintain a constant voltage. Depending on the interpass dwell time used and the height of the wall, the Z offset value needs to be adjusted. This interpass dwell time is defined as the time in which the arc is stopped between the deposition of consecutive layers.. The study of interpass dwell times (30, 60, 90, 120, 180, and 240 s) and the use of a cooling plate (with and without) was conducted as shown in
Table 2 where X means that the experiment was conducted and - no.
Samples were cut for microstructural characterisation and measurement of the effective area in the cross section. After cutting the samples, they were mounted, polished, and etched. Light microscopy images were taken with an Olympus GX51 microscope (Olympus Corp., Tokyo, Japan). Internal defects such us porosity and segregations were also evaluated by measurement along the height of the wall. Edges of walls were discarded for this analysis.
Mechanical characterisation was carried out in X direction (horizontal) and Z direction (vertical). The flat dog-bone tensile test specimens were extracted from the walls, according to the ASTM E8M standard [
32], by using an electron discharge machine as shown in
Figure 5. Six samples were tested for each set of cooling conditions. Tensile tests were performed in a Z100 ZWICK/Roell testing machine (ZwickRoell S.L., Barcelona, Spain) with a maximum load capacity of 100 kN. Specimens were tested at room temperature with a displacement rate of 1.6 mm/min and an extensometer with a gauge length of 25 mm.
4. Discussion
The selected weld bead for the growth of the walls assures a minimum and stable surface waving. Heat input was said to greatly influence increasing the width of the weld bead, whereas the height decreased in a lower way [
30]. Moreover, the circling strategy gave rise to thicker weld beads and hence a more stable and homogenous growth of the wall. This was also observed in previous works with the unitary torch [
8]. Concretely, in this case, the comparable dimensions of the weld bead were 8.0 mm of thickness and 3.7 mm of height.
Regarding manufactured walls with the different cooling conditions, shown in
Figure 10, the one using the lowest interpass dwell time (30 s) without the use of a cooling plate had the lowest height (39.1 mm) and the largest width (16.1 mm). On the contrary, the one using the largest interpass dwell time (90 s) with the use of a cooling plate had the largest height (49.1 mm) and the lowest width (12.7 mm). These results were expected following the same trend as the dimensions of the weld beads (
Table 4).
Cracking has been found due to the heat accumulation which forms low melting point segregations during the cooling stage as shown in
Figure 16. Therefore, we can observe them in a high amount once the wall grows in height because the heat dissipation becomes more difficult and then the interpass temperature increases, so the cooling rate greatly decreases.
An interlayer boundary zone with a further increased occurrence of segregations can be identified (
Figure 11b). In these regions, primary stages and segregations merged at the grain boundaries as a result of comparatively long heat exposure giving rise occasionally to cracks. Even if the cracks are not opened, due to the difference in hardness, grain size, and the formation of segregations, these areas can act as metallurgical notches, thus affecting the mechanical properties in the vertical loading direction, as can be observed in
Table 8.
As stated in other works, the increase in deposition rate requires additional cooling as found with other materials such as steel [
24]. As observed in
Figure 9, the effect of the use of the cooling plate in the interpass temperature can be clearly seen; however, it is more noticeable for shorter interpass dwell times, when the substrate is at higher temperatures. Other kinds of forced cooling [
33] could be of help in decreasing the interpass temperature in a greater amount; however, the obtained microstructure can be affected by obtaining supersaturated solid solution hard phases [
27] and affecting the mechanical properties.
Segregations are detected higher in the wall for an interpass dwell time of 120 s without using the cooling plate. In this case, the height at which the first segregations appear (14 mm) is close to the one measured for an interpass dwell time of 30 s (7 mm) using the cooling plate. This implies a reduction in time of fabrication.
The interpass temperature measured with the unitary torch using 90 s of interpass dwell time between layer deposition is 85 °C [
8] and 175 °C in the same conditions with TWIN torch. Comparing the heat input for the used welding parameters with TWIN torch it is 2.81 KJ/cm versus 0.72 KJ/cm for the unitary torch. Nevertheless, the use of TWIN presents interesting advantages compared with the unitary torch such as the higher deposition rate (2.87 kg/h vs. 0.93 kg/h [
8]) for the same material and without considering the interpass dwell time. Moreover, as shown in
Table 10, with the optimisation of interpass dwell time, the sum of the interpass dwell time needed for the deposition of 20 layers was equal with both torches: 30 min. The higher deposition rate obtained by the TWIN torch comprehends both of the following: higher layer height (2.36 vs. 1.85 mm) and higher layer width (12 vs. 6.2 mm) for the same welding speed.
The use of the interpass dwell time of 240 s makes the advantages of TWIN disappear. This interpass dwell time has been optimised by adjusting it to the temperature of the wall, since closer layers to the substrate can dissipate the heat more easily than the more distant ones and hence the heat accumulates with a growth in height [
13,
34]. In
Table 11 the total interpass dwell time used, and the presence of segregations, can be seen for 35 layered walls. The difference between the optimised wall and the wall with 240 s of interpass dwell time is close to 1 h. This manufacturing time is around 150 s of constant interpass dwell time, this being an acceptable interpass dwell time for an industrial approach. Real parts usually require manufacturing times of each individual layer equivalent or higher than the interpass dwell times. Moreover, a proper interpass dwell time helps increase the effective area as seen in
Table 5 [
27] and the height and thickness per layer grown also increases as shown in
Table 10. This optimisation of interpass dwell time gave rise to a reduction in manufacturing time of 36%.
Thermography was demonstrated to be a useful tool to locate the temperature and segregations to define the limit temperature to avoid segregations. Optimising the setup would help to improve the quality of the obtained results. To do so, the camera should be placed on a zenithal position with respect to the weld. By doing this, the effect of the emissivity could be homogenised. Together with this, a wider range of temperatures should be employed, in order to avoid saturation on the torch. Regarding the extracted data, further analysis should be performed, focusing not only on the final temperature after 30 s, but also on the cooling rate of the curves, which could offer more accurate information about the crack generation.
Regarding the mechanical properties, optimal results were obtained even for the shorter interpass dwell times where segregations could be found in the wall cross section. However, as stated before, these segregations appear from the surface of the wall, whereas the tensile samples were extracted from the medium zone of the wall, and therefore the cracks did not reach the tensile samples. However, in a real part, those defects must be avoided, because the BTF ratio increases as the effective area decreases (
Table 5). Obtained tensile values are similar to the ones revealed for WAAM in other works [
8,
29,
35] and higher compared to the main providers’ material specifications. Moreover, taking into account the average of the obtained mechanical properties (
Table 12) for the different interpass dwell times, they reveal deviations of less than 3 Mpa, which means that the used interpass dwell times do not induce substantial differences in the microstructure apart from the appearance of segregations in the outer zones. In the work from Derekar et al. [
2], for the same material, a slight difference in mechanical properties was observed (4 Mpa) when comparing two interpass temperatures (50 °C and 100 °C), obtaining better properties for 100 °C of interpass temperature due to the larger grain size, and the lower amount and smaller pores. In this work, the interpass temperature is above 100 °C in all the cases, which can induce large grain size and the typical inhomogeneous microstructure due to the continuous thermal cycles of the process, and hence the mechanical properties are not significantly affected by the changes in the interpass dwell time. This was also observed in the work of Köhler et al. [
25].
Anisotropy in elongation is more evident at low interpass dwell times (27% for 90 s of interpass dwell time vs. 5% for 240 s of interpass dwell time) where horizontal orientation obtained higher results in terms of strength and elongation as found in other works [
25,
29,
35] and compared to the unitary torch [
8]. This is principally explained by the heterogeneity and defects between layers [
25]. The lower anisotropy has been obtained for the longest interpass dwell time (240 s), and for the unitary torch without a cooling plate (90 s). This suggests the importance of maintaining a low interpass temperature. In the case of manufacturing with CMT-Twin, it has been observed that there is a need for a cooling plate in addition to longer interpass dwell times to maintain a low interpass temperature. The optimised wall obtained 18% of anisotropy in elongation. This might be improved if a constant interpass temperature is maintained during the wall manufacturing.