Effect of Trajectory Curvature on the Microstructure and Properties of Surfacing Wall Formed with the Process of Wire Arc Additive Manufacturing

Curvature effects are typically present in the process of additive manufacturing (AM), particularly for wire arc additive manufacturing. In this paper, stainless-steel wire was adopted to deposit thin-walled samples with different curvatures. Optical microscopy, SEM, EDS and micro-hardness was used to analyse the microstructure, composition and properties of the samples. The result shows that the bottom region of the thin-walled sample had a mainly planar and cellular crystal microstructure. For the middle region, the microstructure revealed mainly dendrites, and the top layer has equiaxed dendrite morphology. The microhardness value of the bottom was greater than that of the middle, and the microhardness value of the middle was greater than that of the top. Moreover, the grain size of the inner part (direct to curvature radius) was larger than that of the outer part, and the micro-hardness value exhibited an increasing tendency from the inner to the outer side. With enlarging curvature, the degree of grain size differences and micro-hardness variants decreased. Finally, an investigation with a low carbon steel wire showed that it had a similar curvature effect for its AM specimen.


Introduction
Additive manufacturing (AM), also known as rapid forming technology, was proposed by Charles W. Hul in the 1980s [1]. "AM as a process of joining materials to make objects from 3D model data, usually layer upon layer, as opposed to subtractive manufacturing methodologies" [2]. Compared with traditional machining operations and other manufacturing technologies, additive manufacturing technology can quickly and accurately manufacture complex parts with high forming efficiency, which can save design and manufacturing costs, to a certain extent [3][4][5]. The heat sources used in additive metal manufacturing include lasers, electron beams, arcs, etc. [6][7][8][9][10][11][12]. Frequently used AM materials are Polymers, ceramics and metals [13,14]. Recently, AM has been widely used in various industries, such as aerospace. The AM process has been used to produce the central flange of the Chinese-made aircraft C919. The selective laser melting process is also used in the renovation and coating of precision parts [15]. The AM process plays an important role in Industry 4.0, because it saves time and has lower costs [16]. Compared to laser and electron beam AM, wire arc additive manufacturing (WAAM) equipment, with its low cost and high forming efficiency, is suitable for manufacturing large components [17][18][19].
In the process of arc additive manufacturing, it is complicated by the thermal effect of the arc pre-deposited adjacent region being reheated with subsequent welding passes [20,21]. Besides, the heat accumulation during the forming process has different effects on the microstructure and properties of formed parts. Therefore, in industrialized applications of WAAM technology, it is valuable to understand the influence of the arc thermal effect on the microstructure and performance of different zones of the formed parts. Liu et al. [22] prepared single-layer multilayer samples at different deposition rates. The results show that the mechanical properties could be much improved with a suitable deposition rate, because it has a finer and more uniform microstructure. Zhou et al. [23], studied the effect of an external magnetic field on the microstructure and mechanical properties of shaped specimens. They found that the microstructure was more uniform, and the mechanical properties were superior than when utilizing an external magnetic field. Ge et al. [24] studied the thermal history, microstructure and properties of formed parts in different locations, showing that the preheating effect from previously built layers can be effectively used to reduce residual stresses. The cooling rate firstly decreased rapidly, and then kept stable near the top layer. The parts can be a straight wall or a block, and they can also be curved. Thus, it is usually inevitable to have a curved traveling path for additive manufacturing. The presence of a curve may result in a difference in performance between the inside and outside of the curve. For example, under aerospace or high temperature and high-pressure conditions, it may be necessary to carry out some intensive treatment measures on the side with poor performance. Specific and in-depth analysis of the influence of the curvature of the path on the microstructure and properties of fabricated parts, is still lacking. Stainless-steel components with a complex structure are widely used in the industry. Meanwhile, WAAM provides a novel processing method option for them. This paper investigates the effect of trajectory curvature on the microstructure and mechanical properties of thin-walled samples with a 304 stainless-steel wire arc additive manufacturing process. Moreover, low carbon steel wire was used to verify the effect of curvature on WAAM structure and performance.

Experimental
The stainless-steel wire used in the experiment was ER304. The wire diameter was 1.0 mm, and the substrate material was Q235 steel plate. The composition of the welding wire and substrate material is shown in Table 1. The oxide film on the surface of the substrate plate was cleaned with mechanical methods, and then wiped with acetone before welding. The additive manufacturing system used here consisted mainly of the Megmid MIG500 welder (Shenzhen, China) and the ABB1410 robot (Zurich, Switzerland), as shown in Figure 1. The sample was reciprocally deposited without pre-heating before welding, and each pass had a fixed curvature. In the process of arc additive manufacturing, it is complicated by the thermal effect of the arc pre-deposited adjacent region being reheated with subsequent welding passes [20,21]. Besides, the heat accumulation during the forming process has different effects on the microstructure and properties of formed parts. Therefore, in industrialized applications of WAAM technology, it is valuable to understand the influence of the arc thermal effect on the microstructure and performance of different zones of the formed parts. Liu et al. [22] prepared single-layer multilayer samples at different deposition rates. The results show that the mechanical properties could be much improved with a suitable deposition rate, because it has a finer and more uniform microstructure. Zhou et al. [23], studied the effect of an external magnetic field on the microstructure and mechanical properties of shaped specimens. They found that the microstructure was more uniform, and the mechanical properties were superior than when utilizing an external magnetic field. Ge et al. [24] studied the thermal history, microstructure and properties of formed parts in different locations, showing that the preheating effect from previously built layers can be effectively used to reduce residual stresses. The cooling rate firstly decreased rapidly, and then kept stable near the top layer. The parts can be a straight wall or a block, and they can also be curved. Thus, it is usually inevitable to have a curved traveling path for additive manufacturing. The presence of a curve may result in a difference in performance between the inside and outside of the curve. For example, under aerospace or high temperature and high-pressure conditions, it may be necessary to carry out some intensive treatment measures on the side with poor performance. Specific and in-depth analysis of the influence of the curvature of the path on the microstructure and properties of fabricated parts, is still lacking. Stainless-steel components with a complex structure are widely used in the industry. Meanwhile, WAAM provides a novel processing method option for them. This paper investigates the effect of trajectory curvature on the microstructure and mechanical properties of thin-walled samples with a 304 stainless-steel wire arc additive manufacturing process. Moreover, low carbon steel wire was used to verify the effect of curvature on WAAM structure and performance.

Experimental
The stainless-steel wire used in the experiment was ER304. The wire diameter was 1.0 mm, and the substrate material was Q235 steel plate. The composition of the welding wire and substrate material is shown in Table 1. The oxide film on the surface of the substrate plate was cleaned with mechanical methods, and then wiped with acetone before welding. The additive manufacturing system used here consisted mainly of the Megmid MIG500 welder (Shenzhen, China) and the ABB1410 robot (Zurich, Switzerland), as shown in Figure 1. The sample was reciprocally deposited without pre-heating before welding, and each pass had a fixed curvature.  The path curvature of the experimental parts is shown in Figure 2. The sample was cut with a wire-cut electrical discharge machine (DK7750, Star peak, Taizhou, China). The metallographic etching solution for stainless-steel consisted of nitric acid, hydrofluoric acid and glycerine, and their volume ratio was 1:2:1. The sample was etched by swabbing with the solution until the boundary between the layers was clearly displayed. Next, the transverse hardness measurement was performed on the first, fifth and tenth layers, from the left (inside) side to the right (outside), using a Vickers hardness tester (THVS-1MDX-AXY, Qifeng, Chengdu, China). The conditions of the hardness test had a point interval of 0.03 mm, a load of 200 g, and a dwell time of 10 s. Finally, the sample was polished and etched again by swabbing, in order to observe the metallographic structured with a metallographic microscope (Olympus GX51, Tokyo, Japan). The path curvature of the experimental parts is shown in Figure 2. The sample was cut with a wire-cut electrical discharge machine (DK7750, Star peak, Taizhou, China). The metallographic etching solution for stainless-steel consisted of nitric acid, hydrofluoric acid and glycerine, and their volume ratio was 1:2:1. The sample was etched by swabbing with the solution until the boundary between the layers was clearly displayed. Next, the transverse hardness measurement was performed on the first, fifth and tenth layers, from the left (inside) side to the right (outside), using a Vickers hardness tester (THVS-1MDX-AXY, Qifeng, Chengdu, China). The conditions of the hardness test had a point interval of 0.03 mm, a load of 200 g, and a dwell time of 10 s. Finally, the sample was polished and etched again by swabbing, in order to observe the metallographic structured with a metallographic microscope (Olympus GX51, Tokyo, Japan). The traveling path of low carbon steel (ER50-6, the composition of the welding wire is shown in Table 1) was the same as the stainless-steel welding wire, in terms of deposited layers. The metallographic etching solution for low carbon steel contained 4% (v/v) nitric acid alcohol. At the first stage, the investigation was carried out with a large number of basic technological experiments, such as using single-pass single-layer and single-pass multi-layer. We compared and analysed the morphology (weld height, weld width, etc.) of the single-pass single-layer, and the depositing effect and the dimensional accuracy of the single-pass multi-layer forming test, under different technological parameters. Some single-layer single-pass weld shapes with various process parameters are shown in Figure 3. The process parameters were gradually adjusted to get a stable geometric weld for ER304 stainless-steel welding wire, and ER50-6 low carbon steel welding wire in this experiment, as shown in Table 2. The traveling path of low carbon steel (ER50-6, the composition of the welding wire is shown in Table 1) was the same as the stainless-steel welding wire, in terms of deposited layers. The metallographic etching solution for low carbon steel contained 4% (v/v) nitric acid alcohol. At the first stage, the investigation was carried out with a large number of basic technological experiments, such as using single-pass single-layer and single-pass multi-layer. We compared and analysed the morphology (weld height, weld width, etc.) of the single-pass single-layer, and the depositing effect and the dimensional accuracy of the single-pass multi-layer forming test, under different technological parameters. Some single-layer single-pass weld shapes with various process parameters are shown in Figure 3. The process parameters were gradually adjusted to get a stable geometric weld for ER304 stainless-steel welding wire, and ER50-6 low carbon steel welding wire in this experiment, as shown in Table 2.   Figure 4 shows a deposited sample produced by ER304 stainless-steel wire. It can be seen that the bead surface was smooth, and the overlap fusion between the welds was sound, without collapse.   Figure 4 shows a deposited sample produced by ER304 stainless-steel wire. It can be seen that the bead surface was smooth, and the overlap fusion between the welds was sound, without collapse.    Figure  4, of the curvature 0 sample, respectively. The white part in the figures is austenite, and the black part is ferrite. Meanwhile, from the crystal morphology, region A was mainly fine cell crystal. Region B was mainly coarse dendrites and had secondary dendrites. Besides, its dendrite spacing was significantly larger than that of region A. Region C is mainly composed of free-form equiaxial crystals.   Figure 4, of the curvature 0 sample, respectively. The white part in the figures is austenite, and the black part is ferrite. Meanwhile, from the crystal morphology, region A was mainly fine cell crystal. Region B was mainly coarse dendrites and had secondary dendrites. Besides, its dendrite spacing was significantly larger than that of region A. Region C is mainly composed of free-form equiaxial crystals.    Figure  4, of the curvature 0 sample, respectively. The white part in the figures is austenite, and the black part is ferrite. Meanwhile, from the crystal morphology, region A was mainly fine cell crystal. Region B was mainly coarse dendrites and had secondary dendrites. Besides, its dendrite spacing was significantly larger than that of region A. Region C is mainly composed of free-form equiaxial crystals. The ratio of the temperature gradient (G) to the growth rate (R) determines the mode of solidification, and the microstructure after solidification [25]. R = νcosα, and α is related to the distance to the centreline, here ν denotes the welding speed. In Figure 5, the positions of the bottom, middle and top metallographic structures are located on a vertical line, and the distances from the centre line are basically the same. It can be considered that the growth rates (R) of the three parts are similar, and that the value of G/R is approximately depended on G. Due to the heat dissipation of the substrate, the temperature gradient G at the bottom of the additive parts was maximized, and the G/R was also the largest. Hence, it was a mainly planar structure and composed of cellular crystal at the bottom, as shown in Figure 5a,b. As the height of the deposited layer was increased, the temperature gradient G in the middle was smaller than that in the bottom, and the G/R in the middle was also smaller than that of the bottom, due to the preheating effect of the previously formed layer. The middle was mainly dendrites, as shown in Figure 5c,d. Due to the heat accumulation of the formed portion, both the temperature gradient G and the G/R ratio of the top portion were the smallest. In addition, the heat dissipation direction of the molten pool at the top was changed, so the top portion was mainly equiaxed dendrite, as shown in Figure 5e,f. Figure 6a,b are the inside and outside metallographic structures of the fifth layer of the curvature 1 sample (position D and E in Figure 4b). Figure 6c,d are the inside and outside metallographic structures of the fifth layer of the curvature 2 sample. It could be seen that the dendrite spacing and the secondary dendrite spacing in the middle inside site were larger than that of the middle outside. The ratio of the temperature gradient (G) to the growth rate (R) determines the mode of solidification, and the microstructure after solidification [25]. R = νcosα, and α is related to the distance to the centreline, here ν denotes the welding speed. In Figure 5, the positions of the bottom, middle and top metallographic structures are located on a vertical line, and the distances from the centre line are basically the same. It can be considered that the growth rates (R) of the three parts are similar, and that the value of G/R is approximately depended on G. Due to the heat dissipation of the substrate, the temperature gradient G at the bottom of the additive parts was maximized, and the G/R was also the largest. Hence, it was a mainly planar structure and composed of cellular crystal at the bottom, as shown in Figure 5a,b. As the height of the deposited layer was increased, the temperature gradient G in the middle was smaller than that in the bottom, and the G/R in the middle was also smaller than that of the bottom, due to the preheating effect of the previously formed layer. The middle was mainly dendrites, as shown in Figure 5c,d. Due to the heat accumulation of the formed portion, both the temperature gradient G and the G/R ratio of the top portion were the smallest. In addition, the heat dissipation direction of the molten pool at the top was changed, so the top portion was mainly equiaxed dendrite, as shown in Figure 5e,f. Figure 6a,b are the inside and outside metallographic structures of the fifth layer of the curvature 1 sample (position D and E in Figure 4b). Figure 6c,d are the inside and outside metallographic structures of the fifth layer of the curvature 2 sample. It could be seen that the dendrite spacing and the secondary dendrite spacing in the middle inside site were larger than that of the middle outside. The ratio of the temperature gradient (G) to the growth rate (R) determines the mode of solidification, and the microstructure after solidification [25]. R = νcosα, and α is related to the distance to the centreline, here ν denotes the welding speed. In Figure 5, the positions of the bottom, middle and top metallographic structures are located on a vertical line, and the distances from the centre line are basically the same. It can be considered that the growth rates (R) of the three parts are similar, and that the value of G/R is approximately depended on G. Due to the heat dissipation of the substrate, the temperature gradient G at the bottom of the additive parts was maximized, and the G/R was also the largest. Hence, it was a mainly planar structure and composed of cellular crystal at the bottom, as shown in Figure 5a,b. As the height of the deposited layer was increased, the temperature gradient G in the middle was smaller than that in the bottom, and the G/R in the middle was also smaller than that of the bottom, due to the preheating effect of the previously formed layer. The middle was mainly dendrites, as shown in Figure 5c,d. Due to the heat accumulation of the formed portion, both the temperature gradient G and the G/R ratio of the top portion were the smallest. In addition, the heat dissipation direction of the molten pool at the top was changed, so the top portion was mainly equiaxed dendrite, as shown in Figure 5e,f. Figure 6a,b are the inside and outside metallographic structures of the fifth layer of the curvature 1 sample (position D and E in Figure 4b). Figure 6c,d are the inside and outside metallographic structures of the fifth layer of the curvature 2 sample. It could be seen that the dendrite spacing and the secondary dendrite spacing in the middle inside site were larger than that of the middle outside. For curved passes, the radius at the inside and outside sites is different [26], as shown in Figure 7. According to the test conditions, the width of the single-pass multi-layer additive was about 6mm, and the welding speed of the experiment was 5 mm/s. Thus, from the formula

Macroscopic Morphology and Microstructure of Stainless-steel Specimens
vo: actual welding speed of outside; vi: actual welding speed of inside; v: actual welding speed. it was concluded that for , the curvature 1 sample was equal to 5.3 mm/s, and was equal to 4.7 mm /s. Therefore, the actual welding traveling speed on the outside was higher than that of the inside. Under the same welding parameters, higher welding speeds on the outside made the heat input relatively small, and the cooling rate relatively high, resulting in a small pitch of dendrites on the outside, and a finer grain size. Meanwhile, it can be seen in Figure 6 that the difference in grain size between the inside and the outside of the curvature 2 sample is not apparent, compared to that of the curvature 1 sample. There were two reasons for this phenomenon. Firstly, in the process of forming a curved path for arc additive manufacturing, a magnetic field would be generated in the adjacent area. The magnetic field density in the inside is larger than that of the outside, which causes the direction of arc thrust to be directed to the outside [27]. The asymmetry of this force causes the asymmetry of the droplet [28], which results in a transport of heat input to the outside [29]. Whereas For curved passes, the radius at the inside and outside sites is different [26], as shown in Figure 7. According to the test conditions, the width of the single-pass multi-layer additive was about 6mm, and the welding speed of the experiment was 5 mm/s. Thus, from the formula For curved passes, the radius at the inside and outside sites is different [26], as shown in Figure 7. According to the test conditions, the width of the single-pass multi-layer additive was about 6mm, and the welding speed of the experiment was 5 mm/s. Thus, from the formula vo: actual welding speed of outside; vi: actual welding speed of inside; v: actual welding speed. it was concluded that for , the curvature 1 sample was equal to 5.3 mm/s, and was equal to 4.7 mm /s. Therefore, the actual welding traveling speed on the outside was higher than that of the inside. Under the same welding parameters, higher welding speeds on the outside made the heat input relatively small, and the cooling rate relatively high, resulting in a small pitch of dendrites on the outside, and a finer grain size. Meanwhile, it can be seen in Figure 6 that the difference in grain size between the inside and the outside of the curvature 2 sample is not apparent, compared to that of the curvature 1 sample. There were two reasons for this phenomenon. Firstly, in the process of forming a curved path for arc additive manufacturing, a magnetic field would be generated in the adjacent area. The magnetic field density in the inside is larger than that of the outside, which causes the direction of arc thrust to be directed to the outside [27]. The asymmetry of this force causes the asymmetry of the droplet [28], which results in a transport of heat input to the outside [29]. Whereas Therefore, the actual welding traveling speed on the outside was higher than that of the inside. Under the same welding parameters, higher welding speeds on the outside made the heat input relatively small, and the cooling rate relatively high, resulting in a small pitch of dendrites on the outside, and a finer grain size. Meanwhile, it can be seen in Figure 6 that the difference in grain size between the inside and the outside of the curvature 2 sample is not apparent, compared to that of the curvature 1 sample. There were two reasons for this phenomenon. Firstly, in the process of forming a curved path for arc additive manufacturing, a magnetic field would be generated in the adjacent area. The magnetic field density in the inside is larger than that of the outside, which causes the direction of arc thrust to be directed to the outside [27]. The asymmetry of this force causes the asymmetry of the droplet [28], which results in a transport of heat input to the outside [29]. Whereas the effect of heat input, caused by the difference in speed between the inside and outside, was offset by the magnetic arc blow. Secondly, a higher effect of magnetic arc blow could be obtained by the larger curvature. Figure 8 shows the microhardness distribution result. The hardness values of the first and fifth layers were mostly distributed above 200 HV, which were higher than that of the 304 stainless-steel substrate. This indicates that the microstructure of the first and fifth layers was denser and more uniform than that of the 304 stainless-steel substrate. Meanwhile, the hardness of the tenth layer was rather less than 200 HV, which was caused by its slightly lower compactness, than that of 304 stainless-steel. From the overall observation and analysis, the hardness of the first layer was greater than that of the fifth and tenth layers. Moreover, the fifth layer was slightly harder than that of the tenth layer. Generally, hardness is related to the grain size, carbide particle size and their distribution. Due to the conventional heat dissipation of the substrate, the bottom layer of weld metal had the fastest cooling rate, the smallest weld grain, and the highest micro-hardness. Figures 9 and 10 show the elemental distribution mapping of the middle layer and top layer with EDS analysis. It can be seen that the concentration of Cr and C in the grain boundary position was slightly higher than that in the inner region of the grain. The trend of the top layer was more obvious than that of the middle layer. This elemental distribution originates from the relatively slow cooling rate of the top layer. It had sufficient time to precipitate more carbides at the grain boundaries, and cause the carbon content in the inner region of the grains to be relatively small, which slightly decreased the micro-hardness value. However, in the middle layer, the cooling rate was slightly faster, causing more carbide particles to remain within the grains, and less precipitation to the grain boundaries. Hence, the hardness of the middle layer was slightly higher than that of the top layer. the effect of heat input, caused by the difference in speed between the inside and outside, was offset by the magnetic arc blow. Secondly, a higher effect of magnetic arc blow could be obtained by the larger curvature. Figure 8 shows the microhardness distribution result. The hardness values of the first and fifth layers were mostly distributed above 200 HV, which were higher than that of the 304 stainless-steel substrate. This indicates that the microstructure of the first and fifth layers was denser and more uniform than that of the 304 stainless-steel substrate. Meanwhile, the hardness of the tenth layer was rather less than 200 HV, which was caused by its slightly lower compactness, than that of 304 stainless-steel. From the overall observation and analysis, the hardness of the first layer was greater than that of the fifth and tenth layers. Moreover, the fifth layer was slightly harder than that of the tenth layer. Generally, hardness is related to the grain size, carbide particle size and their distribution. Due to the conventional heat dissipation of the substrate, the bottom layer of weld metal had the fastest cooling rate, the smallest weld grain, and the highest micro-hardness. Figures 9 and 10 show the elemental distribution mapping of the middle layer and top layer with EDS analysis. It can be seen that the concentration of Cr and C in the grain boundary position was slightly higher than that in the inner region of the grain. The trend of the top layer was more obvious than that of the middle layer. This elemental distribution originates from the relatively slow cooling rate of the top layer. It had sufficient time to precipitate more carbides at the grain boundaries, and cause the carbon content in the inner region of the grains to be relatively small, which slightly decreased the micro-hardness value. However, in the middle layer, the cooling rate was slightly faster, causing more carbide particles to remain within the grains, and less precipitation to the grain boundaries. Hence, the hardness of the middle layer was slightly higher than that of the top layer.     Figure 11 shows the lateral hardness distribution from inside to outside. The curvature 0 sample was relatively even in terms of hardness distribution, while both the curvature 1 and curvature 2 samples had a rising trend from inside to outside. Besides, it can be seen that the trend of the curvature 2 sample tended to be gentler than that of the curvature 1 sample, as the difference in hardness between the two sides of the curvature 2 sample was smaller than that of the curvature 1 sample. This result also corresponds to the metallographic microstructure shown in Figure 6.  Figure 11 shows the lateral hardness distribution from inside to outside. The curvature 0 sample was relatively even in terms of hardness distribution, while both the curvature 1 and curvature 2 samples had a rising trend from inside to outside. Besides, it can be seen that the trend of the curvature 2 sample tended to be gentler than that of the curvature 1 sample, as the difference in hardness between the two sides of the curvature 2 sample was smaller than that of the curvature 1 sample. This result also corresponds to the metallographic microstructure shown in Figure 6.

Microstructure of the Low Carbon Steel Sample
The metallographic structure of the WAAM low carbon steel sample is shown in Figure 12. When the curvature was small, the difference of grain size between inside and outside was larger. When the curvature increased, the difference of grain size between inside and outside decreased, but the grain size inside was still larger than that found in the outside. The reason that the structure and properties of additive manufacturing parts were affected by curvature, was considered to be the difference in actual welding traveling speed between the inside and outside. With the increasing of curvature, the enhanced effect of the magnetic arc blow reduced the difference in grain size between the inside and outside, owing to a difference in velocity.

Microstructure of the Low Carbon Steel Sample
The metallographic structure of the WAAM low carbon steel sample is shown in Figure 12. When the curvature was small, the difference of grain size between inside and outside was larger. When the curvature increased, the difference of grain size between inside and outside decreased, but the grain size inside was still larger than that found in the outside. The reason that the structure and properties of additive manufacturing parts were affected by curvature, was considered to be the difference in actual welding traveling speed between the inside and outside. With the increasing of curvature, the enhanced effect of the magnetic arc blow reduced the difference in grain size between the inside and outside, owing to a difference in velocity.

Microstructure of the Low Carbon Steel Sample
The metallographic structure of the WAAM low carbon steel sample is shown in Figure 12. When the curvature was small, the difference of grain size between inside and outside was larger. When the curvature increased, the difference of grain size between inside and outside decreased, but the grain size inside was still larger than that found in the outside. The reason that the structure and properties of additive manufacturing parts were affected by curvature, was considered to be the difference in actual welding traveling speed between the inside and outside. With the increasing of curvature, the enhanced effect of the magnetic arc blow reduced the difference in grain size between the inside and outside, owing to a difference in velocity.

Conclusions
This investigation prepared stainless-steel and carbon steel thin-walled specimens with different curvatures, by a wire arc additive manufacturing method. Studies shows that trajectory curvature has an apparent influence on the microstructure and performance of thin-walled samples, in different local regions, which can be summarized as:

•
The top layer has a slow cooling rate. More carbides particles are precipitated at the grain boundaries, rather than within the grains, and the microhardness value is slightly lower than that of the middle layer. • Different trajectory curvature leads to a difference in microstructure and properties for inner and outer regions, which originates from the difference in heat input, due to the difference in the actual welding traveling speed between the inside and outside regions.

•
As curvature increases, the enhanced effect of the magnetic arc blow reduces the difference in heat input, due to the difference in actual welding traveling speed between the inside and outside areas.